Browse the corpus

Background Description of the condition Clinical Features Trachoma is a form of chronic conjunctivitis caused by Chlamydia trachomatis. Following repeated infections, the upper tarsal conjunctiva becomes scarred. As the scar tissue contracts, it shortens the posterior lamella (inner surface) of the upper lid causing the eyelashes to turn in (entropion) and rub against the transparent cornea. This contact between one or more lashes and the surface of the eye is called trichiasis. Trichiasis due to trachoma has a wide spectrum of severity from a single lash touching the eye through to the entire upper lid being rolled in (Rajak 2011). It may result from misdirected or metaplastic eyelashes, in the absence of frank entropion (Rajak 2011). Corneal opacification and the resulting blindness probably develop primarily as a result of this trauma and secondary bacterial corneal infection.

he eye through to the entire upper lid being rolled in (Rajak 2011). It may result from misdirected or metaplastic eyelashes, in the absence of frank entropion (Rajak 2011). Corneal opacification and the resulting blindness probably develop primarily as a result of this trauma and secondary bacterial corneal infection. Epidemiology The most recent World Health Organization (WHO) estimates suggest that about 40 million people have active trachoma and about 8.2 million have trichiasis (Mariotti 2009). The signs of active trachoma are usually most frequently found in young children, with males and females equally affected. The scarring complications become evident in later life. Women are usually more frequently affected by trichiasis than men (West 1991). The rate at which the scarring complications of the disease progress varies, possibly reflecting variation in the pressure of C. trachomatis infection and immunogenetic predisposition of different populations (Hu 2013). Trachoma is most prevalent in hot, dry areas and is associated with poverty (Emerson 2000). The greatest burden of disease is in sub‐Saharan Africa.

the disease progress varies, possibly reflecting variation in the pressure of C. trachomatis infection and immunogenetic predisposition of different populations (Hu 2013). Trachoma is most prevalent in hot, dry areas and is associated with poverty (Emerson 2000). The greatest burden of disease is in sub‐Saharan Africa. Description of the intervention Trachoma control The eradication of blinding trachoma is one of the objectives of the global Vision 2020 programme to eliminate avoidable blindness, led by the WHO and the International Agency for the Prevention of Blindness. In 1997 the WHO launched an initiative on trachoma control based on the 'SAFE' strategy. SAFE stands for Surgery for trichiasis, Antibiotics, Facial cleanliness, and Environmental improvement. Improved facial cleanliness and environmental hygiene are effective at reducing transmission by removing the conditions that promote spread of the disease. Antibiotics reduce the risk of disease transmission by treating the infectious agent. Surgery to correct the lid deformity is the only treatment that is likely to be beneficial in the late stages of the disease; however, once corneal opacification has occurred, management options are very limited. Cochrane reviews of the optimum antibiotic regimen for trachoma (Evans 2011), environmental sanitation (Rabiu 2012), and face‐washing promotion (Ejere 2015), are published on the Cochrane Library.

l in the late stages of the disease; however, once corneal opacification has occurred, management options are very limited. Cochrane reviews of the optimum antibiotic regimen for trachoma (Evans 2011), environmental sanitation (Rabiu 2012), and face‐washing promotion (Ejere 2015), are published on the Cochrane Library. Trichiasis treatment options The primary aim of treatment for trichiasis is to prevent blindness due to trauma from the lashes abrading the cornea (Rajak 2012). Treatments may be divided into non‐surgical treatments and surgical treatments: Non‐surgical treatments epilation (manual removal of eyelashes, usually with forceps); eyelid‐taping (to force eyelashes back to correct position) Surgical treatments Surgical procedures for lash ablation or removal: electrolysis (fine needle used to pass electric current to base of lash follicle); cryotherapy (freezing treatment to the lash follicles) excision of lash‐bearing tissue A wide variety of surgical options are available for the treatment of upper lid entropion (Rajak 2012), and it is likely that certain operations are more successful than others. In trachoma‐endemic countries the most commonly used procedures are: bilamellar tarsal rotation (BLTR): full‐thickness incision through the eyelid, including the scarred tarsal plate, orbicularis oculi and the skin, fixation with everting sutures; posterior lamellar tarsal rotation (PLTR)/Trabut: incision through the scarred tarsal plate and conjunctiva only, leaving the skin and orbicularis oculi intact, fixation with everting sutures;

bilamellar tarsal rotation (BLTR): full‐thickness incision through the eyelid, including the scarred tarsal plate, orbicularis oculi and the skin, fixation with everting sutures; posterior lamellar tarsal rotation (PLTR)/Trabut: incision through the scarred tarsal plate and conjunctiva only, leaving the skin and orbicularis oculi intact, fixation with everting sutures; tarsal advance and rotation: incision of the tarsal plate and rotation of the terminal portion. The upper part of the tarsus is separated from the anterior lamellar, advanced and fixed with sutures. These techniques are illustrated in the WHO's publication ‐ Trichiasis surgery for trachoma (WHO 2013). How the intervention might work The lash treatments directly remove or ablate the follicles. They do not correct any underlying anatomical abnormality such as entropion and are, therefore, generally not suitable if there is a significant degree of entropion. The lid rotation procedures correct the underlying entropion by an incision through the scarred tissue, outward turning of the lid and fixation with sutures. Generally, this will correct the trichiasis in most cases, although in more severe cases this might be insufficient. Post‐operative trichiasis can occur either through inadequate surgery or progressive scarring disease.

ion by an incision through the scarred tissue, outward turning of the lid and fixation with sutures. Generally, this will correct the trichiasis in most cases, although in more severe cases this might be insufficient. Post‐operative trichiasis can occur either through inadequate surgery or progressive scarring disease. Why it is important to do this review Evidence from case series and randomised controlled trials suggests that upper lid surgery can be successful at abolishing trichiasis; however, typically 20% to 40% of eyelids suffer from post‐operative trichiasis by one year (Bog 1993; Bowman 2000; Burton 2005b; Rajak 2011b; Reacher 1990; Reacher 1992; Ward 2005; West 2005); and up to 60% at three years (Khandekar 2001). Therefore, it is important to understand the determinants of a good treatment outcome and identify strategies that lead to this. Risk factors associated with post‐operative trichiasis include the severity of pre‐operative trichiasis (Alemayehu 2004; Burton 2005a; Rajak 2013); chlamydial infection (Zhang 2004); inflammation of the tarsal conjunctiva (Burton 2005a; Burton 2005b; Ward 2005); bacterial infection (Burton 2005a); and left eyes (Merbs 2005; West 2005).

ciated with post‐operative trichiasis include the severity of pre‐operative trichiasis (Alemayehu 2004; Burton 2005a; Rajak 2013); chlamydial infection (Zhang 2004); inflammation of the tarsal conjunctiva (Burton 2005a; Burton 2005b; Ward 2005); bacterial infection (Burton 2005a); and left eyes (Merbs 2005; West 2005). The choice of treatment will depend on factors such as available resources and expertise, location (opportunity for follow up) and how advanced the disease is. The WHO strategy for the control of blinding trachoma calls for lid surgery to be delivered by ophthalmic assistants as well as ophthalmologists (Habtamu 2011), because the numbers of ophthalmologists are insufficient to provide the service. Ophthalmic assistants are usually taught to perform only one type of operation so it is important to ensure that the operation they use is known to be effective. Although surgery generally produces good results, in many settings only a minority of patients with trichiasis will attend for surgery (Bowman 2002; Courtright 1994; West 1994). Delivery of surgery in the community or non‐surgical management of trichiasis may be more acceptable than surgery in a conventional hospital setting (Bowman 2000; Graz 1999; Rajak 2011a). Objectives To assess the effects of interventions for trachomatous trichiasis for people living in endemic settings.

ies, participants with bilateral disease were randomised by eye rather than by individual participant (details below); these studies were included on the basis that the majority of participants were randomised individually. Details of the included studies can be found in the 'Characteristics of included studies' table. For ease of interpretation, studies have been grouped according to the aspect of trichiasis management they address: Surgical technique: Adamu 2002, Dhaliwal 2005, Gower 2013, Reacher 1990 and Reacher 1992 compared different surgical interventions. Epilation: Rajak 2011a compared epilation to surgery Lid taping: Graz 1999 examined lid taping versus epilation for managing trichiasis. Antibiotic treatment: Burton 2005a, West 2006 and Zhang 2006 examined whether peri‐operative azithromycin treatment reduced post‐operative trichiasis versus another or no treatment. Alternative suture materials: Rajak 2011b compared absorbable to non‐absorbable sutures. Surgery setting: Bowman 2000 investigated alternative settings for conducting surgery (health centre versus village). Personnel performing surgery: Alemayehu 2004 compared the outcome of surgery performed by different types of health care personnel. Types of interventions Surgical technique Five studies compared different surgical interventions for trichiasis: Adamu 2002; Dhaliwal 2005; Gower 2013; Reacher 1990 and Reacher 1992.

Although surgery generally produces good results, in many settings only a minority of patients with trichiasis will attend for surgery (Bowman 2002; Courtright 1994; West 1994). Delivery of surgery in the community or non‐surgical management of trichiasis may be more acceptable than surgery in a conventional hospital setting (Bowman 2000; Graz 1999; Rajak 2011a). Objectives To assess the effects of interventions for trachomatous trichiasis for people living in endemic settings. Methods Criteria for considering studies for this review Types of studies We included randomised controlled trials of interventions for trachomatous trichiasis. The unit of randomisation was individuals or clusters, depending on the design of the study. Types of participants Participants in the trials were people with trachomatous trichiasis, defined as one or more eye lashes touching the globe when looking straight ahead. Types of interventions We included trials in which any intervention intended to prevent corneal opacification from prolonged lash‐globe contact was compared to another intervention or to no treatment. Surgical interventions were procedures to correct entropion or ablate the lash roots and non‐surgical interventions were taping of the lid margin or manual removal of the eyelashes (epilation). We included trials that compared: different surgical or non‐surgical interventions; medication to reduce post‐operative trichiasis; any intervention delivered in a hospital setting to the same intervention in a community setting; the same intervention delivered by different types of health care professionals.

Types of interventions We included trials in which any intervention intended to prevent corneal opacification from prolonged lash‐globe contact was compared to another intervention or to no treatment. Surgical interventions were procedures to correct entropion or ablate the lash roots and non‐surgical interventions were taping of the lid margin or manual removal of the eyelashes (epilation). We included trials that compared: different surgical or non‐surgical interventions; medication to reduce post‐operative trichiasis; any intervention delivered in a hospital setting to the same intervention in a community setting; the same intervention delivered by different types of health care professionals. Types of outcome measures Primary outcomes The primary outcome measure for this review was post‐operative trichiasis. This dichotomous outcome was defined as any lash touching the globe in the primary position. The critical points for follow‐up were three months, one year, and two years after treatment. Secondary outcomes Secondary measures were: Visual acuity change Measured by Snellen or logMAR visual acuity charts at one year and two years after treatment. Corneal opacification change Measured by clinical examination or photographic comparisons at one year and two years after treatment. Acceptance of treatment As measured by uptake/attendance for treatment. Adverse effects Any adverse effects, whether minor or severe, were recorded. Quality of life Any qualitative measures of discomfort/patient satisfaction were noted.

Corneal opacification change Measured by clinical examination or photographic comparisons at one year and two years after treatment. Acceptance of treatment As measured by uptake/attendance for treatment. Adverse effects Any adverse effects, whether minor or severe, were recorded. Quality of life Any qualitative measures of discomfort/patient satisfaction were noted. Economic evaluation Where any cost data for interventions were available this was noted and commented on in the context of cost‐effectiveness. No formal cost‐effectiveness evaluation was planned. Search methods for identification of studies Electronic searches We searched CENTRAL (which contains the Cochrane Eyes and Vision Group Trials Register) (2015, Issue 4), Ovid MEDLINE, Ovid MEDLINE In‐Process and Other Non‐Indexed Citations, Ovid MEDLINE Daily, Ovid OLDMEDLINE (January 1946 to May 2015), EMBASE (January 1980 to May 2015), the ISRCTN registry (www.isrctn.com/editAdvancedSearch), ClinicalTrials.gov (www.clinicaltrials.gov) and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (www.who.int/ictrp/search/en). We did not use any date or language restrictions in the electronic searches for trials. We last searched the electronic databases on 7 May 2015. See: Appendices for details of search strategies for CENTRAL (Appendix 1), MEDLINE (Appendix 2), EMBASE (Appendix 3), ISRCTN (Appendix 4), ClinicalTrials.gov (Appendix 5) and the ICTRP (Appendix 6).

Search methods for identification of studies Electronic searches We searched CENTRAL (which contains the Cochrane Eyes and Vision Group Trials Register) (2015, Issue 4), Ovid MEDLINE, Ovid MEDLINE In‐Process and Other Non‐Indexed Citations, Ovid MEDLINE Daily, Ovid OLDMEDLINE (January 1946 to May 2015), EMBASE (January 1980 to May 2015), the ISRCTN registry (www.isrctn.com/editAdvancedSearch), ClinicalTrials.gov (www.clinicaltrials.gov) and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (www.who.int/ictrp/search/en). We did not use any date or language restrictions in the electronic searches for trials. We last searched the electronic databases on 7 May 2015. See: Appendices for details of search strategies for CENTRAL (Appendix 1), MEDLINE (Appendix 2), EMBASE (Appendix 3), ISRCTN (Appendix 4), ClinicalTrials.gov (Appendix 5) and the ICTRP (Appendix 6). Searching other resources We contacted experts and researchers in the field to ask them for details of published, unpublished or ongoing trials. We searched the reference lists of relevant trials. Data collection and analysis Selection of studies Two review authors assessed the titles and abstracts resulting from the searches and selected all titles that referred to treatment for trachomatous trichiasis. Copies of possibly relevant trials were obtained and independently assessed by three review authors according to the 'Criteria for considering studies for this review'. Trials meeting these criteria were also assessed for quality.

searches and selected all titles that referred to treatment for trachomatous trichiasis. Copies of possibly relevant trials were obtained and independently assessed by three review authors according to the 'Criteria for considering studies for this review'. Trials meeting these criteria were also assessed for quality. Data extraction and management We recorded data from included studies in a table under the following headings: methods ‐ including randomisation, intention‐to‐treat analysis; participants ‐ including cluster or individual, country, number; losses to follow up; interventions ‐ including types of surgery or other intervention, setting of intervention (community or clinic); outcomes ‐ including definitions of success or failure, visual acuity change, adverse effects. Authors independently extracted outcome data. One author entered data into RevMan 5, which was checked by a second author. Assessment of risk of bias in included studies Three review authors assessed trial quality according to methods set out in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) using the Cochrane Eyes and Vision Group Review Development Guidelines. In particular, random sequence generation, allocation concealment, performance bias, detection bias (masking of outcome graders to initial treatment), attrition bias (adequacy of follow up) and reporting bias were assessed. Risk of bias was graded as "Low Risk", "High Risk" and "Unclear Risk", according to the guidelines in Section 8 of the Handbook. We resolved disagreements between the authors by discussion.

n bias (masking of outcome graders to initial treatment), attrition bias (adequacy of follow up) and reporting bias were assessed. Risk of bias was graded as "Low Risk", "High Risk" and "Unclear Risk", according to the guidelines in Section 8 of the Handbook. We resolved disagreements between the authors by discussion. Measures of treatment effect The treatment effects were measured by calculating the odds ratio and 95% confidence interval for these. Unit of analysis issues The unit of analysis was generally the individual patient, with the exception of one cluster randomised trial. Dealing with missing data Studies were assessed for missing data and whether missing data were "missing at random". In one instance, data were found to be systematically excluded from the analysis of the primary outcome in the original report (recurrence occurring by three months after surgery). As sufficient information was presented in the report it was possible to reintroduce these primary outcome events here. Assessment of heterogeneity Heterogeneity between studies was assessed for those investigating the use of post‐operative antibiotic. Heterogeneity was considered in several categories: clinical, methodological and statistical. Assessment of reporting biases The studies were assessed in terms of the completeness of data presented.

Assessment of heterogeneity Heterogeneity between studies was assessed for those investigating the use of post‐operative antibiotic. Heterogeneity was considered in several categories: clinical, methodological and statistical. Assessment of reporting biases The studies were assessed in terms of the completeness of data presented. Data synthesis The interventions tested were varied and there was considerable heterogeneity. We present a descriptive summary of the results rather than a single summary statistic. Where the unit of randomisation was a cluster rather than an individual, data were analysed by cluster. The principal outcome measure was post‐operative trichiasis, which was defined as any lash touching the globe in the primary position. Odds ratios were calculated for the different interventions. Visual acuity data were presented as dichotomous data—statistically significant improvement or no improvement from pre‐ to post‐intervention. Data on adverse effects and acceptance of surgery were also presented as dichotomous data. The exception to this was the three studies examining the effect of azithromycin on the outcome of surgery. A meta‐analysis was performed for the dichotomous outcome of post‐operative trichiasis using a Mantel‐Haenszel, random‐effects model. For trials where there were more than two arms these were considered separately in relation to the reference intervention where appropriate, such as different surgical options. However, if two arms utilised the same intervention, they were combined.

st‐operative trichiasis using a Mantel‐Haenszel, random‐effects model. For trials where there were more than two arms these were considered separately in relation to the reference intervention where appropriate, such as different surgical options. However, if two arms utilised the same intervention, they were combined. Subgroup analysis and investigation of heterogeneity No sub‐group analysis was performed. Sensitivity analysis No sensitivity analysis was performed. Results Description of studies Results of the search The search of electronic databases revealed a total of 656 reports. We retrieved nine papers for further assessment. All of these were randomised trials of interventions for trachoma trichiasis (Adamu 2002; Alemayehu 2004; Bowman 2000; Burton 2005a; Dhaliwal 2005; Graz 1999; Reacher 1990; Reacher 1992; West 2006).

arch The search of electronic databases revealed a total of 656 reports. We retrieved nine papers for further assessment. All of these were randomised trials of interventions for trachoma trichiasis (Adamu 2002; Alemayehu 2004; Bowman 2000; Burton 2005a; Dhaliwal 2005; Graz 1999; Reacher 1990; Reacher 1992; West 2006). An update search run in May 2015 identified 654 new records (Figure 1). The Trials Search Co‐ordinator removed 190 duplicate records, screened 464 records and removed 280 references that were not relevant to the scope of the review. We screened the remaining 184 references and discarded 176 reports as not relevant. We obtained eight full‐text reports of trials for assessment. We included four reports of four new studies (Gower 2013; Rajak 2011a; Rajak 2011b; Zhang 2006); and added a new report to Burton 2005a. Previously a study by West 2006 was awaiting assessment: this study has now been included and three new reports of the trial have been added to the review. In the previous version of this review Dhaliwal 2005 was excluded; however we have re‐assessed this study and have now included it in the review.Figure 1 Study flow diagram. Included studies In some of the included studies, participants with bilateral disease were randomised by eye rather than by individual participant (details below); these studies were included on the basis that the majority of participants were randomised individually. Details of the included studies can be found in the 'Characteristics of included studies' table.

Surgery setting: Bowman 2000 investigated alternative settings for conducting surgery (health centre versus village). Personnel performing surgery: Alemayehu 2004 compared the outcome of surgery performed by different types of health care personnel. Types of interventions Surgical technique Five studies compared different surgical interventions for trichiasis: Adamu 2002; Dhaliwal 2005; Gower 2013; Reacher 1990 and Reacher 1992. Reacher 1990 randomly allocated individuals with major trichiasis (six or more lashes touching) to one of five operations: (1) bilamellar tarsal rotation; (2) tarsal advance and rotation; (3) eversion splinting; (4) tarsal advance; or (5) tarsal grooving. Minor trichiasis cases were excluded. Eyes with defective lid closure were also excluded from the study and all received a tarsal advance procedure. Surgery was performed by one of three surgeons. In the Reacher 1992 study participants were again grouped (stratified) according to severity and then randomly allocated: Minor trichiasis – allocated to (1) electrolysis; (2) cryotherapy; or (3) bilamellar tarsal rotation. Major trichiasis – allocation to either (1) bilamellar tarsal rotation; or (2) tarsal advance and rotation. Defective lid closure – allocation to (1) tarsal advance and rotation; or (2) tarsal advance with buccal mucosal membrane graft. Adamu 2002 compared bilamellar tarsal rotation and posterior lamellar tarsal rotation in patients with minor trichiasis, major trichiasis or defective lid closure with trichiasis. Surgery was performed by second‐year ophthalmic residents according to standardised procedures.

Defective lid closure – allocation to (1) tarsal advance and rotation; or (2) tarsal advance with buccal mucosal membrane graft. Adamu 2002 compared bilamellar tarsal rotation and posterior lamellar tarsal rotation in patients with minor trichiasis, major trichiasis or defective lid closure with trichiasis. Surgery was performed by second‐year ophthalmic residents according to standardised procedures. Dhaliwal 2005 compared three operations: (1) terminal tarsal rotation (a variant of the posterior tarsal rotation); (2) tarsal advance and rotation; and (3) tarsal grooving. One surgeon performed all of the procedures. Gower 2013 compared standard bilamellar tarsal rotation surgery to the same procedure performed using an eyelid clamp. Surgery was performed by 18 surgical technicians, who were trained and certified to perform only one form of the operation. Patients were randomly allocated to the surgeon. Epilation Rajak 2011a investigated whether epilation was non‐inferior to posterior lamellar rotation surgery for minor trichiasis. Patients with minor trichiasis were randomised to either immediate surgery or repeated epilation with manufactured epilating forceps. Lid taping Graz 1999 compared manual removal of eyelashes (epilation) with the use of a double‐sided sticking plaster to force eyelashes away from contact with the globe: both interventions were undertaken prior to lid surgery. There were three groups: (1) epilation alone; (2) sticking plaster alone; and (3) sticking plaster for eight weeks then crossover to epilation.

lashes (epilation) with the use of a double‐sided sticking plaster to force eyelashes away from contact with the globe: both interventions were undertaken prior to lid surgery. There were three groups: (1) epilation alone; (2) sticking plaster alone; and (3) sticking plaster for eight weeks then crossover to epilation. Antibiotic treatment Three separate trials have examined whether post‐operative oral azithromycin can reduce post‐operative trichiasis in three different countries: (1) The Gambia (Burton 2005a; (2) Ethiopia (West 2006); and (3) Nepal (Zhang 2006). In the Gambian study all participants underwent posterior lamellar tarsal rotation and were prescribed tetracycline eye ointment twice daily for two weeks (Burton 2005a). Those randomised to the intervention group received a 1 g dose of azithromycin at the time of surgery; adults and children in these households were also given a single dose of azithromycin (adults 1 g and children 20 mg/kg) to reduce the risk of re‐infection. This medication was re‐administered at six months. In the Ethiopian study, participants all underwent bilamellar tarsal rotation surgery (West 2006). They were then randomised to one of three intervention arms: (1) 1g dose of azithromycin for the patient alone; (2) 1 g dose of azithromycin for the patient and single‐dose azithromycin treatment for household members; (3) topical tetracycline (twice per day for six weeks). In the Nepal study, participants all received bilamellar tarsal rotation surgery (Zhang 2006). At the end of surgery they were alternately given either azithromycin or a placebo.

In the Ethiopian study, participants all underwent bilamellar tarsal rotation surgery (West 2006). They were then randomised to one of three intervention arms: (1) 1g dose of azithromycin for the patient alone; (2) 1 g dose of azithromycin for the patient and single‐dose azithromycin treatment for household members; (3) topical tetracycline (twice per day for six weeks). In the Nepal study, participants all received bilamellar tarsal rotation surgery (Zhang 2006). At the end of surgery they were alternately given either azithromycin or a placebo. Alternative suture materials Rajak 2011b compared posterior lamellar rotation performed with silk sutures to the same procedure performed with absorbable polyglactin‐910 sutures. Surgery setting Bowman 2000 compared providing surgery in the participants' own village to surgery provided at the nearest health centre. Posterior lamellar tarsal rotation surgery was performed on all participants by one of five trained nurses or an ophthalmic assistant. Personnel performing surgery Alemayehu 2004 compared post‐operative trichiasis rates following surgery by ophthalmologists and non‐ophthalmologist integrated eye care workers (IECW) in Ethiopia. Subjects with trachomatous trichiasis (TT) were randomised to surgery by either an ophthalmologist or an IECW. Both groups used the bilamellar tarsal rotation procedure.

compared post‐operative trichiasis rates following surgery by ophthalmologists and non‐ophthalmologist integrated eye care workers (IECW) in Ethiopia. Subjects with trachomatous trichiasis (TT) were randomised to surgery by either an ophthalmologist or an IECW. Both groups used the bilamellar tarsal rotation procedure. Types of participants Surgical technique Participants in the Reacher 1990 study were 165 adult Omanis with TT. Participants were grouped according to severity: minor trichiasis, major trichiasis or defective lid closure as defined above. Only those with major trichiasis were eligible for randomisation. Participants who had undergone previous treatment were included. Reacher 1992 recruited 367 individuals diagnosed with TT by the Oman Prevention of Blindness Program. Trichiasis was graded as for the Reacher 1990 paper, and participants were grouped as minor trichiasis, major trichiasis or defective lid closure. Participants who had undergone previous treatment were included. In the study by Adamu 2002, participants were 153 consecutive patients with TT presenting at a teaching hospital in Addis Ababa. Eight children (< 15 years) were included. All participants were graded pre‐operatively as having minor trichiasis, major trichiasis or defective lid closure. Previously operated eyes were excluded from the study. Dhaliwal 2005 recruited 77 consecutive patients (90 eyes) in an eye clinic in India.

In the study by Adamu 2002, participants were 153 consecutive patients with TT presenting at a teaching hospital in Addis Ababa. Eight children (< 15 years) were included. All participants were graded pre‐operatively as having minor trichiasis, major trichiasis or defective lid closure. Previously operated eyes were excluded from the study. Dhaliwal 2005 recruited 77 consecutive patients (90 eyes) in an eye clinic in India. In Gower 2013, which was conducted in Tanzania, 1917 participants were enrolled through screening campaigns. Participants were at least 18 years old, had previously unoperated trichiasis and did not plan to move within 2 years. All degrees of trichiasis severity were eligible. Epilation Rajak 2011a was conducted in Ethiopia. Only adults (> 18 years) with minor trichiasis were eligible. 1300 individuals were identified and recruited through community outreach campaigns; 66% were female. At baseline by chance there was slightly more corneal disease in the epilation arm. Lid taping Graz 1999 randomised a total of 57 consecutive adult patients attending a hospital clinic: n = 21 randomised to sticking tape; n = 18 to epilation; n = 18 to sticking tape followed by epilation. Baseline characteristics were comparable except that five lids (number of participants not stated) in the sticking tape group had trichiasis due to a cause other than trachoma.

t patients attending a hospital clinic: n = 21 randomised to sticking tape; n = 18 to epilation; n = 18 to sticking tape followed by epilation. Baseline characteristics were comparable except that five lids (number of participants not stated) in the sticking tape group had trichiasis due to a cause other than trachoma. Antibiotic treatment In Burton 2005a 451 participants with major trichiasis were enrolled; 70% were female. Baseline characteristics were similar between the two randomisation groups for age, ethnicity and severity of trichiasis. In West 2006 1452 individuals with any degree of TT were recruited; 77% were female. The three arms were balanced in terms of baseline characteristics. Zhang 2006 randomised 109 individuals with TT (both major and minor trichiasis); 73% were female. The baseline characteristics of the two arms were comparable. Alternative Suture Materials Rajak 2011b recruited 1300 adults with major TT; 78% were female. They were identified through community outreach campaigns.

In West 2006 1452 individuals with any degree of TT were recruited; 77% were female. The three arms were balanced in terms of baseline characteristics. Zhang 2006 randomised 109 individuals with TT (both major and minor trichiasis); 73% were female. The baseline characteristics of the two arms were comparable. Alternative Suture Materials Rajak 2011b recruited 1300 adults with major TT; 78% were female. They were identified through community outreach campaigns. Surgery setting Bowman 2000 selected five districts in The Gambia that were known to have high levels of trichiasis and where village‐based surgery previously had not been available. The districts were subdivided to form eight pairs of village clusters. Within each pair, one cluster of villages was randomised to village‐based surgery and the other cluster of villages to health centre‐based surgery. Screening was undertaken by trained ophthalmic nurses. Only participants with major trichiasis (at least five in‐turned lashes) were eligible for inclusion, in accordance with the Gambian national guidelines for surgery. Participants ineligible for village‐based surgery for medical reasons were excluded from the trial and referred for health centre‐based surgery. In all 158 individuals with major trichiasis were recruited. Personnel performing surgery Alemayehu 2004 recruited 982 people with TT; 77% were female and 3% were children. Baseline characteristics of the randomised groups are not described.

Surgery setting Bowman 2000 selected five districts in The Gambia that were known to have high levels of trichiasis and where village‐based surgery previously had not been available. The districts were subdivided to form eight pairs of village clusters. Within each pair, one cluster of villages was randomised to village‐based surgery and the other cluster of villages to health centre‐based surgery. Screening was undertaken by trained ophthalmic nurses. Only participants with major trichiasis (at least five in‐turned lashes) were eligible for inclusion, in accordance with the Gambian national guidelines for surgery. Participants ineligible for village‐based surgery for medical reasons were excluded from the trial and referred for health centre‐based surgery. In all 158 individuals with major trichiasis were recruited. Personnel performing surgery Alemayehu 2004 recruited 982 people with TT; 77% were female and 3% were children. Baseline characteristics of the randomised groups are not described. Types of outcomes In all but one of the studies a successful outcome included the absence of post‐operative trichiasis; this is usually defined as no lashes touching the eye. Specific outcomes for each study are described below. Surgical technique In the Reacher 1990 study a successful outcome was defined as no post‐operative trichiasis (no lashes in contact with the globe after surgery) and complete gentle closure of eyelids. There was no pre‐defined outcome point, and in the group with major trichiasis follow‐up varied from 5 to 11 months.

Types of outcomes In all but one of the studies a successful outcome included the absence of post‐operative trichiasis; this is usually defined as no lashes touching the eye. Specific outcomes for each study are described below. Surgical technique In the Reacher 1990 study a successful outcome was defined as no post‐operative trichiasis (no lashes in contact with the globe after surgery) and complete gentle closure of eyelids. There was no pre‐defined outcome point, and in the group with major trichiasis follow‐up varied from 5 to 11 months. Reacher 1992 defined a successful surgical outcome as no post‐operative trichiasis, no further epilation/surgery during follow up period, complete lid closure, no over‐correction of lid margin, acceptable appearance to patient and examiner, and no onset of phthisis. They also examined the effect on visual acuity and the complication rate. Follow‐up points were not defined in the methodology, but occurred (on average) at 9 and 21 months. Adamu 2002 defined success as no eyelash‐eyeball contact, complete lid closure and no over‐ or under‐correction. Recurrence was defined as eyelash‐eyeball contact in all positions of gaze or inward rotation of the lid margin. Visual acuity was measured pre‐ and post‐operatively. Final follow up was planned at three months. Dhaliwal 2005 used several outcome measures at six months: trichiasis and/or entropion recurrence, palpebral aperture, an acceptable cosmetic appearance.

Adamu 2002 defined success as no eyelash‐eyeball contact, complete lid closure and no over‐ or under‐correction. Recurrence was defined as eyelash‐eyeball contact in all positions of gaze or inward rotation of the lid margin. Visual acuity was measured pre‐ and post‐operatively. Final follow up was planned at three months. Dhaliwal 2005 used several outcome measures at six months: trichiasis and/or entropion recurrence, palpebral aperture, an acceptable cosmetic appearance. In Gower 2013, the primary outcome was the presence of one or more of the following unfavourable outcomes: post‐operative trichiasis; eyelid contour abnormality; or granuloma formation. Participants were re‐examined at six weeks, 12 months and 24 months. Both eyes in bilateral cases were included in the analysis, with an appropriate adjustment. Epilation In Rajak 2011a, the primary outcome measure was "failure" which was defined as either (1) five or more eyelashes touching the globe; or (2) a history of surgery performed in the trial eye at any point during the two‐year follow‐up period (in the case of the surgical arm this would be repeat surgery). Participants were re‐examined every six months for two years. Lid taping Graz 1999 collected data at 1, 4 and 12 weeks and recorded recurrence of trichiasis, visual acuity, patient discomfort and any adverse events.

Epilation In Rajak 2011a, the primary outcome measure was "failure" which was defined as either (1) five or more eyelashes touching the globe; or (2) a history of surgery performed in the trial eye at any point during the two‐year follow‐up period (in the case of the surgical arm this would be repeat surgery). Participants were re‐examined every six months for two years. Lid taping Graz 1999 collected data at 1, 4 and 12 weeks and recorded recurrence of trichiasis, visual acuity, patient discomfort and any adverse events. Antibiotic treatment Outcomes in the Burton 2005a study were assessed at 6 and 12 months post‐operatively. The primary outcome was post‐operative trichiasis. Secondary outcomes included visual acuity and patient perception of improvement by asking whether vision and pain was 'worse', 'same' or 'better'. In West 2006 the primary outcome was post‐operative trichiasis. This was assessed at 2 weeks, 6 weeks, 6 months, and 12 months. Patients were examined again at 2 and 3 years. Zhang 2006 evaluated three outcomes: post‐operative trichiasis; chlamydial infection; and active trachoma. Participants were re‐examined at 3, 6 and 12 months post‐operatively. If the participant developed "surgical failure", which was defined as five or more lashes touching the eye by three months, they were excluded from the analysis.

ed three outcomes: post‐operative trichiasis; chlamydial infection; and active trachoma. Participants were re‐examined at 3, 6 and 12 months post‐operatively. If the participant developed "surgical failure", which was defined as five or more lashes touching the eye by three months, they were excluded from the analysis. Alternative Suture Materials Rajak 2011b The primary outcome measure was the proportion of those individuals seen at the 12‐month follow‐up who were found to have either (1) post‐operative trichiasis; or (2) a history of repeat TT surgery during the first year. Participants were re‐examined every six months for two years. Surgery setting There were three main outcomes in the Bowman 2000 study: (1) uptake of treatment in the local village compared to the health centre; (2) post‐operative trichiasis); (3) complication rate. Other parameters measured included time taken by the patient to travel to the operating room and cost implications associated with presenting for surgery. Personnel performing surgery The first follow‐up in the Alemayehu 2004 study was at seven days post‐operatively: those with post‐operative trichiasis at this assessment were deemed surgical failures and were excluded from the primary analysis; the main outcome was post‐operative trichiasis rate at three and six months. Those with post‐operative trichiasis at three months were excluded from the six‐month assessment. Analysis was by randomised group and also by presenting severity.

ment were deemed surgical failures and were excluded from the primary analysis; the main outcome was post‐operative trichiasis rate at three and six months. Those with post‐operative trichiasis at three months were excluded from the six‐month assessment. Analysis was by randomised group and also by presenting severity. Risk of bias in included studies There are potential biases in several of the studies. These are considered under the following headings: allocation, blinding, incomplete data, and selective reporting. See Figure 2; Figure 3.Figure 2 Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies. Figure 3 Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias in included studies There are potential biases in several of the studies. These are considered under the following headings: allocation, blinding, incomplete data, and selective reporting. See Figure 2; Figure 3.Figure 2 Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies. Figure 3 Risk of bias summary: review authors' judgements about each risk of bias item for each included study. Allocation Surgical technique Adamu 2002 used an unspecified 'Lottery method' for randomisation. In bilateral cases the left eye always received the opposite treatment to that randomly allocated to the right eye, with no analytic adjustment for bilateral cases reported. The risk of bias is unclear given the limited details reported. In Gower 2013 randomisation assignments were created using permuted block sizes of 6 to 12 and placed in opaque envelopes; hence, the risk of bias is expected to be low. Reacher 1990 randomly allocated procedures using random number tables. Limited information is provided on the process and no details are provided on concealment; thus, the risk of bias is unclear. In the Reacher 1992 study, participants were randomly allocated using a computer‐generated sequence; assignments were concealed in opaque envelopes. Risk of bias is expected to be low. Dhaliwal 2005 randomised the study eye by a computer‐generated sequence. However, if a second eye was operated one of the other two options was used. The details are limited and, therefore, the risk of allocation bias is unclear.

uence; assignments were concealed in opaque envelopes. Risk of bias is expected to be low. Dhaliwal 2005 randomised the study eye by a computer‐generated sequence. However, if a second eye was operated one of the other two options was used. The details are limited and, therefore, the risk of allocation bias is unclear. Epilation In Rajak 2011a participants were allocated to arm by a computer‐generated randomisation sequence and assignments were kept in sealed, opaque envelopes, leading to a low risk of bias. Lid Taping Graz 1999 provide no description of the randomisation process. Antibiotic treatment In Burton 2005a separate randomisation sequences were generated for each surgeon using random number tables and blocked in groups of four, and assignments were kept in sequential sealed envelopes, leading to a low risk of allocation bias. In West 2006, participants were randomly assigned to one of three intervention arms using a computer‐generated sequence with a variable block size, and assignments were kept in opaque containers until needed, leading to a low risk of bias. In Zhang 2006 participants all received bilamellar tarsal rotation surgery, and at the end of surgery they were alternately given either azithromycin or a placebo. This randomisation method makes the next assignment predictable, and hence makes the risk of allocation bias high. Alternative Suture Materials In Rajak 2011b the risk of allocation bias is expected to be low since people were allocated to arm by a computer‐generated randomisation sequence and assignments were maintained in opaque envelopes.

Antibiotic treatment In Burton 2005a separate randomisation sequences were generated for each surgeon using random number tables and blocked in groups of four, and assignments were kept in sequential sealed envelopes, leading to a low risk of allocation bias. In West 2006, participants were randomly assigned to one of three intervention arms using a computer‐generated sequence with a variable block size, and assignments were kept in opaque containers until needed, leading to a low risk of bias. In Zhang 2006 participants all received bilamellar tarsal rotation surgery, and at the end of surgery they were alternately given either azithromycin or a placebo. This randomisation method makes the next assignment predictable, and hence makes the risk of allocation bias high. Alternative Suture Materials In Rajak 2011b the risk of allocation bias is expected to be low since people were allocated to arm by a computer‐generated randomisation sequence and assignments were maintained in opaque envelopes. Surgery setting In Bowman 2000, a cluster randomised trial, the communities were paired and randomly allocated before the numbers of TT patients was known, so risk of allocation bias low. Personnel performing surgery Alemayehu 2004 do not provide any details of the randomisation or concealment methods employed; hence, risk of bias is unknown.

Surgery setting In Bowman 2000, a cluster randomised trial, the communities were paired and randomly allocated before the numbers of TT patients was known, so risk of allocation bias low. Personnel performing surgery Alemayehu 2004 do not provide any details of the randomisation or concealment methods employed; hence, risk of bias is unknown. Blinding Surgical technique Adamu 2002 provides no description of methods to mask either the participants or those performing the follow‐up examinations. In Gower 2013 randomisation assignments were placed in opaque envelopes and stored in a locked office until distribution and outcomes were assessed by a masked examiner. Reacher 1990 provides no information on the concealment of treatment allocation sequences. The follow‐up examinations were by a single observer, but it is unclear if he was masked to the allocation. In Reacher 1992 the allocation sequence was concealed in opaque envelopes. Follow‐up examinations were performed by a single masked observer. Dhaliwal 2005 provide no details about masking of the allocation prior to administration or during follow‐up; it is unclear if the observations were made by the same clinician who performed surgery or by a separate clinician. Therefore the risk of bias is unclear. Epilation In Rajak 2011a the randomisation sequence was concealed in opaque envelopes. Observers were independent of, and masked to, the allocation. Lid taping In Graz 1999 masking of the observers was not possible.

Blinding Surgical technique Adamu 2002 provides no description of methods to mask either the participants or those performing the follow‐up examinations. In Gower 2013 randomisation assignments were placed in opaque envelopes and stored in a locked office until distribution and outcomes were assessed by a masked examiner. Reacher 1990 provides no information on the concealment of treatment allocation sequences. The follow‐up examinations were by a single observer, but it is unclear if he was masked to the allocation. In Reacher 1992 the allocation sequence was concealed in opaque envelopes. Follow‐up examinations were performed by a single masked observer. Dhaliwal 2005 provide no details about masking of the allocation prior to administration or during follow‐up; it is unclear if the observations were made by the same clinician who performed surgery or by a separate clinician. Therefore the risk of bias is unclear. Epilation In Rajak 2011a the randomisation sequence was concealed in opaque envelopes. Observers were independent of, and masked to, the allocation. Lid taping In Graz 1999 masking of the observers was not possible. Antibiotic treatment In Burton 2005a allocation was contained in opaque sequentially‐numbered envelopes and administered immediately following surgery by a nurse not involved in surgery or follow‐up assessment. Pre‐operative assessments and the 12‐month follow‐up assessments were made by the same observer. Assessments at six months were by a different observer. Both observers were masked to treatment allocation. Surgery was performed prior to the randomisation. In West 2006 the random treatment allocation was concealed in sequential opaque envelopes. In Zhang 2006 participants and observers were masked to whether they had received azithromycin or a placebo; however, this was on an alternating basis, rather than random allocation.

gery was performed prior to the randomisation. In West 2006 the random treatment allocation was concealed in sequential opaque envelopes. In Zhang 2006 participants and observers were masked to whether they had received azithromycin or a placebo; however, this was on an alternating basis, rather than random allocation. Alternative suture materials In Rajak 2011b the randomisation sequence was concealed in opaque envelopes. Observers were independent of, and masked to, the allocation. Surgery setting In Bowman 2000 it would have been impossible to mask the patients or personnel to the location of surgery. Personnel performing surgery In Alemayehu 2004 the patients and the surgeons were not masked to the allocation. The follow‐up observations were made by ophthalmologists masked to the allocation. Incomplete outcome data Surgical technique The studies by Adamu 2002, Gower 2013, Dhaliwal 2005, Reacher 1990 and Reacher 1992 all present a complete data set with high rates of follow‐up. Epilation Rajak 2011a presents a complete data set with high rates of follow‐up. Lid taping Graz 1999 presents a complete data set with high rates of follow‐up. Antibiotic treatment Burton 2005a presents a complete data set with high follow‐up. A geographically defined sub‐set of the original trial subjects was followed up to 4 years, with 94% follow‐up of those alive at 4 years. West 2006 presents a complete data set with high follow‐up. Similarly Zhang 2006 reports a complete data set with high follow‐up rate. Alternative Suture Materials Rajak 2011b presents a complete data set with high rates of follow‐up.

Antibiotic treatment Burton 2005a presents a complete data set with high follow‐up. A geographically defined sub‐set of the original trial subjects was followed up to 4 years, with 94% follow‐up of those alive at 4 years. West 2006 presents a complete data set with high follow‐up. Similarly Zhang 2006 reports a complete data set with high follow‐up rate. Alternative Suture Materials Rajak 2011b presents a complete data set with high rates of follow‐up. Surgery setting Bowman 2000 presents complete outcome data. Personnel performing surgery Alemayehu 2004 had moderate loss to follow‐up. No comparative data on those lost to follow‐up are provided to determine risk of attrition bias. Selective reporting Surgical technique Adamu 2002 provides limited information of their analytical protocol; therefore, risk of bias is unclear. The studies by Gower 2013, Reacher 1990 and Reacher 1992 all report a clear analytical approach which was followed in the results. Epilation The protocol for Rajak 2011a is published online and was followed. Lid taping Graz 1999 reports a clear analytical approach which was followed in the results. Antibiotic treatment Burton 2005a reports a clear analytical approach which was followed in the results. The protocol for West 2006 is published separately. Zhang 2006 excluded "surgical failures" from the analysis, defined as five or more lashes touching the globe at 3 months. Alternative suture materials The protocol for Rajak 2011b is published online and was followed. Surgery setting Bowman 2000 reports a clear analytical approach.

Antibiotic treatment Burton 2005a reports a clear analytical approach which was followed in the results. The protocol for West 2006 is published separately. Zhang 2006 excluded "surgical failures" from the analysis, defined as five or more lashes touching the globe at 3 months. Alternative suture materials The protocol for Rajak 2011b is published online and was followed. Surgery setting Bowman 2000 reports a clear analytical approach. Personnel performing surgery Alemayehu 2004 provides limited information on the analytical protocol; however, all participants seen at follow‐up were included in the analyses. Other potential sources of bias None identified. Effects of interventions Surgical interventions 1. Bilamellar tarsal rotation surgery compared to tarsal advance and rotation Two studies reported this comparison. In Reacher 1990 the mean months of follow‐up per group ranged from 7.4 to 8.8 months (total range 5 to 11 months); in Reacher 1992 this was reported at 9 and 21 months after surgery. 1.1 Post‐operative trichiasis Eyes receiving bilamellar tarsal rotation surgery had lower odds of post‐operative trichiasis compared to eyes receiving tarsal advance and rotation (OR 0.29, 95% CI 0.16 to 0.50, 260 eyes, I² = 0%) (Analysis 1.1). 1.2 Visual acuity change Visual acuity was not reported in Reacher 1990. In Reacher 1992 eyes receiving surgery for major trichiasis on average had half a line improvement in Snellen acuity (P < 0.001) but the difference between the intervention groups (if any) was not clearly reported. 1.3 Corneal opacification change Not reported.

1.2 Visual acuity change Visual acuity was not reported in Reacher 1990. In Reacher 1992 eyes receiving surgery for major trichiasis on average had half a line improvement in Snellen acuity (P < 0.001) but the difference between the intervention groups (if any) was not clearly reported. 1.3 Corneal opacification change Not reported. 1.4 Acceptance of treatment This was not reported in Reacher 1990. In Reacher 1992 38 people refused their random allocation but it was not reported which groups they were allocated to. 1.5 Adverse effects Bilamellar tarsal rotation surgery was associated with more overcorrection of the entropion but the estimate was imprecise with very wide confidence intervals compatible with no effect, or more overcorrection in the tarsal advance group (OR 2.57, 95% CI 0.28 to 23.25, 312 eyes, I² = 0%) (Analysis 1.2). There were only 3 cases of overcorrection in the two trials but they were all in the bilamellar group. Defective lid closure was also more common in the bilamellar group but again occurred rarely and the estimate of effect was very imprecise (OR 1.90, 95% CI 0.29 to 12.37, 312 eyes, I² = 0%). (Analysis 1.3). 1.6 Quality of life Not reported. 2. Bilamellar tarsal rotation compared to techniques that do not create a full‐thickness incision of the tarsal plate and complete rotation of the lash‐bearing tissue Only one study reported these comparisons (Reacher 1990). The mean months of follow‐up per group ranged from 7.4 to 8.8 months (total range 5 to 11 months)

2. Bilamellar tarsal rotation compared to techniques that do not create a full‐thickness incision of the tarsal plate and complete rotation of the lash‐bearing tissue Only one study reported these comparisons (Reacher 1990). The mean months of follow‐up per group ranged from 7.4 to 8.8 months (total range 5 to 11 months) 2.1 Post‐operative trichiasis Bilamellar tarsal rotation was more effective than techniques that do not create a full‐thickness incision of the tarsal plate and complete rotation of the lash‐bearing tissue such as tarsal grooving and eversion splinting, and a non‐significant trend in the comparison with the tarsal advance procedure (Table 3).Table 1 Bilamellar tarsal rotation compared to techniques that do not create a full‐thickness incision of the tarsal plate and complete rotation of the lash‐bearing tissue Outcome Other technique Follow‐up range 5 to 11 months Bilamellar tarsal rotation n/N Other technique n/N Odds ratio (95% CI) Post‐operative trichiasis Tarsal grooving 10/39 29/33 0.05 (0.01, 0.17) Eversion splinting 10/39 14/21 0.17 (0.05, 0.55) Tarsal advance 10/39 17/38 0.43 (0.16, 1.11) Data from Reacher 1990. n = number of eyes with outcome; N = total number of eyes followed up None of the other review outcomes were reported. 3. Bilamellar tarsal rotation surgery compared to posterior lamellar tarsal rotation One study reported this comparison (Adamu 2002). 256 upper eyelids of 153 people with trichiasis were randomly allocated to bilamellar tarsal rotation (n = 124) or posterior lamellar tarsal rotation (n = 132) and followed up to 3 months.

3. Bilamellar tarsal rotation surgery compared to posterior lamellar tarsal rotation One study reported this comparison (Adamu 2002). 256 upper eyelids of 153 people with trichiasis were randomly allocated to bilamellar tarsal rotation (n = 124) or posterior lamellar tarsal rotation (n = 132) and followed up to 3 months. 3.1 Post‐operative trichiasis After three months there was less post‐operative trichiasis in the bilamellar tarsal rotation group but the confidence intervals were wide and compatible with no effect, or more trichiasis (Table 4).Table 2 Bilamellar tarsal rotation surgery compared to posterior lamellar tarsal rotation Outcome Follow‐up Bilamellar tarsal rotation n/N Posterior lamellar tarsal rotation n/N Odds ratio (95% CI) Post‐operative trichiasis 3 months 12/124 15/132 0.84 (0.37, 1.86) Data from Adamu 2002 n = number of eyes with outcome; N = total number of eyes followed up 3.2 Visual acuity change The authors stated that there was an improvement in vision after surgery but the data were not reported and it was of borderline statistical significance (P = 0.0515). 3.3 Corneal opacification change Not reported. 3.4 Acceptance of treatment Not reported. 3.5 Adverse events Lid‐notching and pyogenic granuloma were more common in the bilamellar than the posterior lamellar tarsal rotation operations (Chi² 9.54, P = 0.002) but no data were reported. 3.6 Quality of life Not reported.

3.2 Visual acuity change The authors stated that there was an improvement in vision after surgery but the data were not reported and it was of borderline statistical significance (P = 0.0515). 3.3 Corneal opacification change Not reported. 3.4 Acceptance of treatment Not reported. 3.5 Adverse events Lid‐notching and pyogenic granuloma were more common in the bilamellar than the posterior lamellar tarsal rotation operations (Chi² 9.54, P = 0.002) but no data were reported. 3.6 Quality of life Not reported. 4. Terminal tarsal rotation compared to tarsal advance and rotation and tarsal grooving One study reported these comparisons (Dhaliwal 2005). The study randomised 77 participants (90 eyes); all 77 participants had six‐month follow‐up data. Pre‐operatively entropion was found to be moderate in 65 eyes and severe in 25 eyes (although these terms were not defined). No eyes had defective lid closure. Bilateral TT was present in 13 people. 4.1 Post‐operative trichiasis After six months there was no significant difference in the rate of post‐operative trichiasis between the three procedures; however, the sample size in each group was insufficient to detect a statistically significant difference at a meaningful level. 4.2 Visual acuity change Not reported. 4.3 Corneal opacification change Not reported. 4.4 Acceptance of treatment Not reported. 4.5 Adverse effects Lid‐notching: (1) terminal tarsal rotation 9 (30%); (2) tarsal advance and rotation 6 (20%); and (3) tarsal grooving 10 (33%)

4.1 Post‐operative trichiasis After six months there was no significant difference in the rate of post‐operative trichiasis between the three procedures; however, the sample size in each group was insufficient to detect a statistically significant difference at a meaningful level. 4.2 Visual acuity change Not reported. 4.3 Corneal opacification change Not reported. 4.4 Acceptance of treatment Not reported. 4.5 Adverse effects Lid‐notching: (1) terminal tarsal rotation 9 (30%); (2) tarsal advance and rotation 6 (20%); and (3) tarsal grooving 10 (33%) Pyogenic granuloma: (1) terminal tarsal rotation 3 (10%); (2) tarsal advance and rotation 5 (17%); and (3) tarsal grooving 3 (10%) 4.6 Quality of life There was no difference between the three groups in terms of the proportion of patients who were satisfied with the cosmetic appearance following surgery: (1) terminal tarsal rotation 28 (93%); (2) tarsal advance and rotation 27 (90%); and (3) tarsal grooving 27 (90%). 5. Bilamellar tarsal rotation with a clamp compared to bilamellar tarsal rotation without a clamp One study (Gower 2013) reported this comparison. The study randomised 1917 participants (3345 eyes) and followed up to two years. 5.1 Post‐operative trichiasis There was more post‐operative trichiasis in the clamp surgery group at two years. After adjustment for correlation between eyes and for surgeon, age, sex and baseline TT severity the adjusted OR was 1.36, 95% CI 0.96 to 1.93. The lower confidence interval includes, but is fairly close to 1 (no effect). 5.2 Visual acuity change Not reported. 5.3 Corneal opacification change Not reported.

5.1 Post‐operative trichiasis There was more post‐operative trichiasis in the clamp surgery group at two years. After adjustment for correlation between eyes and for surgeon, age, sex and baseline TT severity the adjusted OR was 1.36, 95% CI 0.96 to 1.93. The lower confidence interval includes, but is fairly close to 1 (no effect). 5.2 Visual acuity change Not reported. 5.3 Corneal opacification change Not reported. 5.4 Acceptance of treatment Not reported. 5.5 Adverse effects Eyelid contour abnormalities were less frequent in the clamp surgery group (adjusted OR 0.65, 95% CI 0.44 to 0.98, 3343 eyes). Granulomas were less frequent in the clamp surgery group (adjusted OR 0.67, 95% CI 0.46 to 0.97, 3343 eyes). 5.6 Quality of life Not reported. 6. Bilamellar tarsal rotation surgery compared to cryotherapy or electrolysis 6.1 Post‐operative trichiasis Post‐operative trichiasis in 166 eyes with minor trichiasis was reported by Reacher 1992; follow‐up ranged from 1 to 21 months. Bilamellar tarsal rotation was more effective than destruction of the lashes by cryotherapy or electrolysis (Table 5).Table 3 Bilamellar tarsal rotation surgery compared to cryotherapy or electrolysis Outcome Other technique Follow‐up range 1 to 21 months Bilamellar tarsal rotation n/N Other technique n/N Odds ratio (95% CI) Post‐operative trichiasis Cryotherapy 6/52 41/57 0.05 (0.02, 0.14) Electrolysis 6/52 30/57 0.12 (0.04, 0.32) Data from Reacher 1992 n = number of eyes with outcome; N = total number of eyes followed up

Outcome Other technique Follow‐up range 1 to 21 months Bilamellar tarsal rotation n/N Other technique n/N Odds ratio (95% CI) Post‐operative trichiasis Cryotherapy 6/52 41/57 0.05 (0.02, 0.14) Electrolysis 6/52 30/57 0.12 (0.04, 0.32) Data from Reacher 1992 n = number of eyes with outcome; N = total number of eyes followed up 6.2 Visual acuity change Treatment of minor trichiasis was not associated with any improvement in vision but the differential effect of treatment, if any, was not clearly reported. 6.3 Corneal opacification change Not reported. 6.4 Acceptance of treatment Not reported. 6.5 Adverse effects Not reported. 6.6 Quality of life Not reported. 7. Posterior lamellar tarsal rotation surgery compared to epilation One trial reported this comparison (Rajak 2011a). At baseline 1300 participants with minor trichiasis were randomised to surgery or epilation and followed up for two years. 7.1 Post‐operative trichiasis Trichiasis defined as one or more lashes touching the globe or repeat surgery was less frequent in the surgery group over two years (Table 6).Table 4 Posterior lamellar tarsal rotation surgery compared to epilation

7. Posterior lamellar tarsal rotation surgery compared to epilation One trial reported this comparison (Rajak 2011a). At baseline 1300 participants with minor trichiasis were randomised to surgery or epilation and followed up for two years. 7.1 Post‐operative trichiasis Trichiasis defined as one or more lashes touching the globe or repeat surgery was less frequent in the surgery group over two years (Table 6).Table 4 Posterior lamellar tarsal rotation surgery compared to epilation Outcome Follow‐up Posterior lamellar tarsal rotation surgery n/N Epilation n/N Odds ratio (95% CI) Post‐operative trichiasis Two years 114/637 298/641 0.25 (0.19, 0.32) Visual acuity change: deterioration of one or more lines of visual acuity One year 168/620 174/598 0.91 (0.71, 1.16) Two years 207/613 224/603 0.86 (0.68, 1.09) Deterioration in corneal opacification One year 7/620 12/598 0.56 (0.22, 1.43) Two years 25/613 33/603 0.73 (0.43, 1.25) Data from Rajak 2011a. n = number of eyes with outcome; N = total number of eyes followed up 7.2 Visual acuity change There was no evidence of a difference between the two groups in deterioration in visual acuity between baseline and one year or two years; however the confidence intervals include 1, therefore they are consistent with no effect (Table 6).

n = number of eyes with outcome; N = total number of eyes followed up 7.2 Visual acuity change There was no evidence of a difference between the two groups in deterioration in visual acuity between baseline and one year or two years; however the confidence intervals include 1, therefore they are consistent with no effect (Table 6). 7.3 Corneal opacification change Progression in corneal opacification was infrequent in both groups at one and two years. There was a non‐significant trend to less deterioration in corneal opacification between baseline and one or two years in the surgery group; however the wide confidence intervals mean we cannot exclude the possibility that there is less deterioration in the epilation group or no difference between procedures (Table 6). 7.4 Acceptance of treatment All participants accepted their initial randomisation allocation. At two years 185/593 (31%) of individuals who were still epilating accepted the offer of free community‐based surgery. 7.5 Adverse effects In the surgery arm, by two years, 105 (16%) of eyes had developed recurrent trichiasis, granulomas developed in 18 (2.9%) and eyelid contour abnormalities were reported in 29 (4.7%).

7.4 Acceptance of treatment All participants accepted their initial randomisation allocation. At two years 185/593 (31%) of individuals who were still epilating accepted the offer of free community‐based surgery. 7.5 Adverse effects In the surgery arm, by two years, 105 (16%) of eyes had developed recurrent trichiasis, granulomas developed in 18 (2.9%) and eyelid contour abnormalities were reported in 29 (4.7%). 7.6 Quality of life At 12 months 43% of participants in the surgery arm recalled "severe" treatment pain compared to 27% of people in the epilation group. More people in the surgery groups reported better subjective improvement in vision at 12 months (78%) compared to 33% of the epilation group, and less eye pain (44% compared to 70%), and eye watering (51% compared to 63%). All these differences were statistically significant. 8. Absorbable (polyglactin‐910) compared to non‐absorbable (silk) sutures One trial investigated this comparison (Rajak 2011b). 1300 participants with major trichiasis were randomised to absorbable polyglactin‐910 sutures or silk sutures and followed‐up at one year and two years. 8.1 Post‐operative trichiasis Post‐operative trichiasis at one year and two years was similar between the two groups (Table 7).Table 5 Absorbable (polyglactin‐910) compared to non‐absorbable (silk) sutures

8. Absorbable (polyglactin‐910) compared to non‐absorbable (silk) sutures One trial investigated this comparison (Rajak 2011b). 1300 participants with major trichiasis were randomised to absorbable polyglactin‐910 sutures or silk sutures and followed‐up at one year and two years. 8.1 Post‐operative trichiasis Post‐operative trichiasis at one year and two years was similar between the two groups (Table 7).Table 5 Absorbable (polyglactin‐910) compared to non‐absorbable (silk) sutures Outcome Follow‐up Absorbable sutures n/N Silk sutures n/N Odds ratio (95% CI) Post‐operative trichiasis One year 114/608 120/628 0.98 (0.73, 1.30)) Two years 117/609 118/609 0.99 (0.75, 1.32) Visual acuity change: deterioration of one or more lines One year 167/608 146/628 1.25 (0.97, 1.62) Two years 191/609 198/609 0.95 (0.75, 1.21) Deterioration in corneal opacification One year 6/608 11/628 0.56 (0.21, 1.52) Two years 24/609 18/609 1.35 (0.72, 2.51) Data from Rajak 2011b. n = number of eyes with outcome; N = total number of eyes followed up 8.2 Visual acuity change Change (deterioration of one or more lines) in visual acuity between baseline and two years was similar in the two groups (Table 7). 8.3 Corneal opacification change There was less deterioration in corneal opacification between baseline and one year in the absorbable sutures group and more at two years but the confidence intervals were wide and compatible with no effect (Table 7). 8.4 Acceptance of treatment Not reported.

8.2 Visual acuity change Change (deterioration of one or more lines) in visual acuity between baseline and two years was similar in the two groups (Table 7). 8.3 Corneal opacification change There was less deterioration in corneal opacification between baseline and one year in the absorbable sutures group and more at two years but the confidence intervals were wide and compatible with no effect (Table 7). 8.4 Acceptance of treatment Not reported. 8.5 Adverse effects Fewer granulomas were found in the absorbable sutures group at six months (OR 0.63, 95% CI 0.40 to 0.99, 1187 eyes). 8.6 Quality of life Not reported. Non‐surgical interventions 9. Sticking tape compared to epilation One study reported this comparison and followed up to 3 months (Graz 1999). The study randomised 57 participants. 9.1 Post‐operative trichiasis The use of sticking tape alone was significantly more effective at preventing eyelashes from touching the eye compared to epilation alone at three months (Table 8).Table 6 Sticking tape compared to epilation Outcome Follow‐up Sticking tape n/N Epilation n/N Odds ratio (95% CI) Post‐operative trichiasis 3 months 6/21 18/18 0.01 (0, 0.22) Data from Graz 1999. n = number of eyes with outcome; N = total number of eyes followed up 9.2 Visual acuity change Snellen visual acuity was measured using the 'E' optotype, but outcomes were not reported. 9.3 Corneal opacification change Not reported. 9.4 Acceptance of treatment Attendance for treatment was not recorded but it was mentioned in the discussion that there may have been compliance issues. 9.5 Adverse effects Not reported.

9.2 Visual acuity change Snellen visual acuity was measured using the 'E' optotype, but outcomes were not reported. 9.3 Corneal opacification change Not reported. 9.4 Acceptance of treatment Attendance for treatment was not recorded but it was mentioned in the discussion that there may have been compliance issues. 9.5 Adverse effects Not reported. 9.6 Quality of life A patient questionnaire of six closed‐response questions was used to measure levels of discomfort; results were summarised as 'complaint' versus 'no complaint'. Patients found the less successful treatment of epilation significantly more uncomfortable than the sticking tape (P = 0.002); this was only reported on those with unilateral trichiasis. 10. Azithromycin compared to no azithromycin Three studies reported this comparison (Burton 2005a; West 2006; Zhang 2006). Burton 2005a randomised 451 people with major trichiasis only to receive azithromycin or no azithromycin at time of surgery; household members were also treated. West 2006 (STAR trial) randomised participants to three groups: azithromycin to patient only (483); azithromycin to patient and household (485); or tetracycline (484). Zhang 2006 randomised 109 people (148 eyes) with either minor and major trichiasis to azithromycin or placebo at time of surgery.

Burton 2005a randomised 451 people with major trichiasis only to receive azithromycin or no azithromycin at time of surgery; household members were also treated. West 2006 (STAR trial) randomised participants to three groups: azithromycin to patient only (483); azithromycin to patient and household (485); or tetracycline (484). Zhang 2006 randomised 109 people (148 eyes) with either minor and major trichiasis to azithromycin or placebo at time of surgery. 10.1 Post‐operative trichiasis There was heterogeneity between the trials in terms of postoperative trichiasis. West 2006 found a lower recurrence with azithromycin, while Burton 2005a and Zhang 2006 did not find a significant difference. In combining the data from the three trials, overall there was less post‐operative trichiasis at one year in people given azithromycin; however the combined result effect was uncertain as the confidence intervals include 1 (OR 0.85, 95% CI 0.63 to 1.14, 1954 eyes, 3 studies) (Analysis 2.1). At three years in the STAR trial (West 2006) there was less post‐operative trichiasis but again the overall effect was uncertain. A similar pattern was seen at four years follow‐up in Burton 2005a (Analysis 2.2). 10.2 Visual acuity change In Burton 2005a visual acuity improved in 57.6% of eyes by 12 months. There was an overall improvement of 0.14 logMAR (P < 0.0001). Data were not reported by randomised groups.

At three years in the STAR trial (West 2006) there was less post‐operative trichiasis but again the overall effect was uncertain. A similar pattern was seen at four years follow‐up in Burton 2005a (Analysis 2.2). 10.2 Visual acuity change In Burton 2005a visual acuity improved in 57.6% of eyes by 12 months. There was an overall improvement of 0.14 logMAR (P < 0.0001). Data were not reported by randomised groups. In the STAR trial (West 2006) a comparison of the change in visual acuity between baseline and six months on a consecutive subset of 439 study subjects did not find a significant difference in visual acuity outcomes between the azithromycin and control groups but data were not reported. 10.3 Corneal opacification change Not reported. 10.4 Acceptance of treatment In Burton 2005a all study participants accepted their allocated treatment. In the STAR trial (West 2006) and Zhang 2006 this was not reported. 10.5 Adverse effects In Burton 2005a one participant had a post‐operative skin infection that settled on oral antibiotics. Two individuals had defective lid closure of less than 2 mm lagophthalmos. In West 2006 there was a similar rate of adverse events in the azithromycin and control groups with a rate per 100 person of adverse events (death/illness/ocular) at 6 weeks of 2.90 (1.93‐4.19) in the azithromycin group and 3.10 (1.73‐5.11) in the tetracycline group. Adverse effects were not reported in Zhang 2006. 10.6 Quality of life In Burton 2005a 77% reported improvement in vision and 94.3% felt the operated eye was more comfortable but this was not analysed by randomised groups.

In West 2006 there was a similar rate of adverse events in the azithromycin and control groups with a rate per 100 person of adverse events (death/illness/ocular) at 6 weeks of 2.90 (1.93‐4.19) in the azithromycin group and 3.10 (1.73‐5.11) in the tetracycline group. Adverse effects were not reported in Zhang 2006. 10.6 Quality of life In Burton 2005a 77% reported improvement in vision and 94.3% felt the operated eye was more comfortable but this was not analysed by randomised groups. In West 2006 trichiasis surgery had a marked benefit on physical functioning in a sub‐group analysis (Wolle 2011), but this was not analysed by groups allocated to azithromycin or control. Quality of life was not reported in Zhang 2006. Settings and personnel 11. Community‐based compared to health centre‐based surgery One study reported this comparison (Bowman 2000). This was a cluster‐randomised trial. Eight paired clusters of villages were randomised to community or village‐based surgery (86 participants) or health centre‐based surgery (72 participants). 11.1 Post‐operative trichiasis More people in the village‐based surgery group had post‐operative trichiasis at 3 months compared to the health centre‐based surgery group but the confidence intervals were wide and compatible with no effect or less trichiasis (OR 1.44, 95% CI 0.26 to 7.90). 11.2 Visual acuity change Not reported. 11.3 Corneal opacification change Not reported.

11.1 Post‐operative trichiasis More people in the village‐based surgery group had post‐operative trichiasis at 3 months compared to the health centre‐based surgery group but the confidence intervals were wide and compatible with no effect or less trichiasis (OR 1.44, 95% CI 0.26 to 7.90). 11.2 Visual acuity change Not reported. 11.3 Corneal opacification change Not reported. 11.4 Acceptance of treatment In six of the eight pairs of clusters surgical uptake was higher for village‐based surgery; however, this difference might have occurred by chance (difference 20%, 95% CI −9% to 49%). Analysed by individual, 57/86 (66%) in the village‐based clusters attended for surgery compared to 32/72 (44%) in the health centre‐based group (OR 2.46, 95% CI 1.29 to 4.68). 11.5 Adverse effects A total of four events are reported but there was said to be no difference between groups (data not reported). 11.6 Quality of life Not reported. 11.7 Cost The cost of travel was significantly less in the clusters randomised to community‐based surgery (difference between means 10.5 Dalasi, 95% CI 6.07 to 14.93). Journey time to village‐based surgery was significantly less (difference between means 36 minutes, 95% CI 15.37 to 56.63). 12. Ophthalmologist compared to integrated eye care worker (IECW) One study reported this comparison (Alemayehu 2004). Of the 982 randomised, 713 (73%) attended the three‐month outcome assessment: 370/713 (52%) ophthalmologist group; 343/713 (48%) IECW group.

11.7 Cost The cost of travel was significantly less in the clusters randomised to community‐based surgery (difference between means 10.5 Dalasi, 95% CI 6.07 to 14.93). Journey time to village‐based surgery was significantly less (difference between means 36 minutes, 95% CI 15.37 to 56.63). 12. Ophthalmologist compared to integrated eye care worker (IECW) One study reported this comparison (Alemayehu 2004). Of the 982 randomised, 713 (73%) attended the three‐month outcome assessment: 370/713 (52%) ophthalmologist group; 343/713 (48%) IECW group. This study reported a linear trend for increased risk of recurrence with increasing severity of pre‐operative entropion (X² 22, P < 0.001). Randomisation was not stratified according to severity. Data on pre‐operative disease severity is not presented by the two randomisation groups. It is possible that either group may have treated a higher proportion of patients with severe disease, which could influence the outcome. A seven‐day post‐operative check was planned to identify surgical failures (which were then excluded from further analysis) but the outcomes from this assessment were not reported. If there was a significant difference in the number of failures between the groups at that stage, it would affect the interpretation of the results. 12.1 Post‐operative trichiasis There was more post‐operative trichiasis in the ophthalmologist group at 3 months but the confidence intervals were wide and compatible with no effect or less trichiasis (OR 1.32, 95% CI 0.83 to 2.11, 713 eyes).

This study reported a linear trend for increased risk of recurrence with increasing severity of pre‐operative entropion (X² 22, P < 0.001). Randomisation was not stratified according to severity. Data on pre‐operative disease severity is not presented by the two randomisation groups. It is possible that either group may have treated a higher proportion of patients with severe disease, which could influence the outcome. A seven‐day post‐operative check was planned to identify surgical failures (which were then excluded from further analysis) but the outcomes from this assessment were not reported. If there was a significant difference in the number of failures between the groups at that stage, it would affect the interpretation of the results. 12.1 Post‐operative trichiasis There was more post‐operative trichiasis in the ophthalmologist group at 3 months but the confidence intervals were wide and compatible with no effect or less trichiasis (OR 1.32, 95% CI 0.83 to 2.11, 713 eyes). There was a difference in the success rates of the two IECWs: one operated on 184 of whom 12 (6.5%) developed recurrence, the other operated on 159 of whom 22 (13.8%) developed recurrence (OR 2.3, 95% CI 1.1 to 4.8). None of the other review outcomes were reported.

