Browse the corpus

What are the new findings Children's perceived barriers to sport participation change over a short space of time. Perceived barriers may be unrelated to weight status. Perceived barriers may be similar for boys and girls before adolescence.

What are the new findings Children's perceived barriers to sport participation change over a short space of time. Perceived barriers may be unrelated to weight status. Perceived barriers may be similar for boys and girls before adolescence. Introduction Physical activity has been associated with numerous health benefits during childhood and adolescence.1 Sport is one of the most popular forms of physical activity worldwide,2 it contributes 55–65% of daily moderate to vigorous energy expenditure in youth,3 4 and conveys a range of psychosocial health benefits that are over and above those attributable to physical activity.5 6 However, as 20–30% of children in the UK, the USA and Australia do not participate in sport,7–9 and participation rates decline dramatically as children age, it is essential to investigate the barriers preventing participation in youth sport.10 11 Much of the current research focuses on reasons for participation and dropout.11–13 However, the understanding of barriers preventing participation from the child or adolescent's perspective is important. The ecological model of health behaviour was developed to ‘emphasize the environmental and policy concepts of behaviour, while incorporating social and psychological influences’.14 The social–ecological model of physical activity (EMPA) was developed from this, and ‘portrays physical activity behaviour as being influenced by interplay between environmental settings and biological and psychological factors’,15 providing a framework for designing and evaluating physical activity and other health-related behavioural interventions. The model comprises four main domains: intrapersonal (individual beliefs, knowledge, skills, age), social environment (relationships, culture, society), physical environmental (natural or man-made environments) and policy (legislation). Barriers to sports participation may fall into any of these domains.

interventions. The model comprises four main domains: intrapersonal (individual beliefs, knowledge, skills, age), social environment (relationships, culture, society), physical environmental (natural or man-made environments) and policy (legislation). Barriers to sports participation may fall into any of these domains. A review of the relatively small amount of research in the area found that adolescents perceived social and intrapersonal factors to be the most prominent barriers to sports participation.16 However, these studies involved cross-sectional research designs, and there is no research into how perceived barriers to participation change over time in the same children, and data on children's barriers (rather than adolescents’) is lacking.16 Barriers would be expected to change as children develop both physically and socially; what they gain in motor skills and sports knowledge17 may be tempered by a new awareness of social standing and peer influence.18

ver time in the same children, and data on children's barriers (rather than adolescents’) is lacking.16 Barriers would be expected to change as children develop both physically and socially; what they gain in motor skills and sports knowledge17 may be tempered by a new awareness of social standing and peer influence.18 In our recent prospective study in an English birth cohort,19 continued participation in sports clubs between the ages of 9 and 12 years was associated with decreased adiposity. This finding is particularly important given the high prevalence of obesity and associated negative outcomes among youth.9 Additionally, this finding suggests that individuals with increased adiposity may be less likely to participate in sports, and body-related or social barriers may be particularly prominent in this group.20 Similar barriers may be more evident among females,21 as girls and young women are less physically active and take part in less sport than their male peers.22 23 The novel aim of this study was to investigate how perceived barriers to participation in school and outside-school sports clubs change in the same cohort over 3 years. This information would allow interventions to be tailored to the specific needs of children and adolescents at a time of high risk of dropout. Three main hypotheses were tested: (1) perceived barriers will change from 9 to 12 years; (2) overweight children will perceive different barriers to children of healthy weight and (3) girls will perceive different barriers than boys.

to the specific needs of children and adolescents at a time of high risk of dropout. Three main hypotheses were tested: (1) perceived barriers will change from 9 to 12 years; (2) overweight children will perceive different barriers to children of healthy weight and (3) girls will perceive different barriers than boys. Methods Sample The Gateshead Millennium Study is a birth cohort of adolescents born in northeast England between June 1999 and May 2000.24 Briefly, all children born to Gateshead-resident mothers in 34 prespecified recruiting weeks were invited to participate with no exclusion criteria. In total, 1029 babies were recruited to the original cohort, 523 male (50.8%) 506 female (49.2%) predominantly from the white ethnic majority (98%). The metropolitan borough of Gateshead is a mix of urban and rural, and the northeast of England is more deprived than England in general.25 The sample has remained socioeconomically representative of northern England and stable throughout the study; at birth, 15% of the sample was in the most affluent quintile, 20% in the second most affluent quintile, then 23%, 22% and finally 19% of the sample in the least affluent quintile. At follow-up at age 9 years, 18% of the sample were in the most affluent quintile, followed by 22%, 22%, 20% and then 18% in the least affluent quintile.24 The northeast consistently has higher than national average levels of childhood obesity.26

en 23%, 22% and finally 19% of the sample in the least affluent quintile. At follow-up at age 9 years, 18% of the sample were in the most affluent quintile, followed by 22%, 22%, 20% and then 18% in the least affluent quintile.24 The northeast consistently has higher than national average levels of childhood obesity.26 Data for the current analyses were collected at two datasweeps: at ages 8–10 and 11–13 years, referred to as 9 and 12 years.19 For each phase, families who had not previously opted out of the cohort were sent an invitation letter and information leaflet. Informed consent was obtained from the main carer of each child; children provided written assent. Ethical approval for the study was granted by the Newcastle University Ethics Committee.

s.19 For each phase, families who had not previously opted out of the cohort were sent an invitation letter and information leaflet. Informed consent was obtained from the main carer of each child; children provided written assent. Ethical approval for the study was granted by the Newcastle University Ethics Committee. Measures: barriers to sport participation Children completed the Youth Sport Survey (adapted from Godin and Shepherd 198527), about sports clubs they attended both in school and outside school.19 For this study, 72% of children at 9 years (n=421), and 63% of children at 12 years (n=331) took part in a sports club, and participation at 12 years was associated with reduced fat mass index.19 The children also answered the question ‘Do you find it hard to take part in sports clubs for any reason?’ with free text answers for both school sports clubs and outside-school sports clubs. The responses were analysed using content analysis using an inductive thematic approach: the answer was read and the ‘theme’ or ‘subdomain’ associated with it was created. For example, the answer ‘I don't take part in any outside school clubs, because none of my friends do’, was used to create a subdomain of ‘friends don't go’, other similar responses were then placed into this category. Subdomains were then grouped thematically according to a simplified version of EMPA15 that consisted of three main domains: physical environmental, intrapersonal and social environment. A cumulative EMPA count was created for each domain of EMPA. A response could be placed into more than one category of EMPA if it included more than one factor. Subdomains were not created a priori, as we did not know what answers would be given.

in domains: physical environmental, intrapersonal and social environment. A cumulative EMPA count was created for each domain of EMPA. A response could be placed into more than one category of EMPA if it included more than one factor. Subdomains were not created a priori, as we did not know what answers would be given. Data were coded independently by the lead author and a research assistant, who then met to discuss and agree on any discrepancies and those that were difficult to code. Measures: anthropometry At each time point, height was measured to 0.1 cm with a Leicester height measure (Chasmors, London, UK), and weight to 0.1 kg in light indoor clothing. Body mass index (BMI) centiles and BMI z-scores according to age-specific UK 1990 data28 were derived, and children categorised into healthy weight (HW, <85th centile), overweight (OW, ≥85th <95th centile) or obese (OB, ≥95th centile). Stage of puberty was assessed at 12 years using the self-reported Pubertal Development Scale,29 a self-report measure of puberty for young adolescents with good reliability and validity.

and children categorised into healthy weight (HW, <85th centile), overweight (OW, ≥85th <95th centile) or obese (OB, ≥95th centile). Stage of puberty was assessed at 12 years using the self-reported Pubertal Development Scale,29 a self-report measure of puberty for young adolescents with good reliability and validity. Statistical analyses Data were analysed in SPSS V.21 and STATA V.13. Answers were analysed by EMPA domain and overweight status. To assess the change in reported barrier over time, a change variable was created; children either changed domain of EMPA or did not, and within-subjects longitudinal associations were then examined using logistic regression. χ2 Tests were used to assess the association of EMPA domain with overweight status. The influence of puberty at 12 years, and socioeconomic status (SES) (measured by Townsend score, an area-based measure derived from residential postcode,30 and divided into quintiles) was tested with one-way ANOVA. Differences between sexes were tested by logistic regression. Results At 9 years of age, 574 children took part, and at age 12 years, 500 adolescents took part, and 441 children answered the questions at both 9 and 12 years. There were no differences in BMI or BMIz-score between children who took part at both time points and those who did not, but results are restricted to those with data at both 9 and 12 years. Participant characteristics are shown in table 1. Table 1 Participant characteristics, n=441

Results At 9 years of age, 574 children took part, and at age 12 years, 500 adolescents took part, and 441 children answered the questions at both 9 and 12 years. There were no differences in BMI or BMIz-score between children who took part at both time points and those who did not, but results are restricted to those with data at both 9 and 12 years. Participant characteristics are shown in table 1. Table 1 Participant characteristics, n=441 9 Years 12 Years Male (n, %) 210 47.6 210 47.6 Female (n, %) 231 52.4 231 52.4 Body mass index (BMI) (mean, SD) 17.9 2.8 20.5 3.8 BMI z-score* (mean, SD) 0.5 1.1 0.7 1.2 Healthy weight (n, %) 297 67.7 274 62.7 Overweight† (n, %) 80 18.2 82 18.8 Obese† (n, %) 62 14.1 81 18.5 Stage of puberty‡ – – 2.3 0.6 *z-Scores calculated relative to age-specific UK 1990 reference data.28 †Cut-points for population monitoring of overweight were used to categorise the children into healthy weight (<85th centile), overweight (≥85th <95th centile) or obese (≥95th centile). Overweight and obese categories were combined for this analysis. ‡Continuous scale from 1 to 4; prepubertal to postpubertal. Responses to the question ‘Do you find it hard to take part in sports clubs for any reason?’ are described by socioecological domain of EMPA (figure 1). Children who reported no perceived barriers represented a substantial portion of the responses (40–60%), and were included in the analysis to reduce bias.

‡Continuous scale from 1 to 4; prepubertal to postpubertal. Responses to the question ‘Do you find it hard to take part in sports clubs for any reason?’ are described by socioecological domain of EMPA (figure 1). Children who reported no perceived barriers represented a substantial portion of the responses (40–60%), and were included in the analysis to reduce bias. Figure 1 Distribution of responses by domain of social–ecological model of physical activity. Data are per cent of total responses for each domain of social–ecological model of physical activity, for each category of school and outside-school sports clubs, at each age (‘no reported barrier’ has been removed for clarity). The number of subdomains within each domain of EMPA differed by age and by school or outside-school sports clubs; 46 subdomains were identified in total, there were 24, 24, 29 and 28 subdomains for 9 years school, 9 years outside-school, 12 years school, and 12 years outside-school sports clubs, respectively (see online supplementary table S2). 10.1136/bmjsem-2015-000079.supp1Supplementary table 2 Reported barriers to participation in school- and outside-school sports clubs, by domain and sub-domain of the social-ecological model of physical activity*

The number of subdomains within each domain of EMPA differed by age and by school or outside-school sports clubs; 46 subdomains were identified in total, there were 24, 24, 29 and 28 subdomains for 9 years school, 9 years outside-school, 12 years school, and 12 years outside-school sports clubs, respectively (see online supplementary table S2). 10.1136/bmjsem-2015-000079.supp1Supplementary table 2 Reported barriers to participation in school- and outside-school sports clubs, by domain and sub-domain of the social-ecological model of physical activity* Online supplementary table S2 shows how perceived barriers to taking part in sports clubs changed over 3 years. For younger children, the physical environmental domain was prominent, suggesting more practical difficulties, such as a lack of suitable club, for example, ‘[I would like to play] Tennis—isn't a club’ (ID1, boy, 9 years), ‘[I did] Street dance-not many people took part so it stopped’ (ID2, girl, 9 years), a lack of permission or transport: ‘[I would like to do] karate & football—[but there is] no one to take me’ (ID3, boy, 9 years), but also that clubs were only available to older or younger children. Lack of time was an issue at both ages: ‘I do nothing out of school or after school—not enough time’ (ID5, girl, 12 years). By adolescence, the respondents showed a marked change in their responses, with answers predominantly intrapersonal and socially environmental, and displaying a general lack of interest: ‘because … I can't be bothered to stay after school’ (ID4, boy 12 years), and having other interests and priorities, whether friends, family or homework: ‘I always have a lot of things to do at break and lunch, and none of the after school clubs suit me’ (ID6, girl, 12 years).

and displaying a general lack of interest: ‘because … I can't be bothered to stay after school’ (ID4, boy 12 years), and having other interests and priorities, whether friends, family or homework: ‘I always have a lot of things to do at break and lunch, and none of the after school clubs suit me’ (ID6, girl, 12 years). At both ages, there were children and adolescents describing themselves as ‘not sporty’, or disliking sports, and at 12 years worrying about not fitting in: ‘Sometimes I feel as if I won't fit in with others or maybe feel left out’ (ID7, boy, 12 years), ‘I don't feel comfortable and don't like doing things without my friends’ (ID8, girl, 12 years), ‘I am often very shy and rarely speak for myself. I am worried that I will get things mixed up so try to avoid things’ (ID9, girl, 12 years). There were reports about lack of fitness: ‘My chest really hurts when I exercise for too long’ (ID10, girl, 12 years), and also some starkly honest responses, for example, ‘my mum won't pick me up because she doesn't care’ (ID11, girl, 12 years), and: ‘because the internet exists’ (ID12, girl, 12 years).

rl, 12 years). There were reports about lack of fitness: ‘My chest really hurts when I exercise for too long’ (ID10, girl, 12 years), and also some starkly honest responses, for example, ‘my mum won't pick me up because she doesn't care’ (ID11, girl, 12 years), and: ‘because the internet exists’ (ID12, girl, 12 years). To assess within-subject change over time, only the first answer given was used, as the majority of children gave only one answer (15 and 12 children at 9 years gave a second answer for school and outside-school sports clubs, respectively, and 13 and 8 children at 12 years). The within-subjects change over time found that 154 of 435 children identified barriers within the same domain of EMPA for school sports clubs from 9 to 12 years, and 160 of 420 for outside-school sports club (table 2). The distributions were also different between school and outside-school sports clubs at both ages, although the changes in EMPA domain from 9 to 12 years were not statistically significant (χ2 p=0.054 for outside-school sports clubs and p=0.410 for school sports clubs). Table 2 Change in reported barrier to (a) school sports club participation and (b) outside-school sports club participation from age 9 to 12 years, by social-ecological model of physical activity (EMPA) domain

To assess within-subject change over time, only the first answer given was used, as the majority of children gave only one answer (15 and 12 children at 9 years gave a second answer for school and outside-school sports clubs, respectively, and 13 and 8 children at 12 years). The within-subjects change over time found that 154 of 435 children identified barriers within the same domain of EMPA for school sports clubs from 9 to 12 years, and 160 of 420 for outside-school sports club (table 2). The distributions were also different between school and outside-school sports clubs at both ages, although the changes in EMPA domain from 9 to 12 years were not statistically significant (χ2 p=0.054 for outside-school sports clubs and p=0.410 for school sports clubs). Table 2 Change in reported barrier to (a) school sports club participation and (b) outside-school sports club participation from age 9 to 12 years, by social-ecological model of physical activity (EMPA) domain 9 years EMPA domain Physical environment Intrapersonal Social environment No barrier reported Total (a) 12 years EMPA domain Physical environment 7 25 12 32 76 Intrapersonal 6 26 6 25 63 Social environment 3 16 13 16 48 No barrier reported 13 82 48 105 248 Total 29 142 86 178 435 (b) Physical environment 13 18 11 41 83 Intrapersonal 2 12 4 16 34 Social environment 5 8 8 26 47 No barrier reported 15 51 60 130 256 Total 35 83 89 213 420 Analysis by overweight status is shown in figure 2. There were no statistically significant differences between HW and OWOB children at each time point, however, change in perceived barrier was associated with OWOB status at 12 years, after controlling for sex, puberty and baseline OWOB status (OR 1.7, 95% CI to 1.0–3.0, p=0.033). This was due to more OWOB children reporting ‘no perceived barrier’ at 9 years, then a barrier at 12 years, than HW children (26% vs 34%).

, however, change in perceived barrier was associated with OWOB status at 12 years, after controlling for sex, puberty and baseline OWOB status (OR 1.7, 95% CI to 1.0–3.0, p=0.033). This was due to more OWOB children reporting ‘no perceived barrier’ at 9 years, then a barrier at 12 years, than HW children (26% vs 34%). Figure 2 Distribution of responses by domain of social–ecological model of physical activity and overweight status. Data are per cent of total responses for each domain of social–ecological model of physical activity, for each category of school and outside-school sports clubs, at each age, by weight status (‘no reported barrier’ has been removed for clarity). HW, healthy weight; OWOB, overweight or obese. The distribution of answers by EMPA was similar between the sexes, with the exception of outside-school sports clubs at 9 years, where girls gave fewest responses in the intrapersonal domain, but boys’ responses were split equally between intrapersonal and social environment. No statistically significant differences between the sexes were found either cross-sectionally at each age, or in the change in EMPA exhibited longitudinally when tested by χ2 (p>0.05), or by stage of puberty at 12 years (one-way ANOVA p>0.05). There were no associations with SES.

intrapersonal and social environment. No statistically significant differences between the sexes were found either cross-sectionally at each age, or in the change in EMPA exhibited longitudinally when tested by χ2 (p>0.05), or by stage of puberty at 12 years (one-way ANOVA p>0.05). There were no associations with SES. Discussion The current study is novel in using responses from the same group of children, at 9 years and then 12 years of age, and highlights the range of perceived barriers—46 were identified across just a 3-year timespan. The current work has also addressed an important gap in the literature by contributing to the knowledge of younger children's experiences.

sing responses from the same group of children, at 9 years and then 12 years of age, and highlights the range of perceived barriers—46 were identified across just a 3-year timespan. The current work has also addressed an important gap in the literature by contributing to the knowledge of younger children's experiences. Age-related changes There was a clear difference with age; as children, the perceived barriers were predominantly of a physical environmental nature. This is in keeping with children of this age and stage of development; they are not yet overly concerned about their social standing,18 and parental support is vital, as the children may be considered too young to travel home alone, do not have their own money (or parents cannot afford clubs/equipment), may forget to get permission forms completed, or are refused permission: ‘[I would like to go] swimming—Mum won't let me go back’ (ID 13, boy, 9 years). Others have found that parental support, plus access to a variety of clubs, are motivators for young children's participation in sports.16 A lack of time was cited at both ages, and at 9 years, this may have been due to homework or other clubs they enjoyed: ‘[I would like to do] karate—but clashes with piano lessons’ (ID 14, girl, 9 years). Lack of time and competing demands were reported by Canadian parents of children this age,31 and children in Ireland who had never participated in sports clubs provided similar reasons; they struggled to find suitable clubs, with transport, and with feelings of incompetence.32 Participants in the current study also reported that they ‘weren't good at sport’, or that it was ‘too hard’, and there were children with injuries, or who were scared of getting hurt. This could point to exposure to developmentally inappropriate sports,17 or that the children lacked the fundamental movement skills required to perform them. These could be difficult feelings to overcome, perhaps helped by improved teacher and coach training in understanding and assessing fundamental movement skills,33 and communicating their importance to parents.

inappropriate sports,17 or that the children lacked the fundamental movement skills required to perform them. These could be difficult feelings to overcome, perhaps helped by improved teacher and coach training in understanding and assessing fundamental movement skills,33 and communicating their importance to parents. The transition from primary to secondary school (and from childhood to adolescence), marked a clear distinction in barriers, with the most prominent subdomains capturing disinterest. Other researchers have emphasised the need for ‘sampling’ at these ages, where children try different sports with an emphasis on fun and participation, rather than competition.34 In the current study, many children already participated in a sports club, perhaps explaining why many children did not perceive any barriers. However, several responses could be intervention targets, including providing transport home, making clubs free/cheaper, and making clubs available to children of all ages. The results of this study also suggest that interventions to increase sports participation should be age-specific. The 12-year-olds’ concerns relating to their social environment emphasise the importance of friendship groups at this age: ‘I don't take part in any outside school clubs, because none of my friends do’ (ID15, girl, 12 years). This emerges during early adolescence, as they become more aware of what their friends think of them, and the need to feel accepted and similar.18 This has been described in a dance intervention for girls35; the authors suggest emphasising enjoyment and socialisation in recruitment campaigns.35 Peer acceptance and friendship quality are two important dimensions of peer influence that have been linked with increased commitment to sports, greater enjoyment, and improved psychosocial well-being among adolescents.36 37

irls35; the authors suggest emphasising enjoyment and socialisation in recruitment campaigns.35 Peer acceptance and friendship quality are two important dimensions of peer influence that have been linked with increased commitment to sports, greater enjoyment, and improved psychosocial well-being among adolescents.36 37 Some of the findings of this study echo those of Stanley et al38 39 who discussed physical activity (not specifically sport participation) with children aged 10–13 years in Australia. The children mentioned the importance of friends, parental support, lack of time and perceived enjoyment,38 39 highlighting a degree of generalisability of studies to this age group, enabling successful sports interventions developed in one setting to be applied more rapidly in others.

en aged 10–13 years in Australia. The children mentioned the importance of friends, parental support, lack of time and perceived enjoyment,38 39 highlighting a degree of generalisability of studies to this age group, enabling successful sports interventions developed in one setting to be applied more rapidly in others. Weight-related and sex-related barriers The analysis found no clear difference with overweight status, suggesting that children and young adolescents have similar concerns across the weight spectrum. This may help future intervention designs to be more inclusive and is consistent with the finding that there is no difference in participation in organised sports by weight status.40 Similarly, sport participation has previously been shown to differ by SES,19 while other research has found that SES and cultural background, but not BMI, predict dropout from sport.41 There was also no clear association with sex, which is surprising, given that levels of physical activity and sports participation are lower in girls than boys. Allender et al16 reviewed barriers reported by young women, however, few of them were replicated in the current study. This may reflect the younger age of the current participants, and that sports clubs at 9 years may be available to both sexes. Further follow-up will show how perceived barriers change again.

han boys. Allender et al16 reviewed barriers reported by young women, however, few of them were replicated in the current study. This may reflect the younger age of the current participants, and that sports clubs at 9 years may be available to both sexes. Further follow-up will show how perceived barriers change again. Strengths and limitations Strengths include the relatively large sample size, with repeated assessment over 3 years, across the key period of transitions from primary–secondary school and childhood–adolescence. The use of a simple tool to extract information could be perceived as a weakness, but the breadth of answers, and similarity with qualitative studies shows that a simple question could help to inform intervention design. Other limitations include a lack of representation from across the entire socioecological model; information on policy barriers was limited, but might have included ‘lack of clubs’ or ‘expense’, however, more information would be required to decide that. An additional qualitative study with alternative methods, for example, interviews, might have revealed insights that were not possible from the current method. Additional information across the entire range of adolescence would also add value, therefore, it would be useful to follow the cohort further into adolescence and adulthood. Generalisability to other settings may be affected by the predominantly white ethnic background of the participants, and the low SES of the region; more affluent families or regions may have more clubs available to them.

d value, therefore, it would be useful to follow the cohort further into adolescence and adulthood. Generalisability to other settings may be affected by the predominantly white ethnic background of the participants, and the low SES of the region; more affluent families or regions may have more clubs available to them. Setting this research in context The importance of sports participation throughout life is being recognised by individual governments and also internationally, by the WHO Europe network for health-enhancing physical activity (HEPA), The Association For International Sport for All (TAFISA) and the IOC. The recent report by the UK Government ‘Sporting Future: A New Strategy for an Active Nation’42 promises a commitment to non-elite elite sport, and to increasing the funding and coaching available to children from 5 years old, as well as encouraging sport for social and mental health benefits. The current study should feed into the knowledge base for those seeking to increase sports participation in children and adolescents, specifically by understanding the variety of barriers and pressures that children and young people face.

5 years old, as well as encouraging sport for social and mental health benefits. The current study should feed into the knowledge base for those seeking to increase sports participation in children and adolescents, specifically by understanding the variety of barriers and pressures that children and young people face. Conclusions The transition from childhood to adolescence represents a marked change in perceived barriers to participation in sports clubs that may not differ substantially by sex or weight status. Interventions need to be tailored to the specific needs of the age group that is being targeted, and cover as many domains of the socioecological model as possible. Furthermore, interventions should address a narrow range of ages, because a universal intervention is unlikely to be applicable to both 9-year-olds and 12-year-olds. The views of children and adolescents should be sought prior to, and during, intervention design and implementation. The authors specially thank the Gateshead Millennium Study families and children for their participation in the study, and to the schools for their helpful cooperation. The authors appreciate the support of Gateshead Health NHS Foundation Trust and Gateshead Education Authority. The authors warmly thank the research team for their effort. They also thank Kay Mann for additional statistical support.

for their participation in the study, and to the schools for their helpful cooperation. The authors appreciate the support of Gateshead Health NHS Foundation Trust and Gateshead Education Authority. The authors warmly thank the research team for their effort. They also thank Kay Mann for additional statistical support. Contributors: LB was responsible for data acquisition, analysis and interpretation of data; manuscript drafting and revision. JKR was responsible for data acquisition, manuscript drafting and revision. JJR, LG and SAV were responsible for the study conception and design, analysis and interpretation of data, manuscript drafting and revision. AJA, KNP and MSP were responsible for the study conception and design, manuscript drafting and revision. LB is the study guarantor. All authors have had full access to all the data in the study, and take responsibility for the integrity of the data and the accuracy of the data analysis. Funding: This work was supported by the Chief Scientist Office (grant number CZH/4/484), the University of Strathclyde, Gateshead Council, the Mental Health Foundation and Breathe North. The cohort was first established with funding from the Henry Smith Charity and Sport Aiding Research in Kids (SPARKS) and followed-up with grants from NPRI, Gateshead NHS Trust R&D, Northern and Yorkshire NHS R&D, and Northumberland, Tyne and Wear NHS Trust. LB is funded by a Newcastle University Research Fellowship. AJA is supported by a NIHR Professorship. Competing interests: None declared. Ethics approval: Newcastle University Ethics Committee.

Funding: This work was supported by the Chief Scientist Office (grant number CZH/4/484), the University of Strathclyde, Gateshead Council, the Mental Health Foundation and Breathe North. The cohort was first established with funding from the Henry Smith Charity and Sport Aiding Research in Kids (SPARKS) and followed-up with grants from NPRI, Gateshead NHS Trust R&D, Northern and Yorkshire NHS R&D, and Northumberland, Tyne and Wear NHS Trust. LB is funded by a Newcastle University Research Fellowship. AJA is supported by a NIHR Professorship. Competing interests: None declared. Ethics approval: Newcastle University Ethics Committee. Provenance and peer review: Not commissioned; externally peer reviewed. Data sharing statement: Data are available from the lead author on request.

What we already know Musculoskeletal/sport medicine injuries make up approximately 20% of the general practitioner (GP) workload. GPs often do not feel comfortable in managing musculoskeletal/sport medicine injuries, and therefore a high referral burden to secondary care often results. Within the UK health system, there is now an emphasis on shifting patient management from secondary and tertiary care into the community, with subsequent management pressures for community health workers, including GPs. New service models are needed to manage this demand. Background With approximately 20% of general practitioners’ (GPs)/family physicians’ workload being related to musculoskeletal (MSK) conditions,1–3 GPs and primary care staff need to have access to appropriate clinics for investigating and managing MSK conditions. Moreover, primary care staff commonly report low confidence levels in managing common MSK conditions,3 further increasing the need for specialist MSK and, a related discipline, sport and exercise medicine (SEM) knowledge within GP.

eed to have access to appropriate clinics for investigating and managing MSK conditions. Moreover, primary care staff commonly report low confidence levels in managing common MSK conditions,3 further increasing the need for specialist MSK and, a related discipline, sport and exercise medicine (SEM) knowledge within GP. The current health system within the UK dictates that patients with MSK/SEM symptoms are initially reviewed by the GP and then referred on to secondary care when further investigation and management is required. The secondary care options include an appointment in hospital with an orthopaedic specialist or review within orthopaedic Integrated Clinical Assessment and Treatment units (ICATs), typically within a community-based healthcare centre. However, waiting times for these specialties are long and waiting times for elective care are a common source of patient dissatisfaction within healthcare systems.4 What this study adds This study extends the analysis of the implementation of a unique musculoskeletal/sport and exercise medicine into routine GP to 6 months and builds on our previously published work.6 A musculoskeletal and sport and exercise medicine clinic, run by a GP with a specialist interest in these conditions, can be successfully integrated into a local GP clinic. This service model has the potential to reduce referrals to secondary care, produce high patient satisfaction and make sound economical sense with significant cost savings.

A musculoskeletal and sport and exercise medicine clinic, run by a GP with a specialist interest in these conditions, can be successfully integrated into a local GP clinic. This service model has the potential to reduce referrals to secondary care, produce high patient satisfaction and make sound economical sense with significant cost savings. More wide scale studies are required to ensure that these results can be replicated throughout the UK healthcare system and beyond. If GPs had appropriate specialist knowledge in this area, then the patients’ management could largely be within the local GP practice with secondary care input only for a select few patients not able to be managed appropriately by the GP, consequently decreasing the waiting times for elective care in hospitals. This latter approach would be supported by the Transforming Your Care policy in Northern Ireland (NI), with a shift in emphasis now from hospital-based to community-based management options.5 The ICAT service model has illustrated that GPs with a specialist interest can manage common MSK symptoms competently, but this project is developing this idea further by demonstrating that GPs with a specialist interest can manage these patients within their own practices. Aim To develop a reproducible GP-staffed MSK and SEM clinic based within primary care that is economically sound and sustainable within the current National Health Service (NHS) climate, providing high patient satisfaction.

If GPs had appropriate specialist knowledge in this area, then the patients’ management could largely be within the local GP practice with secondary care input only for a select few patients not able to be managed appropriately by the GP, consequently decreasing the waiting times for elective care in hospitals. This latter approach would be supported by the Transforming Your Care policy in Northern Ireland (NI), with a shift in emphasis now from hospital-based to community-based management options.5 The ICAT service model has illustrated that GPs with a specialist interest can manage common MSK symptoms competently, but this project is developing this idea further by demonstrating that GPs with a specialist interest can manage these patients within their own practices. Aim To develop a reproducible GP-staffed MSK and SEM clinic based within primary care that is economically sound and sustainable within the current National Health Service (NHS) climate, providing high patient satisfaction. Methods This quality improvement project was conducted in a Belfast GP practice of approximately 9000 patients and five GP partners. The study methodology and provisional results have been previously reported,6 but the current study develops the initial paper6 further by extending the analysis to 6 months.

Aim To develop a reproducible GP-staffed MSK and SEM clinic based within primary care that is economically sound and sustainable within the current National Health Service (NHS) climate, providing high patient satisfaction. Methods This quality improvement project was conducted in a Belfast GP practice of approximately 9000 patients and five GP partners. The study methodology and provisional results have been previously reported,6 but the current study develops the initial paper6 further by extending the analysis to 6 months. The GP practice uses Egton Medical Information Systems (EMIS) electronic patient records and is based approximately two miles from Belfast city centre. The practice introduced an MSK and SEM clinic staffed by one GP with a specialist interest in MSK and SEM conditions, with appropriate postgraduate qualifications in these areas, and its performance was reviewed over two 3-month time periods between August and October 2014 and then again between January and March 2015. A monthly 4 h clinic was held over these time periods and appointment times were approximately 20 min. All primary care staff within the practice could refer to the clinic any patient with MSK and/or SEM presentations, on whom they wanted some specialist input. Management options included injection therapy, exercise prescription and onward referrals to appropriate colleagues, for example, physiotherapy. No ethical approval was required for this study as it was an audit of a new service within routine GP.

MSK and/or SEM presentations, on whom they wanted some specialist input. Management options included injection therapy, exercise prescription and onward referrals to appropriate colleagues, for example, physiotherapy. No ethical approval was required for this study as it was an audit of a new service within routine GP. The practice's performance during the running of the clinic was compared with the same time periods in 2013 and 2014 to control for other variables except for the introduction of the GP-based MSK and SEM clinic. The variables which were monitored included cases seen in the clinic, waiting times, treatments given and onward referral to colleagues from the clinic. Equipment which was available to the GP within the GP-based MSK and SEM clinic included a clinical room with a computer to allow access to the patients’ GP clinical records, a couch and lighting. There were also facilities to undertake joint injections under appropriate sterile conditions. Within the primary care clinic, there was access to laboratory blood tests (to exclude an inflammatory condition) and imaging, including open access radiology for plain films. If needed, patients could be referred for input to orthopaedic ICATs and hospital specialties, including rheumatology and orthopaedics, using the typical referral methods.

nic, there was access to laboratory blood tests (to exclude an inflammatory condition) and imaging, including open access radiology for plain films. If needed, patients could be referred for input to orthopaedic ICATs and hospital specialties, including rheumatology and orthopaedics, using the typical referral methods. Ten patients were randomly selected to complete a patient satisfaction questionnaire (see online supplementary appendix 1). The patients were approached by the practice administration staff to complete the questionnaire via a paper copy after they had finished their clinic visit. This questionnaire was developed from a recognised patient satisfaction questionnaire for assessing GP services,7 with input from the Patient Experience Group at the Royal Group of Hospitals, Belfast. Results GP-staffed MSK and SEM clinic performance August to October 2014 Thirty-five patients were seen in the GP-staffed MSK and SEM clinic, 14 males and 21 females, between August and October 2014. The age range of patients seen was from 35 to 77 years. The patients were generally referred from the other GPs within the practice, but a minority of referrals were also sent from physiotherapy, podiatry and hospital colleagues. For example, one patient was referred to the clinic from rheumatology due to the long waiting times for joint injections within their department.

The patients were generally referred from the other GPs within the practice, but a minority of referrals were also sent from physiotherapy, podiatry and hospital colleagues. For example, one patient was referred to the clinic from rheumatology due to the long waiting times for joint injections within their department. With regard to management in the GP-based clinic, as well as the steroid injection, the patients were also given advice regarding appropriate conservative management options, including analgesia, muscle stretches and strengthening exercises. The most common management option employed within the clinic was the steroid injection. A source of appropriate patient information used within the clinic was from Arthritis Research UK (ARUK), who provide information leaflets for patients on various MSK symptoms.8 The main joint presenting to the clinic was the shoulder, with the main pathology detected here being within the supraspinatus muscle (table 1). Table 1 Main joints presenting in the musculoskeletal/sport and exercise medicine general practitioner clinic between August and October 2014 Joint with presenting symptom Number of times presenting Shoulder 13 Knee 6 Hip 5 Hand 4 Elbow 4 Foot 4 *Previously published (6).

The main joint presenting to the clinic was the shoulder, with the main pathology detected here being within the supraspinatus muscle (table 1). Table 1 Main joints presenting in the musculoskeletal/sport and exercise medicine general practitioner clinic between August and October 2014 Joint with presenting symptom Number of times presenting Shoulder 13 Knee 6 Hip 5 Hand 4 Elbow 4 Foot 4 *Previously published (6). January to March 2015 Forty-eight patients were seen in the GP-staffed MSK and SEM clinic within this period, 22 males and 26 females. The age range of patients seen was from 26 to 77 years. All patients were referred by a GP in the practice for further assessment of their symptoms. The main joint presenting to the clinic was again the shoulder, with the main clinical diagnosis detected here being supraspinatus bursitis and the primary management option being the local steroid injection (table 2). Table 2 Main joints presenting to the musculoskeletal/sport and exercise medicine general practitioner clinic between January and March 2015

January to March 2015 Forty-eight patients were seen in the GP-staffed MSK and SEM clinic within this period, 22 males and 26 females. The age range of patients seen was from 26 to 77 years. All patients were referred by a GP in the practice for further assessment of their symptoms. The main joint presenting to the clinic was again the shoulder, with the main clinical diagnosis detected here being supraspinatus bursitis and the primary management option being the local steroid injection (table 2). Table 2 Main joints presenting to the musculoskeletal/sport and exercise medicine general practitioner clinic between January and March 2015 Joint with presenting symptom Number of times presenting Shoulder 18 Elbow 7 Knee 7 Foot 6 Hip 6 Hand/wrist 5 Patient satisfaction questionnaire responses Ten participants were randomly selected from the six clinics held to complete a patient satisfaction questionnaire (see online supplementary appendix 1). This included four males and six females of mean age 51 years (to the nearest year). Their responses are summarised in table 3. The patients generally felt that it was easy to get an appointment at the GP-based clinic, with the investigating doctor generally doing enough tests and patients having high levels of confidence in the treating GP. Communication at the clinic was clear, with high patient satisfaction with the services offered. Patients appeared happier to have their conditions managed within their GP surgery than compared with hospital or another community-based health facility and would have prefer to see future specialists within their GP clinic rather than without. There was also a space at the bottom of the questionnaire for free text comments and three different people left a comment, included in online supplementary appendix 2. The patients’ comments illustrated how highly they valued the service at the GP clinic.

future specialists within their GP clinic rather than without. There was also a space at the bottom of the questionnaire for free text comments and three different people left a comment, included in online supplementary appendix 2. The patients’ comments illustrated how highly they valued the service at the GP clinic. Table 3 Patient satisfaction questionnaire responses

future specialists within their GP clinic rather than without. There was also a space at the bottom of the questionnaire for free text comments and three different people left a comment, included in online supplementary appendix 2. The patients’ comments illustrated how highly they valued the service at the GP clinic. Table 3 Patient satisfaction questionnaire responses Question Number of times selected Strongly agree (1) Agree (2) Neither agree nor disagree (3) Disagree (4) Strongly disagree (5) 1. Getting an appointment for the GP-based orthopaedic clinic at a convenient time was easy? 9 1 2. The doctor did enough tests to find out what was wrong with me? 8 1 1 3. I have absolute faith and confidence in the doctor at the GP-based orthopaedic clinic? 9 1 4. The doctor at the GP-based orthopaedic clinic did not tell me enough about the treatment? 1 9 5. The doctor at the GP-based orthopaedic clinic fully explained how the illness and treatment would affect my future health? 6 4 6. Appointments are easy to make whenever I need them at the GP-based orthopaedic clinic? 6 4 7. I felt perfectly satisfied with the way I was treated at the surgery when I attended the GP-based orthopaedic clinic? 9 1 8. The doctor showed a genuine interest in my problems at the GP-based orthopaedic clinic? 9 1 9. The doctor always puts me at ease at the GP-based orthopaedic clinic? 9 1 10. My general experience at the GP-based orthopaedic clinic was very good? 9 1 11. My experience at the GP-based orthopaedic clinic was generally better than if I had been referred to hospital for an orthopaedic outpatient review? 9 1 *Previously published (6).

s puts me at ease at the GP-based orthopaedic clinic? 9 1 10. My general experience at the GP-based orthopaedic clinic was very good? 9 1 11. My experience at the GP-based orthopaedic clinic was generally better than if I had been referred to hospital for an orthopaedic outpatient review? 9 1 *Previously published (6). GP, general practitioner. Referral statistics The practicese referral rates between August and October 2013 and 2014 are included in table 4, whereas the referral rates between January and March 2014 and 2015 are included in table 5. The number of onward referrals made by the treating GP after being seen within the MSK and SEM clinic in the period August to October 2014 was three—two to physiotherapy and one to rheumatology—to exclude an inflammatory cause of the patient's pain, whereas the number of onward referrals from the clinic during the time period January to March 2015 was two—one to occupational therapy and one to orthopaedics—for consideration of surgical management options. This is compared with the 83 referrals which would have been made from the practice if this MSK and SEM clinic did not exist. Comparing referral rates between August and October 2013 and the same time period in 2014, overall referrals from the practice were reduced by 147, orthopaedic referrals were reduced by 2, while rheumatology referrals were reduced by 3, MSK presentations to the practice were reduced by 60, and physiotherapy and X-ray referrals were reduced by 47 and 90, respectively. Table 4 Referral rates for the practice: August to October 2013 and 2014

Referral statistics The practicese referral rates between August and October 2013 and 2014 are included in table 4, whereas the referral rates between January and March 2014 and 2015 are included in table 5. The number of onward referrals made by the treating GP after being seen within the MSK and SEM clinic in the period August to October 2014 was three—two to physiotherapy and one to rheumatology—to exclude an inflammatory cause of the patient's pain, whereas the number of onward referrals from the clinic during the time period January to March 2015 was two—one to occupational therapy and one to orthopaedics—for consideration of surgical management options. This is compared with the 83 referrals which would have been made from the practice if this MSK and SEM clinic did not exist. Comparing referral rates between August and October 2013 and the same time period in 2014, overall referrals from the practice were reduced by 147, orthopaedic referrals were reduced by 2, while rheumatology referrals were reduced by 3, MSK presentations to the practice were reduced by 60, and physiotherapy and X-ray referrals were reduced by 47 and 90, respectively. Table 4 Referral rates for the practice: August to October 2013 and 2014 Practice statistics Overall practic referral rates Orthopaedic referrals MSK presentations within the practice Physiotherapy referrals X-ray referrals Rheumatology referrals August to October 2013 881 55 317 133 319 13 August to October 2014 732 53 257 86 229 10 *Previously published (6). MSK, musculoskeletal.

Table 4 Referral rates for the practice: August to October 2013 and 2014 Practice statistics Overall practic referral rates Orthopaedic referrals MSK presentations within the practice Physiotherapy referrals X-ray referrals Rheumatology referrals August to October 2013 881 55 317 133 319 13 August to October 2014 732 53 257 86 229 10 *Previously published (6). MSK, musculoskeletal. Table 5 Referral rates for the practice: January to March 2014 and 2015 Practice statistics Overall outpatient practice referral rates Orthopaedic referrals Overall community referrals (includes physiotherapy) Physiotherapy referrals X-ray referrals Rheumatology referrals January to March 2014 971 61 233 93 324 16 January to March 2015 819 52 159 52 321 12 Comparing the referral rates between January and March 2014 and the same time period in 2015, overall referrals to outpatients from the practice were reduced by 152, orthopaedic referrals were reduced by 9, while rheumatology referrals were reduced by 4, and physiotherapy and X-ray referral rates were reduced by 41 and 3, respectively.

erral rates between January and March 2014 and the same time period in 2015, overall referrals to outpatients from the practice were reduced by 152, orthopaedic referrals were reduced by 9, while rheumatology referrals were reduced by 4, and physiotherapy and X-ray referral rates were reduced by 41 and 3, respectively. Economic evaluation All the patients referred to the GP-based MSK clinic were seen within 4 weeks. A review of the 2013 orthopaedic ICAT waiting times in NI showed that 5833 patients (49.5%) were seen between 0 and 6 weeks of referral, 2304 patients (19.6%) between 6 and 9 weeks, 1872 patients (15.9%) waiting between 9 and 12 weeks, 1389 patients (11.8%) waiting between 12 and 15 weeks, 236 patients (2%) waiting between 15 and 18 weeks and 144 patients (1.2%) waiting more than 18 weeks to be seen (from a total of 11 778 patients). The cost in NI of a routine hospital orthopaedic outpatient review in 2014 was £213 and the average orthopaedic ICAT cost per attendance was £82 (information obtained through direct communications with the finance department of the Department of Health, Social Services and Public Safety (DHSSPS; NI)), whereas the cost of 1 h of GP-patient contact, including direct care staff costs with qualification costs, for 2013–2014 was £183 (information obtained through direct communications with the finance department of DHSSPS (NI)). Three patients were at least seen per hour and therefore the cost per patient reviewed at the GP-based MSK and SEM clinic was conservatively costed at £61 per patient in 2014. Therefore, if all the clinic's patients were reviewed within a hospital orthopaedic outpatient clinic, the cost to the NI health service would have been £17 679, or within the orthopaedic ICAT system the cost would have been £6806. This is compared with the £5063 which it cost to run the GP-based MSK and SEM clinic, a potential saving of between £1743 and £12 616 per 83 patients reviewed.

orthopaedic outpatient clinic, the cost to the NI health service would have been £17 679, or within the orthopaedic ICAT system the cost would have been £6806. This is compared with the £5063 which it cost to run the GP-based MSK and SEM clinic, a potential saving of between £1743 and £12 616 per 83 patients reviewed. Discussion Thirty-five and then 48 patients were reviewed at the GP-based MSK and SEM clinic in 2014 and 2015, respectively, all within 4 weeks of initial presentation to their own GP. Patient satisfaction with the service was generally very high, with all patients preferring to be reviewed within their own GP surgery rather than being referred to a hospital or another community-based health centre, in keeping with the principles of Transforming Your Care.5 This study extends the analysis of the implementation of a unique MSK/SEM into routine GP to 6 months and builds on our previously published quality improvement work.6

GP surgery rather than being referred to a hospital or another community-based health centre, in keeping with the principles of Transforming Your Care.5 This study extends the analysis of the implementation of a unique MSK/SEM into routine GP to 6 months and builds on our previously published quality improvement work.6 Referral trends Onward orthopaedic referrals from the GP practice were reduced by 11 during the study period compared with the same time periods in 2014 and 2015. MSK presentations to the GP practice in 2014 were reduced by 114 compared with the same time frame in 2013. This statistic was unfortunately not available for the 2015 cohort of patients due to changes within the computer system. This reduction in MSK presentations to the GP surgery could be explained by a more efficient management of these symptoms within the surgery, with patients utilising the GP-based MSK and SEM clinic and receiving appropriate management rather than continually re-presenting to their own GP on multiple occasions. The reduction in X-ray referrals, 93 during the full study period, has potentially significant consequences for patients, with radiation from X-rays being associated with certain adverse health consequences.9 Five referrals were made from the clinic to secondary care (two to physiotherapy and one each to rheumatology, orthopaedics and occupational therapy). However, if this clinic did not exist, then all 83 patients seen in the clinic would have been referred to secondary care for input. This clinic therefore successfully reduced the burden on secondary care for orthopaedic and MSK referrals.

siotherapy and one each to rheumatology, orthopaedics and occupational therapy). However, if this clinic did not exist, then all 83 patients seen in the clinic would have been referred to secondary care for input. This clinic therefore successfully reduced the burden on secondary care for orthopaedic and MSK referrals. Prevalence of various MSK conditions The main presenting conditions/symptoms in this project related to the shoulder, with supraspinatus and subacromial bursitis being the most common area for pathology. However, the back and knee have been reported as the most common body regions causing patients to attend their GP in patients with MSK symptoms,2 whereas other studies concluded that the back and neck were the most common presenting areas.3 This difference may be explained by the fact that the GP-based MSK and SEM clinic was receiving referrals from primary care, and the GPs were therefore filtering out these other MSK presentations within their own clinics. Previous authors have also found that women present more commonly than men for MSK problems,2 in keeping with our findings. This information should also be utilised when teaching GPs about primary care MSK medicine and the common joints which present to GPs. This would allow GPs to feel more confident in managing common MSK and SEM symptoms and therefore reduce the secondary care referral burden.