12.1 Post‐operative trichiasis There was more post‐operative trichiasis in the ophthalmologist group at 3 months but the confidence intervals were wide and compatible with no effect or less trichiasis (OR 1.32, 95% CI 0.83 to 2.11, 713 eyes). There was a difference in the success rates of the two IECWs: one operated on 184 of whom 12 (6.5%) developed recurrence, the other operated on 159 of whom 22 (13.8%) developed recurrence (OR 2.3, 95% CI 1.1 to 4.8). None of the other review outcomes were reported. Discussion Summary of main results Trachomatous trichiasis is a significant ophthalmic public health problem in several regions of the world, particularly in sub‐Saharan Africa. The evidence identified in this review serves to highlight not only what is currently known about the effectiveness of treatment for this condition but also the challenges involved in delivering the necessary care and achieving good long‐term outcomes to prevent blindness.

the world, particularly in sub‐Saharan Africa. The evidence identified in this review serves to highlight not only what is currently known about the effectiveness of treatment for this condition but also the challenges involved in delivering the necessary care and achieving good long‐term outcomes to prevent blindness. Surgical interventions Detailed descriptions of the different surgical procedures are provided elsewhere (Rajak 2012). Some evidence suggests that operations in which the full‐thickness of the tarsal plate is incised and the terminal lash‐bearing tarsus is rotated so that the lid margin is everted are more effective than procedures that do not involve this (Reacher 1990). It was possible to pool the results of Reacher 1990 and Reacher 1992, which found evidence that bilamellar tarsal rotation has lower post‐operative trichiasis rates than tarsal advance and rotation. This unilamellar procedure was performed by placing the sutures through the marginal strip of the tarsal plate and a second set from the upper end of the tarsal plate into the anterior lamella. In Adamu 2002 a variant of the bilamellar tarsal rotation was compared to the posterior lamellar tarsal rotation (also known as the Trabut procedure), in which the sutures were placed above the lashes, similar to the bilamellar technique. The same unilamellar technique as that used by Adamu 2002 was used in a case series in East Africa with similar anatomical success (Bog 1993). It is possible that the difference in results reflects the difference in technique, but the studies also had differing follow‐up periods (times of assessment), which may have influenced the outcomes.

echnique as that used by Adamu 2002 was used in a case series in East Africa with similar anatomical success (Bog 1993). It is possible that the difference in results reflects the difference in technique, but the studies also had differing follow‐up periods (times of assessment), which may have influenced the outcomes. At present both the bilamellar and posterior lamellar tarsal rotation operations are both extensively used in trachoma‐endemic countries. There is currently no conclusive evidence that bilamellar surgery is superior to the posterior lamellar operation. Although cases of overcorrection and exposure were more common following bilamellar lid surgery, the risk was very low and the difference was not statistically significant in any study. It appears that both bilamellar and posterior lamellar tarsal rotation procedures are safe operations. The only common complication was post‐operative trichiasis. The use of a lid clamp in the bilamellar tarsal rotation procedure did not reduce post‐operative trichiasis but was associated with better eyelid contour outcomes and fewer granulomas (Gower 2013). Both lid rotation operations are relatively simple and require little equipment. Ophthalmologists, nurses, ophthalmology trainees and IECWs undertook the surgery in the included studies. It is possible that certain types of surgery are more effective than other types in specific situations. To our knowledge this has not yet been tested in any trial.

e relatively simple and require little equipment. Ophthalmologists, nurses, ophthalmology trainees and IECWs undertook the surgery in the included studies. It is possible that certain types of surgery are more effective than other types in specific situations. To our knowledge this has not yet been tested in any trial. Interventions to treat minor trichiasis by destroying the lashes, such as cryotherapy and electrolysis, appear to have low success rates in preventing recurrent trichiasis compared to bilamellar tarsal rotation surgery (Reacher 1992). As the equipment required is costly and can be difficult to maintain, this strategy is not recommended.

reat minor trichiasis by destroying the lashes, such as cryotherapy and electrolysis, appear to have low success rates in preventing recurrent trichiasis compared to bilamellar tarsal rotation surgery (Reacher 1992). As the equipment required is costly and can be difficult to maintain, this strategy is not recommended. Epilation Epilation is widely practised in most regions that have a high prevalence of trichiasis and it may have a role in the management of minor trichiasis where there are only a few peripheral lashes and a patient declines surgery. The study that investigated this was a non‐inferiority trial comparing enhanced epilation (high‐quality forceps and training) to posterior lamellar tarsal rotation surgery for minor trichiasis cases only (Rajak 2011a). The primary endpoint in that trial was progression to major trichiasis with a prespecified non‐inferiority margin of 10%. The cumulative risk of failure was 13.2% in the epilation group and 2.2% in the surgical group, with a risk difference of 11% (95% CI 8.1% to 13.9%, which includes the 10% non‐inferiority margin). Therefore, the trial provided an inconclusive result relative to the predefined 10% non‐inferiority margin. The proportion having any degree of trichiasis during follow‐up was significantly higher in the epilation group. However, over a two‐year period the important secondary outcome measures of change in visual acuity and corneal opacity showed no significant difference. It is notable that at two years only 31% of those still epilating accepted the offer of surgery.

sis during follow‐up was significantly higher in the epilation group. However, over a two‐year period the important secondary outcome measures of change in visual acuity and corneal opacity showed no significant difference. It is notable that at two years only 31% of those still epilating accepted the offer of surgery. Lid‐taping Double‐sided sticking plaster requires replacement of the plaster every week. As a temporary measure, however, the use of sticking plaster to evert the lashes is useful and in the study by Graz 1999 was superior to, and more comfortable than, epilation. Antibiotic treatment Post‐operative trichiasis was observed more often when the tarsal conjunctiva was inflamed (Burton 2005a; Burton 2005b; Rajak 2011b; West 2005). This clinically apparent inflammatory reaction could arise for a number of reasons: a smouldering immunologically driven process, infection with chlamydia or other bacteria, or mechanical irritation from the lashes (Burton 2010;Burton 2012). It is likely that chronic inflammation is the basis of progressive conjunctival scarring. Adjuvant therapy may reduce the risk of recurrent scarring and trichiasis.

smouldering immunologically driven process, infection with chlamydia or other bacteria, or mechanical irritation from the lashes (Burton 2010;Burton 2012). It is likely that chronic inflammation is the basis of progressive conjunctival scarring. Adjuvant therapy may reduce the risk of recurrent scarring and trichiasis. Three studies with heterogeneous designs, populations and study participants have investigated whether azithromycin can reduce trichiasis recurrence. Azithromycin is a broad spectrum antibiotic with some anti‐inflammatory properties. The first study was conducted in a programmatic context where many different surgeons were operating in The Gambia, a country with a very low prevalence of chlamydial infection, and recruited only major trichiasis cases (Burton 2005a). This study did not find an effect from azithromycin at one or four years, and overall recurrence rates were relatively high with significant inter‐surgeon heterogeneity. The second (and largest) study, which included both major and minor trichiasis from an area of Ethiopia with high chlamydial infection rates and a limited number of surgeons, found a significantly lower trichiasis recurrence rate in the azithromycin arm at one year, but not at three years (West 2006). The third study, from Nepal, did not find an overall significant difference at one year (Zhang 2006). A sub‐group analysis suggested a possible effect for major trichiasis cumulatively to one year. None of the other sub‐groups showed a significant effect. It is not possible to draw a definite conclusion about the impact of azithromycin on post‐operative trichiasis due to the study heterogeneity. The pooled analysis did not find a significant reduction in post‐operative trichiasis. It appears that where post‐operative trichiasis rates are low, there may be some benefit of azithromycin treatment; however, under operational/programmatic conditions with higher overall post‐operative trichiasis rates there was no observed effect. Non‐chlamydial bacteria have been frequently cultured from the conjunctiva of people with trichiasis (Burton 2005a).

is rates are low, there may be some benefit of azithromycin treatment; however, under operational/programmatic conditions with higher overall post‐operative trichiasis rates there was no observed effect. Non‐chlamydial bacteria have been frequently cultured from the conjunctiva of people with trichiasis (Burton 2005a). Alternative suture materials The study comparing silk with absorbable sutures found no difference in the recurrence rates at one or two years (Rajak 2011b). In many settings patients with trichiasis have to travel long distances to obtain surgical services. The typical practice is to use silk sutures, which have to be removed a week to 10 days post‐operatively. This incurs additional transport and time for both the patient and surgical team, adding to the overall cost of surgery. However, in many settings absorbable sutures are substantially more expensive than silk sutures. Occasionally silk sutures are not removed, which can cause granuloma formation and serious complications for the cornea if left in long‐term. Therefore, the use of absorbable sutures may be preferable for cost reasons where these can be afforded; a formal cost benefit analysis is needed to answer this question.

utures. Occasionally silk sutures are not removed, which can cause granuloma formation and serious complications for the cornea if left in long‐term. Therefore, the use of absorbable sutures may be preferable for cost reasons where these can be afforded; a formal cost benefit analysis is needed to answer this question. Surgery setting The uptake of trichiasis surgery is often low (Courtright 1994; West 1994); strategies that increase the proportion of patients who attend for surgery need to be developed. Bowman 2000 showed that it was less costly, and took less time, for participants to attend surgery in the community and that community‐based surgery was as safe and as effective as surgery in a health centre. However the uptake of surgery was only 20% better for village‐based surgery than for health centre‐based surgery (when analysed by cluster). It is not mentioned whether the resource implications for setting up village‐based surgery were greater than those required for health centre‐based surgery and it may be that this needs to be considered in the light of the reasonably small improvement in uptake. The village‐based approach to delivering surgery may work better in certain environments than others and should be considered along with other strategies for further research.

uired for health centre‐based surgery and it may be that this needs to be considered in the light of the reasonably small improvement in uptake. The village‐based approach to delivering surgery may work better in certain environments than others and should be considered along with other strategies for further research. Personnel performing surgery Most regions where trichiasis is prevalent have few ophthalmologists, and other health workers need to be trained to undertake the surgery in order to provide an adequate level of service. The evidence from the Alemayehu 2004 study suggests that surgery performed by a specially‐trained integrated eye worker is as successful as that undertaken by an ophthalmologist. Some caution needs to be applied in interpreting this result as the early recurrences (occurring by seven days) were excluded from the analysis. While this is very encouraging in terms of providing not only man‐power but also a high‐quality service, care needs to be taken to ensure appropriate specialist training is in place. Multiple studies have shown that success rates vary across surgeons (Burton 2005a; West 2005). This finding probably reflects varying levels of training, innate skill and experience, and serves to highlight the need not only for a high standard of basic training but also for ongoing monitoring and support.

is in place. Multiple studies have shown that success rates vary across surgeons (Burton 2005a; West 2005). This finding probably reflects varying levels of training, innate skill and experience, and serves to highlight the need not only for a high standard of basic training but also for ongoing monitoring and support. Secondary outcome measures Improvement in visual acuity or prevention of visual acuity deterioration is the primary long‐term aim of treating trachomatous trichiasis. However, this requires long‐term studies and therefore such evidence would be very difficult to obtain. The studies included here do not provide the opportunity to evaluate whether trichiasis surgery prevents long‐term loss of vision, because surgery has never been compared directly with no treatment. However, several studies have found modest but significant improvement in vision following surgery (Burton 2005a; Rajak 2011b; Reacher 1992; West 2006). Some of the visual impairment prior to surgery may be due to photophobia, lacrimation or irritation, or combinations of these three factors, caused by trichiatic lashes. Correction of the trichiasis should relieve these symptoms. Corneal scarring generally does not improve following surgery, although one study which compared corneal photographs before surgery to those at 2 years did find a degree of improvement in a minority of people (Rajak 2011b).

actors, caused by trichiatic lashes. Correction of the trichiasis should relieve these symptoms. Corneal scarring generally does not improve following surgery, although one study which compared corneal photographs before surgery to those at 2 years did find a degree of improvement in a minority of people (Rajak 2011b). Some trials have reported relatively high rates of eyelid contour abnormalities, which were less frequent with the use of a lid clamp (Rajak 2011b, Gower 2013). Other factors that are likely to affect patient satisfaction with surgery are comfort and appearance. Despite a lack of evidence for improvement or preservation of vision, Bowman 2002 reported 85% of operated patients were pleased with the outcome of surgery and 94% would recommend the operation to others. In the Burton 2005a study, 94% of patients said they were more comfortable a year after surgery, compared to their pre‐operative condition. Careful study of these important subjective factors may help provide a more complete picture of successful treatment and maybe even identify motivators for attendance. Quality of the evidence It is important to recognise the real difficulties of conducting studies in areas where resources and access to health care are limited, but there are several quality issues that must be highlighted in order for the studies to be interpreted appropriately. The grade of the evidence presented by many of the thirteen studies was high; however, there was some variability in the quality either due to small sample size or unclear or sub‐optimal methodology.

but there are several quality issues that must be highlighted in order for the studies to be interpreted appropriately. The grade of the evidence presented by many of the thirteen studies was high; however, there was some variability in the quality either due to small sample size or unclear or sub‐optimal methodology. For several the sample size was relatively low and probably insufficient to address the question (Dhaliwal 2005; Graz 1999; Reacher 1990; Zhang 2006). Most studies randomised participants; for several, however, sufficient information is not provided to determine the risk of bias associated with randomization (Adamu 2002; Alemayehu 2004; Graz 1999; Reacher 1990). In one study patients were assigned on an alternating basis (Zhang 2006), and this approach is vulnerable to bias because the assignment coordinator could adjust the enrolment order based on his/her preferences or beliefs. Participant masking to intervention was not possible in most studies and would not be expected to influence the outcome, since typically an independent observer evaluated a non‐subjective outcome (presence of lashes touching the eye). Masking of outcome assessors is more crucial for unbiased reporting; however, in four studies it is unclear whether masking was done (Adamu 2002; Bowman 2000; Graz 1999; Reacher 1990). While this must be considered a quality concern, the reality may have been that researchers were working with a limited number of eye care professionals and did not have the resources to exclude some staff from delivery of care in order to ensure adequate masking.

one (Adamu 2002; Bowman 2000; Graz 1999; Reacher 1990). While this must be considered a quality concern, the reality may have been that researchers were working with a limited number of eye care professionals and did not have the resources to exclude some staff from delivery of care in order to ensure adequate masking. The length of follow‐up in many of the included trials is quite short, limiting information regarding the long‐term effectiveness of the treatments. Long‐term follow‐up studies indicate that trichiasis continues to return after several years, but at a slower rate than during the first year post‐operatively (Rajak 2013; West 2006). With the exception of Reacher 1992, Burton 2005a, West 2006, Rajak 2011a, Rajak 2011b and Gower 2013, all studies had less than 12 months of follow‐up. Many areas where trachomatous trichiasis remains a significant public health problem are isolated and poor, and some of the population may be semi‐nomadic, making longer follow‐up difficult to achieve. The relatively short duration of follow‐up did, however, mean there were generally high levels of follow‐up. Authors' conclusions The evidence summarised in this review provides some indication of the basis for practice but one must remember that there are weaknesses in some of the data that may make the results unreliable.

The length of follow‐up in many of the included trials is quite short, limiting information regarding the long‐term effectiveness of the treatments. Long‐term follow‐up studies indicate that trichiasis continues to return after several years, but at a slower rate than during the first year post‐operatively (Rajak 2013; West 2006). With the exception of Reacher 1992, Burton 2005a, West 2006, Rajak 2011a, Rajak 2011b and Gower 2013, all studies had less than 12 months of follow‐up. Many areas where trachomatous trichiasis remains a significant public health problem are isolated and poor, and some of the population may be semi‐nomadic, making longer follow‐up difficult to achieve. The relatively short duration of follow‐up did, however, mean there were generally high levels of follow‐up. Authors' conclusions The evidence summarised in this review provides some indication of the basis for practice but one must remember that there are weaknesses in some of the data that may make the results unreliable. There is less risk of recurrence if the full thickness of the tarsal plate is incised and the lash‐bearing tissue is rotated away from the globe to evert the lid margin compared to procedures that do not involve a full‐thickness tarsotomy. This can be achieved by either unilamellar or bilamellar lid surgery. Operations such as tarsal grooving, tarsal advance and eversion splinting are less effective. The bilamellar tarsal rotation is better than either cryotherapy or electrolysis in rendering a patient trichiasis‐free.

involve a full‐thickness tarsotomy. This can be achieved by either unilamellar or bilamellar lid surgery. Operations such as tarsal grooving, tarsal advance and eversion splinting are less effective. The bilamellar tarsal rotation is better than either cryotherapy or electrolysis in rendering a patient trichiasis‐free. The optimal management of minor trichiasis remains uncertain. However, for patients with minor trichiasis who refuse surgery or where surgery is unavailable, epilation is a generally acceptable second‐line alternative treatment with comparable visual outcomes. Post‐operative azithromycin may be associated with reduced risk of post‐operative trichiasis; however, this effect is less certain under programmatic conditions than under high‐quality surgery by a few highly‐skilled integrated eye care workers. Silk and absorbable sutures have comparable outcomes. However, absorbable sutures have practical operational advantages in settings where they are affordable. Local health workers (nurses, medical assistants or non‐ophthalmologist doctors) may be trained to a level where they can perform trichiasis surgery as effectively as an ophthalmologist; however, it is essential that local health workers receive good training, surgical certification and follow‐up supportive supervision in order to help ensure high‐quality surgery is maintained. If uptake for trichiasis surgery is low in central settings, consideration should be given to providing it in patients' own communities.

Local health workers (nurses, medical assistants or non‐ophthalmologist doctors) may be trained to a level where they can perform trichiasis surgery as effectively as an ophthalmologist; however, it is essential that local health workers receive good training, surgical certification and follow‐up supportive supervision in order to help ensure high‐quality surgery is maintained. If uptake for trichiasis surgery is low in central settings, consideration should be given to providing it in patients' own communities. Recurrent trichiasis: the recurrence rate of trichiasis following surgery remains high in most programmatic settings, regardless of the intervention used. Recurrence is strongly associated with tarsal conjunctival inflammation, the cause of which is not well understood. Research is needed to improve understanding of the reasons why trichiasis recurs and to investigate targets for possible adjuvant therapy. Further trials are needed to compare the long‐term results of bilamellar and posterior lamellar tarsal rotation surgery, the two most commonly used operations. Since it is now known that the risk of recurrence is influenced by entropion severity, this must be taken into account in the design of future trials.

therapy. Further trials are needed to compare the long‐term results of bilamellar and posterior lamellar tarsal rotation surgery, the two most commonly used operations. Since it is now known that the risk of recurrence is influenced by entropion severity, this must be taken into account in the design of future trials. Acceptability of treatment: a major obstacle to successful trichiasis surgery is failure to attend for an operation. Further research is required to identify other means of increasing the proportion of people who attend for surgery. It may be possible to identify the perceived benefits of lid surgery to the patient—such as free transport or improved cosmesis—and to use these perceived benefits to persuade others to attend. Pattern of surgical provision: we do not yet know the most effective way of delivering trichiasis surgery: is it through a high‐volume surgical camp, or through a single surgeon working long‐term in a specific district? Further studies are required to determine not only which method will give the best uptake, but also which is most cost effective and is associated with the lowest recurrence rate. Quality of surgery: there is a need for research to understand the determinants of a successful surgical training programme. There is a need to assess methods to audit the outcome of surgery by individual surgeons. This information then needs to be fed back to improve training programmes.

Pattern of surgical provision: we do not yet know the most effective way of delivering trichiasis surgery: is it through a high‐volume surgical camp, or through a single surgeon working long‐term in a specific district? Further studies are required to determine not only which method will give the best uptake, but also which is most cost effective and is associated with the lowest recurrence rate. Quality of surgery: there is a need for research to understand the determinants of a successful surgical training programme. There is a need to assess methods to audit the outcome of surgery by individual surgeons. This information then needs to be fed back to improve training programmes. Quality of life: although it has been shown that most trichiasis patients are happy with the results of their surgery the impact of surgery on quality of life and what might enhance this is a neglected area of research, Acknowledgements The editorial team of the Cochrane Eyes and Vision Group prepared and executed the electronic searches. We are grateful to Amir Bedri Kello and Barbara Hawkins for their comments on the 2015 version of the review.

Quality of life: although it has been shown that most trichiasis patients are happy with the results of their surgery the impact of surgery on quality of life and what might enhance this is a neglected area of research, Acknowledgements The editorial team of the Cochrane Eyes and Vision Group prepared and executed the electronic searches. We are grateful to Amir Bedri Kello and Barbara Hawkins for their comments on the 2015 version of the review. Appendices Appendix 1. CENTRAL search strategy #1 MeSH descriptor: [Trachoma] explode all trees #2 MeSH descriptor: [Trichiasis] this term only #3 MeSH descriptor: [Chlamydia trachomatis] explode all trees #4 MeSH descriptor: [Eye] explode all trees #5 #3 and #4 #6 trachoma* or tracoma* or trichiasis #7 MeSH descriptor: [Entropion] explode all trees #8 entropion #9 #6 or #7 or #8 #10 #1 or #2 or #5 or #9 #11 MeSH descriptor: [Ophthalmologic Surgical Procedures] explode all trees #12 MeSH descriptor: [Eyelids] explode all trees #13 MeSH descriptor: [Eyelid Diseases] explode all trees #14 surg* or operat* or incis* #15 tars* near/3 rotat* #16 tars* near/3 plat* #17 tarsotom* or trabut or ballen or streatfield #18 epilat$ or electrolysis or cryo* #19 (tap* or plaster*) near (eye*) #20 (tap* or plaster*) near (lid*) #21 #11 or #12 or #13 or #14 or #15 or #16 or #17 or #18 or #19 or #20 #22 #10 and #21

#14 surg* or operat* or incis* #15 tars* near/3 rotat* #16 tars* near/3 plat* #17 tarsotom* or trabut or ballen or streatfield #18 epilat$ or electrolysis or cryo* #19 (tap* or plaster*) near (eye*) #20 (tap* or plaster*) near (lid*) #21 #11 or #12 or #13 or #14 or #15 or #16 or #17 or #18 or #19 or #20 #22 #10 and #21 Appendix 2. MEDLINE (Ovid) search strategy 1. randomized controlled trial.pt. 2. (randomized or randomised).ab,ti. 3. placebo.ab,ti. 4. dt.fs. 5. randomly.ab,ti. 6. trial.ab,ti. 7. groups.ab,ti. 8. or/1‐7 9. exp animals/ 10. exp humans/ 11. 9 not (9 and 10) 12. 8 not 11 13. exp trachoma/ 14. trichiasis/ 15. exp chlamydia‐trachomatis/ 16. exp eye/ 17. 15 and 16 18. (trachoma$ or tracoma$ or trichiasis).tw. 19. exp entropion/ 20. entropion.tw. 21. or/18‐20 22. 13 or 14 or 17 or 21 23. exp ophthalmologic surgical procedures/ 24. exp eyelids/ 25. exp eyelid diseases/ 26. (surg$ or operat$ or incis$).tw. 27. (tars$ adj3 rotat$).tw. 28. (tars$ adj3 plate).tw. 29. (tarsotom$ or trabut or ballen or streatfield).tw. 30. (epilat$ or electrolysis or cryo$).tw. 31. ((tap$ or plaster$) adj5 eye$).tw. 32. ((tap$ or plaster$) adj5 lid$).tw. 33. or/23‐32 34. 22 and 33 35. 12 and 34 36. (201307$ or 201308$ or 201309$ or 201310$ or 201311$ or 201312$ or 2014$ or 2015$).ed. 37. 35 and 36 The search filter for trials at the beginning of the MEDLINE strategy is from the published paper by Glanville (Glanville 2006).

Appendix 2. MEDLINE (Ovid) search strategy 1. randomized controlled trial.pt. 2. (randomized or randomised).ab,ti. 3. placebo.ab,ti. 4. dt.fs. 5. randomly.ab,ti. 6. trial.ab,ti. 7. groups.ab,ti. 8. or/1‐7 9. exp animals/ 10. exp humans/ 11. 9 not (9 and 10) 12. 8 not 11 13. exp trachoma/ 14. trichiasis/ 15. exp chlamydia‐trachomatis/ 16. exp eye/ 17. 15 and 16 18. (trachoma$ or tracoma$ or trichiasis).tw. 19. exp entropion/ 20. entropion.tw. 21. or/18‐20 22. 13 or 14 or 17 or 21 23. exp ophthalmologic surgical procedures/ 24. exp eyelids/ 25. exp eyelid diseases/ 26. (surg$ or operat$ or incis$).tw. 27. (tars$ adj3 rotat$).tw. 28. (tars$ adj3 plate).tw. 29. (tarsotom$ or trabut or ballen or streatfield).tw. 30. (epilat$ or electrolysis or cryo$).tw. 31. ((tap$ or plaster$) adj5 eye$).tw. 32. ((tap$ or plaster$) adj5 lid$).tw. 33. or/23‐32 34. 22 and 33 35. 12 and 34 36. (201307$ or 201308$ or 201309$ or 201310$ or 201311$ or 201312$ or 2014$ or 2015$).ed. 37. 35 and 36 The search filter for trials at the beginning of the MEDLINE strategy is from the published paper by Glanville (Glanville 2006). Appendix 3. EMBASE (Ovid) search strategy 1. exp randomized controlled trial/ 2. exp randomization/ 3. exp double blind procedure/ 4. exp single blind procedure/ 5. random$.tw. 6. or/1‐5 7. (animal or animal experiment).sh. 8. human.sh. 9. 7 and 8 10. 7 not 9 11. 6 not 10 12. exp clinical trial/ 13. (clin$ adj3 trial$).tw. 14. ((singl$ or doubl$ or trebl$ or tripl$) adj3 (blind$ or mask$)).tw. 15. exp placebo/ 16. placebo$.tw. 17. random$.tw. 18. exp experimental design/ 19. exp crossover procedure/ 20. exp control group/ 21. exp latin square design/ 22. or/12‐21 23. 22 not 10 24. 23 not 11 25. exp comparative study/ 26. exp evaluation/ 27. exp prospective study/ 28. (control$ or prospectiv$ or volunteer$).tw. 29. or/25‐28 30. 29 not 10

.tw. 17. random$.tw. 18. exp experimental design/ 19. exp crossover procedure/ 20. exp control group/ 21. exp latin square design/ 22. or/12‐21 23. 22 not 10 24. 23 not 11 25. exp comparative study/ 26. exp evaluation/ 27. exp prospective study/ 28. (control$ or prospectiv$ or volunteer$).tw. 29. or/25‐28 30. 29 not 10 Appendix 4. ISRCTN search strategy (Trachoma OR Tracoma OR Trichiasis) Appendix 5. ClinicalTrials.gov search strategy (Trachoma OR Tracoma OR Trichiasis) Appendix 6. ICTRP search strategy Trachoma OR Tracoma OR Trichiasis Data and analyses Comparison 1 Bilamellar tarsal rotation versus tarsal advance and rotation Outcome or subgroup title No. of studies No. of participants Statistical method Effect size 1 One or more lashes touching globe at nine monthsAnalysis 1.1 Comparison 1 Bilamellar tarsal rotation versus tarsal advance and rotation, Outcome 1 One or more lashes touching globe at nine months. 2 260 Odds Ratio (M‐H, Fixed, 95% CI) 0.29 [0.16, 0.50] 2 Overcorrection following surgeryAnalysis 1.2 Comparison 1 Bilamellar tarsal rotation versus tarsal advance and rotation, Outcome 2 Overcorrection following surgery. 2 312 Odds Ratio (M‐H, Fixed, 95% CI) 2.57 [0.28, 23.25] 3 Defective lid closure following surgeryAnalysis 1.3 Comparison 1 Bilamellar tarsal rotation versus tarsal advance and rotation, Outcome 3 Defective lid closure following surgery. 2 312 Odds Ratio (M‐H, Fixed, 95% CI) 1.90 [0.29, 12.37] Comparison 2 Peri‐operative azithromycin versus no azithromycin

2 312 Odds Ratio (M‐H, Fixed, 95% CI) 2.57 [0.28, 23.25] 3 Defective lid closure following surgeryAnalysis 1.3 Comparison 1 Bilamellar tarsal rotation versus tarsal advance and rotation, Outcome 3 Defective lid closure following surgery. 2 312 Odds Ratio (M‐H, Fixed, 95% CI) 1.90 [0.29, 12.37] Comparison 2 Peri‐operative azithromycin versus no azithromycin Outcome or subgroup title No. of studies No. of participants Statistical method Effect size 1 One or more lashes touching the globe at one yearAnalysis 2.1 Comparison 2 Peri‐operative azithromycin versus no azithromycin, Outcome 1 One or more lashes touching the globe at one year. 3 1954 Odds Ratio (M‐H, Random, 95% CI) 0.85 [0.63, 1.14] 2 One or more lashes touching the globe at longer follow‐upAnalysis 2.2 Comparison 2 Peri‐operative azithromycin versus no azithromycin, Outcome 2 One or more lashes touching the globe at longer follow‐up. 2 1589 Odds Ratio (M‐H, Random, 95% CI) 0.80 [0.60, 1.05] 2.1 Three years 1 1322 Odds Ratio (M‐H, Random, 95% CI) 0.76 [0.54, 1.07] 2.2 Four years 1 267 Odds Ratio (M‐H, Random, 95% CI) 0.87 [0.53, 1.41] What's new Last assessed as up‐to‐date: 7 May 2015. Date Event Description 14 December 2016 Amended Amendment to Analysis 2.1 History Protocol first published: Issue 1, 2003 Review first published: Issue 3, 2006

2 1589 Odds Ratio (M‐H, Random, 95% CI) 0.80 [0.60, 1.05] 2.1 Three years 1 1322 Odds Ratio (M‐H, Random, 95% CI) 0.76 [0.54, 1.07] 2.2 Four years 1 267 Odds Ratio (M‐H, Random, 95% CI) 0.87 [0.53, 1.41] What's new Last assessed as up‐to‐date: 7 May 2015. Date Event Description 14 December 2016 Amended Amendment to Analysis 2.1 History Protocol first published: Issue 1, 2003 Review first published: Issue 3, 2006 Date Event Description 15 December 2015 Amended Minor amendment to Analysis 2:1: change from 'subtotals only' to 'no totals' 19 May 2015 New search has been performed Issue 11, 2015: Electronic searches have been updated 19 May 2015 New citation required and conclusions have changed Issue 11, 2015: 6 studies (4 new: Gower 2013; Rajak 2011a; Rajak 2011b; Zhang 2006 and 2 from previous searches: Dhaliwal 2005; West 2006) have been included in this update 29 October 2008 Amended Converted to new review format. 1 March 2006 New citation required and conclusions have changed Substantive amendment Differences between protocol and review The protocol for this review was originally published in 2003 (Yorston 2003). Since that time Cochrane methods have changed and we have incorporated these changes, in particular assessment of risk of bias in included studies. Characteristics of studies Characteristics of included studies [ordered by study ID] Adamu 2002

Differences between protocol and review The protocol for this review was originally published in 2003 (Yorston 2003). Since that time Cochrane methods have changed and we have incorporated these changes, in particular assessment of risk of bias in included studies. Characteristics of studies Characteristics of included studies [ordered by study ID] Adamu 2002 Methods Study Design: parallel group randomised trial of different operations for trichiasis. Eyes: for cases who only had trichiasis affecting one eye only that eye was included. In bilateral cases both eyes were included: surgery type was allocated randomly for the right eye and then the left eye was allocated to the other procedure. Analysis was not adjusted for bilateral cases. Participants Country: Ethiopia Setting: teaching hospital Number of participants: 153 patients, 256 eyes Lost to follow up: 19 Male: 32, Female: 109 (gender unknown for 19 not followed up) Interventions (1) Bilamellar tarsal rotation (2) Posterior lamellar tarsal rotation Outcomes Primary outcome: recurrent trichiasis (one or more lashes touching eye) Secondary outcomes: overcorrection; complications Follow up: 3 months Reported improvement in vision, but no details given Notes The unilamellar operation is different to the one described in Reacher 1990 and Reacher 1992 Funder: not specified Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Unclear risk An unspecified 'Lottery method' used. In bilateral cases the left eye always received the opposite treatment to that randomly allocated to the right eye. Allocation concealment (selection bias) Unclear risk No description. Blinding of participants and personnel (performance bias) All outcomes Unclear risk No description. Blinding of outcome assessment (detection bias) All outcomes Unclear risk No description. Incomplete outcome data (attrition bias) All outcomes Unclear risk No description of the individuals lost to follow‐up. Selective reporting (reporting bias) Unclear risk No study protocol available. Limited information on analytical protocol in methods. Alemayehu 2004

ion bias) All outcomes Unclear risk No description. Incomplete outcome data (attrition bias) All outcomes Unclear risk No description of the individuals lost to follow‐up. Selective reporting (reporting bias) Unclear risk No study protocol available. Limited information on analytical protocol in methods. Alemayehu 2004 Methods Study Design: parallel group randomised trial of different levels of surgical training and background for surgeons performing trichiasis surgery Eyes: in bilateral cases both eyes were included. No adjustment was made in the analysis. Participants Country: Ethiopia Setting: unclear Number of participants: 982 patients (1750 eyes) Lost to follow up: 464 eyes Female: 77.2% (absolute numbers not provided) Interventions (1) Bilamellar tarsal rotation carried out by ophthalmologist (2) Bilamellar tarsal rotation carried out by integrated eye worker Outcomes Primary outcome: recurrent trichiasis (one or more lashes touching eye) Secondary outcomes: severity of recurrence Follow up: 3 months Notes Funder: not specified Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Unclear risk No description: "were randomised" but no further details provided. Allocation concealment (selection bias) Unclear risk No description. Blinding of participants and personnel (performance bias) All outcomes Low risk No masking of participants or surgical personnel, but unlikely to have influenced the outcome. Blinding of outcome assessment (detection bias) All outcomes Low risk Follow‐up by ophthalmologists who were not involved in the surgery and masked to the allocation. Incomplete outcome data (attrition bias) All outcomes High risk Participants were excluded post‐randomisation if surgery failed. 269 patients were lost to follow‐up at 3 months. . Selective reporting (reporting bias) Unclear risk No study protocol available. Limited information on analytical protocol in methods. Bowman 2000

plete outcome data (attrition bias) All outcomes High risk Participants were excluded post‐randomisation if surgery failed. 269 patients were lost to follow‐up at 3 months. . Selective reporting (reporting bias) Unclear risk No study protocol available. Limited information on analytical protocol in methods. Bowman 2000 Methods Study Design: cluster randomised study of surgery provided in the patients' village compared to surgery provided at a local health centre. Eyes: unclear how bilateral cases were analysed. Participants Country: The Gambia Setting: village and health centre‐based surgery Number of participants: 8 pairs of villages (156 participants) Lost to follow up: 0 All participants had major trichiasis pre‐operatively Interventions (1) Posterior lamellar tarsal rotation surgery in village (2) Posterior lamellar tarsal rotation surgery in health centre Outcomes Primary Outcome: surgical uptake Secondary Outcomes: recurrent trichiasis, time taken from home to surgery, cost, complications Follow up: 1 week and 3 months Notes Funder: Sight Savers International Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Low risk Quote: "Clusters were paired by district or geographical proximity. One cluster from each pair was randomly allocated to receive village‐based surgery and the other, health centre‐based surgery." Method of sequence generation not specified. Allocation concealment (selection bias) Low risk All clusters randomly allocated at baseline, before the TT patients were identified. Blinding of participants and personnel (performance bias) All outcomes Low risk Participants and personnel not masked to community allocation. Unlikely to have influenced the outcome. Blinding of outcome assessment (detection bias) All outcomes Unclear risk No description. Incomplete outcome data (attrition bias) All outcomes Low risk Data complete. Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. Burton 2005a

ve influenced the outcome. Blinding of outcome assessment (detection bias) All outcomes Unclear risk No description. Incomplete outcome data (attrition bias) All outcomes Low risk Data complete. Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. Burton 2005a Methods Study Design: parallel group randomised trial of peri‐operative azithromycin treatment Eyes: only one eye per participant was included in the analysis. In bilateral cases, the eye with more severe baseline trichiasis was included. Participants Country: The Gambia Setting: health centres Number of participants: 451 Lost to follow up: 25 Male: 133, Female: 318 A consecutive, geographically defined sub‐set of the original participants was followed‐up at 4 years. Number of participants: 356 Lost to follow up: 90 Male: 103, Female: 253 Major trichiasis cases only Interventions (1) Posterior lamellar tarsal rotation surgery (no placebo) (2) Posterior lamellar tarsal rotation surgery followed by single‐dose azithromycin treatment at baseline and 6 months, administered to the patient and the members of their household Outcomes Primary Outcome: recurrent trichiasis by one year Secondary Outcomes: visual acuity by reduced LogMAR chart, conjunctival infection, corneal opacification Follow up: 12 months and 4 years (subset) Notes Funder: International Trachoma Initiative Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Low risk Quote "Separate randomisation sequences were generated for each surgeon from random number tables and blocked in groups of four" Allocation concealment (selection bias) Low risk Quote "Treatment allocations were concealed in opaque sequentially numbered envelopes" Blinding of participants and personnel (performance bias) All outcomes Low risk Quote "Immediately following surgery a nurse not involved in clinical assessments or surgery administered the randomly allocated treatment." Participants were not masked to allocation, as no placebo was used. Blinding of outcome assessment (detection bias) All outcomes Low risk Quote "Examiners were ‘‘blinded’’ to treatment allocation." Incomplete outcome data (attrition bias) All outcomes Low risk 98% follow‐up of those still alive at 1 year. At 4 years 94% of the surviving patients were re‐examined Selective reporting (reporting bias) Low risk Clear analytical approach in methods. Dhaliwal 2005

e "Examiners were ‘‘blinded’’ to treatment allocation." Incomplete outcome data (attrition bias) All outcomes Low risk 98% follow‐up of those still alive at 1 year. At 4 years 94% of the surviving patients were re‐examined Selective reporting (reporting bias) Low risk Clear analytical approach in methods. Dhaliwal 2005 Methods Study Design: parallel groups randomised to one of three procedures Eyes: in bilateral cases both eyes were separately randomised. No adjustment was made for this in the analysis. Participants Country: India Setting: hospital ophthalmology department Number of participants: 77 patients (90 eyes) Lost to follow up: 0 eyes Female: unclear as % is provided by eye and not person Interventions (1) Terminal tarsotomy and rotation (2) Tarsal advance and rotation (3) Tarsal grooving Outcomes Post‐opearative trichiasis and/or entropion at 6 months Notes Terminal tarsotomy and rotation is different from the PLTR/Trabut Funder: not specified Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Unclear risk Method for the allocation for the second eye is unclear and excluded the method used on the first eye to be operated. Allocation concealment (selection bias) Unclear risk No description. Blinding of participants and personnel (performance bias) All outcomes Unclear risk No description. Blinding of outcome assessment (detection bias) All outcomes Unclear risk No description. Incomplete outcome data (attrition bias) All outcomes Low risk High follow‐up. Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. Gower 2013

nclear risk No description. Blinding of outcome assessment (detection bias) All outcomes Unclear risk No description. Incomplete outcome data (attrition bias) All outcomes Low risk High follow‐up. Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. Gower 2013 Methods Study Design: parallel group randomised controlled trial bilamellar tarsal rotation surgery with or without a TT clamp Eyes: both eyes of bilateral cases were included in the analysis. Appropriate adjustment was made in the analysis for bilateral cases. Participants Country: Tanzania Setting: health centres Number of participants: 1917 Number of Eyes: 3345 Lost to follow up: 51 Male: 490, Female: 1427 Interventions (1) Standard bilamellar tarsal rotation (2) Bilamellar tarsal rotation performed with a TT clamp Outcomes Primary Outcome: was defined as one or more of the following: recurrent trichiasis, eyelid contour abnormality or granuloma. Secondary outcomes: post‐operative trichiasis, eyelid contour abnormality or granuloma individually. Follow‐up: 6 weeks, 12 months, 24 months. Notes 18 different surgical technicians performed the surgery. Surgeons were randomly allocated at the outset to only perform one of the two types of operation. Each recruitment day there were 4 surgical technicians operating – 2 standard procedure and 2 with the TT clamp. Patients with TT were randomly allocated to one of the 4 surgical technicians operating. Funder: Bill and Melinda Gates Foundation Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Low risk Quote "Randomization assignments were created using permuted block sizes of 6 to 12". Allocation concealment (selection bias) Low risk Quote "Randomization assignments were placed in opaque envelopes and stored in a locked office until distribution". Blinding of participants and personnel (performance bias) All outcomes Low risk Participants and surgical technicians were not masked to the allocation; however, this is unlikely to have affected the outcome measure. Blinding of outcome assessment (detection bias) All outcomes Low risk Quote "Examinations were conducted by a single trained examiner who was not involved in enrolment, randomization, or surgical procedures". Incomplete outcome data (attrition bias) All outcomes Low risk 98% follow‐up rate. Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. Graz 1999

conducted by a single trained examiner who was not involved in enrolment, randomization, or surgical procedures". Incomplete outcome data (attrition bias) All outcomes Low risk 98% follow‐up rate. Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. Graz 1999 Methods Study Design: parallel group randomised trial of epilation versus sticking tape. Eyes: in bilateral cases both eyes were treated the same way and the data from both included in the analysis. The authors combined the data from both eyes to give a single 'clinical status' outcome. Participants Country: China Setting: hospital and home Number of participants: 57 (68 eyes) Lost to follow up: 0 Male: 20, Female: 37 Interventions (1) Epilation alone for 12 weeks (2) Sticking plaster for 12 weeks (3) Sticking plaster for 8 weeks, followed by cross‐over to epilation Outcomes Successful outcome: no lashes touching the eye. Secondary outcomes: visual acuity by Snellen E‐chart, corneal status, patient‐reported symptoms by questionnaire. Follow up: 3 months Notes The group that had sticking plaster for 8 weeks followed by epilation has been excluded from our analysis as this group received both interventions. Funder: Swiss Agency for Development and Cooperation Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Unclear risk No description. Allocation concealment (selection bias) Unclear risk No description. Blinding of participants and personnel (performance bias) All outcomes Unclear risk Personnel not masked; performance risk possible. Blinding of outcome assessment (detection bias) All outcomes High risk Observer not masked. Quote "A blinded design was not possible for obvious reasons, but data entry and analysis were blinded." Incomplete outcome data (attrition bias) All outcomes Low risk Data complete. Selective reporting (reporting bias) Low risk Data complete. Rajak 2011a

ent (detection bias) All outcomes High risk Observer not masked. Quote "A blinded design was not possible for obvious reasons, but data entry and analysis were blinded." Incomplete outcome data (attrition bias) All outcomes Low risk Data complete. Selective reporting (reporting bias) Low risk Data complete. Rajak 2011a Methods Study Design: parallel group randomised trial of epilation versus surgery for minor trichiasis. Non‐inferiority trial. Eyes: only one eye per participant was included in the analysis. In bilateral cases, one eye was randomly assigned to be the 'study eye', although both eyes were treated, if indicated. Participants Country: Ethiopia Setting: health centres Number of participants: 1300 Lost to follow up: 81 Male: 537, Female: 863 Minor trichiasis cases only Interventions (1) Posterior lamellar tarsal rotation (2) Epilation with new forceps and training Outcomes Primary Outcome: proportion of individuals at any follow‐up who had “failed” defined as either (1) five or more eyelashes touching the globe or (2) a history of surgery performed in the trial eye at any point during the follow‐up period (in the case of the surgical arm this would be repeat surgery). A pre‐specified non‐inferiority margin was 10%. Secondary Outcomes: corneal opacity change, visual acuity by LogMAR E‐chart, entropion, conjunctival inflammation and symptoms by questionnaire. Follow up at 6, 12, 18 and 24 months. Notes Funder: Band Aid Foundation and The Wellcome Trust. Johnson & Johnson donated all sutures. Tweezerman donated all forceps. Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Low risk Quote "Participants were randomly allocated to the epilation or surgery groups using a 1:1 allocation ratio for each surgeon, using a computer‐generated randomisation sequence with random block sizes. Randomisation was stratified by surgeon because of possible intersurgeon variability." Allocation concealment (selection bias) Low risk Quote "The random allocation sequences for each surgeon were concealed in sequentially numbered, sealed, opaque envelopes, which were colour coded for surgeon and placed in separate containers for each surgeon. The person who prepared these envelopes was independent of all other aspects of the trial." Blinding of participants and personnel (performance bias) All outcomes Low risk Participants were not masked to their allocation, as they either received surgery or practiced epilation.

d in separate containers for each surgeon. The person who prepared these envelopes was independent of all other aspects of the trial." Blinding of participants and personnel (performance bias) All outcomes Low risk Participants were not masked to their allocation, as they either received surgery or practiced epilation. Other than the surgeons performing the operation and the epilation trainer, all other personnel were masked to the allocation. Blinding of outcome assessment (detection bias) All outcomes Low risk Quote "The two individuals responsible for all the clinical outcome measurements were masked to the allocation. At follow‐up, the trichiasis and corneal examination was performed and recorded before the eyelid was everted, so that the examiner was masked to whether surgery had been performed on the tarsal conjunctiva." Incomplete outcome data (attrition bias) All outcomes Low risk 93.5% follow‐up at 24 months. Selective reporting (reporting bias) Low risk Study protocol published online. Rajak 2011b

ecorded before the eyelid was everted, so that the examiner was masked to whether surgery had been performed on the tarsal conjunctiva." Incomplete outcome data (attrition bias) All outcomes Low risk 93.5% follow‐up at 24 months. Selective reporting (reporting bias) Low risk Study protocol published online. Rajak 2011b Methods Study Design: parallel group randomised trial of absorbable versus silk sutures for major trichiasis. Eyes: only one eye per participant was included in the analysis. In bilateral cases, one eye was randomly assigned to be the 'study eye', although both eyes were treated, if indicated. Participants Country: Ethiopia Setting: health centres Number of participants: 1300 Lost to follow up: 82 Male: 285, Female: 1015 Interventions (1) Posterior lamellar tarsal rotation with absorbable sutures (2) Posterior lamellar tarsal rotation with silk sutures Outcomes Primary Outcome: proportion of those individuals seen at the 12‐month follow‐up who were found to have either (1) recurrent trichiasis, defined as one or more lashes touching the eye or clinical evidence of epilation; or (2) a history of repeat TT surgery during the first year. Secondary Outcomes: corneal opacity change, visual acuity by LogMAR E‐chart, entropion, conjunctival inflammation and symptoms by questionnaire. Follow up at 6, 12, 18 and 24 months. Notes Funder: Band Aid Foundation and The Wellcome Trust. Johnson & Johnson donated all sutures. Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Low risk Quote "Participants were randomly allocated to the silk or polyglactin‐910 suture arms using a 1:1 allocation ratio for each surgeon, using a computer‐generated randomisation sequence with random block sizes. Randomisation was stratified by surgeon because of possible inter‐surgeon variability." Allocation concealment (selection bias) Low risk Quote "The random allocation sequences for each surgeon were concealed in sequentially numbered, sealed, opaque envelopes, which were placed in separate containers for each surgeon; a person independent of all other aspects of the trial prepared these envelopes." Blinding of participants and personnel (performance bias) All outcomes Low risk Quote "Participants and surgeons were aware of the suture allocation." Other personnel involved in the study were masked to the allocation.

for each surgeon; a person independent of all other aspects of the trial prepared these envelopes." Blinding of participants and personnel (performance bias) All outcomes Low risk Quote "Participants and surgeons were aware of the suture allocation." Other personnel involved in the study were masked to the allocation. Blinding of outcome assessment (detection bias) All outcomes Low risk Quote "The two individuals who were responsible for all the clinical observations in this trial were masked to the allocation." Incomplete outcome data (attrition bias) All outcomes Low risk 95.1% follow‐up at 12 months. Selective reporting (reporting bias) Low risk Study protocol published online. Reacher 1990

ow risk Quote "The two individuals who were responsible for all the clinical observations in this trial were masked to the allocation." Incomplete outcome data (attrition bias) All outcomes Low risk 95.1% follow‐up at 12 months. Selective reporting (reporting bias) Low risk Study protocol published online. Reacher 1990 Methods Study Design: parallel group randomised trial of five different operations for trachomatous trichiasis. Eyes: only one eye per participant was included in the analysis. Unclear how lids were selected in bilateral cases. Participants Country: Oman Setting: not specified Number of participants: 165 (165 eyes) Lost to follow‐up: 12 Gender: not specified Interventions (1) Bilamellar tarsal rotation (2) Tarsal advance and rotation (3) Eversion splinting (4) Tarsal advance (5) Tarsal grooving Outcomes Primary outcome: "treatment success" defined as no eyelash/eyeball contact in the primary position and complete eyelid closure. Mean follow up: 7.9 months. A follow‐up schedule is not specified beyond 14 days. Notes Three surgeons performed the surgery, with two surgeons performing relatively few of the operations. Funder: not specified Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Unclear risk Limited information. Quote "operations allocated by random number tables". Allocation concealment (selection bias) Unclear risk No comment on concealment of the random sequence. Blinding of participants and personnel (performance bias) All outcomes Unclear risk Participants and surgeon were not masked to the allocation. Blinding of outcome assessment (detection bias) All outcomes Unclear risk Not specified. Incomplete outcome data (attrition bias) All outcomes Low risk 93% follow‐up rate. Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. Reacher 1992

ked to the allocation. Blinding of outcome assessment (detection bias) All outcomes Unclear risk Not specified. Incomplete outcome data (attrition bias) All outcomes Low risk 93% follow‐up rate. Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. Reacher 1992 Methods Study Design: three separate parallel group trials for minor trichiasis, major trichiasis and defective lid closure. Participants Country: Oman Setting: hospital or health centre Number of participants: 367 (172 eyes with minor trichiasis, 200 eyes with major trichiasis, 21 eyes with lid closure defect) Lost to follow up: 24 Interventions Minor trichiasis: (1) Bilamellar tarsal rotation (2) Electrolysis (3) Cryotherapy Major trichiasis: (1) Bilamellar tarsal rotation (2) Tarsal advance and rotation Defective lid closure: (1) Tarsal advance and rotation (2) Tarsal advance with graft Outcomes Primary outcome: "Successful outcome" defined as no eyelash/eyeball contact in the primary position, complete eyelid closure, no overcorrection and a cosmetically acceptable outcome. Notes The group with lid closure defects was too small to be analysed separately and has been excluded from the analysis. Funder: Edna McConnell Clarke Foundation and National Institutes of Health. Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Low risk Quote: "Allocation was made by opening a sequence of sealed envelopes specific for lid grade that contained one of a random series of operations generated by computer." Allocation concealment (selection bias) Low risk Quote: "Allocation was made by opening a sequence of sealed envelopes specific for lid grade that contained one of a random series of operations generated by computer." Blinding of participants and personnel (performance bias) All outcomes Unclear risk Participants and surgeons were not masked to the allocation. Blinding of outcome assessment (detection bias) All outcomes Low risk Masked independent assessor. Incomplete outcome data (attrition bias) All outcomes Unclear risk 96% follow‐up, but very variable time frames (1.2 to 25 months). Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. West 2006

etection bias) All outcomes Low risk Masked independent assessor. Incomplete outcome data (attrition bias) All outcomes Unclear risk 96% follow‐up, but very variable time frames (1.2 to 25 months). Selective reporting (reporting bias) Low risk No study protocol available. Clear analytical approach in methods. West 2006 Methods Design: parallel group randomised trial of peri‐operative azithromycin treatment. Eyes: only one eye per participant was included in the analysis. In bilateral cases, the study eye was determined by the study number (left eye if an odd number). Participants Country: Ethiopia Setting: health centres Number of participants: 1452 Lost to follow‐up: participants were followed until they developed recurrence or until 1 year post‐operatively, whichever came first. 98% of eligible subjects were followed at 1 year. Male: 331, Female: 1121 Number of participants: 1452 Lost to follow‐up: 30 Male: 331, Female: 1121 Interventions (1) Bilamellar tarsal rotation with tetracycline eye ointment (2) Bilamellar tarsal rotation with azithromycin to the patient only (3) Bilamellar tarsal rotation with azithromycin to the patient and their family Outcomes Primary Outcome: post‐operative trichiasis 1 year and 3 years after surgery Secondary Outcomes: visual acuity reported in a subset of the trial participants Notes Funder: National Eye Institute. Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) Low risk Random sequence, with variable block size. Allocation concealment (selection bias) Low risk Allocation concealed in envelopes. Blinding of participants and personnel (performance bias) All outcomes Low risk Participants were not blinded to the allocation. Blinding of outcome assessment (detection bias) All outcomes Low risk Outcome assessments were performed by a masked grader. Incomplete outcome data (attrition bias) All outcomes Low risk 98% follow‐up at one year; 91.1% follow‐up at three years. Selective reporting (reporting bias) Low risk Study protocol published and no evidence of selective reporting of outcomes Zhang 2006

s Low risk Outcome assessments were performed by a masked grader. Incomplete outcome data (attrition bias) All outcomes Low risk 98% follow‐up at one year; 91.1% follow‐up at three years. Selective reporting (reporting bias) Low risk Study protocol published and no evidence of selective reporting of outcomes Zhang 2006 Methods Design: parallel group randomised trial of peri‐operative azithromycin treatment. Eyes: bilateral cases both eyes included. Participants Country: Nepal Setting: field camps Number of participants: 109 (148 eyes operated) Lost to follow‐up: uncertain, as reported by eye rather than person Male: 29, Female: 80 Interventions (1) Bilamellar tarsal rotation with placebo (2) Bilamellar tarsal rotation with azithromycin, single dose Outcomes Primary outcome: recurrent trichiasis (1+ lash touching the eye) Notes Funder: Fight for Sight, International Trachoma Initiative, National Institutes of Health. Risk of bias Bias Authors' judgement Support for judgement Random sequence generation (selection bias) High risk No random sequence. Patients allocated to azithromycin and placebo on an alternating basis. Allocation concealment (selection bias) High risk No concealment. Blinding of participants and personnel (performance bias) All outcomes Low risk Placebo used. Blinding of outcome assessment (detection bias) All outcomes Low risk Examinations by masked graders. Incomplete outcome data (attrition bias) All outcomes Unclear risk Some minor internal inconsistencies data. Selective reporting (reporting bias) High risk Excluded "surgical failures" from the analysis, defined as five+ lashes touching the globe at 3 months. TT = trachomatous trichiasis

Examinations by masked graders. Incomplete outcome data (attrition bias) All outcomes Unclear risk Some minor internal inconsistencies data. Selective reporting (reporting bias) High risk Excluded "surgical failures" from the analysis, defined as five+ lashes touching the globe at 3 months. TT = trachomatous trichiasis Contributions of authors Conceiving the review: DY Designing the review: MB Screening search results: MB Organising retrieval of papers: MB/DH Screening retrieved papers against inclusion criteria: MB/EH/DH/EG Appraising quality of papers: MB/EH/DH/EG Abstracting data from papers: MB/DH/EG Writing to authors of papers for additional information: MB Obtaining and screening data on unpublished studies: MB Data management for the review: MB Entering data into RevMan: MB Analysis of data: MB/EG Interpretation of data: MB/EH/DH/EG Writing the review: MB/EH/DH/EG Sources of support Internal sources International Centre for Eye Health, London School of Hygiene and Tropical Medicine, UK. Grant R21EY023303‐01, National Eye Institute, National Institutes of Health, USA. External sources MJB is supported by the Wellcome Trust, UK. National Institute for Health Research (NIHR), UK. Richard Wormald, Co‐ordinating Editor for the Cochrane Eyes and Vision Group (CEVG) acknowledges financial support for his CEVG research sessions from the Department of Health through the award made by the National Institute for Health Research to Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology for a Specialist Biomedical Research Centre for Ophthalmology.

Group (CEVG) acknowledges financial support for his CEVG research sessions from the Department of Health through the award made by the National Institute for Health Research to Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology for a Specialist Biomedical Research Centre for Ophthalmology. The NIHR also funds the CEVG Editorial Base in London. The views expressed in this publication are those of the authors and not necessarily those of the NIHR, NHS, or the Department of Health.

Background 'Digital health' is an overarching term for the use of information and communication technology (ICT) for health purposes. The term has more recently been used as “a broad umbrella term encompassing eHealth, mHealth, as well as emerging areas such as the use of advanced computing sciences, 'big data,' genomics and artificial intelligence” (WHO 2018). The use of digital technology for health has emerged as an important innovation with the potential to strengthen health systems in many settings. This potential to address health system challenges and to improve the delivery of services has propelled significant investments into digital health, particularly in low‐ and middle‐income countries (LMICs). Governments have access to a broad range of digital health tools, but there are gaps in the evidence on the effectiveness, feasibility, and acceptability of digital health interventions (Aranda‐Jan 2014; Gurol‐Urganci 2013; Vervloet 2012).

s into digital health, particularly in low‐ and middle‐income countries (LMICs). Governments have access to a broad range of digital health tools, but there are gaps in the evidence on the effectiveness, feasibility, and acceptability of digital health interventions (Aranda‐Jan 2014; Gurol‐Urganci 2013; Vervloet 2012). Digital health interventions have shown potential for improving the efficiency and effectiveness of health service delivery and health system functioning, the latter referring to digital tools for strengthening key health systems functions such as leadership and governance, finance, human resource, and health information systems, as well as equipment and medicine supply systems. This includes a wide range of applications for electronic monitoring and evaluation, clinical support decision‐making tools, electronic diagnostics and prescribing systems, increased access to health services in remote areas, co‐ordination and knowledge exchange between different cadres and levels of health workers, electronic management and administration systems, and for improving health service responsiveness and patient‐orientated change interventions aimed at improved patient self‐care and health awareness (Catalani 2013, Naghizadeh 2017).

rdination and knowledge exchange between different cadres and levels of health workers, electronic management and administration systems, and for improving health service responsiveness and patient‐orientated change interventions aimed at improved patient self‐care and health awareness (Catalani 2013, Naghizadeh 2017). Among the most common areas for digital health interventions (perhaps due to the high prevalence of mobile phone use globally, including in low‐resource settings) is patient‐orientated change aimed at improved self‐care and self‐management of health and illness, through for instance digital health reminders of appointment and general health promotion messaging (Gurol‐Urganci 2013). Although there is increasing evidence that digital targeted client communication may improve patient adherence behaviour, less is known about the acceptability, relevance, and usefulness of these interventions from the perspective of the client population. This information is needed to inform practice and policy on optimising the design, implementation, and improvement of digital targeted client communication interventions.

nce behaviour, less is known about the acceptability, relevance, and usefulness of these interventions from the perspective of the client population. This information is needed to inform practice and policy on optimising the design, implementation, and improvement of digital targeted client communication interventions. Description of the topic of interest Within the field of digital health, there are a variety of ways digital technologies may be used for public health purposes. This review focused on digital targeted client communication (DTCC). Digital targeted client communication may be used to transmit health event alerts to specific population groups; deliver health information based on a known health status or demographic; alert and remind about a particular health behaviour; or transmit diagnostic results to clients (WHO 2018). Targeted communication can also be further customised according to an individual’s specific needs, resulting in 'tailored client communication,' whereby message content, timing, and frequency are matched to the needs and preferences of an individual (Hawkins 2008). The communication can be unidirectional and bidirectional, but initial contact is from the health system, as opposed to on‐demand information service and telemedicine, where the client initiates the first contact with the health system (WHO 2018). The purpose of the DTCC would be to improve health and well‐being, healthcare services, and/or the functioning of the health system. Typical interventions include sending brief text messages as a reminder to adhere to health visits and medical treatment, to provide clients with health information, to monitor their progress, and/or to provide medical advice and support.

and well‐being, healthcare services, and/or the functioning of the health system. Typical interventions include sending brief text messages as a reminder to adhere to health visits and medical treatment, to provide clients with health information, to monitor their progress, and/or to provide medical advice and support. Why it is important to do this review Through the World Health Assembly Resolution on Digital Health, Ministries of Health recognised that digital technologies can potentially bring value to the health system, but called for a better understanding of best practices and the promotion of evidence‐based digital health interventions and standards (WHO 2018a). This resolution also highlighted the need to ensure that “digital health solutions complement and enhance existing health service delivery models, strengthen integrated, people‐centred health services and contribute to improved population health, and health equity, including gender equality” and noted the lack of evidence on the impact of digital health in these respects (WHO May 2018).

lth solutions complement and enhance existing health service delivery models, strengthen integrated, people‐centred health services and contribute to improved population health, and health equity, including gender equality” and noted the lack of evidence on the impact of digital health in these respects (WHO May 2018). To address this need, the World Health Organization (WHO) embarked on developing evidence‐based guidelines to inform government‐led investments in digital health interventions for health system strengthening, including mechanisms to bolster access to reproductive, maternal, newborn, child, or adolescent health (RMNCAH) services. This qualitative evidence synthesis is among a series of systematic reviews informing the WHO guidelines on digital interventions for health system strengthening. The scope of this synthesis reflects the WHO’s assessment of global intervention priorities in this area. In addition to contributing to the WHO guideline, the findings of this review will be of interest more generally to programme planners and policymakers when deciding if and how to implement DTCC via mobile device in their setting. This review will complement the two WHO‐commissioned reviews that focus on the effectiveness of targeted digital communication via mobile device (Palmer Ongoing a, Palmer Ongoing).

terest more generally to programme planners and policymakers when deciding if and how to implement DTCC via mobile device in their setting. This review will complement the two WHO‐commissioned reviews that focus on the effectiveness of targeted digital communication via mobile device (Palmer Ongoing a, Palmer Ongoing). Researchers in this field have also suggested that to better understand barriers and facilitators of successful implementation of digital interventions, clients’ and healthcare providers’ perceptions of the safety of the interventions, potential harms, and adverse effects should be assessed and explored (Gurol‐Urganci 2013). Barriers may include privacy concerns (Ahmed 2017), poor access to reliable network coverage, and poor integration into existing health systems (Aranda‐Jan 2014). Perceptions that the technology empowers the user and improves communication may serve as facilitators to successful implementation of digital health interventions (Ahmed 2017). Reviewing and synthesising the qualitative evidence on perceptions and experiences of clients will not only complement the evidence emerging from the effectiveness reviews, but may also enhance our understanding of broader contextual, organisational, technical, social, and individual factors that may be shaping the development, implementation, and responses to targeted digital communication. Objectives To explore clients' perceptions and experiences of targeted digital communication via mobile devices on topics related to reproductive, maternal, newborn, child, or adolescent health (RMNCAH).

Researchers in this field have also suggested that to better understand barriers and facilitators of successful implementation of digital interventions, clients’ and healthcare providers’ perceptions of the safety of the interventions, potential harms, and adverse effects should be assessed and explored (Gurol‐Urganci 2013). Barriers may include privacy concerns (Ahmed 2017), poor access to reliable network coverage, and poor integration into existing health systems (Aranda‐Jan 2014). Perceptions that the technology empowers the user and improves communication may serve as facilitators to successful implementation of digital health interventions (Ahmed 2017). Reviewing and synthesising the qualitative evidence on perceptions and experiences of clients will not only complement the evidence emerging from the effectiveness reviews, but may also enhance our understanding of broader contextual, organisational, technical, social, and individual factors that may be shaping the development, implementation, and responses to targeted digital communication. Objectives To explore clients' perceptions and experiences of targeted digital communication via mobile devices on topics related to reproductive, maternal, newborn, child, or adolescent health (RMNCAH). Methods Criteria for considering studies for this review Topic of interest We focused on clients’ perceptions and experiences of digital targeted client communication (DTCC) via mobile devices in the areas of reproductive, maternal, newborn, child, or adolescent health (RMNCAH).

Objectives To explore clients' perceptions and experiences of targeted digital communication via mobile devices on topics related to reproductive, maternal, newborn, child, or adolescent health (RMNCAH). Methods Criteria for considering studies for this review Topic of interest We focused on clients’ perceptions and experiences of digital targeted client communication (DTCC) via mobile devices in the areas of reproductive, maternal, newborn, child, or adolescent health (RMNCAH). Types of studies We included primary studies that used qualitative study designs such as ethnography, phenomenology, case studies, and grounded theory as well as qualitative process evaluations. We included primary studies that used qualitative methods for data collection (e.g. individual interviews, focus group discussions, diaries, document analysis, open‐ended survey questions, and observation) and that used qualitative methods for data analysis (e.g. thematic analysis, framework analysis or grounded theory). We excluded primary studies that collected data using qualitative methods but did not perform a qualitative analysis (e.g. open‐ended survey questions where the responses are analysed using descriptive statistics only). We included mixed‐methods studies when it was possible to extract data that were collected and analysed using qualitative methods. We included studies regardless of whether they had been carried out alongside studies of the effectiveness of digital health interventions.

are analysed using descriptive statistics only). We included mixed‐methods studies when it was possible to extract data that were collected and analysed using qualitative methods. We included studies regardless of whether they had been carried out alongside studies of the effectiveness of digital health interventions. Types of interventions We included studies exploring clients’ experiences and perceptions of targeted digital communication (e.g. text messages and interactive voice response) accessible via mobile devices (see Table 1). This could include perceptions and experiences of the content of the message, the delivery mechanism itself, the sender, or other aspects tied to this form of communication.

and perceptions of targeted digital communication (e.g. text messages and interactive voice response) accessible via mobile devices (see Table 1). This could include perceptions and experiences of the content of the message, the delivery mechanism itself, the sender, or other aspects tied to this form of communication. We defined 'digital targeted client communication' (DTCC) as the transmission of targeted health content to a specified population, or to individuals within a predefined health or demographic group. This transmitted information can fall along a continuum of tailored (personalised to an individual person’s condition) to standard, general, untailored communication. It can include the transmission of individualised notifications according to a specific individual’s clinical care plan as well as the transmission of predetermined content developed for the identified population group (Hawkins 2008). Eligible individuals need to be identified and subscribed into a system that allows the transmission of the health information via digital device to a number they have requested. Additionally, the timing and content of the transmitted information should be determined by the health system, and not by a client seeking information on‐demand. Examples of targeted client communication could include: providing targeted health education, promotion, or information to clients based on known health or demographic characteristics;

We defined 'digital targeted client communication' (DTCC) as the transmission of targeted health content to a specified population, or to individuals within a predefined health or demographic group. This transmitted information can fall along a continuum of tailored (personalised to an individual person’s condition) to standard, general, untailored communication. It can include the transmission of individualised notifications according to a specific individual’s clinical care plan as well as the transmission of predetermined content developed for the identified population group (Hawkins 2008). Eligible individuals need to be identified and subscribed into a system that allows the transmission of the health information via digital device to a number they have requested. Additionally, the timing and content of the transmitted information should be determined by the health system, and not by a client seeking information on‐demand. Examples of targeted client communication could include: providing targeted health education, promotion, or information to clients based on known health or demographic characteristics; providing alerts, notifications, and reminders to a client based on a clinical care plan or protocol, such as in the case of medication adherence and appointments to see a healthcare provider. In contrast, untargeted client communication is the transmission of health promotion content to the general population or an undefined target population.

providing alerts, notifications, and reminders to a client based on a clinical care plan or protocol, such as in the case of medication adherence and appointments to see a healthcare provider. In contrast, untargeted client communication is the transmission of health promotion content to the general population or an undefined target population. By mobile devices, we mean mobile phones or handheld mobile devices of any kind (but not analogue landline telephones), as well as tablets, personal digital assistants, and smartphones that facilitate communication to a targeted group of clients via different channels including short message service (SMS), voice, interactive voice response, multimedia messages, and social media when used for instant messaging purposes. For a specific list of included and excluded types of delivery mechanisms see Table 1 below. Table 1: List of included and excluded mobile devices and platforms Included Excluded Mobile text messaging (including SMS and USSD) Interactive voice response (IVR) Voice calls and callbacks WhatsApp and other instant messaging services (such as Facebook Messenger) Multimedia messages, including video and audiovisual messages Applications that provide notifications to the client Communication in which the content and timing are predefined Web‐based intervention, if content development is optimised for mobile delivery or training and implementation support is based on the use of mobile devices Applications (apps) that provide targeted client communication, such as notifications to the client

Applications that provide notifications to the client Communication in which the content and timing are predefined Web‐based intervention, if content development is optimised for mobile delivery or training and implementation support is based on the use of mobile devices Applications (apps) that provide targeted client communication, such as notifications to the client Web portals, applications, and websites that do not have a targeted communication component to notify clients Emails alone that did not explicitly state transmission to mobile devices Social media websites such as Facebook, Baidu, and Twitter, unless there is explicit mention of the use of targeted communication or messaging services to individuals We included targeted client communication that aimed to remind or recall; inform and educate; or provide support (Hill 2011; Kaufman 2017; Willis 2013).We included targeted client communication that focused on the health issues identified in Appendix 1. We derived this list of health issues from two key resources by the World Health Organization on Essential Interventions for Reproductive, Maternal, Newborn, Child, and Adolescent Health (RMNCAH) and Family Planning, Safe Abortion Care, Maternal, Newborn, and Child Health (PMNCH) (Partnership for Maternal Newborn Child Health 2011).We included studies where the message was initiated by a governmental or non‐governmental, private, or public organisation and was targeted at individuals or groups.

Health (RMNCAH) and Family Planning, Safe Abortion Care, Maternal, Newborn, and Child Health (PMNCH) (Partnership for Maternal Newborn Child Health 2011).We included studies where the message was initiated by a governmental or non‐governmental, private, or public organisation and was targeted at individuals or groups. We included communication that was one‐way (e.g. triggered by a system to the defined population groups) or two‐way (e.g. allows for discussion or question and answer between the targeted population and the health system). Two‐way or bidrectional communication was included if the first communication was initiated by the health system or healthcare provider to a client's mobile device. Studies of bidirectional communication initiated by clients to contact the health system were included in another review related to telemedicine and client‐to‐provider consultations (Gonçalves‐Bradley 2018, WHO 2019). We included studies where the digital component of the intervention was delivered as part of a wider package, or if we judged it to be the major component of the intervention. The focus of the study needed to be on one of the intervention areas listed in Table 1. Types of participants The review focused on the following population groups as defined in relation to the WHO guideline for which this review was commissioned (WHO 2019).

We included studies where the digital component of the intervention was delivered as part of a wider package, or if we judged it to be the major component of the intervention. The focus of the study needed to be on one of the intervention areas listed in Table 1. Types of participants The review focused on the following population groups as defined in relation to the WHO guideline for which this review was commissioned (WHO 2019). We included studies that focused on the perceptions and experiences of clients. We define clients as “an individual who is a potential or current user of health services; may also be referred to as patient or non‐patient who uses health information and services” (WHO 2019). We included studies that focused on the perceptions and experiences of clients in one or more of the following groups. Adolescent and youth populations (ages 10 to 24 years) that were users/potential users of sexual and reproductive health (SRH) services. Studies that included other population groups were included if participants’ age had been disaggregated or where it was explicitly mentioned that a minimum of 70% of participants were between the ages of 10 and 24 years. Adult users/potential users of SRH (age 18+). Studies that explicitly stated that they also included population groups under 18 years of age were included where it was explicitly mentioned that a minimum of 70% of the participants were above the age of 18 years. Pregnant and postpartum women up to six weeks' postpartum and their partners or others who support them.

Adult users/potential users of SRH (age 18+). Studies that explicitly stated that they also included population groups under 18 years of age were included where it was explicitly mentioned that a minimum of 70% of the participants were above the age of 18 years. Pregnant and postpartum women up to six weeks' postpartum and their partners or others who support them. Pregnant and postpartum women living with HIV up to six weeks' postpartum and their partners or others who support them, with the exception of breastfeeding, for which it was six months' postpartum. Parents and caregivers of children under five years of age. Search methods for identification of studies Electronic searches Information Specialist John Eyers developed the search strategies in consultation with the review authors. We searched the following electronic databases for eligible studies between 3 and 6 July 2017, dependant on the data base (see Appendix 2). MEDLINE (OvidSP) MEDLINE In‐Process & Other Non‐Indexed Citations (OvidSP) Embase (Ovid) World Health Organization Global Health Library POPLINE

Search methods for identification of studies Electronic searches Information Specialist John Eyers developed the search strategies in consultation with the review authors. We searched the following electronic databases for eligible studies between 3 and 6 July 2017, dependant on the data base (see Appendix 2). MEDLINE (OvidSP) MEDLINE In‐Process & Other Non‐Indexed Citations (OvidSP) Embase (Ovid) World Health Organization Global Health Library POPLINE Using guidelines developed by the Cochrane Qualitative Research Methods Group for searching for qualitative evidence (Booth 2011, Harris 2018, Noyes 2015), as well as modified versions of the search for the associated or 'sister' effectiveness reviews (Palmer Ongoing a, Palmer Ongoing), we developed search strategies for each database. There were no language or geographic restrictions on the search. We used 1993 as the cut‐off date for the search, as the first commercial SMS message was sent in December 1992. A similar approach was taken in the related effectiveness reviews (Palmer Ongoing a, Palmer Ongoing). Searching other resources We asked the Guideline Development Group network for the WHO guideline on Digital Health Guidelines for Health System Strengthening to identify and send in any studies that fit the inclusion criteria on 28 June 2017 (WHO 2019).

Using guidelines developed by the Cochrane Qualitative Research Methods Group for searching for qualitative evidence (Booth 2011, Harris 2018, Noyes 2015), as well as modified versions of the search for the associated or 'sister' effectiveness reviews (Palmer Ongoing a, Palmer Ongoing), we developed search strategies for each database. There were no language or geographic restrictions on the search. We used 1993 as the cut‐off date for the search, as the first commercial SMS message was sent in December 1992. A similar approach was taken in the related effectiveness reviews (Palmer Ongoing a, Palmer Ongoing). Searching other resources We asked the Guideline Development Group network for the WHO guideline on Digital Health Guidelines for Health System Strengthening to identify and send in any studies that fit the inclusion criteria on 28 June 2017 (WHO 2019). We sent a public call for papers to global listservs, including Global Digital Health Network and Implementing Best Practices (IBP).We handsearched the database www.mHealthEvidence.org for any studies that met our inclusion criteria on 17 August 2017. This database is designed to bring together literature on digital health from a global perspective to help stakeholders quickly access up‐to‐date, relevant evidence. We searched PubMed for all studies linked to the trials included in the related effectiveness reviews in September and October 2018.

We sent a public call for papers to global listservs, including Global Digital Health Network and Implementing Best Practices (IBP).We handsearched the database www.mHealthEvidence.org for any studies that met our inclusion criteria on 17 August 2017. This database is designed to bring together literature on digital health from a global perspective to help stakeholders quickly access up‐to‐date, relevant evidence. We searched PubMed for all studies linked to the trials included in the related effectiveness reviews in September and October 2018. Data collection, management, and synthesis Selection of studies We collated records identified from different sources into Covidence, a systematic review screening tool (Covidence). We identified duplicates and removed them. Three review authors then independently assessed the titles and abstracts of the identified records to determine potential eligibility, discarding those that were clearly irrelevant to the topic. Review authors HA and EA screened all titles and abstracts, and TT resolved any conflicts. We obtained the the full text of all the papers identified as potentially relevant, and two review authors (TT, EA or HA) independently assessed these for inclusion in the review. NL resolved disagreements. See Characteristics of excluded studies for a list of the excluded studies and the main reasons for exclusion.

Data collection, management, and synthesis Selection of studies We collated records identified from different sources into Covidence, a systematic review screening tool (Covidence). We identified duplicates and removed them. Three review authors then independently assessed the titles and abstracts of the identified records to determine potential eligibility, discarding those that were clearly irrelevant to the topic. Review authors HA and EA screened all titles and abstracts, and TT resolved any conflicts. We obtained the the full text of all the papers identified as potentially relevant, and two review authors (TT, EA or HA) independently assessed these for inclusion in the review. NL resolved disagreements. See Characteristics of excluded studies for a list of the excluded studies and the main reasons for exclusion. Translation of studies in languages other than English Although we searched for languages spoken by at least one member of the review team (French, English, Scandinavian languages), all of the identified or included studies were in the English language. Sampling of studies As qualitative evidence synthesis aims for variation in concepts rather than an exhaustive sample, and because large numbers of studies can impair the quality of the analysis, we purposefully sampled from the 52 articles that met our inclusion criteria. We developed a sampling frame that took into consideration the population group, data richness, and closeness of the study data to the review objective.

Sampling of studies As qualitative evidence synthesis aims for variation in concepts rather than an exhaustive sample, and because large numbers of studies can impair the quality of the analysis, we purposefully sampled from the 52 articles that met our inclusion criteria. We developed a sampling frame that took into consideration the population group, data richness, and closeness of the study data to the review objective. Firstly, we divided the studies that met our inclusion criteria by client group as listed above in the inclusion criteria. As there were a limited number of included studies for pregnant and postpartum women (up to six weeks) (seven studies) and for pregnant and postpartum women (up to six weeks) living with HIV (two studies), all studies were included. Secondly, we assessed the included studies within each client group for data richness, using a scale of 1 to 5 (see Appendix 3), and also looked at how closely the data from the study matched the review objectives. Studies with 'thin data' or that did not provide a close match to our review objective were not sampled. In total, we sampled 35 studies to be included in the analysis (see Table 2 below). Table 2: Sampled studies included in the synthesis per client group Adolescent and youth populations as potential users of SRH services Adult populations as potential users of SRH services Pregnant and postpartum women (up to 6 weeks) Pregnant and postpartum women (up to 6 weeks) living with HIV Parents and other caregivers of children under 5 years of age 12 10 7 2 4

per client group Adolescent and youth populations as potential users of SRH services Adult populations as potential users of SRH services Pregnant and postpartum women (up to 6 weeks) Pregnant and postpartum women (up to 6 weeks) living with HIV Parents and other caregivers of children under 5 years of age 12 10 7 2 4 Data extraction We performed data extraction using a data extraction form designed specifically for this synthesis. We used the form to extract key themes and categories relevant to the synthesis objective using the Supporting the Use of Research Evidence (SURE) framework (SURE Collaboration 2011). We used a second form to extract information about first author, date of publication, language, country of study, context (urban, rural), and participant group to which the intervention was directed. We also extracted information on research method and if theoretical or conceptual frameworks were used. HA extracted data from all the sampled studies. EA double‐checked the data extraction and verified that all relevant data were extracted. Assessment of the methodological limitations of included studies To assess the methodological quality of the included studies, we applied a quality appraisal framework to each study. We used an adaptation of the Critical Appraisal Skills Programme (CASP) quality assessment tool for qualitative studies (CASP 2018). Other reviews of qualitative evidence have also used this tool (Ames 2017; Glenton 2013; Gopinathan 2014; Lewin 2010). The adapted tool that we used included the following eight questions.

e used an adaptation of the Critical Appraisal Skills Programme (CASP) quality assessment tool for qualitative studies (CASP 2018). Other reviews of qualitative evidence have also used this tool (Ames 2017; Glenton 2013; Gopinathan 2014; Lewin 2010). The adapted tool that we used included the following eight questions. Are the setting/s and context described adequately? Is the sampling strategy described, and is this appropriate? Is the data collection strategy described and justified? Is the data analysis described, and is this appropriate? Are the claims made/findings supported by sufficient evidence? Is there evidence of reflexivity? Does the study demonstrate sensitivity to ethical concerns? Any other concerns? HA conducted the initial assessment, and NL and TT reviewed the assessments. We accept that there is no ‘gold standard’ approach for assessing the methodological quality of primary qualitative studies, but believe that this adapted CASP checklist fit our needs in the context of this synthesis. We did not exclude any studies based on our assessment of methodological limitations, but used this information to assess our confidence in the synthesis findings, as part of the GRADE‐CERQual (Confidence in the Evidence from Reviews of Qualitative research) approach (Lewin 2018). Data management and synthesis For our synthesis, we first grouped articles according to client group as defined above. However, since there were only two articles focusing on pregnant and postpartum women living with HIV, we combined these with the other studies focusing on pregnant and postpartum women.