MSK problems,2 in keeping with our findings. This information should also be utilised when teaching GPs about primary care MSK medicine and the common joints which present to GPs. This would allow GPs to feel more confident in managing common MSK and SEM symptoms and therefore reduce the secondary care referral burden. Economic considerations Having a GP-based MSK and SEM clinic has the potential for significant cost savings for the NHS. Managing patients within their own GP practice through utilisation of a GP with a specialist interest in MSK and SEM conditions in this current project had a potential cost saving of between £1743 and £12 616 per 83 patients reviewed. This does not include the reductions in hospital, physiotherapy and X-ray referrals seen within this study as well as fewer MSK presentations to the GP surgery and these cost savings are therefore conservative. This economic saving was achieved with high patient satisfaction and occurs at a time when the Belfast Trust is under significant economic pressure.10 Challenges, lessons and future directions One of the main issues encountered during the process for the GP leading the MSK and SEM clinic was that the Trust does not have guidelines for patients on oral anticoagulants receiving intra-articular injections. This should be remedied by the Trust, particularly with the advent of the newer oral anticoagulant agents, although previous authors have suggested that joint injections with a therapeutic international normalised ratio is safe.11 12

e guidelines for patients on oral anticoagulants receiving intra-articular injections. This should be remedied by the Trust, particularly with the advent of the newer oral anticoagulant agents, although previous authors have suggested that joint injections with a therapeutic international normalised ratio is safe.11 12 With regard to the future development of this service, it is hoped that this model will provide the blueprint for a further roll-out of GP-based MSK and SEM clinics within NI. To enable patients to receive the best possible care within these clinics, it is hoped that GPs with a specialist interest in MSK and SEM will be able to be trained in the use of MSK ultrasound—the ‘stethoscope’ of modern medicine.13 This skill would be able to be utilised as both an investigation and to help with patient management via, for example, ultrasound-guided injections, and the provision of this service should be reviewed to illustrate if it provides a further cost saving to the MSK and SEM service.

the ‘stethoscope’ of modern medicine.13 This skill would be able to be utilised as both an investigation and to help with patient management via, for example, ultrasound-guided injections, and the provision of this service should be reviewed to illustrate if it provides a further cost saving to the MSK and SEM service. A further improvement to the service could be made by providing both platelet-rich plasma (PRP) and whole-blood injections in addition to the current provision of corticosteroid injections, for muscle,14 tendinopathic15 and arthritic conditions,16–18 if supported by an appropriate evidence base.19 This project also provides GPs with information on which MSK and SEM conditions commonly present to GP and which some GPs may have difficulty managing, particularly around the shoulder and knee. GPs and GP trainees should therefore be provided with appropriate education to help address this learning need and further reduce the onward referral from primary care of MSK and SEM conditions.

conditions commonly present to GP and which some GPs may have difficulty managing, particularly around the shoulder and knee. GPs and GP trainees should therefore be provided with appropriate education to help address this learning need and further reduce the onward referral from primary care of MSK and SEM conditions. Strengths and weaknesses The review of referral rates from the GP practice is dependent on the primary care staff coding appropriately and our statistics are therefore limited by the quality of the information which the staff enters into the system. This may therefore lead to underestimating or overestimating referral rates. The statistics available from the primary care computer system, EMIS, for the two study time frames were different due to mandatory changes to the computer system. The diagnoses made were largely clinical and we would have ideally confirmed these with appropriate imaging, but a pragmatic approach was taken to investigate and manage the patients, in keeping with ‘normal’ primary care. This study was performed in one GP practice, which allowed us 100% follow-up of our patients, but our sample size was relatively small with no long-term follow-up. The next stage for the quality improvement project will be to replicate these findings within a larger geographical area, with long-term follow-up of patients, and then present the findings to local commissioners to further extend this trial across different areas of the UK.

elatively small with no long-term follow-up. The next stage for the quality improvement project will be to replicate these findings within a larger geographical area, with long-term follow-up of patients, and then present the findings to local commissioners to further extend this trial across different areas of the UK. Summary With the financial constraints now faced by the NHS and new healthcare policies shifting focus from hospital-based to community-based management options, we present a novel service model for managing MSK and SEM problems in primary care. This model can make sound economic sense and deliver high patient satisfaction within primary care, with low waiting times, helping to reduce the referral burden on secondary care from primary care. We present a reproducible model that can be commissioned as a service by the local clinical commissioning groups and be extended throughout the NI and UK health service as part of Transforming Your Care policy.5 Supplementary Material Supplementary Data The author wishes to acknowledge the STEP medical leadership programme within the Belfast Trust, Northern Ireland, particularly Dr Claire Lundy and her team, for their help throughout this project. The author would also like to acknowledge the help of Dr Nigel Hart and Professor Margaret Cupples with regard to supervision. This project could not have been completed without the support of all the staff and patients at Springfield Road Surgery, Belfast. Twitter: Follow Neil Heron at @neilSportDoc Competing interests: None declared.

Supplementary Material Supplementary Data The author wishes to acknowledge the STEP medical leadership programme within the Belfast Trust, Northern Ireland, particularly Dr Claire Lundy and her team, for their help throughout this project. The author would also like to acknowledge the help of Dr Nigel Hart and Professor Margaret Cupples with regard to supervision. This project could not have been completed without the support of all the staff and patients at Springfield Road Surgery, Belfast. Twitter: Follow Neil Heron at @neilSportDoc Competing interests: None declared. Provenance and peer review: Not commissioned; externally peer reviewed.

Strengths and limitations of this study This study shows that university soccer athletes wearing full equipment and cleats tested on FieldTurf or a firm surface make around two more errors on Modified Balance Error Scoring System (M-BESS) testing as compared to control testing performed barefoot in shorts and a T-shirt on a firm surface. Different individual factors such as gender and the wearing of a brace or tape at the ankles/feet area may affect total M-BESS scores in different field conditions. These findings support the use of M-BESS in field settings for screening concussion if deviations from the baseline are seen. Findings may not be generalisable to athletes from other sports wearing a different set of equipment.

Different individual factors such as gender and the wearing of a brace or tape at the ankles/feet area may affect total M-BESS scores in different field conditions. These findings support the use of M-BESS in field settings for screening concussion if deviations from the baseline are seen. Findings may not be generalisable to athletes from other sports wearing a different set of equipment. Introduction Concussion is a brain injury characterised by an alteration in cerebral function caused by a direct acceleration or deceleration force transmitted to a freely mobile head.1 2 It is estimated that between 1.6 and 3.8 million sports-related concussions occur each year in the USA.3–5 It is a common injury in contact sports such as soccer and North American football.4 Consensus guidelines recommend removing any player who is exhibiting signs or symptoms of a concussion for evaluation.1 2 6 Making the diagnosis of a concussion is not always easy, as players can deny symptoms7 and neurocognitive testing may be similar, or only slightly different from baseline.8 The physical and traditional neurological examinations are often normal. The Balance Error Scoring System (BESS) was developed as an objective test to assess concussed athletes.9 10 This test has been found to be a useful physical examination tool to help differentiate concussed from non-concussed athletes, especially within the first few days following injury.11

minations are often normal. The Balance Error Scoring System (BESS) was developed as an objective test to assess concussed athletes.9 10 This test has been found to be a useful physical examination tool to help differentiate concussed from non-concussed athletes, especially within the first few days following injury.11 Examination of a potentially concussed athlete should ideally occur in a quiet space, removed from the distractions of the athletic environment.12 The diagnosis is often made more difficult by the fact that this assessment must be done in a timely fashion. To overcome the challenges of screening for concussion on the sidelines, a number of clinical assessment tools have been proposed.1 6 13 The Sport Concussion Assessment Tool versions 2 and 3 (SCAT-2 and SCAT-3) are commonly used tools that assess symptoms, physical, neurological, cognitive, balance and coordination examinations.6 13 The balance examination is a critical section of SCAT-2 and SCAT-3. While the traditional BESS testing is completed on a firm surface followed by testing on a foam surface, the modified version of BESS (M-BESS) included in SCAT-2 and SCAT-3 only assesses balance on a firm surface.14–16 The M-BESS requires balancing in three stances, that is, in double leg, single leg and tandem gait performed with eyes closed, hands placed on hips and standing barefoot on a firm surface.17 Balancing in each of the three stances is assessed with a maximum score of 10 points each. Mistakes are subtracted from the score, making the maximum total M-BESS score of 30.17 The normative scores for M-BESS have been published recently,15 16 but clinical judgment prevails as the gold standard for diagnosing concussion.10 18 To date, definitive research data are lacking on what absolute M-BESS scores reliably help rule in, or rule out, a concussion.19 Similarly, the literature on factors affecting M-BESS scores is unavailable. Nonetheless, studies on the traditional version of BESS test performed on firm and foam surfaces indicate that performances can be influenced by gender (females better than males), training, bracing, injury, exercise, fatigue and time since injury.20 21

the literature on factors affecting M-BESS scores is unavailable. Nonetheless, studies on the traditional version of BESS test performed on firm and foam surfaces indicate that performances can be influenced by gender (females better than males), training, bracing, injury, exercise, fatigue and time since injury.20 21 While some have suggested that M-BESS scores are reliable in athletes wearing athletic gear and cleats,19 to date, studies have not evaluated the M-BESS performance in athletes under these conditions.22 23 The knowledge gained by such a study might help sport medicine professionals who have to examine athletes during a game or practice when it is not feasible to have an athlete remove all of their gear and cleats.19 Therefore, the objective of this study is to assess the difference in M-BESS test scores in athletes wearing their protective equipment with cleats in two field conditions as compared to the traditional testing performed in bare feet on a firm surface while wearing shorts and a T-shirt. Methods Setting The study was conducted in the sport medicine clinic at McGill University, Montreal, Québec, Canada, during the 2014 varsity season. The university has male and female varsity soccer teams and a male varsity North American football team. The age range of players usually varies from 18 to 30 years. Each team is followed by a team of therapists/trainers, a sport medicine fellow and a supervising attending physician from the Department of Sport Medicine. Ethics approval was obtained from the Research Ethics Board of McGill University Health Centre.

team. The age range of players usually varies from 18 to 30 years. Each team is followed by a team of therapists/trainers, a sport medicine fellow and a supervising attending physician from the Department of Sport Medicine. Ethics approval was obtained from the Research Ethics Board of McGill University Health Centre. Study design In this cross-sectional study, three observers assessed M-BESS performances in male and female players under three conditions: (1) no protective equipment but wearing shorts and T-shirts in bare feet on a firm surface (control condition); (2) full protective equipment with cleats on FieldTurf; and (3) full protective equipment with cleats on firm surface. Football players removed their helmets for the M-BESS evaluations. FieldTurf is the surface used for all varsity games for football and soccer at McGill University. It is a synthetic surface which consists of polyethylene blend fibres with an infill bottom layer of sand, a middle layer mixture of sand and cryogenic rubber and a top layer of rubber. The fibres are meant to replicate blades of grass, while the infill acts as a cushion. A hard firm surface was used as the other testing condition, as this is the surface normally used for control measurements. It is also usually readily available to most sport medicine professionals during practice or game conditions, either in the locker room or close to the sidelines.

ile the infill acts as a cushion. A hard firm surface was used as the other testing condition, as this is the surface normally used for control measurements. It is also usually readily available to most sport medicine professionals during practice or game conditions, either in the locker room or close to the sidelines. The sample size estimation (n>37) was based on the assumption that M-BESS performances with and without cleats would be highly concordant (intraclass correlation coefficient (ICC) ≈ 0.80 with 95% CI of 0.20) with type I error set at 0.05% and study power at 95%. Participants Sampling was convenience based. This study included athletes from varsity football and soccer teams between the ages of 18 and 30 years. Those who were unfit to practise or play due to any injury or illness (eg, musculoskeletal injury, concussion, infectious illness) or had any known balance or vestibular problems were excluded. All participants signed informed consent and received a monetary compensation ($20) for their time.

e ages of 18 and 30 years. Those who were unfit to practise or play due to any injury or illness (eg, musculoskeletal injury, concussion, infectious illness) or had any known balance or vestibular problems were excluded. All participants signed informed consent and received a monetary compensation ($20) for their time. A total of 60 players were included in this convenience sampling-based study (table 1). Players were called by their coaches during the practice sessions to participate in this study. The study included athletes from varsity football (n=39) and soccer teams (n=21). No players approached refused to participate or were excluded because of current illness, injury (eg, musculoskeletal injury, concussion, infectious illness) or known balance or vestibular problems. For soccer athletes, 10 were females and 11 were males. Average age for the 60 athletes studied was 21.1 years (SD=1.8; range 18–25 years). Average body mass index (BMI) was 27.9 kg/m2 (SD=5.0). A significant proportion of players had a history of lower extremity injury (40.0%, n=24) or past diagnosed concussion (31.7%, n=19), but were not suffering from any of the condition at the time of testing. Approximately one in seven players (13.3%, n=8) had bracing of the ankles/feet and one player (1.7%) had tape wrapped at their ankles/feet during field testing. No players had bracing or taping of their ankles/feet during control testing. Table 1 Characteristics of study participants

A total of 60 players were included in this convenience sampling-based study (table 1). Players were called by their coaches during the practice sessions to participate in this study. The study included athletes from varsity football (n=39) and soccer teams (n=21). No players approached refused to participate or were excluded because of current illness, injury (eg, musculoskeletal injury, concussion, infectious illness) or known balance or vestibular problems. For soccer athletes, 10 were females and 11 were males. Average age for the 60 athletes studied was 21.1 years (SD=1.8; range 18–25 years). Average body mass index (BMI) was 27.9 kg/m2 (SD=5.0). A significant proportion of players had a history of lower extremity injury (40.0%, n=24) or past diagnosed concussion (31.7%, n=19), but were not suffering from any of the condition at the time of testing. Approximately one in seven players (13.3%, n=8) had bracing of the ankles/feet and one player (1.7%) had tape wrapped at their ankles/feet during field testing. No players had bracing or taping of their ankles/feet during control testing. Table 1 Characteristics of study participants Athletes, (n) Per cent Gender Male 50 83.3 Female 10 16.7 Players Soccer 21 35.0 Football 39 65.0 Tape* 1 1.7 Brace* 8 13.3 History of concussion 19 31.7 History of lower extremity injury 24 40.0 Mean SD Age 21.1 1.8 Body mass index 27.9 5.0 *Taping or bracing was only on the ankle/foot area during field testing.

Table 1 Characteristics of study participants Athletes, (n) Per cent Gender Male 50 83.3 Female 10 16.7 Players Soccer 21 35.0 Football 39 65.0 Tape* 1 1.7 Brace* 8 13.3 History of concussion 19 31.7 History of lower extremity injury 24 40.0 Mean SD Age 21.1 1.8 Body mass index 27.9 5.0 *Taping or bracing was only on the ankle/foot area during field testing. Measures Three observers (two fellows in sport medicine and one senior resident in emergency medicine) assessed M-BESS performance in three stances, in three conditions, independently of each other. One author, a sport medicine physician with over 20 years' experience, led an information session on proper M-BESS assessment prior to data collection to ensure uniform knowledge and standardised procedures for the M-BESS assessment. Before each test, the same set of instructions for the M-BESS was read aloud to each athlete, with an opportunity to ask questions or for clarification prior to beginning the test. The control condition was evaluated in a sport medicine clinic, while the other field conditions requiring the athletes to wear their protective equipment occurred on separate days on the sidelines during, or immediately after, team practices.

portunity to ask questions or for clarification prior to beginning the test. The control condition was evaluated in a sport medicine clinic, while the other field conditions requiring the athletes to wear their protective equipment occurred on separate days on the sidelines during, or immediately after, team practices. In each stance, the three observers counted the errors in deviations from the proper stance, that is, moving hands off of iliac crests, opening eyes, a step/stumble or fall, abduction or flexion of the hip beyond 30°, lifting forefoot or heel off testing surface and remaining out of the proper testing position for >5 s. Only one error was counted when multiple errors occurred at the same time. The number of errors in each stance was subtracted from a score of 10 for each of the three stances. The maximum total score for each testing condition was 30. Information about players' age, gender, height, weight, team membership (football or soccer), use of brace or taping, history of lower extremity injury and any past diagnosed concussions was recorded on a separate sheet.

ore of 10 for each of the three stances. The maximum total score for each testing condition was 30. Information about players' age, gender, height, weight, team membership (football or soccer), use of brace or taping, history of lower extremity injury and any past diagnosed concussions was recorded on a separate sheet. Analyses All the information gathered from testing was later entered on a spreadsheet by one of the investigators. A total of 10% of entries were verified by a second investigator with expected error rate of <1%. Mean M-BESS performance scores for each stance and total scores were computed by averaging scores by three observers. Paired t tests were used to assess differences between conditions as: (1) full protective equipment with cleats on FieldTurf versus control; (2) full protective equipment with cleats on firm surface versus control; and (3) full protective equipment with cleats on FieldTurf versus full protective equipment with cleats on firm surface. The above differences in M-BESS scores were also assessed for various player characteristics—for example, by age, BMI categories, team, taping or bracing of ankles/feet, history of lower limb injury and concussion. Further, individual differences for M-BESS, for example, by age, for each of the field conditions were assessed using Student's t test. Last, we assessed the interobserver reliability for three conditions by computing ICCs.22

tegories, team, taping or bracing of ankles/feet, history of lower limb injury and concussion. Further, individual differences for M-BESS, for example, by age, for each of the field conditions were assessed using Student's t test. Last, we assessed the interobserver reliability for three conditions by computing ICCs.22 Results The mean total M-BESS scores and mean scores for the individual position components for each of the three conditions tested are listed in table 2. Single-leg stance, tandem-leg stance and total M-BESS performance scores were significantly different (p<0.001) for both field testing conditions compared to control conditions. Double-leg stance for all conditions tested was almost identical as only two errors occurred in one athlete during the full equipment with cleats on FieldTurf M-BESS testing. The difference in total M-BESS scores was roughly two points lower for athletic equipment with cleats on FieldTurf versus control and for athletic equipment with cleats on firm surface versus control. The mean total M-BESS scores for athletic equipment with cleats on FieldTurf were 26.3 (SD=2.0), athletic equipment with cleats on firm surface were 26.6 (SD=2.1) and control were 28.4 (SD=1.5). There were no differences between the two different field conditions for single-leg stance (p=0.60), tandem-leg stance (p=0.57) and total M-BESS performance scores (p=0.26). Table 2 Mean scores for Modified Balance Error Scoring System (M-BESS) by three observers for three different testing conditions

Results The mean total M-BESS scores and mean scores for the individual position components for each of the three conditions tested are listed in table 2. Single-leg stance, tandem-leg stance and total M-BESS performance scores were significantly different (p<0.001) for both field testing conditions compared to control conditions. Double-leg stance for all conditions tested was almost identical as only two errors occurred in one athlete during the full equipment with cleats on FieldTurf M-BESS testing. The difference in total M-BESS scores was roughly two points lower for athletic equipment with cleats on FieldTurf versus control and for athletic equipment with cleats on firm surface versus control. The mean total M-BESS scores for athletic equipment with cleats on FieldTurf were 26.3 (SD=2.0), athletic equipment with cleats on firm surface were 26.6 (SD=2.1) and control were 28.4 (SD=1.5). There were no differences between the two different field conditions for single-leg stance (p=0.60), tandem-leg stance (p=0.57) and total M-BESS performance scores (p=0.26). Table 2 Mean scores for Modified Balance Error Scoring System (M-BESS) by three observers for three different testing conditions Mean SD Barefoot on firm surface (control) Double legs 10.0 0 Single leg 8.7 1.3 Tandem 9.7 0.5 Total 28.4 1.5 Cleats on FieldTurf*† Double legs 9.9 0.4 Single leg 7.2 1.4 Tandem 9.2 1.2 Total 26.3 2.0 Cleats on firm surface*† Double legs 10.0 0 Single leg 7.3 1.7 Tandem 9.3 0.9 Total 26.6 2.1 N=60 athletes tested. *Wearing full protective equipment with cleats.