We did not exclude any studies based on our assessment of methodological limitations, but used this information to assess our confidence in the synthesis findings, as part of the GRADE‐CERQual (Confidence in the Evidence from Reviews of Qualitative research) approach (Lewin 2018). Data management and synthesis For our synthesis, we first grouped articles according to client group as defined above. However, since there were only two articles focusing on pregnant and postpartum women living with HIV, we combined these with the other studies focusing on pregnant and postpartum women. We conducted an initial framework analysis using the SURE framework to identify themes in the data. We did this within each of the population groups and then looked across population groups. The SURE framework has been used as an analysis framework in other studies and reviews (Glenton 2013; Glenton 2016; Gopinathan 2014; Lewin 2010; Muloliwa Forthcoming; Oku 2017). We used the headings and subheadings from the SURE framework as a starting point for the analysis and then adapted them through an iterative process. Next, within each section of the framework we did a thematic analysis of the extracted data to identify our synthesis findings. For example, data included in the framework under 'health systems constraints ‐ accessibility of care' were thematically synthesised, and findings around access to digital devices and interventions were identified (see Findings 5 to 9 in the Results). Another example is around the framework area of knowledge and skills. Here we adapted the category to look at knowledge and skills in relationship to using a mobile device. A final example of adaptation is under health systems constraints, relationships with norms and standards. We adapted this category to address issues related to privacy and confidentiality. Some areas of the framework were left empty and discarded. Once findings were identified, HA read through all of the sampled studies again to double‐check data extraction. We also went through the findings and identified those where the contributing studies were only/predominantly from high‐income (HIC) or LMIC settings. The same was done for the different client groups. We have indicated this in the detailed description of the relevant findings.

es again to double‐check data extraction. We also went through the findings and identified those where the contributing studies were only/predominantly from high‐income (HIC) or LMIC settings. The same was done for the different client groups. We have indicated this in the detailed description of the relevant findings. We then thematically analysed the 25 identified findings in order to group them into six related overarching categories to provide a narrative for the Findings section. Some categories reflect those within the SURE framework, whereas others have been reorganised to address different issues raised by clients. The six overarching categories related to the general acceptability of and preferences around DTCC; the varying degrees of access to network services, phones, and messages; communication delivery and format preferences; communication content preferences; privacy and confidentiality regarding personal health information; and the perceptions of intervention impact. To create the summary of findings for the Abstract we took all of the findings with moderate or high confidence in the evidence and worked them together into a clear story line. Appraisal of confidence in the review findings Four review authors (HA, CG, SL, NL) used GRADE‐CERQual to assess the confidence that can be placed in each review finding (Lewin 2018). Each finding was assessed by at least two review authors. The GRADE‐CERQual approach assesses confidence in the evidence based on the following four components (Lewin 2018a).

Appraisal of confidence in the review findings Four review authors (HA, CG, SL, NL) used GRADE‐CERQual to assess the confidence that can be placed in each review finding (Lewin 2018). Each finding was assessed by at least two review authors. The GRADE‐CERQual approach assesses confidence in the evidence based on the following four components (Lewin 2018a). Methodological limitations of included studies: the extent to which there are concerns about the design or conduct of the primary studies that contributed evidence to an individual review finding. Coherence of the review finding: an assessment of how clear and cogent the fit is between the data from the primary studies and a review finding that synthesises that data. Adequacy of the data contributing to a review finding: an overall determination of the degree of richness and quantity of data supporting a review finding. Relevance of the included studies to the review question: the extent to which the body of evidence from the primary studies supporting a review finding is applicable to the phenomenon of interest (perspective or population, context, setting) specified in the review question. After assessing each of the four components, we made a judgement about the overall confidence in the evidence supporting each review finding. We judged confidence as high, moderate, low, or very low. The final assessment was based on consensus among the review authors. All findings started as high confidence and were then graded down if there were important concerns regarding any of the CERQual components (Lewin 2018).

vidence supporting each review finding. We judged confidence as high, moderate, low, or very low. The final assessment was based on consensus among the review authors. All findings started as high confidence and were then graded down if there were important concerns regarding any of the CERQual components (Lewin 2018). 'Summary of qualitative findings' tables We concluded the appraisal of confidence in each review finding by drafting a 'Summary of qualitative findings' table that presents the findings and our assessment of confidence in these findings, as well as an explanation of this assessment, based on the GRADE‐CERQual approach. Supplementing the related Cochrane effectiveness reviews with synthesised qualitative findings We explored how the findings from our synthesis related to, and could help to inform, the findings of the two related Cochrane reviews of effectiveness of DTCC (Palmer Ongoing a; Palmer Ongoing). To do this we utilised a matrix approach similar to the one used previously by Candy 2011, Ames 2017, and Munabi‐Babigumira 2017. This approach has also been described by Harden 2018. Our matrix explored whether potential implementation barriers that we identified in our synthesis had been addressed in the programmes evaluated in the related reviews of effectiveness.

to the one used previously by Candy 2011, Ames 2017, and Munabi‐Babigumira 2017. This approach has also been described by Harden 2018. Our matrix explored whether potential implementation barriers that we identified in our synthesis had been addressed in the programmes evaluated in the related reviews of effectiveness. To create the matrix we undertook the following steps: firstly, we selected the synthesis findings that we had assessed as having high or moderate confidence and that presented potential barriers to the implementation of targeted client communication programmes. Secondly, we created 10 questions reflecting these potential barriers, and placed these in a table. Finally, we assessed whether any attempt had been made to address these implementation barriers in the trials that were included in the two related Cochrane Reviews of effectiveness. To carry out this assessment, we examined the publications included in the two Cochrane Reviews of effectiveness (Palmer Ongoing a; Palmer Ongoing). We also performed a further search for additional publications that could be related to the trials. We did this by (1) examining the reference lists of the main trial publication; and (2) searching for each trial in PubMed, and doing an advanced search for 'Similar articles'. The advanced search for ‘Similar articles’ used the first author of the trial to identify possible related studies that had this author as a co‐author, and selected any that appeared to be related to the trial.

publication; and (2) searching for each trial in PubMed, and doing an advanced search for 'Similar articles'. The advanced search for ‘Similar articles’ used the first author of the trial to identify possible related studies that had this author as a co‐author, and selected any that appeared to be related to the trial. Researchers’ reflexivity Within qualitative research, researchers are expected to reflect on their own background and position, and how it will affect the design, analysis, and reporting of their research. Throughout the data synthesis, the authors were aware of their own positions and reflected on how these could influence the data synthesis and study design. Several of the authors have both primary and evidence synthesis research experience in digital health (reporting positive, negative, and neutral findings), and they considered themselves to be agnostic as to the outcome of this evidence synthesis. Results Included studies We screened 9531 abstracts and assessed 142 full‐text articles. Fifty‐two studies met our inclusion criteria. From these 52 studies, we sampled 35 studies for analysis (Figure 1). The 17 studies that met the inclusion criteria but were not sampled into the synthesis can be found in Table 1.Figure 1 PRISMA flow diagram. Table 1 Studies that were included but not sampled

Results Included studies We screened 9531 abstracts and assessed 142 full‐text articles. Fifty‐two studies met our inclusion criteria. From these 52 studies, we sampled 35 studies for analysis (Figure 1). The 17 studies that met the inclusion criteria but were not sampled into the synthesis can be found in Table 1.Figure 1 PRISMA flow diagram. Table 1 Studies that were included but not sampled Study ID Reason not sampled Adetunji 2017 Thin data Ahlers‐Schmidt 2012 Thin data Ahlers‐Schmidt 2013 Thin data Ahlers‐Schmidt 2014 Thin data Anand 2017 Thin data Atukunda 2017 Thin data Baranoski 2014 Thin data Cornelius 2012 Formative study not as close to the research objective as other included studies Datta 2014 Thin data George 2012 Thin data in comparison to other studies from the same setting Graham 2015 Thin data Harris 2013 Thin data Martin 2016 Thin data Montoya 2014 Thin data Montoya 2015 Thin data Smillie 2014 (Kenya) Thin data Swendeman 2015 Thin data All of the sampled studies were published between 2009 and 2017. All of the included studies were published in English. Sixteen of the sampled studies were from high‐income countries: Australia (1), Canada (2), the UK (4), and the USA (9). Nineteen of the sampled studies were from low‐ or middle‐income countries: Cambodia (1), Cameroon (2), Ghana (1), India (2), Kenya (2), Lesotho (1), Nigeria (2), Peru (3), Sierra Leone (1), South Africa (2), and Uganda (2).

studies were from high‐income countries: Australia (1), Canada (2), the UK (4), and the USA (9). Nineteen of the sampled studies were from low‐ or middle‐income countries: Cambodia (1), Cameroon (2), Ghana (1), India (2), Kenya (2), Lesotho (1), Nigeria (2), Peru (3), Sierra Leone (1), South Africa (2), and Uganda (2). Client populations were adolescents and youth (12 studies); adult users/potential users of reproductive health services (10); pregnant and postpartum women (including those living with HIV) (9); and parents and caregivers of children under five years of age (4). The included studies explored seven different methods or combinations of methods for delivering DTCC: app (2), interactive voice response (IVR) (1), IVR + SMS (1), SMS + voice call (1), SMS (27), mobile phone messaging (2), and mobile phones in general (1). Methodological limitations of the included qualitative studies There was poor reporting of the participant voice in some of the included studies. For example, many studies included limited first‐order constructs or data extracts, and these were often not labelled with an identifier of the participant. We also found poor reporting of researcher reflexivity across many of the studies, which limited transparency regarding the role of the researcher. All studies gave some description, even if very brief, of the context, participants, sampling, methods, and analysis.

en not labelled with an identifier of the participant. We also found poor reporting of researcher reflexivity across many of the studies, which limited transparency regarding the role of the researcher. All studies gave some description, even if very brief, of the context, participants, sampling, methods, and analysis. Confidence in the findings Based on our CERQual assessments, we had high confidence in four findings and moderate confidence in nine findings, indicating that the studies were a good representation of the phenomenon of interest. We had several findings where we had low (nine) or very low confidence (three), indicating that the studies were a weaker fit with the representation of the phenomenon of interest. Our main concerns were connected to the methodological limitations of the studies and the relevance and adequacy of the data. Common methodological limitations included a lack of researcher reflexivity as well as poor reporting of ethical considerations, sampling, and representation of the participant’s voice in the findings. The data were often assessed as being only partially relevant, mainly because the included studies represented few regions; had a focus on a certain target population (e.g. youth) or a specific topic (HIV/AIDS); or because many of the included studies explored participants’ perceptions of hypothetical situations or digital health interventions, or both. Finally, our concerns about adequacy were mainly tied to the limited number of studies included in some findings and the thinness of the data contributing to some findings.

S); or because many of the included studies explored participants’ perceptions of hypothetical situations or digital health interventions, or both. Finally, our concerns about adequacy were mainly tied to the limited number of studies included in some findings and the thinness of the data contributing to some findings. The GRADE‐CERQual evidence profile tables supporting the assessment of confidence in each finding can be found in Appendix 4. We start each section of the findings with a link to the 'CERQual summary of qualitative findings' table where a summary assessment of the findings from that section is presented. Findings and categories identified in the data From our synthesis, we developed a set of individual findings, and then organised these findings into six overarching categories related to (1) the general acceptability of and preferences around digital health interventions; (2) the varying degrees of access to network services, phones, and messages; (3) communication delivery and format preferences; (4) communication content preferences; (5) privacy and confidentiality regarding personal health information; and (6) perceptions of intervention impact. Unless specifically addressed in the detailed finding, the data were not specific to HIC or LMIC setting or to a specific client group. For a description of the context and client group in each study contributing to a finding, please refer to the evidence profiles in Appendix 4.

(6) perceptions of intervention impact. Unless specifically addressed in the detailed finding, the data were not specific to HIC or LMIC setting or to a specific client group. For a description of the context and client group in each study contributing to a finding, please refer to the evidence profiles in Appendix 4. General acceptability of and preferences around digital targeted client communication ('Summary of qualitative findings' table for Findings 1 to 4 is shown in Table 2.)Table 2 'Summary of qualitative findings' table for findings related to general acceptability of and preferences around digital health interventions

(6) perceptions of intervention impact. Unless specifically addressed in the detailed finding, the data were not specific to HIC or LMIC setting or to a specific client group. For a description of the context and client group in each study contributing to a finding, please refer to the evidence profiles in Appendix 4. General acceptability of and preferences around digital targeted client communication ('Summary of qualitative findings' table for Findings 1 to 4 is shown in Table 2.)Table 2 'Summary of qualitative findings' table for findings related to general acceptability of and preferences around digital health interventions Finding Overall CERQual assessment Explanation for assessment Contributing studies 1 Overall, participants had a range of views regarding acceptance of the idea of receiving health information through their mobile devices. This was due to factors such as familiarity with the technology, convenience, control, being able to save and re‐read messages later, cost, seeing it as a simple way of providing a reminder for medication or appointments, and the sense that someone was thinking about them and cared enough to send a message Low confidence Due to moderate concerns regarding methodological limitations and relevance Akinfaderin‐Agarau 2012; Brown 2014; Calderón 2017; Cates 2015; Cornelius 2009; Curioso 2009; Evans 2016; French 2016; Gold 2010; Greaney 2014; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Jennings 2013; Lau 2014; Mbuagbaw 2012; Mbuagbaw 2014; Menacho 2013; Missal 2016; Munro 2017; Naughton 2013; Odeny 2014; Perry 2012; Rana 2015; Rodrigues 2015; Sloan 2017; Smillie 2014; Smith 2017; Willoughby 2017; Wright 2011 2 In discussing the pros and cons of digital targeted client communication compared to in‐person meetings with a healthcare provider, some participants perceived interacting with a healthcare provider as preferable, warmer, and something to which they were accustomed. Others also felt that people could receive a faster response using digital communication and that the messages were more convenient and less judgemental. However, some liked having direct access to both healthcare providers and digital targeted client communication Very low confidence Due to minor concerns regarding methodological limitations and serious concerns regarding adequacy and relevance Calderón 2017; Nachega 2016; Naughton 2013; Sloan 2017; Smillie 2014 3 Participants said that they liked two‐way digital communication, as this allowed them to engage directly with a healthcare provider, which they trusted more; to receive answers to their questions and have opportunities for discussion; and to receive a more immediate response.

; Naughton 2013; Sloan 2017; Smillie 2014 3 Participants said that they liked two‐way digital communication, as this allowed them to engage directly with a healthcare provider, which they trusted more; to receive answers to their questions and have opportunities for discussion; and to receive a more immediate response. However, some participants felt that for some topics they would feel uncomfortable talking to a healthcare provider through a digital channel due to issues related to shyness and privacy, and would prefer to use SMS Very low confidence Due to moderate concerns regarding methodological limitations and adequacy and serious concerns regarding relevance Akinfaderin‐Agarau 2012; Calderón 2017; Cates 2015; Jennings 2013; Rana 2015; Rodrigues 2015; Smillie 2014; Smith 2017; Willoughby 2017 4 Some participants expressed a concern that some people might view digital targeted communication from healthcare providers as a replacement for seeking appropriate medical assistance, which might have adverse impacts. While some saw digital health as a way to increase access to care, others noted that text messaging might be seen by poorer people as a cheaper or sufficient healthcare option, which might decrease appropriate health‐seeking behaviour Very low confidence Due to serious concerns regarding relevance and adequacy Willoughby 2017

ome saw digital health as a way to increase access to care, others noted that text messaging might be seen by poorer people as a cheaper or sufficient healthcare option, which might decrease appropriate health‐seeking behaviour Very low confidence Due to serious concerns regarding relevance and adequacy Willoughby 2017 Finding 1: Overall, participants had a range of views regarding acceptance of the idea of receiving health information through their mobile devices. This was due to factors such as familiarity with the technology, convenience, control, being able to save and re‐read messages later, cost, seeing it as a simple way of providing a reminder for medication or appointments, and the sense that someone was thinking about them and cared enough to send a message (low confidence in the evidence).

rs such as familiarity with the technology, convenience, control, being able to save and re‐read messages later, cost, seeing it as a simple way of providing a reminder for medication or appointments, and the sense that someone was thinking about them and cared enough to send a message (low confidence in the evidence). Many studies from a variety of contexts and client groups presented data related to the range of participants’ views regarding the acceptance of DTCC and the factors that influenced this acceptance (Akinfaderin‐Agarau 2012; Brown 2014; Calderón 2017; Cates 2015; Cornelius 2009; Curioso 2009; Evans 2016; French 2016; Gold 2010; Greaney 2014; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Jennings 2013;Lau 2014; Mbuagbaw 2012; Mbuagbaw 2014; Menacho 2013; Missal 2016; Munro 2017; Naughton 2013; Odeny 2014; Perry 2012; Rana 2015; Rodrigues 2015; Sloan 2017; Smillie 2014; Smith 2017; Willoughby 2017; Wright 2011). Many participants had not used mobile devices to access health information previously, but were open to and interested in the idea of digital health interventions being used to deliver up‐to‐date knowledge and information or reminders for appointments or medication (Akinfaderin‐Agarau 2012; Cates 2015; Evans 2016; Gold 2010; Greaney 2014; Hirsch‐Moverman 2017; Jennings 2013; Lau 2014; Mbuagbaw 2014; Odeny 2014). Some participants felt that it was more personal than other methods of delivering health information such as posters and billboards (Gold 2010). Others perceived it as a way of boosting already‐existing interventions or curricula, such as school‐based HIV curricula (Cornelius 2009).

au 2014; Mbuagbaw 2014; Odeny 2014). Some participants felt that it was more personal than other methods of delivering health information such as posters and billboards (Gold 2010). Others perceived it as a way of boosting already‐existing interventions or curricula, such as school‐based HIV curricula (Cornelius 2009). However, some participants felt that DTCC delivered via mobile device were not acceptable for their setting and chose not to participate in the interventions. An example of this was seen in a family planning intervention in Sierra Leone where some husbands had a problem with their partner being called by the healthcare provider or their partner participating in the family planning intervention, or both (Jalloh‐Vos 2014). Others felt that in some situations it was important to still have the opportunity to speak to a person on the phone or in person (face‐to‐face), for example if they were having a strong craving for a cigarette and needed support in that moment (Naughton 2013). Some participants felt that DTCC would be useful for certain population groups such as younger audiences (French 2016; Willoughby 2017; Wright 2011), youth just starting on HIV medication, patients not adhering to medication, and those with less education (Smith 2017).

and needed support in that moment (Naughton 2013). Some participants felt that DTCC would be useful for certain population groups such as younger audiences (French 2016; Willoughby 2017; Wright 2011), youth just starting on HIV medication, patients not adhering to medication, and those with less education (Smith 2017). Familiarity with technology, especially SMS, was one reason that participants put forward for accepting digital health interventions (Brown 2014; Cornelius 2009; French 2016; Jennings 2013; Smillie 2014). Youth and pregnant women described using digital technology as something they already did frequently, that fit their learning styles, and was not a foreign approach (Brown 2014; Cornelius 2009; French 2016; Munro 2017). Others mentioned that they were already using SMS to request healthcare providers to call them back or to set up health appointments (Jennings 2013).

technology as something they already did frequently, that fit their learning styles, and was not a foreign approach (Brown 2014; Cornelius 2009; French 2016; Munro 2017). Others mentioned that they were already using SMS to request healthcare providers to call them back or to set up health appointments (Jennings 2013). Participants also noted that messages delivered through digital mechanisms were very convenient and in some cases cost‐effective. Some felt that they would be very beneficial for families who lived a long distance from the health facility and would save them time and money (Calderón 2017). Messages were perceived as being quick and much easier than going to the health facility to get a pamphlet, going to a doctor’s appointment (Calderón 2017; Jennings 2013; Smillie 2014; Smith 2017), or searching for information online or in books (Lau 2014). Messages were perceived as easily accessible, providing immediate guidance or support, and not taking up much time or attention (French 2016; Munro 2017; Naughton 2013; Perry 2012; Sloan 2017). Clients also liked the fact that the messages were often free. Challenges to convenience included the cost of receiving messages and phone calls and the difficulty of maintaining privacy and confidentiality in some settings when discussing sensitive information via call or SMS (Jennings 2013; Perry 2012).

Sloan 2017). Clients also liked the fact that the messages were often free. Challenges to convenience included the cost of receiving messages and phone calls and the difficulty of maintaining privacy and confidentiality in some settings when discussing sensitive information via call or SMS (Jennings 2013; Perry 2012). Participants liked that they could save and re‐read messages as well as have control over receiving, keeping, or deleting the information (Brown 2014; Evans 2016; French 2016; Munro 2017). Others felt that the intervention gave them some control over their own care and health information (Jennings 2013; Munro 2017). Some participants felt that DTCC delivered via a mobile device was an acceptable way of providing reminders for medication taking or appointments (Curioso 2009; Lau 2014; Mbuagbaw 2012; Mbuagbaw 2014; Rana 2015; Rodrigues 2015). However, others felt that these reminders were not necessary and could be detrimental to patient independence. Some participants only wanted them sent to patient groups who needed help adhering to medication or when they were preoccupied or fatigued (Hirsch‐Moverman 2017; Mbuagbaw 2014; Rana 2015; Rodrigues 2015). Finally, some participants liked DTCC that delivered messages to their mobile phones, experiencing it as supportive and making them feel that someone was thinking of and cared about them (Greaney 2014; Lau 2014; Munro 2017; Naughton 2013; Rana 2015; Sloan 2017).

Some participants felt that DTCC delivered via a mobile device was an acceptable way of providing reminders for medication taking or appointments (Curioso 2009; Lau 2014; Mbuagbaw 2012; Mbuagbaw 2014; Rana 2015; Rodrigues 2015). However, others felt that these reminders were not necessary and could be detrimental to patient independence. Some participants only wanted them sent to patient groups who needed help adhering to medication or when they were preoccupied or fatigued (Hirsch‐Moverman 2017; Mbuagbaw 2014; Rana 2015; Rodrigues 2015). Finally, some participants liked DTCC that delivered messages to their mobile phones, experiencing it as supportive and making them feel that someone was thinking of and cared about them (Greaney 2014; Lau 2014; Munro 2017; Naughton 2013; Rana 2015; Sloan 2017). Finding 2: In discussing the pros and cons of DTCC compared to in‐person meetings with a healthcare provider, some participants perceived interacting with a healthcare provider as preferable, warmer, and something to which they were accustomed. Others also felt that people could receive a faster response using digital communication and that the messages were more convenient and less judgemental. However, some liked having direct access to both healthcare providers and DTCC (very low confidence).

preferable, warmer, and something to which they were accustomed. Others also felt that people could receive a faster response using digital communication and that the messages were more convenient and less judgemental. However, some liked having direct access to both healthcare providers and DTCC (very low confidence). A few studies, from both LMIC and HIC contexts (Calderón 2017; Nachega 2016; Naughton 2013; Sloan 2017; Smillie 2014), described a range of participants’ preferences for digital health interventions compared to in‐person visits to healthcare providers. The majority of studies in this finding looked at the perspectives of pregnant and postpartum women and parents. Some clients liked having direct access to both healthcare providers and to digital health interventions, as each played a different role (Naughton 2013; Sloan 2017; Smillie 2014). Some felt that the digital health interventions were more convenient, reliable, flexible, and faster and provided more frequent support (Nachega 2016; Naughton 2013; Sloan 2017; Smillie 2014). Clients who were pregnant and trying to quit smoking often preferred the SMS interventions, as they felt healthcare providers judged them and made them feel uncomfortable (Sloan 2017). Clients in some studies liked the digital health interventions but still felt it was important to have access to in‐person visits with healthcare providers or speaking with someone when needed (Calderón 2017; Naughton 2013; Smillie 2014).

t healthcare providers judged them and made them feel uncomfortable (Sloan 2017). Clients in some studies liked the digital health interventions but still felt it was important to have access to in‐person visits with healthcare providers or speaking with someone when needed (Calderón 2017; Naughton 2013; Smillie 2014). Finding 3: Participants said that they liked two‐way digital communication as this allowed them to engage directly with a healthcare provider, which they trusted more; to receive answers to their questions and have opportunities for discussion; and to receive a more immediate response. However, some participants felt that for some topics they would feel uncomfortable talking to a healthcare provider through a digital channel, due to issues related to shyness and privacy, and would prefer to use SMS (very low confidence).

and have opportunities for discussion; and to receive a more immediate response. However, some participants felt that for some topics they would feel uncomfortable talking to a healthcare provider through a digital channel, due to issues related to shyness and privacy, and would prefer to use SMS (very low confidence). Some studies from both LMIC and HIC contexts found that participants wanted or liked to have the option of engaging directly with healthcare providers through DTCC in order to receive answers to their questions (Akinfaderin‐Agarau 2012; Calderón 2017; Cates 2015; Jennings 2013; Rana 2015; Rodrigues 2015; Smillie 2014; Smith 2017; Willoughby 2017). In general, participants felt that these types of two‐way communication options would be useful and provide them with answers to their questions when they needed them, as well as allow them to maintain contact with their healthcare providers in between appointments if questions or concerns were to arise (Rana 2015). One participant in one study stated that two‐way communication would not be acceptable, as the person on the other end would then be informed of his HIV status (Rodrigues 2015).

as allow them to maintain contact with their healthcare providers in between appointments if questions or concerns were to arise (Rana 2015). One participant in one study stated that two‐way communication would not be acceptable, as the person on the other end would then be informed of his HIV status (Rodrigues 2015). Some participants preferred voice calls for engaging with healthcare providers above communicating with them through SMS (Akinfaderin‐Agarau 2012). There were different reasons for this. Some participants felt that they could ask detailed questions and receive detailed answers as well as discuss the various problems or challenges they were facing (Akinfaderin‐Agarau 2012; Rodrigues 2015), or that the service could be more trusted because they spoke to someone directly (Akinfaderin‐Agarau 2012). Others thought that it would provide more opportunity for discussion and ensure that the message was well received by the intended recipient, and that they could receive an immediate response (Jennings 2013). However, some participants explained that they preferred SMS services, as they would allow the participants to be more open and to ask about issues they would be too shy to bring up when speaking directly with someone (Akinfaderin‐Agarau 2012; Smillie 2014). Some participants felt that they would not feel comfortable talking to someone in person, especially if that person was new or unknown to them (Smillie 2014). SMS was also viewed as advantageous for brief and relatively confidential receipt of information (Jennings 2013).

omeone (Akinfaderin‐Agarau 2012; Smillie 2014). Some participants felt that they would not feel comfortable talking to someone in person, especially if that person was new or unknown to them (Smillie 2014). SMS was also viewed as advantageous for brief and relatively confidential receipt of information (Jennings 2013). Finding 4: Some participants expressed a concern that some people might view digital targeted communication from healthcare providers as a replacement to seeking appropriate medical assistance, which might have adverse impacts. While some saw digital health as a way to increase access to care, others noted that text messaging might be seen by poorer people as a cheaper or sufficient healthcare option, which might decrease appropriate health‐seeking behaviour (very low confidence). Participants in one study from the USA exploring college students' views on receiving SMS for sexual health promotion expressed concern that other people might become over‐reliant on digital health interventions because they were seen as a cheaper option than going to the doctor (Willoughby 2017). There was a worry that people would use the digital communication intervention instead of seeking appropriate medical attention. The participants thought that this could especially be the case for people with few resources.

rventions because they were seen as a cheaper option than going to the doctor (Willoughby 2017). There was a worry that people would use the digital communication intervention instead of seeking appropriate medical attention. The participants thought that this could especially be the case for people with few resources. Varying degrees of access to network services, phones, and messages ('Summary of qualitative findings' table for Findings 5 to 9 is shown in Table 3.)Table 3 'Summary of qualitative findings' table for findings related to the varying degrees of access to network services, phones, and messages

rventions because they were seen as a cheaper option than going to the doctor (Willoughby 2017). There was a worry that people would use the digital communication intervention instead of seeking appropriate medical attention. The participants thought that this could especially be the case for people with few resources. Varying degrees of access to network services, phones, and messages ('Summary of qualitative findings' table for Findings 5 to 9 is shown in Table 3.)Table 3 'Summary of qualitative findings' table for findings related to the varying degrees of access to network services, phones, and messages Finding Overall CERQual assessment Explanation for assessment Contributing studies 5 Participants reported varying degrees of access to network services, including cell networks (for calls and SMS) and internet. In addition, some had poor access to electricity to charge their phones. These factors were reported to be barriers to using the digital targeted client communication. High confidence Due to minor concerns regarding methodological limitations Akinfaderin‐Agarau 2012; Cornelius 2009; Flax 2017; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Mbuagbaw 2012; Mbuagbaw 2014; Smillie 2014 6 Participants reported varying degrees of access to mobile devices. For instance, some had no phone; some had lost or broken their phone; some could not afford to purchase airtime; some had changed their number or sim card; or for some access to the phone was controlled by another person. These factors were reported to be barriers to using the digital targeted client communication. Moderate confidence Due to minor concerns regarding methodological limitations and relevance Akinfaderin‐Agarau 2012; Entsieh 2015; Flax 2017; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Jennings 2013; Menacho 2013; Missal 2016; Rana 2015; Smillie 2014 7 Some participants, particularly women and adolescents, had their access to phones controlled or restricted by others, especially if they had to share or borrow a phone. They noted that they would often have to explain why they wanted to use the phone, and who they wanted to call, to allay suspicions about this communication. They mentioned that this was a barrier to accessing digital targeted client communication and made it difficult to keep their messages private. Moderate confidence Due to minor concerns regarding methodological limitations, coherence, adequacy, and relevance Akinfaderin‐Agarau 2012; Flax 2017; Jalloh‐Vos 2014; Rana 2015 8 Participants believed that the cost of participating in digital targeted client communication should be free or very low, as cost could present a barrier to participation, particularly for young people and those on lower incomes.

, adequacy, and relevance Akinfaderin‐Agarau 2012; Flax 2017; Jalloh‐Vos 2014; Rana 2015 8 Participants believed that the cost of participating in digital targeted client communication should be free or very low, as cost could present a barrier to participation, particularly for young people and those on lower incomes. Participants felt that there should be little or no charge for costs such as joining the digital health intervention, downloading applications (apps), or for sending and receiving mobile messages/phone calls. High confidence Due to minor concerns regarding relevance Akinfaderin‐Agarau 2012; Calderón 2017; Cornelius 2009; Menacho 2013; Mitchell 2016; Perry 2012; Rana 2015; Smith 2017 9 Participants’ ability to access digital communication was sometimes limited by their language skills and their personal level of literacy or techno‐literacy, or both. Moderate confidence Due to minor concerns regarding relevance and moderate concerns regarding methodological limitations Akinfaderin‐Agarau 2012; Calderón 2017; Curioso 2009; Greaney 2014; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Mbuagbaw 2014; Rodrigues 2015; Smillie 2014 Finding 5: Participants reported varying degrees of access to network services, including cell networks (for calls and SMS) and the internet. In addition, some participants had poor access to electricity to charge their phones. These factors were reported to be barriers to using the DTCC (high confidence).

Participants felt that there should be little or no charge for costs such as joining the digital health intervention, downloading applications (apps), or for sending and receiving mobile messages/phone calls. High confidence Due to minor concerns regarding relevance Akinfaderin‐Agarau 2012; Calderón 2017; Cornelius 2009; Menacho 2013; Mitchell 2016; Perry 2012; Rana 2015; Smith 2017 9 Participants’ ability to access digital communication was sometimes limited by their language skills and their personal level of literacy or techno‐literacy, or both. Moderate confidence Due to minor concerns regarding relevance and moderate concerns regarding methodological limitations Akinfaderin‐Agarau 2012; Calderón 2017; Curioso 2009; Greaney 2014; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Mbuagbaw 2014; Rodrigues 2015; Smillie 2014 Finding 5: Participants reported varying degrees of access to network services, including cell networks (for calls and SMS) and the internet. In addition, some participants had poor access to electricity to charge their phones. These factors were reported to be barriers to using the DTCC (high confidence). Studies from a range of income settings found that issues related to network services and electricity acted as a barrier to people’s use of DTCC (Akinfaderin‐Agarau 2012; Cornelius 2009; Flax 2017; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Mbuagbaw 2012; Mbuagbaw 2014; Smillie 2014). Lack of network or internet coverage meant that some participants could not participate in the intervention or did not receive some of the messages (Akinfaderin‐Agarau 2012; Cornelius 2009; Flax 2017; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Mbuagbaw 2012; Mbuagbaw 2014; Smillie 2014). For instance, when network coverage was poor, some participants in a Nigerian study recommended that SMS was the best option as they were more likely to be transmitted when the network was unstable, whereas voice calls would not connect or would be dropped (Akinfaderin‐Agarau 2012). Participants in a study from Canada described living in mountainous areas with no network coverage (Smillie 2014). Participants in low‐income settings also mentioned that not being able to charge their phone due to power outages or lack of access to electricity was a barrier to participating in digital health interventions (Akinfaderin‐Agarau 2012; Hirsch‐Moverman 2017; Mbuagbaw 2014).

areas with no network coverage (Smillie 2014). Participants in low‐income settings also mentioned that not being able to charge their phone due to power outages or lack of access to electricity was a barrier to participating in digital health interventions (Akinfaderin‐Agarau 2012; Hirsch‐Moverman 2017; Mbuagbaw 2014). Finding 6: Participants reported varying degrees of access to mobile devices. For instance, some participants had no phone; some had lost or broken their phone; some could not afford to purchase airtime; some had changed their number or sim card; or for some access to the phone was controlled by another person. These factors were reported to be barriers to using the DTCC (moderate confidence). Some studies, the majority from LMIC settings in Africa (Akinfaderin‐Agarau 2012; Entsieh 2015; Flax 2017; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Jennings 2013; Menacho 2013; Missal 2016; Rana 2015; Smillie 2014), found that access to functioning mobile phones was a barrier to participants’ use of DTCC. Some participants reported not owning a phone (Akinfaderin‐Agarau 2012; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Rana 2015), and others had lost or broken their phone (Flax 2017; Smillie 2014). For some, not owning a phone caused feelings of jealousy and unhappiness and forced them to borrow a phone if they wanted to participate in the digital health intervention (Jalloh‐Vos 2014).

rau 2012; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Rana 2015), and others had lost or broken their phone (Flax 2017; Smillie 2014). For some, not owning a phone caused feelings of jealousy and unhappiness and forced them to borrow a phone if they wanted to participate in the digital health intervention (Jalloh‐Vos 2014). Cost was also a barrier to participation for some participants, as they could not afford the airtime or credit needed to receive or send SMS or phone calls (Jalloh‐Vos 2014; Jennings 2013; Smillie 2014), although in some studies, participants received free airtime, which removed this access barrier (Hirsch‐Moverman 2017). In other cases, participants changed residence, changed sim cards, or had multiple sim cards and were no longer able to be reached by the DTCC (Missal 2016; Rana 2015; Smillie 2014). For some participants, their access to a phone was controlled by others. This could be because they could not afford to purchase a phone themselves (Jalloh‐Vos 2014), or because physical access to the mobile phone was controlled by another person (see Finding 18) (Akinfaderin‐Agarau 2012; Jalloh‐Vos 2014; Rana 2015). This group was mainly comprised of women and adolescents, and this is discussed further in Findings 7 and 18.

t afford to purchase a phone themselves (Jalloh‐Vos 2014), or because physical access to the mobile phone was controlled by another person (see Finding 18) (Akinfaderin‐Agarau 2012; Jalloh‐Vos 2014; Rana 2015). This group was mainly comprised of women and adolescents, and this is discussed further in Findings 7 and 18. Finding 7: Some participants, particularly women and adolescents, had their access to phones controlled or restricted by others, especially if they had to share or borrow a phone. They noted that they would often have to explain why they wanted to use the phone, and who they wanted to call, to allay suspicions about this communication. They mentioned that this was a barrier to accessing DTCC and made it difficult to keep their messages private (moderate confidence).

d to share or borrow a phone. They noted that they would often have to explain why they wanted to use the phone, and who they wanted to call, to allay suspicions about this communication. They mentioned that this was a barrier to accessing DTCC and made it difficult to keep their messages private (moderate confidence). A few studies from LMIC settings in Africa found that some participants, particularly women and adolescents, had their access to phones controlled or restricted by others, especially if they had to share or borrow a phone (Akinfaderin‐Agarau 2012; Flax 2017; Jalloh‐Vos 2014; Rana 2015). They mentioned that this was a barrier to accessing DTCC and made it difficult to keep their messages private. In some contexts, women had their mobile phone use controlled by their husbands or other family members. The women would often have to explain why they wanted to use the phone and who they wanted to call (Akinfaderin‐Agarau 2012; Jalloh‐Vos 2014; Rana 2015), for example to allay suspicion that they were talking to their boyfriends or having an affair. For some women, this would mean having to find an alternative phone to use if they did not want their husband to know they were using a digital health service. For example, in one study, some women did not want their husbands knowing they were receiving information on family planning (Jalloh‐Vos 2014). In some settings, women and girls were also viewed by their society as not having time to use phones due to greater domestic obligations than their male counterparts (Akinfaderin‐Agarau 2012). Youth in one study also reported facing restrictions related to using phones at school (Rana 2015).

planning (Jalloh‐Vos 2014). In some settings, women and girls were also viewed by their society as not having time to use phones due to greater domestic obligations than their male counterparts (Akinfaderin‐Agarau 2012). Youth in one study also reported facing restrictions related to using phones at school (Rana 2015). In one study, women in a women’s group all shared a single phone. They elected one group member to control the phone and share the messages. This group member was then responsible for distributing the messages from the DTCC. In this context, the majority of the participants accepted this form of phone sharing and believed it was functional (Flax 2017). In all studies contributing to this finding, participants felt that sharing a phone or having access to their phone controlled by someone else delayed the delivery of the message (Flax 2017; Rana 2015), and decreased the privacy and confidentiality around their personal information (Akinfaderin‐Agarau 2012; Jalloh‐Vos 2014; Rana 2015). Finding 8: Participants believed that the cost of participating in DTCC should be free or very low, as cost could present a barrier to participation, particularly for young people and those on lower incomes. Participants felt that there should be little or no charge for costs such as joining the digital health intervention, downloading applications (apps), or for sending and receiving mobile messages/phone calls (high confidence).

present a barrier to participation, particularly for young people and those on lower incomes. Participants felt that there should be little or no charge for costs such as joining the digital health intervention, downloading applications (apps), or for sending and receiving mobile messages/phone calls (high confidence). Participants in several studies felt it was important for digital health interventions to have little or no cost, as these costs could present a barrier to participation (Akinfaderin‐Agarau 2012; Calderón 2017; Cornelius 2009; Menacho 2013; Mitchell 2016; Perry 2012; Rana 2015; Smith 2017). This was especially important to young people (Akinfaderin‐Agarau 2012; Perry 2012), and those with lower incomes (Calderón 2017; Cornelius 2009; Mitchell 2016; Rana 2015). If the intervention could not be offered at no cost to the client, then participants felt that the interventions should be very low cost and that cheaper options should be used, for example SMS instead of voice calls (Akinfaderin‐Agarau 2012). In some cases, messages sent to participants were free, but if a participant wanted to reply they had to use their own airtime. Some participants thought that this would prevent people from using the bi‐directional functions within digital health interventions (Rana 2015). Finding 9: Participants’ ability to access digital communication was sometimes limited by their language skills and their personal level of literacy and/or techno‐literacy (moderate confidence).

Participants in several studies felt it was important for digital health interventions to have little or no cost, as these costs could present a barrier to participation (Akinfaderin‐Agarau 2012; Calderón 2017; Cornelius 2009; Menacho 2013; Mitchell 2016; Perry 2012; Rana 2015; Smith 2017). This was especially important to young people (Akinfaderin‐Agarau 2012; Perry 2012), and those with lower incomes (Calderón 2017; Cornelius 2009; Mitchell 2016; Rana 2015). If the intervention could not be offered at no cost to the client, then participants felt that the interventions should be very low cost and that cheaper options should be used, for example SMS instead of voice calls (Akinfaderin‐Agarau 2012). In some cases, messages sent to participants were free, but if a participant wanted to reply they had to use their own airtime. Some participants thought that this would prevent people from using the bi‐directional functions within digital health interventions (Rana 2015). Finding 9: Participants’ ability to access digital communication was sometimes limited by their language skills and their personal level of literacy and/or techno‐literacy (moderate confidence). Some studies, the majority from LMIC settings (Akinfaderin‐Agarau 2012; Calderón 2017; Curioso 2009; Greaney 2014; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Mbuagbaw 2014; Rodrigues 2015; Smillie 2014), found that participants’ ability to access digital health messages was sometimes limited by their language skills (Akinfaderin‐Agarau 2012; Jalloh‐Vos 2014; Mbuagbaw 2014; Smillie 2014), or not understanding how to use the technology (Calderón 2017; Curioso 2009; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Rodrigues 2015; Smillie 2014). One study from the USA found that Latina women needing cancer screening receiving interactive voice recordings as reminders understood how to access messages but that the language used in the messages was not familiar to them (Greaney 2014), as illustrated in the following quote.

os 2014; Rodrigues 2015; Smillie 2014). One study from the USA found that Latina women needing cancer screening receiving interactive voice recordings as reminders understood how to access messages but that the language used in the messages was not familiar to them (Greaney 2014), as illustrated in the following quote. “'If your response is ‘yes’, then press the star button.’ ‘Star’ is what Americans say, but on the telephone there is no star; it is an asterisk. It all depends on who you are speaking with. If the person you are speaking with understands that that is a star then let’s press the star, but if the person understands it’s an asterisk then he/she will begin to look for a star” (Greaney 2014). Participants with literacy issues often preferred voice calls to SMS, as they could talk with the caller and ask for clarifying information (Akinfaderin‐Agarau 2012; Jalloh‐Vos 2014). Participants in two studies said they had learned or were willing to learn how to text in order to participate in digital health interventions (Calderón 2017; Smillie 2014). Communication delivery and format preferences ('Summary of qualitative findings' table for Findings 10 to 13 is shown in Table 4.)Table 4 'Summary of qualitative findings' table for findings related to communication delivery and format preferences

Participants with literacy issues often preferred voice calls to SMS, as they could talk with the caller and ask for clarifying information (Akinfaderin‐Agarau 2012; Jalloh‐Vos 2014). Participants in two studies said they had learned or were willing to learn how to text in order to participate in digital health interventions (Calderón 2017; Smillie 2014). Communication delivery and format preferences ('Summary of qualitative findings' table for Findings 10 to 13 is shown in Table 4.)Table 4 'Summary of qualitative findings' table for findings related to communication delivery and format preferences Finding Overall CERQual assessment Explanation for assessment Contributing studies 10 Participants often had preferences for how often health messages were sent, the time of day they were sent, and the duration of the digital targeted client communication. However, there was variation in what most participants felt was appropriate timing and frequency, and these preferences were often linked to the health issue on which the messaging was focused; whether people had their own phone or had to share a phone; and the participant’s particular circumstances. Participants were particularly concerned about being bombarded with too many messages; whether the timing of the messages was convenient for them; and/or whether messages arrived in connection with the behaviour the message was trying to target. Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Calderón 2017; Cornelius 2009; Evans 2016; French 2016; Gold 2010; Greaney 2014; Jennings 2013; Mbuagbaw 2012; Menacho 2013; Missal 2016; Mitchell 2016; Munro 2017; Naughton 2013; Odeny 2014; Rana 2015; Rodrigues 2015; Sloan 2017; Smillie 2014; Smith 2017; Ware 2016; Willoughby 2017; Wright 2011 11 Participants had different preferences for various delivery channels available for sharing information through digital targeted client communication, including mobile messaging, interactive voice response, or speaking with a healthcare provider. These preferences were influenced by a number of factors including cost, convenience, the ability to store messages and re‐read them, familiarity with the channel, personal preferences, the nature of the content being delivered, the nature of the topic, language and literacy considerations, and the ability to have a discussion with a real‐life person.

by a number of factors including cost, convenience, the ability to store messages and re‐read them, familiarity with the channel, personal preferences, the nature of the content being delivered, the nature of the topic, language and literacy considerations, and the ability to have a discussion with a real‐life person. Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Akinfaderin‐Agarau 2012; Cates 2015; Curioso 2009; Greaney 2014; Jennings 2013; Missal 2016; Mitchell 2016; Naughton 2013; Odeny 2014; Rana 2015; Rodrigues 2015; Smillie 2014; Willoughby 2017 12 Participants appreciated personalised health information and discussed their preferences for options to make interventions more relevant to individuals. This could include sender‐based personalisation or receiver‐based options. Reasons for these preferences included engaging the user, enhancing credibility, increasing feelings of ownership, control over their personal information and feelings of privacy. Preferences for tailoring included making digital health messages personalised by using an individual's name; allowing participants to choose the content, topic, and language of their messages; providing information relevant to the participant's setting (local information); allowing them to select the timing and frequency of the message; providing personalised reminders (e.g. for vaccination or medication); and allowing participants to have control over privacy settings. Low confidence Due to minor concerns regarding methodological limitations and serious concerns regarding relevance Calderón 2017; Evans 2016; French 2016; Goldenberg 2015; Hirsch‐Moverman 2017; Jennings 2013; Munro 2017; Naughton 2013; Odeny 2014; Sloan 2017; Ware 2016; Willoughby 2017 13 Participants mentioned various message formats that they preferred. These included a preference for short, concise, personalised, clear, and direct messages in a language they could understand and in full text rather than "text speak". Low confidence Due to minor concerns regarding methodological limitations and serious concerns regarding relevance Akinfaderin‐Agarau 2012; Calderón 2017; Cates 2015; Curioso 2009; Evans 2016; French 2016; Gold 2010; Greaney 2014; Lau 2014; Menacho 2013; Missal 2016; Munro 2017; Naughton 2013; Odeny 2014; Perry 2012; Rana 2015; Smillie 2014; Willoughby 2017

garding methodological limitations and serious concerns regarding relevance Akinfaderin‐Agarau 2012; Calderón 2017; Cates 2015; Curioso 2009; Evans 2016; French 2016; Gold 2010; Greaney 2014; Lau 2014; Menacho 2013; Missal 2016; Munro 2017; Naughton 2013; Odeny 2014; Perry 2012; Rana 2015; Smillie 2014; Willoughby 2017 Finding 10: Participants often had preferences for how often health messages were sent, the time of day they were sent, and the duration of the DTCC. However, there was variation in what most participants felt was appropriate timing and frequency, and these preferences were often linked to the health issue on which the messaging was focused; whether people had their own phone or had to share a phone; and the participant’s particular circumstances. Participants were particularly concerned about being bombarded with too many messages; whether the timing of the messages was convenient for them; and/or whether messages arrived in connection with the behaviour the message was trying to target (moderate confidence).

one; and the participant’s particular circumstances. Participants were particularly concerned about being bombarded with too many messages; whether the timing of the messages was convenient for them; and/or whether messages arrived in connection with the behaviour the message was trying to target (moderate confidence). Many studies discussed and presented participants’ preferences related to timing, frequency of messages, and duration of digital health projects (Calderón 2017; Cornelius 2009, Evans 2016; French 2016; Gold 2010; Greaney 2014; Jennings 2013; Mbuagbaw 2012; Menacho 2013; Missal 2016; Mitchell 2016; Munro 2017; Naughton 2013; Odeny 2014; Rana 2015; Rodrigues 2015; Sloan 2017; Smillie 2014; Smith 2017; Ware 2016; Willoughby 2017; Wright 2011). However, within and across studies and client groups, there was no consensus as to the ideal timing, frequency, or duration, as this was linked to personal preferences, contextual factors (such as attending school), and the behaviour or information the text message was trying to target.

; Willoughby 2017; Wright 2011). However, within and across studies and client groups, there was no consensus as to the ideal timing, frequency, or duration, as this was linked to personal preferences, contextual factors (such as attending school), and the behaviour or information the text message was trying to target. With regard to frequency, participants did not want in general to feel pestered or bombarded by too many messages (Evans 2016; French 2016; Gold 2010; Willoughby 2017), but described a fine balance between feeling bombarded and not receiving enough information to reinforce the messages. For example, participants receiving medication reminders were open to receiving multiple texts a day, whereas those receiving more general/less tailored information wanted messages much less frequently. However, in a number of studies no clear consensus emerged on the optimal frequency of the messages (Evans 2016; Gold 2010; Rana 2015). Participants' preferences for message frequency could also be linked to owning a phone or having to share (Jennings 2013). Those who had to share a phone wanted messages only a few times a week with the ability to stop the messages if the phone owner was away, whereas those who owned their own phone were open to daily messages (Jennings 2013).

for message frequency could also be linked to owning a phone or having to share (Jennings 2013). Those who had to share a phone wanted messages only a few times a week with the ability to stop the messages if the phone owner was away, whereas those who owned their own phone were open to daily messages (Jennings 2013). With regard to the timing of delivery of the messages, preferences for timing varied among population groups. It was important to participants that messages arrive when they could be seen and accessed, for example not late at night, when they would not be seen until the next day (Rana 2015; Ware 2016). Some adolescents and young adults thought during school hours would be fine (Cornelius 2009; French 2016), whereas others believed that this could cause problems based on restrictions around phone use (Cornelius 2009; Rana 2015; Wright 2011). Some participants felt that it was important that the message arrive in connection with the behaviour it was targeting (French 2016; Gold 2010; Menacho 2013; Naughton 2013; Rana 2015; Ware 2016), or in good time before an appointment (Odeny 2014), for example on a Friday night before going out to remind them of condom use (French 2016). Participants also liked the option of tailoring the timing of messages to fit their lives (French 2016). With regard to duration, no consensus emerged on how long the intervention should last (Evans 2016).

With regard to the timing of delivery of the messages, preferences for timing varied among population groups. It was important to participants that messages arrive when they could be seen and accessed, for example not late at night, when they would not be seen until the next day (Rana 2015; Ware 2016). Some adolescents and young adults thought during school hours would be fine (Cornelius 2009; French 2016), whereas others believed that this could cause problems based on restrictions around phone use (Cornelius 2009; Rana 2015; Wright 2011). Some participants felt that it was important that the message arrive in connection with the behaviour it was targeting (French 2016; Gold 2010; Menacho 2013; Naughton 2013; Rana 2015; Ware 2016), or in good time before an appointment (Odeny 2014), for example on a Friday night before going out to remind them of condom use (French 2016). Participants also liked the option of tailoring the timing of messages to fit their lives (French 2016). With regard to duration, no consensus emerged on how long the intervention should last (Evans 2016). Finding 11: Participants had different preferences for various delivery channels available for sharing information through DTCC, including mobile messaging, interactive voice response, or speaking with a healthcare provider. These preferences were influenced by a number of factors including cost, convenience, the ability to store messages and re‐read them, familiarity with the channel, personal preferences, the nature of the content being delivered, the nature of the topic, language and literacy considerations, and the ability to have a discussion with a real‐life person (moderate confidence).

s including cost, convenience, the ability to store messages and re‐read them, familiarity with the channel, personal preferences, the nature of the content being delivered, the nature of the topic, language and literacy considerations, and the ability to have a discussion with a real‐life person (moderate confidence). Some studies presented data related to participants’ preferences for the delivery channel used to share information for digital health interventions (Akinfaderin‐Agarau 2012; Cates 2015; Curioso 2009; Greaney 2014; Jennings 2013; Mitchell 2016; Missal 2016; Naughton 2013; Odeny 2014; Rana 2015; Rodrigues 2015; Smillie 2014; Willoughby 2017). Some participants believed that different delivery channels would meet different needs and have different purposes (Willoughby 2017). Participants from a number of studies had a preference for information delivered by SMS (Akinfaderin‐Agarau 2012; Curioso 2009; Jennings 2013; Mitchell 2016; Naughton 2013; Odeny 2014; Rana 2015; Smillie 2014). Reasons for preferring SMS included lower cost for the participant (Akinfaderin‐Agarau 2012); the messages were brief (Jennings 2013; Naughton 2013); confidential and protected privacy (Akinfaderin‐Agarau 2012; Smillie 2014); could be kept for reference and re‐reading later (Curioso 2009; Jennings 2013); were easier to share (Akinfaderin‐Agarau 2012; Naughton 2013); and to understand, for example due to literacy issues or not understanding the accent of the person calling (Akinfaderin‐Agarau 2012; Smillie 2014).

‐Agarau 2012; Smillie 2014); could be kept for reference and re‐reading later (Curioso 2009; Jennings 2013); were easier to share (Akinfaderin‐Agarau 2012; Naughton 2013); and to understand, for example due to literacy issues or not understanding the accent of the person calling (Akinfaderin‐Agarau 2012; Smillie 2014). However, participants in a few studies raised concerns with text messaging or preferred other delivery channels. They thought, for example, that text messages were more appropriate for younger audiences (Willoughby 2017), or that the highly convenient nature of the SMS was a negative that would only have a short‐term impact, as people would delete them or ignore them (Naughton 2013; Rana 2015). Finally, others who were unfamiliar with SMS technology felt that they were too passive and unfamiliar and so they ignored them (Rodrigues 2015). A few studies presented data that participants preferred interactive voice response (IVR) to receive digital health interventions (Greaney 2014; Missal 2016; Rodrigues 2015). Often this was related to low literacy levels in a community or that participants found them easier, more interactive, and that they attracted more attention than an SMS, meaning that they would not be easily missed (Rodrigues 2015). However, some participants felt that they should be able to repeat or re‐read the message at a time more convenient to them, for example through linked delivery on platforms such as voice, video, or SMS (Missal 2016).

y attracted more attention than an SMS, meaning that they would not be easily missed (Rodrigues 2015). However, some participants felt that they should be able to repeat or re‐read the message at a time more convenient to them, for example through linked delivery on platforms such as voice, video, or SMS (Missal 2016). A number of studies presented data where participants discussed their preferences for speaking directly with a healthcare provider on the phone to other delivery platforms (Akinfaderin‐Agarau 2012; Greaney 2014; Jennings 2013; Odeny 2014; Rana 2015; Rodrigues 2015). Participants talked about why they preferred speaking directly to a healthcare provider. These reasons included that phone calls were clear and could aid understanding in people with low literacy levels (Akinfaderin‐Agarau 2012); they trusted speaking to a healthcare provider (Akinfaderin‐Agarau 2012); they could engage in discussions (Jennings 2013); and they felt it was faster and more immediate to receive an answer to their questions (Akinfaderin‐Agarau 2012; Jennings 2013; Rodrigues 2015). Participants also discussed other delivery channels that they liked. These included an app (Mitchell 2016), written letters (Greaney 2014), voicemails, Facebook, and reminder alarms (Rana 2015).

A number of studies presented data where participants discussed their preferences for speaking directly with a healthcare provider on the phone to other delivery platforms (Akinfaderin‐Agarau 2012; Greaney 2014; Jennings 2013; Odeny 2014; Rana 2015; Rodrigues 2015). Participants talked about why they preferred speaking directly to a healthcare provider. These reasons included that phone calls were clear and could aid understanding in people with low literacy levels (Akinfaderin‐Agarau 2012); they trusted speaking to a healthcare provider (Akinfaderin‐Agarau 2012); they could engage in discussions (Jennings 2013); and they felt it was faster and more immediate to receive an answer to their questions (Akinfaderin‐Agarau 2012; Jennings 2013; Rodrigues 2015). Participants also discussed other delivery channels that they liked. These included an app (Mitchell 2016), written letters (Greaney 2014), voicemails, Facebook, and reminder alarms (Rana 2015). Finding 12: Participants appreciated personalised health information and discussed their preferences for options to make interventions more relevant to individuals. This could include sender‐based personalisation or receiver‐based options. Reasons for these preferences included engaging the user, enhancing credibility, increasing feelings of ownership, control over their personal information, and feelings of privacy. Preferences for tailoring included making digital health messages personalised by using an individual's name; allowing participants to choose the content, topic, and language of their messages; providing information relevant to the participant's setting (local information); allowing them to select the timing and frequency of the message; providing personalised reminders (e.g. for vaccination or medication); and allowing participants to have control over privacy settings (low confidence).

ge of their messages; providing information relevant to the participant's setting (local information); allowing them to select the timing and frequency of the message; providing personalised reminders (e.g. for vaccination or medication); and allowing participants to have control over privacy settings (low confidence). Some studies discussed participants’ thoughts about personalised or customised messaging and their preferences for options to make interventions more relevant to individuals (Calderón 2017; Evans 2016; French 2016; Goldenberg 2015; Hirsch‐Moverman 2017; Jennings 2013; Munro 2017; Naughton 2013; Odeny 2014; Sloan 2017; Ware 2016; Willoughby 2017). Some participants felt that if the messages were not tailored to individual users this could cause them to disengage from interventions or cause problems in their personal lives (Jennings 2013; Munro 2017; Naughton 2013, ). For example, participants in one study on HIV messaging in Kenya to prevent mother‐to‐child transmission suggested that two different sets of messages be developed for those who had and those who had not disclosed their HIV status to their partner. This would help women avoid risking disclosure of their status (Jennings 2013). Participants in another study discussed the importance of using personal pronouns to make messages more relevant to the user. They felt that naming functions within the digital health intervention as, for example, “ MY test plan”, would help participants to take ownership of these functions (Goldenberg 2015).

gs 2013). Participants in another study discussed the importance of using personal pronouns to make messages more relevant to the user. They felt that naming functions within the digital health intervention as, for example, “ MY test plan”, would help participants to take ownership of these functions (Goldenberg 2015). Participants in several studies mentioned ways in which the people receiving communication could personalise or customise digital health interventions that they believed would be important. These included the following. Being able to request the time of day the message(s) would be sent (French 2016; Hirsch‐Moverman 2017; Jennings 2013; Sloan 2017; Ware 2016; Willoughby 2017). Being able to select the frequency of the messages (Hirsch‐Moverman 2017; Willoughby 2017). Being able to customise the app to meet personal needs, such as reminders and privacy settings (Goldenberg 2015). Being able to select a preferred language (Jennings 2013; Ware 2016). Being able to personalise or select message content (Jennings 2013; Sloan 2017; Ware 2016; Willoughby 2017). Participants in a number of studies mentioned ways in which the people designing and sending communication could personalise or customise digital health interventions that they believed would be important. These included the following. Using unique access codes in order to prevent children and other adults from unintentionally gaining entry into discussed personal health information (Calderón 2017). Using the individual’s name (Evans 2016). Delivering the message in different languages (Evans 2016).

Participants in a number of studies mentioned ways in which the people designing and sending communication could personalise or customise digital health interventions that they believed would be important. These included the following. Using unique access codes in order to prevent children and other adults from unintentionally gaining entry into discussed personal health information (Calderón 2017). Using the individual’s name (Evans 2016). Delivering the message in different languages (Evans 2016). Including community‐specific information (Evans 2016; Munro 2017). Being able to, for example, text “STOP” if they wanted to stop receiving messages (French 2016). Explaining the reason for the appointment as well as the date (e.g. the specific vaccinations to be received on that day) (Odeny 2014). Finding 13: Participants mentioned various message formats that they preferred. These included a preference for short, concise, personalised, clear, and direct messages in a language they could understand and in full text rather than "text speak" (low confidence).

Explaining the reason for the appointment as well as the date (e.g. the specific vaccinations to be received on that day) (Odeny 2014). Finding 13: Participants mentioned various message formats that they preferred. These included a preference for short, concise, personalised, clear, and direct messages in a language they could understand and in full text rather than "text speak" (low confidence). A number of studies presented data related to participants’ preferences for message format (Akinfaderin‐Agarau 2012; Calderón 2017; Cates 2015; Curioso 2009; Evans 2016; French 2016; Gold 2010; Greaney 2014; Lau 2014; Menacho 2013; Missal 2016; Munro 2017; Naughton 2013; Odeny 2014; Perry 2012; Rana 2015; Smillie 2014; Willoughby 2017). In general, participants liked or wanted short and concise messages that were easy to understand and factual, especially from text messages (Calderón 2017; Curioso 2009; Evans 2016; French 2016; Gold 2010; Greaney 2014; Lau 2014; Menacho 2013; Odeny 2014; Perry 2012; Rana 2015; Wright 2011). Most participants preferred text messages that were written out in full and did not use abbreviations or slang, as these felt more professional and were more representative of how they thought a health professional would write (Cates 2015; French 2016; Naughton 2013; Willoughby 2017).

rry 2012; Rana 2015; Wright 2011). Most participants preferred text messages that were written out in full and did not use abbreviations or slang, as these felt more professional and were more representative of how they thought a health professional would write (Cates 2015; French 2016; Naughton 2013; Willoughby 2017). In one study (Missal 2016), participants commented on their experiences with interactive voice response (IVR). Most found the content and language of the message useful. However, some found the message too fast, short, and sometimes the voice was not clear enough. They were unable to request for the message to be played again if they had missed or misunderstood the content. Communication content preferences ('Summary of qualitative findings' table for Fndings 14 to 17 is shown in Table 5.)Table 5 'Summary of qualitative findings' table for findings related to communication content preferences

In one study (Missal 2016), participants commented on their experiences with interactive voice response (IVR). Most found the content and language of the message useful. However, some found the message too fast, short, and sometimes the voice was not clear enough. They were unable to request for the message to be played again if they had missed or misunderstood the content. Communication content preferences ('Summary of qualitative findings' table for Fndings 14 to 17 is shown in Table 5.)Table 5 'Summary of qualitative findings' table for findings related to communication content preferences Finding Overall CERQual assessment Explanation for assessment Contributing studies 14 Participants’ perceptions of who sent the digital health communication could influence their trust in and perception of the credibility and value of the digital targeted client communication and the information it provides. Participants said they wanted a known, identified phone number; messages sent from a reliable, trusted, credible source such as health professionals or official sources; and in some cases to feel like the messages were sent by a person (even if sent from an automated service). However, some participants, such as those with stigmatised health conditions, preferred an unmarked sender to protect their privacy. Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Akinfaderin‐Agarau 2012; Brown 2014; Calderón 2017; Cates 2015; Evans 2016; Greaney 2014; Lau 2014; Mbuagbaw 2012; Menacho 2013; Missal 2016; Naughton 2013; Rana 2015; Rodrigues 2015; Smillie 2014; Willoughby 2017 15 Participants said that the tone of digital health communication mattered to them. Their preferences varied but included a tone that was: motivational, friendly, encouraging, polite, respectful, congratulatory, personalised, upbeat, positive, humorous, and relatable. Some participants highlighted that they did not like feeling pressured, lectured to, shamed, or frightened by digital health messages. Low confidence Due to minor concerns regarding methodological limitations and serious concerns regarding relevance Cates 2015; Curioso 2009; Evans 2016; French 2016; Gold 2010; Jennings 2013; Menacho 2013; Munro 2017; Naughton 2013; Odeny 2014; Perry 2012; Rana 2015; Sloan 2017; Wright 2011 16 Participants had preferences regarding the content they receive through digital targeted client communication. They wanted varied content that provided new knowledge and reminders, as well as explanations, solutions, and suggestions about health issues. They were interested in content related to health, illness, and treatments and practical topics such as health facility location and transportation.

rgeted client communication. They wanted varied content that provided new knowledge and reminders, as well as explanations, solutions, and suggestions about health issues. They were interested in content related to health, illness, and treatments and practical topics such as health facility location and transportation. They wanted this information to be relevant and acceptable to their personal circumstances and local setting. Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Brown 2014; Calderón 2017; Cornelius 2009; Entsieh 2015; French 2016; Gold 2010; Greaney 2014; Jalloh‐Vos 2014; Jennings 2013; Mbuagbaw 2014; Missal 2016; Mitchell 2016; Munro 2017; Nachega 2016; Odeny 2014; Perry 2012; Sloan 2017; Smith 2017 17 Some participants felt that including elements in the mobile‐based platform in which participants are asked for a response (e.g. via knowledge quizzes or multiple‐choice questions or a practical tool allowing access to additional information, such as a nutrition calculator) could increase the engagement of users with the intervention, its content, and provide additional information to them. In one study, participants suggested that it would be helpful if the response was quick, simple, and convenient. Low confidence Due to minor concerns regarding methodological limitations, moderate concerns regarding adequacy, and serious concerns regarding relevance Cornelius 2009; Munro 2017; Naughton 2013; Wright 2011

n one study, participants suggested that it would be helpful if the response was quick, simple, and convenient. Low confidence Due to minor concerns regarding methodological limitations, moderate concerns regarding adequacy, and serious concerns regarding relevance Cornelius 2009; Munro 2017; Naughton 2013; Wright 2011 Finding 14: Participants’ perceptions of who sent the digital health communication could influence their trust in and perception of the credibility and value of the DTCC and the information it provides. Participants said they wanted a known, identified phone number; messages sent from a reliable, trusted, credible source such as health professionals or official sources; and in some cases to feel like the messages were sent by a person (even if sent from an automated service). However, some participants, such as those with stigmatised health conditions, preferred an unmarked sender to protect their privacy (moderate confidence).

source such as health professionals or official sources; and in some cases to feel like the messages were sent by a person (even if sent from an automated service). However, some participants, such as those with stigmatised health conditions, preferred an unmarked sender to protect their privacy (moderate confidence). Several studies found that participants’ perceptions of who the sender of the message was could influence their trust in and perception of the digital health intervention’s credibility and value (Akinfaderin‐Agarau 2012; Brown 2014; Calderón 2017; Cates 2015; Evans 2016; Greaney 2014; Mbuagbaw 2012; Menacho 2013; Missal 2016; Naughton 2013; Lau 2014; Rana 2015; Rodrigues 2015; Smillie 2014; Willoughby 2017). Participants in many studies identified that the sender should be known and identifiable. If this was not the case many stated that they were more likely to ignore or delete the message (Akinfaderin‐Agarau 2012; Greaney 2014; Smillie 2014), as their phones were already receiving many messages linked with telemarketing (Menacho 2013; Missal 2016). If a sender was unknown, it was also felt that this would decrease the credibility of the message. However, some participants preferred an unmarked sender in order to protect their privacy. This was in the case, for example, of a stigmatised health condition such as HIV, where if the sender's phone number were identifiable, a person’s HIV status may be inadvertently revealed (Rana 2015). These participants felt that a solution would be to allow participants to choose if they wanted the number or name of the sender to be attached to the message (Rana 2015).

health condition such as HIV, where if the sender's phone number were identifiable, a person’s HIV status may be inadvertently revealed (Rana 2015). These participants felt that a solution would be to allow participants to choose if they wanted the number or name of the sender to be attached to the message (Rana 2015). Participants also wanted the messages to come from a reliable, trusted, and credible source (Brown 2014; Cates 2015; Evans 2016; Lau 2014; Willoughby 2017). They were interested in the credentials or education of the person creating or responding to the message. Many felt that it was important that the information be written and sent by health professionals and from official sources (Cates 2015; Greaney 2014; Lau 2014; Mbuagbaw 2012). In some cases, if this was unclear, this led them to worry about the accuracy of message content and the intentions of the proposed intervention (Calderón 2017). If the participant knew and trusted the source, they also felt that their data would be protected (Evans 2016).

2015; Greaney 2014; Lau 2014; Mbuagbaw 2012). In some cases, if this was unclear, this led them to worry about the accuracy of message content and the intentions of the proposed intervention (Calderón 2017). If the participant knew and trusted the source, they also felt that their data would be protected (Evans 2016). Many participants felt that it was important that the messages felt like they were coming from an actual person rather than from an automated system ( Lau 2014; Naughton 2013; Rodrigues 2015; Willoughby 2017). Some participants felt that since the messages were written by a health professional, the computer’s role in the automated sending of the messages was irrelevant (Naughton 2013). These participants suggested that messages which were not repeated and were delivered at various times of day using the participant’s name would feel more like they were sent by a person (Naughton 2013). Finding 15: Participants said that the tone of digital health communication mattered to them. Their preferences varied but included a tone that was: motivational, friendly, encouraging, polite, respectful, congratulatory, personalised, upbeat, positive, humorous, and relatable. Some participants highlighted that they did not like feeling pressured, lectured, shamed, or frightened by digital health messages (low confidence).

nces varied but included a tone that was: motivational, friendly, encouraging, polite, respectful, congratulatory, personalised, upbeat, positive, humorous, and relatable. Some participants highlighted that they did not like feeling pressured, lectured, shamed, or frightened by digital health messages (low confidence). A number of studies, the majority from HIC contexts (Cates 2015; Curioso 2009; Evans 2016; French 2016; Gold 2010; Jennings 2013; Menacho 2013; Munro 2017; Naughton 2013; Odeny 2014; Perry 2012; Rana 2015; Sloan 2017; Wright 2011), presented data that discussed participants’ preferences related to the tone of the messaging used in DTCC delivered via mobile devices. In general, there was consensus across studies that messages should be polite and respectful. Participants indicated that the tone of the message could influence their acceptance of the message (Cates 2015; French 2016; Jennings 2013); their trust in the message content (French 2016); the credibility of the message; and their engagement with the messages and the digital health intervention (Cates 2015; Curioso 2009; French 2016; Gold 2010; Jennings 2013; Menacho 2013; Munro 2017; Wright 2011).

of the message (Cates 2015; French 2016; Jennings 2013); their trust in the message content (French 2016); the credibility of the message; and their engagement with the messages and the digital health intervention (Cates 2015; Curioso 2009; French 2016; Gold 2010; Jennings 2013; Menacho 2013; Munro 2017; Wright 2011). Participants liked when the tone of the message was, for example, motivational (Curioso 2009; Rana 2015), encouraging (Curioso 2009; Jennings 2013; Munro 2017; Naughton 2013; Odeny 2014; Sloan 2017), upbeat, positive, and reassuring (Evans 2016; Gold 2010; Munro 2017; Perry 2012), friendly (French 2016; Sloan 2017), polite and respectful (Evans 2016; Odeny 2014; Wright 2011), humorous (Gold 2010; Menacho 2013; Wright 2011), or supportive (Jennings 2013; Munro 2017; Odeny 2014; Sloan 2017). Participants did not like it when messages were used as a warning, to scare people, or were based in fear (Cates 2015; Evans 2016; Munro 2017), or if they felt pressured, told off, shamed, lectured, or patronised by the content (French 2016; Munro 2017). Finding 16: Participants had preferences regarding the content they receive through DTCC. They wanted varied content that provided new knowledge and reminders, as well as explanations, solutions, and suggestions about health issues. They were interested in content related to health, illness, and treatments and practical topics such as health facility location and transportation. They wanted this information to be relevant and acceptable to their personal circumstances and local setting (moderate confidence).

s, and suggestions about health issues. They were interested in content related to health, illness, and treatments and practical topics such as health facility location and transportation. They wanted this information to be relevant and acceptable to their personal circumstances and local setting (moderate confidence). A number of studies presented data related to participants’ preferences regarding the content they received through digital health interventions (Brown 2014; Calderón 2017; Cornelius 2009; Entsieh 2015; French 2016; Gold 2010; Greaney 2014; Jalloh‐Vos 2014; Jennings 2013; Mbuagbaw 2014; Mitchell 2016; Missal 2016; Munro 2017; Nachega 2016; Odeny 2014; Perry 2012; Sloan 2017; Smith 2017). Participants expressed preferences for varied content (Brown 2014; Entsieh 2015; Missal 2016; Perry 2012), that provided them with new knowledge (Brown 2014; Calderón 2017; Cornelius 2009; Entsieh 2015; French 2016; Gold 2010; Jalloh‐Vos 2014; Missal 2016; Munro 2017; Odeny 2014; Perry 2012; Sloan 2017; Smith 2017), or a reminder to take a medication or to reinforce something they knew already (Akinfaderin‐Agarau 2012; Gold 2010; Greaney 2014; Mbuagbaw 2014; Nachega 2016). Participants liked when the content of digital health interventions gave explanations, solutions, and suggestions about health issues (Brown 2014; Entsieh 2015; French 2016; Gold 2010; Greaney 2014; Jalloh‐Vos 2014; Missal 2016; Munro 2017; Perry 2012). An example of such practical advice is the suggestion that mothers who struggled with breastfeeding (or the idea of breastfeeding) could pump their milk and give it to their infants in a bottle instead of switching to formula milk (Brown 2014), as expressed in the quote below from a teenage mother in the USA.

ry 2012). An example of such practical advice is the suggestion that mothers who struggled with breastfeeding (or the idea of breastfeeding) could pump their milk and give it to their infants in a bottle instead of switching to formula milk (Brown 2014), as expressed in the quote below from a teenage mother in the USA. “I was kind of grossed out by actually breast feeding so I had decided to just use formula. I don't know why I was so grossed out, I just was. But then you said in one of the messages that you can pump and feed the breast milk through a bottle, and the baby still gets all those benefits. It makes sense, but I just never thought of it. Because of that message, I started to do that and my baby still gets breast milk, but otherwise I would have given up” (Brown 2014). Finally, participants were interested in content related to health, illness, and treatments and practical topics such as health facility location and transportation (Brown 2014; Calderón 2017; Greaney 2014; Jalloh‐Vos 2014; Jennings 2013; Missal 2016; Mitchell 2016; Munro 2017; Perry 2012). Many wanted this information to be relevant and acceptable to their personal circumstances and local setting. For example, participants from a few studies expressed an interest in information related to the location of health facilities, including transport information, and contact information for the health facility closest to them to be included in the content (Greaney 2014; Mitchell 2016; Perry 2012).

ces and local setting. For example, participants from a few studies expressed an interest in information related to the location of health facilities, including transport information, and contact information for the health facility closest to them to be included in the content (Greaney 2014; Mitchell 2016; Perry 2012). Finding 17: Some participants felt that including elements in the mobile‐based platform in which participants are asked for a response (e.g. via knowledge quizzes or multiple‐choice questions or a practical tool allowing access to additional information, such as a nutrition calculator) could increase the engagement of users with the intervention, its content, and provide additional information to them. In one study, participants suggested that it would be helpful if the response was quick, simple, and convenient (low confidence).

owing access to additional information, such as a nutrition calculator) could increase the engagement of users with the intervention, its content, and provide additional information to them. In one study, participants suggested that it would be helpful if the response was quick, simple, and convenient (low confidence). Participants in a few studies, all from HIC contexts (Cornelius 2009; Munro 2017; Naughton 2013; Wright 2011), felt that bi‐directional communication or content that somehow engaged clients (e.g. using quizzes or replies) would be more useful than one‐way communication. Participants mentioned that for this to be the case, the interaction should be made convenient and allow for quick and simple responses. Such interactive options would help keep participants' attention and prolong engagement with the intervention. Participants in one study from the UK, women who had smoked during a previous pregnancy and receive SMS support for smoking cessation, also felt that when the communication asked for a reply they would think more about the content of the messages they were receiving (Naughton 2013), as noted by one participant below. “An interactive text inviting a reply would make ‘you think more about the text message’ otherwise ‘you don't have to do anything with it so you read it and then forget about it’” (Naughton 2013).