Mean SD Barefoot on firm surface (control) Double legs 10.0 0 Single leg 8.7 1.3 Tandem 9.7 0.5 Total 28.4 1.5 Cleats on FieldTurf*† Double legs 9.9 0.4 Single leg 7.2 1.4 Tandem 9.2 1.2 Total 26.3 2.0 Cleats on firm surface*† Double legs 10.0 0 Single leg 7.3 1.7 Tandem 9.3 0.9 Total 26.6 2.1 N=60 athletes tested. *Wearing full protective equipment with cleats. †Football players removed their helmets for the M-BESS evaluation. Table 3 shows the effects of player characteristics on mean total M-BESS scores for control versus the two field conditions. Overall, the significant differences between control and the two field conditions persisted in almost all player characteristics except two conditions. For the subset of players with a BMI≥30 kg/m2 (who were all male football players), there was no difference between control testing and both field testing conditions for mean total M-BESS scores. Also, male soccer players had no statistical difference in mean total M-BESS scores between control testing and testing with athletic equipment and cleats on firm surface. Table 3 Association of player characteristics with differences in total Modified Balance Error Scoring System (M-BESS) scores for three different testing conditions

Table 3 shows the effects of player characteristics on mean total M-BESS scores for control versus the two field conditions. Overall, the significant differences between control and the two field conditions persisted in almost all player characteristics except two conditions. For the subset of players with a BMI≥30 kg/m2 (who were all male football players), there was no difference between control testing and both field testing conditions for mean total M-BESS scores. Also, male soccer players had no statistical difference in mean total M-BESS scores between control testing and testing with athletic equipment and cleats on firm surface. Table 3 Association of player characteristics with differences in total Modified Balance Error Scoring System (M-BESS) scores for three different testing conditions Barefoot on firm surface (control) Cleats* on FieldTurf Cleats* on firm surface M-BESS M-BESS M-BESS Athletes, (n) M SD M SD M SD Gender (only soccer players) Male 11 29.1 1.2 27.3† 1.6 27.8‡ 2.0 Female 10 28.5 1.0 24.9† 1.9 26.2† 1.9 Players (only male players) Football§ 39 28.2 1.6 26.4† 2.0 26.4† 2.2 Age (years) 18–21 39 28.4 1.6 26.5† 2.0 26.8† 2.1 22–30 21 28.4 1.3 26.0† 2.1 26.2† 2.1 Body mass index (kg/m2) <25 24 28.8 1.2 26.1† 2.1 26.8† 2.0 25 to <30 24 28.4 1.4 26.5† 2.1 26.4† 2.3 ≥30 12 27.8 1.9 26.3‡ 1.9 26.5‡ 2.1 Taping or braces¶ No 51 28.6 1.3 26.5† 2.0 26.9† 2.0 Yes  9 27.6 1.9 25.2† 2.2 24.6† 1.7 History of concussion No 41 28.4 1.4 26.1† 2.0 26.5† 2.1 Yes 19 28.5 1.6 26.8† 2.1 26.9† 2.3 History of lower limb injury No 36 28.2 1.6 26.5† 1.8 26.8† 2.0 Yes 24 28.8 1.2 26.1† 2.4 26.3† 2.3 p Values refer to differences from control M-BESS values.

o 51 28.6 1.3 26.5† 2.0 26.9† 2.0 Yes  9 27.6 1.9 25.2† 2.2 24.6† 1.7 History of concussion No 41 28.4 1.4 26.1† 2.0 26.5† 2.1 Yes 19 28.5 1.6 26.8† 2.1 26.9† 2.3 History of lower limb injury No 36 28.2 1.6 26.5† 1.8 26.8† 2.0 Yes 24 28.8 1.2 26.1† 2.4 26.3† 2.3 p Values refer to differences from control M-BESS values. *Wearing full protective equipment with cleats. †Statistically different from control (p<0.01). ‡Statistically different from control (0.01≤p<0.05). §Football players removed their helmets for the M-BESS evaluation. ¶Taping or bracing was only on the ankle/foot area during field testing. M, mean. There was no significant difference in the mean total M-BESS scores between male football and soccer players for all conditions tested (p=0.06 or higher). In fact, we found no significant difference in M-BESS scores between player groups except for two cases. First, female soccer players performed significantly worse than male soccer players for cleats on FieldTurf condition (24.9 vs 27.3, p=0.005). Second, players who had taping or bracing of their ankles/feet had significantly lower scores than other players for cleats on firm surface condition (24.6 vs 26.9, p=0.002). A moderate-to-high interobserver reliability (0.60≤ICC≤0.75) was observed for total M-BESS scores under three conditions (table 4). The interobserver reliability was higher for barefoot on firm surface condition as compared to the other two field conditions. Table 4 Interobserver (n=3) reliability of Modified Balance Error Scoring System (M-BESS) scores for three different testing conditions

A moderate-to-high interobserver reliability (0.60≤ICC≤0.75) was observed for total M-BESS scores under three conditions (table 4). The interobserver reliability was higher for barefoot on firm surface condition as compared to the other two field conditions. Table 4 Interobserver (n=3) reliability of Modified Balance Error Scoring System (M-BESS) scores for three different testing conditions Intraclass correlation coefficient 95% CIs Barefoot on firm surface (control) Single leg 0.75 0.64 to 0.84 Tandem 0.71 0.56 to 0.81 Total 0.75 0.63 to 0.84 Cleats on FieldTurf*† Single leg 0.53 0.36 to 0.68 Tandem 0.52 0.36 to 0.37 Total 0.60 0.44 to 0.73 Cleats on firm surface*† Single leg 0.61 0.46 to 0.74 Tandem 0.67 0.54 to 0.79 Total 0.68 0.54 to 0.79 Intraclass correlation coefficients were not estimated for double-leg stance because there were nearly no balancing errors. *Wearing full protective equipment with cleats. †Football players removed their helmets for the M-BESS evaluation. Discussion This study quantifies the differences in M-BESS scores that should be expected if the test is performed in real-world field conditions for university athletes playing football or soccer. Findings suggest that, as compared to control testing, mean M-BESS scores were around 2 points lower when performed with protective equipment and cleats on FieldTurf or firm surface.

cores that should be expected if the test is performed in real-world field conditions for university athletes playing football or soccer. Findings suggest that, as compared to control testing, mean M-BESS scores were around 2 points lower when performed with protective equipment and cleats on FieldTurf or firm surface. These findings are relevant for health professionals in sports, perhaps as well as coaches, who both have the responsibility of assuring the safety of the players.24 This study extends the scope of M-BESS assessments, and it shows that M-BESS testing is feasible in different field conditions. Findings indicate to what extent scores should be adjusted when evaluating football or soccer athletes wearing protective equipment with cleats on two different surfaces.

ety of the players.24 This study extends the scope of M-BESS assessments, and it shows that M-BESS testing is feasible in different field conditions. Findings indicate to what extent scores should be adjusted when evaluating football or soccer athletes wearing protective equipment with cleats on two different surfaces. While this study did not assess acutely concussed athletes, it suggests that non-concussed athletes tested on FieldTurf or a firm surface wearing their protective equipment and cleats should be expected to make on average two more mistakes during M-BESS testing as compared to control testing. If athletes are to be tested in field conditions during game or practice situations, it would seem prudent to gather baseline scores while they are wearing their protective equipment and cleats either on a firm or their regular playing surface. This would allow sport medicine professionals to compare an athlete's postinjury M-BESS scores with their own baseline score performed under similar conditions and on similar surfaces. These precautions may help in avoiding misinterpretation of M-BESS scores in field conditions that are different from control testing, as was observed in this study.

e professionals to compare an athlete's postinjury M-BESS scores with their own baseline score performed under similar conditions and on similar surfaces. These precautions may help in avoiding misinterpretation of M-BESS scores in field conditions that are different from control testing, as was observed in this study. Equipment versus cleats In this study, during field testing conditions, football athletes were wearing shoulder pads, hip pads, thigh pads, a jersey, pants and cleats (helmets were removed for testing), while soccer athletes were wearing only shin pads, a jersey, shorts and cleats. Given the fact there were no significant differences in mean M-BESS scores between football and soccer players for any of the three testing conditions, differences from control conditions for both groups of athletes are likely due to the wearing of cleats during field testing conditions. While football and soccer players practise and play on the same FieldTurf, they do wear different styles of cleats. The type of cleats and length of studs on the cleats were not assessed in this study, but it may be an area of future study. Inter-rater reliability This study noted that inter-rater reliability decreased in conditions other than the control condition. This indicates that sport medicine professionals using M-BESS in field conditions should receive periodic group training to ensure as much homogeneity as possible when interpreting M-BESS errors in athletes wearing protective equipment and cleats in field conditions.

ased in conditions other than the control condition. This indicates that sport medicine professionals using M-BESS in field conditions should receive periodic group training to ensure as much homogeneity as possible when interpreting M-BESS errors in athletes wearing protective equipment and cleats in field conditions. Limitations This study has several limitations. First, only players from two sports were recruited. Therefore, findings might not be applicable to other sports using other equipment—for example, lacrosse and baseball.25 Second, there were only 10 female soccer players and 11 male soccer players in the study. Any differences between males and females, field testing versus control testing for soccer players, etc, may be better identified by a larger study with more soccer players, and in particular, more females included in the study group. Third, the M-BESS performance was measured during training sessions, which might have different effects on fatigue levels as compared to a competitive match.23 The findings might therefore be biased towards observing small differences. Nonetheless, an attempt was made to recruit players towards the end of the training session to account for the effects of fatigue. Fourth, testing was not performed on any natural grass surfaces. FieldTurf is a fairly flat consistent surface, whereas natural grass may be uneven and the firmness may vary under different weather conditions. Fifth, there are a range of values that may be seen in normal testing, in all conditions, as evidenced by our SDs. While a larger study may help to lower SDs and detect differences in the effect of different player characteristics on testing which this study was unable to find, it is likely that a range of normal values will always exist when testing a large group of athletes. While data do exist as to what should constitute an obvious abnormal traditional BESS score done on firm and foam surfaces, similar data values do not exist for the M-BESS scoring system.19 As per current practice, the findings suggest the responsibility of interpreting M-BESS scores relies on individual sport medicine professionals. As mentioned previously, having a baseline score under the same field conditions may be helpful when interpreting an individual athlete's postinjury M-BESS scores. Last, these tests were performed on non-concussed athletes. The differences in M-BESS scores in field conditions versus control for concussed athletes have not been determined.

sly, having a baseline score under the same field conditions may be helpful when interpreting an individual athlete's postinjury M-BESS scores. Last, these tests were performed on non-concussed athletes. The differences in M-BESS scores in field conditions versus control for concussed athletes have not been determined. While it may be logical to assume that M-BESS scores with significantly higher than a two-point difference between control and field condition testing may be due to other effects, such as a concussion, the fact that a range of normal values exist for M-BESS testing must be taken into consideration. Conclusion This study showed that total M-BESS performance scores in university North American football and soccer athletes wearing protective equipment with cleats in field settings were roughly two points less than the control tests performed in barefoot conditions on a firm surface. These findings may make M-BESS more accessible to sport medicine professionals who often face the reality of not being able to test an injured player in an ideal clinical setting during an ongoing match. The authors are grateful to players, coaches, therapists and trainers of McGill University football and soccer teams for facilitating this work. Contributors: AMA, JSD and JAB conceived the study. AMA, SAJ, JAB collected the data. AMA, JAB and JSD analysed the data. All authors were involved in interpretation of analysis. AMA wrote the first draft. All authors critically reviewed and approved the manuscript before submission.

The authors are grateful to players, coaches, therapists and trainers of McGill University football and soccer teams for facilitating this work. Contributors: AMA, JSD and JAB conceived the study. AMA, SAJ, JAB collected the data. AMA, JAB and JSD analysed the data. All authors were involved in interpretation of analysis. AMA wrote the first draft. All authors critically reviewed and approved the manuscript before submission. Funding: This work was partially funded by the McGill University Emergency Medicine Research Committee. Competing interests: None declared. Patient consent: Obtained. Ethics approval: This study was approved by the McGill University Health Centre Research Ethics Board. Provenance and peer review: Not commissioned; externally peer reviewed. Data sharing statement: No additional data are available.

What are the new findings? Standing in a rocker-sole shoe reduced postural stability compared with standing barefoot, whereas standing in a flat-sole shoe did not influence postural stability. Long-term use of rocker-sole or flat-sole shoes do not influence postural stability in barefoot standing. How might it impact on clinical practice in the near future? This study questions the belief that balance rehabilitation, especially when delivered in standing using rocker-sole shoes, will result in a long-term influence on postural control in people with chronic low back pain (CLBP). Treatment approaches directed towards influencing or ‘normalising’ altered CoP parameters may not be appropriate for people with CLBP. Introduction Differences in postural control during standing have been reported in people with chronic low back pain (CLBP).1–9 During more challenging standing conditions, defined as standing on compliant ground with visual occlusion, people with CLBP demonstrate increased centre of pressure (CoP) displacements and velocities, thought to indicate a reduced ability to maintain postural stability.10 These differences in postural control have been proposed as underpinning mechanisms in the presence and recurrent nature of CLBP.7 11

th visual occlusion, people with CLBP demonstrate increased centre of pressure (CoP) displacements and velocities, thought to indicate a reduced ability to maintain postural stability.10 These differences in postural control have been proposed as underpinning mechanisms in the presence and recurrent nature of CLBP.7 11 Greater CoP displacements, interpreted as increased postural instability, are reported during standing wearing rocker-sole compared with traditional flat-sole shoes,12–14 suggesting rocker-sole shoes may act as a balance training device. Rehabilitation with proprioceptive or balance training has demonstrated clinical benefits in people with functional ankle instability and anterior cruciate ligament-deficient knees15 16 and is recommended as a CLBP treatment.17 To the authors' knowledge, no published study has investigated the short-term and long-term influence of rocker-sole shoes on postural stability in people with CLBP. Hence, the following hypotheses were investigated: H1:Standing in rocker-sole shoes will promote a greater postural instability than standing in flat-sole shoes in the anteroposterior direction compared with barefoot standing. H2:Individuals presenting with CLBP who wear rocker-sole shoes as part of their rehabilitation programme will improve their barefoot standing stability in the anteroposterior direction in the shorter (6 weeks) and longer term (6 months) against those who wear standard flat-sole trainers. Methods This randomised trial with repeated measures recruited participants from a study investigating the influence of footwear on CLBP.18

H2:Individuals presenting with CLBP who wear rocker-sole shoes as part of their rehabilitation programme will improve their barefoot standing stability in the anteroposterior direction in the shorter (6 weeks) and longer term (6 months) against those who wear standard flat-sole trainers. Methods This randomised trial with repeated measures recruited participants from a study investigating the influence of footwear on CLBP.18 Participant recruitment, consent and randomisation Following ethical approval from Outer North London Research Ethics Committee (REC: 10/H0724/7), 20 participants, previously consented and block randomised in a clinical study investigating the effects of footwear on CLBP,18 were invited to take part by CSM. Inclusion criteria were: aged 18–65 years, with a 3-month or greater history of LBP. Exclusion criteria were as the main trial,18 excluding constant LBP, specific spinal diagnosis inappropriate for physiotherapy interventions (eg, spinal fracture of infection); any condition inappropriate for exercise physiotherapy (eg, severe cardiovascular or metabolic disease) or for wearing rocker-sole footwear (eg, Morton's neuroma, peripheral neuropathy); and participants who had previously used rocker-sole shoes. Interventions On consenting and entering the current study, participants were already allocated either the rocker-sole (Masai Barefoot Technology (MBT) Chapa Caviar, Masai GB Limited, London, UK) or the flat-sole shoe (Gel 1140, ASICS, Warrington, UK) (figure 1).18 Figure 1 Study shoes: rocker-sole shoe (top); flat-sole shoe (bottom).

Interventions On consenting and entering the current study, participants were already allocated either the rocker-sole (Masai Barefoot Technology (MBT) Chapa Caviar, Masai GB Limited, London, UK) or the flat-sole shoe (Gel 1140, ASICS, Warrington, UK) (figure 1).18 Figure 1 Study shoes: rocker-sole shoe (top); flat-sole shoe (bottom). Participants had been fitted with their allocated footwear type and taught how to walk in their shoes (see online supplementary appendix 1). They were instructed not to wear their allocated shoes prior to baseline biomechanical assessment, then wear them for a minimum of 2 hours/day while standing or walking for the study duration. Between baseline and 6-week assessment, participants attended a 4-week LBP exercise group (fulfilling methods of the main clinical study participants were recruited from18). 10.1136/bmjsem-2016-000170.supp1supplementary appendix Data collection Data collection occurred at the ‘One Small Step Gait Laboratory’, Guys' Hospital, London. Demographic, back-pain disability (Roland-Morris Questionnaire) and pain scores (numerical rating scale) were recorded at baseline. Biomechanical assessment Participants were assessed wearing short trousers and vest or no top. Participants' anthropometric measurements (pelvic width; leg length; knee width; ankle width; height; and weight) were recorded to inform the mechanical model formulated for each participant in Vicon's Nexus (1.8.1) motion capture software (Vicon Motions systems, Oxford, UK).

wearing short trousers and vest or no top. Participants' anthropometric measurements (pelvic width; leg length; knee width; ankle width; height; and weight) were recorded to inform the mechanical model formulated for each participant in Vicon's Nexus (1.8.1) motion capture software (Vicon Motions systems, Oxford, UK). Participants were assessed barefoot and shod, with their feet on adjacent force plates (FP5000, AMTI, Massachusetts, USA), during four posture-challenging standing conditions involving manipulation of visual input and support surface: (1) firm surface, eyes-open; (2) firm surface, eyes-closed; (3) compliant surface, eyes-open; (4) compliant surface, eyes-closed. Compliant surface was achieved by placing an Airex™ cushion (48.5×40.0×6.4 cm, 0.7 kg, high density (50 kg/m3), closed-cell foam) (l-group, St Louis, Missouri, USA) over each force plate (figure 2). Figure 2 Participant standing on foam cushions overlying force plates. Barefoot assessment Participants stood barefoot, feet approximately pelvis width apart and were instructed to keep their eyes focused on a red sticker at eye height on a tripod 3 m in front of them.19 Participants were assessed for three 40 s trials (shown to produce acceptable reliability20) for each standing condition. The middle 30 s of each trial was analysed to avoid possible initial sway errors and effects of participant fatigue or anticipation of a trial ending.

eye height on a tripod 3 m in front of them.19 Participants were assessed for three 40 s trials (shown to produce acceptable reliability20) for each standing condition. The middle 30 s of each trial was analysed to avoid possible initial sway errors and effects of participant fatigue or anticipation of a trial ending. Each participant received the same instructions at the start of each trial:When I say ‘Go’ I want you to stand and maintain your balance until you hear the instruction to rest. Each trial will last for 40 seconds. Focus on the red sticker on the tripod ahead of you. Keep your arms relaxed by your sides. A rest period of 20 s occurred between each 40 s trial. Sufficient trials were performed to enable three valid sets of data to be recorded. A test was invalidated if the participant: (1) moved their foot position during the test; (2) changed their arm starting position or (3) opened their eyes during an eyes-closed task. Shod assessment Study shoes were then put on. The shod assessment protocol was conducted as described in the Barefoot assessment section. Shod assessment protocol was conducted by AS; shoes were concealed from CSM to maintain assessor blinding in the main trial.18 Outcome measures The following postural stability primary outcomes were assessed at baseline, 6 weeks and 6 months: (1) root mean squared error and (2) velocity of the CoP in the anteroposterior direction (CoPRMSE AP and CoPVEL AP, respectively). Equations, demonstrating how CoP data were calculated, are presented in online supplementary appendix 2.

postural stability primary outcomes were assessed at baseline, 6 weeks and 6 months: (1) root mean squared error and (2) velocity of the CoP in the anteroposterior direction (CoPRMSE AP and CoPVEL AP, respectively). Equations, demonstrating how CoP data were calculated, are presented in online supplementary appendix 2. 10.1136/bmjsem-2016-000170.supp2supplementary appendix Sample size A sample size calculation was not conducted due to the lack of reported data of minimal clinically important difference for the primary outcome measures (CoP parameters). Data extraction Industry-standard motion capture files (.c3d) containing force data were extracted. Force plate data were filtered with a low-pass (10 Hz) Butterworth filter. CoP parameters (CoPRMSE AP and COPVEL AP) were calculated using a proprietary program writer Visual Basic for Application (Microsoft Excel, Reading, UK).

raction Industry-standard motion capture files (.c3d) containing force data were extracted. Force plate data were filtered with a low-pass (10 Hz) Butterworth filter. CoP parameters (CoPRMSE AP and COPVEL AP) were calculated using a proprietary program writer Visual Basic for Application (Microsoft Excel, Reading, UK). Data analysis The primary analysis was by intention-to-treat, including all eligible randomised participants who provided follow-up data. Two-way mixed model (between–within) analysis of variances were conducted with one within-subject (assessment time points) and one between-group factor (footwear type) to compare the influence of footwear type over time and one within-subject (standing condition) and one between-group factor (footwear type) to compare baseline data between groups. Analysis of variance used data from participants with full data sets (rocker-sole group n=13, flat-sole group n=7 for baseline comparisons and immediate effect of footwear; rocker-sole group n=11, flat-sole group n=5 for long-term follow-up). Macuhly test of sphericity assumption and Levene's test of equality of variances assumption were considered for within-subject and between-subject effects, respectively. The α level for determining statistical significance was set at 0.05. Data were analysed using IBM SPSS V.20.0.0 (IBM, Armonk, New York, USA). Results are presented as means (SDs) unless otherwise stated.

t of equality of variances assumption were considered for within-subject and between-subject effects, respectively. The α level for determining statistical significance was set at 0.05. Data were analysed using IBM SPSS V.20.0.0 (IBM, Armonk, New York, USA). Results are presented as means (SDs) unless otherwise stated. Results Twenty participants (from 38 who showed interest in the study) were recruited into the study from June 2010 to November 2010 (the final 6 months of main study recruitment18). Seven participants had been prerandomised to receive the flat-sole and 13 to receive the rocker-sole shoe.18 There were no differences between the groups in demographic or outcome measures (table 1) at baseline. Table 1 Baseline characteristics of the study participants Flat-sole group (n=7) Rocker-sole group (n=13) p Value Gender Male 3 (42.9%)* 6 (46.2%)* 0.89† Female 4 (57.1%)* 7 (53.8%)* Age (years) 37.9 (13.0) 42.6 (12.5) 0.43 Weight (kg) 82.4 (22.0) 70.3 (11.3) 0.12 Height (cm) 173.8 (7.3) 173.5 (9.5) 0.95 Roland Morris Disability Questionnaire (0–24; 0=best) 7.9 (1.8) 5.7 (3.3) 0.13 Numerical rating score for pain (0–10; 0=best) 6.3 (1.5) 5.7 (1.7) 0.48 Summary measures represent means (SD). *Summary measures represent numbers (percentages). †Data analysed with independent t-test or χ2 test. Baseline barefoot CoP parameters are presented in table 2. There were no differences between the groups in CoPRMSE AP, CoPVEL AP for any of the four standing conditions (F(3,51)=0.31, p=0.82, η2=0.02; F(1.76,29.94)=0.15, p=0.83, η2=0.01, respectively).

*Summary measures represent numbers (percentages). †Data analysed with independent t-test or χ2 test. Baseline barefoot CoP parameters are presented in table 2. There were no differences between the groups in CoPRMSE AP, CoPVEL AP for any of the four standing conditions (F(3,51)=0.31, p=0.82, η2=0.02; F(1.76,29.94)=0.15, p=0.83, η2=0.01, respectively). Table 2 Barefoot anteroposterior centre of pressure and postural strategy parameters at baseline Standing condition Group CoPRMSE AP (mm) CoPVEL AP (mm/s) Eyes open firm surface Flat-sole shoe 4.80 (2.47) 7.33 (2.01) Rocker-sole shoe 4.39 (1.84) 7.19 (1.13) Eyes closed firm surface Flat-sole shoe 4.98 (1.87) 7.54 (1.44) Rocker-sole shoe 4.05 (1.26) 7.50 (1.12) Eyes open compliant surface Flat-sole shoe 10.06 (2.87) 11.89 (1.18) Rocker-sole shoe 8.63 (2.61) 12.67 (4.38) Eyes closed Compliant surface Flat-sole shoe 11.06 (2.86) 17.94 (4.32) Rocker-sole shoe 10.62 (2.66) 17.75 (4.12) Summary measures represent means (SD). AP, anteroposterior; RMSE, root mean squared error; VEL, velocity. Participant attrition and retention during the study are presented in figure 3. At 6 months, 16 (80%) participants were reassessed. Figure 3 Flow of participants through trial.

Standing condition Group CoPRMSE AP (mm) CoPVEL AP (mm/s) Eyes open firm surface Flat-sole shoe 4.80 (2.47) 7.33 (2.01) Rocker-sole shoe 4.39 (1.84) 7.19 (1.13) Eyes closed firm surface Flat-sole shoe 4.98 (1.87) 7.54 (1.44) Rocker-sole shoe 4.05 (1.26) 7.50 (1.12) Eyes open compliant surface Flat-sole shoe 10.06 (2.87) 11.89 (1.18) Rocker-sole shoe 8.63 (2.61) 12.67 (4.38) Eyes closed Compliant surface Flat-sole shoe 11.06 (2.86) 17.94 (4.32) Rocker-sole shoe 10.62 (2.66) 17.75 (4.12) Summary measures represent means (SD). AP, anteroposterior; RMSE, root mean squared error; VEL, velocity. Participant attrition and retention during the study are presented in figure 3. At 6 months, 16 (80%) participants were reassessed. Figure 3 Flow of participants through trial. Comparison of CoP parameters when standing barefoot and standing shod Standing in rocker-sole shoes, with eyes-open on firm surface, resulted in a mean increase in CoPRMSE AP of 6.41 mm (t(12)=7.77, p<0.01) and CoPVEL AP of 4.10 mm/s (t(12)=7.14, p<0.01) when compared with standing barefoot (table 3). There was no difference in CoPRMSE AP or CoPVEL AP when standing in flat-sole shoes compared with barefoot (table 3). Table 3 Sagittal plane centre of pressure parameters during barefoot and shod standing, with eyes open on firm surface

Comparison of CoP parameters when standing barefoot and standing shod Standing in rocker-sole shoes, with eyes-open on firm surface, resulted in a mean increase in CoPRMSE AP of 6.41 mm (t(12)=7.77, p<0.01) and CoPVEL AP of 4.10 mm/s (t(12)=7.14, p<0.01) when compared with standing barefoot (table 3). There was no difference in CoPRMSE AP or CoPVEL AP when standing in flat-sole shoes compared with barefoot (table 3). Table 3 Sagittal plane centre of pressure parameters during barefoot and shod standing, with eyes open on firm surface Flat-sole shoe group (n=7) Rocker-sole shoe group (n=13) CoPRMSE AP (mm) CoPVEL AP (mm/s) CoPRMSE AP (mm) CoPVEL AP (mm/s) Barefoot 4.78 (2.26) 7.03 (2.00) 4.39 (1.84) 7.19 (1.13) Shod 5.61 (2.33) 7.11 (1.27) 10.79 (3.01) 11.28 (1.93) Difference between means 0.84 (2.03) 0.07 (1.20) 6.41 (2.97)* 4.10 (2.07)* Summary measures represent means (SD) or percentages where indicated (%). *Significant difference within groups between barefoot and shoe conditions (p<0.01). AP, anteroposterior; RMSE, root mean squared error; VEL, velocity.

Flat-sole shoe group (n=7) Rocker-sole shoe group (n=13) CoPRMSE AP (mm) CoPVEL AP (mm/s) CoPRMSE AP (mm) CoPVEL AP (mm/s) Barefoot 4.78 (2.26) 7.03 (2.00) 4.39 (1.84) 7.19 (1.13) Shod 5.61 (2.33) 7.11 (1.27) 10.79 (3.01) 11.28 (1.93) Difference between means 0.84 (2.03) 0.07 (1.20) 6.41 (2.97)* 4.10 (2.07)* Summary measures represent means (SD) or percentages where indicated (%). *Significant difference within groups between barefoot and shoe conditions (p<0.01). AP, anteroposterior; RMSE, root mean squared error; VEL, velocity. Influence of long-term shoe wear on barefoot sagittal plane CoP parameters Neither the rocker-sole nor the flat-sole group demonstrated change in CoPRMSE AP or CoPVEL AP when assessed barefoot during the most challenging standing condition (eyes-closed, compliant ground), at any follow-up point (rocker-sole group F(2,20)=2.28, p=0.13, η2=0.19 and F(2,20)=2.69, p=0.09, η2=0.21, respectively; flat-sole group F(2,8)=1.89, p=0.21, η2=0.32 and F(2,8)=0.27, p=0.70, η2=0.06, respectively) (table 4). Furthermore, there were no differences between-groups in CoPRMSE AP or CoPVEL AP at any follow-up point during the most challenging standing condition (F(2,28)=1.80, p=0.19, η2=0.11 and F(2,28)=0.28, p=0.76, η2=0.02). Table 4 Change in barefoot centre of pressure parameters during standing, eyes closed on compliant surface at reassessment points

Influence of long-term shoe wear on barefoot sagittal plane CoP parameters Neither the rocker-sole nor the flat-sole group demonstrated change in CoPRMSE AP or CoPVEL AP when assessed barefoot during the most challenging standing condition (eyes-closed, compliant ground), at any follow-up point (rocker-sole group F(2,20)=2.28, p=0.13, η2=0.19 and F(2,20)=2.69, p=0.09, η2=0.21, respectively; flat-sole group F(2,8)=1.89, p=0.21, η2=0.32 and F(2,8)=0.27, p=0.70, η2=0.06, respectively) (table 4). Furthermore, there were no differences between-groups in CoPRMSE AP or CoPVEL AP at any follow-up point during the most challenging standing condition (F(2,28)=1.80, p=0.19, η2=0.11 and F(2,28)=0.28, p=0.76, η2=0.02). Table 4 Change in barefoot centre of pressure parameters during standing, eyes closed on compliant surface at reassessment points Assessment Centre of pressure parameter Baseline 6 weeks 6 months p Value Flat-sole shoe group (n=5) CoPRMSE AP (mm) 10.80 (2.85) 10.70 (3.40) 9.29 (1.95) 0.21 CoPVEL AP (mm/s) 21.61 (3.48) 20.66 (4.82) 20.19 (5.85) 0.70 Rocker-sole shoe group (n=11) CoPRMSE AP (mm) 10.43 (2.85) 9.35 (2.62) 9.75 (3.08) 0.13 CoPVEL AP (mm/s) 17.85 (4.59) 15.28 (3.64) 15.77 (4.26) 0.09 Summary measures represent means (SD). No difference in COPRMSE AP or CoPVEL AP was found for the three less challenging standing conditions assessed within-shoe or between-shoe groups at any follow-up point.

Assessment Centre of pressure parameter Baseline 6 weeks 6 months p Value Flat-sole shoe group (n=5) CoPRMSE AP (mm) 10.80 (2.85) 10.70 (3.40) 9.29 (1.95) 0.21 CoPVEL AP (mm/s) 21.61 (3.48) 20.66 (4.82) 20.19 (5.85) 0.70 Rocker-sole shoe group (n=11) CoPRMSE AP (mm) 10.43 (2.85) 9.35 (2.62) 9.75 (3.08) 0.13 CoPVEL AP (mm/s) 17.85 (4.59) 15.28 (3.64) 15.77 (4.26) 0.09 Summary measures represent means (SD). No difference in COPRMSE AP or CoPVEL AP was found for the three less challenging standing conditions assessed within-shoe or between-shoe groups at any follow-up point. Influence of long-term shoe wear on postural control assessed when shod When standing in study shoes, with eyes-open on firm surface, no significant differences were observed in CoPRMSE AP or CoPVEL AP for either shoe group at any reassessment point (rocker-sole group: F(2,20)=1.35, p=0.28, η2=0.12, and F(2,20)=1.84, p=0.19, η2=0.15, respectively; flat-sole group: F(2,8)=0.74, p=0.51, η2=0.16, F(2,8)=0.63, p=0.56, η2=0.14). Furthermore, while wearing study shoes, there were no differences between-groups in change in CoPRMSE AP or CoPVEL AP at any reassessment point (F(2,28)=1.18, p=0.32, η2=0.08, and F(2,28)=0.37, p=0.70, η2=0.03, respectively) (table 5). Table 5 Change over time in anteroposterior centre of pressure parameters during shod standing, eyes open on firm surface

Influence of long-term shoe wear on postural control assessed when shod When standing in study shoes, with eyes-open on firm surface, no significant differences were observed in CoPRMSE AP or CoPVEL AP for either shoe group at any reassessment point (rocker-sole group: F(2,20)=1.35, p=0.28, η2=0.12, and F(2,20)=1.84, p=0.19, η2=0.15, respectively; flat-sole group: F(2,8)=0.74, p=0.51, η2=0.16, F(2,8)=0.63, p=0.56, η2=0.14). Furthermore, while wearing study shoes, there were no differences between-groups in change in CoPRMSE AP or CoPVEL AP at any reassessment point (F(2,28)=1.18, p=0.32, η2=0.08, and F(2,28)=0.37, p=0.70, η2=0.03, respectively) (table 5). Table 5 Change over time in anteroposterior centre of pressure parameters during shod standing, eyes open on firm surface Assessment Centre of pressure parameter Baseline 6 weeks 6 months p Value Flat-sole shoe group (n=5) CoPRMSE AP (mm) 5.20 (1.52) 6.03 (2.95) 5.29 (2.22) 0.51 CoPVEL AP (mm/s) 7.28 (2.04) 6.22 (1.16) 6.29 (1.97) 0.56 Rocker-sole shoe group (n=11) CoPRMSE AP (mm) 10.17 (2.84) 9.54 (2.79) 11.07 (3.89) 0.28 CoPVEL AP (mm/s) 9.39 (2.24) 9.10 (3.25) 8.24 (1.81) 0.19 Summary measures represent means (SD).

le shoe group (n=5) CoPRMSE AP (mm) 5.20 (1.52) 6.03 (2.95) 5.29 (2.22) 0.51 CoPVEL AP (mm/s) 7.28 (2.04) 6.22 (1.16) 6.29 (1.97) 0.56 Rocker-sole shoe group (n=11) CoPRMSE AP (mm) 10.17 (2.84) 9.54 (2.79) 11.07 (3.89) 0.28 CoPVEL AP (mm/s) 9.39 (2.24) 9.10 (3.25) 8.24 (1.81) 0.19 Summary measures represent means (SD). Discussion This study investigated the influence of rocker-sole shoes on postural stability in people with CLBP. The results were concordant with Hypothesis 1; that is, that the wearing of rocker-sole shoes provides a less stable surface to stand on than flat-sole shoes. However, the results do not support Hypothesis 2; there were no differences in barefoot CoP parameters within-groups or between-groups during barefoot trials at 6 weeks or 6 months, compared with baseline, for any standing condition. Furthermore, there were no changes from baseline in CoP parameters in the rocker-sole group when shod at 6 weeks and 6 months. These findings suggest that adaptation of the postural control system did not occur following long-term wear of rocker-sole shoes. Alternatively, the outcomes assessed were not appropriate to detect any potential training effect offered by the rocker-shoes.

rs in the rocker-sole group when shod at 6 weeks and 6 months. These findings suggest that adaptation of the postural control system did not occur following long-term wear of rocker-sole shoes. Alternatively, the outcomes assessed were not appropriate to detect any potential training effect offered by the rocker-shoes. Anteroposterior CoP parameters The current study demonstrated similar barefoot baseline CoP parameters between shoe groups. When compared with the findings of other studies investigating CLBP with the same outcome measures under similar protocols, this study demonstrated increased postural stability during less challenging standing conditions,6 11 21 and reduced postural stability during more challenging standing conditions.11 21 22 These differences may be due to a number of methodological and demographic differences reported to influence outcome, namely: number of trials;10 trial durations;10 participant age;23–26 body weight;27 28 body height27 28 and gender.25 However, the consistent increase in CoP parameters from stable to more challenging standing conditions in the current study concurs with other studies.7 11

hic differences reported to influence outcome, namely: number of trials;10 trial durations;10 participant age;23–26 body weight;27 28 body height27 28 and gender.25 However, the consistent increase in CoP parameters from stable to more challenging standing conditions in the current study concurs with other studies.7 11 A reduction in a CoP parameter is interpreted as an improvement in postural stability.10 It was hypothesised that due to the increased proprioceptive input from wearing rocker shoes,12 a greater reduction in barefoot and shod postural excursion may occur at reassessment in the rocker-sole compared with the flat-sole group. However, neither group demonstrated a significant change in CoP parameters at any follow-up compared with baseline when barefoot or shod. This lack of change suggests that the rocker-sole footwear either (1) provided an additional postural challenge; however, the type of challenge did not result in long-term improvements in sensorimotor function, (2) provided an appropriate postural challenge but ‘dosage’ was insufficient for a training effect to occur or (3) influenced proprioceptive deficits; however, improvements were not detected.

additional postural challenge; however, the type of challenge did not result in long-term improvements in sensorimotor function, (2) provided an appropriate postural challenge but ‘dosage’ was insufficient for a training effect to occur or (3) influenced proprioceptive deficits; however, improvements were not detected. The first explanation, suggesting that the increased postural challenge from rocker-sole shoes does not influence long-term improvements in sensorimotor function compared with wearing flat-sole shoes, concurs with the findings of other studies.29 30 Nigg et al29 investigated the influence of rocker-sole footwear on balance in golfers with LBP and in people with knee osteoarthritis.30 In support of the current study findings, Nigg et al29 concluded that no differences in balance performance were detected between the intervention (rocker-sole group) or control group (normal shoes) at 6 and 12 weeks.29 The current study adds to Nigg et al's conclusions by demonstrating that longer term use of rocker-sole shoes (6 months) has no further influence on postural stability.

t no differences in balance performance were detected between the intervention (rocker-sole group) or control group (normal shoes) at 6 and 12 weeks.29 The current study adds to Nigg et al's conclusions by demonstrating that longer term use of rocker-sole shoes (6 months) has no further influence on postural stability. The second explanation suggests a greater postural challenge may have resulted in a measured training effect. When compared with standing barefoot, the rocker shoes demonstrated a 57–146% increase in the CoP parameters assessed. Introducing additional postural challenge in an attempt to increase the CoP parameters further may not only be unsafe or impractical in a CLBP population, but may also, in the absence of evidence to support a relationship between increased postural challenge and change in CoP parameters or clinical change, be inappropriate. The third explanation suggests that the null hypothesis was incorrectly accepted and study conclusions are incorrect. This may have been due to an underpowered sample, poor reliability of the outcome variables or an insensitivity to detect genuine changes in postural control. The reliability of the outcome variables may be improved by increasing the duration and number of trials. However, of the numerous CoP parameters regularly reported in research assessing postural stability, the two parameters chosen in the current study have been reported as highly reliable.10

nges in postural control. The reliability of the outcome variables may be improved by increasing the duration and number of trials. However, of the numerous CoP parameters regularly reported in research assessing postural stability, the two parameters chosen in the current study have been reported as highly reliable.10 Although changes in CoP parameters have been suggested as appropriate outcome measures to detect clinical change,31 to the authors knowledge, measurements of the SE of CoP parameters, during challenging standing conditions, have yet to be reported in the literature for people with CLBP. The differences in postural instability outcomes during challenging standing conditions for both shoes types in the current study are less than the reported SEs of the same CoP parameters assessed in reliability studies investigating elderly participants (who also demonstrate poor postural stability).32 Changes in CoP parameters following an intervention may be too small to reliably determine whether change in postural stability has occurred.

udy are less than the reported SEs of the same CoP parameters assessed in reliability studies investigating elderly participants (who also demonstrate poor postural stability).32 Changes in CoP parameters following an intervention may be too small to reliably determine whether change in postural stability has occurred. The clinical study investigating the effects of rocker-sole footwear on CLBP,18 from which the current participants were recruited, demonstrated clinically important statistically significant reductions in disability and pain (in rocker-sole and flat-sole shoe groups) at follow-up; however, the current study demonstrates no change in postural parameters. This study and the findings of Kuukkanen and Malkia33 (who in the presence of improvement in function in patients with LBP, found no improvement in postural stability at 6 months following an exercise intervention) suggest that CoP parameters may be insensitive to real changes in postural control or that there may be no significant changes in control. If the latter, the use of any mechanical indices as outcome measures would be inappropriate; if the former, alternative mechanical outcome measures need to be developed and tested.

suggest that CoP parameters may be insensitive to real changes in postural control or that there may be no significant changes in control. If the latter, the use of any mechanical indices as outcome measures would be inappropriate; if the former, alternative mechanical outcome measures need to be developed and tested. Limitations A systematic review investigating acceptable reliability for CoP parameters in asymptomatic individuals, published subsequently to the start of the current study, recommended a minimum trial duration of 90 s—a greater duration than that applied in this clinical trial.34 However, in the current study, prolonged standing may have aggravated symptoms, and negatively influenced attrition rates. The authors recognise the small sample size of this study may have resulted in a type II error. Although the study sample is small (n=20), when compared with participants in the clinical study18 from which study participants were recruited (n=115), there were similar reductions in pain and disability at 6-week and 6-month follow-up (disability: rocker-sole group F(2,106)=0.20, p=0.82, η2=0.001; flat-sole group, F(1.53,73.4)=0.24, p=0.73, η2=0.01; pain: rocker-sole group, F(1.70,90.10)=0.01, p=0.99, η2 <0.01; flat-sole group, F(2,96)=1.04, p=0.36, η2=0.02), suggesting that this subgroup was a representative sample of a larger CLBP population, hence reducing the likelihood of a type II error. It is unclear what effect either shoe type may have on CoP parameters in people with more severe CLBP, greater postural instability at baseline or if worn for >6 months.

The authors recognise the small sample size of this study may have resulted in a type II error. Although the study sample is small (n=20), when compared with participants in the clinical study18 from which study participants were recruited (n=115), there were similar reductions in pain and disability at 6-week and 6-month follow-up (disability: rocker-sole group F(2,106)=0.20, p=0.82, η2=0.001; flat-sole group, F(1.53,73.4)=0.24, p=0.73, η2=0.01; pain: rocker-sole group, F(1.70,90.10)=0.01, p=0.99, η2 <0.01; flat-sole group, F(2,96)=1.04, p=0.36, η2=0.02), suggesting that this subgroup was a representative sample of a larger CLBP population, hence reducing the likelihood of a type II error. It is unclear what effect either shoe type may have on CoP parameters in people with more severe CLBP, greater postural instability at baseline or if worn for >6 months. Conclusions This is the first randomised trial with long-term follow-up comparing the influence of rocker-sole and flat-sole shoes on standing CoP parameters in a CLBP population. Long-term use of rocker-sole or flat-sole shoes in addition to attendance to a 4-week exercise group does not appear to influence barefoot postural control, as determined by CoP parameters, during standing in people with CLBP.

influence of rocker-sole and flat-sole shoes on standing CoP parameters in a CLBP population. Long-term use of rocker-sole or flat-sole shoes in addition to attendance to a 4-week exercise group does not appear to influence barefoot postural control, as determined by CoP parameters, during standing in people with CLBP. The authors would like to thank all participants for their contributions to this study. The authors would also like to thank Tanya Forster, Andrew Lewis and Jonathan Noble for their assistance during the data collection and analysis process. The authors thank all physiotherapy departments who participated in this trial, namely: Balance Performance Physiotherapy, Clapham, London, UK, SW4; Chelsea and Westminster Hospital, Chelsea, London, UK, SW10; Queen Mary's Hospital, Roehampton, UK, SW14; and St George's Hospital, Tooting, UK SW18. Twitter: Follow Catharine MacRae at @Sian_MacRae Contributors: CSM was the primary investigator, involved in all aspects of the study, including methodology, data collection, analysis and interpretation of data, and was the primary author of the article. All authors contributed to methodology, data interpretation, and editing of the manuscript for publication. All authors approved the final revision of the submitted manuscript. In addition, JSL received grant funding for the study, AS contributed to data collection. Funding: The clinical study from which participants in the current study were recruited was funded by a Masai GB Limited project grant.

Contributors: CSM was the primary investigator, involved in all aspects of the study, including methodology, data collection, analysis and interpretation of data, and was the primary author of the article. All authors contributed to methodology, data interpretation, and editing of the manuscript for publication. All authors approved the final revision of the submitted manuscript. In addition, JSL received grant funding for the study, AS contributed to data collection. Funding: The clinical study from which participants in the current study were recruited was funded by a Masai GB Limited project grant. Disclaimer: The corresponding author had full access to all data in the study and had final responsibility for the decision to submit for publication. Patient consent: Obtained. Ethics approval: Outer North London Research Ethics Committee. Provenance and peer review: Not commissioned; externally peer reviewed.