Participants in a few studies, all from HIC contexts (Cornelius 2009; Munro 2017; Naughton 2013; Wright 2011), felt that bi‐directional communication or content that somehow engaged clients (e.g. using quizzes or replies) would be more useful than one‐way communication. Participants mentioned that for this to be the case, the interaction should be made convenient and allow for quick and simple responses. Such interactive options would help keep participants' attention and prolong engagement with the intervention. Participants in one study from the UK, women who had smoked during a previous pregnancy and receive SMS support for smoking cessation, also felt that when the communication asked for a reply they would think more about the content of the messages they were receiving (Naughton 2013), as noted by one participant below. “An interactive text inviting a reply would make ‘you think more about the text message’ otherwise ‘you don't have to do anything with it so you read it and then forget about it’” (Naughton 2013). Privacy and confidentiality regarding personal health information ('Summary of qualitative findings' table for Findings 18 to 19 is shown in Table 6.)Table 6 'Summary of qualitative findings' table for findings related to privacy and confidentiality regarding personal health information

“An interactive text inviting a reply would make ‘you think more about the text message’ otherwise ‘you don't have to do anything with it so you read it and then forget about it’” (Naughton 2013). Privacy and confidentiality regarding personal health information ('Summary of qualitative findings' table for Findings 18 to 19 is shown in Table 6.)Table 6 'Summary of qualitative findings' table for findings related to privacy and confidentiality regarding personal health information Finding Overall CERQual assessment Explanation for assessment Contributing studies 18 Some participants with health issues that are often seen as stigmatised or very personal (e.g. HIV, family planning, and abortion care) worried that their confidential health information would be disclosed or their identity traced due to their participation in digital targeted client communication. In general, people’s perceptions of information delivery channels (SMS, interactive voice response, voice call) were influenced by how confidential they felt the delivery channels to be. High confidence Due to minor concerns regarding methodological limitations Akinfaderin‐Agarau 2012; Calderón 2017; Cates 2015; Curioso 2009; Evans 2016; French 2016; Goldenberg 2015; Greaney 2014; Jalloh‐Vos 2014; Jennings 2013; Mbuagbaw 2012; Mbuagbaw 2014; Menacho 2013; Mitchell 2016; Nachega 2016; Odeny 2014; Perry 2012; Rana 2015; Rodrigues 2015; Smith 2017; Willoughby 2017 19 Some participants proposed strategies to address their concerns regarding confidentiality and privacy. These strategies for communication included neutral, coded, or discreet language; access codes; communication that does not disclose the sender; coming from a trusted sender; and the ability to tailor and control content, timing, and frequency of their messages. High confidence Due to minor concerns regarding methodological limitations Calderón 2017; Curioso 2009; Evans 2016; French 2016; Goldenberg 2015; Greaney 2014; Mbuagbaw 2012; Menacho 2013; Odeny 2014; Rana 2015; Rodrigues 2015; Smith 2017; Willoughby 2017

or and control content, timing, and frequency of their messages. High confidence Due to minor concerns regarding methodological limitations Calderón 2017; Curioso 2009; Evans 2016; French 2016; Goldenberg 2015; Greaney 2014; Mbuagbaw 2012; Menacho 2013; Odeny 2014; Rana 2015; Rodrigues 2015; Smith 2017; Willoughby 2017 Below we present findings specifically related to privacy and confidentiality. However, this theme is also touched upon in a number of other findings including Finding 7. Finding 18: Some participants with health issues that are often seen as stigmatised or very personal (e.g. HIV, family planning, and abortion care) worried that their confidential health information would be disclosed or their identity traced due to their participation in DTCC. In general, people’s perceptions of information delivery channels (SMS, interactive voice response, voice call) were influenced by how confidential they felt the delivery channels to be (high confidence).

eir confidential health information would be disclosed or their identity traced due to their participation in DTCC. In general, people’s perceptions of information delivery channels (SMS, interactive voice response, voice call) were influenced by how confidential they felt the delivery channels to be (high confidence). A number of studies, the majority from LMIC settings (Akinfaderin‐Agarau 2012; Calderón 2017; Cates 2015; Curioso 2009; Evans 2016; French 2016; Goldenberg 2015; Greaney 2014; Jalloh‐Vos 2014; Jennings 2013; Mbuagbaw 2012; Mbuagbaw 2014; Menacho 2013; Mitchell 2016; Nachega 2016; Odeny 2014; Perry 2012; Rana 2015; Rodrigues 2015; Smith 2017; Willoughby 2017), found that some participants had concerns about the extent to which their privacy and personal information were sufficiently protected. This was especially true for those who were dealing with health conditions that are often seen as stigmatised or very personal. People participating in interventions related to HIV and AIDS expressed the strongest concerns, especially for interventions that sent HIV testing reminders or reminders to take medication (Akinfaderin‐Agarau 2012; Curioso 2009; Evans 2016; Goldenberg 2015; Jennings 2013; Mbuagbaw 2012; Mbuagbaw 2014; Menacho 2013; Nachega 2016; Odeny 2014; Rana 2015; Rodrigues 2015). These participants worried that the SMS or phone conversation would reveal their status to people who picked up their phones or who overheard their conversations. However, in one study from Kenya, participants felt that receiving an SMS or phone call would protect their privacy more than a face‐to‐face appointment, as conversations in a health facility were easily overheard by others. Some participants in a family planning digital communication study in Sierra Leone feared that participating in the intervention would compromise their privacy because their husbands or family members would find out they were using family planning methods (Jalloh‐Vos 2014). Similarly, a participant in a postabortion care intervention in Cambodia was worried that her medical history of abortion would be discovered by her family if she was to receive a phone call in their presence or if someone else were to answer her phone (Smith 2017).

were using family planning methods (Jalloh‐Vos 2014). Similarly, a participant in a postabortion care intervention in Cambodia was worried that her medical history of abortion would be discovered by her family if she was to receive a phone call in their presence or if someone else were to answer her phone (Smith 2017). Some adolescents participating in interventions related to sexual and reproductive health felt that if others saw the messages it might be embarrassing; it may cause their parents to ask them or they may be suspected of having a disease (Akinfaderin‐Agarau 2012; Perry 2012; Willoughby 2017). Some participants had more general worries that their private information, such as banking details and personal health information, would be disclosed or shared (Calderón 2017; Goldenberg 2015; Mitchell 2016). Participants in a few studies expressed opinions on how confidential they felt different delivery channels were. Some believed that texting provided more privacy than receiving a voice call (Cates 2015; Curioso 2009; French 2016; Menacho 2013; Perry 2012). Some participants felt that text messages were easier, more confidential, and more readily available (Curioso 2009). Others felt that they had more control over text messaging as they could prevent the messages from appearing on their phone screens, could lock their phones, and could delete messages (French 2016).

12). Some participants felt that text messages were easier, more confidential, and more readily available (Curioso 2009). Others felt that they had more control over text messaging as they could prevent the messages from appearing on their phone screens, could lock their phones, and could delete messages (French 2016). Finding 19: Some participants proposed strategies to address their concerns regarding confidentiality and privacy. These strategies for communication included neutral, coded, or discreet language; access codes; communication that does not disclose the sender; coming from a trusted sender; and the ability to tailor and control content, timing, and frequency of their messages (high confidence).

regarding confidentiality and privacy. These strategies for communication included neutral, coded, or discreet language; access codes; communication that does not disclose the sender; coming from a trusted sender; and the ability to tailor and control content, timing, and frequency of their messages (high confidence). In some studies, many from LMIC settings (Calderón 2017; Curioso 2009; Evans 2016; French 2016; Greaney 2014; Goldenberg 2015; Mbuagbaw 2012; Menacho 2013; Odeny 2014; Rana 2015; Rodrigues 2015; Smith 2017; Willoughby 2017), participants presented strategies that could be implemented to address their concerns regarding privacy and confidentiality. Some participants felt that the best way to protect their privacy would be to use neutral or coded language (Curioso 2009; Goldenberg 2015; Mbuagbaw 2012; Menacho 2013; Odeny 2014; Rana 2015; Willoughby 2017). This was especially true for those receiving messages about sensitive topics such as HIV and family planning. Others felt that privacy could be protected by not disclosing the sender of the message and making sure the message was sent from a trusted source (Evans 2016; Greaney 2014). Some participants suggested that digital health interventions use access codes or passwords that participants would have to enter to gain access to messages (Calderón 2017; French 2016; Goldenberg 2015; Rana 2015). Finally, many participants believed that some of their concerns about privacy could be addressed by allowing them to tailor and control various aspects of the interventions such as the content, frequency, and timing of their messages (Goldenberg 2015; Rodrigues 2015; Smith 2017).

ench 2016; Goldenberg 2015; Rana 2015). Finally, many participants believed that some of their concerns about privacy could be addressed by allowing them to tailor and control various aspects of the interventions such as the content, frequency, and timing of their messages (Goldenberg 2015; Rodrigues 2015; Smith 2017). Perceptions of intervention impact ('Summary of qualitative findings' table for Findings 20 to 25 is shown in Table 7.)Table 7 'Summary of qualitative findings' table for findings related to perceptions of programme impact

ench 2016; Goldenberg 2015; Rana 2015). Finally, many participants believed that some of their concerns about privacy could be addressed by allowing them to tailor and control various aspects of the interventions such as the content, frequency, and timing of their messages (Goldenberg 2015; Rodrigues 2015; Smith 2017). Perceptions of intervention impact ('Summary of qualitative findings' table for Findings 20 to 25 is shown in Table 7.)Table 7 'Summary of qualitative findings' table for findings related to perceptions of programme impact Finding Overall CERQual assessment Explanation for assessment Contributing studies 20 Some participants thought that participating in digital targeted client communication had influenced their behaviour, whilst others did not. Reasons given for the changes in behaviour included receiving new knowledge; receiving strategies on how to initiate discussion with a partner or healthcare provider; being motivated or reassured by the intervention; and being reminded, for example, to take medication or make an appointment. Some participants who believed that the intervention did not have any influence on their behaviour found that the digital health interventions were not relevant to them. Low confidence Due to minor concerns regarding relevance and adequacy and moderate concerns regarding methodological limitations Brown 2014; Entsieh 2015; French 2016; Gold 2010; Greaney 2014; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Jennings 2013; Lau 2014; Missal 2016; Munro 2017; Rodrigues 2015; Sloan 2017; Smillie 2014; Smith 2017; Ware 2016 21 Some participants suggested that the effects of the messaging may not be sustained over time, as they and others would become bored with or fatigued by the messages, especially if the content was not varied enough. Low confidence Due to moderate concerns regarding relevance and adequacy Cornelius 2009; Curioso 2009; Evans 2016; Gold 2010; Menacho 2013; Mitchell 2016; Rana 2015; Willoughby 2017 22 Some participants were concerned about becoming over‐reliant on digital reminders and thought that this might make them dependent on digital targeted communication for undertaking some health tasks. They were concerned that in the absence of these reminders they would adhere poorly to care plans. Low confidence Due to minor concerns regarding methodological limitations, moderate concerns about relevance, and serious concerns about adequacy Jalloh‐Vos 2014; Mbuagbaw 2012; Rana 2015 23 Some participants felt that digital health interventions could save them time and money by giving them access to health care via their mobile phones.

o minor concerns regarding methodological limitations, moderate concerns about relevance, and serious concerns about adequacy Jalloh‐Vos 2014; Mbuagbaw 2012; Rana 2015 23 Some participants felt that digital health interventions could save them time and money by giving them access to health care via their mobile phones. This was especially relevant to participants who faced barriers in attending health care because of distance to a health facility and a lack of time and or financial means. Low confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding adequacy and relevance Calderón 2017; Smith 2017 24 Some participants felt that digital health interventions provided them with feelings of support and connectedness, as they felt that someone was taking the time to send them messages. A few participants felt that in some cases the sense of caring and support that they received from healthcare providers through digital health interventions had a positive influence on their relationship with their healthcare provider. Moderate confidence Due to moderate concerns regarding methodological limitations and relevance Brown 2014; Calderón 2017; Entsieh 2015; Jalloh‐Vos 2014; Lau 2014; Mbuagbaw 2014; Munro 2017; Nachega 2016; Rana 2015; Rodrigues 2015; Sloan 2017; Smillie 2014; Smith 2017; Ware 2016; Wright 2011 25 Participants described how they shared digital communication content more broadly with friends, family, and community members. Many participants felt that the information would be useful to others. Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Calderón 2017; Cornelius 2009; Flax 2017; French 2016; Gold 2010; Jennings 2013; Perry 2012; Smith 2017; Wright 2011

ty members. Many participants felt that the information would be useful to others. Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Calderón 2017; Cornelius 2009; Flax 2017; French 2016; Gold 2010; Jennings 2013; Perry 2012; Smith 2017; Wright 2011 Finding 20: Some participants thought that participating in DTCC had influenced their behaviour, while others did not. Reasons given for the changes in behaviour included receiving new knowledge; receiving strategies on how to initiate discussion with a partner or healthcare provider; being motivated or reassured by the intervention; and being reminded, for example, to take medication or make an appointment. Some participants who believed that the intervention did not have any influence on their behaviour found that the digital health interventions were not relevant to them (low confidence). A number of studies found that participants thought that taking part in DTCC had influenced their behaviour (Brown 2014; Entsieh 2015; French 2016; Gold 2010; Greaney 2014; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Jennings 2013; Lau 2014; Missal 2016; Munro 2017; Rodrigues 2015; Sloan 2017; Smillie 2014; Smith 2017; Ware 2016), while others felt that it had no impact.

ght that taking part in DTCC had influenced their behaviour (Brown 2014; Entsieh 2015; French 2016; Gold 2010; Greaney 2014; Hirsch‐Moverman 2017; Jalloh‐Vos 2014; Jennings 2013; Lau 2014; Missal 2016; Munro 2017; Rodrigues 2015; Sloan 2017; Smillie 2014; Smith 2017; Ware 2016), while others felt that it had no impact. In some studies participants felt that the targeted client communication had had a positive impact on their behaviour. New knowledge about child and maternal health such as vaccinations and breastfeeding and antenatal care, Brown 2014; Entsieh 2015, risk of sexually transmitted infections (Gold 2010), the husband’s role and how he can support his wife during pregnancy and delivery (Missal 2016), and family planning methods (Smith 2017), had influenced participant actions, including for testing or choosing a particular contraceptive method. Digital targeted client communication also provided some participants with specific strategies, for example how to discuss a health topic with a significant other. Some studies found that this led to participants talking to their partners about testing for sexually transmitted diseases (French 2016; Jennings 2013); preparing for labour and delivery (Missal 2016); or discussing questions with a healthcare provider (Munro 2017). Participants felt that without these strategies they might not have taken the steps to have these conversations about their health.

ut testing for sexually transmitted diseases (French 2016; Jennings 2013); preparing for labour and delivery (Missal 2016); or discussing questions with a healthcare provider (Munro 2017). Participants felt that without these strategies they might not have taken the steps to have these conversations about their health. Messages also motivated or reassured participants about their health decisions. Some participants felt a sense of confidence in their parenting decisions, as the messages they received validated their parenting choices (Brown 2014); or made them feel like the difficult conversations about disease testing were the right thing to do (French 2016; Jennings 2013); or motivated them to quit smoking (Sloan 2017), or take their medications on time (Smillie 2014). Messages also made participants feel less apprehensive about procedures such as sexually transmitted infection testing (Gold 2010), or continuing contraceptive use after side effects occurred (Smith 2017). In other cases the messages motivated women to visit the health facility more frequently (Jalloh‐Vos 2014), or husbands to support their wives during pregnancy (Missal 2016).

procedures such as sexually transmitted infection testing (Gold 2010), or continuing contraceptive use after side effects occurred (Smith 2017). In other cases the messages motivated women to visit the health facility more frequently (Jalloh‐Vos 2014), or husbands to support their wives during pregnancy (Missal 2016). Some participants felt that the reminders delivered through digital health interventions helped them remember to take their medication (Hirsch‐Moverman 2017; Smith 2017; Rodrigues 2015; Ware 2016); get tested (French 2016; Jennings 2013); improve their use of condoms (French 2016); adhere to or engage in treatment (Jennings 2013; Lau 2014); schedule or attend appointments (Greaney 2014; Jalloh‐Vos 2014; Smillie 2014); or by just keeping sexual health “up in their mind” (Gold 2010). In some studies (Gold 2010; Greaney 2014; Hirsch‐Moverman 2017; Rodrigues 2015; Smillie 2014,), participants believed that the intervention had no impact on their behaviour, whether it be remembering to take medications (Hirsch‐Moverman 2017; Rodrigues 2015; Smillie 2014); testing for sexually transmitted infections and using condoms (Gold 2010); or scheduling an appointment for screening (Greaney 2014). Finding 21: Some participants suggested that the effects of the messaging may not be sustained over time, as they and others would become bored with or fatigued by the messages, especially if the content was not varied enough (low confidence).

In some studies (Gold 2010; Greaney 2014; Hirsch‐Moverman 2017; Rodrigues 2015; Smillie 2014,), participants believed that the intervention had no impact on their behaviour, whether it be remembering to take medications (Hirsch‐Moverman 2017; Rodrigues 2015; Smillie 2014); testing for sexually transmitted infections and using condoms (Gold 2010); or scheduling an appointment for screening (Greaney 2014). Finding 21: Some participants suggested that the effects of the messaging may not be sustained over time, as they and others would become bored with or fatigued by the messages, especially if the content was not varied enough (low confidence). Some studies found that some participants thought the effects of digital health messaging may not be sustainable, as they or other users might become fatigued by the messages (Curioso 2009; Cornelius 2009; Evans 2016; Gold 2010; Menacho 2013; Mitchell 2016; Rana 2015; Willoughby 2017). Participants thought that this message fatigue would most likely occur if the frequency, content, and topics were not varied enough (Evans 2016; Gold 2010; Menacho 2013). Finding 22: Some participants were concerned about becoming over‐reliant on digital reminders and thought that this might make them dependent on digital targeted communication for undertaking some health tasks. They were concerned that, in the absence of these reminders, they would adhere poorly to care plans (low confidence).

22: Some participants were concerned about becoming over‐reliant on digital reminders and thought that this might make them dependent on digital targeted communication for undertaking some health tasks. They were concerned that, in the absence of these reminders, they would adhere poorly to care plans (low confidence). A few studies, from LMIC settings in Africa (Jalloh‐Vos 2014; Mbuagbaw 2012; Rana 2015), found that some participants were concerned about over‐reliance on digital reminders, and that in their absence adherence might become problematic. Two studies explored perceptions related to adherence to HIV medication, and another family planning and antenatal/postpartum care. These concerns were important to participants that mentioned the problem of reliance, as they believed that the digital health interventions would eventually end and people needed to remember themselves (Rana 2015). For example, in Sierra Leone, both women and men mentioned that if the nurse did not call to remind them about family planning they would forget, and that it was the nurse's job to remind them to come (Jalloh‐Vos 2014). Finding 23: Some participants felt that digital health interventions could save them time and money by giving them access to health care via their mobile phones. This was especially relevant to participants who faced barriers in attending health care because of distance to a health facility and a lack of time and or financial means (low confidence).

elt that digital health interventions could save them time and money by giving them access to health care via their mobile phones. This was especially relevant to participants who faced barriers in attending health care because of distance to a health facility and a lack of time and or financial means (low confidence). Two studies from LMIC settings found that participants believed that digital health interventions could potentially save them time and money by giving them access to health care through their mobile phones instead of having to go to the health facility (Calderón 2017; Smith 2017). In Peru (Calderón 2017), women over 18 who had at least one child believed that this would be useful to families living in resource‐poor communities, where families often cannot afford the cost or time related to transport to the health facility or the consultation. They felt that the intervention would be used more by families that live further away from a health facility, described as follows. “People like me who work all day could use it, we have to use it, because they then already have someone to ask, you will not be with that doubt ‘do I give or not give the medication to the baby?’ or ‘how much medication do I have to give him/her?’ I don’t know, so when we have someone to call or to send text messages to, you will not hesitate to do it as they say” (Calderón 2017).

, because they then already have someone to ask, you will not be with that doubt ‘do I give or not give the medication to the baby?’ or ‘how much medication do I have to give him/her?’ I don’t know, so when we have someone to call or to send text messages to, you will not hesitate to do it as they say” (Calderón 2017). Women in Cambodia also felt that the digital health intervention saved them time and money, as they could receive their family planning counselling over the phone instead of having to travel to the health facility. This saved them both the fees associated with transport and the consultation with the healthcare provider (Smith 2017). Finding 24: Some participants felt that digital health interventions provided them with feelings of support and connectedness, as they felt that someone was taking the time to send them messages. A few participants felt that in some cases the sense of caring and support that they received from healthcare providers through digital health interventions had a positive influence on their relationship with their healthcare provider (moderate confidence).

y felt that someone was taking the time to send them messages. A few participants felt that in some cases the sense of caring and support that they received from healthcare providers through digital health interventions had a positive influence on their relationship with their healthcare provider (moderate confidence). A number of studies, the majority from LMIC settings (Brown 2014; Calderón 2017; Entsieh 2015; Jalloh‐Vos 2014; Lau 2014; Mbuagbaw 2014; Munro 2017; Nachega 2016; Rana 2015; Rodrigues 2015; Sloan 2017; Smillie 2014; Smith 2017; Ware 2016; Wright 2011), presented data related to participants’ reflections around the caring and supportive nature of DTCC. In some cases, the participants in these studies felt that someone was interested in their situation, was invested in their well‐being, and cared about them. This led some participants to feel encouraged and to have increased self‐confidence and feelings of self‐worth. For others, the messages provided support, guidance, and information, often giving a sense of direction, reassurance, and motivation to participants. Support was presented in various ways, for example to take medications or providing counselling. In some cases, this increased dialogue between healthcare workers and participants had a positive influence on their relationship, for example by making each group more aware of the other’s expectations (Jalloh‐Vos 2014; Smillie 2014; Smith 2017).

various ways, for example to take medications or providing counselling. In some cases, this increased dialogue between healthcare workers and participants had a positive influence on their relationship, for example by making each group more aware of the other’s expectations (Jalloh‐Vos 2014; Smillie 2014; Smith 2017). Finding 25: Participants described how they shared digital communication content more broadly with friends, family, and community members. Many participants felt that the information would be useful to others (moderate confidence). Some studies found that participants shared messages they had received or thought they would share messages with friends, family, colleagues, or neighbours (Calderón 2017; Cornelius 2009; Flax 2017; French 2016; Jennings 2013; Perry 2012; Smith 2017; Wright 2011). Many participants felt that it was important to share this information, as they had found the messages helpful and believed the information would be useful to others (Calderón 2017; Flax 2017; French 2016; Gold 2010; Jennings 2013; Perry 2012; Smith 2017). Participants in one study felt that sharing messages and experiences could help to create a sense of community (Wright 2011). In another study, participants were excited about the possibility of creating message content to share with friends and family (Cornelius 2009).

Some studies found that participants shared messages they had received or thought they would share messages with friends, family, colleagues, or neighbours (Calderón 2017; Cornelius 2009; Flax 2017; French 2016; Jennings 2013; Perry 2012; Smith 2017; Wright 2011). Many participants felt that it was important to share this information, as they had found the messages helpful and believed the information would be useful to others (Calderón 2017; Flax 2017; French 2016; Gold 2010; Jennings 2013; Perry 2012; Smith 2017). Participants in one study felt that sharing messages and experiences could help to create a sense of community (Wright 2011). In another study, participants were excited about the possibility of creating message content to share with friends and family (Cornelius 2009). Finally, one study sent SMSes to pre‐existing community‐based women’s groups about maternal and child health topics. Not all of the women in the group were pregnant or had children. However, these participants mentioned that they still found the content interesting and discussed it with other group members, family, and neighbours (Flax 2017). Supplementing the Cochrane Reviews of effectiveness with synthesised qualitative findings – matrix results As described in the Methods section, we used a matrix approach to explore whether potential implementation barriers that we had identified in our synthesis (see Table 8 below) had been addressed in the programmes evaluated in the related Cochrane Reviews of effectiveness (Palmer Ongoing a; Palmer Ongoing).

trix results As described in the Methods section, we used a matrix approach to explore whether potential implementation barriers that we had identified in our synthesis (see Table 8 below) had been addressed in the programmes evaluated in the related Cochrane Reviews of effectiveness (Palmer Ongoing a; Palmer Ongoing). The relevance of our synthesis findings to the reviews of effectiveness was strengthened by the fact that the studies included in the qualitative synthesis and those included in the effectiveness reviews came from similar settings. Around half of the studies in both the synthesis and the reviews were from low‐income countries (primarily African countries), whilst the remaining studies where from high‐income countries (primarily the USA). In addition, around 10 of the qualitative studies included in our synthesis appear to have been carried out to inform the development of some of the trials. Table 8 presents an overview of our matrix assessment (a more detailed version can be found in Appendix 5, and a list of all of the studies included in the analysis is shown in Appendix 6). In summary, we found that some of the included trials did describe efforts to address some of the potential barriers to implementation that we had identified in the qualitative synthesis. However, most of the trials referred to only a small number of barriers, and in some trials potential participants were actively excluded if they were dealing with these barriers.

trials did describe efforts to address some of the potential barriers to implementation that we had identified in the qualitative synthesis. However, most of the trials referred to only a small number of barriers, and in some trials potential participants were actively excluded if they were dealing with these barriers. One barrier identified in our synthesis was related to situations where the target group did not own a functioning mobile phone. In more than half of the trials, trialists dealt with this barrier by making mobile phone ownership a condition for trial participation. Between 0.3% and 63% of eligible participants were excluded from trial participation because they did not own a phone or could not receive text messages. Some trialists did attempt to find solutions to this problem, however. For instance, some trialists only required that participants had access to a phone, for instance through family, friends, or neighbours (Brown 2016; Gibson 2017; Kassaye 2016; Odeny 2014); in other trials participants were provided with phones (e.g. Cook 2015; Ingersoll 2015); and in one trial, local healthcare providers were given phones to share with women who had no access to phones (Joshi 2015).

, for instance through family, friends, or neighbours (Brown 2016; Gibson 2017; Kassaye 2016; Odeny 2014); in other trials participants were provided with phones (e.g. Cook 2015; Ingersoll 2015); and in one trial, local healthcare providers were given phones to share with women who had no access to phones (Joshi 2015). Other barriers identified in our synthesis were tied to poor access to electricity to charge phones and poor access to network services. However, trialists rarely referred to these issues. If mentioned, this was usually because access to specific networks or to stable electricity was a condition for participation in the trial. Trials that did attempt to deal with these barriers included one trial based in the USA (Ingersoll 2015). Here, assessments carried out prior to the trial showed that many people had inconsistent cellular or internet service, but that they could usually receive text messages. The trialists therefore built a texting system rather than one that would require a consistent cellular signal or internet access (Ingersoll 2015). Another exception was a Kenya‐based trial (Pop‐Eleches 2011). As many participants in this trial needed to charge their phones at fee‐based charging stations in nearby markets, the trialists provided financial support to cover these fees (Pop‐Eleches 2011).

onsistent cellular signal or internet access (Ingersoll 2015). Another exception was a Kenya‐based trial (Pop‐Eleches 2011). As many participants in this trial needed to charge their phones at fee‐based charging stations in nearby markets, the trialists provided financial support to cover these fees (Pop‐Eleches 2011). Another barrier was tied to the expenses associated with the trial interventions. This was also confirmed as a problem by some trialists, who described poor participation because of cost barriers (Ahlers‐Schmidt 2012; Lester 2010; Norton 2014). Around one‐third of the trials described efforts to address this problem in some way, for instance by offering vouchers to participants (e.g. Lund 2012; Omole 2018), or by offering participants free access to various internet sites in return for receiving mobile advertising (Gold 2011).

2012; Lester 2010; Norton 2014). Around one‐third of the trials described efforts to address this problem in some way, for instance by offering vouchers to participants (e.g. Lund 2012; Omole 2018), or by offering participants free access to various internet sites in return for receiving mobile advertising (Gold 2011). Our synthesis also pointed to problems in receiving messages when people changed their phone numbers or sim cards. Again, this issue was rarely referred to in the trials. When mentioned, this was usually as an explanation for participant loss to follow‐up or as a condition for trial participation. Examples of trials that did try to address this issue include a trial from India, where participants with new phones were able to update their contact details with their local health worker (Joshi 2015); a USA‐based trial (Bull 2016), where participants were asked to update information about their phone number every month (Devine 2014); and a trial from Ecuador, where trialists gathered several different contact avenues to maximise their ability to contact participants (Maslowsky 2016).

ocal health worker (Joshi 2015); a USA‐based trial (Bull 2016), where participants were asked to update information about their phone number every month (Devine 2014); and a trial from Ecuador, where trialists gathered several different contact avenues to maximise their ability to contact participants (Maslowsky 2016). Two related barriers were tied to situations where people had to share phones or had their phones controlled by others, and situations where these people, as well as people with their own phones, were concerned about privacy issues when receiving messages. Very few trialists discussed situations whereby people’s phones might be controlled by others, and at least two‐thirds did not refer to privacy issues. However, several trialists did describe efforts to address issues of privacy and confidentiality. These included trials targeting people about oral contraception (Castano 2012), or HIV treatment and prevention (Garofalo 2016; Mbuagbaw 2012; Ybarra 2016) where trialists ensured that messages did not refer to the recipient’s name, their health status, or the name of the medication; encouraged recipients to delete messages; and gave them information about privacy settings on their phones.

2012), or HIV treatment and prevention (Garofalo 2016; Mbuagbaw 2012; Ybarra 2016) where trialists ensured that messages did not refer to the recipient’s name, their health status, or the name of the medication; encouraged recipients to delete messages; and gave them information about privacy settings on their phones. Other barriers identified in our synthesis were created by language problems, low literacy, or limited techno‐literacy. At least a third of the trials made no reference to these issues. Another third actively excluded people who did not speak one or two mainstream languages, and several trials also excluded people who could not read these languages or who were illiterate. Some trialists described efforts to overcome these problems, including designing text messages so that they scored low on readability scales ( Bigna 2014; Stockwell 2014); offering text messages in the local language (Kassaye 2016); or sending voice messages or pictorial messages instead of text messages (de Costa 2012; Smith 2015). Our synthesis also found that people’s trust in and perceptions of the information they received was influenced by their perceptions of the sender. Attempts to address this issue were rarely described by trialists. Exceptions included a UK‐based trial, Cooper 2015, that highlighted in advertising materials that the intervention was supported by the National Health Service (NHS) (Emery 2018); and a USA‐based trial (Bull 2016), where messages were linked to celebrities that the target audience said were most interesting to them (Devine 2014).

included a UK‐based trial, Cooper 2015, that highlighted in advertising materials that the intervention was supported by the National Health Service (NHS) (Emery 2018); and a USA‐based trial (Bull 2016), where messages were linked to celebrities that the target audience said were most interesting to them (Devine 2014). The final issue we assessed in our matrix was the extent to which the target group had been given an opportunity to offer feedback about their needs, preferences, and experiences regarding the intervention. More than half of the trialists described collecting user feedback to develop or improve the intervention, although it was not always clear how information gathered before, during, or after the trial had influenced the intervention or would influence future versions of the intervention. Discussion Summary of main results The acceptability of digital targeted client communication (DTCC) is mixed. Some clients described DTCC in positive terms, as providing them with support and connectedness; giving them a feeling that someone cares about them; and, in some cases, having a positive influence on their relationship with their healthcare provider. However, clients who are dealing with health conditions that are often stigmatised or very personal (e.g. HIV, family planning, and abortion care) worry that their confidential health information will be disclosed or their identity traced.

aving a positive influence on their relationship with their healthcare provider. However, clients who are dealing with health conditions that are often stigmatised or very personal (e.g. HIV, family planning, and abortion care) worry that their confidential health information will be disclosed or their identity traced. Clients' perceptions and experiences of DTCC can be influenced by a number of factors. Participants believed that there should be little or no charge to participate in digital health interventions. They wanted messages that were from a trusted sender and that were polite and encouraging. They did not like feeling pressured, lectured to, or frightened. They wanted varied information that arrived at a time and frequency that was convenient for them. Content preferences included new knowledge, reminders, solutions, and suggestions about health issues presented in a clear, short, and personalised way. In general, barriers to participating in digital health interventions included problems with network connectivity, access to electricity, device usability, and issues tied to data confidentiality and privacy.

Clients' perceptions and experiences of DTCC can be influenced by a number of factors. Participants believed that there should be little or no charge to participate in digital health interventions. They wanted messages that were from a trusted sender and that were polite and encouraging. They did not like feeling pressured, lectured to, or frightened. They wanted varied information that arrived at a time and frequency that was convenient for them. Content preferences included new knowledge, reminders, solutions, and suggestions about health issues presented in a clear, short, and personalised way. In general, barriers to participating in digital health interventions included problems with network connectivity, access to electricity, device usability, and issues tied to data confidentiality and privacy. Some of our findings addressed issues related to gender, equity, and human rights. For example, access to healthcare services via digital devices may be particularly helpful to clients with caring or work responsibilities; clients who live far from health facilities; and poorer people. However, access to digital health interventions may be particularly difficult for others who speak minority languages or who have low literacy skills or low digital literacy skills, who for example do not know how to open, read, or send a text message. Participation may also be difficult for clients with poor access to network services, electricity, or ownership of mobile phones. For clients, particularly women and adolescents, who have to share or borrow a phone or who have phone access controlled by others, it can be difficult to receive messages or keep them private. Clients with stigmatised health conditions such as HIV were also concerned about what would happen if their privacy was not protected, for example if the fact that they were participating in the digital health intervention led to the disclosure of their HIV status.

be difficult to receive messages or keep them private. Clients with stigmatised health conditions such as HIV were also concerned about what would happen if their privacy was not protected, for example if the fact that they were participating in the digital health intervention led to the disclosure of their HIV status. Overall completeness and applicability of evidence The sampling approach we used in this review (see above) aimed to achieve a maximum variation of target populations, settings, delivery mechanisms, and content focus of the targeted digital health intervention. We found studies that represented clients from all of the client groups included in the scope of this synthesis. However, studies from parents and pregnant and postpartum women and their partners were not as common. For example, we found very few studies researching DTCC for HIV‐positive pregnant and postpartum women. We are therefore less certain whether these client groups have the same perspectives and expectations of targeted digital health interventions. A majority of the studies covered topics related to sexual and reproductive health for young people and adults, with a focus on sexually transmitted infections and specifically HIV/AIDS. We are therefore less certain about whether participants receiving digital health interventions related to other topics have the same preferences for information and intervention delivery, such as vaccination and general child health.

nd adults, with a focus on sexually transmitted infections and specifically HIV/AIDS. We are therefore less certain about whether participants receiving digital health interventions related to other topics have the same preferences for information and intervention delivery, such as vaccination and general child health. The studies included in this review come from a wide variety of contexts and settings. There are a number of included studies from poorly resourced healthcare systems as well as high‐income settings. The range of settings included in this review highlighted access issues as well as issues related to gender and equity across all settings. The majority of the studies included in this synthesis looked at digital health interventions where the content was or would be delivered by SMS. We have only a few studies that explore perceptions related to other delivery methods such as interactive voice response (IVR). The data collection methods and study designs of the included studies may in some cases limit the applicability and completeness of the data reported. One example of this is the large number of studies that presented hypothetical digital health interventions (18 studies), both with and without examples of content, and asked participants to reflect on their preferences if they would participate. It is unclear whether these preferences based on hypothetical thinking would remain the same once the participant had actually experienced participating in the intervention and receiving the messages.

with and without examples of content, and asked participants to reflect on their preferences if they would participate. It is unclear whether these preferences based on hypothetical thinking would remain the same once the participant had actually experienced participating in the intervention and receiving the messages. All of the included studies made use of individual or group interviews and focus group discussions as their main method of data collection. None used long‐term ethnographic methods or field observation. While interviews and focus groups allow researchers to collect data on what people say, observational methods also allow researchers to collect data on what people do and why. This would have been appropriate for understanding how clients engaged with and used the various digital health interventions and allowed researchers to compare actions with interview and focus group data. Interviews and focus group discussions seem to be the most commonly used research methods amongst qualitative researchers exploring issues related to health. This could be because they are less time‐consuming than longer‐term ethnographic methods.

owed researchers to compare actions with interview and focus group data. Interviews and focus group discussions seem to be the most commonly used research methods amongst qualitative researchers exploring issues related to health. This could be because they are less time‐consuming than longer‐term ethnographic methods. Agreements and disagreements with other studies or reviews The review team identified one published qualitative evidence synthesis that had a similar scope to this review (O’Connor 2016). The O'Connor synthesis had a much broader scope and explored factors affecting client engagement and recruitment to digital health interventions. Also, these interventions did not have to be targeted at a specific audience and were open to any health intervention delivered by a digital technology. We are aware of an ongoing related qualitative evidence synthesis considering healthcare workers’ experiences with and perceptions of targeted digital communication via mobile device (Odendaal 2015). A number of reviews of the effect of mHealth programs in general have been carried out (Anglada‐Martinez 2015; Ahmed 2017; Aranda‐Jan 2014; Catalani 2013; Cole‐Lewis 2010; Free 2013; Free 2016; Gurman 2012; Krishna 2009; Lee 2016; Peiris 2014). Two overviews of systematic reviews looking at text messaging (Hall 2015), and the impact of mHealth interventions (Marcolino 2018), have also been done. However, many of these reviews and overviews address health issues beyond the scope of our synthesis, such as obesity and chronic illness (Bacigalupo 2013; Hamine 2015; Peiris 2014).

systematic reviews looking at text messaging (Hall 2015), and the impact of mHealth interventions (Marcolino 2018), have also been done. However, many of these reviews and overviews address health issues beyond the scope of our synthesis, such as obesity and chronic illness (Bacigalupo 2013; Hamine 2015; Peiris 2014). In 2015, a review was published that explored the adoption of mHealth in low resource environments (Chib 2015). This review found that the majority of studies in these environments concentrated principally on pilot projects focused mostly on the introduction and implementation of new interventions. This finding is similar to the types of studies we found in this synthesis, where approximately half of the included studies were projects using hypothetical examples to develop targeted digital health interventions. A further six studies were pilot projects for new targeted digital health interventions. We only identified one study that explored client experiences related to an ongoing project at scale, Mobile Midwife in Ghana (Entsieh 2015). Other systematic reviews have identified a similar large number of pilot studies (Catalani 2013; Gurman 2012). In the absence of studies of large‐scale implementation (where the targeted digital communication has become a routine part of care), it is difficult to say if people’s experiences and perceptions of smaller studies are transferable to interventions delivered at scale.

ber of pilot studies (Catalani 2013; Gurman 2012). In the absence of studies of large‐scale implementation (where the targeted digital communication has become a routine part of care), it is difficult to say if people’s experiences and perceptions of smaller studies are transferable to interventions delivered at scale. Our qualitative evidence synthesis highlighted issues of access to mobile phone technology, networks, and electricity. Other studies have highlighted this issue as well (Aranda‐Jan 2014; Bukachi 2007). As discussed in the findings of our matrix analysis, future trials should take these access issues into consideration when defining participant inclusion criteria. At this time, little is known about how targeted digital health communication interventions are used or perceived by those who do not have access to phones, networks, or electricity, as the large majority of existing trials and qualitative studies include participants who have phones and access to networks and electricity. Our findings highlight a potential lack of digital literacy among clients in many settings, which could also affect the feasibility and acceptance of targeted digital health communication interventions. Similarly, the O’Connor 2016 review concluded that "more investment is also needed to improve computer literacy and ensure technologies are accessible and affordable for those who wish to sign up to them”.

gs, which could also affect the feasibility and acceptance of targeted digital health communication interventions. Similarly, the O’Connor 2016 review concluded that "more investment is also needed to improve computer literacy and ensure technologies are accessible and affordable for those who wish to sign up to them”. Finally, issues related to tailoring and personalisation of timing, format, content, and privacy were clearly described in the findings of this review. One other review, Gurman 2012, mentions that less than half of the interventions included described targeting or tailoring the content. A meta‐analysis of tailored print health behaviour change interventions found that tailored interventions were more effective than non‐tailored interventions for health promotion (Noar 2007). The O’Connor 2016 review also found that interventions that are personalised when possible should be a focus when creating digital health interventions. To support this thinking, research has found that successful intervention design demands a user‐centred and iterative approach to developing new digital behaviour change interventions (Yardley 2016). Tailoring was also identified as a core interactive design feature in effective e‐health interventions (Morrison 2012).

To support this thinking, research has found that successful intervention design demands a user‐centred and iterative approach to developing new digital behaviour change interventions (Yardley 2016). Tailoring was also identified as a core interactive design feature in effective e‐health interventions (Morrison 2012). Summary of integrating the findings from this synthesis with the findings of relevant Cochrane effectiveness reviews We used a matrix approach to explore how the findings from our synthesis related to, or could help to inform, the findings of the two related Cochrane Reviews of effectiveness of DTCC (Palmer Ongoing a; Palmer Ongoing).

rating the findings from this synthesis with the findings of relevant Cochrane effectiveness reviews We used a matrix approach to explore how the findings from our synthesis related to, or could help to inform, the findings of the two related Cochrane Reviews of effectiveness of DTCC (Palmer Ongoing a; Palmer Ongoing). Our synthesis and the two intervention reviews were designed to complement each other, and used similar inclusion criteria where possible. The included studies in the synthesis and the two reviews were therefore broadly similar in terms of population groups, delivery mechanism, setting, and publication date. All three syntheses/reviews included studies with the same population groups (i.e. adolescents and adults that are users or potential users of reproductive health services; pregnant and postpartum women, including women living with HIV; and parents of children under five). In all three syntheses/reviews, the majority of programmes used text messages to communicate with their target audiences. Slightly more than half of the 35 studies in the synthesis were from LMICs. Slightly more than half of the 68 studies in the intervention reviews were also from LMICs. The studies sampled in our synthesis were published between 2009 and 2017, whilst the trials included in the intervention reviews were published between 2006 and 2018.

han half of the 35 studies in the synthesis were from LMICs. Slightly more than half of the 68 studies in the intervention reviews were also from LMICs. The studies sampled in our synthesis were published between 2009 and 2017, whilst the trials included in the intervention reviews were published between 2006 and 2018. Our matrix shows that most of the potential barriers to implementation raised by participants in the qualitative research were not referred to in most of the trials. It is possible that trialists did attempt to find solutions but did not report these in their publications. It is also possible that our search strategy failed to identify all relevant publications, and that these solutions were reported elsewhere. For instance, many of the trialists did not describe how they dealt with privacy issues when participants received messages in the papers we examined. However, it is possible that some privacy issues were addressed to some extent during scientific ethics review and approval processes.

solutions were reported elsewhere. For instance, many of the trialists did not describe how they dealt with privacy issues when participants received messages in the papers we examined. However, it is possible that some privacy issues were addressed to some extent during scientific ethics review and approval processes. Some of the included trials did describe efforts to address some of these issues. Where this occurred, it would be useful to explore the impact, transferability, and potential sustainability of these efforts. Some of the solutions described by trialists are likely to be sustainable, including efforts to increase accessibility of messages through audio, pictures, and local languages; or protecting people’s privacy though anonymising messages. However, other solutions may be less sustainable outside of a trial context or may be unacceptable for other reasons. These include the distribution of free phones or covering the costs of recharging phones. Other solutions, such as sharing of phones, may also lead to concerns about privacy.

hough anonymising messages. However, other solutions may be less sustainable outside of a trial context or may be unacceptable for other reasons. These include the distribution of free phones or covering the costs of recharging phones. Other solutions, such as sharing of phones, may also lead to concerns about privacy. In several trials, trialists simply excluded participants who were dealing with the barriers we identified in our synthesis. The exclusion of participants who do not own their own phones; who are likely to change their phone number or sim card; or because of language, literacy, or techno‐literacy issues is problematic. Whilst the size of these challenges varies from setting to setting, people that experience these challenges may be the same people who need health services the most, but who access them the least. Interventions that specifically aim to increase people’s access to and use of healthcare services should therefore make a particular effort to address these challenges.

om setting to setting, people that experience these challenges may be the same people who need health services the most, but who access them the least. Interventions that specifically aim to increase people’s access to and use of healthcare services should therefore make a particular effort to address these challenges. Table 8 below presents an overview of the findings of our matrix analysis. The full table can be found in Appendix 5. The table starts with 10 questions identified from the qualitative synthesis findings. The table then indicates how many studies from each of the client groups addressed the question (Y), or not (N); if the description was unclear (?) or if the topic of the question was not mentioned in the intervention study (NM). The table ends by displaying the total number of studies in numbers and percentages that addressed the question (Y), or not (N); if the description was unclear (?) or if the topic of the question was not mentioned in the intervention study (NM).

he topic of the question was not mentioned in the intervention study (NM). The table ends by displaying the total number of studies in numbers and percentages that addressed the question (Y), or not (N); if the description was unclear (?) or if the topic of the question was not mentioned in the intervention study (NM). Table 8: Integrating findings from this synthesis with the findings of relevant Cochrane effectiveness reviews Have the trialists described efforts to address situations where members of the target group: 1. do not own a functioning mobile device; 2. have poor access to network services; 3. have poor access to electricity to charge mobile devices; 4. want to avoid expenses associated with the intervention, such as paying for airtime; 5. change their phone numbers or sim cards; 6. have access to the phone controlled by someone else; 7. have low literacy, differing language skills, or limited techno‐literacy; 8. have concerns about privacy and confidentiality; 9. perceive different sources as more or less reliable, trusted, and credible; 10. have members of the target group been given an opportunity to offer feedback about their needs, preferences, and experiences regarding the intervention. 1 2 3 4 5 6 7 8 9 10 Adolescents (N = 13) Y 3 1 1 5 2 1 2 6 1 7 N 9 1 3 1 5 1 ? 1 2 1 2 3 NM 1 12 12 6 6 10 4 7 12 2 Adults (N = 27) Y 7 2 1 12 2 8 10 1 17 N 15 2 1 3 7 1 5 1 ? 1 5 2 1 1 NM 5 22 25 12 20 24 12 14 25 9 Pregnant and postpartum women (N = 11) Y 2 1 4 3 2 1 5 N 7 1 4 1 ? 3 2 NM 2 10 10 7 8 11 2 11 10 3 Parents (N = 14) Y 2 5 7 N 10 3 3 ? 1 1 1 NM 2 14 14 14 10 14 5 13 14 7 Mothers living with HIV (N = 3) Y 2 2 2 1 2 N ? 1 1 NM 1 3 3 1 3 2 1 1 3 1 Total for all population groups (N = 68) Y 16 (24%) 4 (5%) 2 (3%) 23 (35%) 5 (7%) 3 (4%) 19 (27%) 17 (25%) 3 (4%) 38 (57%) N 41 (60%) 2 (3%) 2 (3%) 4 (5%) 13 (19%) 2 (3%) 17 (25%) 1 (1%) 2 (3%) ? 1 (1%) 1 (1%) 3 (4%) 2 (3%) 11 (15%) 4 (5%) 1 (1%) 6 (8%) NM 11 (16%) 61 (91%) 64 (94%) 40 (59%) 47 (70%) 61 (90%) 24 (34%) 46 (69%) 64 (95%) 22 (32%) N = No; Y = Yes; ? = Unclear; NM = Not mentioned

5 (7%) 3 (4%) 19 (27%) 17 (25%) 3 (4%) 38 (57%) N 41 (60%) 2 (3%) 2 (3%) 4 (5%) 13 (19%) 2 (3%) 17 (25%) 1 (1%) 2 (3%) ? 1 (1%) 1 (1%) 3 (4%) 2 (3%) 11 (15%) 4 (5%) 1 (1%) 6 (8%) NM 11 (16%) 61 (91%) 64 (94%) 40 (59%) 47 (70%) 61 (90%) 24 (34%) 46 (69%) 64 (95%) 22 (32%) N = No; Y = Yes; ? = Unclear; NM = Not mentioned Review author reflexivity As part of the synthesis process, we reflected on how our backgrounds and positions might have influenced our choice of review topic, study selection, data extraction, analysis, and interpretation of data. Our backgrounds are in health systems research, social sciences, epidemiology, and nursing, and while working on the synthesis we were employed by government research institutions in Norway, South Africa and the USA (HA, CG, SL, NL, EA) and by the WHO (TT). The synthesis was commissioned to inform a WHO guideline, specifically to address guideline questions regarding the acceptability and feasibility of digital health interventions. Three of us were key members of the WHO guideline technical team (TT, SL, CG). Some of us had been involved in primary research related to digital health interventions, whilst others (CG, HA) had no previous work experience on this topic. Before working on the synthesis, our viewpoints regarding digital health interventions ranged from being neutral to these types of interventions to a slight skepticism of digital interventions as a magic bullet in solving health issues. All of us started the process believing that the implementation of digital health interventions should be informed by robust evidence of effectiveness, acceptability, and feasibility. Whilst working on the synthesis, we became more convinced of the importance of supporting evidence‐based decision‐making in digital health ‐ especially having seen from the studies included in this review, as well as from several other reviews commissioned for the WHO guideline, the range of challenges and constraints in implementing these interventions at scale, and in ways that protect the privacy of participants.

based decision‐making in digital health ‐ especially having seen from the studies included in this review, as well as from several other reviews commissioned for the WHO guideline, the range of challenges and constraints in implementing these interventions at scale, and in ways that protect the privacy of participants. Authors' conclusions Implications for practice The following questions, derived from our findings, may help health system or programme managers when implementing or planning for digital targeted client communication strategies to address issues of importance to their target population. It is important to consider local contextual factors including gender, age, cultural group, and education when implementing new digital targeted client communication strategies. Do clients own or have access to a functioning mobile device? If not, have solutions to access issues been considered? Do clients have access to network services in the area where they will be engaging with the digital health intervention? If not, have solutions to access issues been considered? Do clients have access to electricity to charge mobile devices? If not, have solutions to access issues been considered? Is participating in the digital health intervention free or of very limited cost to ensure that there are no barriers to participation? If not, have solutions to access issues been considered? Have solutions been considered for when clients change their phone numbers or sim cards in order to maintain intervention engagement and avoid losing contact?

Is participating in the digital health intervention free or of very limited cost to ensure that there are no barriers to participation? If not, have solutions to access issues been considered? Have solutions been considered for when clients change their phone numbers or sim cards in order to maintain intervention engagement and avoid losing contact? Have solutions been considered for when clients have their access to a phone controlled by someone else? Have solutions been considered for tailoring or changing intervention content to engage clients who have low literacy, differing language skills, or limited digital literacy? Have solutions been considered for tailoring or changing intervention content to ensure the privacy and confidentiality of clients and to avoid any harms that a break in this privacy may cause? Has an attempt been made to explore how clients perceive different sources of digital health interventions as more or less reliable, trusted, and credible? Has an attempt been made to use those sources that are perceived as trusted, reliable, and credible to send digital health messages? Have members of the client target group been given an opportunity to offer feedback about their needs, preferences, and experiences regarding the intervention during intervention development, implementation, and evaluation? Implications for research These implications have been derived from the CERQual assessment and the overview of the studies included in this review.

Have members of the client target group been given an opportunity to offer feedback about their needs, preferences, and experiences regarding the intervention during intervention development, implementation, and evaluation? Implications for research These implications have been derived from the CERQual assessment and the overview of the studies included in this review. There is a need for better reporting of context, sampling, methods, and researcher reflexivity in qualitative studies. Future qualitative studies should report their methods clearly and include reflection on the researchers' roles in the study and how this may have impacted on the process and results of the study. More detail concerning setting and participants is also needed to identify underlying cultural or social phenomena (shared values or beliefs) that mediate the influence of communications, as these need to be addressed when designing targeted digital health interventions. A better representation of the participant’s voice in the data in the studies included in this synthesis could have improved our confidence in some of the findings. For example, in some studies quotes were not labelled with a participant identifier, so we were unable to determine if the quotes came from multiple participants or the same participant.

OR programme OR programmes OR support OR supporting OR supported OR norm OR norms OR leader OR leaders OR leadership OR advocate OR advocated OR advocacy OR information OR action OR need OR needs OR influence OR influences OR influenced OR influencing OR participate OR participates OR participated OR participation ) )) OR All Fields:((cell OR cellular OR mobile) AND (phone OR phones OR telephone OR telephones OR technology OR technologies OR device OR devices)) OR smartphone OR smartphones OR smart‐phone OR smart‐phones OR cellphone OR cellphones OR mobiles OR mhealth OR m‐health OR "mobile health" OR ehealth OR e‐health OR "electronic health" OR telemedicine OR tele‐medicine OR telehealth OR tele‐health OR telecare OR tele‐care OR telenursing OR tele‐nursing OR telepsychiatry OR tele‐psychiatry OR telemonitor OR telemonitoring OR tele‐monitor OR tele‐monitoring OR teleconsult OR teleconsulting OR tele‐consult OR tele‐consulting OR telecounsel OR telecounseling OR tele‐counsel OR tele‐counseling OR telecoach OR telecoaching OR tele‐coach OR tele‐coaching OR videoconference OR videoconferences OR videoconferencing OR video‐conference OR video‐conferences OR video‐conferencingOR webcast OR webcasts OR webcasting OR web‐cast OR web‐casts OR web‐casting OR ((text OR texts OR texting OR short OR voice OR multimedia OR multi‐media OR electronic OR instant) AND (message OR messages OR messaging) OR instant messenger OR texting OR texted OR texter OR texters OR ((sms OR mms) AND (service OR services OR message OR messages OR messaging)) OR interactive voice response OR interactive voice responses OR ivr OR voice call OR voice calls OR callback OR voice over internet OR voip OR "mobile app" OR "mobile application" OR "mobile applications" OR (social AND (media OR network* OR networks OR networking)) OR ((medical OR clinical OR health OR healthcare OR nurse OR nurses OR nursing) AND informatics)) AND (qualitative OR interview OR interviews OR interviewing OR focus group OR focus groups OR questionnaire OR questionnaires OR ethnography OR ethnographic OR perception OR perceptions OR perceive OR perceives OR perceived OR opinion OR opinions OR attitude OR attitudes OR view OR views OR experience OR experiences OR experienced OR experiencing OR sceptic OR sceptical OR skeptic OR skeptical OR dilemma OR dilemmas OR "social mobilisation" OR "social mobilization" OR compliant OR compliance OR refuse OR refusal OR refused OR refusing OR feeling OR feelings OR impression OR impressions OR belief O

participant’s voice in the data in the studies included in this synthesis could have improved our confidence in some of the findings. For example, in some studies quotes were not labelled with a participant identifier, so we were unable to determine if the quotes came from multiple participants or the same participant. Research about digital targeted client communication should aim to include a broader spectrum of participants in relation to phone ownership, literacy, and ability to use a smartphone. Researchers could also focus on exploring why some digital health interventions do or do not influence participants' actions and behaviour. More research is needed on the public’s preferences around the details of timing, amount, and content of digital health interventions from people who have actually participated in digital health interventions. There is a large body of hypothetical studies, and there are some studies that evaluate or discuss participants' experiences after participating in a pilot project or a research trial. However, we found only one study that interviewed participants involved in a digital health intervention that was being delivered on a national scale, in Ghana (Entsieh 2015). Trials assessing the effectiveness of digital targeted client communication interventions should consider the issues identified in this qualitative evidence synthesis (Table 8 above) and should ensure that the design and assessment of the intervention are properly reported, for instance following existing reporting guidelines for digital health interventions (Agarwal 2016).

lient communication interventions should consider the issues identified in this qualitative evidence synthesis (Table 8 above) and should ensure that the design and assessment of the intervention are properly reported, for instance following existing reporting guidelines for digital health interventions (Agarwal 2016). Acknowledgements We would like to acknowledge the contributions of Garrett Mehl in developing the protocol and Janan Dietrich for the searching and screening of other relevant studies linked to the matrix analysis. Appendices Appendix 1. List of targeted client communication topics For adolescent and youth populations as potential users of SRH services For adult populations as potential users of SRH services For pregnant and postpartum women (up to 6 weeks) For pregnant and postpartum women (up to 6 weeks) living with HIV Parents and other caregivers of children under 5 years of age Family planning/ contraception Sexual violence Prevention, diagnosis, and treatment of STIs, including HIV Screening for cervical and breast cancer Folic acid fortification Infertility Safe abortion HPV vaccination Comprehensive sexual education Puberty Family planning/ contraception Sexual violence Prevention, diagnosis, and treatment of STIs, including HIV Screening for cervical and breast cancer Infertility Safe abortion Antenatal care Birth preparedness Skilled attendant at birth Emergency obstetric care Postpartum care Kangaroo Mother Care Tetanus immunization Anemia prevention and control STI testing and treatment in pregnancy Sexual violence Malaria prevention and treatment Smoking cessation during pregnancy Antenatal care Birth preparedness

Screening for cervical and breast cancer Infertility Safe abortion Antenatal care Birth preparedness Skilled attendant at birth Emergency obstetric care Postpartum care Kangaroo Mother Care Tetanus immunization Anemia prevention and control STI testing and treatment in pregnancy Sexual violence Malaria prevention and treatment Smoking cessation during pregnancy Antenatal care Birth preparedness Skilled attendant at birth Emergency obstetric care Postpartum care Kangaroo mother care Tetanus immunisation Anaemia prevention and control STI testing and treatment in pregnancy Sexual violence Malaria prevention and treatment ARV adherence Early infant diagnosis Retention of mother and infant pairs in eMTCT care Postnatal care Immunisation Breastfeeding Integrated management of newborn and childhood illnesses (IMNCI) Water, sanitation, and hygiene (WASH) Management of diarrhoeal illnesses, oral rehydration solution, zinc Growth monitoring and nutrition Early infant diagnosis in HIV‐exposed children; ARV therapy for HIV‐exposed and HIV‐infected children Early childhood development Abbreviations: ARV: antiretroviral; eMTCT: elimination of mother‐to‐child transmission; HPV: human papillomavirus; SRH: sexual and reproductive health; STI: sexually transmitted infection Appendix 2. Search strategy Ovid MEDLINE® In‐Process & Other Non‐Indexed Citations, Ovid MEDLINE® Daily and Ovid MEDLINE® 1946 to Present ‐ Searched 3 July 2017 1 Family Planning Services/ (23966) 2 Contraception/ (18495) 3 Reproductive behavior/ or Contraception behavior/ (8083) 4 exp Contraceptive agents/ (68229) 5 exp Contraceptive Devices/ (23779)

Appendix 2. Search strategy Ovid MEDLINE® In‐Process & Other Non‐Indexed Citations, Ovid MEDLINE® Daily and Ovid MEDLINE® 1946 to Present ‐ Searched 3 July 2017 1 Family Planning Services/ (23966) 2 Contraception/ (18495) 3 Reproductive behavior/ or Contraception behavior/ (8083) 4 exp Contraceptive agents/ (68229) 5 exp Contraceptive Devices/ (23779) 6 (condom* or (OC adj pill) or (depot medroxyprogest* or NET‐EN or NET EN or Mesigyna or Cyclofem) or (intrauterine system or intra‐uterine system or IUS or intrauterine device* or intra‐uterine device* or IUD*) or (vasectomy or sterilisation or sterilization or (tubal adj ligation)) or ((vaginal adj ring) or cycletel or cycle‐tel or ((abstain or abstinen*) adj2 (sex* or intercourse)) or lactational amenorr*)).ti,ab,kw. (55490) 7 (contracept* or family planning or (birth adj (control or regulat* or spacing)) or planned parenthood or ((population or fertility) adj (regulat* or control))).ti,ab,kw. (87955) 8 Pregnancy in Adolescence/ (7459) 9 (pregnan* adj2 (adolescen* or teen* or schoolchild*)).ti,ab,kw. (6428) 10 Pregnancy, unplanned/ or Pregnancy, unwanted/ (3799) 11 (pregnan* adj3 (prevent* or interrupt* or unplanned or unwanted or mistimed)).ti,ab,kw. (12797) 12 exp Sexually Transmitted Diseases/di, dt, ep, pc, px, tm [Diagnosis, Drug Therapy, Epidemiology, Prevention & Control, Psychology, Transmission] (202486) 13 (sexually transmi* or STI or STIs or STD or STDs or venereal).ti,ab,kw. (43203) 14 exp HIV Infections/di, dt, ep, pc, px, tm [Diagnosis, Drug Therapy, Epidemiology, Prevention & Control, Psychology, Transmission] (171444)

12 exp Sexually Transmitted Diseases/di, dt, ep, pc, px, tm [Diagnosis, Drug Therapy, Epidemiology, Prevention & Control, Psychology, Transmission] (202486) 13 (sexually transmi* or STI or STIs or STD or STDs or venereal).ti,ab,kw. (43203) 14 exp HIV Infections/di, dt, ep, pc, px, tm [Diagnosis, Drug Therapy, Epidemiology, Prevention & Control, Psychology, Transmission] (171444) 15 HIV Seropositivity/dt, ep, pc, px, tm [Drug Therapy, Epidemiology, Prevention & Control, Psychology, Transmission] (8236) 16 (Anti‐HIV Agents/ or Antiretroviral Therapy, Highly Active/) and Medication Adherence/ (1671) 17 (hiv or hiv‐1* or hiv‐2* or hiv1 or hiv2 or human immunodeficiency virus or human immunedeficiency virus or human immuno‐deficiency virus or human immune‐deficiency virus or (human immun* and deficiency virus) or acquired immunodeficiency syndrome or acquired immunedeficiency syndrome or acquired immuno‐deficiency syndrome or acquired immune‐deficiency syndrome or (acquired immun* and deficiency syndrome)).ti,ab,kw. (303793) 18 ((antiretroviral* or anti‐retroviral* or ARV*) adj2 (complian* or adheren*)).ti,ab,kw. (2178) 19 (Anti‐HIV Agents/ or Antiretroviral Therapy, Highly Active/) and (Infant, Premature/ or Infant, Newborn/ or Infant, Low Birth Weight/ or Infant, Extremely Low Birth Weight/ or Infant, Small for Gestational Age/ or Infant/ or Infant, Very Low Birth Weight/ or Infant, Postmature/ or Infant, Extremely Premature/ or Child/ or Child, Preschool/ or Adolescent/) (8310)

Highly Active/) and (Infant, Premature/ or Infant, Newborn/ or Infant, Low Birth Weight/ or Infant, Extremely Low Birth Weight/ or Infant, Small for Gestational Age/ or Infant/ or Infant, Very Low Birth Weight/ or Infant, Postmature/ or Infant, Extremely Premature/ or Child/ or Child, Preschool/ or Adolescent/) (8310) 20 ((antiretroviral* or anti‐retroviral* or ARV*) and (infant* or newborn* or neonat* or child* or schoolchild* or adolescen* or teen*)).ti,ab,kw. (7745) 21 Papillomavirus Infections/pc [Prevention & Control] (4874) 22 Papillomavirus Vaccines/ad, tu [Administration & Dosage, Therapeutic Use] (3397) 23 Human Papillomavirus Recombinant Vaccine Quadrivalent, Types 6, 11, 16, 18/ad, tu [Administration & Dosage, Therapeutic Use] (63) 24 ((hpv or papilloma virus* or papillomavirus*) adj2 (vaccinat* or revaccinat* or immuniz* or immunis* or immunother* or inoculat* or innoculat* or prophyla*)).ti,ab,kw. (4505) 25 Domestic Violence/ or Spouse Abuse/ or Intimate Partner Violence/ or Rape/ (18827) 26 (((sexual or domestic or spouse* or intimate partner) adj3 (violen* or abus*)) or rape).ti,ab,kw. (29317) 27 Puberty/ (12795) 28 (pubert* or pubescen*).ti,ab,kw. (35567) 29 Menstruation/ (15555) 30 (menstruat* or menstrual*).ti,ab,kw. (46593) 31 Abortion, Legal/ (7381) 32 Abortion, Induced/ (26890) 33 (abort* or miscarr* or (pregnan* adj2 terminat*)).ti,ab,kw. (90341) 34 Infertility/ (13584) 35 Reproductive Techniques, Assisted/ (8167) 36 Fertilization in Vitro/ (28895) 37 (infertil* or assisted reproductive technolog* or in vitro fertili* or in‐vitro fertili* or IVF).ti,ab,kw. (77929)

31 Abortion, Legal/ (7381) 32 Abortion, Induced/ (26890) 33 (abort* or miscarr* or (pregnan* adj2 terminat*)).ti,ab,kw. (90341) 34 Infertility/ (13584) 35 Reproductive Techniques, Assisted/ (8167) 36 Fertilization in Vitro/ (28895) 37 (infertil* or assisted reproductive technolog* or in vitro fertili* or in‐vitro fertili* or IVF).ti,ab,kw. (77929) 38 Sexual behavior/ or Sex work/ or Safe sex/ or Unsafe sex/ (58649) 39 (sex* adj (protected or unprotected or safe or unsafe or risk* or behavio*)).ti,ab,kw. (30802) 40 (Contact tracing/ or Disease notification/) and Sexual partners/ (481) 41 (partner* adj3 (notifi* or tracing or report*)).ti,ab,kw. (4188) 42 Prenatal Care/ (24157) 43 (((antenatal or ante‐natal or prenatal or pre‐natal or antepartum or ante‐partum) adj3 (care or service* or counsel* or test*)) or (birth adj3 prepar*)).ti,ab,kw. (23850) 44 Maternal Health Services/ (12513) 45 ((maternal or mother*) adj3 (health or service* or care)).ti,ab,kw. (23535) 46 Reproductive Health/ (2179) 47 (reproductive adj2 (health or care or service*)).ti,ab,kw. (11640) 48 Midwifery/ (17799) 49 (midwi* or skilled birth or skilled attendan*).ti,ab,kw. (21866) 50 Obstetric Labor Complications/ (16607) 51 Pregnancy Complications/ (85521) 52 ((obstetric* or pregnan* or labour or labor or parturition) adj3 (emergenc* or complication*)).ti,ab,kw. (19569) 53 Postnatal Care/ (4855) 54 Perinatal Care/ (3758) 55 Postpartum Period/ (22402) 56 ((postnatal or post‐natal or perinatal or peri‐natal or postpartum or post‐partum) adj2 (care or service*)).ti,ab,kw. (5608) 57 Maternal Nutritional Physiological Phenomena/ (3266)

52 ((obstetric* or pregnan* or labour or labor or parturition) adj3 (emergenc* or complication*)).ti,ab,kw. (19569) 53 Postnatal Care/ (4855) 54 Perinatal Care/ (3758) 55 Postpartum Period/ (22402) 56 ((postnatal or post‐natal or perinatal or peri‐natal or postpartum or post‐partum) adj2 (care or service*)).ti,ab,kw. (5608) 57 Maternal Nutritional Physiological Phenomena/ (3266) 58 Prenatal Nutritional Physiological Phenomena/ (1578) 59 Breast Feeding/ (34309) 60 (breast feed* or breast fed or breastfeed* or breastfed).ti,ab,kw. (36816) 61 (Infant, Premature/ or Infant, Newborn/ or Infant, Low Birth Weight/ or Infant, Extremely Low Birth Weight/ or Infant, Small for Gestational Age/ or Infant/ or Infant, Very Low Birth Weight/ or Infant, Postmature/ or Infant, Extremely Premature/) and Early Diagnosis/ (2250) 62 (early adj1 diagnos* adj2 (infant* or neonat* or newborn*)).ti,ab,kw. (378) 63 diagnosis.fs. and (infant* or neonat* or newborn*).ti,ab,kw. (83646) 64 *"Infectious Disease Transmission, Vertical"/ (8460) 65 ((mother‐to‐child transmi* adj3 (prevent* or eliminat*)) or emtct or pmtct or (vertical adj transmi*)).ti,ab,kw. (7627) 66 (Immunization/ or Immunization, passive/ or Immunization schedule/ or Immunization, secondary/ or Immunization Programs/ or Vaccination/ or Mass vaccination/) and (Infant, Premature/ or Infant, Newborn/ or Infant, Low Birth Weight/ or Infant, Extremely Low Birth Weight/ or Infant, Small for Gestational Age/ or Infant/ or Infant, Very Low Birth Weight/ or Infant, Postmature/ or Infant, Extremely Premature/ or Child/ or Child, Preschool/ or Adolescent/ or Pregnancy/) (43689)

) and (Infant, Premature/ or Infant, Newborn/ or Infant, Low Birth Weight/ or Infant, Extremely Low Birth Weight/ or Infant, Small for Gestational Age/ or Infant/ or Infant, Very Low Birth Weight/ or Infant, Postmature/ or Infant, Extremely Premature/ or Child/ or Child, Preschool/ or Adolescent/ or Pregnancy/) (43689) 67 ((immuniz* or immunis* or vaccinat*) and (infant* or newborn* or neonat* or child* or adolescen* or teen*)).ti,ab,kw. (45220) 68 Child health services/ or Maternal‐child health services/ (19872) 69 "Delivery of Health Care, Integrated"/ (10591) 70 ((integrat* adj3 (health care or healthcare or management or treat* or service*) adj3 (child* or schoolchild* or infant* or neonat* or newborn or adolescen* or teen*)) or IMNCI).ti,ab,kw. (886) 71 (Diarrhea/di, dt, ep, pc, th, tm or Diarrhea, Infantile/di, dt, ep, pc, th, tm) and (Infant, Premature/ or Infant, Newborn/ or Infant, Low Birth Weight/ or Infant, Extremely Low Birth Weight/ or Infant, Small for Gestational Age/ or Infant/ or Infant, Very Low Birth Weight/ or Infant, Postmature/ or Infant, Extremely Premature/ or Child/ or Child, Preschool/ or Adolescent/) (10175) 72 (diarrh* and (infant* or newborn* or neonat* or child* or schoolchild* or adolescen* or teen*)).ti,ab,kw. (25409) 73 Hand Hygiene/ or Hand Disinfection/ (5710) 74 Water Supply/ (30922) 75 Drinking Water/ (5420) 76 Sanitation/ (6553) 77 (handwash* or hand‐wash* or (wash* adj1 hand*) or hand hygiene or hand‐hygiene or soap or water suppl* or sanitation or sanitary or drinking water or potable water).ti,ab,kw. (79227)

72 (diarrh* and (infant* or newborn* or neonat* or child* or schoolchild* or adolescen* or teen*)).ti,ab,kw. (25409) 73 Hand Hygiene/ or Hand Disinfection/ (5710) 74 Water Supply/ (30922) 75 Drinking Water/ (5420) 76 Sanitation/ (6553) 77 (handwash* or hand‐wash* or (wash* adj1 hand*) or hand hygiene or hand‐hygiene or soap or water suppl* or sanitation or sanitary or drinking water or potable water).ti,ab,kw. (79227) 78 Fluid Therapy/ and (Infant, Premature/ or Infant, Newborn/ or Infant, Low Birth Weight/ or Infant, Extremely Low Birth Weight/ or Infant, Small for Gestational Age/ or Infant/ or Infant, Very Low Birth Weight/ or Infant, Postmature/ or Infant, Extremely Premature/ or Child/ or Child, Preschool/) (4512) 79 (oral rehydration adj (solution* or salt* or therapy)).ti,ab,kw. (2171) 80 Child Development/ or Adolescent Development/ (43788) 81 ((child* or schoolchild* or adolescen* or teen*) adj2 (develop* or progress*)).ti,ab,kw. (48148) 82 Breast Neoplasms/di, dg, pc or (Breast Neoplasms/ and Mass Screening/) (59944) 83 Uterine Cervical Neoplasms/di, dg, pc [Diagnosis, Diagnostic Imaging, Prevention & Control] (22428) 84 (((breast or cervix or cervical) adj (neoplasm* or cancer*)) and (screen* or diagnos*)).ti,ab,kw. (70367) 85 Folic Acid/ad, tu, th [Administration & Dosage, Therapeutic Use, Therapy] (8348) 86 Folic Acid Deficiency/dt, pc, th [Drug Therapy, Prevention & Control, Therapy] (803) 87 (folic acid adj (fortif* or supplement* or treat* or therap*)).ti,ab,kw. (3090) 88 Sex Education/ (8462) 89 (sex* adj (educat* or "health promot*")).ti,ab,kw. (8482) 90 Pregnancy in Adolescence/ (7459)

85 Folic Acid/ad, tu, th [Administration & Dosage, Therapeutic Use, Therapy] (8348) 86 Folic Acid Deficiency/dt, pc, th [Drug Therapy, Prevention & Control, Therapy] (803) 87 (folic acid adj (fortif* or supplement* or treat* or therap*)).ti,ab,kw. (3090) 88 Sex Education/ (8462) 89 (sex* adj (educat* or "health promot*")).ti,ab,kw. (8482) 90 Pregnancy in Adolescence/ (7459) 91 Kangaroo‐Mother Care Method/ (220) 92 (kangaroo adj2 (mother or infant or care)).ti,ab,kw. (540) 93 (Anemia/dt, pc or Anemia, Hypochromic/dt, pc or Anemia, Iron‐Deficiency/dt, pc) and Pregnancy/ (1420) 94 ((maternal or mother* or pregnan*) adj2 (nutrition* or folate or folic or iron or anaemi* or anemi*)).ti,ab,kw. (8929) 95 (Malaria/di, dt, pc or Malaria, Falciparum/di, dt, pc or Malaria, Vivax/di, dt, pc) and (Pregnancy/ or Pregnancy Complications, Parasitic/) (1999) 96 ((malaria* or falciparum or vivax) adj3 (pregnan* or mother* or maternal or postpartum or post partum)).ti,ab,kw. (2187) 97 Smoking Cessation/ and (Pregnancy/ or Pregnancy in Adolescence/) (1515) 98 (((smoking or smoker* or cigarette or tobacco) adj3 (ceas* or cessation or stop* or discontinu*)) and (pregnan* or maternal or mother*)).ti,ab,kw. (1895) 99 Mental health/ or Mental disorders/ or Mental health services/ or Community mental health services/ (206414) 100 Maternal behavior/ or Mother‐child relations/ or Parenting/ or Paternal behavior/ (38878) 101 Depression, Postpartum/ (4465) 102 (((mental or behavio*) adj3 (health or disorder*)) or postpartum depression or post‐partum depression).ti,ab,kw. (186868) 103 or/1‐102 (1885303) 104 Cell Phones/ (7022) 105 Smartphone/ (1162)

99 Mental health/ or Mental disorders/ or Mental health services/ or Community mental health services/ (206414) 100 Maternal behavior/ or Mother‐child relations/ or Parenting/ or Paternal behavior/ (38878) 101 Depression, Postpartum/ (4465) 102 (((mental or behavio*) adj3 (health or disorder*)) or postpartum depression or post‐partum depression).ti,ab,kw. (186868) 103 or/1‐102 (1885303) 104 Cell Phones/ (7022) 105 Smartphone/ (1162) 106 MP3‐Player/ (167) 107 Computers, Handheld/ (3076) 108 ((cell* or mobile*) adj1 (phone* or telephone* or technolog* or device*)).ti,ab,kw. (12925) 109 (handheld or hand‐held).ti,ab,kw. (9885) 110 (smartphone* or smart‐phone* or cellphone* or mobiles).ti,ab,kw. (5371) 111 ((personal adj1 digital) or (PDA adj3 (device* or assistant*)) or MP3 player* or MP4 player*).ti,ab,kw. (1291) 112 (samsung or nokia).ti,ab,kw. (808) 113 (windows adj3 (mobile* or phone*)).ti,ab,kw. (42) 114 android.ti,ab,kw. (1508) 115 (ipad* or i‐pad* or ipod* or i‐pod* or iphone* or i‐phone*).ti,ab,kw. (1927) 116 (tablet* adj3 (device* or computer*)).ti,ab,kw. (974) 117 Telemedicine/ (16542) 118 Videoconferencing/ or Webcasts as topic/ (1479) 119 Text Messaging/ (1630) 120 Telenursing/ (173) 121 (mhealth or m‐health or "mobile health" or ehealth or e‐health or "electronic health").ti,ab,kw. (15150) 122 (telemedicine or tele‐medicine or telehealth or tele‐health or telecare or tele‐care or telenursing or tele‐nursing or telepsychiatry or tele‐psychiatry or telemonitor* or tele‐monitor* or teleconsult* or tele‐consult* or telecounsel* or tele‐counsel* or telecoach* or tele‐coach*).ti,ab,kw. (13744)