What are the new findings? Evaluation and reporting of recruitment processes in sports interventions is scarce resulting in a lack of evidence for suitable engagement mechanisms, particularly in hard-to-reach groups. Active-only recruitment approaches achieved more representative samples for their target population than passive or combined techniques. It is of concern, however, that these approaches may be more vulnerable to limited participant engagement, thus requiring additional components such as motivational interviewing to encourage participation. How might it impact on clinical practice in the near future? For future sporting interventions, techniques involving active recruitment may be particularly appropriate to recruit hard-to-reach participants or to better achieve target participant demographic composition. Introduction Physical inactivity is a global public health problem and the fourth leading cause of global mortality, resulting in over 5.3 million deaths a year worldwide.1 It costs £0.9 billion to the National Health Service in the UK alone.2 Sport presents a possible means of promoting physical activity (PA) and health. However, little is known about how best to engage inactive individuals in sport to increase PA.3

lobal mortality, resulting in over 5.3 million deaths a year worldwide.1 It costs £0.9 billion to the National Health Service in the UK alone.2 Sport presents a possible means of promoting physical activity (PA) and health. However, little is known about how best to engage inactive individuals in sport to increase PA.3 It is recognised that PA and sports participation are unequally distributed across society with gender, age, disability, education and socioeconomic status (SES) as determinants.4 Certain subgroups of the population are therefore more likely to be inactive and stand to gain significant health benefits from increasing their PA.5 These groups are frequently considered ‘hard to reach’, and as such, it is important to understand how they may be successfully engaged in health-promoting interventions.3 Nevertheless, historically there has been a significant recruitment bias in PA interventions that have predominantly recruited white, middle-class, middle-aged women, resulting in under-represented male, socioeconomically disadvantaged and minority ethnic populations.6 In light of this, sports interventions are becoming increasingly targeted in attempts to reduce health inequalities and engage underserved populations.7

predominantly recruited white, middle-class, middle-aged women, resulting in under-represented male, socioeconomically disadvantaged and minority ethnic populations.6 In light of this, sports interventions are becoming increasingly targeted in attempts to reduce health inequalities and engage underserved populations.7 In general, academic journals have prioritised the publication of intervention findings above the evaluation and reporting of recruitment processes and outcomes.8 9 This is limiting because, independent of intervention efficacy, the viability of a programme will be determined by its ability to recruit sufficient numbers of eligible participants.10 This limitation is further exacerbated by the fact that there has been a general lack of evaluation of sports programmes, thus limiting the evidence base on how to engage often hard-to-reach inactive populations in sport.3 11 We argue that a better understanding of recruitment procedures and their effectiveness is needed to inform those wishing to successfully replicate or adapt interventions, particularly when informing policy or practice.12 This review has therefore been undertaken to provide evidence for the role of planning, implementation and reporting of recruitment processes for sports interventions promoting positive PA behaviour change.

those wishing to successfully replicate or adapt interventions, particularly when informing policy or practice.12 This review has therefore been undertaken to provide evidence for the role of planning, implementation and reporting of recruitment processes for sports interventions promoting positive PA behaviour change. Method This review systematically identifies and evaluates the effectiveness of recruitment techniques used in interventions aimed at increasing PA using sport. Although the definition of recruitment varies between studies, we have defined ‘recruitment’ to be those who enrol on an intervention independent of whether they participate. We further define recruitment effectiveness as engaging sufficient numbers of target populations to (1) register for the intervention, (2) participate, (3) complete the intervention or any follow-up, and (4) achieve long-term positive PA behaviour change. Consequently, for the purpose of this review, retention is also considered a component of recruitment effectiveness. From a methodological perspective, the two primary challenges were the identification of reports on PA interventions using sport and the subsequent identification of the recruitment methods used. Hence, the adopted methodology was designed to particularly address these issues. Inclusion criteria and analysis methodology were previously specified and documented in a protocol registered as CRD42015015815 (available at http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42015015815).

Method This review systematically identifies and evaluates the effectiveness of recruitment techniques used in interventions aimed at increasing PA using sport. Although the definition of recruitment varies between studies, we have defined ‘recruitment’ to be those who enrol on an intervention independent of whether they participate. We further define recruitment effectiveness as engaging sufficient numbers of target populations to (1) register for the intervention, (2) participate, (3) complete the intervention or any follow-up, and (4) achieve long-term positive PA behaviour change. Consequently, for the purpose of this review, retention is also considered a component of recruitment effectiveness. From a methodological perspective, the two primary challenges were the identification of reports on PA interventions using sport and the subsequent identification of the recruitment methods used. Hence, the adopted methodology was designed to particularly address these issues. Inclusion criteria and analysis methodology were previously specified and documented in a protocol registered as CRD42015015815 (available at http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42015015815). Data sources To identify potential studies, searches were conducted using electronic databases, clinical trials registers, grey literature and snowballing from reference lists. This involved a systematic search of the following electronic databases: CINAHL, MEDLINE, EMBASE, PsycINFO and SPORTDiscus. Then the two primary organisations in England contributing to the delivery of sports programmes for PA and health (Sport England, the national body for sport in England, and UKActive, a not-for-profit health body for the PA sector in the UK) were contacted to identify additional grey literature. Titles and abstracts of identified material were checked against inclusion and exclusion criteria for suitability. Full articles were then acquired and assessed for inclusion. Snowballing was employed whereby, following inclusion, the reference lists of papers were searched and further articles considered for inclusion. The search was completed in November 2015.

were checked against inclusion and exclusion criteria for suitability. Full articles were then acquired and assessed for inclusion. Snowballing was employed whereby, following inclusion, the reference lists of papers were searched and further articles considered for inclusion. The search was completed in November 2015. The search terms used were developed using those of earlier systematic reviews on the evaluation of participation outcomes of sporting interventions13 and recruitment into walking interventions10 combined with an extensive list of sporting activities recognised by Sport England.14 Details of the search syntax used for electronic databases are provided in the online supplementary appendix. For clinical trials registers, ‘sport’ was the only search term within the title, using interventional studies in adults or seniors as limiters. The details of the full inclusion and exclusion criteria are provided in  online supplementary table 1. Pragmatic considerations meant the search was limited to papers published in English. All programmes recruiting adults into interventions involving sport and reporting PA or participation outcomes were included. 10.1136/bmjsem-2017-000231.supp1Supplementary data 10.1136/bmjsem-2017-000231.supp4Supplementary data

The search terms used were developed using those of earlier systematic reviews on the evaluation of participation outcomes of sporting interventions13 and recruitment into walking interventions10 combined with an extensive list of sporting activities recognised by Sport England.14 Details of the search syntax used for electronic databases are provided in the online supplementary appendix. For clinical trials registers, ‘sport’ was the only search term within the title, using interventional studies in adults or seniors as limiters. The details of the full inclusion and exclusion criteria are provided in  online supplementary table 1. Pragmatic considerations meant the search was limited to papers published in English. All programmes recruiting adults into interventions involving sport and reporting PA or participation outcomes were included. 10.1136/bmjsem-2017-000231.supp1Supplementary data 10.1136/bmjsem-2017-000231.supp4Supplementary data Study selection The first reviewer conducted a review of all identified studies to exclude duplicates and studies that clearly did not meet the inclusion criteria, for example, interventions looking at elite sport performance-related outcomes. To check the inclusion process, 15% of review articles were randomly selected at the abstract screening stage and screened by the second reviewer, and all papers were found to have been correctly excluded. Where duplicate studies presented, the journal article reporting the most recruitment data was analysed and other articles excluded unless multiple papers were found to report on distinctly different aspects of the intervention and still meet the inclusion criteria.

ll papers were found to have been correctly excluded. Where duplicate studies presented, the journal article reporting the most recruitment data was analysed and other articles excluded unless multiple papers were found to report on distinctly different aspects of the intervention and still meet the inclusion criteria. Data extraction A data extraction table was developed by both authors guided by the protocols of other reviews looking at recruitment into health interventions for guidance.15–17 The resultant table summarised the characteristics of the study, population, sport intervention, recruitment, retention and outcomes. Corresponding data were then extracted from all included papers by the first and second reviewers and were transcribed into the table using Microsoft Excel. Data synthesis We anticipated considerable heterogeneity between identified interventions and their recruitment approaches and therefore planned to employ a narrative synthesis of results. Extracted information was also used to synthesise additional data relating to the study quality, efficiency and effectiveness following the methodology of a similar review.10

heterogeneity between identified interventions and their recruitment approaches and therefore planned to employ a narrative synthesis of results. Extracted information was also used to synthesise additional data relating to the study quality, efficiency and effectiveness following the methodology of a similar review.10 Assessment of quality of recruitment reporting Due to the specific focus of this review, we did not attempt to assess the general quality of studies but rather focused on the quality of reporting of recruitment. This was assessed using criteria previously developed by Foster et al 10 in their review of recruitment into walking intervention studies. Two reviewers independently assessed the quality of recruitment reporting in the studies regarding where the population was recruited, who conducted the recruitment, the time spent planning and preparing the recruitment, the time spent conducting the recruitment and the target population to be recruited. Each criterion was given a value of zero (absent or inadequately described) or one (explicitly described and present). Using the nomenclature of Foster et al,10 studies scoring three or below overall were considered ‘low quality’ while those that scored between four and five were considered ‘high quality’.

e recruited. Each criterion was given a value of zero (absent or inadequately described) or one (explicitly described and present). Using the nomenclature of Foster et al,10 studies scoring three or below overall were considered ‘low quality’ while those that scored between four and five were considered ‘high quality’. Assessment of efficiency We also adopted the methodology of Foster et al 10 to evaluate the efficiency of recruitment processes within included studies. This involved calculating the recruitment rates and efficiency ratios, where possible, for each included study. The following values were sought: the total number of potential participants who could be eligible for study (‘pool’), potential participants invited to participate in the study (‘invited’), potential participants who responded to the invitation (‘responded’) and participants who were assessed as eligible to participate and began the programme (‘started’). Where possible, ratios were calculated for each stage, for example, by dividing the number of participants who ‘started’ the study by the total ‘invited’ the proportion taking up the intervention. Furthermore, a weekly rate of recruitment was calculated for those studies also providing recruitment duration.

‘started’). Where possible, ratios were calculated for each stage, for example, by dividing the number of participants who ‘started’ the study by the total ‘invited’ the proportion taking up the intervention. Furthermore, a weekly rate of recruitment was calculated for those studies also providing recruitment duration. Results Study characteristics Twenty-three papers representing 22 interventions met our inclusion criteria. Figure 1 reports the flow of studies through the review process. Characteristics of included studies are presented in online supplementary table 2, ranked by quality score. Each included paper is referenced in the results and discussion sections in superscript using their reference citation. Full references for included papers are therefore listed in the bibliography. Studies were located in the UK (n=10),18–27 USA (n=5),28–32 Canada (n=3),33–35 Norway (n=2),36 37 Italy (n=1),38 Switzerland (n=1)39 and South America (n=1).40 Nearly all the studies were quantitative experimental studies in design, with 9 randomised controlled trials,18 27–30 32 33 35 40 2 non-randomised controlled trials25 37 and 11 before-and-after studies.19 21–24 26 31 34 36 38 39 Of these, four were mixed methods and incorporated some qualitative design.19 21 28 34 We found only one qualitative study reporting on recruitment approaches.20 Figure 1 Review flow chart.

Results Study characteristics Twenty-three papers representing 22 interventions met our inclusion criteria. Figure 1 reports the flow of studies through the review process. Characteristics of included studies are presented in online supplementary table 2, ranked by quality score. Each included paper is referenced in the results and discussion sections in superscript using their reference citation. Full references for included papers are therefore listed in the bibliography. Studies were located in the UK (n=10),18–27 USA (n=5),28–32 Canada (n=3),33–35 Norway (n=2),36 37 Italy (n=1),38 Switzerland (n=1)39 and South America (n=1).40 Nearly all the studies were quantitative experimental studies in design, with 9 randomised controlled trials,18 27–30 32 33 35 40 2 non-randomised controlled trials25 37 and 11 before-and-after studies.19 21–24 26 31 34 36 38 39 Of these, four were mixed methods and incorporated some qualitative design.19 21 28 34 We found only one qualitative study reporting on recruitment approaches.20 Figure 1 Review flow chart. Intervention characteristics There were a wide variety of intervention designs and sports demonstrated across the included studies (see online supplementary table 3). Twelve interventions involved multiple physical activities, all or some of which involved participating in sport,18–27 36 37 while 11 offered an intervention using only one sporting activity.28–35 38–40 The main activities reportedly used, in isolation or combination, were dance (n=8),23 25 29 32 34 37 38 40 football (n=7),18–21 23 26 37 exercise classes (n=7),19 24 25 27 31 37 38 running/jogging (n=6),18 23 24 33 38 39 swimming (n=5)23 24 36–38 and yoga (n=2).28 30 A range of settings were used to deliver interventions, including professional football stadia and facilities,18–21 26 leisure or sports facilities22 25 28 31 33 34 and non-sporting community sites such as schools and churches.27 29 36 38 The average reported intervention duration was 23.7 weeks (SD ±27.7 weeks, range 8–104 weeks).

ttings were used to deliver interventions, including professional football stadia and facilities,18–21 26 leisure or sports facilities22 25 28 31 33 34 and non-sporting community sites such as schools and churches.27 29 36 38 The average reported intervention duration was 23.7 weeks (SD ±27.7 weeks, range 8–104 weeks). Characteristics of the participants Sample sizes (N started) of the studies ranged from 15 to 160 018 participants. Twenty studies reported participant ages18–22 25 27–40 with a mean age of 51.3 years (SD ±6.3 years) and range from 18 to 70 years (see online supplementary table 4). Six out of 22 studies that reported gender focused on recruiting female-only participants,24 29 32 34 35 39 and five studies recruited only men.18 20 21 26 33 From the remaining 12 studies that did not recruit sex-specific groups, 67% (SD ±16.2%) of participants were women. Thirteen studies reported ethnicity.18–22 25 29–35 Three studies reported targeting a single specific ethnic group: African-Americans,29 South Asians34 and ‘coloured’ (a term used by the authors) ethnicities.32 Of the remaining studies, 10 reported other ethnicity data; 80% of these participants were white Caucasian (SD ±16%, range 54%–100%).18–22 25 30 31 33 35 Sociodemographic data (SES or income groups, education) were not consistently reported. An area-based index of multiple deprivation was reported in two studies, one of which reported the highest proportion of participants in the two least deprived quintiles (5=25.1%, 4=22.2%)18 and the other in the two most deprived quintiles (1=25.7%, 2=20.29%).27 Average household income was below the poverty threshold in one study26 and indicative of a relatively high socioeconomic status in another.32 Seven studies reported employment status.20 21 28 30 34 36 38 In six of the seven studies, the majority of participants were employed.20 21 28 30 34 36

es (1=25.7%, 2=20.29%).27 Average household income was below the poverty threshold in one study26 and indicative of a relatively high socioeconomic status in another.32 Seven studies reported employment status.20 21 28 30 34 36 38 In six of the seven studies, the majority of participants were employed.20 21 28 30 34 36 Overview of recruitment reporting Five studies were classified as ‘high’ quality18 28 29 33 34 and the remaining 18 classified as ‘low’ quality in relation to recruitment reporting (see online supplementary table 5). All studies reported the setting where the recruitment of participants took place, but eight did not report who conducted the recruitment.25–27 32 36–38 40 None of the included studies reported the time spent planning or preparing their recruitment, although six studies reported the time spent conducting recruitment.18 25 28 29 33 34 All studies detailed the target population for recruitment (see online supplementary table 8).

o conducted the recruitment.25–27 32 36–38 40 None of the included studies reported the time spent planning or preparing their recruitment, although six studies reported the time spent conducting recruitment.18 25 28 29 33 34 All studies detailed the target population for recruitment (see online supplementary table 8). Recruitment data reported Because none of the studies reported the time spent planning or preparing the recruitment, no studies covered all stages of the recruitment process. All of the studies reported a specific target group and some details of where recruitment was conducted, although these were often non-specific (see online supplementary tables 6 and 8). Most popular were community settings (n=14) such as community centres,31 34 sports clubs,18–21 26 places of worship29 and locally distributed advertising/media.27 30 Medical or care settings were also popular (n=6).24 25 30 36 38 40 Universities22 35 and workplaces were also used.33 39 Seventeen studies reported who conducted the study recruitment. Most frequently reported recruiters were research staff (n=5), which reflects the evaluative nature of much of the literature.18 28 29 35 39 Seven studies reported the time spent on implementing recruitment,18 25 28 29 33 34 37 which averaged as 52 weeks (SD ±71 weeks, range 4–156 weeks) (see online supplementary table 6).

frequently reported recruiters were research staff (n=5), which reflects the evaluative nature of much of the literature.18 28 29 35 39 Seven studies reported the time spent on implementing recruitment,18 25 28 29 33 34 37 which averaged as 52 weeks (SD ±71 weeks, range 4–156 weeks) (see online supplementary table 6). Recruitment planning and implementation The reporting of recruitment methods was inconsistent and varied across studies (see online supplementary tables 6 and 7). The exact number of recruitment methods used was generally not disclosed and difficult to infer from the recruitment description. Five studies relied on one method of recruitment only,26 28 30 36 38 while 14 studies used two or more approaches.18 21 23–25 27–29 31–33 35 37 40 Recruitment approaches were categorised as ‘passive’ or ‘active’. ‘Passive’ recruitment techniques prompt potential participants to identify themselves for the programme,41 whereas ‘active’ techniques require those involved in the programme to initiate contact with a potential participant (eg, health professional referrals).42 No relationship between the quality of recruitment reporting and the number of recruitment strategies used was observed. We did observe that while a number of studies used only passive techniques (n=6),27 28 30 35 39 40 most used a mixture of active and passive (n=13)18–20 22–25 29 32–34 37 and a small number used only active methods (n=3) (see online supplementary table 7).31 36 38

the number of recruitment strategies used was observed. We did observe that while a number of studies used only passive techniques (n=6),27 28 30 35 39 40 most used a mixture of active and passive (n=13)18–20 22–25 29 32–34 37 and a small number used only active methods (n=3) (see online supplementary table 7).31 36 38 Recruitment rates and efficiencies We were unable to extract all of the values for the ‘pool, invited, responded and started’ participation levels required to calculate efficiencies across the recruitment process as set out by the Foster et al 10 review (see online supplementary table 9). We were, however, able to calculate a weekly recruitment rate using the final number of participants divided by the time spent recruiting in weeks for six studies (mean 13 participants per week, SD ±14, range <1 to 37 participants per week). Only one study reported the volume of participant uptake grouped by recruitment method.18

wever, able to calculate a weekly recruitment rate using the final number of participants divided by the time spent recruiting in weeks for six studies (mean 13 participants per week, SD ±14, range <1 to 37 participants per week). Only one study reported the volume of participant uptake grouped by recruitment method.18 Physical activity outcomes Physical activity outcomes were reported both directly, using validated measures of PA, and indirectly through the reporting of attendance or participation across the included studies (see online supplementary table 10). Of those studies reporting change in PA, 14 reported significant increases in PA between baseline and the end of the intervention.18 20 21 25 28 29 32 33 35–40 Only five of these, however, reported maintaining a significant increase in PA from baseline at post-intervention follow-up32 36 38–40 with four reverting to non-significant differences.18 28 29 35 Furthermore, one study reported a significant increase in PA at the post-intervention follow-up but not at the end of the intervention30 and another27 had no significant changes in PA to report at the end of the intervention. In the studies that did not quantify PA, records of attendance were used to represent PA participation during the intervention period.

icant increase in PA at the post-intervention follow-up but not at the end of the intervention30 and another27 had no significant changes in PA to report at the end of the intervention. In the studies that did not quantify PA, records of attendance were used to represent PA participation during the intervention period. Recruitment target, exclusion and study retention Sample size calculations were referenced in five studies,18 27 29 36 37 of which two indicated they were used to provide a target sample size for recruitment,18 29 which was successfully achieved by one18 and not the other.29 Fourteen studies indicated that there was a screening process to determine eligibility within the recruitment pathway.18 23 26 28–32 34–36 38–40 However, only five reported the proportion of recruits found to be ineligible at this stage (mean 18%, SD ±17.9%).18 28 29 32 39 The number of recruited participants who did not attend the intervention was reported in six studies (mean 19%, SD ±22.9%).26 29 31 34 35 40 The cost of recruitment per participant was calculated in one study at £20.32.25 It was not clear in the majority of studies whether incentives for recruitment (n=0) and retention (n=2)18 31 were offered. Attendance was reported in a variety of ways in 15 studies.18 22 24–26 28 29 31–35 38–40 The average reported attendance was 77% (n=9, SD ±12.4%, range 57%–100%).

culated in one study at £20.32.25 It was not clear in the majority of studies whether incentives for recruitment (n=0) and retention (n=2)18 31 were offered. Attendance was reported in a variety of ways in 15 studies.18 22 24–26 28 29 31–35 38–40 The average reported attendance was 77% (n=9, SD ±12.4%, range 57%–100%). Retention figures were reported in 16 studies.18 19 21 23 25 26 29–31 33 35–40 Retention reporting could either refer to participation in follow-up (study retention) or participation in PA or sport (PA retention) as part of the intervention or beyond. The average reported study retention rate at first follow-up was 82% (n=13, SD ±14.8, range 49–100),18 19 21 26 29 30 33 35 36 38–40 whereas the average reported PA retention rate was 28.5% (n=2, SD ±13.4, range 19–38).23 31 Number and reason for dropouts was reported in 11 studies.18 25 26 29 30 33–36 39 40 Most commonly cited reasons for dropout were illness or injury, work, unexpected commitments, lack of time, relocation or travel, and disliking the intervention.

e reported PA retention rate was 28.5% (n=2, SD ±13.4, range 19–38).23 31 Number and reason for dropouts was reported in 11 studies.18 25 26 29 30 33–36 39 40 Most commonly cited reasons for dropout were illness or injury, work, unexpected commitments, lack of time, relocation or travel, and disliking the intervention. Additional comments reported regarding recruitment and retention Comments relating to recruitment highlighted word of mouth19 21 22 24 25 and social media22–24 as valuable recruiters. Additionally, the role of recruitment partnerships19 20 23 24 as well as active,24 passive25 and multiple23–25 recruitment mechanisms were discussed. The design of promotional materials was highlighted in a number of studies.22–24 The intervention setting,20 21 appeal of the activity18 19 21 22 25 and opportunities for socialising19 23 were also important for recruitment of the target group. Several studies commented on the successes20 21 and challenges24 27 of reaching the target group.