121 (mhealth or m‐health or "mobile health" or ehealth or e‐health or "electronic health").ti,ab,kw. (15150) 122 (telemedicine or tele‐medicine or telehealth or tele‐health or telecare or tele‐care or telenursing or tele‐nursing or telepsychiatry or tele‐psychiatry or telemonitor* or tele‐monitor* or teleconsult* or tele‐consult* or telecounsel* or tele‐counsel* or telecoach* or tele‐coach*).ti,ab,kw. (13744) 123 (videoconferenc* or video‐conferenc* or webcast* or web‐cast*).ti,ab,kw. (2476) 124 (((text* or short or voice or multimedia or multi‐media or electronic or instant) adj1 messag*) or instant messenger).ti,ab,kw. (3373) 125 (texting or texted or texter* or ((sms or mms) adj (service* or messag*)) or interactive voice response* or IVR or voice call* or callback* or voice over internet or VOIP).ti,ab,kw. (2519) 126 (Facebook or Twitter or Whatsapp* or Skyp* or YouTube or "You Tube" or Google Hangout*).ti,ab,kw. (4017) 127 Mobile Applications/ (2148) 128 "mobile app*".ti,ab,kw. (1671) 129 Social Media/ (3725) 130 (social adj (media or network*)).ti,ab,kw. (15779) 131 Reminder Systems/ (3035) 132 (remind* adj3 (text* or system* or messag*)).ti,ab,kw. (1388) 133 Electronic Mail/ (2399) 134 (electronic mail* or email* or e‐mail or webmail).ti,ab,kw. (11345) 135 Medical informatics/ or Medical informatics applications/ (12810) 136 Nursing informatics/ or Public health informatics/ (2455) 137 ((medical or clinical or health or healthcare or nurs*) adj3 informatics).ti,ab,kw. (5000) 138 Multimedia/ (1784) 139 Hypermedia/ (398) 140 Blogging/ (807)

134 (electronic mail* or email* or e‐mail or webmail).ti,ab,kw. (11345) 135 Medical informatics/ or Medical informatics applications/ (12810) 136 Nursing informatics/ or Public health informatics/ (2455) 137 ((medical or clinical or health or healthcare or nurs*) adj3 informatics).ti,ab,kw. (5000) 138 Multimedia/ (1784) 139 Hypermedia/ (398) 140 Blogging/ (807) 141 (multimedia or multi‐media or hypermedia or hyper‐media or blog* or vlog* or weblog* or web‐log*).ti,ab,kw. (6134) 142 Interactive Tutorial/ (247) 143 Computer‐Assisted Instruction/ (11225) 144 ((interactive or computer‐assisted) adj1 (tutor* or technolog* or learn* or instruct* or software or communication)).ti,ab,kw. (2208) 145 or/104‐144 (132547) 146 103 and 145 (15436) 147 Qualitative Research/ (34710) 148 Interview/ (27953) 149 (theme$ or thematic).mp. (73915) 150 qualitative.af. (183873) 151 Nursing Methodology Research/ (16939) 152 questionnaire$.mp. (600933) 153 ethnological research.mp. (7) 154 ethnograph$.mp. (8388) 155 ethnonursing.af. (139) 156 phenomenol$.af. (21166) 157 (grounded adj (theor$ or study or studies or research or analys?s)).af. (9080) 158 (life stor$ or women* stor$).mp. (1094) 159 (emic or etic or hermeneutic$ or heuristic$ or semiotic$).af. or (data adj1 saturat$).tw. or participant observ$.tw. (18540) 160 (social construct$ or (postmodern$ or post‐structural$) or (post structural$ or poststructural$) or post modern$ or post‐modern$ or feminis$ or interpret$).mp. (455073) 161 (action research or cooperative inquir$ or co operative inquir$ or co‐operative inquir$).mp. (3372) 162 (humanistic or existential or experiential or paradigm$).mp. (124350)

160 (social construct$ or (postmodern$ or post‐structural$) or (post structural$ or poststructural$) or post modern$ or post‐modern$ or feminis$ or interpret$).mp. (455073) 161 (action research or cooperative inquir$ or co operative inquir$ or co‐operative inquir$).mp. (3372) 162 (humanistic or existential or experiential or paradigm$).mp. (124350) 163 (field adj (study or studies or research)).tw. (13705) 164 human science.tw. (251) 165 biographical method.tw. (15) 166 theoretical sampl$.af. (530) 167 ((purpos$ adj4 sampl$) or (focus adj group$)).af. (47292) 168 (life world or life‐world or conversation analys?s or personal experience$ or theoretical saturation).mp. (13506) 169 ((lived or life) adj experience$).mp. (8082) 170 cluster sampl$.mp. (5645) 171 observational method$.af. (586) 172 content analysis.af. (18829) 173 (constant adj (comparative or comparison)).af. (3524) 174 ((discourse$ or discurs$) adj3 analys?s).tw. (1693) 175 narrative analys?s.af. (871) 176 heidegger$.tw. (585) 177 colaizzi$.tw. (513) 178 spiegelberg$.tw. (77) 179 (van adj manen$).tw. (320) 180 (van adj kaam$).tw. (40) 181 (merleau adj ponty$).tw. (187) 182 husserl$.tw. (222) 183 foucault$.tw. (687) 184 (corbin$ adj2 strauss$).tw. (254) 185 glaser$.tw. (884) 186 or/147‐185 (1448447) 187 146 and 186 (4951) 188 limit 187 to yr="1993 ‐Current" (4863) Embase (Ovid) 1980 to 2017 Week 27 ‐ Searched 5 July 2017 1 family planning/ (34549) 2 contraception/ (41073) 3 reproductive behavior/ (901) 4 contraceptive behavior/ (2733) 5 exp contraceptive agent/ (129445) 6 exp contraceptive device/ (37567)

185 glaser$.tw. (884) 186 or/147‐185 (1448447) 187 146 and 186 (4951) 188 limit 187 to yr="1993 ‐Current" (4863) Embase (Ovid) 1980 to 2017 Week 27 ‐ Searched 5 July 2017 1 family planning/ (34549) 2 contraception/ (41073) 3 reproductive behavior/ (901) 4 contraceptive behavior/ (2733) 5 exp contraceptive agent/ (129445) 6 exp contraceptive device/ (37567) 7 (condom* or (OC adj pill) or (depot medroxyprogest* or NET‐EN or NET EN or Mesigyna or Cyclofem) or (intrauterine system or intra‐uterine system or IUS or intrauterine device* or intra‐uterine device* or IUD*) or (vasectomy or sterilisation or sterilization or (tubal adj ligation)) or ((vaginal adj ring) or cycletel or cycle‐tel or ((abstain or abstinen*) adj2 (sex* or intercourse)) or lactational amenorr*)).ti,ab,kw. (58260) 8 (contracept* or family planning or (birth adj (control or regulat* or spacing)) or planned parenthood or ((population or fertility) adj (regulat* or control))).ti,ab,kw. (83356) 9 adolescent pregnancy/ (8624) 10 (pregnan* adj2 (adolescen* or teen* or schoolchild*)).ti,ab,kw. (6466) 11 unplanned pregnancy/ (4221) 12 unwanted pregnancy/ (3016) 13 (pregnan* adj3 (prevent* or interrupt* or unplanned or unwanted or mistimed)).ti,ab,kw. (14324) 14 exp sexually transmitted disease/di, dt, ep, pc [Diagnosis, Drug Therapy, Epidemiology, Prevention] (36065) 15 (sexually transmi* or STI or STIs or STD or STDs or venereal).ti,ab,kw. (52215) 16 exp Human immunodeficiency virus infection/di, dt, ep, pc [Diagnosis, Drug Therapy, Epidemiology, Prevention] (161341)

13 (pregnan* adj3 (prevent* or interrupt* or unplanned or unwanted or mistimed)).ti,ab,kw. (14324) 14 exp sexually transmitted disease/di, dt, ep, pc [Diagnosis, Drug Therapy, Epidemiology, Prevention] (36065) 15 (sexually transmi* or STI or STIs or STD or STDs or venereal).ti,ab,kw. (52215) 16 exp Human immunodeficiency virus infection/di, dt, ep, pc [Diagnosis, Drug Therapy, Epidemiology, Prevention] (161341) 17 (hiv or hiv‐1* or hiv‐2* or hiv1 or hiv2 or human immunodeficiency virus or human immunedeficiency virus or human immuno‐deficiency virus or human immune‐deficiency virus or (human immun* and deficiency virus) or acquired immunodeficiency syndrome or acquired immunedeficiency syndrome or acquired immuno‐deficiency syndrome or acquired immune‐deficiency syndrome or (acquired immun* and deficiency syndrome)).ti,ab,kw. (369396) 18 ((antiretroviral* or anti‐retroviral* or ARV*) adj2 (complian* or adheren*)).ti,ab,kw. (2552) 19 (antiretroviral therapy/ or highly active antiretroviral therapy/) and medication compliance/ (675) 20 (antiretroviral therapy/ or highly active antiretroviral therapy/) and (child/ or infant/ or adolescent/ or newborn/) (4120) 21 ((antiretroviral* or anti‐retroviral* or ARV*) and (infant* or newborn* or neonat* or child* or schoolchild* or adolescen* or teen*)).ti,ab,kw. (10070) 22 papillomavirus infection/pc [Prevention] (2211) 23 Wart virus vaccine/ad, dt [Drug Administration, Drug Therapy] (6026) 24 ((hpv or papilloma virus* or papillomavirus*) adj2 (vaccinat* or revaccinat* or immuniz* or immunis* or immunother* or inoculat* or innoculat* or prophyla*)).ti,ab,kw. (5876)

21 ((antiretroviral* or anti‐retroviral* or ARV*) and (infant* or newborn* or neonat* or child* or schoolchild* or adolescen* or teen*)).ti,ab,kw. (10070) 22 papillomavirus infection/pc [Prevention] (2211) 23 Wart virus vaccine/ad, dt [Drug Administration, Drug Therapy] (6026) 24 ((hpv or papilloma virus* or papillomavirus*) adj2 (vaccinat* or revaccinat* or immuniz* or immunis* or immunother* or inoculat* or innoculat* or prophyla*)).ti,ab,kw. (5876) 25 domestic violence/ or battered woman/ or family violence/ or exp partner violence/ (19133) 26 statutory rape/ or acquaintance rape/ or rape/ or marital rape/ (7024) 27 (((sexual or domestic or spouse* or intimate partner) adj3 (violen* or abus*)) or rape).ti,ab,kw. (35500) 28 puberty/ or menarche/ (31168) 29 (pubert* or pubescen*).ti,ab,kw. (44006) 30 menstruation/ (18624) 31 (menstruat* or menstrual*).ti,ab,kw. (51352) 32 abortion/ or imminent abortion/ or recurrent abortion/ or septic abortion/ or spontaneous abortion/ (60269) 33 (abort* or miscarr* or (pregnan* adj2 terminat*)).ti,ab,kw. (100665) 34 infertility/ (36051) 35 infertility therapy/ or in vitro fertilization/ (18622) 36 (infertil* or assisted reproductive technolog* or in vitro fertili* or in‐vitro fertili* or IVF).ti,ab,kw. (111090) 37 sexual behavior/ or adolescent sexual behavior/ or casual sex/ or prostitution/ or exp safe sex/ or sexual practice/ or exp unsafe sex/ (105056) 38 (sex* adj (protected or unprotected or safe or unsafe or risk* or behavio*)).ti,ab,kw. (31641) 39 contact examination/ (3153) 40 (partner* adj3 (notifi* or tracing or report*)).ti,ab,kw. (5253)

37 sexual behavior/ or adolescent sexual behavior/ or casual sex/ or prostitution/ or exp safe sex/ or sexual practice/ or exp unsafe sex/ (105056) 38 (sex* adj (protected or unprotected or safe or unsafe or risk* or behavio*)).ti,ab,kw. (31641) 39 contact examination/ (3153) 40 (partner* adj3 (notifi* or tracing or report*)).ti,ab,kw. (5253) 41 prenatal care/ or prenatal screening/ (39325) 42 (((antenatal or ante‐natal or prenatal or pre‐natal or antepartum or ante‐partum) adj3 (care or service* or counsel* or test*)) or (birth adj3 prepar*)).ti,ab,kw. (29229) 43 maternal health service/ (427) 44 ((maternal or mother*) adj3 (health or service* or care or welfare)).ti,ab,kw. (26499) 45 reproductive health/ (13050) 46 (reproductive adj2 (health or care or service*)).ti,ab,kw. (14894) 47 midwife/ or nurse midwife/ (27896) 48 (midwi* or skilled birth or skilled attendan*).ti,ab,kw. (23691) 49 labor complication/ (9201) 50 pregnancy complication/ (71151) 51 ((obstetric* or pregnan* or labour or labor or parturition) adj3 (emergenc* or complication*)).ti,ab,kw. (30187) 52 postnatal care/ or newborn care/ (16297) 53 perinatal care/ (12694) 54 maternal care/ or maternal welfare/ (26845) 55 maternal nutrition/ (9782) 56 puerperium/ (31726) 57 ((postnatal or post‐natal or perinatal or peri‐natal or postpartum or post‐partum) adj2 (care or service*)).ti,ab,kw. (6938) 58 breast feeding/ (42834) 59 (breast feed* or breast fed or breastfeed* or breastfed).ti,ab,kw. (42934) 60 early diagnosis/ and (exp infant/ or newborn/) (5547) 61 (early adj1 diagnos* adj2 (infant* or neonat* or newborn*)).ti,ab,kw. (504)

57 ((postnatal or post‐natal or perinatal or peri‐natal or postpartum or post‐partum) adj2 (care or service*)).ti,ab,kw. (6938) 58 breast feeding/ (42834) 59 (breast feed* or breast fed or breastfeed* or breastfed).ti,ab,kw. (42934) 60 early diagnosis/ and (exp infant/ or newborn/) (5547) 61 (early adj1 diagnos* adj2 (infant* or neonat* or newborn*)).ti,ab,kw. (504) 62 diagnosis.fs. and (infant* or neonat* or newborn*).ti,ab,kw. (103248) 63 vertical transmission/ (12627) 64 ((mother‐to‐child transmi* adj3 (prevent* or eliminat*)) or emtct or pmtct or (vertical adj transmi*)).ti,ab,kw. (9576) 65 (immunization/ or mass immunization/ or vaccination/) and (exp infant/ or newborn/ or exp child/ or adolescent/ or pregnancy/) (43651) 66 ((immuniz* or immunis* or vaccinat*) and (infant* or newborn* or neonat* or child* or adolescen* or teen* or pregnan*)).ti,ab,kw. (55411) 67 child health care/ or early childhood intervention/ or maternal child health care/ (36574) 68 integrated health care system/ (9135) 69 ((integrat* adj3 (health care or healthcare or management or treat* or service*) adj3 (child* or schoolchild* or infant* or neonat* or newborn or adolescen* or teen*)) or IMNCI).ti,ab,kw. (976) 70 diarrhea/di, dm, dt, ep, pc, th [Diagnosis, Disease Management, Drug Therapy, Epidemiology, Prevention, Therapy] (21647) 71 infantile diarrhea/di, dm, dt, ep, pc, th [Diagnosis, Disease Management, Drug Therapy, Epidemiology, Prevention, Therapy] (1735) 72 (diarrhea/di, dm, dt, ep, pc, th or infantile diarrhea/di, dm, dt, ep, pc, th) and (exp infant/ or newborn/ or exp child/ or adolescent/ or pregnancy/) (7720)

70 diarrhea/di, dm, dt, ep, pc, th [Diagnosis, Disease Management, Drug Therapy, Epidemiology, Prevention, Therapy] (21647) 71 infantile diarrhea/di, dm, dt, ep, pc, th [Diagnosis, Disease Management, Drug Therapy, Epidemiology, Prevention, Therapy] (1735) 72 (diarrhea/di, dm, dt, ep, pc, th or infantile diarrhea/di, dm, dt, ep, pc, th) and (exp infant/ or newborn/ or exp child/ or adolescent/ or pregnancy/) (7720) 73 (diarrh* and (infant* or newborn* or neonat* or child* or schoolchild* or adolescen* or teen*)).ti,ab,kw. (29397) 74 hand washing/ or hand disinfection/ (11356) 75 water supply/ (32264) 76 drinking water/ (38027) 77 sanitation/ (12723) 78 (handwash* or hand‐wash* or (wash* adj1 hand*) or hand hygiene or hand‐hygiene or soap or water suppl* or sanitation or sanitary or drinking water or potable water).ti,ab,kw. (93230) 79 oral rehydration therapy/ (2408) 80 (oral rehydration adj (solution* or salt* or therapy)).ti,ab,kw. (2242) 81 child development/ or adolescent development/ (43677) 82 ((child* or schoolchild* or adolescen* or teen*) adj2 (develop* or progress*)).ti,ab,kw. (53455) 83 breast cancer/di, dm, dt, pc [Diagnosis, Disease Management, Drug Therapy, Prevention] (95144) 84 breast cancer/ and cancer screening/ (15395) 85 uterine cervix cancer/di, dm, dt, pc [Diagnosis, Disease Management, Drug Therapy, Prevention] (16779) 86 (((breast or cervix or cervical) adj (neoplasm* or cancer*)) and (screen* or diagnos*)).ti,ab,kw. (107649) 87 folic acid/ad, dt [Drug Administration, Drug Therapy] (11535) 88 folic acid deficiency/dm, dt, pc, th [Disease Management, Drug Therapy, Prevention, Therapy] (1168)

85 uterine cervix cancer/di, dm, dt, pc [Diagnosis, Disease Management, Drug Therapy, Prevention] (16779) 86 (((breast or cervix or cervical) adj (neoplasm* or cancer*)) and (screen* or diagnos*)).ti,ab,kw. (107649) 87 folic acid/ad, dt [Drug Administration, Drug Therapy] (11535) 88 folic acid deficiency/dm, dt, pc, th [Disease Management, Drug Therapy, Prevention, Therapy] (1168) 89 (folic acid adj (fortif* or supplement* or treat* or therap*)).ti,ab,kw. (3883) 90 sexual education/ (10473) 91 (sex* adj (educat* or "health promot*")).ti,ab,kw. (8361) 92 kangaroo care/ (706) 93 (kangaroo adj2 (mother or infant or care)).ti,ab,kw. (718) 94 (anemia/dt, pc or iron deficiency anemia/dt, pc) and pregnancy/ (1207) 95 ((maternal or mother* or pregnan*) adj2 (nutrition* or folate or folic or iron or anaemi* or anemi*)).ti,ab,kw. (9114) 96 (malaria/di, dm, dt, pc or malaria, falciparum/di, dm, dt, pc or malaria, vivax/di, dm, dt, pc) and (pregnancy/ or pregnancy complication/) (1486) 97 ((malaria* or falciparum or vivax) adj3 (pregnan* or mother* or maternal or postpartum or post partum)).ti,ab,kw. (2651) 98 smoking cessation/ and (pregnancy/ or adolescent pregnancy/) (1931) 99 (((smoking or smoker* or cigarette or tobacco) adj3 (ceas* or cessation or stop* or discontinu*)) and (pregnan* or maternal or mother*)).ti,ab,kw. (2272) 100 mental health/ or community mental health/ or mental health service/ (147416) 101 maternal behavior/ or parental behavior/ or paternal behavior/ (22210) 102 puerperal depression/ (8330)

99 (((smoking or smoker* or cigarette or tobacco) adj3 (ceas* or cessation or stop* or discontinu*)) and (pregnan* or maternal or mother*)).ti,ab,kw. (2272) 100 mental health/ or community mental health/ or mental health service/ (147416) 101 maternal behavior/ or parental behavior/ or paternal behavior/ (22210) 102 puerperal depression/ (8330) 103 (((mental or behavio*) adj3 (health or disorder*)) or postpartum depression or post‐partum depression or postnatal depression or post‐natal depression).ti,ab,kw. (235408) 104 or/1‐103 (2249116) 105 mobile phone/ or smartphone/ (15726) 106 mp3 player/ (160) 107 ((cell* or mobile*) adj1 (phone* or telephone* or technolog* or device*)).ti,ab,kw. (16368) 108 (handheld or hand‐held).ti,ab,kw. (12981) 109 (smartphone* or smart‐phone* or cellphone* or mobiles).ti,ab,kw. (7417) 110 ((personal adj1 digital) or (PDA adj3 (device* or assistant*)) or MP3 player* or MP4 player*).ti,ab,kw. (1683) 111 (samsung or nokia).ti,ab,kw. (1425) 112 (windows adj3 (mobile* or phone*)).ti,ab,kw. (61) 113 android.ti,ab,kw. (2314) 114 (ipad* or i‐pad* or ipod* or i‐pod* or iphone* or i‐phone*).ti,ab,kw. (3494) 115 (tablet* adj3 (device* or computer*)).ti,ab,kw. (1535) 116 telemedicine/ or telecardiology/ or teleconsultation/ or teledermatology/ or telediagnosis/ or telemonitoring/ or telepathology/ or telepsychiatry/ or teleradiotherapy/ or telesurgery/ or teletherapy/ (27082) 117 videoconferencing/ or webcast/ (2779) 118 text messaging/ (2815) 119 telenursing/ (201) 120 (mhealth or m‐health or "mobile health" or ehealth or e‐health or "electronic health").ti,ab,kw. (19120)

116 telemedicine/ or telecardiology/ or teleconsultation/ or teledermatology/ or telediagnosis/ or telemonitoring/ or telepathology/ or telepsychiatry/ or teleradiotherapy/ or telesurgery/ or teletherapy/ (27082) 117 videoconferencing/ or webcast/ (2779) 118 text messaging/ (2815) 119 telenursing/ (201) 120 (mhealth or m‐health or "mobile health" or ehealth or e‐health or "electronic health").ti,ab,kw. (19120) 121 (telemedicine or tele‐medicine or telehealth or tele‐health or telecare or tele‐care or telenursing or tele‐nursing or telepsychiatry or tele‐psychiatry or telemonitor* or tele‐monitor* or teleconsult* or tele‐consult* or telecounsel* or tele‐counsel* or telecoach* or tele‐coach*).ti,ab,kw. (17396) 122 (videoconferenc* or video‐conferenc* or webcast* or web‐cast*).ti,ab,kw. (3291) 123 (((text* or short or voice or multimedia or multi‐media or electronic or instant) adj1 messag*) or instant messenger).ti,ab,kw. (4409) 124 (texting or texted or texter* or ((sms or mms) adj (service* or messag*)) or interactive voice response* or IVR or voice call* or callback* or voice over internet or VOIP).ti,ab,kw. (3476) 125 (Facebook or Twitter or Whatsapp* or Skyp* or YouTube or "You Tube" or Google Hangout*).ti,ab,kw. (5764) 126 mobile application/ (4307) 127 "mobile app*".ti,ab,kw. (1967) 128 social media/ (8882) 129 (social adj (media or network*)).ti,ab,kw. (20412) 130 reminder system/ (2115) 131 (remind* adj3 (text* or system* or messag*)).ti,ab,kw. (1924) 132 e‐mail/ (14658) 133 (electronic mail* or email* or e‐mail or webmail).ti,ab,kw. (22106) 134 medical informatics/ (17675) 135 nursing informatics/ (1272)

127 "mobile app*".ti,ab,kw. (1967) 128 social media/ (8882) 129 (social adj (media or network*)).ti,ab,kw. (20412) 130 reminder system/ (2115) 131 (remind* adj3 (text* or system* or messag*)).ti,ab,kw. (1924) 132 e‐mail/ (14658) 133 (electronic mail* or email* or e‐mail or webmail).ti,ab,kw. (22106) 134 medical informatics/ (17675) 135 nursing informatics/ (1272) 136 ((medical or clinical or health or healthcare or nurs*) adj3 informatics).ti,ab,kw. (6985) 137 multimedia/ (3162) 138 hypermedia/ (368) 139 blogging/ (135) 140 (multimedia or multi‐media or hypermedia or hyper‐media or blog* or vlog* or weblog* or web‐log*).ti,ab,kw. (8936) 141 teaching/ (79979) 142 ((interactive or computer‐assisted) adj1 (tutor* or technolog* or learn* or instruct* or software or communication)).ti,ab,kw. (3036) 143 or/105‐142 (250953) 144 104 and 143 (27712) 145 qualitative analysis/ or qualitative research/ (91734) 146 interview/ (163414) 147 (theme$ or thematic).mp. (90768) 148 qualitative.af. (231824) 149 nursing methodology research/ (14443) 150 questionnaire$.mp. (730111) 151 ethnological research.mp. (8) 152 ethnograph$.mp. (9657) 153 ethnonursing.af. (103) 154 phenomenol$.af. (25758) 155 (grounded adj (theor$ or study or studies or research or analys?s)).af. (10886) 156 (life stor$ or women* stor$).mp. (1335) 157 (emic or etic or hermeneutic$ or heuristic$ or semiotic$).af. or (data adj1 saturat$).tw. or participant observ$.tw. (20697) 158 (social construct$ or (postmodern$ or post‐structural$) or (post structural$ or poststructural$) or post modern$ or post‐modern$ or feminis$ or interpret$).mp. (395594)

156 (life stor$ or women* stor$).mp. (1335) 157 (emic or etic or hermeneutic$ or heuristic$ or semiotic$).af. or (data adj1 saturat$).tw. or participant observ$.tw. (20697) 158 (social construct$ or (postmodern$ or post‐structural$) or (post structural$ or poststructural$) or post modern$ or post‐modern$ or feminis$ or interpret$).mp. (395594) 159 (action research or cooperative inquir$ or co operative inquir$ or co‐operative inquir$).mp. (4053) 160 (humanistic or existential or experiential or paradigm$).mp. (151958) 161 (field adj (study or studies or research)).tw. (15406) 162 human science.tw. (267) 163 biographical method.tw. (19) 164 theoretical sampl$.af. (692) 165 ((purpos$ adj4 sampl$) or (focus adj group$)).af. (52066) 166 (life world or life‐world or conversation analys?s or personal experience$ or theoretical saturation).mp. (38320) 167 ((lived or life) adj experience$).mp. (10021) 168 cluster sampl$.mp. (7024) 169 observational method$.af. (1937) 170 content analysis.af. (23013) 171 (constant adj (comparative or comparison)).af. (4144) 172 ((discourse$ or discurs$) adj3 analys?s).tw. (1875) 173 narrative analys?s.af. (1004) 174 heidegger$.tw. (644) 175 colaizzi$.tw. (563) 176 spiegelberg$.tw. (95) 177 (van adj manen$).tw. (338) 178 (van adj kaam$).tw. (34) 179 (merleau adj ponty$).tw. (193) 180 husserl$.tw. (266) 181 foucault$.tw. (707) 182 (corbin$ adj2 strauss$).tw. (266) 183 glaser$.tw. (828) 184 or/145‐183 (1691925) 185 144 and 184 (8529) 186 limit 185 to yr="1993 ‐Current" (8389) 187 limit 186 to embase (3633)

176 spiegelberg$.tw. (95) 177 (van adj manen$).tw. (338) 178 (van adj kaam$).tw. (34) 179 (merleau adj ponty$).tw. (193) 180 husserl$.tw. (266) 181 foucault$.tw. (707) 182 (corbin$ adj2 strauss$).tw. (266) 183 glaser$.tw. (828) 184 or/145‐183 (1691925) 185 144 and 184 (8529) 186 limit 185 to yr="1993 ‐Current" (8389) 187 limit 186 to embase (3633) WHO Global Health Library – Searched 6 July 2017 (tw:((cell* OR mobile*) AND (phone* OR telephone* OR technolog* OR device*)) OR smartphone* OR smart‐phone* OR cellphone* OR mobiles OR mhealth OR m‐health OR "mobile health" OR ehealth OR e‐health OR "electronic health" OR telemedicine OR tele‐medicine OR telehealth OR tele‐health OR telecare OR tele‐care OR telenursing OR tele‐nursing OR telepsychiatry OR tele‐psychiatry OR telemonitor* OR tele‐monitor* OR teleconsult* OR tele‐consult* OR telecounsel* OR tele‐counsel* OR telecoach* OR tele‐coach* OR videoconferenc* OR video‐conferenc* OR webcast* OR web‐cast* OR ((text* OR short OR voice OR multimedia OR multi‐media OR electronic OR instant) AND messag*) OR instant messenger OR texting OR texted OR texter* OR ((sms OR mms) AND (service* OR messag*)) OR interactive voice response* OR ivr OR voice call* OR callback* OR voice over internet OR voip OR "mobile app*" OR (social AND (media OR network*)) OR ((medical OR clinical OR health OR healthcare OR nurs*) AND informatics)) OR mj:("Telemedicine" OR "Cell Phones" OR "Internet" OR "Mobile Applications" OR "Medical Informatics" OR "Information Technology" OR "Smartphone")) AND (instance:"ghl") AND ( db:("LILACS" OR "WPRIM" OR "WHOLIS" OR "IMEMR" OR "AIM") AND year_cluster:("2015" OR "2013" OR "2014" OR "2005" OR "2007" OR "2006" OR "2011" OR "2012" OR "2009" OR "2001" OR "2003" OR "2010" OR "2016" OR "2008" OR "2002" OR "2000" OR "1998" OR "2004" OR "1999" OR "1997" OR "1996" OR "1995" OR "1993" OR "2017") AND tw:(qualitative OR interview* OR focus group* OR questionnaire* OR ethnograph* OR perception* OR perceiv* OR opinion* OR attitude* OR view* OR experienc* OR sceptic* OR skeptic* OR dilemma* OR "social mobilisation" OR "social mobilization" OR complian* OR refus* OR feeling* OR impression* OR belief* OR trust OR accept* OR knowledge OR comprehension OR understanding OR aware* OR ( (communit* OR social OR patient* ) AND ( behavi* OR integrat* OR program* OR support* OR norm OR norms OR leader* OR advoca* OR information OR action OR need OR needs OR influenc* OR complian* OR participat* ) )) = 1411 hits

n* OR belief* OR trust OR accept* OR knowledge OR comprehension OR understanding OR aware* OR ( (communit* OR social OR patient* ) AND ( behavi* OR integrat* OR program* OR support* OR norm OR norms OR leader* OR advoca* OR information OR action OR need OR needs OR influenc* OR complian* OR participat* ) )) = 1411 hits POPLINE – Searched 6 July 2017 Keyword:(TEXT MESSAGING OR MOBILE DEVICES OR INFORMATION COMMUNICATION TECHNOLOGY OR CELLULAR PHONE) AND All Fields: (qualitative OR interview OR interviews OR interviewing OR focus group OR focus groups OR questionnaire OR questionnaires OR ethnography OR ethnographic OR perception OR perceptions OR perceive OR perceives OR perceived OR opinion OR opinions OR attitude OR attitudes OR view OR views OR experience OR experiences OR experienced OR experiencing OR sceptic OR sceptical OR skeptic OR skeptical OR dilemma OR dilemmas OR "social mobilisation" OR "social mobilization" OR compliant OR compliance OR refuse OR refusal OR refused OR refusing OR feeling OR feelings OR impression OR impressions OR belief OR beliefs OR trust OR accept OR acceptance OR accepting OR accepted OR knowledge OR comprehension OR understanding OR aware OR awareness OR ( (community OR communities OR social OR patient OR patients OR client OR clients ) AND ( behavior OR behaviour OR behaviors OR behaviours OR behavioral OR behavioural OR integrate OR integration OR program OR programs OR programme OR programmes OR support OR supporting OR supported OR norm OR norms OR leader OR leaders OR leadership OR advocate OR advocated OR advocacy OR information OR action OR need OR needs OR influence OR influences OR influenced OR influencing OR participate OR participates OR participated OR participation ) ))

OR programme OR programmes OR support OR supporting OR supported OR norm OR norms OR leader OR leaders OR leadership OR advocate OR advocated OR advocacy OR information OR action OR need OR needs OR influence OR influences OR influenced OR influencing OR participate OR participates OR participated OR participation ) )) OR

iews OR experience OR experiences OR experienced OR experiencing OR sceptic OR sceptical OR skeptic OR skeptical OR dilemma OR dilemmas OR "social mobilisation" OR "social mobilization" OR compliant OR compliance OR refuse OR refusal OR refused OR refusing OR feeling OR feelings OR impression OR impressions OR belief O R beliefs OR trust OR accept OR acceptance OR accepting OR accepted OR knowledge OR comprehension OR understanding OR aware OR awareness OR ( (community OR communities OR social OR patient OR patients OR client OR clients ) AND ( behavior OR behaviour OR behaviors OR behaviours OR behavioral OR behavioural OR integrate OR integration OR program OR programs OR programme OR programmes OR support OR supporting OR supported OR norm OR norms OR leader OR leaders OR leadership OR advocate OR advocated OR advocacy OR information OR action OR need OR needs OR influence OR influences OR influenced OR influencing OR participate OR participates OR participated OR participation ) )) – 1381 hits (1993‐2017)

es OR support OR supporting OR supported OR norm OR norms OR leader OR leaders OR leadership OR advocate OR advocated OR advocacy OR information OR action OR need OR needs OR influence OR influences OR influenced OR influencing OR participate OR participates OR participated OR participation ) )) – 1381 hits (1993‐2017) Appendix 3. Data richness scale table Score Measure Example 1 Very little qualitative data presented that relate to the synthesis objective. Those findings that are presented are fairly descriptive. For example, a mixed‐methods study using open‐ended survey questions or a more detailed qualitative study where only part of the data relates to the synthesis objective 2 Some qualitative data presented that relate to the synthesis objective For example, a limited number of qualitative findings from a mixed‐methods or qualitative study 3 A reasonable amount of qualitative data that relate to the synthesis objective For example, a typical qualitative research article in a journal with a smaller word limit and often using simple thematic analysis 4 A good amount and depth of qualitative data that relate to the synthesis objective For example, a qualitative research article in a journal with a larger word count that includes more context and setting descriptions and a more in‐depth presentation of the findings 5 A large amount and depth of qualitative data that relate in depth to the synthesis objective For example, from a detailed ethnography or a published qualitative article with the same objectives as the synthesis

count that includes more context and setting descriptions and a more in‐depth presentation of the findings 5 A large amount and depth of qualitative data that relate in depth to the synthesis objective For example, from a detailed ethnography or a published qualitative article with the same objectives as the synthesis Appendix 4. CERQual evidence profiles Finding 1: Overall, participants had a range of views regarding acceptance of the idea of receiving health information through their mobile devices. This was due to factors such as familiarity with the technology, convenience, control, being able to save and re‐read messages later, cost, seeing it as a simple way of providing a reminder for medication or appointments, and the sense that someone was thinking about them and cared enough to send a message.

le devices. This was due to factors such as familiarity with the technology, convenience, control, being able to save and re‐read messages later, cost, seeing it as a simple way of providing a reminder for medication or appointments, and the sense that someone was thinking about them and cared enough to send a message. Assessment for each CERQual component Methodological limitations Moderate concerns about methodological limitations due to poor reporting of participant voices in the findings and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Moderate concerns about relevance due to a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy No or very minor concerns about adequacy Overall CERQual assessment Low confidence Due to moderate concerns regarding methodological limitations and relevance Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Brown 2014 USA; single, adolescent mothers; health promotion information weekly via SMS during the first 6 months postpartum; pilot or implementation study with participation in an mHealth programme Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Cates 2015 USA; middle school students designing text messages to promote HPV vaccine; hypothetical with an example of messages being used Cornelius 2009 USA; African‐American adolescents; SMS to support HIV/AIDS curriculum; hypothetical with no examples of programme content Curioso 2009 Peru; HIV‐positive adults receiving ART; SMS related to HIV/AIDS; hypothetical with no examples of programme content Evans 2016 UK; African communities; SMS‐based HIV mHealth programme; hypothetical with examples and with no examples of programme content French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Gold 2010 Australia; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Hirsch‐

an RCT or pilot RCT Gold 2010 Australia; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Hirsch‐ Moverman 2017 Lesotho; HIV patients; SMS to provide real‐time adherence support to people on HIV and TB treatment; part of an RCT or pilot RCT Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Lau 2014 South Africa; pregnant women; SMS for antenatal health promotion; part of an RCT or pilot RCT Mbuagbaw 2012 Cameroon; HIV‐positive patients; SMS for HIV drug adherence; part of an RCT or pilot RCT Mbuagbaw 2014 Cameroon; individuals living with HIV or involved in HIV support work; community‐owned text messaging programme to support people living with HIV; hypothetical with no examples of programme content Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery; pilot or implementation study with participation in an mHealth programme Munro 2017 Canada; pregnant or have given birth in the last 12 months; SMS Text4baby programme about prenatal education; hypothetical with an example of messages being used Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Perry 2012 USA; adolescents aged 15 to 20; SMS with preventative sexual health messages; evaluation or formative research on an existing mHealth programme that the participants have been using Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RC

participants have been using Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RC T Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Anti retroviral therapy (ARV); Human Papillomavirus (HPV); Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT); Tuberculosis (TB)

example of messages being used Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Anti retroviral therapy (ARV); Human Papillomavirus (HPV); Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT); Tuberculosis (TB) Finding 2: In discussing the pros and cons of digital targeted client communication compared to in‐person meetings with a healthcare provider, some participants perceived interacting with a healthcare provider as preferable, warmer, and something to which they were accustomed. Others also felt that people could receive a faster response using digital communication and that the messages were more convenient and less judgemental. However, some liked having direct access to both healthcare providers and digital targeted client communication. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of researcher reflexivity Coherence No or very minor concerns about coherence Relevance Serious concerns about relevance due to a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention and partial relevance of the target group Adequacy Serious concerns about adequacy due to thin data from a small number of studies Overall CERQual assessment Very low confidence Due to minor concerns regarding methodological limitations and serious concerns regarding adequacy and relevance Contributing studies Study Context Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Nachega 2016 South Africa; HIV‐infected pregnant women; SMS about ART adherence to prevent PMTCT; hypothetical with no examples of programme content Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Anti retroviral therapy (ARV); Prevention of mother‐to‐child transmission (PMTCT)

ancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Anti retroviral therapy (ARV); Prevention of mother‐to‐child transmission (PMTCT) Finding 3: Participants said that they liked 2‐way digital communication, as this allowed them to engage directly with a healthcare provider, which they trusted more; to receive answers to their questions and have opportunities for discussion; and to receive a more immediate response. However, some participants felt that for some topics they would feel uncomfortable talking to a healthcare provider through a digital channel, due to issues related to shyness and privacy, and would prefer to use SMS.

rs to their questions and have opportunities for discussion; and to receive a more immediate response. However, some participants felt that for some topics they would feel uncomfortable talking to a healthcare provider through a digital channel, due to issues related to shyness and privacy, and would prefer to use SMS. Assessment for each CERQual component Methodological limitations Moderate concerns about methodological limitations due to poor reporting of sampling, ethical considerations, and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Serious concerns about relevance due to a large number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy Moderate concerns about adequacy due to thin data Overall CERQual assessment Very low confidence Due to moderate concerns regarding methodological limitations and adequacy and serious concerns regarding relevance Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Cates 2015 USA; middle school students designing text messages to promote HPV vaccine; hypothetical with an example of messages being used Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Anti retroviral therapy (ART); Human Papillomavirus (HPV);

omen who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Anti retroviral therapy (ART); Human Papillomavirus (HPV); Prevention of mother‐to‐child transmission (PMTCT) Finding 4: Some participants expressed a concern that some people might view digital targeted communication from healthcare providers as a replacement for seeking appropriate medical assistance, which might have adverse impacts. Whilst some saw digital health as a way to increase access to care, others noted that text messaging might be seen by poorer people as a cheaper or sufficient healthcare option, which might decrease appropriate health‐seeking behaviour. Assessment for each CERQual component Methodological limitations No or very minor concerns Coherence No or very minor concerns Relevance Serious concerns due to data from only 1 setting Adequacy Serious concerns due to data from only 1 setting Overall CERQual assessment Very low confidence Due to serious concerns regarding relevance and adequacy Contributing studies Study Context Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used

Adequacy Serious concerns due to data from only 1 setting Overall CERQual assessment Very low confidence Due to serious concerns regarding relevance and adequacy Contributing studies Study Context Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Finding 5: Participants reported varying degrees of access to network services, including cell networks (for calls and SMS) and internet. In addition, some had poor access to electricity to charge their phones. These factors were reported to be barriers to using the digital targeted client communication.

Adequacy Serious concerns due to data from only 1 setting Overall CERQual assessment Very low confidence Due to serious concerns regarding relevance and adequacy Contributing studies Study Context Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Finding 5: Participants reported varying degrees of access to network services, including cell networks (for calls and SMS) and internet. In addition, some had poor access to electricity to charge their phones. These factors were reported to be barriers to using the digital targeted client communication. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of sampling (unclear how participants were recruited in several studies) and researcher reflexivity Coherence No or very minor concerns about coherence Relevance No or very minor concerns about relevance Adequacy No or very minor concerns about adequacy Overall CERQual assessment High confidence Due to minor concerns regarding methodological limitations Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Cornelius 2009 USA; African‐American adolescents; SMS to support HIV/AIDS curriculum; hypothetical with no examples of programme content Flax 2017 Nigeria; women of all ages belonging to microcredit financing groups; received weekly cell phone breastfeeding text and voice messages to a shared phone; part of an RCT or pilot RCT Hirsch‐Moverman 2017 Lesotho; HIV patients; SMS to provide real‐time adherence support to people on HIV and TB treatment; part of an RCT or pilot RCT Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Mbuagbaw 2012 Cameroon; HIV‐positive patients; SMS for HIV drug adherence; part of an RCT or pilot RCT Mbuagbaw 2014 Cameroon; individuals living with HIV or involved in HIV support work; community‐owned text messaging programme to support people living with HIV; hypothetical with no examples of programme content Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Randomized control trial (RCT); Tuberculosis (TB)

ext messaging programme to support people living with HIV; hypothetical with no examples of programme content Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Randomized control trial (RCT); Tuberculosis (TB) Finding 6: Participants reported varying degrees of access to mobile devices. For instance, some had no phone; some had lost or broken their phone; some could not afford to purchase airtime; some had changed their number or sim card; or for some access to the phone was controlled by another person. These factors were reported to be barriers to using the digital targeted client communication. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of sampling (unclear how participants were recruited in several studies) and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Minor concerns about relevance due to a focus on study populations that may have limited access to mobile phone ownership, e.g.

itations due to poor reporting of sampling (unclear how participants were recruited in several studies) and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Minor concerns about relevance due to a focus on study populations that may have limited access to mobile phone ownership, e.g. due to age, gender, socio‐economic status, or health condition (partial relevance) Adequacy No or very minor concerns about adequacy Overall CERQual assessment Moderate confidence Due to minor concerns regarding methodological limitations and relevance Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Entsieh 2015 Ghana; pregnant and nursing mothers aged 20 to 35; “Mobile Midwife” app; qualitative research on an existing programme implemented at scale Flax 2017 Nigeria; Women of all ages belonging to micro credit financing groups; received weekly cell phone breastfeeding text and voice messages to a shared phone; Part of a RCT or pilot RCT Hirsch‐Moverman 2017 Lesotho; HIV patients; SMS to provide real‐time adherence support to people on HIV and TB treatment; part of an RCT or pilot RCT Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery; pilot or implementation study with participation in an mHealth programme Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Prevention of mother‐to‐child transmission (PMTCT); Tuberculosis (TB)

iving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Prevention of mother‐to‐child transmission (PMTCT); Tuberculosis (TB) Finding 7: Some participants, particularly women and adolescents, had their access to phones controlled or restricted by others, especially if they had to share or borrow a phone. They noted that they would often have to explain why they wanted to use the phone, and who they wanted to call, to allay suspicions about this communication. They mentioned that this was a barrier to accessing digital targeted client communication and made it difficult to keep their messages private. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of sampling (unclear how participants were recruited in several studies) and researcher reflexivity Coherence Minor concerns about coherence, as the majority of participants in 1 study did not see phone sharing as a problem Relevance Minor concerns about relevance due to a focus on study populations that may have limited access to mobile phone ownership, e.g. due to age, gender, SES, or health condition (partial relevance) Adequacy Minor concerns about adequacy due to a limited number of studies Overall CERQual assessment Moderate confidence Due to minor concerns regarding methodological limitations, coherence, adequacy, and relevance Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Flax 2017 Nigeria; Women of all ages belonging to micro credit financing groups; received weekly cell phone breastfeeding text and voice messages to a shared phone; Part of a RCT or pilot RCT Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content

CT or pilot RCT Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Finding 8: Participants believed that the cost of taking part in digital targeted client communication should be free or very low, as cost could present a barrier to participation, particularly for young people and those on lower incomes. Participants felt that there should be little or no charge for costs such as joining the digital health intervention, downloading applications (apps), or for sending and receiving mobile messages/phone calls.

ow, as cost could present a barrier to participation, particularly for young people and those on lower incomes. Participants felt that there should be little or no charge for costs such as joining the digital health intervention, downloading applications (apps), or for sending and receiving mobile messages/phone calls. Assessment for each CERQual component Methodological limitations No or very minor concerns about methodological limitations Coherence No or very minor concerns about coherence Relevance Minor concerns about relevance due to partial relevance in relation to participant group (adolescents focus) and/or in low or middle‐income settings where cost may be particularly important Adequacy No or very minor concerns about adequacy Overall CERQual assessment High confidence Due to minor concerns regarding relevance Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Cornelius 2009 USA; African‐American adolescents; SMS to support HIV/AIDS curriculum; hypothetical with no examples of programme content Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Mitchell 2016 USA; men who have sex with men; an app to motivate for HIV testing; hypothetical with no examples of programme content Perry 2012 USA; adolescents aged 15 to 20; SMS with preventative sexual health messages; evaluation or formative research on an existing mHealth programme that the participants have been using Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT

participants have been using Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Finding 9: Participants’ ability to access digital communication was sometimes limited by their language skills and their personal level of literacy or techno‐literacy, or both.

participants have been using Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Finding 9: Participants’ ability to access digital communication was sometimes limited by their language skills and their personal level of literacy or techno‐literacy, or both. Assessment for each CERQual component Methodological limitations Moderate concerns about methodological limitations due to poor reporting of sampling, participant voices in the findings, and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Minor concerns about relevance due to partial relevance of study population (populations that are more likely to have literacy and language challenges) Adequacy No or very minor concerns about adequacy Overall CERQual assessment Moderate confidence Due to minor concerns regarding relevance and moderate concerns regarding methodological limitations Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Curioso 2009 Peru; HIV‐positive adults receiving ART; SMS related to HIV/AIDS; hypothetical with no examples of programme content Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Hirsch‐Moverman 2017 Lesotho; HIV patients; SMS to provide real‐time adherence support to people on HIV and TB treatment; part of an RCT or pilot RCT Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Mbuagbaw 2014 Cameroon; individuals living with HIV or involved in HIV support work; community‐owned text messaging programme to support people living with HIV; hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Anti retrovira

ndia; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Anti retrovira l therapy (ART); Tuberculosis (TB) Finding 10: Participants often had preferences for how often health messages were sent, the time of day they were sent, and the duration of the digital targeted client communication. However, there was variation in what most participants felt was appropriate timing and frequency, and these preferences were often linked to the health issue on which the messaging was focused; whether people had their own phone or had to share a phone; and the participant’s particular circumstances. Participants were particularly concerned about being bombarded with too many messages; whether the timing of the messages was convenient for them; and/or whether messages arrived in connection with the behaviour the message was trying to target.

PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Mbuagbaw 2012 Cameroon; HIV‐positive patients; SMS for HIV drug adherence; part of an RCT or pilot RCT Mbuagbaw 2014 Cameroon; individuals living with HIV or involved in HIV support work; community‐owned text messaging programme to support people living with HIV; hypothetical with no examples of programme content Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Mitchell 2016 USA; men who have sex with men; an app to motivate for HIV testing; hypothetical with no examples of programme content Nachega 2016 South Africa; HIV‐infected pregnant women; SMS about ART adherence to prevent PMTCT; hypothetical with no examples of programme content Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Perry 2012 USA; adolescents aged 15 to 20; SMS with preventative sexual health messages; evaluation or formative research on an existing mHealth programme that the participants have been using Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Anti retroviral therapy (ARV); Human Papillomavirus (HPV); Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT)

e or had to share a phone; and the participant’s particular circumstances. Participants were particularly concerned about being bombarded with too many messages; whether the timing of the messages was convenient for them; and/or whether messages arrived in connection with the behaviour the message was trying to target. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of researcher reflexivity Coherence No or very minor concerns about coherence Relevance Moderate concerns about relevance due to a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy No or very minor concerns about adequacy Overall CERQual assessment Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Contributing studies Study Context Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Cornelius 2009 USA; African‐American adolescents; SMS to support HIV/AIDS curriculum; hypothetical with no examples of programme content Evans 2016 UK; African communities; SMS‐based HIV mHealth programme; hypothetical with examples and with no examples of programme content French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Gold 2010 Australia; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Mbuagbaw 2012 Cameroon; HIV‐positive patients; SMS for HIV drug adherence; part of an RCT or pilot RCT Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery; pilot or implementation study with participation in an mHealth programme Mitchell 2016 USA; men who have sex with men; an app to motivate for H

es and with no examples of programme content Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery; pilot or implementation study with participation in an mHealth programme Mitchell 2016 USA; men who have sex with men; an app to motivate for H IV testing; hypothetical with no examples of programme content Munro 2017 Canada; pregnant or have given birth in the last 12 months; SMS Text4baby programme about prenatal education; hypothetical with an example of messages being used Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Ware 2016 Uganda; HIV‐positive patients initiating ART; SMS for ART adherence; part of an RCT or pilot RCT Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT)

USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT) Finding 11: Participants had different preferences for various delivery channels available for sharing information through digital targeted client communication, including mobile messaging, interactive voice response, or speaking with a healthcare provider. These preferences were influenced by a number of factors including cost, convenience, the ability to store messages and re‐read them, familiarity with the channel, personal preferences, the nature of the content being delivered, the nature of the topic, language and literacy considerations, and the ability to have a discussion with a real‐life person. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of researcher reflexivity Coherence No or very minor concerns about coherence Relevance Moderate concerns about relevance due to a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention. However, they may still have had experience with the communication channel outside of an mHealth programme that they could draw on.

vance due to a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention. However, they may still have had experience with the communication channel outside of an mHealth programme that they could draw on. Adequacy No or very minor concerns about adequacy Overall CERQual assessment Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Cates 2015 USA; middle school students designing text messages to promote HPV vaccine; hypothetical with an example of messages being used Curioso 2009 Peru; HIV‐positive adults receiving ART; SMS related to HIV/AIDS; hypothetical with no examples of programme content Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery; pilot or implementation study with participation in an mHealth programme Mitchell 2016 USA; men who have sex with men; an app to motivate for HIV testing; hypothetical with no examples of programme content Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Willoughby 2017 USA; college students; SMS for sexual health promotion

interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Willoughby 2017 USA; college students; SMS for sexual health promotion ; hypothetical with an example of messages being used Anti retroviral therapy (ARV); Human Papillomavirus (HPV); Prevention of mother‐to‐child transmission (PMTCT) Finding 12: Participants appreciated personalised health information and discussed their preferences for options to make interventions more relevant to individuals. This could include sender‐based personalisation or receiver‐based options. Reasons for these preferences included engaging the user, enhancing credibility, increasing feelings of ownership, control over their personal information, and feelings of privacy. Preferences for tailoring included making digital health messages personalised by using an individual's name; allowing participants to choose the content, topic, and language of their messages; providing information relevant to the participant's setting (local information); allowing them to select the timing and frequency of the message; providing personalised reminders (e.g. for vaccination or medication); and allowing participants to have control over privacy settings.

topic, and language of their messages; providing information relevant to the participant's setting (local information); allowing them to select the timing and frequency of the message; providing personalised reminders (e.g. for vaccination or medication); and allowing participants to have control over privacy settings. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of researcher reflexivity Coherence No or very minor concerns about coherence Relevance Serious concerns about relevance due to a large number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy No or very minor concerns about adequacy Overall CERQual assessment Low confidence Due to minor concerns regarding methodological limitations and serious concerns regarding relevance Contributing studies Study Context Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Evans 2016 UK; African communities; SMS‐based HIV mHealth programme; hypothetical with examples and with no examples of programme content French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Goldenberg 2015 USA; men who have sex with men; SMS on HIV testing reminders; hypothetical with examples and with no examples of programme content Hirsch‐Moverman 2017 Lesotho; HIV patients; SMS to provide real‐time adherence support to people on HIV and TB treatment; part of an RCT or pilot RCT Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Munro 2017 Canada; pregnant or have given birth in the last 12 months; SMS Text4baby programme about prenatal education; hypothetical with an example of messages being used Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Ware 2016 Uganda; HIV‐positive patients initiating

ogramme Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Ware 2016 Uganda; HIV‐positive patients initiating ART; SMS for ART adherence; part of an RCT or pilot RCT Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT); Tuberculosis (TB) Finding 13: Participants mentioned various message formats that they preferred. These included a preference for short, concise, personalised, clear, and direct messages in a language they could understand and in full text rather than "text speak".

ART; SMS for ART adherence; part of an RCT or pilot RCT Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT); Tuberculosis (TB) Finding 13: Participants mentioned various message formats that they preferred. These included a preference for short, concise, personalised, clear, and direct messages in a language they could understand and in full text rather than "text speak". Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of participant voices in the findings and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Serious concerns about relevance due to partial relevance of study population (several of the studies were among adolescents) and a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy No or very minor concerns about adequacy Overall CERQual assessment Low confidence Due to minor concerns regarding methodological limitations and serious concerns regarding relevance Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Cates 2015 USA; middle school students designing text messages to promote HPV vaccine; hypothetical with an example of messages being used Curioso 2009 Peru; HIV‐positive adults receiving ART; SMS related to HIV/AIDS; hypothetical with no examples of programme content Evans 2016 UK; African communities; SMS‐based HIV mHealth programme; hypothetical with examples and with no examples of programme content French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Gold 2010 Australia; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Lau 2014 South Africa; pregnant women; SMS for antenatal health promotion; part of an RCT or pilot RCT Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of

voice call reminding of screening; hypothetical with an example of messages being used Lau 2014 South Africa; pregnant women; SMS for antenatal health promotion; part of an RCT or pilot RCT Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery; pilot or implementation study with participation in an mHealth programme Munro 2017 Canada; pregnant or have given birth in the last 12 months; SMS Text4baby programme about prenatal education; hypothetical with an example of messages being used Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Perry 2012 USA; adolescents aged 15 to 20; SMS with preventative sexual health messages; evaluation or formative research on an existing mHealth programme that the participants have been using Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Anti retroviral therapy (ART); Human Papiloma Virus (HPV); Randomized control trial (RCT)

V as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Anti retroviral therapy (ART); Human Papiloma Virus (HPV); Randomized control trial (RCT) Finding 14: Participants’ perceptions of who sent the digital health communication could influence their trust in and perception of the credibility and value of the digital targeted client communication and the information it provides. Participants said they wanted a known, identified phone number; messages sent from a reliable, trusted, credible source such as health professionals or official sources; and in some cases to feel like the messages were sent by a person (even if sent from an automated service). However, some participants, such as those with stigmatised health conditions, preferred an unmarked sender to protect their privacy.

ble, trusted, credible source such as health professionals or official sources; and in some cases to feel like the messages were sent by a person (even if sent from an automated service). However, some participants, such as those with stigmatised health conditions, preferred an unmarked sender to protect their privacy. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of participant voices in the findings and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Moderate concerns about relevance due to a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy No or very minor concerns about adequacy Overall CERQual assessment Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Brown 2014 USA; single, adolescent mothers; health promotion information weekly via SMS during the first 6 months postpartum; pilot or implementation study with participation in an mHealth programme Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Cates 2015 USA; middle school students designing text messages to promote HPV vaccine; hypothetical with an example of messages being used Evans 2016 UK; African communities; SMS‐based HIV mHealth programme; hypothetical with examples and with no examples of programme content Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Lau 2014 South Africa; pregnant women; SMS for antenatal health promotion; part of an RCT or pilot RCT Mbuagbaw 2012 Cameroon; HIV‐positive patients; SMS for HIV drug adherence; part of an RCT or pilot RCT Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for

r HIV drug adherence; part of an RCT or pilot RCT Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery; pilot or implementation study with participation in an mHealth programme Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Human Papiloma Virus (HPV) Finding 15: Participants said that the tone of digital health communication mattered to them. Their preferences varied but included a tone that was: motivational, friendly, encouraging, polite, respectful, congratulatory, personalised, upbeat, positive, humorous, and relatable. Some participants highlighted that they did not like feeling pressured, lectured to, shamed, or frightened by digital health messages.

Their preferences varied but included a tone that was: motivational, friendly, encouraging, polite, respectful, congratulatory, personalised, upbeat, positive, humorous, and relatable. Some participants highlighted that they did not like feeling pressured, lectured to, shamed, or frightened by digital health messages. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of researcher reflexivity Coherence No or very minor concerns about coherence Relevance Serious concerns about relevance due to partial relevance of study population (several of the studies were among adolescents) and a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy No or very minor concerns about adequacy Overall CERQual assessment Low confidence Due to minor concerns regarding methodological limitations and serious concerns regarding relevance Contributing studies Study Context Cates 2015 USA; middle school students designing text messages to promote HPV vaccine; hypothetical with an example of messages being used Curioso 2009 Peru; HIV‐positive adults receiving ART; SMS related to HIV/AIDS; hypothetical with no examples of programme content Evans 2016 UK; African communities; SMS‐based HIV mHealth programme; hypothetical with examples and with no examples of programme content French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Gold 2010 Australia; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Munro 2017 Canada; pregnant or have given birth in the last 12 months; SMS Text4baby programme about prenatal education; hypothetical with an example of messages being used Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Perry 2012 USA; adolescents aged

aughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Perry 2012 USA; adolescents aged 15 to 20; SMS with preventative sexual health messages; evaluation or formative research on an existing mHealth programme that the participants have been using Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Anti retroviral therapy (ARV); Human Papillomavirus (HPV); Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT) Finding 16: Participants had preferences regarding the content they receive through digital targeted client communication. They wanted varied content that provided new knowledge and reminders, as well as explanations, solutions, and suggestions about health issues. They were interested in content related to health, illness, and treatments and practical topics such as health facility location and transportation. They wanted this information to be relevant and acceptable to their personal circumstances and local setting.

ort for postabortion care; part of an RCT or pilot RCT Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Anti retroviral therapy (ARV); Human Papillomavirus (HPV); Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT) Finding 19: Some participants proposed strategies to address their concerns regarding confidentiality and privacy. These strategies for communication included neutral, coded, or discreet language; access codes; communication that does not disclose the sender; coming from a trusted sender; and the ability to tailor and control content, timing, and frequency of their messages.

explanations, solutions, and suggestions about health issues. They were interested in content related to health, illness, and treatments and practical topics such as health facility location and transportation. They wanted this information to be relevant and acceptable to their personal circumstances and local setting. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of researcher reflexivity Coherence No or very minor concerns about coherence Relevance Moderate concerns about relevance due to a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy No or very minor concerns about adequacy Overall CERQual assessment Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Contributing studies Study Context Brown 2014 USA; single, adolescent mothers; health promotion information weekly via SMS during the first 6 months postpartum; pilot or implementation study with participation in an mHealth programme Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Cornelius 2009 USA; African‐American adolescents; SMS to support HIV/AIDS curriculum; hypothetical with no examples of programme content Entsieh 2015 Ghana; pregnant and nursing mothers aged 20 to 35; “Mobile Midwife” app; qualitative research on an existing programme implemented at scale French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Gold 2010 Australia; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Mbuagbaw 2014 Cameroon; individuals living with HIV or involved in HIV support work; community‐owned text messaging programme to support p

T or pilot RCT Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Mbuagbaw 2014 Cameroon; individuals living with HIV or involved in HIV support work; community‐owned text messaging programme to support p eople living with HIV; hypothetical with no examples of programme content Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery; pilot or implementation study with participation in an mHealth programme Mitchell 2016 USA; men who have sex with men; an app to motivate for HIV testing; hypothetical with no examples of programme content Munro 2017 Canada; pregnant or have given birth in the last 12 months; SMS Text4baby programme about prenatal education; hypothetical with an example of messages being used Nachega 2016 South Africa; HIV‐infected pregnant women; SMS about ART adherence to prevent PMTCT; hypothetical with no examples of programme content Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Perry 2012 USA; adolescents aged 15 to 20; SMS with preventative sexual health messages; evaluation or formative research on an existing mHealth programme that the participants have been using Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Anti retroviral therapy (ARV); Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT)

during pregnancy; part of an RCT or pilot RCT Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Anti retroviral therapy (ARV); Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT) Finding 17: Some participants felt that including elements in the mobile‐based platform in which participants are asked for a response (e.g. via knowledge quizzes or multiple‐choice questions or a practical tool allowing access to additional information, such as a nutrition calculator) could increase the engagement of users with the intervention, its content, and provide additional information to them. In one study, participants suggested that it would be helpful if the response was quick, simple, and convenient. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations, as most studies were fairly well conducted and reported (the lack of reflexivity in 3 of the studies is not a serious concern because of the focus of the finding) Coherence No or very minor concerns about coherence Relevance Serious concerns about relevance due to a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention; all of the studies were conducted in high‐income countries, and most of the studies were on adolescent and youth populations Adequacy Moderate concerns about adequacy due to the small number of studies and thin data Overall CERQual assessment Low confidence Due to minor concerns regarding methodological limitations, moderate concerns regarding adequacy, and serious concerns regarding relevance Contributing studies Study Context Cornelius 2009 USA; African‐American adolescents; SMS to support HIV/AIDS curriculum; hypothetical with no examples of programme content Munro 2017 Canada; pregnant or have given birth in the last 12 months; SMS Text4baby programme about prenatal education; hypothetical with an example of messages being used Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used

s being used Naughton 2013 UK; women who smoked during a recent pregnancy; SMS for smoking cessation during pregnancy; pilot or implementation study with participation in an mHealth programme Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Finding 18: Some participants with health issues that are often seen as stigmatised or very personal (e.g. HIV, family planning, and abortion care) worried that their confidential health information would be disclosed or their identity traced due to their participation in digital targeted client communication. In general, people’s perceptions of information delivery channels (SMS, interactive voice response, voice call) were influenced by how confidential they felt the delivery channels to be.

l health information would be disclosed or their identity traced due to their participation in digital targeted client communication. In general, people’s perceptions of information delivery channels (SMS, interactive voice response, voice call) were influenced by how confidential they felt the delivery channels to be. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of researcher reflexivity Coherence No or very minor concerns about coherence Relevance No or very minor concerns about relevance Adequacy No or very minor concerns about adequacy Overall CERQual assessment High confidence Due to minor concerns regarding methodological limitations Contributing studies Study Context Akinfaderin‐Agarau 2012 Nigeria; adolescent girls and young women; using mobile phones to provide sexual and reproductive health information and services; hypothetical with no examples of programme content Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Cates 2015 USA; middle school students designing text messages to promote HPV vaccine; hypothetical with an example of messages being used Curioso 2009 Peru; HIV‐positive adults receiving ART; SMS related to HIV/AIDS; hypothetical with no examples of programme content Evans 2016 UK; African communities; SMS‐based HIV mHealth programme; hypothetical with examples and with no examples of programme content French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Goldenberg 2015 USA; men who have sex with men; SMS on HIV testing reminders; hypothetical with examples and with no examples of programme content Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Mbuagbaw 2012 Cameroon; HIV‐positive patients; SMS for HIV drug adherence; part of an RCT or pilot RCT Mbuagbaw 2014 Cameroon; individuals living with HIV or involved in HIV support work; community‐owned

ess their concerns regarding confidentiality and privacy. These strategies for communication included neutral, coded, or discreet language; access codes; communication that does not disclose the sender; coming from a trusted sender; and the ability to tailor and control content, timing, and frequency of their messages. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of participant voices in the findings and of researcher reflexivity Coherence No or very minor concerns about coherence Relevance No or very minor concerns about relevance Adequacy No or very minor concerns about adequacy Overall CERQual assessment High confidence Due to minor concerns regarding methodological limitations Contributing studies Study Context Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Curioso 2009 Peru; HIV‐positive adults receiving ART; SMS related to HIV/AIDS; hypothetical with no examples of programme content Evans 2016 UK; African communities; SMS‐based HIV mHealth programme; hypothetical with examples and with no examples of programme content French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Goldenberg 2015 USA; men who have sex with men; SMS on HIV testing reminders; hypothetical with examples and with no examples of programme content Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Mbuagbaw 2012 Cameroon; HIV‐positive patients; SMS for HIV drug adherence; part of an RCT or pilot RCT Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Odeny 2014 Kenya; women; SMS for early infant HIV testing; hypothetical with no examples of programme content Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Willoughby 2017 USA; college studen

n the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Willoughby 2017 USA; college studen ts; SMS for sexual health promotion; hypothetical with an example of messages being used Anti retroviral therapy (ARV); Randomized control trial (RCT) Finding 20: Some participants thought that participating in digital targeted client communication had influenced their behaviour, whilst others did not. Reasons given for the changes in behaviour included receiving new knowledge; receiving strategies on how to initiate discussion with a partner or healthcare provider; being motivated or reassured by the intervention; and being reminded, for example, to take medication or make an appointment. Some participants who believed that the intervention did not have any influence on their behaviour found that the digital health interventions were not relevant to them.

or healthcare provider; being motivated or reassured by the intervention; and being reminded, for example, to take medication or make an appointment. Some participants who believed that the intervention did not have any influence on their behaviour found that the digital health interventions were not relevant to them. Assessment for each CERQual component Methodological limitations Moderate concerns about methodological limitations due to poor reporting of participant voice in the findings, ethical considerations, and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Minor concerns about relevance due to the fact that a large group of the studies were tied to pregnancy and childbirth, which can in itself influence behaviour change Adequacy Minor concerns about adequacy due to thin data in some studies Overall CERQual assessment Low confidence Due to minor concerns regarding relevance and adequacy and moderate concerns regarding methodological limitations Contributing studies Study Context Brown 2014 USA; single, adolescent mothers; health promotion information weekly via SMS during the first 6 months postpartum; pilot or implementation study with participation in an mHealth programme Entsieh 2015 Ghana; pregnant and nursing mothers aged 20 to 35; “Mobile Midwife” app; qualitative research on an existing programme implemented at scale French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Gold 2010 Australia; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Greaney 2014 USA; Latina women over the age of 21 needing cancer screening; interactive voice call reminding of screening; hypothetical with an example of messages being used Hirsch‐Moverman 2017 Lesotho; HIV patients; SMS to provide real‐time adherence support to people on HIV and TB treatment; part of an RCT or pilot RCT Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Lau 2014 South Africa; pregnant women; SMS for antenatal health promotion; part of an RCT or pilot RCT Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery;

; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Lau 2014 South Africa; pregnant women; SMS for antenatal health promotion; part of an RCT or pilot RCT Missal 2016 India; husbands of pregnant women 12 to 20 weeks along; voice messages about antenatal care and preparing for delivery; pilot or implementation study with participation in an mHealth programme Munro 2017 Canada; pregnant or have given birth in the last 12 months; SMS Text4baby programme about prenatal education; hypothetical with an example of messages being used Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Ware 2016 Uganda; HIV‐positive patients initiating ART; SMS for ART adherence; part of an RCT or pilot RCT Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT); Tuberculosis (TB)

bortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Ware 2016 Uganda; HIV‐positive patients initiating ART; SMS for ART adherence; part of an RCT or pilot RCT Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT); Tuberculosis (TB) Finding 21: Some participants suggested that the effects of the messaging may not be sustained over time, as they and others would become bored with or fatigued by the messages, especially if the content was not varied enough. Assessment for each CERQual component Methodological limitations No or very minor concerns about methodological limitations Coherence No or very minor concerns about coherence Relevance Moderate concerns due to the fact that in the majority of studies participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention, and a focus on HIV/AIDS (partial relevance) Adequacy Moderate concerns due to thin data in some of the included studies Overall CERQual assessment Low confidence Due to moderate concerns regarding relevance and adequacy Contributing studies Study Context Cornelius 2009 USA; African‐American adolescents; SMS to support HIV/AIDS curriculum; hypothetical with no examples of programme content Curioso 2009 Peru; HIV‐positive adults receiving ART; SMS related to HIV/AIDS; hypothetical with no examples of programme content Evans 2016 UK; African communities; SMS‐based HIV mHealth programme; hypothetical with examples and with no examples of programme content Gold 2010 Australia; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Menacho 2013 Peru; men who have sex with men; SMS to motivate for HIV testing; hypothetical with examples and with no examples of programme content Mitchell 2016 USA; men who have sex with men; an app to motivate for HIV testing; hypothetical with no examples of programme content Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Anti retroviral therapy (ART)

th no examples of programme content Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Willoughby 2017 USA; college students; SMS for sexual health promotion; hypothetical with an example of messages being used Anti retroviral therapy (ART) Finding 22: Some participants were concerned about becoming over‐reliant on digital reminders and thought that this might make them dependent on digital targeted communication for undertaking some health tasks. They were concerned that in the absence of these reminders they would adhere poorly to care plans. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of methods in 1 study Coherence No or very minor concerns about coherence Relevance Moderate concerns about relevance due to the fact that all of the studies are from 1 region; 2 focus on 1 health issue (HIV); and 1 is hypothetical Adequacy Serious concerns about adequacy due to thin data from a small number of studies Overall CERQual assessment Low confidence Due to minor concerns regarding methodological limitations, moderate concerns about relevance, and serious concerns about adequacy Contributing studies Study Context Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Mbuagbaw 2012 Cameroon; HIV‐positive patients; SMS for HIV drug adherence; part of an RCT or pilot RCT Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Anti retroviral therapy (ART)

al care and family planning; part of an RCT or pilot RCT Mbuagbaw 2012 Cameroon; HIV‐positive patients; SMS for HIV drug adherence; part of an RCT or pilot RCT Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Anti retroviral therapy (ART) Finding 23: Some participants felt that digital health interventions could save them time and money by giving them access to health care via their mobile phones. This was especially relevant to participants who faced barriers in attending health care because of distance to a health facility and a lack of time and/or financial means. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of participant voices in the findings, researcher reflexivity, and unclear ethical considerations in 1 study Coherence No or very minor concerns about coherence Relevance Moderate concerns about relevance due to partial relevance of setting and populations who may be particularly affected by lack of time and funds, and distance Adequacy Moderate concerns about adequacy due to a limited number of studies Overall CERQual assessment Low confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding adequacy and relevance Contributing studies Study Context Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT

text Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Finding 24: Some participants felt that digital health interventions provided them with feelings of support and connectedness, as they felt that someone was taking the time to send them messages. A few participants felt that in some cases the sense of caring and support that they received from healthcare providers through digital health interventions had a positive influence on their relationship with their healthcare provider.

connectedness, as they felt that someone was taking the time to send them messages. A few participants felt that in some cases the sense of caring and support that they received from healthcare providers through digital health interventions had a positive influence on their relationship with their healthcare provider. Assessment for each CERQual component Methodological limitations Moderate concerns about methodological limitations due to poor reporting of participant voice in the findings, ethical considerations, and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Moderate concerns about relevance due to a fair number of studies where participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy No or very minor concerns about adequacy Overall CERQual assessment Moderate confidence Due to moderate concerns regarding methodological limitations and relevance Contributing studies Study Context Brown 2014 USA; single, adolescent mothers; health promotion information weekly via SMS during the first 6 months postpartum; pilot or implementation study with participation in an mHealth programme Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Entsieh 2015 Ghana; pregnant and nursing mothers aged 20 to 35; “Mobile Midwife” app; qualitative research on an existing programme implemented at scale Jalloh‐Vos 2014 Sierra Leone; pregnant and postpartum women and their partners; mobile phone intervention for antenatal care and family planning; part of an RCT or pilot RCT Lau 2014 South Africa; pregnant women; SMS for antenatal health promotion; part of an RCT or pilot RCT Mbuagbaw 2014 Cameroon; individuals living with HIV or involved in HIV support work; community‐owned text messaging programme to support people living with HIV; hypothetical with no examples of programme content Munro 2017 Canada; pregnant or have given birth in the last 12 months; SMS Text4baby programme about prenatal education; hypothetical with an example of messages being used Nachega 2016 South Africa; HIV‐infected pregnant women; SMS about ART adherence to prevent PMTCT; hypothetical with no examples of programme content Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in

ed Nachega 2016 South Africa; HIV‐infected pregnant women; SMS about ART adherence to prevent PMTCT; hypothetical with no examples of programme content Rana 2015 Uganda; HIV‐positive youth receiving ART; SMS for HIV‐positive youth; Hypothetical with no examples of programme content Rodrigues 2015 India; participants in the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Ware 2016 Uganda; HIV‐positive patients initiating ART; SMS for ART adherence; part of an RCT or pilot RCT Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Anti retroviral therapy (ART); Prevention of mother‐to‐child transmission (PMTCT) Finding 25: Participants described how they shared digital communication content more broadly with friends, family, and community members. Many participants felt that the information would be useful to others.

the intervention arm of the trial; interactive voice recordings and SMS for HIV ART adherence; part of an RCT or pilot RCT Sloan 2017 UK; women who had received the MiQuit intervention during pregnancy; SMS for smoking cessation during pregnancy; part of an RCT or pilot RCT Smillie 2014 Canada; HIV‐positive people; SMS about HIV as part of the WelTel BC trial; pilot or implementation study with participation in an mHealth programme Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Ware 2016 Uganda; HIV‐positive patients initiating ART; SMS for ART adherence; part of an RCT or pilot RCT Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Anti retroviral therapy (ART); Prevention of mother‐to‐child transmission (PMTCT) Finding 25: Participants described how they shared digital communication content more broadly with friends, family, and community members. Many participants felt that the information would be useful to others. Assessment for each CERQual component Methodological limitations Minor concerns about methodological limitations due to poor reporting of participant voices in the findings and researcher reflexivity Coherence No or very minor concerns about coherence Relevance Moderate concerns about relevance due to the fact that in a fair number of studies participants did not experience an mHealth intervention but were asked to comment about their preferences regarding a hypothetical intervention Adequacy No or very minor concerns about adequacy Overall CERQual assessment Moderate confidence Due to minor concerns regarding methodological limitations and moderate concerns regarding relevance Contributing studies Study Context Calderon 2017 Peru; women over 18 who had at least 1 child; SMS‐based mHealth programme on child health; hypothetical with no examples of programme content Cornelius 2009 USA; African‐American adolescents; SMS to support HIV/AIDS curriculum; hypothetical with no examples of programme content Flax 2017 Nigeria; women of all ages belonging to microcredit financing groups; received weekly cell phone breastfeeding text and voice messages to a shared phone; part of an RCT or pilot RCT French 2016 UK; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Gold 2010 Australia; young people aged 16 to 24; SMS on sexually transmitted infections; part of an RCT or pilot RCT Jennings 2013 Kenya; HIV‐positive women enrolled in PMTCT and their male partners; SMS reminder for PMTCT testing; hypothetical with an example of messages being used Perry 2012 USA; adolescents aged 15 to 20; SMS with preventative sexual health messages; evaluation or formative research on an existing mHealth programme that the participants have been using Smith 2017 Cambodia; women who had received an abortion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT)

tion; mobile phone voice messaging and counsellor support for postabortion care; part of an RCT or pilot RCT Wright 2011 USA; African‐American men aged 16 to 20; SMS for HIV prevention; hypothetical with an example of messages being used Prevention of mother‐to‐child transmission (PMTCT); Randomized control trial (RCT) Appendix 5. Expanded matrix table Have the trialists described efforts to address situations where members of the target group: 1. do not own a functioning mobile device; 2. have poor access to network services; 3. have poor access to electricity to charge mobile devices; 4. want to avoid expenses associated with the intervention, such as paying for airtime; 5. change their phone numbers or sim cards; 6. have access to the phone controlled by someone else; 7. have low literacy, differing language skills, or limited techno‐literacy; 8. have concerns about privacy and confidentiality; 9. perceive different sources as more or less reliable, trusted, and credible; 10. have members of the target group been given an opportunity to offer feedback about their needs, preferences, and experiences regarding the intervention. Adolescents (N = 13) Trial 1 2 3 4 5 6 7 8 9 10 1 Belzer 2015 (USA) Y Y N Y 2 Bull 2016 (USA) Y Y Y Y Y Y Y Y Y 3 Castano 2012 (USA) N Y ? Y Y N 4 Delamere 2006 (Ireland) 5 Garofalo 2016 (USA) N Y Y Y Y Y 6 Gold 2011 (Australia) N Y Y Y 7 Jeffries 2016 (USA) N N ? 8 Lim 2012 (Australia) N N N Y 9 McCarthy 2016 (UK) N N N Y Y 10 Reed 2014 (USA) N Y 11 Rokicki 2017 (Ghana) Y ? ? ? Y 12 Suffoletto 2013 (USA) N ? N ? 13 Ybarra 2017 (USA) N N N N ? Y ? Adults (N = 27) Trial 1 2 3 4 5 6 7 8 9 10 14 Abdul 2013 (Malaysia) N 15 Barnabas 2016 (South Africa/Uganda) N Y Y Y Y 16 Constant 2014 (South Africa) N Y N Y ? Y 17 Cook 2015 (USA) Y Y N Y 18 de Costa 2012 (Brazil) N Y N N Y Y 19 de Tolly 2012 (South Africa) N Y N Y 20 Downing 2013 (Australia) N N 21 Gerdts 2015 (Colombia) 22 Hou 2010 (USA) N Y ? Y 23 Huang 2013 (China) N Y 24 Ingersoll 2015 (USA) Y Y Y Y Y Y 25 Joseph 2016 (Mozambique) N N N N Y Y 26 Lee 2016 (USA) Y Y Y 27 Leiby 2016 (Zambia) N Y Y 28 Lester 2010 (Kenya) Y Y Y Y Y 29 Mbuagbaw 2012 (Cameroon) N N ? Y Y Y 30 Mugo 2016 (Kenya) Y Y ? Y 31 Norton 2014 (USA) N N N N 32 Nsagha 2016 (Cameroon) N ? ? 33 Odeny 2012 (Kenya) N Y N ? Y 34 Pop‐Eleches 2011 (Kenya) Y Y Y Y Y 35 Ruan 2017 (China) N N N Y Y 36 Russell 2012 (USA) 37 Rutland 2012 (UK) 38 Shet 2014 (India) Y N Y Y Y 39 Smith 2015 (Cambodia) N N N Y ? ? Y 40 Young 2015 (Peru) ?

o 2016 (Kenya) Y Y ? Y 31 Norton 2014 (USA) N N N N 32 Nsagha 2016 (Cameroon) N ? ? 33 Odeny 2012 (Kenya) N Y N ? Y 34 Pop‐Eleches 2011 (Kenya) Y Y Y Y Y 35 Ruan 2017 (China) N N N Y Y 36 Russell 2012 (USA) 37 Rutland 2012 (UK) 38 Shet 2014 (India) Y N Y Y Y 39 Smith 2015 (Cambodia) N N N Y ? ? Y 40 Young 2015 (Peru) ? Y Y Y Pregnant and postpartum women (N = 11) Trial 1 2 3 4 5 6 7 8 9 10 41 Cooper 2015 (England) N Y ? Y Y 42 Evans 2014 (USA) N N Y 43 Jareethum 2008 (Thailand) N ? 44 Joshi 2015 (India) Y Y Y Y 45 Kamau‐Mbuthia 2013 (Kenya) 46 Lund 2012 (Zanzibar) N Y N Y N N 47 Maslowsky 2016 (Ecuador) Y Y Y Y 48 McConnell 2016 (Kenya) Y Y ? 49 Moniz 2013 (USA) N ? 50 Omole 2018 (Nigeria) N Y N ? 51 Yudin 2017 (Canada) N N Y Parents (N = 14) Trial 1 2 3 4 5 6 7 8 9 10 52 Ahlers‐Schmidt 2012a (USA) N N Y 23 Bangure 2015 (Zimbabwe) N ? Y 54 Bigna 2015 (Cameroon) N N ? Y 55 Brown 2016 (Nigeria) Y Y 56 Domek 2016 (Guatemala) N Y Y 57 Eze 2015 (Nigeria) N N Y 58 Gibson 2017 (Kenya) Y Y 59 Haji 2016 (Kenya) N Y 60 Hannan 2016 (USA) 61 Hofstetter 2015 (USA) N N Y 62 Jimenez 2017 (USA) ? Y 63 Niederhauser 2015 (USA) N N 64 Sharma 2011 (India) N N 65 Stockwell 2014 (USA) N Y Mothers living with HIV/AIDS (N = 3) Trial 1 2 3 4 5 6 7 8 9 10 66 Kassaye 2016 (Kenya) Y Y ? Y ? Y 67 Kebaya 2015 (Kenya) 68 Odeny 2014 (Kenya) Y Y Y Y Y N = No; Y = Yes; Blank = Not mentioned; ? = Unclear Appendix 6. List of all studies included in the matrix analysis Studies included in the matrix analysis Adolescents (N = 13) Belzer 2015

Y Y Y Pregnant and postpartum women (N = 11) Trial 1 2 3 4 5 6 7 8 9 10 41 Cooper 2015 (England) N Y ? Y Y 42 Evans 2014 (USA) N N Y 43 Jareethum 2008 (Thailand) N ? 44 Joshi 2015 (India) Y Y Y Y 45 Kamau‐Mbuthia 2013 (Kenya) 46 Lund 2012 (Zanzibar) N Y N Y N N 47 Maslowsky 2016 (Ecuador) Y Y Y Y 48 McConnell 2016 (Kenya) Y Y ? 49 Moniz 2013 (USA) N ? 50 Omole 2018 (Nigeria) N Y N ? 51 Yudin 2017 (Canada) N N Y Parents (N = 14) Trial 1 2 3 4 5 6 7 8 9 10 52 Ahlers‐Schmidt 2012a (USA) N N Y 23 Bangure 2015 (Zimbabwe) N ? Y 54 Bigna 2015 (Cameroon) N N ? Y 55 Brown 2016 (Nigeria) Y Y 56 Domek 2016 (Guatemala) N Y Y 57 Eze 2015 (Nigeria) N N Y 58 Gibson 2017 (Kenya) Y Y 59 Haji 2016 (Kenya) N Y 60 Hannan 2016 (USA) 61 Hofstetter 2015 (USA) N N Y 62 Jimenez 2017 (USA) ? Y 63 Niederhauser 2015 (USA) N N 64 Sharma 2011 (India) N N 65 Stockwell 2014 (USA) N Y Mothers living with HIV/AIDS (N = 3) Trial 1 2 3 4 5 6 7 8 9 10 66 Kassaye 2016 (Kenya) Y Y ? Y ? Y 67 Kebaya 2015 (Kenya) 68 Odeny 2014 (Kenya) Y Y Y Y Y N = No; Y = Yes; Blank = Not mentioned; ? = Unclear Appendix 6. List of all studies included in the matrix analysis Studies included in the matrix analysis Adolescents (N = 13) Belzer 2015 Belzer ME, Kolmodin M, Clark LF, Huang J, Olson J, Kahana SY, et al. Acceptability and feasibility of a cell phone support intervention for youth living with HIV with nonadherence to antiretroviral therapy. AIDS Patient Care & STDs 2015;29(6):338‐45.

Appendix 6. List of all studies included in the matrix analysis Studies included in the matrix analysis Adolescents (N = 13) Belzer 2015 Belzer ME, Kolmodin M, Clark LF, Huang J, Olson J, Kahana SY, et al. Acceptability and feasibility of a cell phone support intervention for youth living with HIV with nonadherence to antiretroviral therapy. AIDS Patient Care & STDs 2015;29(6):338‐45. Belzer ME, Naar‐King S, Olson J, Sarr M, Thornton S, Kahana SY, et al. The use of cell phone support for non‐adherent HIV‐infected youth and young adults: an initial randomized and controlled intervention trial. AIDS and Behavior 2014;18(4):686‐96. Bull 2016 Bull S, Devine S, Schmiege SJ, Pickard L, Campbell J, Shlay JC, et al. Text messaging, teen outreach program, and sexual health behavior: a cluster randomized trial. American Journal of Public Health 2016;106(S1):S117‐24. Bull S, Devine S, Schmiege SJ, Hammes A, Pickard L, Shlay JC. Text messaging and teen sexual health behavior: long‐term follow‐up of a cluster randomized trial. Computers, Informatics, Nursing 2017;35(11):549‐53. Devine S, Bull S, Dreisbach S, Shlay J. Enhancing a teen pregnancy prevention program with text messaging: engaging minority youth to develop TOP® Plus Text. Journal of Adolescent Health 2014;54(3):S78‐83. Devine S, Leeds C, Shlay JC, Leytem A, Beum R, Bull S. Methods to assess youth engagement in a text messaging supplement to an effective teen pregnancy program. Journal of Biomedical Informatics 2015;56:379‐86.

Devine S, Bull S, Dreisbach S, Shlay J. Enhancing a teen pregnancy prevention program with text messaging: engaging minority youth to develop TOP® Plus Text. Journal of Adolescent Health 2014;54(3):S78‐83. Devine S, Leeds C, Shlay JC, Leytem A, Beum R, Bull S. Methods to assess youth engagement in a text messaging supplement to an effective teen pregnancy program. Journal of Biomedical Informatics 2015;56:379‐86. Bull S, Devine S, Schmiege SJ, Pickard L, Campbell J, Shlay JC. Text messaging, teen outreach program, and sexual health behavior: a cluster randomized trial. American Journal of Public Health 2016;106(S1):S117‐24. Castano 2012 Castano PM, Bynum JY, Andres R, Lara M, Westhoff C. Effect of daily text messages on oral contraceptive continuation: a randomized controlled trial. Obstetrics & Gynecology 2012;119(1):14‐20. Hall K, Castano P, Westhoff C. Oral contraceptive knowledge modestly associated with oral contraceptive continuation among young, urban women. Contraception 2011;84(3):320‐1. Hall KS, Castano PM, Westhoff CL. The influence of oral contraceptive knowledge on oral contraceptive continuation among young women. Journal of Women's Health 2014;23(7):596‐601. Hall KS, Westhoff CL, Castano PM. The impact of an educational text message intervention on young urban women's knowledge of oral contraception. Contraception 2013;87(4):449‐54. Delamere 2006 Delamere S, Dooley S, Harrington L, King A, Mulcahy F. Safer sex text messages: evaluating a health education intervention in an adolescent population. Sexually Transmitted Infections 2006;82:A27. Garofalo 2016

Hall KS, Westhoff CL, Castano PM. The impact of an educational text message intervention on young urban women's knowledge of oral contraception. Contraception 2013;87(4):449‐54. Delamere 2006 Delamere S, Dooley S, Harrington L, King A, Mulcahy F. Safer sex text messages: evaluating a health education intervention in an adolescent population. Sexually Transmitted Infections 2006;82:A27. Garofalo 2016 Garofalo R, Kuhns LM, Hotton A, Johnson A, Muldoon A, Rice D. A randomized controlled trial of personalized text message reminders to promote medication adherence among HIV‐positive adolescents and young adults. AIDS and Behavior 2016;20(5):1049‐59. Dowshen N, Kuhns LM, Gray C, Lee S, Garofalo R. Feasibility of interactive text message response (ITR) as a novel, real‐time measure of adherence to antiretroviral therapy for HIV+ youth. AIDS and Behavior 2013;17(6):2237‐43. Dowshen N, Kuhns LM, Johnson A, Holoyda BJ, Garofalo R. Improving adherence to antiretroviral therapy for youth living with HIV/AIDS: a pilot study using personalized, interactive, daily text message reminders. Journal of Medical Internet Research 2012;14(2):e51. Gold 2011 Gold J, Aitken CK, Dixon HG, Lim MS, Gouillou M, Spelman T, et al. A randomised controlled trial using mobile advertising to promote safer sex and sun safety to young people. Health Education Research 2011;26(5):782‐94. Jeffries 2016 Jeffries C, Ross P, Matoff‐Stepp S, Thompson R, Harris JL, Uhrig JD, et al. Ucare4life: mobile texting to improve HIV care continuum outcomes for minority youth. Topics in antiviral medicine 2016;24:427. Lim 2012

Gold J, Aitken CK, Dixon HG, Lim MS, Gouillou M, Spelman T, et al. A randomised controlled trial using mobile advertising to promote safer sex and sun safety to young people. Health Education Research 2011;26(5):782‐94. Jeffries 2016 Jeffries C, Ross P, Matoff‐Stepp S, Thompson R, Harris JL, Uhrig JD, et al. Ucare4life: mobile texting to improve HIV care continuum outcomes for minority youth. Topics in antiviral medicine 2016;24:427. Lim 2012 Lim MS, Hocking JS, Aitken CK, Fairley CK, Jordan L, Lewis JA, et al. Impact of text and email messaging on the sexual health of young people: a randomised controlled trial. Journal of Epidemiology & Community Health 2012;66(1):69‐74. McCarthy 2016 Free C, McCarthy O, French RS, Wellings K, Michie S, Roberts I. Can text messages increase safer sex behaviours in young people? Intervention development and pilot randomised controlled trial. Health Technology Assessment 2016;20(57):1‐82. McCarthy OL, French RS, Baraitser P, Roberts I, Rathod SD, Devries K, et al. Safetxt: a pilot randomised controlled trial of an intervention delivered by mobile phone to increase safer sex behaviours in young people. BMJ Open 2016;6(12):e013045. French RS, McCarthy O, Baraitser P, Wellings K, Bailey JV, Free C. Young people’s views and experiences of a mobile phone texting intervention to promote safer sex behavior. JMIR mHealth and uHealth 2016;4(2):e26. Reed 2014 Reed JL, Huppert JS, Taylor RG, Gillespie GL, Byczkowski TL, Kahn JA, et al. Improving sexually transmitted infection results notification via mobile phone technology. Journal of Adolescent Health 2014;55(5):690‐7.

French RS, McCarthy O, Baraitser P, Wellings K, Bailey JV, Free C. Young people’s views and experiences of a mobile phone texting intervention to promote safer sex behavior. JMIR mHealth and uHealth 2016;4(2):e26. Reed 2014 Reed JL, Huppert JS, Taylor RG, Gillespie GL, Byczkowski TL, Kahn JA, et al. Improving sexually transmitted infection results notification via mobile phone technology. Journal of Adolescent Health 2014;55(5):690‐7. Rokicki 2017 Rokicki S, Cohen J, Salomon JA, Fink G. Impact of a text‐messaging program on adolescent reproductive health: a cluster‐randomized trial in Ghana. American Journal of Public Health 2017;107(2):298‐305. Rokicki S, Fink G. Assessing the reach and effectiveness of mHealth: evidence from a reproductive health program for adolescent girls in Ghana. BMC Public Health 2017;17(1):969. Suffoletto 2013 Suffoletto B, Akers A, McGinnis KA, Calabria J, Wiesenfeld HC, Clark DB. A sex risk reduction text‐message program for young adult females discharged from the emergency department. Journal of Adolescent Health 2013;53(3):387‐93. Ybarra 2017 Ybarra ML, Prescott TL, Phillips GL Jr, Bull SS, Parsons JT, Mustanski B. Pilot RCT results of an mHealth HIV prevention program for sexual minority male adolescents. Pediatrics 2017;140(1):e20162999. Prescott TL, Phillips G Jr, DuBois LZ, Bull SS, Mustanski B, Ybarra ML. Reaching adolescent gay, bisexual, and queer men online: development and refinement of a national recruitment strategy. Journal of Medical Internet Research 2016;18(8):e200.

Ybarra ML, Prescott TL, Phillips GL Jr, Bull SS, Parsons JT, Mustanski B. Pilot RCT results of an mHealth HIV prevention program for sexual minority male adolescents. Pediatrics 2017;140(1):e20162999. Prescott TL, Phillips G Jr, DuBois LZ, Bull SS, Mustanski B, Ybarra ML. Reaching adolescent gay, bisexual, and queer men online: development and refinement of a national recruitment strategy. Journal of Medical Internet Research 2016;18(8):e200. Ybarra ML, Prescott TL, Philips GL, Bull SS, Parsons JT, Mustanski B. Iteratively developing an mHealth HIV prevention program for sexual minority adolescent men. AIDS and Behavior 2016;20(6):1157‐72. Ybarra ML, Prescott TL, Phillips GL Jr, Parsons JT, Bull SS, Mustanski B. Ethical considerations in recruiting online and implementing a text messaging–based HIV prevention program with gay, bisexual, and queer adolescent males. Journal of Adolescent Health 2016;59(1):44‐9. Adults (N = 27) Abdul Rashid 2013 Abdul Rashid RM, Mohamed M, Hamid ZA, Dahlui M. Is the phone call the most effective method for recall in cervical cancer screening? Results from a randomised control trial. Asian Pacific Journal of Cancer Prevention 2013;14(10):5901‐4. Rashid RM, Ramli S, John J, Dahlui M. Cost effective analysis of recall methods for cervical cancer screening in Selangor ‐ results from a prospective randomized controlled trial. Asian Pacific Journal of Cancer Prevention 2014;15(13):5143‐7.

Abdul Rashid RM, Mohamed M, Hamid ZA, Dahlui M. Is the phone call the most effective method for recall in cervical cancer screening? Results from a randomised control trial. Asian Pacific Journal of Cancer Prevention 2013;14(10):5901‐4. Rashid RM, Ramli S, John J, Dahlui M. Cost effective analysis of recall methods for cervical cancer screening in Selangor ‐ results from a prospective randomized controlled trial. Asian Pacific Journal of Cancer Prevention 2014;15(13):5143‐7. Rashid RMA, Dahlui M. Cost effective analysis of different types of recall on patients' response rate in a pap smear screening program. Journal of Health and Translational Medicine 2013;16:73. Barnabas 2016 Barnabas RV, van R, Tumwesigye E, Brantley J, Baeten JM, van H, et al. Uptake of antiretroviral therapy and male circumcision after community‐based HIV testing and strategies for linkage to care versus standard clinic referral: a multisite, open‐label, randomised controlled trial in South Africa and Uganda. Lancet HIV 2016;3(5):e212‐20. Gilbert HN, Wyatt MA, Asiimwe S, Turyamureeba B, Tumwesigye E, Van Rooyen H, et al. Messaging circumstances and economic pressures as influences on linkage to medical male circumcision following community‐based HIV testing for men in rural southwest Uganda: a qualitative study. AIDS Research and Treatment 2018;2018:Article ID 8387436.

t MA, Asiimwe S, Turyamureeba B, Tumwesigye E, Van Rooyen H, et al. Messaging circumstances and economic pressures as influences on linkage to medical male circumcision following community‐based HIV testing for men in rural southwest Uganda: a qualitative study. AIDS Research and Treatment 2018;2018:Article ID 8387436. Ware NC, Wyatt MA, Asiimwe S, Turyamureeba B, Tumwesigye E, Van Rooyen H, et al. How home HIV testing and counselling with follow‐up support achieves high testing coverage and linkage to treatment and prevention: a qualitative analysis from Uganda. Journal of the International AIDS Society 2016;19(1):20929. Constant 2014 Constant D, de Tolly K, Harries J, Myer L. Mobile phone messages to provide support to women during the home phase of medical abortion in South Africa: a randomised controlled trial. Contraception 2014;90(3):226‐33. de Tolly K, Constant D. Integrating mobile phones into medical abortion provision: intervention development, use, and lessons learned from a randomized controlled trial. JMIR mHealth and uHealth 2014;2(1):e5. Constant D, de Tolly K, Harries J, Myer L. Assessment of completion of early medical abortion using a text questionnaire on mobile phones compared to a self‐administered paper questionnaire among women attending four clinics, Cape Town, South Africa. Reproductive Health Matters 2014;22(Suppl 44):83‐93. Cook 2015 Cook PF, Carrington JM, Schmiege SJ, Starr W, Reeder B. A counselor in your pocket: feasibility of mobile health tailored messages to support HIV medication adherence. Patient Preference & Adherence 2015;9:1353‐66. da Costa 2012

Constant D, de Tolly K, Harries J, Myer L. Assessment of completion of early medical abortion using a text questionnaire on mobile phones compared to a self‐administered paper questionnaire among women attending four clinics, Cape Town, South Africa. Reproductive Health Matters 2014;22(Suppl 44):83‐93. Cook 2015 Cook PF, Carrington JM, Schmiege SJ, Starr W, Reeder B. A counselor in your pocket: feasibility of mobile health tailored messages to support HIV medication adherence. Patient Preference & Adherence 2015;9:1353‐66. da Costa 2012 da Costa TM, Barbosa BJ, Gomes e Costa DA, Sigulem D, de Fátima Marin H, Filho AC, et al. Results of a randomized controlled trial to assess the effects of a mobile SMS‐based intervention on treatment adherence in HIV/AIDS‐infected Brazilian women and impressions and satisfaction with respect to incoming messages. International Journal of Medical Informatics 2012;81(4):257‐69. de Tolly 2012 de Tolly K, Skinner D, Nembaware V, Benjamin P. Investigation into the use of short message services to expand uptake of human immunodeficiency virus testing, and whether content and dosage have impact. Telemedicine Journal and e‐Health 2012;18(1):18‐23. Downing 2013 Downing SG, Cashman C, McNamee H, Penney D, Russell DB, Hellard ME. Increasing chlamydia test of re‐infection rates using SMS reminders and incentives. Sexually Transmitted Infections 2013;89(1):16‐9. Gerdts 2015

de Tolly K, Skinner D, Nembaware V, Benjamin P. Investigation into the use of short message services to expand uptake of human immunodeficiency virus testing, and whether content and dosage have impact. Telemedicine Journal and e‐Health 2012;18(1):18‐23. Downing 2013 Downing SG, Cashman C, McNamee H, Penney D, Russell DB, Hellard ME. Increasing chlamydia test of re‐infection rates using SMS reminders and incentives. Sexually Transmitted Infections 2013;89(1):16‐9. Gerdts 2015 Gerdts C, Moseson H, Mora M, DePineres T. Alternative follow‐up options for medical abortion in Colombia: a pilot randomized controlled trial testing the feasibility of text‐messages. Contraception 2015;92(4):373. Hou 2010 Hou MY, Hurwitz S, Kavanagh E, Fortin J, Goldberg AB. Using daily text‐message reminders to improve adherence with oral contraceptives: a randomized controlled trial. Obstetrics and Gynecology 2010;116(3):633‐40. Huang 2013 Huang D, Sangthong R, McNeil E, Chongsuvivatwong V, Zheng W, Yang X. Effects of a phone call intervention to promote adherence to antiretroviral therapy and quality of life of HIV/AIDS patients in Baoshan, China: a randomized controlled trial. AIDS Research & Treatment 2013;2013:580974. Ingersoll 2015 Ingersoll KS, Dillingham RA, Hettema JE, Conaway M, Freeman J, Reynolds G, et al. Pilot RCT of bidirectional text messaging for ART adherence among nonurban substance users with HIV. Health Psychology 2015;34S:1305‐15. Joseph Davey 2016

Huang D, Sangthong R, McNeil E, Chongsuvivatwong V, Zheng W, Yang X. Effects of a phone call intervention to promote adherence to antiretroviral therapy and quality of life of HIV/AIDS patients in Baoshan, China: a randomized controlled trial. AIDS Research & Treatment 2013;2013:580974. Ingersoll 2015 Ingersoll KS, Dillingham RA, Hettema JE, Conaway M, Freeman J, Reynolds G, et al. Pilot RCT of bidirectional text messaging for ART adherence among nonurban substance users with HIV. Health Psychology 2015;34S:1305‐15. Joseph Davey 2016 Joseph Davey D, Nhavoto JA, Augusto O, Ponce W, Traca D, Nguimfack A, et al. SMSaude: Evaluating mobile phone text reminders to improve retention in HIV care for patients on antiretroviral therapy in Mozambique. Journal of Acquired Immune Deficiency Syndromes 2016;73(2):e23‐30. Lee 2016 Lee HY, Le C, Ghebre R, Yee D. Mobile phone multimedia messaging intervention for breast cancer screening. Cancer Research 2016;76(4):Abstract P3‐08‐03. Lee H, Ghebre R, Le C, Jang YJ, Sharratt M, Yee D. Mobile phone multilevel and multimedia messaging intervention for breast cancer screening: pilot randomized controlled trial. JMIR mHealth and uHealth 2017;5(11):e154. Lee H, Lee M, Gao Z, Sadak K. Development and evaluation of culturally and linguistically tailored mobile app to promote breast cancer screening. Journal of Clinical Medicine 2018;7(8):181. Leiby 2016

Lee H, Ghebre R, Le C, Jang YJ, Sharratt M, Yee D. Mobile phone multilevel and multimedia messaging intervention for breast cancer screening: pilot randomized controlled trial. JMIR mHealth and uHealth 2017;5(11):e154. Lee H, Lee M, Gao Z, Sadak K. Development and evaluation of culturally and linguistically tailored mobile app to promote breast cancer screening. Journal of Clinical Medicine 2018;7(8):181. Leiby 2016 Leiby K, Connor A, Tsague L, Sapele C, Kaonga A, Kakaire J, et al. The impact of SMS‐based interventions on VMMC uptake in Lusaka Province, Zambia: a randomized controlled trial. Journal of Acquired Immune Deficiency Syndromes 2016;72(Suppl 4):S264‐72. Lester 2010 Chi BH, Stringer JS. Mobile phones to improve HIV treatment adherence. Lancet 2010;376(9755):1807‐8. Lester RT, Ritvo P, Mills EJ, Kariri A, Karanja S, Chung MH, et al. Effects of a mobile phone short message service on antiretroviral treatment adherence in Kenya (WelTel Kenya1): a randomised trial. Lancet 2010;376(9755):1838‐45. Memetovic J, Kop ML, Karanja S, Kimani J, Ngugi EN, Ritvo P, et al. Perceived stigma and disclosure of HIV status: their association with clinical and communication outcomes in the WelTel Kenya1 SMS randomized controlled trial. Canadian Journal of Infectious Diseases and Medical Microbiology 2013;24:107a. Lester RT, Mills EJ, Kariri A, Ritvo P, Chung M, Jack W, et al. The HAART cell phone adherence trial (WelTel Kenya1): a randomized controlled trial protocol. Trials 2009;10(1):87.

Memetovic J, Kop ML, Karanja S, Kimani J, Ngugi EN, Ritvo P, et al. Perceived stigma and disclosure of HIV status: their association with clinical and communication outcomes in the WelTel Kenya1 SMS randomized controlled trial. Canadian Journal of Infectious Diseases and Medical Microbiology 2013;24:107a. Lester RT, Mills EJ, Kariri A, Ritvo P, Chung M, Jack W, et al. The HAART cell phone adherence trial (WelTel Kenya1): a randomized controlled trial protocol. Trials 2009;10(1):87. Smillie K, Borek NV, Kop MLvd, Lukhwaro A, Li N, Karanja S, et al. Mobile health for early retention in HIV care: a qualitative study in Kenya (WelTel Retain). African Journal of AIDS Research 2014;13(4):331‐8. Mbuagbaw 2012 Mbuagbaw L, Thabane L, Ongolo‐Zogo P, Lester RT, Mills EJ, Smieja M, et al. The Cameroon Mobile Phone SMS (CAMPS) trial: a randomized trial of text messaging versus usual care for adherence to antiretroviral therapy. PLOS ONE 2012;7(12):e46909. Mbuagbaw L, Bonono‐Momnougui RC, Thabane L. Considerations in using text messages to improve adherence to highly active antiretroviral therapy: a qualitative study among clients in Yaounde, Cameroon. HIV/AIDS (Auckland, NZ) 2012;4:45. Mbuagbaw L, Thabane L, Ongolo‐Zogo P. Opening communication channels with people living with HIV using mobile phone text messaging: insights from the CAMPS trial. BMC Research Notes 2013;6(1):131.

Mbuagbaw L, Bonono‐Momnougui RC, Thabane L. Considerations in using text messages to improve adherence to highly active antiretroviral therapy: a qualitative study among clients in Yaounde, Cameroon. HIV/AIDS (Auckland, NZ) 2012;4:45. Mbuagbaw L, Thabane L, Ongolo‐Zogo P. Opening communication channels with people living with HIV using mobile phone text messaging: insights from the CAMPS trial. BMC Research Notes 2013;6(1):131. Mbuagbaw L, Thabane L, Ongolo‐Zogo P, Lang T. The challenges and opportunities of conducting a clinical trial in a low resource setting: the case of the Cameroon mobile phone SMS (CAMPS) trial, an investigator initiated trial. Trials 2011;12(1):145. Mbuagbaw L, Thabane L, Ongolo‐Zogo P, Lester RT, Mills E, Volmink J, et al. The Cameroon mobile phone SMS (CAMPS) trial: a protocol for a randomized controlled trial of mobile phone text messaging versus usual care for improving adherence to highly active anti‐retroviral therapy. Trials 2011;12(1):5. Mugo 2016 Mugo PM, Wahome EW, Gichuru E, Mwashigadi G, Thiong'o AN, Prins HA, et al. Effect of SMS, phone‐call, and in‐person reminders on repeat HIV test uptake in Kenya. Topics in antiviral medicine 2016;24:420. Mugo PM, Wahome EW, Gichuru EN, Mwashigadi GM, Thiong'o AN, Prins HA, et al. Effect of text message, phone call, and in‐person appointment reminders on uptake of repeat HIV testing among outpatients screened for acute HIV infection in Kenya: a randomized controlled trial. PLOS ONE 2016;11(4):e0153612. Norton 2014

Mugo PM, Wahome EW, Gichuru E, Mwashigadi G, Thiong'o AN, Prins HA, et al. Effect of SMS, phone‐call, and in‐person reminders on repeat HIV test uptake in Kenya. Topics in antiviral medicine 2016;24:420. Mugo PM, Wahome EW, Gichuru EN, Mwashigadi GM, Thiong'o AN, Prins HA, et al. Effect of text message, phone call, and in‐person appointment reminders on uptake of repeat HIV testing among outpatients screened for acute HIV infection in Kenya: a randomized controlled trial. PLOS ONE 2016;11(4):e0153612. Norton 2014 Norton BL, Person AK, Castillo C, Pastrana C, Subramanian M, Stout JE. Barriers to using text message appointment reminders in an HIV clinic. Telemedicine Journal and e‐health 2014;20(1):86‐9. Nsagha 2016 Nsagha DS, Lange I, Fon PN, Nguedia Assob JC, Tanue EA. A randomized controlled trial on the usefulness of mobile text phone messages to improve the quality of care of HIV and AIDS patients in Cameroon. Open AIDS Journal 2016;10:93‐103. Odeny 2012 Odeny TA, Bailey RC, Bukusi EA, Simoni JM, Tapia KA, Yuhas K, et al. Effect of text messaging to deter early resumption of sexual activity after male circumcision for HIV prevention: a randomized controlled trial. Journal of Acquired Immune Deficiency Syndromes 2014;65(2):e50‐7. Odeny TA, Bailey RC, Bukusi EA, Simoni JM, Tapia KA, Yuhas K, et al. Text messaging to improve attendance at post‐operative clinic visits after adult male circumcision for HIV prevention: a randomized controlled trial. PLOS ONE 2012;7(9):e43832. Pop‐Eleches 2011

Odeny TA, Bailey RC, Bukusi EA, Simoni JM, Tapia KA, Yuhas K, et al. Effect of text messaging to deter early resumption of sexual activity after male circumcision for HIV prevention: a randomized controlled trial. Journal of Acquired Immune Deficiency Syndromes 2014;65(2):e50‐7. Odeny TA, Bailey RC, Bukusi EA, Simoni JM, Tapia KA, Yuhas K, et al. Text messaging to improve attendance at post‐operative clinic visits after adult male circumcision for HIV prevention: a randomized controlled trial. PLOS ONE 2012;7(9):e43832. Pop‐Eleches 2011 Pop‐Eleches C, Thirumurthy H, Habyarimana JP, Zivin JG, Goldstein MP, de Walque D, et al. Mobile phone technologies improve adherence to antiretroviral treatment in a resource‐limited setting: a randomized controlled trial of text message reminders. AIDS 2011;25(6):825‐34. Rutland 2012 Rutland E, Roe H, Weaver A. Health promotional messages in short message service (SMS) follow‐up of GU medicine clinic defaulters; a tool to improve subsequent attendance rates? Sexually Transmitted Infections 2012;88(Suppl 1):A4. Shet 2014 De Costa A, Shet A, Kumarasamy N, Ashorn P, Eriksson B, Bogg L, et al. Design of a randomized trial to evaluate the influence of mobile phone reminders on adherence to first line antiretroviral treatment in South India ‐ the HIVIND study protocol. BMC Medical Research Methodolgy 2010;10:25. Erratum: Effect of mobile telephone reminders on treatment outcome in HIV: evidence from a randomised controlled trial in India. BMJ (Online) 2014;349(g5978):no pagination.

De Costa A, Shet A, Kumarasamy N, Ashorn P, Eriksson B, Bogg L, et al. Design of a randomized trial to evaluate the influence of mobile phone reminders on adherence to first line antiretroviral treatment in South India ‐ the HIVIND study protocol. BMC Medical Research Methodolgy 2010;10:25. Erratum: Effect of mobile telephone reminders on treatment outcome in HIV: evidence from a randomised controlled trial in India. BMJ (Online) 2014;349(g5978):no pagination. Rodrigues R, Poongulali S, Balaji K, Atkins S, Ashorn P, De Costa A. 'The phone reminder is important, but will others get to know about my illness?' Patient perceptions of an mHealth antiretroviral treatment support intervention in the HIVIND trial in South India. BMJ Open 2015;5(11):e007574. Shet A, De Costa A, Kumarasamy N, Rodrigues R, Rewari BB, Ashorn P, et al. Effect of mobile telephone reminders on treatment outcome in HIV: evidence from a randomised controlled trial in India. BMJ 2014;349:g5978. Rodrigues R, Shet A, Antony J, Sidney K, Arumugam K, Krishnamurthy S, et al. Supporting adherence to antiretroviral therapy with mobile phone reminders: results from a cohort in South India. PLOS ONE 2012;7(8):e40723. Shet A, Arumugam K, Rodrigues R, Rajagopalan N, Shubha K, Raj T, et al. Designing a mobile phone‐based intervention to promote adherence to antiretroviral therapy in South India. AIDS and Behavior 2010;14(3):716‐20.

Rodrigues R, Shet A, Antony J, Sidney K, Arumugam K, Krishnamurthy S, et al. Supporting adherence to antiretroviral therapy with mobile phone reminders: results from a cohort in South India. PLOS ONE 2012;7(8):e40723. Shet A, Arumugam K, Rodrigues R, Rajagopalan N, Shubha K, Raj T, et al. Designing a mobile phone‐based intervention to promote adherence to antiretroviral therapy in South India. AIDS and Behavior 2010;14(3):716‐20. Shet A, De Costa A, Kumarasamy N, Rodrigues R, Rewari BB, Ashorn P, et al. Effect of mobile telephone reminders on treatment outcome in HIV: evidence from a randomised controlled trial in India. BMJ 2014;349:g5978. Rodrigues R, Poongulali S, Balaji K, Atkins S, Ashorn P, De C. 'The phone reminder is important, but will others get to know about my illness?' Patient perceptions of an mHealth antiretroviral treatment support intervention in the HIVIND trial in South India. BMJ Open 2015;5(11):e007574. Smith 2015 Smith C, Ngo TD, Gold J, Edwards P, Vannak U, Sokhey L, et al. Effect of a mobile phone‐based intervention on post‐abortion contraception: a randomized controlled trial in Cambodia. Bulletin of the World Health Organization 2015;93(12):842‐50A. Smith C, Edwards P, Free C. Assessing the validity and reliability of self‐report data on contraception use in the MObile Technology for Improved Family Planning (MOTIF) randomised controlled trial. Reproductive Health 2018;15(1):50. Smith C, Jarvis C, Free C. Assessing loss to follow‐up in the MObile Technology for Improved Family Planning (MOTIF) randomised controlled trial. Trials 2017;18(1):577.

Smith C, Edwards P, Free C. Assessing the validity and reliability of self‐report data on contraception use in the MObile Technology for Improved Family Planning (MOTIF) randomised controlled trial. Reproductive Health 2018;15(1):50. Smith C, Jarvis C, Free C. Assessing loss to follow‐up in the MObile Technology for Improved Family Planning (MOTIF) randomised controlled trial. Trials 2017;18(1):577. Smith C, Ly S, Uk V, Warnock R, Edwards P, Free C. Process evaluation of a mobile phone‐based intervention to support post‐abortion contraception in Cambodia. Contraception and Reproductive Medicine 2017;2(1):16. Smith C, Ly S, Uk V, Warnock R, Free C. Women’s views and experiences of a mobile phone‐based intervention to support post‐abortion contraception in Cambodia. Reproductive Health 2017;14(1):72. Smith C, Ngo TD, Edwards P, Free C. MObile Technology for Improved Family Planning: update to randomised controlled trial protocol. Trials 2014;15(1):440. Smith C, Vannak U, Sokhey L, Ngo TD, Gold J, Free C. Mobile Technology for Improved Family Planning (MOTIF): the development of a mobile phone‐based (mHealth) intervention to support post‐abortion family planning (PAFP) in Cambodia. Reproductive Health 2015;13(1):1. Smith C, Vannak U, Sokhey L, Ngo TD, Gold J, Khut K, et al. MObile Technology for Improved Family Planning services (MOTIF): study protocol for a randomised controlled trial. Trials 2013;14(1):427. Young 2015

Smith C, Vannak U, Sokhey L, Ngo TD, Gold J, Free C. Mobile Technology for Improved Family Planning (MOTIF): the development of a mobile phone‐based (mHealth) intervention to support post‐abortion family planning (PAFP) in Cambodia. Reproductive Health 2015;13(1):1. Smith C, Vannak U, Sokhey L, Ngo TD, Gold J, Khut K, et al. MObile Technology for Improved Family Planning services (MOTIF): study protocol for a randomised controlled trial. Trials 2013;14(1):427. Young 2015 Young SD, Cumberland WG, Nianogo R, Menacho LA, Galea JT, Coates T. The HOPE social media intervention for global HIV prevention in Peru: a cluster randomised controlled trial. Lancet HIV 2015;2(1):e27‐32. Chiu CJ, Menacho L, Fisher C, Young SD. Ethics issues in social media–based HIV prevention in low‐ and middle‐income countries. Cambridge Quarterly of Healthcare Ethics 2015;24(3):303‐10. Garett R, Menacho L, Young SD. Ethical issues in using social media to deliver an HIV prevention intervention: results from the HOPE Peru Study. Prevention Science 2017;18(2):225‐32. Jaganath D, Gill HK, Cohen AC, Young SD. Harnessing Online Peer Education (HOPE): integrating C‐POL and social media to train peer leaders in HIV prevention. AIDS Care 2012;24(5):593‐600. Krueger EA, Chiu CJ, Menacho LA, Young SD. HIV testing among social media‐using Peruvian men who have sex with men: correlates and social context. AIDS Care 2016;28(10):1301‐5.

Jaganath D, Gill HK, Cohen AC, Young SD. Harnessing Online Peer Education (HOPE): integrating C‐POL and social media to train peer leaders in HIV prevention. AIDS Care 2012;24(5):593‐600. Krueger EA, Chiu CJ, Menacho LA, Young SD. HIV testing among social media‐using Peruvian men who have sex with men: correlates and social context. AIDS Care 2016;28(10):1301‐5. Young SD, Holloway I, Jaganath D, Rice E, Westmoreland D, Coates T. Project HOPE: online social network changes in an HIV prevention randomized controlled trial for African American and Latino men who have sex with men. American Journal of Public Health 2014;104(9):1707‐12. Young SD, Jaganath D. Online social networking for HIV education and prevention: a mixed methods analysis. Sexually Transmitted Diseases 2013;40(2). Menacho LA, Blas MM, Alva IE, Orellana ER. Short text messages to motivate HIV testing among men who have sex with men: a qualitative study in Lima, Peru. Open AIDS Journal 2013;7:1. Pregnant and postpartum women (N = 11) Cooper 2015 Cooper S, Foster K, Naughton F, Leonardi‐Bee J, Sutton S, Ussher M, et al. Pilot study to evaluate a tailored text message intervention for pregnant smokers (MiQuit): study protocol for a randomised controlled trial. Trials 2015;16:29. Naughton F, Cooper S, Foster K, Emery J, Leonardi‐Bee J, Sutton S, et al. Large multi‐centre pilot randomized controlled trial testing a low‐cost, tailored, self‐help smoking cessation text message intervention for pregnant smokers (MiQuit). Addiction 2017;112(7):1238‐49.

Cooper S, Foster K, Naughton F, Leonardi‐Bee J, Sutton S, Ussher M, et al. Pilot study to evaluate a tailored text message intervention for pregnant smokers (MiQuit): study protocol for a randomised controlled trial. Trials 2015;16:29. Naughton F, Cooper S, Foster K, Emery J, Leonardi‐Bee J, Sutton S, et al. Large multi‐centre pilot randomized controlled trial testing a low‐cost, tailored, self‐help smoking cessation text message intervention for pregnant smokers (MiQuit). Addiction 2017;112(7):1238‐49. Emery JL, Coleman T, Sutton S, Cooper S, Leonardi‐Bee J, Jones M, et al. Uptake of tailored text message smoking cessation support in pregnancy when advertised on the internet (MiQuit): observational study. Journal of Medical Internet Research 2018;20(4):e146. Naughton F, Cooper S, Bowker K, Campbell K, Sutton S, Leonardi‐Bee J, et al. Adaptation and uptake evaluation of an SMS text message smoking cessation programme (MiQuit) for use in antenatal care. BMJ Open 2015;5(10):e008871. Naughton F, Prevost AT, Gilbert H, Sutton S. Randomized controlled trial evaluation of a tailored leaflet and SMS text message self‐help intervention for pregnant smokers (MiQuit). Nicotine & Tobacco Research 2012;14(5):569‐77. Sloan M, Hopewell S, Coleman T, Cooper S, Naughton F. Smoking cessation support by text message during pregnancy: a qualitative study of views and experiences of the MiQuit intervention. Nicotine & Tobacco Research 2017;19(5):572‐7. Evans 2014

Naughton F, Prevost AT, Gilbert H, Sutton S. Randomized controlled trial evaluation of a tailored leaflet and SMS text message self‐help intervention for pregnant smokers (MiQuit). Nicotine & Tobacco Research 2012;14(5):569‐77. Sloan M, Hopewell S, Coleman T, Cooper S, Naughton F. Smoking cessation support by text message during pregnancy: a qualitative study of views and experiences of the MiQuit intervention. Nicotine & Tobacco Research 2017;19(5):572‐7. Evans 2014 Evans WD, Wallace B, Szekely D, Nielsen P, Murray E, Abroms L, et al. Initial outcomes from a 4‐week follow‐up study of the Text4baby program in the military women's population: randomized controlled trial. Journal of Medical Internet Research 2014;16(5):e131. Evans W, Nielsen PE, Szekely DR, Bihm JW, Murray EA, Snider J, et al. Dose‐response effects of the text4baby mobile health program: randomized controlled trial. JMIR mHealth and uHealth 2015;3(1):e12. Evans WD, Abroms LC, Poropatich R, Nielsen PE, Wallace JL. Mobile health evaluation methods: the Text4baby case study. Journal of Health Communication 2012;17(Suppl 1):22‐9. Gazmararian JA, Elon L, Yang B, Graham M, Parker R. Text4baby program: an opportunity to reach underserved pregnant and postpartum women? Maternal and Child Health Journal 2014;18(1):223‐32. Jareethum 2008 Jareethum R, Titapant V, Chantra T, Sommai V, Chuenwattana P, Jirawan C. Satisfaction of healthy pregnant women receiving short message service via mobile phone for prenatal support: a randomized controlled trial. Chotmaihet Thangphaet [Journal of the Medical Association of Thailand] 2008;91(4):458‐63.

Gazmararian JA, Elon L, Yang B, Graham M, Parker R. Text4baby program: an opportunity to reach underserved pregnant and postpartum women? Maternal and Child Health Journal 2014;18(1):223‐32. Jareethum 2008 Jareethum R, Titapant V, Chantra T, Sommai V, Chuenwattana P, Jirawan C. Satisfaction of healthy pregnant women receiving short message service via mobile phone for prenatal support: a randomized controlled trial. Chotmaihet Thangphaet [Journal of the Medical Association of Thailand] 2008;91(4):458‐63. Joshi 2015 Joshi S, Patil N, Hegde A. Impact of mHealth initiative on utilization of antenatal care services in rural Maharashtra, India. Indian Journal of Maternal and Child Health 2015;17(2):1‐7. Kamau‐Mbuthia 2013 Kamau‐Mbuthia E, Mbugua S, Webb G, Kalungu S, Sarange C, Lou W, et al. Cell phone based peer counseling to support exclusive breastfeeding is associated with more frequent help and decreased breastfeeding problems. Annals of Nutrition & Metabolism 2013;63:196‐7. Webb G, Kamau‐Mbuthia E, Mbugua S, Kalungu S, Sarange C, Lou W, et al. Infant medication, illness and growth in a randomized controlled trial of exclusive breastfeeding support in Kenya. Annals of Nutrition & Metabolism 2013;63:752. Lund 2012 Lund S, Hemed M, Nielsen BB, Said A, Said K, Makungu MH, et al. Mobile phones as a health communication tool to improve skilled attendance at delivery in Zanzibar: a cluster‐randomised controlled trial. BJOG: An International Journal of Obstetrics & Gynaecology 2012;119(10):1256‐64.

Webb G, Kamau‐Mbuthia E, Mbugua S, Kalungu S, Sarange C, Lou W, et al. Infant medication, illness and growth in a randomized controlled trial of exclusive breastfeeding support in Kenya. Annals of Nutrition & Metabolism 2013;63:752. Lund 2012 Lund S, Hemed M, Nielsen BB, Said A, Said K, Makungu MH, et al. Mobile phones as a health communication tool to improve skilled attendance at delivery in Zanzibar: a cluster‐randomised controlled trial. BJOG: An International Journal of Obstetrics & Gynaecology 2012;119(10):1256‐64. Lund S, Nielsen BB, Hemed M, Boas IM, Said A, Said K, et al. Mobile phones improve antenatal care attendance in Zanzibar: a cluster randomized controlled trial. BMC Pregnancy & Childbirth 2014;14:29. Lund S, Nielsen BB, Hemed M, Said A, Said K, Makungu MH, et al. Mobile phones as a health communication tool to improve maternal and perinatal health in Zanzibar: a cluster randomised controlled trial. Tropical Medicine & International Health 2013;18:22. Lund S, Rasch V, Hemed M, Boas IM, Said A, Said K, et al. Mobile phone intervention reduces perinatal mortality in Zanzibar: secondary outcomes of a cluster randomized controlled trial. JMIR mHealth and uHealth 2014;2(1):e15. Maslowsky 2016 Maslowsky J, Frost S, Hendrick CE, Trujillo Cruz FO, Merajver SD. Effects of postpartum mobile phone‐based education on maternal and infant health in Ecuador. International Journal of Gynaecology & Obstetrics 2016;134(1):93‐8. McConnell 2016

Lund S, Rasch V, Hemed M, Boas IM, Said A, Said K, et al. Mobile phone intervention reduces perinatal mortality in Zanzibar: secondary outcomes of a cluster randomized controlled trial. JMIR mHealth and uHealth 2014;2(1):e15. Maslowsky 2016 Maslowsky J, Frost S, Hendrick CE, Trujillo Cruz FO, Merajver SD. Effects of postpartum mobile phone‐based education on maternal and infant health in Ecuador. International Journal of Gynaecology & Obstetrics 2016;134(1):93‐8. McConnell 2016 McConnell M, Ettenger A, Rothschild CW, Muigai F, Cohen J. Can a community health worker administered postnatal checklist increase health‐seeking behaviors and knowledge?: Evidence from a randomized trial with a private maternity facility in Kiambu County, Kenya. BMC Pregnancy Childbirth 2016;16(1):136. Moniz 2013 Moniz MH, Hasley S, Meyn LA, Beigi RH. Improving influenza vaccination rates in pregnancy through text messaging: a randomized controlled trial. Obstetrics & Gynecology 2013;121(4):734‐40. Omole 2018 Omole O, Ijadunola MY, Olotu E, Omotoso O, Bello B, Awoniran O, et al. The effect of mobile phone short message service on maternal health in south‐west Nigeria. International Journal of Health Planning and Management 2018;33(1):155‐70. Yudin 2017 Yudin MH, Mistry N, De Souza LR, Besel K, Patel V, Blanco Mejia S, et al. Text messages for influenza vaccination among pregnant women: a randomized controlled trial. Vaccine 2017;35(5):842‐8. Parents (N = 14) Ahlers‐Schmidt 2012a

Omole O, Ijadunola MY, Olotu E, Omotoso O, Bello B, Awoniran O, et al. The effect of mobile phone short message service on maternal health in south‐west Nigeria. International Journal of Health Planning and Management 2018;33(1):155‐70. Yudin 2017 Yudin MH, Mistry N, De Souza LR, Besel K, Patel V, Blanco Mejia S, et al. Text messages for influenza vaccination among pregnant women: a randomized controlled trial. Vaccine 2017;35(5):842‐8. Parents (N = 14) Ahlers‐Schmidt 2012a Ahlers‐Schmidt CR, Chesser AK, Nguyen T, Brannon J, Hart TA, Williams KS, et al. Feasibility of a randomized controlled trial to evaluate Text Reminders for Immunization Compliance in Kids (TRICKs). Vaccine 2012;30(36):5305‐9. Bangure 2015 Bangure D, Chirundu D, Gombe N, Marufu T, Mandozana G, Tshimanga M, et al. Effectiveness of short message services reminder on childhood immunization programme in Kadoma, Zimbabwe ‐ a randomized controlled trial, 2013. BMC Public Health 2015;15:137. Bigna 2015 Bigna JJ, Noubiap JJ, Kouanfack C, Plottel CS, Koulla‐Shiro S. Effect of mobile phone reminders on follow‐up medical care of children exposed to or infected with HIV in Cameroon (MORE CARE): a multicentre, single‐blind, factorial, randomised controlled trial. Lancet Infectious Diseases 2014;14(7):600‐8. Bigna JJ, Noubiap JJ, Plottel CS, Kouanfack C, Koulla‐Shiro S. Barriers to the implementation of mobile phone reminders in pediatric HIV care: a pre‐trial analysis of the Cameroonian MORE CARE study. BMC Health Services Research 2014;14:523.

Bigna JJ, Noubiap JJ, Kouanfack C, Plottel CS, Koulla‐Shiro S. Effect of mobile phone reminders on follow‐up medical care of children exposed to or infected with HIV in Cameroon (MORE CARE): a multicentre, single‐blind, factorial, randomised controlled trial. Lancet Infectious Diseases 2014;14(7):600‐8. Bigna JJ, Noubiap JJ, Plottel CS, Kouanfack C, Koulla‐Shiro S. Barriers to the implementation of mobile phone reminders in pediatric HIV care: a pre‐trial analysis of the Cameroonian MORE CARE study. BMC Health Services Research 2014;14:523. Bigna JJ, Noubiap JJ, Plottel CS, Kouanfack C, Koulla‐Shiro S. Factors associated with non‐adherence to scheduled medical follow‐up appointments among Cameroonian children requiring HIV care: a case‐control analysis of the usual‐care group in the MORE CARE trial. Infectious Diseases of Poverty 2014;3(1):44. Bigna JJ. Automated text message reminders to promote good health. Lancet Infectious Diseases 2015;15(1):19‐20. Mbuagbaw L. Mobile phone reminders for paediatric HIV follow‐up care. Lancet Infectious Diseases 2014;14(7):540‐1. Brown 2016 Brown VB, Oluwatosin OA, Akinyemi JO, Adeyemo AA. Effects of community health nurse‐led intervention on childhood routine immunization completion in primary health care centers in Ibadan, Nigeria. Journal of Community Health 2016;41(2):265‐73. Brown VB, Oluwatosin A, Ogundeji MO. Experiences, perceptions and preferences of mothers towards childhood immunization reminder/recall in Ibadan, Nigeria: a cross‐sectional study. Pan African Medical Journal 2015;20:243.

Brown VB, Oluwatosin OA, Akinyemi JO, Adeyemo AA. Effects of community health nurse‐led intervention on childhood routine immunization completion in primary health care centers in Ibadan, Nigeria. Journal of Community Health 2016;41(2):265‐73. Brown VB, Oluwatosin A, Ogundeji MO. Experiences, perceptions and preferences of mothers towards childhood immunization reminder/recall in Ibadan, Nigeria: a cross‐sectional study. Pan African Medical Journal 2015;20:243. Brown VB, Oluwatosin OA. Feasibility of implementing a cellphone‐based reminder/recall strategy to improve childhood routine immunization in a low‐resource setting: a descriptive report. BMC Health Services Research 2017;17(2):703. Domek 2016 Domek GJ, Contreras‐Roldan IL, O'Leary ST, Bull S, Furniss A, Kempe A, et al. SMS text message reminders to improve infant vaccination coverage in Guatemala: a pilot randomized controlled trial. Vaccine 2016;34(21):2437‐43. Domek GJ, Contreras‐Roldan IL, Asturias EJ, Bronsert M, Ventura GAB, O’Leary ST, et al. Characteristics of mobile phone access and usage in rural and urban Guatemala: assessing feasibility of text message reminders to increase childhood immunizations. mHealth 2018;4:9. Eze 2015 Eze GU, Adeleye OO. Enhancing routine immunization performance using innovative technology in an urban area of Nigeria. West African Journal of Medicine 2015;34(1):3‐10. Gibson 2017

Domek GJ, Contreras‐Roldan IL, Asturias EJ, Bronsert M, Ventura GAB, O’Leary ST, et al. Characteristics of mobile phone access and usage in rural and urban Guatemala: assessing feasibility of text message reminders to increase childhood immunizations. mHealth 2018;4:9. Eze 2015 Eze GU, Adeleye OO. Enhancing routine immunization performance using innovative technology in an urban area of Nigeria. West African Journal of Medicine 2015;34(1):3‐10. Gibson 2017 Gibson DG, Ochieng B, Kagucia EW, Were J, Hayford K, Moulton LH, et al. Mobile phone‐delivered reminders and incentives to improve childhood immunisation coverage and timeliness in Kenya (M‐SIMU): a cluster randomised controlled trial. Lancet Global Health 2017;5(4):e428‐38. Haji 2016 Haji A, Lowther S, Ngan'ga Z, Gura Z, Tabu C, Sandhu H, et al. Reducing routine vaccination dropout rates: evaluating two interventions in three Kenyan districts, 2014. BMC Public Health 2016;16:152. Hannan 2016 Hannan J, Brooten D, Page T, Galindo A, Torres M. Low‐income first‐time mothers: effects of APN follow‐up using mobile technology on maternal and infant outcomes. Global Pediatric Health 2016;3:DOI: 10.1177/2333794X16660234. Hofstetter 2015 Hofstetter AM, DuRivage N, Vargas CY, Camargo S, Vawdrey DK, Fisher A, et al. Text message reminders for timely routine MMR vaccination: a randomized controlled trial. Vaccine 2015;33(43):5741‐6. Jimenez 2017 Jimenez ME, DuRivage NE, Bezpalko O, Suh A, Wade R, Blum NJ, et al. A pilot randomized trial of a video patient decision aid to facilitate early intervention referrals from primary care. Clinical Pediatrics 2017;56(3):268‐77.

Hofstetter AM, DuRivage N, Vargas CY, Camargo S, Vawdrey DK, Fisher A, et al. Text message reminders for timely routine MMR vaccination: a randomized controlled trial. Vaccine 2015;33(43):5741‐6. Jimenez 2017 Jimenez ME, DuRivage NE, Bezpalko O, Suh A, Wade R, Blum NJ, et al. A pilot randomized trial of a video patient decision aid to facilitate early intervention referrals from primary care. Clinical Pediatrics 2017;56(3):268‐77. Niederhauser 2015 Niederhauser V, Johnson M, Tavakoli AS. Vaccines4Kids: assessing the impact of text message reminders on immunization rates in infants. Vaccine 2015;33(26):2984‐9. Sharma 2011 Sharma R, Hebbal M, Ankola AV, Murugabupathy V. Mobile‐phone text messaging (SMS) for providing oral health education to mothers of preschool children in Belgaum City. Journal of Telemedicine & Telecare 2011;17(8):432‐6. Stockwell 2015 Stockwell MS, Hofstetter AM, DuRivage N, Barrett A, Fernandez N, Vargas CY, et al. Text message reminders for second dose of influenza vaccine: a randomized controlled trial. Pediatrics 2015;135(1):e83‐91. Mothers living with HIV/AIDS (N = 3) Kassaye 2016 Kassaye SG, Ong'ech J, Sirengo M, Kose J, Matu L, McOdida P, et al. Cluster‐randomized controlled study of SMS text messages for prevention of mother‐to‐child transmission of HIV in rural Kenya. AIDS Research & Treatment 2016;2016:1289328. Jennings L, Ong’ech J, Simiyu R, Sirengo M, Kassaye S. Exploring the use of mobile phone technology for the enhancement of the prevention of mother‐to‐child transmission of HIV program in Nyanza, Kenya: a qualitative study. BMC Public Health 2013;13(1):1131. Kebaya 2014

Kassaye SG, Ong'ech J, Sirengo M, Kose J, Matu L, McOdida P, et al. Cluster‐randomized controlled study of SMS text messages for prevention of mother‐to‐child transmission of HIV in rural Kenya. AIDS Research & Treatment 2016;2016:1289328. Jennings L, Ong’ech J, Simiyu R, Sirengo M, Kassaye S. Exploring the use of mobile phone technology for the enhancement of the prevention of mother‐to‐child transmission of HIV program in Nyanza, Kenya: a qualitative study. BMC Public Health 2013;13(1):1131. Kebaya 2014 Kebaya L, Nduati R, Wamalwa D, Kariuki N, Bashir A. Efficacy of mobile phone use on adherence to nevirapine prophylaxis and retention in care among the HIV‐exposed infants in PMTCT: a randomised controlled trial. Archives of Disease in Childhood 2014;99:A329. Kebaya LM, Wamalwa D, Kariuki N, Admani B, Nduati RW. Efficacy of mobile phone use on adherence to nevirapine prophylaxis and retention in care among HIV‐exposed infants. Topics in Antiviral Medicine 2015;23:407. Odeny 2014 Odeny TA, Bukusi EA, Cohen CR, Yuhas K, Camlin CS, McClelland RS. Texting improves testing: a randomized trial of two‐way SMS to increase postpartum prevention of mother‐to‐child transmission retention and infant HIV testing. AIDS 2014;28(15):2307‐12. Odeny TA, Newman M, Bukusi EA, McClelland RS, Cohen CR, Camlin CS. Developing content for a mHealth intervention to promote postpartum retention in prevention of mother‐to‐child HIV transmission programs and early infant diagnosis of HIV: a qualitative study. PLOS ONE 2014;9(9):e106383.

Odeny TA, Bukusi EA, Cohen CR, Yuhas K, Camlin CS, McClelland RS. Texting improves testing: a randomized trial of two‐way SMS to increase postpartum prevention of mother‐to‐child transmission retention and infant HIV testing. AIDS 2014;28(15):2307‐12. Odeny TA, Newman M, Bukusi EA, McClelland RS, Cohen CR, Camlin CS. Developing content for a mHealth intervention to promote postpartum retention in prevention of mother‐to‐child HIV transmission programs and early infant diagnosis of HIV: a qualitative study. PLOS ONE 2014;9(9):e106383. What's new Date Event Description 16 October 2019 Amended Typographical error corrected in abstract. Differences between protocol and review None Characteristics of studies Characteristics of included studies [ordered by study ID] Akinfaderin‐Agarau 2012 Country Nigeria Participant group For adolescent and youth populations as potential users of sexual and reproductive health services For adult populations as potential users of sexual and reproductive health services Intervention channel Mobile phones Notes Brown 2014 Country USA Participant group For pregnant and postpartum women (up to 6 weeks) Parents and other caregivers of children under 5 years of age Intervention channel SMS text blasts Notes Calderón 2017 Country Peru Participant group Parents and other caregivers of children under 5 years of age Intervention channel SMS Notes Cates 2015 Country USA Participant group For adolescent and youth populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Cornelius 2009

Country USA Participant group For pregnant and postpartum women (up to 6 weeks) Parents and other caregivers of children under 5 years of age Intervention channel SMS text blasts Notes Calderón 2017 Country Peru Participant group Parents and other caregivers of children under 5 years of age Intervention channel SMS Notes Cates 2015 Country USA Participant group For adolescent and youth populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Cornelius 2009 Country USA Participant group For adolescent and youth populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Curioso 2009 Country Peru Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Entsieh 2015 Country Ghana Participant group For pregnant and postpartum women (up to 6 weeks) Parents and other caregivers of children under 5 years of age Intervention channel App “Mobile midwife” Notes Evans 2016 Country UK Participant group For adolescent and youth populations as potential users of sexual and reproductive health services For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Flax 2017 Country Nigeria Participant group Unclear. Pregnant and breastfeeding women as well as women without children Intervention channel SMS Notes French 2016 Country UK Participant group For adolescent and youth populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Gold 2010

Country Nigeria Participant group Unclear. Pregnant and breastfeeding women as well as women without children Intervention channel SMS Notes French 2016 Country UK Participant group For adolescent and youth populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Gold 2010 Country Australia Participant group For adolescent and youth populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Goldenberg 2015 Country USA Participant group For adolescent and youth populations as potential users of sexual and reproductive health services For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Greaney 2014 Country USA Participant group For adolescent and youth populations as potential users of sexual and reproductive health services For adult populations as potential users of sexual and reproductive health services Intervention channel Interactive voice response messages Notes Hirsch‐Moverman 2017 Country Lesotho Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Jalloh‐Vos 2014 Country Sierra Leone Participant group Pregnant and postpartum women and adults for family planning Intervention channel SMS and voice call Notes Jennings 2013 Country Kenya Participant group For pregnant and postpartum women (up to 6 weeks) Intervention channel SMS Notes Lau 2014

Country Lesotho Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Jalloh‐Vos 2014 Country Sierra Leone Participant group Pregnant and postpartum women and adults for family planning Intervention channel SMS and voice call Notes Jennings 2013 Country Kenya Participant group For pregnant and postpartum women (up to 6 weeks) Intervention channel SMS Notes Lau 2014 Country South Africa Participant group For pregnant and postpartum women (up to 6 weeks) Intervention channel SMS Notes Mbuagbaw 2012 Country Cameroon Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Mbuagbaw 2014 Country Cameroon Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Menacho 2013 Country Peru Participant group For adolescent and youth populations as potential users of sexual and reproductive health services For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Missal 2016 Country India Participant group For pregnant and postpartum women (up to 6 weeks) Intervention channel SMS Notes Mitchell 2016 Country USA Participant group For adolescent and youth populations as potential users of sexual and reproductive health services For adult populations as potential users of sexual and reproductive health services Intervention channel An app Notes Munro 2017

Country India Participant group For pregnant and postpartum women (up to 6 weeks) Intervention channel SMS Notes Mitchell 2016 Country USA Participant group For adolescent and youth populations as potential users of sexual and reproductive health services For adult populations as potential users of sexual and reproductive health services Intervention channel An app Notes Munro 2017 Country Canada Participant group For pregnant and postpartum women (up to 6 weeks) Parents and other caregivers of children under 5 years of age Intervention channel SMS “Text4baby program” Notes Nachega 2016 Country South Africa Participant group For pregnant and postpartum women (up to 6 weeks) who were HIV‐positive Intervention channel SMS Notes Naughton 2013 Country UK Participant group For pregnant and postpartum women (up to 6 weeks) Intervention channel SMS Notes Odeny 2014 Country Kenya Participant group For pregnant and postpartum women (up to 6 weeks) Intervention channel SMS Notes Perry 2012 Country USA Participant group For adolescent and youth populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Rana 2015 Country Uganda Participant group For adolescent and youth populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Rodrigues 2015 Country India Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel Interactive voice recordings and SMS Notes Sloan 2017

Country Uganda Participant group For adolescent and youth populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Rodrigues 2015 Country India Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel Interactive voice recordings and SMS Notes Sloan 2017 Country UK Participant group For pregnant and postpartum women (up to 6 weeks) Intervention channel SMS Notes Smillie 2014 Country Canada Participant group For adolescent and youth populations as potential users of sexual and reproductive health services For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Smith 2017 Country Cambodia Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel Mobile phone voice messaging and counsellor support Notes Ware 2016 Country Uganda Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS combined with real‐time adherence monitoring Notes Willoughby 2017 Country USA Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Wright 2011 Country USA Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes

Country Uganda Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS combined with real‐time adherence monitoring Notes Willoughby 2017 Country USA Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Wright 2011 Country USA Participant group For adult populations as potential users of sexual and reproductive health services Intervention channel SMS Notes Characteristics of excluded studies [ordered by study ID] Study Reason for exclusion Abbass‐Dick 2017 Wrong study design Abensur 2011 Wrong study design Abramson 2015 Wrong delivery mechanism Acevedo 1998 Wrong delivery mechanism Adanikin 2014 Wrong study design Agarwal 2014 Wrong study design Ahlers‐Schmidt 2012 Wrong study design Alexander 2014 Wrong study population Aradhya 2013 Wrong study design Asiodu 2015 Wrong delivery mechanism Associates 2015 Wrong study design Associates 2015a Wrong study design Azih 2012 Wrong study design Bacchus 2016 Wrong delivery mechanism Barnett 2016 Wrong topic of interest Belzer 2015 Wrong study design Beratarrechea 2015 Language not spoken by review team members Biediger‐Friedman 2016 Wrong topic of interest Birukila 2017 Wrong study design Blas 2013 Wrong topic of interest Brayboy 2017 Wrong study design Brinkel 2017 Wrong intervention (telemedicine) Broom 2015 Wrong study design Brüll 2016 Wrong delivery mechanism Bull 2010 Wrong delivery mechanism Campbell 2015 Wrong study design Chang 2013 Wrong topic of interest Chib 2013 Wrong study design Cordova 2015 Wrong topic of interest Cormick 2012 Wrong study design Cormick 2015 Wrong study design Curioso 2007 Wrong topic of interest Dean 2012 Wrong intervention Devine 2014 Wrong topic of interest Fletcher 2016 Wrong topic of interest Fornos 2014 Wrong topic of interest Gatwood 2014 Wrong study population Goldenberg 2014 Wrong delivery mechanism Hearn 2014 Wrong delivery mechanism Herbec 2014 Wrong delivery mechanism Hmone 2016 Wrong topic of interest Holloway 2017 Wrong delivery mechanism Horvath 2016 Wrong delivery mechanism Huberty 2015 Wrong topic of interest Huq 2014 Wrong delivery mechanism Irons 2015 Wrong study design Jamison 2013 Wrong intervention Kharbanda 2009 Wrong study population Kharbanda 2011 Wrong study population Labacher 2013 Wrong study design Lee 2014 Wrong study design Levine 2008 Wrong study design Lewis 2013 Wrong study design Maar 2016 Wrong study population Marsh 2014 Wrong delivery mechanism Mbuagbaw 2013 Wrong study design Michell 2014 Wrong study design Moskowitz 2009 Wrong delivery mechanism Muessig 2013 Wrong delivery mechanism O'donnell 2016 Wro

Lee 2014 Wrong study design Levine 2008 Wrong study design Lewis 2013 Wrong study design Maar 2016 Wrong study population Marsh 2014 Wrong delivery mechanism Mbuagbaw 2013 Wrong study design Michell 2014 Wrong study design Moskowitz 2009 Wrong delivery mechanism Muessig 2013 Wrong delivery mechanism O'donnell 2016 Wro ng study design Price 2009 Wrong study design Prieto 2016 Wrong intervention Puccio 2006 Wrong study design Pérez 2015 Wrong study design Ramanathan 2013 Wrong intervention Ranney 2014 Wrong topic of interest Redfern 2016 Wrong patient population Saranto 2009 Wrong study design Schnall 2014 Wrong delivery mechanism Schnall 2016 Wrong delivery mechanism Senn 2017 Wrong intervention Senn 2017a Wrong study design Shet 2010 Wrong study design Sidney 2012 Wrong study design Siedner 2012 Wrong study design Skeels 2006 Wrong topic of interest Smith 2015 Wrong study design Soltani 2012 Wrong topic of interest Soltani 2015 Wrong study design Suwamaru 2012 Wrong study design Thomas 2017 Wrong topic of interest Thompson 2016 Wrong study design Toefy 2016 Wrong study design Tran 2012 Wrong study design Uhrig 2012 Wrong study design Vahdat 2013 Wrong study design Vyas 2012 Wrong patient population WHO 2014 Wrong study design Wilkinson 2017 Wrong study design Ybarra 2016 Wrong study design Contributions of authors Heather Ames: involved in all steps of the process Claire Glenton: protocol development, CERQual, matrix analysis, write‐up Simon Lewin: protocol development, CERQual, reviewing the write‐up Tigest Tamrat: protocol development, search and screen, write‐up

ng study design Price 2009 Wrong study design Prieto 2016 Wrong intervention Puccio 2006 Wrong study design Pérez 2015 Wrong study design Ramanathan 2013 Wrong intervention Ranney 2014 Wrong topic of interest Redfern 2016 Wrong patient population Saranto 2009 Wrong study design Schnall 2014 Wrong delivery mechanism Schnall 2016 Wrong delivery mechanism Senn 2017 Wrong intervention Senn 2017a Wrong study design Shet 2010 Wrong study design Sidney 2012 Wrong study design Siedner 2012 Wrong study design Skeels 2006 Wrong topic of interest Smith 2015 Wrong study design Soltani 2012 Wrong topic of interest Soltani 2015 Wrong study design Suwamaru 2012 Wrong study design Thomas 2017 Wrong topic of interest Thompson 2016 Wrong study design Toefy 2016 Wrong study design Tran 2012 Wrong study design Uhrig 2012 Wrong study design Vahdat 2013 Wrong study design Vyas 2012 Wrong patient population WHO 2014 Wrong study design Wilkinson 2017 Wrong study design Ybarra 2016 Wrong study design Contributions of authors Heather Ames: involved in all steps of the process Claire Glenton: protocol development, CERQual, matrix analysis, write‐up Simon Lewin: protocol development, CERQual, reviewing the write‐up Tigest Tamrat: protocol development, search and screen, write‐up Eliud Akama: search and screen, data extraction, reviewing the write‐up Natalie Leon: protocol development, search and screen, data synthesis, matrix analysis, CERQual, write‐up Sources of support Internal sources No sources of support supplied External sources WHO Department of Reproductive Health and Research, Switzerland.

Tigest Tamrat: protocol development, search and screen, write‐up Eliud Akama: search and screen, data extraction, reviewing the write‐up Natalie Leon: protocol development, search and screen, data synthesis, matrix analysis, CERQual, write‐up Sources of support Internal sources No sources of support supplied External sources WHO Department of Reproductive Health and Research, Switzerland. This work was funded by the UNDP‐UNFPA‐UNICEF‐WHO‐World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), a cosponsored program executed by the World Health Organization (WHO). Declarations of interest Heather Ames: no declaration of interest Claire Glenton: no declaration of interest Simon Lewin: no declaration of interest Tigest Tamrat: no declaration of interest Eliud Akama: no declaration of interest Natalie Leon: no declaration of interest

Background Description of the condition Genital infections caused by Chlamydia trachomatis serovars D-K are the most prevalent bacterial sexually transmitted infection worldwide, with an estimated 106 million people being infected in 2008 (WHO 2012). In this protocol we use the term 'chlamydia' to describe these infections. Chlamydia is the most common notifiable infection in the USA, with 1,307,893 infections reported in 2011 compared with 309,341 cases of gonorrhoea, which is the second most common notifiable condition (CDC 2010). Chlamydia is also the most commonly reported infection in Europe (ECDC 2011), Australia (DoHA 2011) and Canada (PHAC 2006). Chlamydia is most common in young sexually active adults. The prevalence of chlamydia has been estimated to be about 3% to 5% in nationally representative samples of sexually experienced women and men aged 25 years and under in high-income countries (Fenton 2001; Klavs 2004; Miller 2004; Goulet 2010; Bozicevic 2011).