nal materials was highlighted in a number of studies.22–24 The intervention setting,20 21 appeal of the activity18 19 21 22 25 and opportunities for socialising19 23 were also important for recruitment of the target group. Several studies commented on the successes20 21 and challenges24 27 of reaching the target group. Facilitators of retention discussed included social support,19 21 25 26 34 38 39 variety of activities,19 21 22 24 group cohesion,34 39 40 fun/enjoyment,21 25 34 coaching,24 31 routine,19 21 accessibility of delivery site,21 competition,22 timing of sessions,24 affordability,24 use of incentives,24 availability of progression opportunities,24 high programme satisfaction34 and higher baseline self-motivations towards PA.39 Barriers to retention included dropout or non-attendance in the early stages of the intervention,18 25 if individual activity intervention rather than group was used,38 the appeal of the activity21 and degree of competition.21

rtunities,24 high programme satisfaction34 and higher baseline self-motivations towards PA.39 Barriers to retention included dropout or non-attendance in the early stages of the intervention,18 25 if individual activity intervention rather than group was used,38 the appeal of the activity21 and degree of competition.21 Discussion The effectiveness of any PA or sporting intervention is limited by the impact the intervention has on its participants and by the effectiveness of its recruitment of eligible participants to take part in the intervention.10 This systematic review showed that the evidence on how best to recruit is sparse due to the absence of generalisable findings and insufficient reporting of recruitment methodology and process outcomes in sports-based PA interventions. Furthermore, there is an absence of evidence linking specific recruitment methods to more successful long-term behaviour changes due to a lack of clarity surrounding recruitment channel outcomes. Lack of reported information meant we were also unable to asses the cost per person of recruitment for different recruitment strategies. This would be of particular interest as there is an inevitable cost–effect trade-off to be considered when designing recruitment mechanisms.

surrounding recruitment channel outcomes. Lack of reported information meant we were also unable to asses the cost per person of recruitment for different recruitment strategies. This would be of particular interest as there is an inevitable cost–effect trade-off to be considered when designing recruitment mechanisms. We know that reaching priority groups for PA studies, such as the inactive or unhealthy, socioeconomically disadvantaged, ethnic and other minority groups, is challenging.9 Our review demonstrates that current recruitment strategies engage predominantly white, middle-class, middle-aged women unless they are clearly designed to target specific demographic characteristics, such as gender or ethnicity. Furthermore, this recruitment bias for particular populations is supported by an earlier review of wider PA interventions.6 Our review also found that despite targeting some of these demographic characteristics successfully, interventions tended to achieve unrepresentative levels for the remaining untargeted characteristics such as socioeconomic status. For example, Vahabi and Damba34 recruited predominantly Indian women in a South Asian Bollywood dancing intervention. However, most of these women were in full-time employment and had a university degree or higher, suggesting they were of a relatively high socioeconomic status. These findings indicate that typical recruitment techniques adopted by sports PA interventions are not reaching those most in need. Furthermore, targeting specific isolated characteristics may not be sufficient to alleviate recruitment bias towards particular demographic characteristics, thus limiting the generalisability of findings for policy and practice.

ruitment techniques adopted by sports PA interventions are not reaching those most in need. Furthermore, targeting specific isolated characteristics may not be sufficient to alleviate recruitment bias towards particular demographic characteristics, thus limiting the generalisability of findings for policy and practice. The use of active-only recruitment techniques31 36 38 (whereby those involved in the study or programme make the first contact with a participant) appeared to achieve a participant sample with more representative demographic characteristics than passive approaches (ie, prompt potential participants to identify themselves) not targeting specific demographics. This finding is supported by Mutrie et al’s paper,9 which found that active techniques were less likely to encourage self-selection and introduce recruitment bias than passive techniques. For example, a recruitment and intervention partnership targeting low-income demographics via community health services (active)31 successfully represented at-risk populations (52% women, 54% white, average income less than 100% of the federal poverty level). However, this study had a 63% non-use attrition rate, thus demonstrating a low conversion to participation. This supports existing evidence that place-based strategies (ie, recruiting participants from a location where they already aggregate) result in more representative samples, but a lower overall participation rate.9 43 There is, however, evidence from the Mangeri et al paper38 that techniques such as motivational interviewing and offering group intervention activities may encourage higher levels of participation in actively recruited participants. These observations highlight the difficulties and tensions faced by sporting interventions for reaching those most at risk or need. Due to insufficient monitoring and evaluation of recruitment processes, we were unable to determine which specific methods were more effective at engaging particular populations.

icipants. These observations highlight the difficulties and tensions faced by sporting interventions for reaching those most at risk or need. Due to insufficient monitoring and evaluation of recruitment processes, we were unable to determine which specific methods were more effective at engaging particular populations. Setting In addition to recruitment mode, the reach of a programme is determined by the appeal of characteristics of the intervention to target participants. Further, research processes such as having to complete questionnaires or wear an accelerometer may also influence uptake and impact recruitment to the programme in general. This review highlights the diversity and effectiveness of a subgroup of interventions delivered in professional football (soccer) clubs for engaging target populations in PA and health-promoting activities.18–21 26 The professional football context in which these programmes were delivered was reported as a powerful draw for participants.19–21 44 Furthermore, the range of populations successfully targeted and recruited by these interventions demonstrates the wide reach of professional sports settings. Using football club facilities, events, branding and media channels were all thought to contribute to the engagement of target groups, extending beyond the fan base of the host club.45 These programmes, therefore, highlight the importance of delivering interventions in contexts that are appealing to their target group and that use pre-existing interests and communities to achieve successful engagement.

to contribute to the engagement of target groups, extending beyond the fan base of the host club.45 These programmes, therefore, highlight the importance of delivering interventions in contexts that are appealing to their target group and that use pre-existing interests and communities to achieve successful engagement. Reporting The quality of recruitment reporting across included studies was generally classified as poor, with a distinct lack of information on how much time was spent planning and implementing recruitment. The majority of interventions reported using multiple, simultaneous recruitment methods, but only one reported the effectiveness of individual approaches.18 Instead, recruitment was largely generalised to a location or setting, thus making it unclear which method or combination of methods were most effective at recruiting particular target groups and achieving long-term behaviour change.

cruitment methods, but only one reported the effectiveness of individual approaches.18 Instead, recruitment was largely generalised to a location or setting, thus making it unclear which method or combination of methods were most effective at recruiting particular target groups and achieving long-term behaviour change. Higher level overviews of target populations were generally well reported. However, few studies described the pool of potential participants, the number of individuals invited to participate, how many responded and how many started the intervention, all of which were required to calculate recruitment efficiency in accordance with the Foster et al review.10 In fact, comparable levels of reporting were seen between studies included in our review and Foster et al’s,10 thus restricting both reviews’ ability to compute and compare efficiency ratios across included studies. Further to this, we also sought to investigate some process outcomes that are of interest when looking to translate research into practice: target sample size and the extent to which it was achieved, the eligibility and representativeness of recruited participants as well as their participation and retention in the intervention. Unfortunately, these were also poorly reported, meaning there is limited process-based evidence to inform recruitment designs that facilitate powered and representative sample sizes for rigorous evaluation.

nd representativeness of recruited participants as well as their participation and retention in the intervention. Unfortunately, these were also poorly reported, meaning there is limited process-based evidence to inform recruitment designs that facilitate powered and representative sample sizes for rigorous evaluation. The need for improved reporting highlighted in this review may in part be driven by a previous lack of agreement on a standard framework for recruitment reporting. In England, a standard evaluation framework (SEF) has recently been introduced to guide programmes to collect and evaluate information relating to PA interventions.12 The SEF advocates the monitoring and reporting of a range of recruitment variables as part of its essential criteria for evaluating PA interventions. These include method of recruitment, target population characteristics, measures of the flow of participants through the intervention and dates of crucial time points such as first point of contact and follow-ups. Following adequate execution and reporting of the essential recruitment information, such as those set out in the SEF, the future application of data extraction and scoring procedures attempted within this review would likely produce a greater insight into recruitment processes for sporting interventions aiming to promote PA. We are unaware of similar guidance in other international settings, although these may exist. The level of measurement and reporting observed in this review, however, suggests that standard evaluation of recruitment variables is not being used elsewhere. In the UK, the SEF is a relatively recent document having been published in 2012, and as such its impact may yet to be seen in the academic literature. This may also be the case for other similar guidance, yet if comparable guidance has been in place for longer, our findings raise the question of the extent of its adoption and application.

a relatively recent document having been published in 2012, and as such its impact may yet to be seen in the academic literature. This may also be the case for other similar guidance, yet if comparable guidance has been in place for longer, our findings raise the question of the extent of its adoption and application. Strengths and limitations of this review The strength of this review is that it used protocols, previously shown to be successful in other reviews, to systematically identify and analyse 23 sport PA intervention papers using a comprehensive search strategy and extensive data extraction table. We also attempted to compute a set of metrics to describe recruitment efficiency that were informed by similar systematic reviews investigating sport or recruitment.

reviews, to systematically identify and analyse 23 sport PA intervention papers using a comprehensive search strategy and extensive data extraction table. We also attempted to compute a set of metrics to describe recruitment efficiency that were informed by similar systematic reviews investigating sport or recruitment. Regarding limitations, there are limited published or publicly available evaluations of sporting interventions published in English as well as a reluctance from editors to publish articles where recruitment is the prime focus.8 This restricted conclusions relating to a number of processes of interest for this review. In line with the objectives of this study, we assessed the quality of recruitment reporting as our risk of bias tool. An implication of this was that we did not assess the overall quality of the included study. The lack of rigorous recruitment monitoring and reporting meant that the quality of extracted information on recruitment was not enough to allow a meta-analysis to be taken. A further issue in this review was that a number of the programmes considered were designed as research studies and consequently include additional research processes otherwise absent in typical sports interventions. There is evidence that characteristics of the population recruited can be influenced by recruitment protocols, intervention and research design characteristics.9 Research protocols may therefore impact recruitment differently to unevaluated sports programmes or those with a lighter-touch evaluation. Consequently, the generalisability of our results was limited by the difficulties of separating the effect of the research and evaluation from overall programme recruitment. Furthermore, extracted values for participant outcomes were limited to only those who participated in the reported evaluation and may not be representative of the complete participant population.

was limited by the difficulties of separating the effect of the research and evaluation from overall programme recruitment. Furthermore, extracted values for participant outcomes were limited to only those who participated in the reported evaluation and may not be representative of the complete participant population. Conclusions Overall, this review emphasises the need for robust evaluation design and reporting of sports intervention and recruitment processes to permit future evidence-based interventions. There is a growing evidence base for the benefits of sport for physical inactivity and health. However, due to inadequate reporting and evaluation, the mechanisms for achieving effective recruitment and engagement in sport, particularly in hard-to-reach groups, are still unclear. There is a notable tendency of sporting interventions to recruit white, more affluent, middle-aged women. Simply targeting isolated demographic characteristics, such as gender or ethnicity, appears insufficient to recruit a sample representative of the population for the remaining, untargeted characteristics. Combinations of active and passive methods were commonly used, yet active-only recruitment approaches achieved more representative samples for their target population. It is of concern, however, that active-only recruitment may be more vulnerable to limited participant engagement, thus requiring additional components such as motivational interviewing to encourage participation. Independent of recruitment mode, creating an intervention and context that reflect the interests and motivations of the target audience, such as local professional football facilities, presents a promising area.

gagement, thus requiring additional components such as motivational interviewing to encourage participation. Independent of recruitment mode, creating an intervention and context that reflect the interests and motivations of the target audience, such as local professional football facilities, presents a promising area. 10.1136/bmjsem-2017-000231.supp2 10.1136/bmjsem-2017-000231.supp3Supplementary data We would like to thank the authors for their correspondence and contribution to the collected data. We would also like to thank Dr Charlie Foster for his input and advice that helped inform the systematic review protocol, Ryan Love for acting as a second reviewer, William Jones, faculty librarian, for his help in the search strategy and Professor Marc Suhrcke, Professor Richard Fordham, Dr Victoria Warburton and Dr Lee Beaumont for their guidance and support. Funding from Sport England is gratefully acknowledged. Contributors: Charlie Foster gave input and advice that helped inform the systematic review protocol. Ryan Love acted as a second reviewer. William Jones, faculty librarian, helped with the search strategy. Professor Marc Suhrcke, Professor Richard Fordham, Dr Victoria Warburton and Dr Lee Beaumont provided guidance and support.

Contributors: Charlie Foster gave input and advice that helped inform the systematic review protocol. Ryan Love acted as a second reviewer. William Jones, faculty librarian, helped with the search strategy. Professor Marc Suhrcke, Professor Richard Fordham, Dr Victoria Warburton and Dr Lee Beaumont provided guidance and support. Funding: Funding from Sport England is gratefully acknowledged. The work was undertaken under the auspices of the Centre for Diet and Activity Research (CEDAR), a UKCRC Public Health Research Centre of Excellence (RES-590-28-0002) that is funded by the British Heart Foundation, Cancer Research UK, Economic and Social Research Council, Medical Research Council, the National Institute for Health Research and the Wellcome Trust. Competing interests: None declared. Provenance and peer review: Not commissioned; externally peer reviewed.

Background Researchers, policy-makers and practitioners concur that regular physical activity (PA) benefits persons of all ages and backgrounds. It has positive effects on mental health, physical health and longevity for both individuals and populations.1–4 A recent aim of major sporting events has been to secure a legacy of increased PA or participation in sport following the event.5 Major multi-sport games have failed to achieve an inherent, substantial PA legacy.6 Measures that could help address this lack of legacy include (1) producing a clear strategy to increase participation and (2) de-emphasising the sporting element and promoting PA more generally (for example, walking) rather than simply the sport being played.7 8 Golf can provide a novel and suitable narrative to provide a link between sport, walking and potential health benefits.9 Golf playing and spectating is particularly popular in middle-aged and older adults in North America, Europe and Asia in particular.10 This demographic typically have lower levels of PA compared with younger adults and children.11–13

ble narrative to provide a link between sport, walking and potential health benefits.9 Golf playing and spectating is particularly popular in middle-aged and older adults in North America, Europe and Asia in particular.10 This demographic typically have lower levels of PA compared with younger adults and children.11–13 Collectively, tournaments in the USA alone can draw over 10 million spectators per year.14 Those watching the action at several hundred tournaments on six continents worldwide may have the opportunity to gain health-enhancing physical activity (HEPA) on the available square miles of playing arena.11 Indeed, the existing literature suggests that golf spectators rate perceived ‘health benefits’ and ‘exercise’ as important considerations in attending tournaments,15–17 with Lyu and Lee segmenting the motivations of spectators into ‘excitement seekers’, ‘exercise seekers’, ‘interest seekers’ and ‘escape seekers’.17 Our recent scoping review identified knowledge gaps, namely that no studies have characterised the effects of spectating at golf tournaments on PA knowledge or PA levels.9 18 We aim to contribute to these knowledge gaps. We first address critical feasibility questions and assess the extent to which spectating delivers opportunities for PA. Our research questions were the following:Is studying spectator PA through pedometer measured step counts feasible at a professional golf tournament? What reasons do spectators at a European Tour event identify for their attendance?

Collectively, tournaments in the USA alone can draw over 10 million spectators per year.14 Those watching the action at several hundred tournaments on six continents worldwide may have the opportunity to gain health-enhancing physical activity (HEPA) on the available square miles of playing arena.11 Indeed, the existing literature suggests that golf spectators rate perceived ‘health benefits’ and ‘exercise’ as important considerations in attending tournaments,15–17 with Lyu and Lee segmenting the motivations of spectators into ‘excitement seekers’, ‘exercise seekers’, ‘interest seekers’ and ‘escape seekers’.17 Our recent scoping review identified knowledge gaps, namely that no studies have characterised the effects of spectating at golf tournaments on PA knowledge or PA levels.9 18 We aim to contribute to these knowledge gaps. We first address critical feasibility questions and assess the extent to which spectating delivers opportunities for PA. Our research questions were the following:Is studying spectator PA through pedometer measured step counts feasible at a professional golf tournament? What reasons do spectators at a European Tour event identify for their attendance? Can spectators gain a relevant dose of PA (measured by step count) while attending a professional golf tournament?

Our research questions were the following:Is studying spectator PA through pedometer measured step counts feasible at a professional golf tournament? What reasons do spectators at a European Tour event identify for their attendance? Can spectators gain a relevant dose of PA (measured by step count) while attending a professional golf tournament? Methods We conducted a cross-sectional study consisting of two linked elements: a questionnaire completed by spectators on entering the course and a measure of step count from the time a spectator entered the venue until the time they exited. Ethical approval was granted (15 July 2016) by the Moray House School of Education Ethics Committee at the University of Edinburgh. Data collection Data were collected on all days of tournament play (4–7 August 2016) at the European Tour Paul Lawrie Matchplay event in Scotland. The European Tour meteorological service recorded temperatures of between 18°C and 21°C (highs) and 9°C–13°C (lows). Winds were light to moderate, except on the final day of play where 40–45 miles/hour gusts were experienced. Rain fell for <10% of the duration of play. Spectators attending the event were approached by one of six trained researchers who invited spectators to read a two-page participant information sheet, detailing the purposes of the study as they arrived. Those willing to partake were assessed against the inclusion and exclusion criteria stated in table 1 below by a researcher. Table 1 Inclusion and exclusion criteria Inclusion criteria Exclusion criteria Spectators at the European Tour Paul Lawrie Matchplay Aged ≥18 years

Spectators attending the event were approached by one of six trained researchers who invited spectators to read a two-page participant information sheet, detailing the purposes of the study as they arrived. Those willing to partake were assessed against the inclusion and exclusion criteria stated in table 1 below by a researcher. Table 1 Inclusion and exclusion criteria Inclusion criteria Exclusion criteria Spectators at the European Tour Paul Lawrie Matchplay Aged ≥18 years Able to walk (walking aids permitted) Unstable cardiovascular disease not reported Non-spectators (for example staff, marshals, players, caddies) Spectators that had taken part in the study on previous days Aged under 18 years Inability to walk Reported unstable cardiovascular disease (critical aortic stenosis, unstable angina, myocardial infarction within 6 weeks—a medical doctor was part of the research team and could provide individual case advice)

Non-spectators (for example staff, marshals, players, caddies) Spectators that had taken part in the study on previous days Aged under 18 years Inability to walk Reported unstable cardiovascular disease (critical aortic stenosis, unstable angina, myocardial infarction within 6 weeks—a medical doctor was part of the research team and could provide individual case advice) Those eligible were invited to sign a consent form, and following this completed a baseline questionnaire. This questionnaire was devised from a review of relevant previous studies15–17 19 and was refined following discussion with the research team and officials from the European Tour golf. The full questionnaire is shown in the online supplementary appendix 1 and included seven demographic items, eight items including a free text option assessing reasons for spectating, and three items assessing self-reported current PA levels and interest in becoming more physically active. These last three items were facilitated by a member of the research team using a validated tool (Scot-PASQ; NHS Health Scotland, UK). 10.1136/bmjsem-2017-000244.supp1Supplementary Appendix 1

Those eligible were invited to sign a consent form, and following this completed a baseline questionnaire. This questionnaire was devised from a review of relevant previous studies15–17 19 and was refined following discussion with the research team and officials from the European Tour golf. The full questionnaire is shown in the online supplementary appendix 1 and included seven demographic items, eight items including a free text option assessing reasons for spectating, and three items assessing self-reported current PA levels and interest in becoming more physically active. These last three items were facilitated by a member of the research team using a validated tool (Scot-PASQ; NHS Health Scotland, UK). 10.1136/bmjsem-2017-000244.supp1Supplementary Appendix 1 Following this, a researcher fitted a Silva Ex Step (Silva, Stockholm, Sweden) pedometer to the lateral aspect of the right hip region of each participant, noting the time this was fitted. Participants were asked to check the pedometer was registering steps after 1–2 min, and if not it was repositioned to an adjacent position. The European Tour works with a Scottish charity that champions walking ‘Paths for All’. Paths for All recommended the Silva Ex Step as having high usability compared with other devices. A brief validation of five Silva Ex Step devices was performed with <5% difference for all devices noted compared with Actigraph (Pensacola, Florida, USA). Paths for All also offered all spectators information relating to spectating and health, as is standard at Scottish-based European Tour events.

with other devices. A brief validation of five Silva Ex Step devices was performed with <5% difference for all devices noted compared with Actigraph (Pensacola, Florida, USA). Paths for All also offered all spectators information relating to spectating and health, as is standard at Scottish-based European Tour events. The participant then spectated for a length of time of their choosing and in a manner of their choosing. Prior to exiting the venue, participants returned the pedometer to a member of the research team who checked and recorded the number of steps taken and the time returned. Data analysis With regard to feasibility, we decided, rather than to specify in advance a hypothesis to determine feasibility, that we would assess feasibility on a subjective basis based on response, recruitment, compliance and the human and equipment resources required. Pedometer failure is a recognised issue in step-count studies. We had specified criteria for inclusion and exclusion of data. When pedometers were returned, the values were entered into the database, and the researcher assessed them for face validity. The participant sometimes offered information unprompted that the pedometer had failed. Where there was clear error, the result was excluded.

cified criteria for inclusion and exclusion of data. When pedometers were returned, the values were entered into the database, and the researcher assessed them for face validity. The participant sometimes offered information unprompted that the pedometer had failed. Where there was clear error, the result was excluded. Statistical Package for the Social Science V.22 software was used for data management and analysis. Variables were assessed for normality with means or medians reported as appropriate. We used independent samples t-tests to explore any possible differences in step counts by age and gender. The association between minutes spectating and steps taken was tested using Pearson correlation coefficient.

gement and analysis. Variables were assessed for normality with means or medians reported as appropriate. We used independent samples t-tests to explore any possible differences in step counts by age and gender. The association between minutes spectating and steps taken was tested using Pearson correlation coefficient. Results Feasibility/spectator characteristics European Tour figures show the 2016 Paul Lawrie Matchplay was attended by 1500 paying spectators. Approximately 600 spectators in total were approached to take part in the study. A total of 339 spectators were recruited to the study and agreed to complete the questionnaire. Of those who did not agree to take part, most indicated that they were in a hurry to go and watch the golf. Of these 339 participants, 329 collected step count data and returned the pedometer (97.2%). Twenty (6.1%) pedometers failed to register accurate readings. Participants recruited and completing the study represented 22.6% of the eligible tournament population. While not part of the study, researchers were approached by marshals, children, golf caddies, professional players and returning spectators requesting literature relating to golf and health and/or pedometers to monitor their step count highlighting interest in this topic beyond direct participants. The baseline characteristics of participants are shown in table 2. Approximately two-thirds of participants were men, with men between 40 and 59 years old most strongly represented. Table 2 Baseline characteristics of participants

Results Feasibility/spectator characteristics European Tour figures show the 2016 Paul Lawrie Matchplay was attended by 1500 paying spectators. Approximately 600 spectators in total were approached to take part in the study. A total of 339 spectators were recruited to the study and agreed to complete the questionnaire. Of those who did not agree to take part, most indicated that they were in a hurry to go and watch the golf. Of these 339 participants, 329 collected step count data and returned the pedometer (97.2%). Twenty (6.1%) pedometers failed to register accurate readings. Participants recruited and completing the study represented 22.6% of the eligible tournament population. While not part of the study, researchers were approached by marshals, children, golf caddies, professional players and returning spectators requesting literature relating to golf and health and/or pedometers to monitor their step count highlighting interest in this topic beyond direct participants. The baseline characteristics of participants are shown in table 2. Approximately two-thirds of participants were men, with men between 40 and 59 years old most strongly represented. Table 2 Baseline characteristics of participants Age (years) Men Women Total 18–39 49 18 67 40–59 105 46 151 ≥60 68 43 111 Total 222 107 329 Reasons for attendance Within the baseline questionnaire, participants were asked to rate reasons for spectating on a scale of 1 (of no importance) to 10 (of extremely high importance).