. Chlamydia is most common in young sexually active adults. The prevalence of chlamydia has been estimated to be about 3% to 5% in nationally representative samples of sexually experienced women and men aged 25 years and under in high-income countries (Fenton 2001; Klavs 2004; Miller 2004; Goulet 2010; Bozicevic 2011). C. trachomatis is a gram negative obligate intracellular bacterium, which infects columnar epithelium in the lower genital tract in women and men and can also infect the rectum, pharynx, conjunctiva (Stamm 2008) and placenta (Rours 2011). Chlamydia infection causes complications, most commonly resulting from spread from the lower to the upper genital tract. Upper genital tract infection occurs in both sexes but is more common and has more severe consequences in women (Stamm 2008). In women, chlamydia ascends to the upper genital tract in approximately 10% of cases to cause symptomatic pelvic inflammatory disease (PID) (Oakeshott 2010; Herzog 2012). The resulting tubal damage can then cause ectopic pregnancy, tubal infertility and chronic pelvic pain (Paavonen 2008). Although about 45% of tubal infertility might be attributable to chlamydia infection (Price 2012), the probability of tubal infertility in women who have had chlamydia is estimated to be only 1% to 4% (Land 2010; Kavanagh 2013). Chlamydia infection in pregnancy is associated with preterm labour (Rours 2011) and can infect the neonate, causing ophthalmia neonatorum and atypical pneumonia (Kohlhoff 2008). C. trachomatis can cause epididymo-orchitis in men, but its role in prostatitis and male infertility is not well-established (Stamm 2008). Chlamydia can also cause Reiter's syndrome in men (Stamm 2008) and is a co-factor for HIV infection, increasing both susceptibility and infectiousness (Fleming 1999).

ohlhoff 2008). C. trachomatis can cause epididymo-orchitis in men, but its role in prostatitis and male infertility is not well-established (Stamm 2008). Chlamydia can also cause Reiter's syndrome in men (Stamm 2008) and is a co-factor for HIV infection, increasing both susceptibility and infectiousness (Fleming 1999). Uncomplicated genital chlamydia infections are usually asymptomatic in both women and men (Stamm 2008) and untreated infections last more than a year on average (Althaus 2010). C. trachomatis can be treated with tetracyclines (usually doxycycline) or macrolide (usually azithromycin) antibiotics with short-term microbiological cure rates of 90% to 95% (Manhart 2013). Immunity after chlamydia infection is incomplete and repeated chlamydia infection is common (Batteiger 2010a). In studies of women enrolled from primary care and sexual health clinics and followed up prospectively, about 25% of women treated for chlamydia had the infection detected again in the year after treatment (Scott LaMontagne 2007; Walker 2012). There are several reasons for repeated detection of chlamydia. In one prospective study amongst young women in the USA, Batteiger et al. combined information about sexual behaviour and genotype from 183 women with more than one episode of chlamydia infection to estimate that about 66% of infections were probably acquired from a new partner, 17% were re-infections from untreated or inadequately-treated sexual partners, 14% were probable antibiotic treatment failures and 3% persisted without treatment (Batteiger 2010b). There is some evidence to suggest that immunity after natural clearance of chlamydia infection lasts longer than immunity after antibiotic treatment (Geisler 2013).

om untreated or inadequately-treated sexual partners, 14% were probable antibiotic treatment failures and 3% persisted without treatment (Batteiger 2010b). There is some evidence to suggest that immunity after natural clearance of chlamydia infection lasts longer than immunity after antibiotic treatment (Geisler 2013). Description of the intervention Screening of sexually active young adults is the only way to detect most chlamydia infections because of the lack of symptoms or clinical signs in most infected people. Screening is a process of identifying apparently healthy people who may be at increased risk of a disease or condition. They can then be offered information, further tests and appropriate treatment to reduce their risk and/or any complications arising from the disease or condition (UKNSC 2013). There are two goals of screening for genital chlamydia infection: first, to control the transmission of chlamydia and reduce the prevalence of infection in the population; and second to reduce the risk of complications, especially reproductive tract complications in women (Meyers 2007; NCSP 2010). Screening is a programme, not a test (Raffle 2007). This means that screening includes the whole system of events needed to reach the endpoint of reducing the risk of disease or complications. For chlamydia infection, screening includes offering a test to diagnose C. trachomatis, treating people with a positive test, partner notification to identify and treat sexual partners and repeated screening to detect and treat newly acquired infection or re-infection.

of reducing the risk of disease or complications. For chlamydia infection, screening includes offering a test to diagnose C. trachomatis, treating people with a positive test, partner notification to identify and treat sexual partners and repeated screening to detect and treat newly acquired infection or re-infection. The target group for chlamydia screening is usually defined by age and sex. For example, chlamydia screening in the USA is recommended for women aged 25 years and under (CDC 2010), in Australia for women under 25 years (RACGP 2007), and in the UK for women and men aged 25 years and under (NCSP 2010). Whilst behavioural and demographic factors can be used to identify groups at higher risk of chlamydia infection (Stergachis 1993; Gotz 2005), risk factors differ between populations and selective criteria can be difficult to apply in practice.

2007), and in the UK for women and men aged 25 years and under (NCSP 2010). Whilst behavioural and demographic factors can be used to identify groups at higher risk of chlamydia infection (Stergachis 1993; Gotz 2005), risk factors differ between populations and selective criteria can be difficult to apply in practice. Chlamydia screening can be offered systematically, using a population register to invite people in the target age group (van den Broek 2012). More commonly, screening is recommended as an opportunistic activity to be offered to eligible young adults attending healthcare services (RACGP 2007; CDC 2010; NCSP 2010). Repeated screening is recommended in some countries, given the frequency of repeated chlamydia and the fact that young adults may change sexual partners over time. In England, the National Chlamydia Screening Programme recommends a screening test every year or after a change of sexual partner (NCSP 2010). Visits for cervical cancer screening in young women can be used as an opportunity to offer chlamydia screening in some countries, where the target age groups and screening frequency overlap. In the UK, however, cervical cancer screening is only recommended for women over 25 years.

change of sexual partner (NCSP 2010). Visits for cervical cancer screening in young women can be used as an opportunity to offer chlamydia screening in some countries, where the target age groups and screening frequency overlap. In the UK, however, cervical cancer screening is only recommended for women over 25 years. How the intervention might work The way in which chlamydia screening might work depends on the goal of screening. To reduce chlamydia prevalence and incidence, the coverage of screening has to be high enough to identify and treat prevalent cases of chlamydia and to interrupt chains of chlamydia transmission in the population. Screening also has to be frequent enough to prevent repeated infections because of the limited immunity after treatment. Mathematical models show that chlamydia screening reduces prevalence over time; in several models, screening of 30% or more of the target population each year is needed to reduce chlamydia prevalence markedly (Regan 2008; Althaus 2012).

ugh to prevent repeated infections because of the limited immunity after treatment. Mathematical models show that chlamydia screening reduces prevalence over time; in several models, screening of 30% or more of the target population each year is needed to reduce chlamydia prevalence markedly (Regan 2008; Althaus 2012). There are two ways in which screening for chlamydia might work to prevent reproductive tract complications (Peterman 2009; Herzog 2013). First, direct prevention of PID occurs if screening detects and treats an endocervical chlamydia infection in an individual woman before the infection ascends in the genital tract to cause PID and subsequent tubal damage. The effectiveness of screening depends on the timing of progression from lower to upper genital tract infection. If PID occurs immediately, or shortly after the initial lower genital tract infection, there is no opportunity for screening to work (Smith 2007; Herzog 2012). Randomised controlled trials (RCTs) have shown that the incidence of clinically diagnosed PID is lower in women actively invited for chlamydia screening compared to those receiving usual care (Scholes 1996; Ostergaard 2000; Oakeshott 2010; Andersen 2011). Women infected with chlamydia who are enrolled into trials have persisting prevalent infections with an unknown date of infection. The trial findings and supportive evidence from mathematical modelling studies suggest, therefore, that PID development can occur during the course of infection (Herzog 2012). Second, prevention of the transmission of chlamydia through screening and treatment has an indirect effect on the risk of PID because the risk of becoming infected with chlamydia in the first place falls.

modelling studies suggest, therefore, that PID development can occur during the course of infection (Herzog 2012). Second, prevention of the transmission of chlamydia through screening and treatment has an indirect effect on the risk of PID because the risk of becoming infected with chlamydia in the first place falls. Prevention of PID should lead to a reduction in the incidence of ectopic pregnancy and tubal infertility if tubal scarring is prevented. It is, however, very difficult to measure the impact of chlamydia screening on these outcomes because women in the age groups at highest risk of chlamydia infection are usually using contraception. In one RCT, the incidence of ectopic pregnancy and infertility after 11 years of follow-up were similar in women who had received a single invitation to be screened for chlamydia and women who received usual care (Andersen 2002).

in the age groups at highest risk of chlamydia infection are usually using contraception. In one RCT, the incidence of ectopic pregnancy and infertility after 11 years of follow-up were similar in women who had received a single invitation to be screened for chlamydia and women who received usual care (Andersen 2002). There are also potential harms of chlamydia screening. First, a woman who has been treated for chlamydia becomes susceptible and is at risk of repeated infection and PID. It has been suggested that the risk of PID is higher with subsequent chlamydia infections (Hillis 1997), possibly because repeated exposure to C. trachomatis antigens can cause immune mediated tubal damage (Brunham 2005). Second, being diagnosed with a sexually transmitted infection can have a negative emotional and psychological impact on the infected person (Mills 2006; Gottlieb 2011). In one study in the USA, sexual partnerships broke down for 33% of women with a positive chlamydia test result compared with 11% of those receiving a negative result (Gottlieb 2011). Third, the experience of screening can cause anxiety. In a study in the UK, however, chlamydia screening did not increase anxiety or depression and did not reduce self-esteem (Campbell 2006).

n for 33% of women with a positive chlamydia test result compared with 11% of those receiving a negative result (Gottlieb 2011). Third, the experience of screening can cause anxiety. In a study in the UK, however, chlamydia screening did not increase anxiety or depression and did not reduce self-esteem (Campbell 2006). Why it is important to do this review Screening for chlamydia infection is widely recommended (RACGP 2007; CDC 2010; NCSP 2010; Low 2012) and widely practised. Rates of chlamydia testing amongst young adults are high (4000 to 9000 per 100,000 population) in several high-income countries (Bender 2011). There is a strong rationale for early detection and treatment of chlamydia infection in asymptomatic young adults to reduce both transmission and complications (Low 2013). Widespread screening for asymptomatic chlamydia infection has the potential to cause harm, however, especially if the rate of repeated infection after treatment outweighs the benefits of detecting and treating asymptomatic infections, or if receiving a diagnosis of chlamydia results in the breakdown of a relationship (O'Farrell 2013). There are few data about long-term trends in chlamydia prevalence in countries that recommend chlamydia screening. In the USA, repeated cross-sectional studies show that chlamydia prevalence fell between 1999 and 2008 in 14 to 39 year olds as a whole, but not in 15 to 25 year old women, who are the target population for screening (Datta 2012).

long-term trends in chlamydia prevalence in countries that recommend chlamydia screening. In the USA, repeated cross-sectional studies show that chlamydia prevalence fell between 1999 and 2008 in 14 to 39 year olds as a whole, but not in 15 to 25 year old women, who are the target population for screening (Datta 2012). There is a systematic review of the effectiveness of chlamydia screening interventions in studies published up to 2007 (Low 2009). There are new RCTs showing that a one-off screening invitation could reduce the incidence of PID one year later (Scholes 1996; Ostergaard 2000). Also, we know that there are new completed trials with PID (Oakeshott 2010; Andersen 2011) and transmission (van den Broek 2012) as endpoints, and at least one ongoing trial (Hocking 2012). It is therefore important to develop a Cochrane review about this issue. Objectives To assess the effects and safety of chlamydia screening in pregnant and non-pregnant women and in men, compared with standard care, on chlamydia transmission and on complications of infection. Methods Criteria for considering studies for this review Types of studies Randomised controlled trials, non-randomised controlled trials.

Objectives To assess the effects and safety of chlamydia screening in pregnant and non-pregnant women and in men, compared with standard care, on chlamydia transmission and on complications of infection. Methods Criteria for considering studies for this review Types of studies Randomised controlled trials, non-randomised controlled trials. If there are no RCTs addressing a primary outcome of chlamydia screening, we will include non-randomised studies. Chlamydia screening is a complex population-based intervention, one aim of which is to reduce chlamydia prevalence in a population. This is an outcome that is unlikely to be studied in RCTs (Cochrane 2011). Cluster-randomisation is technically possible, but our previous systematic review did not find any RCTs that examined the effect of chlamydia screening on chlamydia transmission (Low 2009). With cluster allocation, trials will be eligible if the groups receive the intervention during different time periods, as long as baseline and outcome data were collected prospectively using the same criteria throughout the trial period. The risk of bias in methods of allocation will be assessed and results from randomised and non-randomised study designs will be analysed separately. We will exclude cohort studies, case-control studies, and interrupted time-series studies. Types of participants Women and men (heterosexual or men who have sex with men) aged over 13 years in any setting. The minimum age group is arbitrary but aims to identify only studies of sexually transmitted chlamydia infections.

With cluster allocation, trials will be eligible if the groups receive the intervention during different time periods, as long as baseline and outcome data were collected prospectively using the same criteria throughout the trial period. The risk of bias in methods of allocation will be assessed and results from randomised and non-randomised study designs will be analysed separately. We will exclude cohort studies, case-control studies, and interrupted time-series studies. Types of participants Women and men (heterosexual or men who have sex with men) aged over 13 years in any setting. The minimum age group is arbitrary but aims to identify only studies of sexually transmitted chlamydia infections. Types of interventions Intervention: Screening for sexually transmitted genital chlamydia infection, defined as the offer of a test to apparently healthy people to identify those at increased risk of chlamydia infection. This definition is adapted from the UK National Screening Committee (UKNSC 2013). We will include any test used to diagnose genital chlamydia infection. Comparison: Inactive control (no offer of screening or standard care).

Types of interventions Intervention: Screening for sexually transmitted genital chlamydia infection, defined as the offer of a test to apparently healthy people to identify those at increased risk of chlamydia infection. This definition is adapted from the UK National Screening Committee (UKNSC 2013). We will include any test used to diagnose genital chlamydia infection. Comparison: Inactive control (no offer of screening or standard care). Types of outcome measures Eligible trials must include at least one of the pre-specified primary outcomes. The primary outcomes will be measures of morbidity that a chlamydia screening programme aims to prevent. One criterion for assessing the effectiveness of a screening programme is that ‘There should be evidence from high quality Randomised Controlled Trials that the screening programme is effective in reducing mortality or morbidity’ (http://www.screening.nhs.uk/criteria). Primary outcomes We include one primary outcome for each goal of chlamydia screening: Outcome for C. trachomatis transmission: Prevalence of chlamydia infection in women and men at least 12 months after the start of the screening intervention. Prevalence is estimated as the number of positive chlamydia tests divided by the number of people tested.

Primary outcomes We include one primary outcome for each goal of chlamydia screening: Outcome for C. trachomatis transmission: Prevalence of chlamydia infection in women and men at least 12 months after the start of the screening intervention. Prevalence is estimated as the number of positive chlamydia tests divided by the number of people tested. Outcomes for reproductive tract morbidity: Incidence of upper genital tract infection in women and men in the 12 months after the offer of screening. Pelvic inflammatory disease (women) or epididymitis (men) are clinical diagnoses, made using clinical criteria defined in advance by the authors. Examples include criteria published by the US Centers for Disease Control and Prevention (CDC 2010), or Hager and Eschenbach (Hager 1983). Outcome for chlamydia infection in pregnancy: Incidence of preterm delivery. Preterm delivery is defined as delivery at a gestational age of less than 37 weeks, with subgroups of gestational ages less than 32 weeks and less than 35 weeks (Rours 2011). Secondary outcomes Outcomes measured in all participants: Proportion of participants receiving the intervention, defined as the number tested for chlamydia divided by the number eligible and invited to take part. Harms of screening, including psychological distress, partner violence, relationship breakdown, using definitions described by the authors. Outcomes measured in women who were not pregnant during the trial or in men: Prevalence of chronic female pelvic pain, defined as patient-reported pain in the lower abdomen or pelvis lasting at least six months (Paavonen 2008).

Harms of screening, including psychological distress, partner violence, relationship breakdown, using definitions described by the authors. Outcomes measured in women who were not pregnant during the trial or in men: Prevalence of chronic female pelvic pain, defined as patient-reported pain in the lower abdomen or pelvis lasting at least six months (Paavonen 2008). Prevalence of female or male infertility, defined using a clinical definition of lack of pregnancy despite unprotected intercourse for 12 months or more (Paavonen 2008). Outcomes measured in women who were pregnant during the trial, or in their infant: Incidence of C. trachomatis neonatal conjunctivitis, defined as C. trachomatis isolated from the conjunctiva by culture or detected by nucleic acid amplification test (Kohlhoff 2008). Incidence of C. trachomatis neonatal pneumonitis, defined as signs of lower respiratory tract infection presenting between 4 and 12 weeks with C. trachomatis isolated from the nasopharynx by culture or detected by nucleic acid amplification test (Kohlhoff 2008). The following outcome will not be included: uptake of chlamydia screening. Screening uptake is an intermediate outcome. The relationship between uptake of screening and the primary outcomes has not been quantified so, for a given level of screening uptake, it is not possible to predict the expected reduction in chlamydia prevalence or incidence of pelvic inflammatory disease.

mydia screening. Screening uptake is an intermediate outcome. The relationship between uptake of screening and the primary outcomes has not been quantified so, for a given level of screening uptake, it is not possible to predict the expected reduction in chlamydia prevalence or incidence of pelvic inflammatory disease. Search methods for identification of studies We will attempt to identify trials meeting the inclusion criteria irrespective of their language, publication date and publication status (published, unpublished, in press, and in progress). We will use both electronic searching in bibliographic databases and handsearching, as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). The results of all searches will be downloaded and managed using Endnote bibliographic software. Duplicate records of the same study will be deleted.

Search methods for identification of studies We will attempt to identify trials meeting the inclusion criteria irrespective of their language, publication date and publication status (published, unpublished, in press, and in progress). We will use both electronic searching in bibliographic databases and handsearching, as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). The results of all searches will be downloaded and managed using Endnote bibliographic software. Duplicate records of the same study will be deleted. Electronic searches We will contact the Trials Search Coordinator (TSC) of the Sexually Transmitted Infections Cochrane Review Group in order to implement a comprehensive search strategy which seeks to capture as many relevant trials as possible in electronic databases. For this purpose, we will use a combination of controlled vocabulary (MeSH, Emtree, DeCS, including exploded terms) and free-text terms (considering spelling variants, plurals, synonyms, acronyms and abbreviations) for “genital Chlamydia infection”and “screening”, with field labels, truncation, proximity operators and boolean operators. The sensitivity of the search strategies will be improved by including keywords from relevant trials detected by earlier searches. We present the search strategies in Appendix 1 (Electronic search strategies). Specifically, we will search in the following electronic databases: MEDLINE, Ovid platform: inception to present. MEDLINE In-Process & Other Non-Indexed Citations, Ovid platform: inception to present.

Electronic searches We will contact the Trials Search Coordinator (TSC) of the Sexually Transmitted Infections Cochrane Review Group in order to implement a comprehensive search strategy which seeks to capture as many relevant trials as possible in electronic databases. For this purpose, we will use a combination of controlled vocabulary (MeSH, Emtree, DeCS, including exploded terms) and free-text terms (considering spelling variants, plurals, synonyms, acronyms and abbreviations) for “genital Chlamydia infection”and “screening”, with field labels, truncation, proximity operators and boolean operators. The sensitivity of the search strategies will be improved by including keywords from relevant trials detected by earlier searches. We present the search strategies in Appendix 1 (Electronic search strategies). Specifically, we will search in the following electronic databases: MEDLINE, Ovid platform: inception to present. MEDLINE In-Process & Other Non-Indexed Citations, Ovid platform: inception to present. MEDLINE Daily Update, Ovid platform: inception to present. EMBASE.com: inception to present. The Cochrane Central Register of Controlled Trials (CENTRAL), Ovid platform: inception to present. LILACS, iAHx interface: inception to present. CINAHL: inception to present. Database of Abstracts of Reviews of Effects (DARE): inception to present. PsycINFO: inception to present.

MEDLINE Daily Update, Ovid platform: inception to present. EMBASE.com: inception to present. The Cochrane Central Register of Controlled Trials (CENTRAL), Ovid platform: inception to present. LILACS, iAHx interface: inception to present. CINAHL: inception to present. Database of Abstracts of Reviews of Effects (DARE): inception to present. PsycINFO: inception to present. We will searching MEDLINE using the Cochrane highly sensitive search strategy for identifying RCTs: sensitivity and precision maximizing version (2008 revision), Ovid format (Higgins 2011a). The LILACS search strategy will be combined with the RCT filter of the iAHx interface. These searches will be updated within 6 months before publication of the review. Searching other resources We will attempt to identify additional relevant trials using the following methods: 1. Searching in the Sexually Transmitted Infections (STI) Cochrane Review Group's Specialised Register, which includes RCTs and controlled clinical trials, from 1944 to 2012, located through: Electronic searching in MEDLINE, EMBASE and CENTRAL.

Searching other resources We will attempt to identify additional relevant trials using the following methods: 1. Searching in the Sexually Transmitted Infections (STI) Cochrane Review Group's Specialised Register, which includes RCTs and controlled clinical trials, from 1944 to 2012, located through: Electronic searching in MEDLINE, EMBASE and CENTRAL. Handsearching in those journals not indexed in MEDLINE or EMBASE (according to the journals’ master list of the STI Cochrane Review Group): Anatolian Journal of Obstetrics & Gynecology, Current Medical Literature Gynecology & Obstetrics, Current Obstetrics and Gynecology Reports, ISRN Obstetrics and Gynecology, Journal of South Asian Federation of Obstetrics & Gynecology, Obstetrics and Gynecology International, Obstetrics Gynaecology and Reproductive Medicine, Sexual Science: the newsletter of the Society for the Scientific Study of Sexuality and Sexualities. 2. Searching trials registers: WHO International Clinical Trials Registry Platform (ICTRP) portal (http://apps.who.int/trialsearch/): inception to present. ClinicalTrials.gov (http://clinicaltrials.gov/): inception to present. 3. Searching Web of Science®: inception to present. 4. Searching for grey literature in System for Information on Grey Literature in Europe “OpenGrey”(http://www.opengrey.eu/): inception to present.

WHO International Clinical Trials Registry Platform (ICTRP) portal (http://apps.who.int/trialsearch/): inception to present. ClinicalTrials.gov (http://clinicaltrials.gov/): inception to present. 3. Searching Web of Science®: inception to present. 4. Searching for grey literature in System for Information on Grey Literature in Europe “OpenGrey”(http://www.opengrey.eu/): inception to present. 5. Contacting authors of all RCTs identified by other methods. A comprehensive list of trials included in the review along with the criteria for considering studies will be sent to the first author of each included study, asking for any additional studies published or unpublished that might be relevant. 6. Handsearching conference proceeding abstracts from the following events: The International Society for Sexually Transmitted Diseases Research - ISSTDR (http://www.isstdr.org/): 2007, 2009 and 2011. The British Association for Sexual Health and HIV - BASHH (http://www.bashh.org/): 2004, 2006, 2007 and 2009. International Congress on Infectious Diseases - ICID (http://www.isid.org/): 2010 and 2012. The International Union against Sexually Transmitted Infections - IUSTI (http://www.iusti.org/): 2011 and 2012. International Society for Infectious Diseases - ISID (http://www.isid.org/): 2011. International Meeting on Emerging Diseases and Surveillance - IMED (http://www.isid.org/): 2007, 2009 and 2011. Interscience Conference on Antimicrobial Agents and Chemotherapy - ICAAC (http://www.icaac.org/): 2011 and 2012. The International Federation of Gynecology and Obstetrics - FIGO (http://www.figo2012.org/home/): 2012.

International Meeting on Emerging Diseases and Surveillance - IMED (http://www.isid.org/): 2007, 2009 and 2011. Interscience Conference on Antimicrobial Agents and Chemotherapy - ICAAC (http://www.icaac.org/): 2011 and 2012. The International Federation of Gynecology and Obstetrics - FIGO (http://www.figo2012.org/home/): 2012. 7. Handsearching previous systematic reviews and other relevant publications on the same topic. 8. Handsearching reference lists of all relevant RCTs identified by other methods. Data collection and analysis Selection of studies Two review authors (NL, SR) will review titles and abstracts of articles identified by the search strategy independently, using a pilot-tested form to document potential eligibility. Disagreements will be discussed. We will obtain the full text manuscripts of all articles agreed as being potentially eligible by both review authors, articles about which the authors still disagree after discussion, and articles with no abstract if there is insufficient information available from the title or publication type to make a decision. If the disagreement cannot be resolved by discussion, the full text of the article will be obtained. The abstracts of articles identified through searching other resources will be assessed using the same criteria as for studies identified through electronic database searches.

Data collection and analysis Selection of studies Two review authors (NL, SR) will review titles and abstracts of articles identified by the search strategy independently, using a pilot-tested form to document potential eligibility. Disagreements will be discussed. We will obtain the full text manuscripts of all articles agreed as being potentially eligible by both review authors, articles about which the authors still disagree after discussion, and articles with no abstract if there is insufficient information available from the title or publication type to make a decision. If the disagreement cannot be resolved by discussion, the full text of the article will be obtained. The abstracts of articles identified through searching other resources will be assessed using the same criteria as for studies identified through electronic database searches. Two independent review authors will examine full text articles using a pilot-tested form to assess eligibility for inclusion. Studies identified by both authors as being eligible for inclusion will be included in the review. Where there are discrepancies, the authors will discuss the article and reach a consensus decision. If there is no agreement, a third independent author will adjudicate to make a final decision about eligibility.

clusion. Studies identified by both authors as being eligible for inclusion will be included in the review. Where there are discrepancies, the authors will discuss the article and reach a consensus decision. If there is no agreement, a third independent author will adjudicate to make a final decision about eligibility. We will use a flow chart to document the numbers of articles assessed and included or excluded at each stage, with a summary of reasons for exclusion. The flow chart will show the total number of studies included in the review and the total number of articles pertaining to these studies. We will record briefly the characteristics of studies excluded from the review if readers might expect them to have been included. Data extraction and management We will develop and pilot standardised forms to extract data about: Study location and setting Trial design and power calculation Ethical approval Inclusion and exclusion criteria Baseline characteristics of trial participants including sex, age, sexual orientation, pregnancy status for women, diagnostic test used to detect C. trachomatis Types of intervention: opportunistic or systematic invitation for screening; number of screening rounds, screening interval Types of comparison group: usual care, alternative screening method Types of outcome: primary, secondary Reporting of methodological characteristics (see next section, Assessment of risk of bias in included studies for details) We will extract the following numerical data: Number of people assessed for eligibility Numbers randomised to intervention and comparison groups

Types of comparison group: usual care, alternative screening method Types of outcome: primary, secondary Reporting of methodological characteristics (see next section, Assessment of risk of bias in included studies for details) We will extract the following numerical data: Number of people assessed for eligibility Numbers randomised to intervention and comparison groups Numbers receiving screening in intervention and comparison groups (at each screening round if multiple rounds) Numbers included in analyses in intervention and comparison groups Numbers with outcomes in intervention and comparison groups One author (SR) will extract data about study characteristics and a second will check these details. They will resolve discrepancies by discussion or a third independent author will adjudicate. Two appropriately authors (from all co-authors) will extract and enter numerical data independently from each included study into Epidata using a structured form. If there are multiple publications relating to the same study, data items can be extracted from different publications. If there are discrepancies between publications about a data item, we will use the data presented in the main trial publication (the publication that includes the results for the primary outcome) or the first chronological publication reporting that data item. Articles in languages other than English will either be translated first and then duplicate data extraction conducted as above or, if there are two review authors who understand the language of publication, they will extract the data directly.

Two appropriately authors (from all co-authors) will extract and enter numerical data independently from each included study into Epidata using a structured form. If there are multiple publications relating to the same study, data items can be extracted from different publications. If there are discrepancies between publications about a data item, we will use the data presented in the main trial publication (the publication that includes the results for the primary outcome) or the first chronological publication reporting that data item. Articles in languages other than English will either be translated first and then duplicate data extraction conducted as above or, if there are two review authors who understand the language of publication, they will extract the data directly. The two files will be compared using the validation function available in Epidata. Discrepancies in data extraction or data entry will be resolved by consensus. If there is no agreement a third independent author will adjudicate to make a final decision. The agreed data will be entered into Review Manager 5 (RevMan) software. If there are insufficient details given to allow the extraction of numerical data, the study will be included and the results described.

The two files will be compared using the validation function available in Epidata. Discrepancies in data extraction or data entry will be resolved by consensus. If there is no agreement a third independent author will adjudicate to make a final decision. The agreed data will be entered into Review Manager 5 (RevMan) software. If there are insufficient details given to allow the extraction of numerical data, the study will be included and the results described. Assessment of risk of bias in included studies We will assess the methods reported to have been used in the design and execution of all included trials. The assessment will determine whether there is a risk of bias that would over- or underestimate the effect of the intervention on one or more outcomes (Higgins 2011a). This assessment relies on reports of methods described by trial authors in publications and, where available, trial protocols. For any trial, the findings of the assessment can only say whether there is a risk of biased results, and cannot determine whether the results themselves are or are not biased. For both randomised and non-randomised trials we will assess the risk of five specific sources of bias: selection bias, performance bias, detection bias, attrition bias and reporting bias; and will record any other biases related to a particular trial. For RCTs we will use the Cochrane Collaboration's ‘Risk of bias’ tool and criteria in the Cochrane Handbook (Table 8.5.d) to assess these in the relevant domains of the reported methods and results (Cochrane 2011).

For both randomised and non-randomised trials we will assess the risk of five specific sources of bias: selection bias, performance bias, detection bias, attrition bias and reporting bias; and will record any other biases related to a particular trial. For RCTs we will use the Cochrane Collaboration's ‘Risk of bias’ tool and criteria in the Cochrane Handbook (Table 8.5.d) to assess these in the relevant domains of the reported methods and results (Cochrane 2011). Selection bias is only the domain for which there are important differences in assessing the risk of bias in randomised and non-randomised controlled trials. For non-randomised controlled trials we will use the UK National Insitute of Health and Care Excellence (NICE) 'methodology checklist' for cohort studies to assess the risk of selection bias (NICE 2012). The NICE methodology checklist format follows that of the Cochrance Collaboration tool, with criteria to assess bias in each domain and a choice of low, high or unclear risk of bias. We will use the Cochrane risk of bias tool to assess non-randomised controlled trials for risks of performance, detection, attrition and reporting biases. Assessors will record whether there is a low, high or unclear risk of bias in each domain of each included trial and give a justification for their decision. For each included trial there will be two independent assessors, including at least one expert in trial methodology (NL) and at least one expert in chlamydia screening (HG). They will resolve discrepancies by discussion. If they cannot agree, a third author will adjudicate.

and give a justification for their decision. For each included trial there will be two independent assessors, including at least one expert in trial methodology (NL) and at least one expert in chlamydia screening (HG). They will resolve discrepancies by discussion. If they cannot agree, a third author will adjudicate. The domains and their source are summarised here: (1a) Random sequence generation (possible selection bias, Cochrane ‘Risk of bias’ tool) Selection bias could occur if allocation to intervention or control groups can be predicted in advance and if participants or clusters of participants are enrolled selectively. The method used to generate the allocation sequence should be unpredictable and should balance prognostic factors, on average, across intervention and comparison groups. We will assess the method as being at: low risk of bias (adequate description of a truly random process, e.g. random number tables, computer generated random numbers); high risk of bias (explicit description of an allocation process that is not truly random, e.g. odd or even dates of birth of individuals, clusters of participants selected for implementation of the intervention with subsequent enrolment of comparison groups); unclear risk of bias (description that does not include enough information to decide whether sequence generation was truly random or not). (1b) Allocation concealment (possible selection bias, Cochrane ‘Risk of bias’ tool)

high risk of bias (explicit description of an allocation process that is not truly random, e.g. odd or even dates of birth of individuals, clusters of participants selected for implementation of the intervention with subsequent enrolment of comparison groups); unclear risk of bias (description that does not include enough information to decide whether sequence generation was truly random or not). (1b) Allocation concealment (possible selection bias, Cochrane ‘Risk of bias’ tool) Selection bias can occur if participants or clusters of participants are selectively enrolled and allocated to a particular group and if their characteristics are associated with the outcome. If the sequence has been randomly generated, selective enrolment can occur if the next assignment is known before allocation. Concealment of the allocation sequence up to the point of assignment prevents selective assignment to a particular intervention group. We will assess the methods of allocation concealment as: low risk of bias (adequate description of a process that prevented foreknowledge of allocation up to the point at which assignment was recorded, e.g. telephone or central randomisation); high risk of bias (description of a process that meant that those assigning participants or clusters of participants knew or could predict the allocation in advance); unclear risk of bias (insufficient details to be able to decide whether the allocation was concealed or not). (1c) Systematic differences between comparison groups (possible selection bias, NICE 'methodology checklist')

high risk of bias (description of a process that meant that those assigning participants or clusters of participants knew or could predict the allocation in advance); unclear risk of bias (insufficient details to be able to decide whether the allocation was concealed or not). (1c) Systematic differences between comparison groups (possible selection bias, NICE 'methodology checklist') In a non-randomised trial, selection bias can occur because of the lack of a random allocation sequence and concealed allocation. If the person assigning individuals or clusters to a particular group knows about the distribution of factors associated with the outcome, they might introduce selection bias. We will assess the risk of selection bias as: low risk of bias ((a) the reason for participant allocation to treatment groups is not expected to affect the outcomes of the study, (b) there were attempts made within the design or analysis to balance the comparison groups for potential confounders and (c) the groups were comparable at baseline for all known major confounders and prognostic factors); high risk of bias (any of (a) to (c) above not fulfilled); unclear risk of bias (insufficient details to be able to decide whether there was a risk of systematic differences between comparison groups). (2) Blinding of participants and personnel (possible performance bias)

low risk of bias ((a) the reason for participant allocation to treatment groups is not expected to affect the outcomes of the study, (b) there were attempts made within the design or analysis to balance the comparison groups for potential confounders and (c) the groups were comparable at baseline for all known major confounders and prognostic factors); high risk of bias (any of (a) to (c) above not fulfilled); unclear risk of bias (insufficient details to be able to decide whether there was a risk of systematic differences between comparison groups). (2) Blinding of participants and personnel (possible performance bias) Screening is an intervention that involves systematic differences in the delivery of a health service. Personnel who offer chlamydia screening tests might offer other sexual health information, advice or interventions, such as condoms, that could affect participants’ risk of chlamydia infection or another outcome. Such information and interventions could also be considered a part of the screening programme, however. Trial participants or clusters of participants in an inactive ‘usual care’ control group might also be considered blinded if they do not know that they are part of a trial. For each included trial, we will describe the intervention. We will consider studies as being at low risk of bias if participants were blinded or if the lack of blinding would be unlikely to affect results for a particular outcome. (3) Blinding of outcome assessment (possible detection bias)

Screening is an intervention that involves systematic differences in the delivery of a health service. Personnel who offer chlamydia screening tests might offer other sexual health information, advice or interventions, such as condoms, that could affect participants’ risk of chlamydia infection or another outcome. Such information and interventions could also be considered a part of the screening programme, however. Trial participants or clusters of participants in an inactive ‘usual care’ control group might also be considered blinded if they do not know that they are part of a trial. For each included trial, we will describe the intervention. We will consider studies as being at low risk of bias if participants were blinded or if the lack of blinding would be unlikely to affect results for a particular outcome. (3) Blinding of outcome assessment (possible detection bias) For chlamydia screening interventions, adequate descriptions of blinding of those assessing the outcomes are important. We will group outcomes that are objectively assessed, e.g. chlamydia test results obtained from automated diagnostic systems, and those that are subjective, e.g. clinical diagnosis of pelvic inflammatory disease.

a screening interventions, adequate descriptions of blinding of those assessing the outcomes are important. We will group outcomes that are objectively assessed, e.g. chlamydia test results obtained from automated diagnostic systems, and those that are subjective, e.g. clinical diagnosis of pelvic inflammatory disease. The incidence of clinically diagnosed pelvic inflammatory disease is a primary outcome of chlamydia screening interventions. The main symptom is lower abdominal pain, which is common and non-specific. Knowledge of group assignment could influence the interpretation of symptoms by both trial participants and personnel delivering the intervention in unpredictable ways. For example, healthcare providers who know whether a woman has been screened for chlamydia might be more likely to assign a diagnosis of pelvic inflammatory disease to a woman who presents with abdominal pain because of increased awareness of the complications of chlamydia infection. On the other hand, they might be reassured if the test was negative or if treatment had been given and then interpret abdominal pain with or without accompanying signs as resulting from another cause. Women who have accepted or declined screening might also modify their assessment of symptoms or their health-seeking behaviour. For subjective outcomes, we will assess methods as follows:

or if treatment had been given and then interpret abdominal pain with or without accompanying signs as resulting from another cause. Women who have accepted or declined screening might also modify their assessment of symptoms or their health-seeking behaviour. For subjective outcomes, we will assess methods as follows: low risk of bias (adequate description of assessment that reduced the risk of bias, e.g. uniform assessment of all trials participants by an independent assessor blinded to allocation, or assessment of diagnoses by an independent assessment panel blinded to allocation); high risk of bias (assessment of outcomes by personnel who knew the group assignment); unclear risk of bias (insufficient information to determine whether outcome assessment was blinded or not). (4) Incomplete outcome data (possible attrition bias due to the amount, nature and handling of incomplete outcome data)

low risk of bias (adequate description of assessment that reduced the risk of bias, e.g. uniform assessment of all trials participants by an independent assessor blinded to allocation, or assessment of diagnoses by an independent assessment panel blinded to allocation); high risk of bias (assessment of outcomes by personnel who knew the group assignment); unclear risk of bias (insufficient information to determine whether outcome assessment was blinded or not). (4) Incomplete outcome data (possible attrition bias due to the amount, nature and handling of incomplete outcome data) For each outcome or class of outcomes we will describe the completeness of data and exclusions from analysis in each included trial. We will state whether analyses were conducted and reported according to intention-to-treat or not. Where reported we will state numbers included in the analysis as a proportion of the totals randomised to intervention and comparison groups, reasons for attrition or exclusion, and whether missing data were balanced across groups or were related to outcomes. Where sufficient information is reported, or can be supplied by the trial authors, we will re-include missing data in our analyses. We will use a cut-off of 20% to assign trials with missing outcome data as being at low or high risk of bias. In addition, we will assess methods as being at: low risk of bias (e.g. no missing outcome data; missing outcome data balanced across groups);

For each outcome or class of outcomes we will describe the completeness of data and exclusions from analysis in each included trial. We will state whether analyses were conducted and reported according to intention-to-treat or not. Where reported we will state numbers included in the analysis as a proportion of the totals randomised to intervention and comparison groups, reasons for attrition or exclusion, and whether missing data were balanced across groups or were related to outcomes. Where sufficient information is reported, or can be supplied by the trial authors, we will re-include missing data in our analyses. We will use a cut-off of 20% to assign trials with missing outcome data as being at low or high risk of bias. In addition, we will assess methods as being at: low risk of bias (e.g. no missing outcome data; missing outcome data balanced across groups); high risk of bias (e.g. numbers or reasons for missing data imbalanced across groups; ‘as treated’ analysis done with substantial departure of intervention received from that assigned at randomisation); unclear risk of bias (insufficient information about missing data or exclusions from analysis). (5) Selective outcome reporting (possible reporting bias) Where available, we will assess the trial protocol and trial registration documents as well as articles or publications resulting from a trial. We will describe the documents available for each included study and assess the methods as follows: low risk of bias (adequate description that all pre-specified outcomes and all expected outcomes of interest to the review have been reported);

Where available, we will assess the trial protocol and trial registration documents as well as articles or publications resulting from a trial. We will describe the documents available for each included study and assess the methods as follows: low risk of bias (adequate description that all pre-specified outcomes and all expected outcomes of interest to the review have been reported); high risk of bias (explicit evidence that not all pre-specified outcomes have been reported, that one or more reported primary outcomes were not pre-specified, that outcomes of interest are reported incompletely and cannot be used in the review, or that there are no results for a key outcome that would have been expected to have been reported); unclear risk of bias (insufficient information to decide whether selective reporting bias is likely or not). (6) Other biases For each included trial we will describe other potential sources of bias. For example, the effects of chlamydia screening can be assessed in cluster-randomised trials. We will describe design-specific risks of bias in domains such as recruitment, baseline imbalances and appropriate statistical analysis. We will report whether these are likely to result in a low, high or unclear risk of bias. (7) Overall risk of bias

For each included trial we will describe other potential sources of bias. For example, the effects of chlamydia screening can be assessed in cluster-randomised trials. We will describe design-specific risks of bias in domains such as recruitment, baseline imbalances and appropriate statistical analysis. We will report whether these are likely to result in a low, high or unclear risk of bias. (7) Overall risk of bias For each primary outcome we will assess the overall level of evidence provided by the included trials using the GRADE approach, as incorporated in Review Manager 5. We will produce a 'Summary of findings' table (Higgins 2011b). The level of evidence is summarised as high, moderate, low or very low. We will give justifications for changing the level of evidence depending on findings about: study limitations; consistency of results; directness of evidence; imprecision; publication bias. Measures of treatment effect All pre-specified primary and secondary outcomes are dichotomous. The treatment effect or harmful effect for each, comparing the outcome in those receiving the screening intervention with the control group, can be expressed as a relative risk (RR) with 95% confidence intervals (CI). An advantage of the RR is that it can be interpreted easily for both high and low event rates. We will also calculate the risk difference (RD, 95% CI), the actual difference in the event rate between intervention and control groups. We will use the risk difference to calculate the number needed to treat to benefit (NNTB) or number needed to treat to harm (NNTH).

erpreted easily for both high and low event rates. We will also calculate the risk difference (RD, 95% CI), the actual difference in the event rate between intervention and control groups. We will use the risk difference to calculate the number needed to treat to benefit (NNTB) or number needed to treat to harm (NNTH). For the primary outcome of chlamydia prevalence, we will report the overall effect estimate at the level of the cluster, and state whether the analysis has taken into account the correlation between individuals within a cluster. We will not combine estimates from individually and cluster-randomised trials. Unit of analysis issues If a trial involves more than one intervention group, we will describe all of the groups in the ‘Characteristics of included studies’ table. We will analyse only those relevant to the pre-specified primary and secondary outcomes; however. If necessary, we will combine results of multiple intervention or comparison groups so that only single pair wise comparisons are made. Cluster-randomised trials of chlamydia screening interventions might measure the effect of the intervention in a geographic area or a school community. In trials of chlamydia screening, the intervention affects not only individuals who are screened and treated (direct effect), but their sexual partners and members of the same sexual network (indirect effect). The indirect effect of screening can reduce the level of repeated exposure to infection of individuals within a cluster.

of chlamydia screening, the intervention affects not only individuals who are screened and treated (direct effect), but their sexual partners and members of the same sexual network (indirect effect). The indirect effect of screening can reduce the level of repeated exposure to infection of individuals within a cluster. Dealing with missing data We will report the percentage of observations with missing data in each included trial. We will use sensitivity analysis to explore the effect of including or excluding trials with high levels of missing data. For each outcome we will attempt to analyse data according to the intention-to-treat principle, with all participants included in the group to which they were randomised and exclusion only of participants with missing outcome data. Assessment of heterogeneity We will report statistical heterogeneity in results between studies using I2, tau2 and Chi2 statistics obtained from analyses in Review Manager 5. We will use the I2 statistic to quantify the percentage of variability between the results that is due to heterogeneity rather than sampling error (Higgins 2002). We will take into account the fact that I2 values are affected by the number of studies, the magnitude and direction of effects in individual trials, and the strength of evidence of heterogeneity. In general, we will consider I2 values less than 40% as showing little evidence of statistical heterogeneity.

iggins 2002). We will take into account the fact that I2 values are affected by the number of studies, the magnitude and direction of effects in individual trials, and the strength of evidence of heterogeneity. In general, we will consider I2 values less than 40% as showing little evidence of statistical heterogeneity. Assessment of reporting biases We will look for evidence of publication and other reporting biases using funnel plots that plot the effect size against precision. If there are more than 10 studies in a meta-analysis we will use statistical tests of funnel plot asymmetry for continuous (Egger 1997) or binary (Harbord 2005) endpoints. Data synthesis We will use narrative syntheses to describe the results of trials where there are too few studies for meta-analysis or where we consider that meta-analysis is not clinically meaningful. We will use forest plots to display results of trials examining the same outcome.

Assessment of reporting biases We will look for evidence of publication and other reporting biases using funnel plots that plot the effect size against precision. If there are more than 10 studies in a meta-analysis we will use statistical tests of funnel plot asymmetry for continuous (Egger 1997) or binary (Harbord 2005) endpoints. Data synthesis We will use narrative syntheses to describe the results of trials where there are too few studies for meta-analysis or where we consider that meta-analysis is not clinically meaningful. We will use forest plots to display results of trials examining the same outcome. Where appropriate, we will combine data using meta-analysis conducted in Review Manager 5. If there are trials that examine the same intervention and are assumed to measure the same underlying effect in similar populations, we will use a fixed-effect model. If there is clinical heterogeneity or statistical evidence of substantial statistical heterogeneity we will use a random-effects model to estimate the average treatment effect across trials. The results will be presented as the summary RR (95% CI) with I2 and tau2 estimates. For meta-analyses with at least three studies combined using a random-effects model we will also calculate a prediction interval to examine the range of effect estimates that might be expected in different settings or populations (Riley 2011). We will not combine results from randomised and non-randomised trials in the same meta-analysis, but will compare these in a sensitivity analysis.

effects model we will also calculate a prediction interval to examine the range of effect estimates that might be expected in different settings or populations (Riley 2011). We will not combine results from randomised and non-randomised trials in the same meta-analysis, but will compare these in a sensitivity analysis. Subgroup analysis and investigation of heterogeneity If there is evidence of substantial heterogeneity (I2 greater than 40%) for the primary outcome measures and if there are enough trials, we will use subgroup analyses to explore it. We will explore the following subgroups: Sex of the patient Level of sexual behaviour risk of the study population (high risk, low risk) Uptake of the intervention (greater or less than 50%) Intensity of the intervention (single offer, multiple screening rounds) For fixed-effect models based on inverse variance meta-analysis, we will use tests of interaction to examine differences between groups. For random-effects models and fixed-effect models using methods other than inverse variance we will inspect confidence intervals for the subgroup estimates. Sensitivity analysis We will conduct sensitivity analyses to investigate the influence of methodological aspects of the review that might influence the results. We pre-specify the following sensitivity analyses: The treatment effect for pelvic inflammatory disease incidence in RCTs assessed as being at low versus high risk of detection bias, i.e. blinded versus non-blinded assessment. The treatment effect for each primary outcome in RCTs assessed at being at low versus high risk of selection bias.

Sensitivity analysis We will conduct sensitivity analyses to investigate the influence of methodological aspects of the review that might influence the results. We pre-specify the following sensitivity analyses: The treatment effect for pelvic inflammatory disease incidence in RCTs assessed as being at low versus high risk of detection bias, i.e. blinded versus non-blinded assessment. The treatment effect for each primary outcome in RCTs assessed at being at low versus high risk of selection bias. The treatment effect for chlamydia prevalence in RCTs versus non-randomised studies. The treatment effect for each primary outcome in intention-to-treat versus per protocol study populations. We will explore additional factors in sensitivity analysis if they arise during analysis. None Appendices Appendix 1. Electronic search strategies MEDLINE and CENTRAL (Ovid platform) 1 exp Mass Screening/ 2 screening$.tw. 3 tested.tw. 4 testing.tw. 5 or/1-4 6 exp Chlamydia/ 7 chlam?di$.tw. 8 exp Chlamydia trachomatis/ 9 exp Chlamydia Infections/ 10 or/6-9 11 randomized controlled trial.pt. 12 controlled clinical trial.pt. 13 randomized.ab. 14 placebo.ab. 15 clinical trials as topic.sh. 16 randomly.ab. 17 trial.ti. 18 or/11-17 19 exp animals/ not humans.sh. 20 18 not 19 21 5 and 10 and 20 Note: the CENTRAL search strategy does not include the terms #11 to #20. EMBASE.com #1 'mass screening'/exp #2 screening*:ti,ab #3 'screening'/exp #4 tested:ti,ab #5 testing:ti,ab #6 #1 OR #2 OR #3 OR #4 OR #5 #7 'Chlamydia'/exp #8 chlamydi*:ti,ab #9 chlamidi*:ti,ab #10 'Chlamydia trachomatis'/exp #11 'chlamydiasis'/exp #12 #7 OR #8 OR #9 OR #10 OR #11 #13 'randomized controlled trial'/exp

Note: the CENTRAL search strategy does not include the terms #11 to #20. EMBASE.com #1 'mass screening'/exp #2 screening*:ti,ab #3 'screening'/exp #4 tested:ti,ab #5 testing:ti,ab #6 #1 OR #2 OR #3 OR #4 OR #5 #7 'Chlamydia'/exp #8 chlamydi*:ti,ab #9 chlamidi*:ti,ab #10 'Chlamydia trachomatis'/exp #11 'chlamydiasis'/exp #12 #7 OR #8 OR #9 OR #10 OR #11 #13 'randomized controlled trial'/exp #14 'single blind procedure'/exp #15 'double blind procedure'/exp #16 'crossover procedure'/exp #17 #13 OR #14 OR #15 OR #16 #18 random*:ab,ti #19 placebo*:ab,ti #20 allocat*:ab,ti #21 crossover*:ab,ti #22 'cross over':ab,ti #23 trial:ti #24 (doubl* NEXT/1 blind*):ab,ti #25 #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 #26 #17 OR #25 #27 'animal'/de #28 'animal experiment'/de #29 'nonhuman'/de #30 #27 OR #28 OR #29 #31 'human'/de #32 #30 AND #31 #33 #30 NOT #32 #34 #26 NOT #33 #35 #6 AND #12 AND #34 AND [embase]/lim LILACS (iAHx interface) (mh:(tamizaje masivo)) OR (ti:(tamizaje)) OR (ab:(tamizaje)) OR (ti:(tamización)) OR (ab:(tamización)) OR (mh:(cribado)) OR (ti:(cribado)) OR (ab:(cribado)) AND (mh:(chlamydia)) OR (ti:(chlamydi*)) OR (ab:(chlamydi*)) OR (ti:(chlamidi*)) OR (ab:(chlamidi*)) OR (ti:(clamidia*)) OR (ab:(clamidia*)) OR (mh:(chlamydia trachomatis)) OR (mh:(infecciones por chlamydia)) AND db:("LILACS") AND type_of_study:("clinical_trials") Other resources “Screening”and “Chlamydia”in title, abstract and keywords. Contributions of authors Nicola Low drafted the protocol. Shelagh Redmond conducted preliminary electronic databases searches. Helen Ward, Anneli Uusküla, Jan van Bergen, Hannelore Götz commented on and suggested revisions to the protocol.

LILACS (iAHx interface) (mh:(tamizaje masivo)) OR (ti:(tamizaje)) OR (ab:(tamizaje)) OR (ti:(tamización)) OR (ab:(tamización)) OR (mh:(cribado)) OR (ti:(cribado)) OR (ab:(cribado)) AND (mh:(chlamydia)) OR (ti:(chlamydi*)) OR (ab:(chlamydi*)) OR (ti:(chlamidi*)) OR (ab:(chlamidi*)) OR (ti:(clamidia*)) OR (ab:(clamidia*)) OR (mh:(chlamydia trachomatis)) OR (mh:(infecciones por chlamydia)) AND db:("LILACS") AND type_of_study:("clinical_trials") Other resources “Screening”and “Chlamydia”in title, abstract and keywords. Contributions of authors Nicola Low drafted the protocol. Shelagh Redmond conducted preliminary electronic databases searches. Helen Ward, Anneli Uusküla, Jan van Bergen, Hannelore Götz commented on and suggested revisions to the protocol. Declarations of interest Berit Andersen is co-author of a completed trial that will be considered for inclusion in the review. Jan van Bergen, Hannelore Götz and Nicola Low were co-authors on a completed trial that will be considered for inclusion in the review (Chlamydia Screening Implementation project, The Netherlands). Nicola Low is a co-investigator on an ongoing trial that will be considered for inclusion in the review (Australian Chlamydia Control Effectiveness Pilot. ACCEPt). Authors of included studies will not be involved in assessing and extracting data of their own studies. Sources of support Internal sources No sources of support supplied External sources European Centre for Disease Prevention and Control, Sweden. Financial support for staff to conduct review as part of a project Chlamydia Control in Europe (ECDC/2011/027).

Background Tuberculosis (TB) is an infectious airborne disease caused by Mycobacterium tuberculosis bacteria and is the second most common cause of death from an infectious disease in adults (HIV/AIDS being first). TB predominantly affects the lungs (pulmonary TB) but can affect other parts of the body, such as the brain or the spine. Active TB disease is confirmed by the presence of viable TB bacilli. The symptoms of pulmonary TB include a persistent cough (for at least two weeks), fever, night sweats, weight loss, chills, haemoptysis and fatigue. In 2012, an estimated 8.6 million people developed TB and 1.3 million people died from TB (WHO 2013a). TB that is drug sensitive (also referred to as drug-susceptible TB) is the most common type of TB and may be effectively treated with a standardized regimen of first-line anti-TB drugs (WHO 2013a). However, TB bacilli may become drug resistant, meaning that first-line anti-TB drugs can no longer kill the bacilli. Drug resistance usually develops because of inappropriate or incorrect use of first-line drugs but new cases are increasingly caused by person-to-person transmission (Streicher 2011; Zhao 2012).

2013a). However, TB bacilli may become drug resistant, meaning that first-line anti-TB drugs can no longer kill the bacilli. Drug resistance usually develops because of inappropriate or incorrect use of first-line drugs but new cases are increasingly caused by person-to-person transmission (Streicher 2011; Zhao 2012). The emergence of drug-resistant TB (DR-TB) threatens to destabilise global TB control. In 2012, approximately 4% of new TB cases were multidrug resistant (WHO 2013a). Therapy for DR-TB requires treatment for more than 12 months and is toxic and expensive. In South Africa in 2011, the treatment of approximately 8000 cases of DR-TB, which comprised only 2.2% of the total TB burden, consumed 32% of the country's annual national TB budget of US$218 million (Pooran 2013). Fifty percent to 75% of patients experience unfavourable outcomes, such as death, treatment failure, or adverse drug reactions (Dheda 2010a; Dheda 2010b). There are two standardized definitions of DR-TB: multidrug-resistant TB (MDR-TB) and extensively drug-resistant TB (XDR-TB). MDR-TB is caused by M. tuberculosis which, when tested microbiologically in the laboratory, is resistant to rifampicin and isoniazid. These drugs are two of the most effective and widely-used anti-TB drugs that form part of the standardized first-line regimen for drug-susceptible TB. Patients with MDR-TB are commonly treated with drugs belonging to the fluoroquinolone (FQ) and second-line injectable anti-TB drug (SLID) classes. The FQ drugs include ofloxacin and moxifloxacin and the SLIDs include amikacin and kanamycin (two aminoglycoside drugs) and capreomycin (a cyclic peptide drug). XDR-TB is caused by M. tuberculosis resistant to isoniazid, rifampicin, plus any FQ and at least one of the three SLIDs (amikacin, kanamycin or capreomycin). Hence, patients with XDR-TB are resistant to both first-line and second-line drugs.

kanamycin (two aminoglycoside drugs) and capreomycin (a cyclic peptide drug). XDR-TB is caused by M. tuberculosis resistant to isoniazid, rifampicin, plus any FQ and at least one of the three SLIDs (amikacin, kanamycin or capreomycin). Hence, patients with XDR-TB are resistant to both first-line and second-line drugs. In South Africa, 80% of MDR-TB is thought to be spread via person-to-person transmission (Streicher 2011) and the same is likely true of MDR-TB and XDR-TB in China (Zhao 2012). Modelling studies (Basu 2007; Basu 2009; Dowdy 2008) have shown that, through the expansion of capacity to rapidly diagnose DR-TB, patient cure rates will be improved through the earlier initiation of appropriate and effective TB treatment. Importantly, once a patient is placed on effective treatment, their infectiousness dramatically declines within one to two weeks (Menzies 1997). However, the exact "infectiousness period" for DR-TB remains unclear. Early treatment initiation may therefore help curtail the spread of DR-TB through the disruption of person-to-person transmission. Thus, there is an urgent need for rapid tests that allow the early detection of drug resistance and the selection of appropriate TB drugs.

"infectiousness period" for DR-TB remains unclear. Early treatment initiation may therefore help curtail the spread of DR-TB through the disruption of person-to-person transmission. Thus, there is an urgent need for rapid tests that allow the early detection of drug resistance and the selection of appropriate TB drugs. Conventional tests for detecting TB drug resistance, referred to as drug susceptibility testing (DST), are traditionally 'phenotypic', in that bacteria in biological fluid from the patient (usually sputum) is inoculated into a culture medium containing the drug of interest and the presence (indicating resistance) or absence (indicating susceptibility) of M. tuberculosis growth is detected (Heysell 2012). Such testing is commonly performed indirectly, in that the pure bacterial culture or isolate grown from the original patient specimen is re-inoculated into drug-containing media. As the growth of M. tuberculosis typically takes between two to six weeks for the initial culture, there is often a significant time delay (two to six months) associated with the diagnosis of DR-TB, especially if re-inoculation is required. These delays are often further exacerbated by the technical and infrastructure requirements of testing, a lack of standardised methodologies for certain drugs (which cause unclear results that require repeating) (Richter 2009), as well as patient-associated difficulties, such as loss to follow-up. Recently, new tests for drug resistance such as the Genotype®MTBDRsl test (henceforth called MTBDRsl) that are rapid (potentially offering a turn-around time of one to two days) and 'genotypic' (as they detect the presence of specific mutations known to be associated with drug resistance) have offered considerable promise for the diagnosis of DR-TB.

sistance such as the Genotype®MTBDRsl test (henceforth called MTBDRsl) that are rapid (potentially offering a turn-around time of one to two days) and 'genotypic' (as they detect the presence of specific mutations known to be associated with drug resistance) have offered considerable promise for the diagnosis of DR-TB. One of the challenges in this Cochrane Review is the choice of the reference standard used to determine the presence or absence of the target conditions (described below). Phenotypic culture-based DST is the most widely used reference standard for drug resistance and is recommended by the WHO (WHO 2007). However, phenotypic culture-based DST is acknowledged to be imperfect and the results are dependent on the concentration of drug used. Genetic sequencing is widely considered to be the best reference standard for testing for the presence of drug resistance; but due to the technical aspects, costs and time associated with this method, it is rarely feasible to perform it on all samples suspected of DR-TB or in all regions of the TB genome that might be associated with resistance. Furthermore, not all genetic determinants or mechanisms of resistance may be known for a particular drug. We discuss the strengths and limitations of the different reference standards further below. Target condition being diagnosed We considered the following three target conditions: resistance of M. tuberculosis to FQs; resistance of M. tuberculosis to SLIDs; and XDR-TB.

One of the challenges in this Cochrane Review is the choice of the reference standard used to determine the presence or absence of the target conditions (described below). Phenotypic culture-based DST is the most widely used reference standard for drug resistance and is recommended by the WHO (WHO 2007). However, phenotypic culture-based DST is acknowledged to be imperfect and the results are dependent on the concentration of drug used. Genetic sequencing is widely considered to be the best reference standard for testing for the presence of drug resistance; but due to the technical aspects, costs and time associated with this method, it is rarely feasible to perform it on all samples suspected of DR-TB or in all regions of the TB genome that might be associated with resistance. Furthermore, not all genetic determinants or mechanisms of resistance may be known for a particular drug. We discuss the strengths and limitations of the different reference standards further below. Target condition being diagnosed We considered the following three target conditions: resistance of M. tuberculosis to FQs; resistance of M. tuberculosis to SLIDs; and XDR-TB. Index test(s) The GenoType® MTBDRsl assay (MTBDRsl, Hain Life Sciences) detects mutations in the gyrA gene (encoding the A-subunit of DNA gyrase), the rrs gene (encoding the 16S rRNA complex) and the embB gene (which, together with the genes embA and embC, codes for arabinosyltransferase) of the TB-causing M. tuberculosis complex species (which includes M. tuberculosis,M. africanum,M. bovis subsp. bovis,M. bovis subsp. caprae,M. bovis subsp. BCG,M. microti,M. canetti andM. pinnipedii) (Hain Life Sciences 2012a). The presence of mutations in these genes is associated with resistance to the FQs (including ofloxacin and levofloxacin), SLIDs (including kanamycin, amikacin and capreomycin) and ethambutol, respectively. Since ethambutol is a first-line TB drug, we did not determine the accuracy of MTBDRsl assay for ethambutol resistance in this review.

tions in these genes is associated with resistance to the FQs (including ofloxacin and levofloxacin), SLIDs (including kanamycin, amikacin and capreomycin) and ethambutol, respectively. Since ethambutol is a first-line TB drug, we did not determine the accuracy of MTBDRsl assay for ethambutol resistance in this review. The assay can be performed either on a patient specimen (direct testing) or on a culture grown from the patient specimen (indirect testing). The type of testing, direct or indirect, is dependent on the quantity of TB in the patient specimen. The manufacturer recommends that the assay is performed directly on the specimen if the specimen contains bacilli that can be seen using a light microscope and an acid-fast stain (smear-positive) (1).

indirect testing). The type of testing, direct or indirect, is dependent on the quantity of TB in the patient specimen. The manufacturer recommends that the assay is performed directly on the specimen if the specimen contains bacilli that can be seen using a light microscope and an acid-fast stain (smear-positive) (1). Figure 1 Clinical pathway diagram showing how molecular drug susceptibility testing (DST), which may use the MTBDRsl assay, is applied. A patient with suspected TB or suspected drug-resistant TB supplies a biological specimen (usually sputum), which is examined by smear microscopy and cultured. If acid-fast bacilli are observed under the microscope (smear-positive), a molecular DST can be performed directly on the specimen. If acid-fast bacilli are not observed (smear-negative), molecular DST can only be performed with acceptable accuracy on the culture isolate grown from the specimen. A molecular test for first-line drug resistance (for example, the MTBDRplus assay) is performed first and, only if resistance to the first-line drugs is indicated, the specimen is tested further for resistance to the second-line drugs using the MTBDRsl assay. Where molecular testing is not available, phenotypic testing for drug resistance may be performed on culture-positive isolates. Although phenotypic testing is being replaced by molecular-based methods in some settings, it is still usually performed in research studies seeking to measure the accuracy of the molecular test. Furthermore, some research studies also use gene sequencing as a reference standard or any specimens with discordant molecular DST-culture results.

ting is being replaced by molecular-based methods in some settings, it is still usually performed in research studies seeking to measure the accuracy of the molecular test. Furthermore, some research studies also use gene sequencing as a reference standard or any specimens with discordant molecular DST-culture results. The assay procedure is comprised of three sequential steps when using direct decontaminated patient material (decontaminated using the standard N-acetyl-cysteine and sodium hydroxide (NALC/NaOH) method), culture isolates in liquid media or when picking colonies from solid media. These steps are: (1) mycobacterial genomic DNA is extracted from the patient specimen or culture isolate; (2) regions within the gyrA, rrs and embB genes are selectively amplified using a multiplex polymerase chain reaction (PCR) assay; and (3) the amplification products are detected on a nitrocellulose membrane strip by reverse hybridisation and visualised using a streptavidin-conjugated alkaline phosphatase colour reaction. The observed bands, each corresponding to a specific probe, can be used to determine the drug susceptibility profile of the analysed specimen (an example is shown in 2). The extraction can also be done indirectly on blood cultures, where a Middlebrook slant is inoculated prior to picking the colonies from the agar after incubation for a period of time. Figure 2 Examples of different GenoType® MTBDRsl strip readouts (Hain Life Sciences 2012b).

The assay procedure is comprised of three sequential steps when using direct decontaminated patient material (decontaminated using the standard N-acetyl-cysteine and sodium hydroxide (NALC/NaOH) method), culture isolates in liquid media or when picking colonies from solid media. These steps are: (1) mycobacterial genomic DNA is extracted from the patient specimen or culture isolate; (2) regions within the gyrA, rrs and embB genes are selectively amplified using a multiplex polymerase chain reaction (PCR) assay; and (3) the amplification products are detected on a nitrocellulose membrane strip by reverse hybridisation and visualised using a streptavidin-conjugated alkaline phosphatase colour reaction. The observed bands, each corresponding to a specific probe, can be used to determine the drug susceptibility profile of the analysed specimen (an example is shown in 2). The extraction can also be done indirectly on blood cultures, where a Middlebrook slant is inoculated prior to picking the colonies from the agar after incubation for a period of time. Figure 2 Examples of different GenoType® MTBDRsl strip readouts (Hain Life Sciences 2012b). A template is supplied by the manufacturer to help read the strips Appendix 1 where the banding patterns are scored by eye, transcribed and manually fed into the Laboratory Information System (LIS). In high-volume settings, the GenoScan®, an automated reader, can be incorporated to interpret the banding patterns automatically and give a suggested interpretation (an example output of the machine is shown in Appendix 2. If the operator agrees with the interpretation, the results are automatically downloaded into the LIS, thus eliminating possible transcription errors. It is important to note that the automated reader only provides a suggested result and requires manual confirmation of the result after the operator has visually inspected the banding pattern. Nonetheless, the test manual provides fairly straightforward instruction with little room for variation in interpretation, even human interpretation. The entire assay procedure can be completed in five hours. The assay can also be performed on DNA from pure isolates taken from cultured patient specimens. Once a diagnosis of MDR-TB has been established, the MTBDRsl can also be used to confirm a diagnosis of XDR-TB.

iation in interpretation, even human interpretation. The entire assay procedure can be completed in five hours. The assay can also be performed on DNA from pure isolates taken from cultured patient specimens. Once a diagnosis of MDR-TB has been established, the MTBDRsl can also be used to confirm a diagnosis of XDR-TB. figure 2 shows an example of different MTBDRsl results. The assay consists of two internal controls (a conjugate control for confirmation of the colorimetric reaction used to visualise bands and an amplification control to ensure that nucleic acid amplification reaction has occurred) plus a control for each gene locus (gyrA, rrs, embB). The two internal controls plus the locus control for the gene of interest should always be positive; otherwise the assay cannot be evaluated for that particular drug. Of note is that a result can be indeterminate for one gene but valid for another (on the basis of only the gene-specific locus control failing). A band for the detection of the M. tuberculosis complex (the "TUB" band) is included. Should the wild-type or mutant probes appear whilst the locus control for a specific gene is less intense than that of the amplification control band (AC band) and the TUB band is interpretable, the locus probes should be considered secondary to that of the other probes for the gene in question and can thus be considered for interpretation.

r mutant probes appear whilst the locus control for a specific gene is less intense than that of the amplification control band (AC band) and the TUB band is interpretable, the locus probes should be considered secondary to that of the other probes for the gene in question and can thus be considered for interpretation. An earlier version of the MTBDRsl manual (version 1) stated that if the locus band was absent but other non-control bands were present (even together with their accompanying gene locus control bands) the assay should be considered non-evaluable (Hain Life Sciences 2012a). However, the most recent version of the manual (version 2; Hain Life Sciences 2012b) states: "in rare cases the TUB zone may be negative while an evaluable resistance pattern is developed. If so, the presence of a strain belonging to the MTB complex must be suspected and the assay should be repeated". Upon inspection, most of these are nontuberculous mycobacteria and thus if the TUB band is not present, it is suggested to use the GenoType® CM/AS kit for the identification of other common mycobacteria, or additional species should the GenoType® CM/AS kit fail to produce a positive identification for any of the 17 species covered by the GenoType® CM/AS kit (Hain Life Sciences 2012b).

thus if the TUB band is not present, it is suggested to use the GenoType® CM/AS kit for the identification of other common mycobacteria, or additional species should the GenoType® CM/AS kit fail to produce a positive identification for any of the 17 species covered by the GenoType® CM/AS kit (Hain Life Sciences 2012b). Clinical pathway figure 1 illustrates the clinical pathway. Depending on the setting, DST is either performed on all patients with confirmed TB or only on patients who are clinically suspected of having DR-TB (for example, if the patient's symptoms have failed to improved on first-line therapy, or if they still have viable bacilli in their sputum after an extended period of treatment. As mentioned above, the manufacturer recommends that if the patient specimen (usually sputum) is smear-positive the assay be performed directly on the specimen (direct testing). If smear-negative, it is recommended that the assay be performed on the culture isolate grown from the patient specimen (indirect testing). DST for resistance to the second-line drugs is only performed if resistance to the first-line drugs is confirmed. Where routine molecular (genotypic) testing is well established, phenotypic DST is not usually performed. However, we expected research studies evaluating the accuracy of molecular DSTs, such as the MTBDRsl assay, to almost always include phenotypic DST as a reference standard. Furthermore, we also expected some studies to use genetic sequencing to resolve any discordant index test-reference standard results.

ly performed. However, we expected research studies evaluating the accuracy of molecular DSTs, such as the MTBDRsl assay, to almost always include phenotypic DST as a reference standard. Furthermore, we also expected some studies to use genetic sequencing to resolve any discordant index test-reference standard results. Prior test(s) As detailed in 1, patients who received MTBDRsl testing will first have received (i) smear microscopy, (ii) liquid culture (if smear-negative), and (iii) phenotypic or genotypic DST for resistance to first-line drugs. Role of index test(s) MTBDRsl would be used as an initial test replacing phenotypic culture-based DST as the initial test.

Prior test(s) As detailed in 1, patients who received MTBDRsl testing will first have received (i) smear microscopy, (ii) liquid culture (if smear-negative), and (iii) phenotypic or genotypic DST for resistance to first-line drugs. Role of index test(s) MTBDRsl would be used as an initial test replacing phenotypic culture-based DST as the initial test. Rationale Second-line TB drugs are used to treat patients with TB that is resistant to the most effective and widely used first-line drugs. To ensure that the most appropriate and least toxic drugs are provided to patients as quickly as possible, it is critical to know whether a patient has resistance to FQs alone, resistance to SLIDs alone, or resistance to both FQs and SLIDs (XDR-TB) as this will guide the selection of drugs. In addition, the presence of XDR-TB has major prognostic implications for the patient and for infection control. The conventional method for the diagnosis of drug resistance (phenotypic culture-based testing) is vulnerable to contamination and the culture can lose viability, meaning it cannot be tested. This method is also slow and can take several months. The resulting diagnostic delay results in unnecessary morbidity, mortality and increased transmission, which is a major driver of new TB cases. There is a need for rapid assays to improve time-to-diagnosis and new molecular assays, such as the MTBDRsl assay, present a promising potential solution.

take several months. The resulting diagnostic delay results in unnecessary morbidity, mortality and increased transmission, which is a major driver of new TB cases. There is a need for rapid assays to improve time-to-diagnosis and new molecular assays, such as the MTBDRsl assay, present a promising potential solution. Objectives To assess and compare the diagnostic accuracy of MTBDRsl for the detection of resistance to FQs in patient specimens (using direct testing) and culture isolates (using indirect testing) confirmed as TB positive. To assess and compare the diagnostic accuracy of MTBDRsl for the detection of resistance to SLIDs in patient specimens (using direct testing) and culture isolates (using indirect testing) confirmed as TB positive. To assess and compare the diagnostic accuracy of MTBDRsl for the detection of XDR-TB in patient specimens (using direct testing) and culture isolates (using indirect testing) confirmed as TB positive.

To assess and compare the diagnostic accuracy of MTBDRsl for the detection of resistance to SLIDs in patient specimens (using direct testing) and culture isolates (using indirect testing) confirmed as TB positive. To assess and compare the diagnostic accuracy of MTBDRsl for the detection of XDR-TB in patient specimens (using direct testing) and culture isolates (using indirect testing) confirmed as TB positive. Secondary objectives We planned to investigate heterogeneity in relation to the reference standard (culture-based DST compared with (1) genetic sequencing, (2) culture-based DST and genetic sequencing, and (3) culture-based DST followed by genetic sequencing with discordant results) and individual drugs within a drug class (for example, ofloxacin and moxifloxacin within the FQ class). We also pre-specified in the protocol investigations of heterogeneity in relation to HIV status, condition of the specimens (fresh or frozen, volume of specimen), patient population (patients suspected of having MDR-TB or XDR-TB) and whether WHO-recommended critical drug concentrations were used for culture-based reference testing.

pecified in the protocol investigations of heterogeneity in relation to HIV status, condition of the specimens (fresh or frozen, volume of specimen), patient population (patients suspected of having MDR-TB or XDR-TB) and whether WHO-recommended critical drug concentrations were used for culture-based reference testing. Methods Criteria for considering studies for this review Types of studies We included all studies that determined the diagnostic accuracy of the index test in comparison with a defined reference standard, including case-control designs, in which cases and controls were sampled from the same patient population. We only included studies from which data could be extracted for true positives (TP), true negatives (TN), false positives (FP) and false negatives (FN). We excluded unpublished studies reported only in abstracts. Participants We included patients and specimens from patients of any age who were thought to have resistance to any of the second-line TB drugs, as well as patients and patient specimens with confirmed MDR-TB from all settings, irrespective of background burden and patient population. Index tests We included studies that evaluated the MTBDRsl assay. Target conditions We considered three target conditions: Resistance to any of the FQs. The FQs include ofloxacin, levofloxacin and moxifloxacin. We excluded ciprofloxacin because this drug is infrequently used in DST. Resistance to any of the SLIDs. The SLIDs include two aminoglycosides, kanamycin and amikacin, and one cyclic peptide, capreomycin. XDR-TB.

Target conditions We considered three target conditions: Resistance to any of the FQs. The FQs include ofloxacin, levofloxacin and moxifloxacin. We excluded ciprofloxacin because this drug is infrequently used in DST. Resistance to any of the SLIDs. The SLIDs include two aminoglycosides, kanamycin and amikacin, and one cyclic peptide, capreomycin. XDR-TB. For the FQs, the presence of mutations in each of the genes probed by the MTBDRsl assay has very high concordance with resistance to all drugs within that drug class. For example, a mutation in the gyrA usually means a strain is resistant to each of the FQs: ofloxacin, levofloxacin and moxifloxacin (Sirgel 2012a). The same holds true for the rrs gene and the two aminoglycosides, kanamycin and amikacin (Sirgel 2012b). Evidence is mixed regarding the level of concordance between resistance to the two aminoglycosides and capreomycin arising from mutations in the rrs gene. We acknowledge that determining resistance to all three SLIDs together, and thus including capreomycin with the aminoglycosides, may be a limitation. However, the index test results are reported in this manner. We discuss this issue further in the Discussion. Reference standards The following reference standards were used to define the target conditions: Phenotypic culture-based DST: solid culture or a commercial liquid culture system (BACTEC 460, MGIT 960 and MGIT Manual System, Becton Dickinson, USA) incorporating the drug of interest. Genetic sequencing of the gyrA or rrs genes, or both.

Reference standards The following reference standards were used to define the target conditions: Phenotypic culture-based DST: solid culture or a commercial liquid culture system (BACTEC 460, MGIT 960 and MGIT Manual System, Becton Dickinson, USA) incorporating the drug of interest. Genetic sequencing of the gyrA or rrs genes, or both. Two reference standards used together: phenotypic culture-based DST and genetic sequencing of the same samples. If a specimen was resistant according to phenotypic culture-based DST or had a mutation in the gyrA or rrs genes, the specimen was classified as having the target condition. If both phenotypic culture-based DST and genetic sequencing indicated susceptibility, the specimen was classified as not having the target condition. Two reference standards used sequentially: phenotypic culture-based DST followed by selective testing by genetic sequencing of samples with discordant results (also referred to as discrepant analysis). Discordant results may be either index test positive/phenotypic culture-based DST negative or index test negative/phenotypic culture-based DST positive.

ntially: phenotypic culture-based DST followed by selective testing by genetic sequencing of samples with discordant results (also referred to as discrepant analysis). Discordant results may be either index test positive/phenotypic culture-based DST negative or index test negative/phenotypic culture-based DST positive. There are strengths and limitations to each of the reference standards. As mentioned, phenotypic culture-based DST is the conventional reference standard, but it is considered to be imperfect and is dependent on the drug concentration threshold used to define resistance. Genetic sequencing is considered to be more accurate than phenotypic culture-based DST; however, this is only if it targets all known resistance determining regions, which are not completely defined for the FQs and the SLIDs. Therefore, genetic sequencing can miss mutations that may cause drug resistance which fall outside of the targeted genes. Furthermore, genetic sequencing is usually applied only to culture isolates when results for the index test and the culture-based reference test do not agree. In this latter situation, there is potential for verification bias because the same reference standard is not being used to verify all index test results. We carried out separate analyses for the different reference standards, described below. In our primary analysis we used culture-based DST as the reference standard. We expected all or nearly all included studies to report results using this reference standard.

There are strengths and limitations to each of the reference standards. As mentioned, phenotypic culture-based DST is the conventional reference standard, but it is considered to be imperfect and is dependent on the drug concentration threshold used to define resistance. Genetic sequencing is considered to be more accurate than phenotypic culture-based DST; however, this is only if it targets all known resistance determining regions, which are not completely defined for the FQs and the SLIDs. Therefore, genetic sequencing can miss mutations that may cause drug resistance which fall outside of the targeted genes. Furthermore, genetic sequencing is usually applied only to culture isolates when results for the index test and the culture-based reference test do not agree. In this latter situation, there is potential for verification bias because the same reference standard is not being used to verify all index test results. We carried out separate analyses for the different reference standards, described below. In our primary analysis we used culture-based DST as the reference standard. We expected all or nearly all included studies to report results using this reference standard. Search methods for identification of studies We attempted to identify all relevant studies regardless of language and publication status (published, unpublished, in press and ongoing). We searched for unpublished data as a means of ensuring the sensitivity of the search for published literature. Unpublished data in this field may provide misleading results as the data set is incomplete. While unpublished sources were searched, we did not include unpublished data in the review. We did not apply date restrictions to the searches.

data as a means of ensuring the sensitivity of the search for published literature. Unpublished data in this field may provide misleading results as the data set is incomplete. While unpublished sources were searched, we did not include unpublished data in the review. We did not apply date restrictions to the searches. Electronic searches Vittoria Lutje (VL), the Information Specialist for the Cochrane Infectious Diseases Group, performed literature searching up to 30 January 2014. To identify all relevant studies, she searched the following databases using the search terms and strategy described in Appendix 3: Cochrane Infectious Diseases Group Specialized Register; MEDLINE (Pubmed, 1966 to January 2014); EMBASE OVID (1980 to January 2014); ISI Web of Knowledge (Science Citation Index - Expanded (1900 to present), Conference Proceedings Citation Index- Science (CPCI-S) (1990 to present) and BIOSIS Previews (1926 to January 2014)); MEDION (http://www.mediondatabase.nl/); LILACS (http://lilacs.bvsalud.org/en/; 1982 to January 2014); and SCOPUS (1995 to January 2014). VL also searched the metaRegister of Controlled Trials (mRCT; http://www.controlled-trials.com/) and the search portal of the World Health Organization (WHO) International Clinical Trials Registry Platform (www.who.int/trialsearch), to identify ongoing trials, and ProQuest Dissertations & Theses A&I to identify relevant dissertations.

the metaRegister of Controlled Trials (mRCT; http://www.controlled-trials.com/) and the search portal of the World Health Organization (WHO) International Clinical Trials Registry Platform (www.who.int/trialsearch), to identify ongoing trials, and ProQuest Dissertations & Theses A&I to identify relevant dissertations. Searching other resources We reviewed reference lists of included articles and any relevant review articles identified through the above methods. We contacted the assay manufacturer (Hain Life Sciences) to identify unpublished studies. We contacted researchers at the Foundation for Innovative New Diagnostics (FIND), members of the StopTB Partnership's New Diagnostics Working Group and other experts in the field of TB diagnostics for information on ongoing or unpublished studies. Data collection and analysis Selection of studies Two review authors (GT and JP) independently scrutinized titles and abstracts identified by electronic literature searching to identify potentially eligible studies. We selected all citations identified as suitable during this screen for full-text review. The same two review authors then independently reviewed full-text papers for study eligibility using the predefined inclusion and exclusion criteria. For full text articles, we resolved any discrepancies by discussion with a third review author (KRS). We maintained a list of excluded studies and their reasons for exclusion.

ame two review authors then independently reviewed full-text papers for study eligibility using the predefined inclusion and exclusion criteria. For full text articles, we resolved any discrepancies by discussion with a third review author (KRS). We maintained a list of excluded studies and their reasons for exclusion. Data extraction and management Two review authors (GT and JP) independently extracted a set of data from each study using a piloted data extraction form. We resolved any discrepancies by discussion. Based on the pilot, we finalized the data extraction form. We then independently extracted data on the following characteristics: Details of study: first author; publication year; country where testing was performed; setting (primary care laboratory, hospital laboratory, reference laboratory); study design; manner of participant selection; number of participants enrolled; number of participants for whom results available; industry sponsorship. Characteristics of participants: age (mean, SD; median, interquartile range; age range); HIV status; smear status; history of TB; known MDR-TB, pre-XDR-TB or XDR-TB status. Target conditions: resistance to FQs; resistance to SLIDs; XDR-TB. Reference standards: name and manufacturer; type; percentage of patients whose reference standard was 'uninterpretable' (for example, contaminated, sequencing failed). Details of specimen: type (such as expectorated sputum, induced sputum or culture isolate); condition (fresh or frozen); definition of a positive smear; type of testing (direct testing or indirect testing).

Reference standards: name and manufacturer; type; percentage of patients whose reference standard was 'uninterpretable' (for example, contaminated, sequencing failed). Details of specimen: type (such as expectorated sputum, induced sputum or culture isolate); condition (fresh or frozen); definition of a positive smear; type of testing (direct testing or indirect testing). Details of outcomes: the number of TP, TN, FP and FN results; number of indeterminate assay results. Time to treatment initiation: defined as the time from specimen collection until patient starts treatment. Time to diagnosis: defined as the time from specimen collection until there is an available TB result in lab or clinic, if the assay was performed in a clinic. We assigned country income status (high-income or low- and middle-income) as classified by the World Bank List of Economies (World Bank 2014). We contacted authors of primary studies for missing data or clarifications. We entered all data into a database manager (Microsoft Excel 2012). For one study that tested the same panel of TB isolates in multiple centres, we selected one centre that provided results in the middle range (neither the best nor the worst results). Whenever possible, we extracted data that used a single patient as the unit of analysis (one MTBDRsl result per one specimen from one patient).

For one study that tested the same panel of TB isolates in multiple centres, we selected one centre that provided results in the middle range (neither the best nor the worst results). Whenever possible, we extracted data that used a single patient as the unit of analysis (one MTBDRsl result per one specimen from one patient). When culture-based DST was performed using more than one drug from the FQs (ofloxacin, moxifloxacin or levofloxacin) or SLIDs (amikacin, kanamycin or capreomycin), we extracted data (TP, TN, FP, FN) for each drug and for each class overall. We also extracted data for the SLIDs as a class overall if culture-based DST was performed using only one drug. No studies reported on the number of 'no TB' or 'no result' results obtained from MTBDRsl, therefore we only reported the proportion of 'indeterminate' results. In the 2 x 2 tables of TP, FP, FN and TN, we based the results of the index test on categorical assay results defined by the visual readout of the MTBDRsl strip. Possible results for the Genotype® MTBDRsl assay (as defined by the product manual) Sensitive to either FQs or SLIDs (referred to as 'aminoglycosides/cyclic peptides'), or both (conjugation and amplification bands present; TUB band present; gene locus band present; all wild type (wt) bands for each gene present; no mutation bands present). In the case of susceptibility to both drug classes, the test would indicate susceptibility for each, rather than having a single composite readout specifying XDR-TB.

on and amplification bands present; TUB band present; gene locus band present; all wild type (wt) bands for each gene present; no mutation bands present). In the case of susceptibility to both drug classes, the test would indicate susceptibility for each, rather than having a single composite readout specifying XDR-TB. Resistant to either FQs or SLIDs, or both (conjugation and amplification bands present; TUB band present; gene locus band present; all, none or some wt bands for each gene present; all, none or some mutation bands present with similar intensity to amplification control). In the case of resistance to both drug classes, the test would indicate resistance for each, rather than having a composite readout. Indeterminate (faint bands) or no result (no conjugation or amplification bands present, no locus band present for the gene of interest). No TB (negative for MTB complex irrespective of locus control band). No result (failure of any one of the control bands, as well as the TUB band).

Resistant to either FQs or SLIDs, or both (conjugation and amplification bands present; TUB band present; gene locus band present; all, none or some wt bands for each gene present; all, none or some mutation bands present with similar intensity to amplification control). In the case of resistance to both drug classes, the test would indicate resistance for each, rather than having a composite readout. Indeterminate (faint bands) or no result (no conjugation or amplification bands present, no locus band present for the gene of interest). No TB (negative for MTB complex irrespective of locus control band). No result (failure of any one of the control bands, as well as the TUB band). Assignment of results to the fluoroquinolones, second-line injectable drugs or both categories MTBDRsl detects the presence of mutations in genes that cause drug resistance for drug classes (ie FQs, SLIDs or both), not to individual drugs within these classes (ofloxacin, moxifloxacin and levofloxacin in the case of the FQs; amikacin, kanamycin and capreomycin in the case of SLIDs). Thus, one study might use phenotypic DST for detection of kanamycin resistance and another study might use phenotypic DST for detection of amikacin resistance as reference standards to confirm SLID resistance. In such a scenario, if the phenotypic DST was positive for resistance and the MTBDRsl result was concordant, we classified the index case result as true-positive. We adopted the same approach for the FQs. Similarly, if the index tests reported resistance to a SLID and, in the case of genetic sequencing being used as a reference standard, the presence of mutations known to be associated with drug resistance to the SLIDs was confirmed, we recorded this as a concordant result positive for resistance to SLIDs. A similar approach was used for the FQs that used genetic sequencing as a reference standard.

genetic sequencing being used as a reference standard, the presence of mutations known to be associated with drug resistance to the SLIDs was confirmed, we recorded this as a concordant result positive for resistance to SLIDs. A similar approach was used for the FQs that used genetic sequencing as a reference standard. Assessment of methodological quality We appraised the quality of the included studies with the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool (Whiting 2011; Appendix 4). QUADAS-2 consists of four domains: patient selection, index test, reference standard, and flow and timing. We assessed all domains for the potential for risk of bias and the first three domains for concerns regarding applicability. We used signalling questions in each domain to form judgments about the risk of bias. One review author (GT) piloted the tool with two included studies and finalized the tool based on experience gained from the pilot testing. Two review authors then independently assessed methodological quality of included studies with the finalized tool. Statistical analysis and data synthesis We performed descriptive analysis for key variables (such as country income status and number of study participants) of the primary studies using Stata version 12.0 and displayed key study characteristics in Characteristics of included studies.

Assessment of methodological quality We appraised the quality of the included studies with the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool (Whiting 2011; Appendix 4). QUADAS-2 consists of four domains: patient selection, index test, reference standard, and flow and timing. We assessed all domains for the potential for risk of bias and the first three domains for concerns regarding applicability. We used signalling questions in each domain to form judgments about the risk of bias. One review author (GT) piloted the tool with two included studies and finalized the tool based on experience gained from the pilot testing. Two review authors then independently assessed methodological quality of included studies with the finalized tool. Statistical analysis and data synthesis We performed descriptive analysis for key variables (such as country income status and number of study participants) of the primary studies using Stata version 12.0 and displayed key study characteristics in Characteristics of included studies. We used the reference standard 'culture' in our primary analyses. We stratified these analyses first by target condition (FQ resistance, SLID resistance or XDR-TB) and second by type of MTBDRsl testing (indirect testing or direct testing). Within each stratum (for example, FQ resistance by indirect testing), we plotted estimates of the studies' observed sensitivities and specificities in forest plots with 95% confidence intervals (CI) and in receiver-operating characteristic (ROC) space using Review Manager (RevMan). Where adequate data were available, we combined data using meta-analysis. We performed the majority of meta-analyses by fitting the bivariate random-effects model (Macaskill 2010; Reitsma 2005) using Stata version 11 with the metandi and xtmelogit commands. We compared models with separate and identical variance terms using likelihood ratio tests to determine the best fitting model. In situations in which there were fewer than four studies, we determined summary estimates of sensitivity and specificity by simplifying the bivariate model to two univariate random-effects logistic regression models. When it was not possible to fit the model and we observed little heterogeneity, we determined summary estimates of sensitivity and specificity separately using a fixed-effect model (Zamora 2006). We presented meta-analysis summaries in tables and ROC space.

l to two univariate random-effects logistic regression models. When it was not possible to fit the model and we observed little heterogeneity, we determined summary estimates of sensitivity and specificity separately using a fixed-effect model (Zamora 2006). We presented meta-analysis summaries in tables and ROC space. We compared results from studies of direct testing with results from studies of indirect testing by adding a covariate for the type of testing to the model. We assessed the significance of the difference in test accuracy between studies using direct testing and studies using indirect testing by a likelihood ratio test comparing models with and without covariate terms. For these comparative analyses, we first included all studies with relevant data and then included only those studies that made direct comparisons between direct and indirect testing with the same participants, where such studies existed. We present the results according to the stated objectives, under the appropriate subheadings in the Results section for each condition: Estimates of the diagnostic accuracy of MTBDRsl using phenotypic culture-based DST as a reference standard, and Investigations of heterogeneity for each testing method. Approach to uninterpretable (indeterminate) MTBDRsl results We excluded indeterminate test results from the analyses for determination of sensitivity and specificity. We determined the proportion of indeterminate MTBDRsl results among the primary studies for each target condition and provided results separately for indirect and direct testing.

le (indeterminate) MTBDRsl results We excluded indeterminate test results from the analyses for determination of sensitivity and specificity. We determined the proportion of indeterminate MTBDRsl results among the primary studies for each target condition and provided results separately for indirect and direct testing. Investigations of heterogeneity Within each stratum (for example SLID resistance), we investigated heterogeneity through visual examination of forest plots of sensitivity and specificity. Then, if sufficient studies were available, we explored the possible influence of the following pre-specified categorical covariates: reference standard (culture, genetic sequencing, culture and genetic sequencing, culture followed by genetic sequencing) and individual drug (amikacin, kanamycin and capreomycin). We determined variation in sensitivity and specificity by adding covariate terms to the meta-analysis models described above. The significance of the difference in test accuracy (for example, between studies using culture versus those using genetic sequencing as the reference standard) was assessed by a likelihood ratio test comparing models with and without covariate terms.

adding covariate terms to the meta-analysis models described above. The significance of the difference in test accuracy (for example, between studies using culture versus those using genetic sequencing as the reference standard) was assessed by a likelihood ratio test comparing models with and without covariate terms. We had also planned to investigate the effect of HIV status, the condition of the specimen (fresh or frozen), sample volume, the drug concentration used for culture-based DST (WHO-recommended or not) and patient population (patients thought to have MDR-TB or XDR-TB) on summary estimates of sensitivity and specificity in a meta-regression analysis by adding covariate terms to the bivariate model. However, there were insufficient data for these additional analyses. Sensitivity analyses For our primary analysis using the culture-based DST reference standard, we performed sensitivity analyses for four QUADAS-2 signalling questions to explore whether the results we found were robust with respect to the methodological quality of the studies. We used the following questions: Was a consecutive or random sample of patients/specimens enrolled? Was a case-control design avoided? Were the reference standard results interpreted without knowledge of the results of the index test? Were the index test results interpreted without knowledge of the results of the reference standard? We did not exclude any studies based on these answers.

Was a consecutive or random sample of patients/specimens enrolled? Was a case-control design avoided? Were the reference standard results interpreted without knowledge of the results of the index test? Were the index test results interpreted without knowledge of the results of the reference standard? We did not exclude any studies based on these answers. Assessment of reporting bias We did not undertake a formal assessment of publication bias of data included in this review using methods such as funnel plots or regression tests because such techniques have not been found to be helpful for determining publication bias within diagnostic test accuracy studies (Macaskill 2010; Tatsioni 2005). Other analyses We had intended to summarize two patient outcomes, time-to-diagnosis and time-to-treatment initiation; however time-to-diagnosis was the only one described in the included studies.

Assessment of reporting bias We did not undertake a formal assessment of publication bias of data included in this review using methods such as funnel plots or regression tests because such techniques have not been found to be helpful for determining publication bias within diagnostic test accuracy studies (Macaskill 2010; Tatsioni 2005). Other analyses We had intended to summarize two patient outcomes, time-to-diagnosis and time-to-treatment initiation; however time-to-diagnosis was the only one described in the included studies. Results Results of the search Our search identified 630 titles (Table 5; Table 6; 3). We did not add any additional titles after reference review or contact with experts. After we removed duplicates, 262 titles remained of which we excluded 140 titles based on a review of title, or abstract, or both. We retrieved full text articles for 41 citations, of which we excluded 20, leaving 21 unique studies included in the review and meta-analysis (3). We have listed the reasons for exclusion of studies in the Characteristics of excluded studies section. One of the 21 studies (Ignatyeva 2012) evaluated a panel of isolates at four different sites in Eastern Europe and we extracted data for the one site that neither performed the best or the worst. Figure 3 Study flow diagram.

Results Results of the search Our search identified 630 titles (Table 5; Table 6; 3). We did not add any additional titles after reference review or contact with experts. After we removed duplicates, 262 titles remained of which we excluded 140 titles based on a review of title, or abstract, or both. We retrieved full text articles for 41 citations, of which we excluded 20, leaving 21 unique studies included in the review and meta-analysis (3). We have listed the reasons for exclusion of studies in the Characteristics of excluded studies section. One of the 21 studies (Ignatyeva 2012) evaluated a panel of isolates at four different sites in Eastern Europe and we extracted data for the one site that neither performed the best or the worst. Figure 3 Study flow diagram. Methodological quality of included studies 4 and 5 show the quality assessment of the 21 included studies. In the patient selection domain, we considered 10 studies (48%) to be at low risk of bias because participants were enrolled consecutively or randomly and the study design was cross-sectional. We considered nine studies to be at high risk of bias because (1) there was a case-control design (five studies: Brossier 2010; Hillemann 2009; Ignatyeva 2012; Kiet 2010; Miotto 2012); (2) enrolment was by convenience (three studies: Barnard 2012; Lacoma 2012; Lopez-Roa 2012); or (3) the study had both a case-control design and convenience sampling (one study: van Ingen 2010). We considered two studies to have unclear risk of bias because it was unclear how patients were selected (Chikamatsu 2012; Fan 2011). With regard to applicability (patient characteristics and setting), we judged 15 studies (71%) to include the appropriate patients and settings to address the review question and six studies to have a high concern about applicability (Brossier 2010; Hillemann 2009; Ignatyeva 2012; Kiet 2010; Miotto 2012; van Ingen 2010). In the index test domain, we considered two studies at high risk of bias as the index test results were not interpreted without knowledge of the results of the reference standard (Chikamatsu 2012; Kiet 2010) and seven studies at unknown risk of bias because information about blinding was unavailable (Brossier 2010; Fan 2011; Ferro 2013; Hillemann 2009; Lopez-Roa 2012; Surcouf 2011; Tukvadze 2014). In all but two studies (Brossier 2010; Tukvadze 2014), the use, conduct and interpretation of the index test was considered applicable. In the reference standard domain, we judged eleven studies (52%) to be at low risk of bias because the reference standard was appropriate and the results were interpreted without knowledge of the results of the MTBDRsl assay (Ajbani 2012; Barnard 2012; Chikamatsu 2012; Hillemann 2009; Huang 2011; Ignatyeva 2012;Jin 2013; Lopez-Roa 2012 Miotto 2012; Tukvadze 2014; Zivanovic 2012. We judged applicability to be of low concern for all studies in the reference standard domain.

re interpreted without knowledge of the results of the MTBDRsl assay (Ajbani 2012; Barnard 2012; Chikamatsu 2012; Hillemann 2009; Huang 2011; Ignatyeva 2012;Jin 2013; Lopez-Roa 2012 Miotto 2012; Tukvadze 2014; Zivanovic 2012. We judged applicability to be of low concern for all studies in the reference standard domain. In the flow and timing domain, we considered 16 studies (76%) to be of low concern for risk of bias because all patients were accounted for in the analysis, information about uninterpretable results was provided and all patients had the same reference tests performed. We considered five studies to have unclear risk of bias in the flow and timing domain because discrepant analysis was performed (Ajbani 2012; Barnard 2012; Kiet 2010; Lacoma 2012; Lopez-Roa 2012) meaning that not all patients received culture-based and sequencing reference testing. Also, we considered one study (Ferro 2013) to have unclear risk of bias because not all patients were accounted for in the analyses. We noted industry involvement in seven (33%) studies and this included: i) donation of MTBDRsl tests (four studies: Hillemann 2009; Miotto 2012; Surcouf 2011; Ferro 2013); ii) preferred pricing of MTBDRsl tests (one study: Barnard 2012); iii) financial support for non-test related study costs (one study: Said 2012); and iv) involvement in the design, analysis or manuscript production (one study: Ajbani 2012). Figure 4 Risk of bias and applicability concerns graph: review authors' judgements about each domain presented as percentages across included studies.

In the flow and timing domain, we considered 16 studies (76%) to be of low concern for risk of bias because all patients were accounted for in the analysis, information about uninterpretable results was provided and all patients had the same reference tests performed. We considered five studies to have unclear risk of bias in the flow and timing domain because discrepant analysis was performed (Ajbani 2012; Barnard 2012; Kiet 2010; Lacoma 2012; Lopez-Roa 2012) meaning that not all patients received culture-based and sequencing reference testing. Also, we considered one study (Ferro 2013) to have unclear risk of bias because not all patients were accounted for in the analyses. We noted industry involvement in seven (33%) studies and this included: i) donation of MTBDRsl tests (four studies: Hillemann 2009; Miotto 2012; Surcouf 2011; Ferro 2013); ii) preferred pricing of MTBDRsl tests (one study: Barnard 2012); iii) financial support for non-test related study costs (one study: Said 2012); and iv) involvement in the design, analysis or manuscript production (one study: Ajbani 2012). Figure 4 Risk of bias and applicability concerns graph: review authors' judgements about each domain presented as percentages across included studies. Figure 5 Risk of bias and applicability concerns summary: review authors' judgements about each domain for each included study.

In the flow and timing domain, we considered 16 studies (76%) to be of low concern for risk of bias because all patients were accounted for in the analysis, information about uninterpretable results was provided and all patients had the same reference tests performed. We considered five studies to have unclear risk of bias in the flow and timing domain because discrepant analysis was performed (Ajbani 2012; Barnard 2012; Kiet 2010; Lacoma 2012; Lopez-Roa 2012) meaning that not all patients received culture-based and sequencing reference testing. Also, we considered one study (Ferro 2013) to have unclear risk of bias because not all patients were accounted for in the analyses. We noted industry involvement in seven (33%) studies and this included: i) donation of MTBDRsl tests (four studies: Hillemann 2009; Miotto 2012; Surcouf 2011; Ferro 2013); ii) preferred pricing of MTBDRsl tests (one study: Barnard 2012); iii) financial support for non-test related study costs (one study: Said 2012); and iv) involvement in the design, analysis or manuscript production (one study: Ajbani 2012). Figure 4 Risk of bias and applicability concerns graph: review authors' judgements about each domain presented as percentages across included studies. Figure 5 Risk of bias and applicability concerns summary: review authors' judgements about each domain for each included study. Findings Of the 21 included studies, eight reported on MTBDRsl testing for resistance to FQs, SLIDs and XDR-TB, 12 reported on testing for resistance to FQs and SLIDs, and one reported on testing for resistance to FQs only. Fourteen studies reported on MTBDRsl performance when done directly on patient specimens, five studies reported on MTBDRsl when performed indirectly on isolates grown from the specimens and a further two studies contained information on both testing methods. Of the 21 studies, 11 used only phenotypic culture-based DST, seven used sequencing and culture on all specimens, three used culture followed by the sequencing of discrepant results and one used sequencing alone. The median (interquartile range (IQR)) number of participants in each study was 100 (50.75, 229.5). The proportion of patients screened with resistance to a FQ, SLID, or XDR-TB (according to phenotypic culture-based testing) were 30% (95% CI 28 to 32), 32% (95% CI 30 to 34), or 15% (95% CI 13 to 17), respectively.

one. The median (interquartile range (IQR)) number of participants in each study was 100 (50.75, 229.5). The proportion of patients screened with resistance to a FQ, SLID, or XDR-TB (according to phenotypic culture-based testing) were 30% (95% CI 28 to 32), 32% (95% CI 30 to 34), or 15% (95% CI 13 to 17), respectively. We presented key characteristics for the 21 studies in the Characteristics of included studies section. The majority (15 studies, 71%) were of cross-sectional study design. One study (Barnard 2012) included extrapulmonary specimens that we excluded from the analysis. Eleven studies (58%) were located in low-income or middle-income countries. All studies but two (Fan 2011, written in Chinese, and Chikamatsu 2012, written in Japanese) were in English. I. Fluoroquinolone resistance detection A. Estimates of the diagnostic accuracy of MTBDRsl using phenotypic culture-based DST as a reference standard 1. Indirect testing We present forest plots of MTBDRsl sensitivity and specificity when performed indirectly for the detection of FQ resistance for 16 studies (1766 participants) that used phenotypic culture-based DST as a reference standard in 6. For individual studies, sensitivity estimates ranged from 57% to 100% and specificity estimates ranged from 77% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 83.1% (95% CI 78.7 to 86.7) and 97.7% (95% CI 94.3 to 99.1), respectively.

phenotypic culture-based DST as a reference standard in 6. For individual studies, sensitivity estimates ranged from 57% to 100% and specificity estimates ranged from 77% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 83.1% (95% CI 78.7 to 86.7) and 97.7% (95% CI 94.3 to 99.1), respectively. Figure 6 Forest plots of MTBDRsl sensitivity and specificity when performed indirectly or directly for FQ resistance detection and using phenotypic culture-based DST as a reference standard. The individual studies are ordered by decreasing sensitivity. TP = true positive; FP = false positive; FN = false negative; TN = true negative. Values between brackets are the 95% CIs of sensitivity and specificity. The figure shows the estimated sensitivity and specificity of the study (blue square) and its 95% CI (black horizontal line). 2. Direct testing In 6 we show forest plots of MTBDRsl sensitivity and specificity when performed directly for the detection of resistance to FQs for seven studies (1033 participants) that used phenotypic culture-based DST as a reference standard. For individual studies, sensitivity estimates ranged from 50% to 100% and specificity estimates ranged from 91% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 85.1% (95% CI 71.9 to 92.7) and 98.2% (95% CI 96.8 to 99.0), respectively.

used phenotypic culture-based DST as a reference standard. For individual studies, sensitivity estimates ranged from 50% to 100% and specificity estimates ranged from 91% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 85.1% (95% CI 71.9 to 92.7) and 98.2% (95% CI 96.8 to 99.0), respectively. 3. Comparison of indirect versus direct testing (i) Diagnostic accuracy We present results comparing indirect and direct MTBDRsl testing for detection of FQ resistance in Table 3, Table 4 and 7. There was no evidence of a statistically significant difference in MTBDRsl accuracy between indirect and direct testing and using culture-based DST as a reference standard when the test was performed in different populations (indirect comparison, P = 0.549). Direct comparisons within the same population were not possible because no studies performed direct and indirect MTBDRsl testing on specimens or isolates from the same patients. Figure 7 Summary plots of MTBDRsl sensitivity and specificity comparing detection of fluoroquinolone resistance by indirect and direct testing. The solid circles correspond to the summary estimates of sensitivity and specificity and are shown with 95% confidence regions (dotted lines) and 95% prediction regions (dashed lines).

Summary plots of MTBDRsl sensitivity and specificity comparing detection of fluoroquinolone resistance by indirect and direct testing. The solid circles correspond to the summary estimates of sensitivity and specificity and are shown with 95% confidence regions (dotted lines) and 95% prediction regions (dashed lines). (ii) Indeterminate rates For indirect testing for FQ resistance, four (0.2%) of 1766 MTBDRsl results were indeterminate (three culture DST resistant and one culture DST sensitive), whereas for direct testing 20 (1.9%) of 1033 were MTBDRsl indeterminate (P < 0.001; 14 were culture DST-sensitive and six did not report a culture-based DST result). B. Investigations of heterogeneity 1. Indirect testing (i) Type of reference standard We present MTBDRsl accuracy estimates for detection of FQ resistance against different reference standards in Table 7 and Appendix 5. Reference standard is genetic sequencing: For individual studies (seven in total), sensitivity estimates ranged from 85% to 100% and specificity estimates ranged from 92% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 99.3% (95% CI 85.9 to 100.0) and 99.7% (95% CI 92.0 to 100.0), respectively. The accuracy using this reference standard was higher than when culture-based DST was used (P < 0.001 for indirect statistical comparisons, Table 3; P < 0.001 for direct statistical comparisons, Table 4). Five studies sequenced the gyrA gene and two sequenced gyrA and gyrB. Reference standard is culture-based DST and genetic sequencing (ie both investigations performed in all isolates):

For individual studies (seven in total), sensitivity estimates ranged from 85% to 100% and specificity estimates ranged from 92% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 99.3% (95% CI 85.9 to 100.0) and 99.7% (95% CI 92.0 to 100.0), respectively. The accuracy using this reference standard was higher than when culture-based DST was used (P < 0.001 for indirect statistical comparisons, Table 3; P < 0.001 for direct statistical comparisons, Table 4). Five studies sequenced the gyrA gene and two sequenced gyrA and gyrB. Reference standard is culture-based DST and genetic sequencing (ie both investigations performed in all isolates): For individual studies (seven in total), sensitivity estimates ranged from 74% to 91% and specificity estimates ranged from 99% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 82.0% (95% CI 77.7 to 85.6) and 99.8% (95% CI 98.5 to 100.0), respectively. The accuracy using this reference standard was higher than when culture-based DST was used (P < 0.001 for indirect comparisons, Table 3; P < 0.001 for direct comparisons, Table 4). Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results:

For individual studies (seven in total), sensitivity estimates ranged from 74% to 91% and specificity estimates ranged from 99% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 82.0% (95% CI 77.7 to 85.6) and 99.8% (95% CI 98.5 to 100.0), respectively. The accuracy using this reference standard was higher than when culture-based DST was used (P < 0.001 for indirect comparisons, Table 3; P < 0.001 for direct comparisons, Table 4). Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results: For individual studies (three in total), sensitivity estimates ranged from 73% to 100% and specificity estimates ranged from 94% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 83.7% (95% CI 74.2 to 90.8) and 99.7% (95% CI 98.4 to 100.0), respectively. Comparisons between accuracy estimates using this reference standard and culture-based DST were not possible given the small number of studies in the former group.

m 94% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 83.7% (95% CI 74.2 to 90.8) and 99.7% (95% CI 98.4 to 100.0), respectively. Comparisons between accuracy estimates using this reference standard and culture-based DST were not possible given the small number of studies in the former group. (ii) Drugs used in the culture-based DST We present MTBDRsl accuracy estimates for detection of resistance to ofloxacin and moxifloxacin against a phenotypic culture-based reference standard in Table 3, Table 4 and Appendix 6. For ofloxacin resistance, sensitivity estimates ranged from 70% to 100% and specificity estimates ranged from 91% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 82.9% (95% CI 79.5 to 87.1) and 98.2% (95% CI 96.1 to 99.1), respectively. For moxifloxacin resistance, sensitivity estimates ranged from 57% to 100% and specificity estimates from 77% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 91.4% (95% CI 64.7 to 98.4) and 90.6% (95% CI 79.3 to 96.1), respectively. The accuracy of MTBDRsl when performed indirectly was not different for ofloxacin versus moxifloxacin (indirect comparison, P = 0.091). Appendix 7 presents a summary ROC plot of sensitivity versus specificity comparing test performance for detection of resistance to the individual FQ drugs.

79.3 to 96.1), respectively. The accuracy of MTBDRsl when performed indirectly was not different for ofloxacin versus moxifloxacin (indirect comparison, P = 0.091). Appendix 7 presents a summary ROC plot of sensitivity versus specificity comparing test performance for detection of resistance to the individual FQ drugs. (iii) Drug concentration used in culture-based DST Nine studies used the WHO-recommended critical concentration of ofloxacin, whereas two did not (Jin 2013; Kiet 2010). Ferro 2013 used the WHO-recommended critical concentration for low level moxifloxacin resistance whereas Lacoma 2012 used the concentration recommended for high level resistance. Two studies (Fan 2011; van Ingen 2010) did not used the recommended critical concentration of moxifloxacin. Comparisons between accuracy estimates for each drug according to concentration were not possible given the small number of studies. 2. Direct testing (i) Type of reference standard Reference standard is genetic sequencing: No studies performed direct MTBDRsl testing and used genetic sequencing as a reference standard. Reference standard is culture-based DST and genetic sequencing (ie both investigations performed in all isolates): No studies performed direct MTBDRsl testing and used both phenotypic culture-based DST and genetic sequencing (performed in all isolates) as a reference standard. Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results:

Reference standard is culture-based DST and genetic sequencing (ie both investigations performed in all isolates): No studies performed direct MTBDRsl testing and used both phenotypic culture-based DST and genetic sequencing (performed in all isolates) as a reference standard. Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results: Two studies reported MTBDRsl sensitivity and specificity when performed directly for the detection of resistance to FQs, with phenotypic culture-based DST and genetic testing performed only on discrepant results as a reference standard. The reported sensitivities were 91% and 96% and the reported specificities were 98% and 99%. (ii) Drugs used in the culture-based DST Sensitivity estimates for MTBDRsl for ofloxacin resistance by direct testing against a phenotypic culture-based reference standard for three studies ranged from 89% to 100%. Specificity estimates from 98% to 100%. No studies performed MTBDRsl by direct testing for moxifloxacin resistance. (iii) Drug concentration used in culture-based DST All three studies in this category used the WHO-recommended critical concentration for ofloxacin.

(ii) Drugs used in the culture-based DST Sensitivity estimates for MTBDRsl for ofloxacin resistance by direct testing against a phenotypic culture-based reference standard for three studies ranged from 89% to 100%. Specificity estimates from 98% to 100%. No studies performed MTBDRsl by direct testing for moxifloxacin resistance. (iii) Drug concentration used in culture-based DST All three studies in this category used the WHO-recommended critical concentration for ofloxacin. II. SLID resistance detection A. Estimates of the diagnostic accuracy of MTBDRsl using phenotypic culture-based DST as a reference standard 1. Indirect testing We present forest plots of MTBDRsl sensitivity and specificity when performed indirectly for the detection of resistance to SLIDs for 14 studies (1637 participants) that used phenotypic culture-based DST as a reference standard in 8. For individual studies, sensitivity estimates ranged from 25% to 100% and specificity estimates ranged from 86% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 76.9% (95% CI 61.1 to 87.6) and 99.5% (95% CI 97.1 to 99.9), respectively.

phenotypic culture-based DST as a reference standard in 8. For individual studies, sensitivity estimates ranged from 25% to 100% and specificity estimates ranged from 86% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 76.9% (95% CI 61.1 to 87.6) and 99.5% (95% CI 97.1 to 99.9), respectively. Figure 8 Forest plots of MTBDRsl sensitivity and specificity for SLID resistance detection when performed indirectly or directly and using phenotypic culture-based DST as a reference standard. The individual studies are ordered by decreasing sensitivity. TP = true positive; FP = false positive; FN = false negative; TN = true negative. Values between brackets are the 95% CIs of sensitivity and specificity. The figure shows the estimated sensitivity and specificity of the study (blue square) and its 95% CI (black horizontal line)

es are ordered by decreasing sensitivity. TP = true positive; FP = false positive; FN = false negative; TN = true negative. Values between brackets are the 95% CIs of sensitivity and specificity. The figure shows the estimated sensitivity and specificity of the study (blue square) and its 95% CI (black horizontal line) 2. Direct testing In 8 we show forest plots of MTBDRsl sensitivity and specificity when performed directly for the detection of resistance to SLIDs for six studies (947 participants) that used phenotypic culture-based DST as a reference standard. For individual studies, sensitivity estimates ranged from 9% to 100%, with one study from Eastern Europe reporting low sensitivity (Kontsevaya 2013). Specificity estimates ranged from 67% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 94.4% (95% CI 25.2 to 99.9) and 98.2% (95% CI 88.9 to 99.7), respectively. When the study from Eastern Europe that reported low sensitivity (Kontsevaya 2013) was removed (Appendix 8), the pooled sensitivity increased to 98.0% (95% CI 39.6 to 100.0), while the pooled specificity decreased to 97.8% (95% CI 86.4 to 99.7).

94.4% (95% CI 25.2 to 99.9) and 98.2% (95% CI 88.9 to 99.7), respectively. When the study from Eastern Europe that reported low sensitivity (Kontsevaya 2013) was removed (Appendix 8), the pooled sensitivity increased to 98.0% (95% CI 39.6 to 100.0), while the pooled specificity decreased to 97.8% (95% CI 86.4 to 99.7). 3. Comparison of indirect versus direct testing (i) Diagnostic accuracy We present results comparing indirect and direct MTBDRsl testing for SLID resistance in Table 3, Table 4 and 9. The pooled sensitivity for direct testing (94.4%, 95% CI 25.2 to 99.9) was similar to the pooled estimate for indirect testing (76.9%, 95% CI 61.1 to 87.6) when the test was performed in different populations using all studies (indirect comparisons, P = 0.451). The pooled specificity was lower (indirect comparisons, P = 0.005) for direct testing (98.2%, 95% CI 88.9 to 99.7) when compared to indirect testing (99.5%, 95% CI 97.1 to 99.9). Figure 9 Summary plots of MTBDRsl sensitivity and specificity comparing detection of resistance for second-line injectable drugs by indirect and direct testing. The solid circles correspond to the summary estimates of sensitivity and specificity and are shown with 95% confidence regions (dotted lines) and 95% prediction regions (dashed lines).

MTBDRsl sensitivity and specificity comparing detection of resistance for second-line injectable drugs by indirect and direct testing. The solid circles correspond to the summary estimates of sensitivity and specificity and are shown with 95% confidence regions (dotted lines) and 95% prediction regions (dashed lines). (ii) Indeterminate rates For indirect testing for SLID resistance, three (0.4%) of 774 MTBDRsl results were indeterminate (one culture DST resistant and two culture DST sensitive; three studies did not report these), whereas for direct testing 53 (6.1%) of 873 were MTBDRsl indeterminate (four were culture DST resistant, 22 were culture DST susceptible and 27 did not have a culture-based DST result; one study did not report indeterminate results) (P < 0.001). B. Investigations of heterogeneity 1. Indirect testing (i) Type of reference standard We present MTBDRsl accuracy estimates for detection of SLID resistance against different reference standards in Table 7 and Appendix 9. Reference standard is genetic sequencing: For individual studies (six in total), sensitivity estimates ranged from 62% to 100% and specificity estimates ranged from 96% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 97.8% (95% CI 77.0 to 99.7) and 99.5% (95% CI 94.5 to 100.0), respectively. The accuracy using this reference standard was higher than when culture-based DST was used (P = 0.017 for indirect statistical comparisons, Table 3; P = 0.045 for direct statistical comparisons, Table 4). All six studies sequenced only the rrs gene.

(95% CI 77.0 to 99.7) and 99.5% (95% CI 94.5 to 100.0), respectively. The accuracy using this reference standard was higher than when culture-based DST was used (P = 0.017 for indirect statistical comparisons, Table 3; P = 0.045 for direct statistical comparisons, Table 4). All six studies sequenced only the rrs gene. Reference standard is culture-based DST and genetic sequencing (ie both investigations performed in all isolates): For individual studies (seven in total), sensitivity estimates ranged from 30% to 85% and specificity estimates ranged from 99% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 56.7% (95% CI 40.8 to 71.3) and 99.9% (95% CI 99.2 to 100.0), respectively. The accuracy using this reference standard was higher than when culture-based DST was used (P = 0.008 for indirect comparisons, Table 3). Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results: For individual studies (three in total), sensitivity estimates ranged from 34% to 100% and specificity estimates ranged from 95% to 100%. We did not determine summary estimates because there were only three studies and the sensitivity was variable.

Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results: For individual studies (three in total), sensitivity estimates ranged from 34% to 100% and specificity estimates ranged from 95% to 100%. We did not determine summary estimates because there were only three studies and the sensitivity was variable. (ii) Drugs used in the culture-based DST We present MTBDRsl accuracy estimates for detection of resistance to amikacin, kanamycin and capreomycin by indirect testing against a phenotypic culture-based reference standard in Table 7 and 10. For amikacin resistance, sensitivity estimates ranged from 80% to 100% and specificity estimates ranged from 97% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 87.9% (95% CI 82.1 to 92.0) and 99.5% (95% CI 97.5 to 99.9), respectively. For kanamycin resistance, sensitivity estimates ranged from 25% to 100% and specificity estimates from 86% to 100%. The pooled sensitivity and specificity were 66.9% (95% CI 44.1 to 83.8) and 98.6% (95% CI 96.1 to 99.5). For capreomycin resistance, sensitivity estimates ranged from 21% to 100% and specificity estimates from 86% to 100%. The pooled sensitivity and specificity were 79.5% (95% CI 58.3 to 91.4) and 95.8% (95% CI 93.4 to 97.3). 11 presents a summary ROC plot of sensitivity versus specificity comparing test performance for detection of resistance to the individual SLIDs by indirect testing.

o 100% and specificity estimates from 86% to 100%. The pooled sensitivity and specificity were 79.5% (95% CI 58.3 to 91.4) and 95.8% (95% CI 93.4 to 97.3). 11 presents a summary ROC plot of sensitivity versus specificity comparing test performance for detection of resistance to the individual SLIDs by indirect testing. Figure 10 Forest plots of MTBDRsl sensitivity and specificity when performed indirectly for the detection of resistance to amikacin (Ak), kanamycin (Kn) and capreomycin (Cm) using culture as a reference standard. The individual studies are ordered by decreasing sensitivity. TP = true positive; FP = false positive; FN = false negative; TN = true negative. Values between brackets are the 95% CIs of sensitivity and specificity. The figure shows the estimated sensitivity and specificity of the study (blue square) and its 95% CI (black horizontal line). Figure 11 Summary plots of MTBDRsl sensitivity and specificity comparing indirect detection of resistance for amikacin (Ak), kanamycin (Kn) and capreomycin (Cm) using culture as a reference standard. The solid circles correspond to the summary estimates of sensitivity and specificity and are shown with 95% confidence regions (dotted lines) and 95% prediction regions (dashed lines).

omparing indirect detection of resistance for amikacin (Ak), kanamycin (Kn) and capreomycin (Cm) using culture as a reference standard. The solid circles correspond to the summary estimates of sensitivity and specificity and are shown with 95% confidence regions (dotted lines) and 95% prediction regions (dashed lines). (iii) Drug concentration used in culture-based DST Five studies used the WHO-recommended critical concentration of amikacin, whereas four did not. Two studies (Huang 2011; Ferro 2013) used the WHO-recommended critical concentration of kanamycin, whereas seven did not. Seven studies used the WHO-recommended critical concentration of capreomycin, two did not (Brossier 2010; Huang 2011) and one (Jin 2013) did not report the critical concentration used. Comparisons between accuracy estimates according to drug concentration were not possible given the small number of studies and participants. 2. Direct testing (i) Type of reference standard Reference standard is genetic sequencing: No studies performed direct MTBDRsl testing and used genetic sequencing as a reference standard. Reference standard is culture-based DST and genetic sequencing (ie both investigations performed in all isolates): No studies performed direct MTBDRsl testing and used both phenotypic culture-based DST and genetic sequencing (performed in all isolates) as a reference standard. Reference standard is culture-based DST followed by genetic sequencing of discrepant index test culture-based DST results:

Reference standard is culture-based DST and genetic sequencing (ie both investigations performed in all isolates): No studies performed direct MTBDRsl testing and used both phenotypic culture-based DST and genetic sequencing (performed in all isolates) as a reference standard. Reference standard is culture-based DST followed by genetic sequencing of discrepant index test culture-based DST results: We found two studies, both of which reported perfect sensitivity and specificity: 100% (95% CI 85 to 100) and 100% (95% CI 97 to 100) for Ajbani 2012 and 100% (95% CI 92 to 100) and 100% (95% CI 98 to 100) for Barnard 2012, respectively.

Reference standard is culture-based DST followed by genetic sequencing of discrepant index test culture-based DST results: We found two studies, both of which reported perfect sensitivity and specificity: 100% (95% CI 85 to 100) and 100% (95% CI 97 to 100) for Ajbani 2012 and 100% (95% CI 92 to 100) and 100% (95% CI 98 to 100) for Barnard 2012, respectively. (ii) Drugs used in the culture-based DST We present MTBDRsl accuracy estimates for detection of resistance to amikacin, kanamycin and capreomycin by direct testing against a phenotypic culture-based reference standard in 12. For amikacin resistance, sensitivity estimates ranged from 64% to 100% and specificity estimates ranged from 89% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 97.3% (95% CI 55.1 to 99.9) and 99.3% (95% CI 92.3 to 99.9), respectively. For kanamycin resistance, sensitivity estimates ranged from 9% to 100% and specificity estimates from 91% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 72.1% (95% CI 9.5 to 98.5) and 98.8% (95% CI 89.3 to 99.9), respectively. For capreomycin resistance, sensitivity estimates ranged from 57% to 100% and specificity estimates from 90% to 100%. The pooled sensitivity and specificity were 68.7% (95% CI 55.4 to 79.5) and 97.0% (95% CI 89.6 to 99.2). 13 presents a summary ROC plot of sensitivity versus specificity comparing test performance for detection of resistance to the individual SLIDs by direct testing.

to 100% and specificity estimates from 90% to 100%. The pooled sensitivity and specificity were 68.7% (95% CI 55.4 to 79.5) and 97.0% (95% CI 89.6 to 99.2). 13 presents a summary ROC plot of sensitivity versus specificity comparing test performance for detection of resistance to the individual SLIDs by direct testing. Figure 12 Forest plots of MTBDRsl sensitivity and specificity when performed directly for the detection of resistance to amikacin (Ak), kanamycin (Kn) and capreomycin (Cm) using culture as a reference standard. The individual studies are ordered by decreasing sensitivity. TP = true positive; FP = false positive; FN = false negative; TN = true negative. Between brackets are the 95% CI of sensitivity and specificity. The figure shows the estimated sensitivity and specificity of the study (blue square) and its 95% CI (black horizontal line). Figure 13 Summary plots of MTBDRsl sensitivity and specificity comparing direct detection of resistance for amikacin (Ak), kanamycin (Kn) and capreomycin (Cm) using culture as a reference standard. The solid circles correspond to the summary estimates of sensitivity and specificity and are shown with 95% confidence regions (dotted lines) and 95% prediction regions (dashed lines).

comparing direct detection of resistance for amikacin (Ak), kanamycin (Kn) and capreomycin (Cm) using culture as a reference standard. The solid circles correspond to the summary estimates of sensitivity and specificity and are shown with 95% confidence regions (dotted lines) and 95% prediction regions (dashed lines). (iii) Drug concentration used in culture-based DST Three studies in this category used the WHO-recommended critical concentration for amikacin. One study (Barnard 2012) used a method (culture on Middlebrook 7H11 media) for which the WHO does not recommend a critical concentration. Two studies used the WHO-recommended critical concentration for kanamycin (Ajbani 2012; Tukvadze 2014) and two (Kontsevaya 2013; Miotto 2012) did not. All four studies used the WHO-recommended critical concentration for capreomycin. III. XDR-TB detection A. Estimates of the diagnostic accuracy of MTBDRsl using phenotypic culture-based DST as a reference standard 1. Indirect testing In figure 14 we present forest plots of MTBDRsl sensitivity and specificity for XDR-TB for eight studies (880 participants) that used phenotypic culture-based DST as a reference standard. For individual studies, sensitivity estimates ranged from 20% to 100% and specificity estimates ranged from 96% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 70.9% (95% CI 42.9 to 88.8) and 98.8% (95% CI 96.1 to 99.6), respectively.

used phenotypic culture-based DST as a reference standard. For individual studies, sensitivity estimates ranged from 20% to 100% and specificity estimates ranged from 96% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 70.9% (95% CI 42.9 to 88.8) and 98.8% (95% CI 96.1 to 99.6), respectively. Figure 14 Forest plots of MTBDRsl sensitivity and specificity when performed indirectly and directly for the detection of XDR-TB using phenotypic culture-based DST as a reference standard. The individual studies are ordered by decreasing sensitivity. TP = true positive; FP = false positive; FN = false negative; TN = true negative. Between brackets are the 95% CI of sensitivity and specificity. The figure shows the estimated sensitivity and specificity of the study (blue square) and its 95% CI (black horizontal line). 2. Direct testing We show forest plots of MTBDRsl sensitivity and specificity for XDR-TB for three studies (664 participants) that used phenotypic culture-based DST as a reference standard in 14. We observed considerable heterogeneity and did not calculate pooled estimates. The test yielded sensitivities and specificities of 92% (95% CI 75 to 99) and 100% (95% CI 99 to 100) for Barnard 2012, 14% (95% CI 5 to 35) and 100% (95% CI 93 to 100) for Kontsevaya 2013 and 100% (95% CI 16 to 100) and 94% (95% CI 84 to 99) for Miotto 2012.

erable heterogeneity and did not calculate pooled estimates. The test yielded sensitivities and specificities of 92% (95% CI 75 to 99) and 100% (95% CI 99 to 100) for Barnard 2012, 14% (95% CI 5 to 35) and 100% (95% CI 93 to 100) for Kontsevaya 2013 and 100% (95% CI 16 to 100) and 94% (95% CI 84 to 99) for Miotto 2012. 3. Comparison of indirect versus direct testing (i) Diagnostic accuracy We present results for indirect MTBDRsl testing for XDR-TB in Table 5. The pooled sensitivity was 70.9% (95% CI 42.9 to 88.8) and the pooled specificity was 98.8% (95% CI 96.1 to 99.6). We were unable to compare these estimates with those for direct testing, because we did not calculate pooled estimates for direct testing as there were only three studies with considerable heterogeneity. (ii) Indeterminate rates For indirect testing for XDR-TB, one (0.1%) of 880 MTBDRsl results was indeterminate (one culture DST sensitive), whereas for direct testing 12 (1.8%) of 644 were MTBDRsl indeterminate (five were culture DST susceptible and seven did not report a culture-based DST result) (P < 0.001). B. Investigations of heterogeneity 1. Indirect testing We present MTBDRsl accuracy estimates for detection of XDR-TB against different reference standards in Table 5. (i) Type of reference standard Reference standard is genetic sequencing: For individual studies (three in total), sensitivity estimates were all 100% and specificity estimates ranged from 95% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 100% (95% CI 94.6 to 100) and 97.5% (95% CI 95.6 to 98.7), respectively.

(i) Type of reference standard Reference standard is genetic sequencing: For individual studies (three in total), sensitivity estimates were all 100% and specificity estimates ranged from 95% to 100%. In the meta-analysis, the pooled sensitivity and specificity were 100% (95% CI 94.6 to 100) and 97.5% (95% CI 95.6 to 98.7), respectively. Reference standard is culture-based DST and genetic sequencing (ie both investigations performed in all isolates): We found two studies. Sensitivity and specificity estimates for Jin 2013 were 56% (95% CI 45 to 67) and 99% (95% CI 96 to 100), respectively, and 71% (95% CI 44 to 90) and 99% (95% CI 95 to 100), respectively for Miotto 2012. The pooled sensitivity and specificity were 58.8% (95% CI 49.1 to 67.9) and 98.8% (95% CI 96.8 to 99.5), respectively. Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results: No studies performed indirect MTBDRsl testing when performed indirectly for XDR-TB and used phenotypic culture-based DST and genetic sequencing for discordant analysis as a reference standard.

We found two studies. Sensitivity and specificity estimates for Jin 2013 were 56% (95% CI 45 to 67) and 99% (95% CI 96 to 100), respectively, and 71% (95% CI 44 to 90) and 99% (95% CI 95 to 100), respectively for Miotto 2012. The pooled sensitivity and specificity were 58.8% (95% CI 49.1 to 67.9) and 98.8% (95% CI 96.8 to 99.5), respectively. Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results: No studies performed indirect MTBDRsl testing when performed indirectly for XDR-TB and used phenotypic culture-based DST and genetic sequencing for discordant analysis as a reference standard. (ii) Drugs used in the culture-based DST One (Kiet 2010) of the eight studies that performed indirect testing for XDR-TB and used cultured-based DST as a reference standard used ofloxacin and kanamycin. Two studies (Hillemann 2009; Zivanovic 2012) used ofloxacin, amikacin and capreomycin. One study (Miotto 2012) used ofloxacin, amikacin and kanamycin. One study (Chikamatsu 2012) used levofloxacin, amikacin, kanamycin and capreomycin. One study (Ignatyeva 2012) used ofloxacin, amikacin, kanamycin and capreomycin. One study (Jin 2013) used ofloxacin, kanamycin and capreomycin. One study (van Ingen 2010) used moxifloxacin, amikacin and ofloxacin. As all but two studies used a different combination of drugs, we did not compare test performance according to drugs used in the culture-based DST.

, amikacin, kanamycin and capreomycin. One study (Jin 2013) used ofloxacin, kanamycin and capreomycin. One study (van Ingen 2010) used moxifloxacin, amikacin and ofloxacin. As all but two studies used a different combination of drugs, we did not compare test performance according to drugs used in the culture-based DST. (iii) Drug concentration used in culture-based DST Four studies in this category used the WHO-recommended critical concentration for ofloxacin (Hillemann 2009; Ignatyeva 2012; Miotto 2012; Zivanovic 2012) and two did not (Jin 2013; Kiet 2010). van Ingen 2010 used moxifloxacin but did not use the WHO-recommended critical concentration. For the study that used levofloxacin (Chikamatsu 2012) the WHO does not recommend a critical concentration for the type of culture used (Ogawa culture). For the six studies that used amikacin, four used the WHO-recommended critical concentration (Hillemann 2009; Ignatyeva 2012; Miotto 2012; Zivanovic 2012), one did not report the concentration used (Chikamatsu 2012) and one used a type of culture-based testing (Middlebrook 7H10 media) for which the WHO did not specify a recommended critical concentration (van Ingen 2010). Of the five studies that used kanamycin, three did not use the WHO-recommended critical concentration (Jin 2013; Kiet 2010; Miotto 2012), one did not report the concentration used (Chikamatsu 2012) and one used a type of culture-based testing (MGIT 960) for which the WHO did not specify a recommended critical concentration (Ignatyeva 2012). Of the six studies that used capreomycin, five used the WHO-recommended critical concentration (Hillemann 2009; Ignatyeva 2012; Miotto 2012; van Ingen 2010; Zivanovic 2012) and two did not report the concentration used (Chikamatsu 2012; Jin 2013).

O did not specify a recommended critical concentration (Ignatyeva 2012). Of the six studies that used capreomycin, five used the WHO-recommended critical concentration (Hillemann 2009; Ignatyeva 2012; Miotto 2012; van Ingen 2010; Zivanovic 2012) and two did not report the concentration used (Chikamatsu 2012; Jin 2013). 2. Direct testing (i) Type of reference standard Reference standard is genetic sequencing: No studies performed direct MTBDRsl testing for XDR-TB and used genetic sequencing as a reference standard. Reference standard is culture-based DST and genetic sequencing (ie both investigations performed in all isolates): No studies performed direct MTBDRsl testing and used both phenotypic culture-based DST and genetic sequencing (performed in all isolates) as a reference standard. Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results: We found a single study (Miotto 2012) that used phenotypic culture-based DST and performed genetic testing only on discrepant results. This study reported a sensitivity of 92% (95% CI 75 to 99) and a specificity of 100% (95% CI 99 to 100).

Reference standard is culture-based DST followed by genetic sequencing of discrepant index test-culture-based DST results: We found a single study (Miotto 2012) that used phenotypic culture-based DST and performed genetic testing only on discrepant results. This study reported a sensitivity of 92% (95% CI 75 to 99) and a specificity of 100% (95% CI 99 to 100). Sensitivity analyses We undertook sensitivity analyses by limiting inclusion in the meta-analyses to: studies with consecutive or random selection of samples, studies with cross-sectional design, studies where index test results were blinded to reference standard results, and studies where reference standard results were blinded to index test results. Table 6 contains sensitivity analyses for the FQs. For the SLIDs (Table 7), using culture-based DST as the reference standard and direct testing, the pooled sensitivity estimate was lower when we dropped studies that enrolled patients by convenience. However, in all the analyses for the detection SLID resistance by direct testing, we found wide 95% CIs suggesting less precision around the pooled estimates. The other sensitivity analyses made no difference to any of the findings.

ivity estimate was lower when we dropped studies that enrolled patients by convenience. However, in all the analyses for the detection SLID resistance by direct testing, we found wide 95% CIs suggesting less precision around the pooled estimates. The other sensitivity analyses made no difference to any of the findings. Other analyses Only four studies described the effect of MTBDRsl on time-to-diagnosis. Lopez-Roa 2012 reported it to have a time-to-diagnosis of eight hours, compared to DST using the agar proportion method (21 days) or the MGIT 960 method (eight days). Said 2012 stated that MTBDRsl had a median time-to-diagnosis of two days, compared to 11 days for the agar proportion method. Tukvadze 2014 noted a median time-to-diagnosis using MTBDRsl of 10 days, versus 70 to 104 days for culture-based DST. Barnard 2012 reported it to have a median turn-around-time of one day (after the diagnosis of first-line resistance), whereas the median turn-around-time for phenotypic culture-based DST was 31 days.

d. Tukvadze 2014 noted a median time-to-diagnosis using MTBDRsl of 10 days, versus 70 to 104 days for culture-based DST. Barnard 2012 reported it to have a median turn-around-time of one day (after the diagnosis of first-line resistance), whereas the median turn-around-time for phenotypic culture-based DST was 31 days. Discussion This systematic review found that, when used indirectly on culture isolates, MTBDRsl had higher pooled sensitivity for detection of FQ resistance (83.1%) than for detection of SLID resistance (76.9%). When used directly on smear-positive sputum specimens, MTBDRsl had lower pooled sensitivity for FQ resistance (85.1%) than for SLID resistance (94.4%); however the pooled sensitivity for detection of SLID resistance was imprecise (95% CI 25.2 to 99.9). When SLID resistance was analysed for individual drugs, the pooled sensitivity was highest for amikacin (87.9%). For detection of resistance to both FQs and SLIDs, pooled specificity was high (= 97%). For detection of XDR-TB by indirect testing, the pooled sensitivity of MTBDRsl was 70.9% and the pooled specificity was 98.9%. The average sensitivities and specificities of MTBDRsl for detection of resistance to FQs and SLIDs and XDR-TB included in the meta-analyses are given in the 'Summary of Findings' tables (Table 1; Table 2; Table 3) and Table 3 Table 4 and Table 5.

d sensitivity of MTBDRsl was 70.9% and the pooled specificity was 98.9%. The average sensitivities and specificities of MTBDRsl for detection of resistance to FQs and SLIDs and XDR-TB included in the meta-analyses are given in the 'Summary of Findings' tables (Table 1; Table 2; Table 3) and Table 3 Table 4 and Table 5. When MTBDRsl accuracy was compared according to whether the test was performed directly or indirectly, the sensitivities were similar for FQ resistance and SLID resistance. Indirect MTBDRsl testing for SLID resistance had slightly superior specificity compared to direct testing (99.5% versus 98.2%). Indirect testing for both FQ and SLID resistance had a lower rate of uninterpretable MTBDRsl results than direct testing.

the sensitivities were similar for FQ resistance and SLID resistance. Indirect MTBDRsl testing for SLID resistance had slightly superior specificity compared to direct testing (99.5% versus 98.2%). Indirect testing for both FQ and SLID resistance had a lower rate of uninterpretable MTBDRsl results than direct testing. We performed subgroup analyses in which we compared the accuracy of MTBDRsl against different reference standards comprised of phenotypic culture-based DST (the traditional gold standard) or genetic sequencing. We looked at MTBDRsl accuracy against each type of reference standard alone or in combination (where either all specimens received both culture-based DST and sequencing). When used indirectly on culture isolates for detection of FQ resistance, MTBDRsl had higher pooled sensitivity against genetic sequencing than against culture-based DST (99.3% versus 83.1%). This suggests that MTBDRsl is sensitive for detecting FQ resistance caused by mutations in gyrA (the only gene that is targeted by MTBDRsl for detection of FQ resistance). However, against culture-based DST, MTBDRsl sensitivity for FQ resistance was only 83.1% suggesting that just less than one in five cases may be caused by mutations outside of gyrA, such as in gyrB, a gene which is not targeted by MTBDRsl. Only two studies (Brossier 2010; Huang 2011) performed genetic sequencing for both gyrA and gyrB and they reported sensitivity estimates of 84.6% and 100.0%, respectively.

ggesting that just less than one in five cases may be caused by mutations outside of gyrA, such as in gyrB, a gene which is not targeted by MTBDRsl. Only two studies (Brossier 2010; Huang 2011) performed genetic sequencing for both gyrA and gyrB and they reported sensitivity estimates of 84.6% and 100.0%, respectively. Similarly, we found higher pooled sensitivity for SLID resistance when MTBDRsl was evaluated against genetic sequencing rather than culture-based DST (97.0% versus 76.9%). In this case, both genetic sequencing and MTBDRsl only target the rrs gene for resistance to SLIDs. This approach can potentially miss mutations outside of this region that are responsible for SLID resistance. Using culture-based DST (sensitivity 76.9%), it appears that around one in four cases of SLID-resistant TB may be caused by mutations outside of rrs. The prevalence of these non-rrs mutations, which can occur in regions such as tlyA, eis and gidB (Georghiou 2012), appears to be most pronounced for kanamycin given the reduced sensitivity (66.9%) of MTBDRsl for resistance to this drug compared to the other SLIDs (sensitivity of 87.9% and 79.5% for amikacin and capreomycin, respectively, against culture-based DST). The sensitivity of MTBDRsl for SLID resistance, and in particular kanamycin resistance, is likely to vary according to the genetic background of TB strains, where some may have a greater frequency of resistance-causing mutations that fall outside of rrs and different levels of cross-resistance within the SLIDs. When we excluded a large study from Eastern Europe (Kontsevaya 2013), both the sensitivity of MTBDRsl and the precision of our pooled estimate improved.

TB strains, where some may have a greater frequency of resistance-causing mutations that fall outside of rrs and different levels of cross-resistance within the SLIDs. When we excluded a large study from Eastern Europe (Kontsevaya 2013), both the sensitivity of MTBDRsl and the precision of our pooled estimate improved. We are aware of an unpublished Foundation for Innovative and New Diagnostics-sponsored evaluation of MTBDRsl at the University of Cape Town (K. Dheda, personal communication). For the direct detection of drug resistance compared to culture-based DST as a reference standard, this work reported a sensitivity and specificity of 79.2% (38/48) and 86.5% (45/52), respectively, for the detection of resistance to ofloxacin, and a sensitivity and specificity of 72.9% (35/48) and 94.2% (49/52) respectively for the detection of resistance to amikacin. When performed indirectly on culture isolates, the sensitivity and specificity for the detection of ofloxacin resistance were 72.3% (115/159) and 99.0% (100/101) respectively, and 76.6% (125/157) and 98.0% (99/101) respectively for amikacin resistance, when compared to culture-based DST as a reference standard.

ikacin. When performed indirectly on culture isolates, the sensitivity and specificity for the detection of ofloxacin resistance were 72.3% (115/159) and 99.0% (100/101) respectively, and 76.6% (125/157) and 98.0% (99/101) respectively for amikacin resistance, when compared to culture-based DST as a reference standard. MTBDRsl is the only commercially-available rapid molecular test for the detection of resistance to the FQs, SLIDs and XDR-TB. Alternative phenotypic methods of DST for TB may take several weeks (Barnard 2012; Lopez-Roa 2012) to several months (Said 2012; Tukvadze 2014). This lengthy turnaround time, during which the patient may be on ineffective therapy and contribute to ongoing TB transmission, is further exacerbated by the need to first grow a M. tuberculosis isolate (which itself may take two to six weeks). Two systematic reviews of MTBDRsl exist (Feng 2013; WHO 2013b). As in our review, WHO 2013b used a random-effects meta-analysis model and arrived at similar summary estimates, generally within three percentage points of those described in our review. Feng 2013 used a fixed-effects model and reported accuracy estimates for kanamycin and capreomycin resistance that were substantially lower than the ones we found. Our review included additional studies not included in these previous reviews. Key questions remain regarding test accuracy and potential sources of heterogeneity, including risk of bias assessment, type of testing (indirect versus direct testing) and reference standard (for example, culture-based DST versus genetic sequencing). We address several of these questions in this review. Although we intended to investigate whether the observed test accuracy varied between studies according to HIV infection, specimen condition (frozen versus fresh), specimen type (induced sputum or extrapulmonary specimen), the drug concentration used in culture-based DST (studies that used WHO-recommended concentrations versus those that did not) or population (patients suspected of having MDR-TB or XDR-TB), there were unfortunately insufficient data to perform these additional analyses for each target condition. We were also unable to examine sources of heterogeneity for detection of XDR-TB due to insufficient data.

ed concentrations versus those that did not) or population (patients suspected of having MDR-TB or XDR-TB), there were unfortunately insufficient data to perform these additional analyses for each target condition. We were also unable to examine sources of heterogeneity for detection of XDR-TB due to insufficient data. More comparative diagnostic accuracy data are needed from strains from different geographic regions (for example, Eastern Europe), where resistance-causing mutations that fall outside of the genes targeted by MTBDRsl are less common than in drug-resistant strains from South Africa, for example. Such future research should include genetic sequencing as a reference standard that targets all known resistance-determining mutations and not just those detectable using MTBDRsl. Summary of main results The main results are presented in the 'Summary of findings' tables: When used indirectly on culture isolates, MTBDRsl detected 83.1% of FQ-resistant cases with high specificity (97.7%) when culture-based DST was used as a reference standard. When evaluated against genetic testing as a reference standard, the sensitivity and specificity were 99.3% and 99.7%, respectively. When used directly on smear-positive sputum specimens, MTBDRsl detected 85.1% of FQ-resistant cases with high specificity (98.2%).

When used indirectly on culture isolates, MTBDRsl detected 83.1% of FQ-resistant cases with high specificity (97.7%) when culture-based DST was used as a reference standard. When evaluated against genetic testing as a reference standard, the sensitivity and specificity were 99.3% and 99.7%, respectively. When used directly on smear-positive sputum specimens, MTBDRsl detected 85.1% of FQ-resistant cases with high specificity (98.2%). When used indirectly on culture isolates, MTBDRsl detected 76.9% of SLID resistant cases with high specificity (99.5%) when culture-based DST was used as a reference standard. The pooled sensitivities for resistance to amikacin, kanamycin and capreomycin were 87.9%, 66.9% and 79.5%, respectively. The sensitivity and specificity for SLID resistance evaluated against genetic testing as a reference standard were 97.0% and 99.5%, respectively. When used directly on smear-positive sputum specimens, MTBDRsl detected 94.4% of SLID-resistant cases with high specificity (98.2%). The pooled sensitivities for resistance to amikacin, kanamycin and capreomycin were 97.3%, 72.1% and 68.7%, respectively. When used indirectly on culture isolates, MTBDRsl detected 70.9% of XDR-TB cases with high specificity (98.8%) when culture-based DST was used as a reference standard. The proportion of indeterminate results was lower when MTBDRsl was performed indirectly rather than directly (0.2% versus 1.9% for FQ resistance, P < 0.001; 0.4% versus 6.1% for SLID resistance, P < 0.001; 0.1% versus 1.8% for XDR-TB resistance, P = 0.002).

When used indirectly on culture isolates, MTBDRsl detected 70.9% of XDR-TB cases with high specificity (98.8%) when culture-based DST was used as a reference standard. The proportion of indeterminate results was lower when MTBDRsl was performed indirectly rather than directly (0.2% versus 1.9% for FQ resistance, P < 0.001; 0.4% versus 6.1% for SLID resistance, P < 0.001; 0.1% versus 1.8% for XDR-TB resistance, P = 0.002). Application of the meta-analysis to a hypothetical cohort 'Summary of findings' tables (Table 1; Table 2; Table 3) summarize the review findings by applying the results to a hypothetical cohort of 1000 individuals with MDR-TB thought to have resistance to a FQ, or SLID, or both. We present scenarios, based on WHO estimates (WHO 2013a), with the prevalence of FQ resistance varying from 12% to 17% to 21%, that of SLID resistance varying from 15% to 23% to 30%, and that of XDR-TB varying from 8% to 11%. The consequences of FP results are likely patient anxiety, morbidity from additional testing, possible delay in further diagnostic evaluation, and prolonged and unnecessary treatment with drugs that may have lower bacteriocidal activity than second-line regimens and often have serious side effects. The consequences of FN results are an increased risk of patient morbidity and mortality, and continued risk of community transmission of drug-resistant TB.

uation, and prolonged and unnecessary treatment with drugs that may have lower bacteriocidal activity than second-line regimens and often have serious side effects. The consequences of FN results are an increased risk of patient morbidity and mortality, and continued risk of community transmission of drug-resistant TB. 1A. Indirect testing for fluoroquinolone resistance, MTBDRsl performed on culture isolates FQ resistance prevalence of 12%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 120 patients have FQ-resistant TB, then MTBDRsl would be expected to miss between 20 and 26 cases and falsely diagnose between eight and 50 cases. FQ resistance prevalence of 17%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 170 patients have FQ-resistant TB, then MTBDRsl would be expected to miss 23 and 26 cases and falsely diagnose between seven and 47 cases. FQ resistance prevalence of 21%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 210 patients have FQ-resistant TB, then MTBDRsl would be expected to miss between 28 and 45 cases and falsely diagnose between seven and 45 cases.

FQ resistance prevalence of 17%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 170 patients have FQ-resistant TB, then MTBDRsl would be expected to miss 23 and 26 cases and falsely diagnose between seven and 47 cases. FQ resistance prevalence of 21%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 210 patients have FQ-resistant TB, then MTBDRsl would be expected to miss between 28 and 45 cases and falsely diagnose between seven and 45 cases. 1B. Direct testing for fluoroquinolone resistance, MTBDRsl performed on sputum specimens FQ resistance prevalence of 12%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 120 patients have FQ-resistant TB, then MTBDRsl would be expected to miss between nine and 34 cases and falsely diagnose between nine and 28 cases. FQ resistance prevalence of 17%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 170 patients have FQ-resistant TB, then MTBDRsl would be expected to miss between 12 and 48 cases and falsely diagnose between eight and 27 cases. FQ resistance prevalence of 21%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 210 patients have FQ-resistant TB, then MTBDRsl would be expected to miss between 15 and 59 cases and falsely diagnose between eight and 25 cases.

FQ resistance prevalence of 17%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 170 patients have FQ-resistant TB, then MTBDRsl would be expected to miss between 12 and 48 cases and falsely diagnose between eight and 27 cases. FQ resistance prevalence of 21%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 210 patients have FQ-resistant TB, then MTBDRsl would be expected to miss between 15 and 59 cases and falsely diagnose between eight and 25 cases. 2A. Indirect testing for SLID resistance, MTBDRsl performed on culture isolates SLID resistance prevalence of 15%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 150 patients have SLID resistant TB, then MTBDRsl would be expected to miss between 19 and 58 cases and falsely diagnose between one and 25 cases. SLID resistance prevalence of 23%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 230 patients have SLID resistant TB, then MTBDRsl would be expected to miss between 29 and 89 cases and falsely diagnose between one and 22 cases. SLID resistance prevalence of 30%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 300 patients have SLID resistant TB, then MTBDRsl would be expected to miss between 37 and 117 cases and falsely diagnose between one and 20 cases.

SLID resistance prevalence of 23%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 230 patients have SLID resistant TB, then MTBDRsl would be expected to miss between 29 and 89 cases and falsely diagnose between one and 22 cases. SLID resistance prevalence of 30%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 300 patients have SLID resistant TB, then MTBDRsl would be expected to miss between 37 and 117 cases and falsely diagnose between one and 20 cases. 2B. Direct testing for SLID resistance, MTBDRsl performed on sputum specimens SLID resistance prevalence of 15%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 150 patients have SLID resistant TB, then MTBDRsl would be expected to miss between zero and 112 cases and falsely diagnose between three and 94 cases. SLID resistance prevalence of 23%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 230 patients have SLID resistant TB, then MTBDRsl would be expected to miss between zero and 172 cases and falsely diagnose between two and 85 cases. SLID resistance prevalence of 30%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 300 patients have SLID resistant TB, then MTBDRsl would be expected to miss between zero and 224 cases and falsely diagnose between two and 78 cases.

SLID resistance prevalence of 23%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 230 patients have SLID resistant TB, then MTBDRsl would be expected to miss between zero and 172 cases and falsely diagnose between two and 85 cases. SLID resistance prevalence of 30%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 300 patients have SLID resistant TB, then MTBDRsl would be expected to miss between zero and 224 cases and falsely diagnose between two and 78 cases. 3A. Indirect testing for XDR-TB resistance, MTBDRsl performed on culture isolates XDR-TB prevalence of 8%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 80 patients have XDR-TB, then MTBDRsl would be expected to miss between nine and 46 cases and falsely diagnose between four and 36 cases. XDR-TB prevalence of 11%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 110 patients have XDR-TB, then MTBDRsl would be expected to miss between 12 and 63 cases and falsely diagnose between four and 35 cases. Strengths and weaknesses of the review The results of this Cochrane Review are based on strict and careful searching, study inclusion and data extraction. The strength of this review is that it allows an assessment of different methods of testing (indirect versus direct) and different reference standards.

XDR-TB prevalence of 11%: if the pooled estimates for MTBDRsl are applied to a hypothetical cohort of 1000 patients with TB, where 110 patients have XDR-TB, then MTBDRsl would be expected to miss between 12 and 63 cases and falsely diagnose between four and 35 cases. Strengths and weaknesses of the review The results of this Cochrane Review are based on strict and careful searching, study inclusion and data extraction. The strength of this review is that it allows an assessment of different methods of testing (indirect versus direct) and different reference standards. Completeness of evidence This is a reasonably complete data set. We included any non-English studies we found. We excluded seven studies that were not diagnostic accuracy studies and 13 studies that were conference abstracts (however, several of these were included in the form of full-length published papers). We did not include unpublished data. Studies of diagnostic test accuracy tend to be poorly indexed (Whiting 2005) and we may therefore have missed some studies despite the comprehensive search.

studies that were conference abstracts (however, several of these were included in the form of full-length published papers). We did not include unpublished data. Studies of diagnostic test accuracy tend to be poorly indexed (Whiting 2005) and we may therefore have missed some studies despite the comprehensive search. Accuracy of the reference standards used For our primary analysis, we used phenotypic culture-based DST. This was the most frequently deployed reference standard in the included studies. Although considered to be the 'gold standard' for drug-resistant TB, culture-based DST is not 100% accurate for detection of drug resistance, in particular with respect to detection of second-line drug resistance. We also determined MTBDRsl accuracy using genetic testing (gene sequencing of loci known to be associated with drug resistance) and both genetic testing and culture-based DST as reference standards. Many TB experts consider genetic sequencing to be the best available reference standard, provided it encompasses all the possible resistance determining regions. In addition, we determined the accuracy of MTBDRsl against a fourth reference standard, where genetic testing was only performed as part of a discrepant analysis in culture-DST-MTBDRsl discordant specimens. However, in most cases we were unable to determine summary estimates due to the small number of studies and therefore were unable to compare MTBDRsl accuracy estimates using this reference standard with those obtained using culture-based DST, or genetic sequencing, or both, as the reference standard.