Table 2 Baseline characteristics of participants Age (years) Men Women Total 18–39 49 18 67 40–59 105 46 151 ≥60 68 43 111 Total 222 107 329 Reasons for attendance Within the baseline questionnaire, participants were asked to rate reasons for spectating on a scale of 1 (of no importance) to 10 (of extremely high importance). Median and mode values showing spectators’ stated reasons for attendance are shown in Table 3. ‘Fresh air’ (rated median 9 out of 10) then ‘watching star players’, ‘exercise/physical activity’, ‘time with friends and family’ and ‘atmosphere’ (all median 8 out of 10) were rated the most important reasons for attending (table 3). Table 3 Reasons for attendance at the Paul Lawrie Matchplay as rated by participants on entry to the venue Watch star players Learn from star players Non-golfing entertainment Atmosphere Fresh air Exercise/physical activity Time with friends/family No of respondents 338 337 333 332 333 334 334 Median 8.00 7.00 5.00 8.00 9.00 8.00 8.00 IQR 2 4 5 2 3 4 3 Mode 8 8 1 8 10 10 10 In terms of the importance of reasons for attendance, exercise and physical activity was of interest to this paper on spectator health. The relative percentage for spectator rating of importance of exercise/physical activity as a reason for attending is displayed in figure 1. Figure 1 Participant rating (1–10) of ‘exercise/physical activity’ as a reason for attendance on entry to the venue. Measured spectator PA

Watch star players Learn from star players Non-golfing entertainment Atmosphere Fresh air Exercise/physical activity Time with friends/family No of respondents 338 337 333 332 333 334 334 Median 8.00 7.00 5.00 8.00 9.00 8.00 8.00 IQR 2 4 5 2 3 4 3 Mode 8 8 1 8 10 10 10 In terms of the importance of reasons for attendance, exercise and physical activity was of interest to this paper on spectator health. The relative percentage for spectator rating of importance of exercise/physical activity as a reason for attending is displayed in figure 1. Figure 1 Participant rating (1–10) of ‘exercise/physical activity’ as a reason for attendance on entry to the venue. Measured spectator PA Table 4 shows the mean/median number of steps taken by spectators, stratified by gender. The independent samples t-test revealed a statistically significant difference by gender with men taking approximately 1858 more steps on the day they attended (95% CI 784 to 2933, p<0.001). There were no important differences in step counts by age group. Table 4 Mean/median number of steps taken by gender Gender Measure Statistic Men Mean 12 172.5 95% CI for mean Lower bound 11 586.6 Upper bound 12 758.4 Median 11 362.5 SD 4327.6 Minimum 1576 Maximum 25 312 Women Mean 10 314.1 95% CI for mean Lower bound 9361.9 Upper bound 11 266.2 Median 10 039.0 SD 4724.2 Minimum 310 Maximum 25 098 Figure 2 displays the number of steps taken categorised into (1) inactive, (2) low active and (3) meeting moderate to vigorous physical activity guidelines.20 Figure 2 Number of spectators: <5000 steps, 5000–7500 steps and >7500 steps.

Gender Measure Statistic Men Mean 12 172.5 95% CI for mean Lower bound 11 586.6 Upper bound 12 758.4 Median 11 362.5 SD 4327.6 Minimum 1576 Maximum 25 312 Women Mean 10 314.1 95% CI for mean Lower bound 9361.9 Upper bound 11 266.2 Median 10 039.0 SD 4724.2 Minimum 310 Maximum 25 098 Figure 2 displays the number of steps taken categorised into (1) inactive, (2) low active and (3) meeting moderate to vigorous physical activity guidelines.20 Figure 2 Number of spectators: <5000 steps, 5000–7500 steps and >7500 steps. Number of steps An increasing number of minutes spectating had a moderate association with total steps taken (r=0.67). This shows that on average, participants attending for longer accrued more steps. Questionnaire data from the SCOT PAS-Q items collected as participants entered the venue highlighted that 89.3% reported meeting the aerobic moderate to vigorous physical activity guidelines in the previous week, while 68.1% reported being ‘interested in being more physically active’. Discussion Principal findings Feasibility This study indicates that it is feasible to study adult spectator PA (by pedometer-measured step counts) at a professional golf tournament. Approximately 56% of spectators approached agreed to participate, and of these, 97.2% returned questionnaire and step-count data.

Questionnaire data from the SCOT PAS-Q items collected as participants entered the venue highlighted that 89.3% reported meeting the aerobic moderate to vigorous physical activity guidelines in the previous week, while 68.1% reported being ‘interested in being more physically active’. Discussion Principal findings Feasibility This study indicates that it is feasible to study adult spectator PA (by pedometer-measured step counts) at a professional golf tournament. Approximately 56% of spectators approached agreed to participate, and of these, 97.2% returned questionnaire and step-count data. It should be noted that it was not practical to engage with every spectator who entered the venue. Spectators can typically access the venue through more than one entrance and often arrive in groups having largely travelled by coach transfer. A larger number of researchers would be needed to engage with a larger volume and proportion of attending spectators. Spectator reasons for attending Spectators rated a number of reasons for attending this professional golf tournament as highly important. ‘Watch star players’, ‘atmosphere’, ‘fresh air’, ‘exercise/physical activity’ and ‘time with friends and family’ all scored median and mode values of 8 out of 10 or greater. Importantly, for this work, obtaining exercise/PA can be a motivation for attending for participants at this event. The median rating was 8 out of 10, with a mode of 10, representing ‘of extremely high importance’.

activity’ and ‘time with friends and family’ all scored median and mode values of 8 out of 10 or greater. Importantly, for this work, obtaining exercise/PA can be a motivation for attending for participants at this event. The median rating was 8 out of 10, with a mode of 10, representing ‘of extremely high importance’. The extent to which spectating delivers an opportunity for PA This is the first published study to measure golf spectator PA by step count. Data show participants took a mean of 11 589 (SD 4531, range 25 002) and a median of 11 086 steps. Through spectating alone, 82.9% of participants met Tudor-Locke et al’s daily guidelines indicative of a ‘physically active lifestyle’ from activity while spectating with 94.8% of spectators meeting either ‘low active’ or ‘physically active’ lifestyle.20 As may be expected, an increasing number of minutes spectating had a positive association with increased total steps taken. There were no apparent differences in step counts by age group, but there was a statistically significant and potentially clinically relevant difference by gender with male participants taking approximately 1858 more steps per day than female participants. Comparison to the literature and explanation for findings Spectator reasons for attending A large body of research has assessed spectator motivations for attendance at sporting events, but most of these pertain to team-based sports,14 with data specific to golf limited.14–17 McDonald et al found clear spectator motivation differences between golf spectators and spectators of other sports.21

ns for attending A large body of research has assessed spectator motivations for attendance at sporting events, but most of these pertain to team-based sports,14 with data specific to golf limited.14–17 McDonald et al found clear spectator motivation differences between golf spectators and spectators of other sports.21 Watching star players is the most powerful motivator for golf spectator attendance in most previous studies conducted,14–17 and the current study supports the importance spectators place on this. Robinson et al argue that the prime marketing focus for events should be on specific well-recognised golfers playing.14 However, spectators in our sample rate at least equally highly other reasons for spectating including ‘fresh air’, ‘spending time with friends and family’ and ‘exercise/physical activity’. These data support Lyu and Lee’s assertion that factors such as these offer attractive marketing angles to tournament organisers/promoters, with the aim of increasing spectator volume and engagement.17 These factors were not probed as explicitly in Robinson et al’s study of US spectators, with the questionnaire employed not golf specific. It is known motivations for golf spectators are different to team sports, being broader and less homogeneous.21

Watching star players is the most powerful motivator for golf spectator attendance in most previous studies conducted,14–17 and the current study supports the importance spectators place on this. Robinson et al argue that the prime marketing focus for events should be on specific well-recognised golfers playing.14 However, spectators in our sample rate at least equally highly other reasons for spectating including ‘fresh air’, ‘spending time with friends and family’ and ‘exercise/physical activity’. These data support Lyu and Lee’s assertion that factors such as these offer attractive marketing angles to tournament organisers/promoters, with the aim of increasing spectator volume and engagement.17 These factors were not probed as explicitly in Robinson et al’s study of US spectators, with the questionnaire employed not golf specific. It is known motivations for golf spectators are different to team sports, being broader and less homogeneous.21 Spectator attitudes towards changing exercise/PA Evidence from North America, Asia and Europe is consistent and growing that exercise/PA can be a motivator for attending golf tournaments.15–17 Golf tournaments and their spectators are heterogeneous, and some may be more motivated than others by PA benefits based on individual, cultural, climactic and tournament differences. They may also be likely to be meeting minimum PA levels already. Our study did not find significant age-related and gender-related differences in attitudes of spectators towards exercise/PA. The literature broadly supports a greater emphasis of these benefits by event promoters,15–17 which may be beneficial in terms of engagement with spectators, local communities and funding organisations.

y did not find significant age-related and gender-related differences in attitudes of spectators towards exercise/PA. The literature broadly supports a greater emphasis of these benefits by event promoters,15–17 which may be beneficial in terms of engagement with spectators, local communities and funding organisations. PA gained while spectating There are no previous published studies that measured the levels of PA attained by golf spectators. Unpublished data (obtained via personal correspondence, Event Scotland) from the 2014 Ryder Cup, Gleneagles, UK, show over 20 000 spectators tagged every checkpoint at locations on course, indicating they had walked 8 kilometres each, or 100 000 miles collectively. At the 2016 Shenzhen Open, Shenzhen, China, 6500 spectators completed a ‘health walk’ intervention, of 10 km each, adding up to a distance seven times the length of the Great Wall of China (personal communication, Shenzhen Open).

ns on course, indicating they had walked 8 kilometres each, or 100 000 miles collectively. At the 2016 Shenzhen Open, Shenzhen, China, 6500 spectators completed a ‘health walk’ intervention, of 10 km each, adding up to a distance seven times the length of the Great Wall of China (personal communication, Shenzhen Open). Step counting using pedometers is a well-established method of measuring PA by the general public, researchers and policy-makers.20 Data showed that 82.9% of participants met Tudor-Locke et al’s moderate to vigorous physical activity daily guidelines (>7500 steps) from activity while spectating alone, when measured by step count. This is the first study to report PA levels in golf spectators. The self-reported interest in exercise/PA as a reason for attending may be important in explaining the high level of PA achieved. For some, attending the event may represent a deliberate attempt to gain HEPA, while others gain incidental HEPA through their desire to observe particular golfers or the course.16 17 Female step count may be lower than male spectators due to factors that may include footwear choice. Equivalent studies of spectator populations’ PA at other tournaments would likely be influenced (positively or negatively) by factors including but not limited to ambient weather conditions, cultural factors, type of tournament and terrain/walkability of the golf course.

due to factors that may include footwear choice. Equivalent studies of spectator populations’ PA at other tournaments would likely be influenced (positively or negatively) by factors including but not limited to ambient weather conditions, cultural factors, type of tournament and terrain/walkability of the golf course. Recommendations for practice/policy and research Recent strategies from the Department of Culture, Media and Sport and Sport England among others have highlighted the value of spectating at sporting venues and the potential for inspiration and increasing PA.22 23 Increasingly, sports organisations/franchises, governing bodies for sport, stadia operators and others are being encouraged to develop practices and policies that promote improved public health for fans and communities. These include efforts relating to healthy eating, alcohol consumption, tobacco use and sustainability as well as promoting PA. This study confirms it is feasible to study spectator PA and attitudes towards PA in a golf setting. Response rates were good, and compliance rates among participants were exceedingly high. We showed that a reasonable sample size can be achieved with a team of six trained researchers. This will be important information for future work and potential power calculations for sample size requirements. A well-structured questionnaire and collaboration with the tournament organisers are also highly recommended.

We showed that a reasonable sample size can be achieved with a team of six trained researchers. This will be important information for future work and potential power calculations for sample size requirements. A well-structured questionnaire and collaboration with the tournament organisers are also highly recommended. Golf spectating does offer an opportunity for PA in this setting and population. Attendance can thus be encouraged, and spectators can be supported to do so in an active fashion in promotional efforts ahead of and during each professional golf event. Golf tournament event planning, marketing efforts, golf course choice and architecture should reflect this. Fans/spectators can receive public health benefits, while tournament organisers/sponsors may realise revenue and corporate and social responsibility benefits. With two-thirds of participants indicating an interest to be more physically active, it may be an opportunity for intervention in a ‘contemplative’ population. While the participants were largely already meeting the guidelines, it should be noted that this is a minimum level of PA and more is better, and that maintenance of PA is critical in adult and ageing populations. Research priorities for the future includeAssessing what methods for providing PA information/intervention (eg, big screen, leaflet, poster, email, direct conversation) are welcomed by spectators.

Golf spectating does offer an opportunity for PA in this setting and population. Attendance can thus be encouraged, and spectators can be supported to do so in an active fashion in promotional efforts ahead of and during each professional golf event. Golf tournament event planning, marketing efforts, golf course choice and architecture should reflect this. Fans/spectators can receive public health benefits, while tournament organisers/sponsors may realise revenue and corporate and social responsibility benefits. With two-thirds of participants indicating an interest to be more physically active, it may be an opportunity for intervention in a ‘contemplative’ population. While the participants were largely already meeting the guidelines, it should be noted that this is a minimum level of PA and more is better, and that maintenance of PA is critical in adult and ageing populations. Research priorities for the future includeAssessing what methods for providing PA information/intervention (eg, big screen, leaflet, poster, email, direct conversation) are welcomed by spectators. Investigating whether the spectating experience could be used as a teachable moment to raise awareness of personal PA behaviour, national guidelines and the benefits of PA and influence behavioural change. Further study of spectator PA levels in different contexts, and with a larger and more representative sample, which may allow a better estimation of accrued PA, and potential gender and age differences.

Investigating whether the spectating experience could be used as a teachable moment to raise awareness of personal PA behaviour, national guidelines and the benefits of PA and influence behavioural change. Further study of spectator PA levels in different contexts, and with a larger and more representative sample, which may allow a better estimation of accrued PA, and potential gender and age differences. Using qualitative methods to undertake an in-depth exploration of why exercise/PA is valued or not valued by spectators, and exploring the barriers to and facilitators of active spectating at professional golf tournaments among senior tournament decision makers. Studying opportunities for other sports/events to explore spectator PA. Strengths and limitations This study was conducted with a pragmatic design and approach. Strengths include a novel approach in raising awareness of PA through sport and demonstrating public health benefits of sporting events that have thus far been elusive. It also demonstrated the feasibility of conducting research with spectators at professional sporting events in collaboration with event organisers, governing bodies and athlete ambassadors. Research co-produced in this way may help implementation/scale up and assist impact and future intervention delivery in this manner. It is the first to objectively report PA accrued while spectating, while other findings are consistent with previous work describing spectator attitudes to exercise/PA.15–17

ambassadors. Research co-produced in this way may help implementation/scale up and assist impact and future intervention delivery in this manner. It is the first to objectively report PA accrued while spectating, while other findings are consistent with previous work describing spectator attitudes to exercise/PA.15–17 A number of limitations are evident. Although approximately 600 spectators were approached, those who agreed to wear a pedometer and take part in the study may be more interested in PA and be more physically active than those who declined leading to a selection bias. Observed results may be susceptible to bias; individuals may have modified their responses and behaviours (for example walked more or less) based on what they believe is socially desirable and awareness of their behaviours being observed (Hawthorne effect). Twenty individuals had conclusive proof of pedometer error (for example from GPS/other pedometer), and their step counts were excluded. A smaller number of individuals expressed an opinion that the pedometer had underestimated their step count, but were included due to lack of objective evidence to support, which may have led to an underestimate of their and the observed population’s step count. Step-count data were collected from entry to exit of venue, but did not capture participant PA during the other parts of their day. These limitations and sample size mandate caution in generalising to golf spectators more generally, particularly in different contexts.

of their and the observed population’s step count. Step-count data were collected from entry to exit of venue, but did not capture participant PA during the other parts of their day. These limitations and sample size mandate caution in generalising to golf spectators more generally, particularly in different contexts. Conclusions Encouraging people to be more active more often is a public health imperative. A key element of generating increased PA in relation to a sporting event may be to de-emphasise participation in the sport itself and promote PA more generally. Evidence from this study showed that spectators’ rate ‘exercise/physical activity’ as an important reason for attending the golf tournament and that spectating can provide HEPA. The authors thank Paths for All, the European Tour Golf, and 4sports for their input and collaboration at the Paul Lawrie Matchplay tournament. Tournament host Paul Lawrie, as a player, provided leadership in highlighting potential opportunities for players to promote physical activity among spectators. We also thank Jack Luscombe for his assistance with data collection. Contributors: ADM, KT, CS, PK, LG and NM contributed to the development of the research questions and study design. ADM, KT, CS, SAG and HS collected and extracted the data. All authors developed the first and subsequent drafts of the manuscript. All authors reviewed and approved the final manuscript. Funding: This work was supported by an unrestricted grant from the World Golf Foundation, with support from the Medical Research Council and the Chief Scientist Office.

Contributors: ADM, KT, CS, PK, LG and NM contributed to the development of the research questions and study design. ADM, KT, CS, SAG and HS collected and extracted the data. All authors developed the first and subsequent drafts of the manuscript. All authors reviewed and approved the final manuscript. Funding: This work was supported by an unrestricted grant from the World Golf Foundation, with support from the Medical Research Council and the Chief Scientist Office. Competing interests: ADM and RH received an unrestricted grant from the World Golf Foundation to fund this research. The World Golf Foundation agreed to publish findings whether positive, negative, or no associations or effects were found. RH and ADM are remunerated for clinical work for the European Tour Golf. Patient consent: Obtained. Ethics approval: School of Education, University of Edinburgh. Provenance and peer review: Not commissioned; externally peer reviewed.

Key points We found differential protein expression of alarmins in healthy and diseased human supraspinatus tendons, before and after treatment. The cell types expressing S100A9, interleukin-33 (IL-33) and hypoxia-inducible factor 1α (HIF-1α) included macrophages and tendon stromal cells. S100A9 and HIF-1α may have pro-inflammatory effects in tendon disease, nuclear IL-33 may be protective against pro-inflammatory stimuli and high-mobility group box 1 may have a potential role in tendon healing. Improving understanding of the role of alarmins in tendon disease will facilitate identification of potential therapeutic targets for patients with tendinopathy. Background Tendon disease represents a significant clinical challenge and is associated with substantial disability among ageing and athletic populations.1 Supraspinatus tendinopathy is one of the most common orthopaedic conditions presenting to clinicians with an estimated prevalence of 4%–26%.2 The importance of inflammation in the pathogenesis of tendon disease is a subject of ongoing debate. Over the past two decades, tendon disease has been characterised as a ‘degenerative condition’.3 More recently, the identification of key immune cell populations and signalling molecules as important regulators in the initiation and propagation of other musculoskeletal diseases, including rheumatoid arthritis (RA) and spondyloarthropathy, has led to a re-emergence of interest in the role of inflammation in tendon disease.4–7

ly, the identification of key immune cell populations and signalling molecules as important regulators in the initiation and propagation of other musculoskeletal diseases, including rheumatoid arthritis (RA) and spondyloarthropathy, has led to a re-emergence of interest in the role of inflammation in tendon disease.4–7 The term ‘tendinopathy’ has been popularised to reinforce the notion that the underlying disease mechanisms are not fully understood.8 However, there is growing support for the contribution of immune cells and inflammatory mediators to the development of tendinopathy.9–13 Recent research showed that numbers of macrophages and mast cells are significantly increased in tendinopathic compared with healthy tendon tissues, and that inflammatory and fibrotic cytokines, peptides and growth factors have altered expression profiles in diseased compared with healthy tendons.14 15 We recently reported that inflammatory signatures changed throughout different stages of severity in tissue samples from patients with supraspinatus tendinopathy.13 However, the mechanisms by which inflammation is perceived by resident tendon cells remains poorly understood.

diseased compared with healthy tendons.14 15 We recently reported that inflammatory signatures changed throughout different stages of severity in tissue samples from patients with supraspinatus tendinopathy.13 However, the mechanisms by which inflammation is perceived by resident tendon cells remains poorly understood. Alarmins, endogenous molecules released on tissue damage, are key effectors in the activation of the immune system that may be important in the pathogenesis of tendon disease.16 Alarmins are structurally diverse host proteins with intracellular or intra nuclear functions, which act in both host-defence and tissue repair.17 These molecules ‘alarm’ the immune system by upregulating signalling pathways including nuclear factor-κβ (NF-κβ), interferons (IFNs) and cyclooxygenases, which have downstream effects on tendon repair processes.17 Studies of alarmins in RA have implicated these molecules in the failure of acute inflammatory resolution, resulting in dysregulated processes and development of chronic inflammation.18 There is a paucity of research on alarmins in tendon disease.19 20 To our knowledge, no study has previously investigated alarmin protein expression in non-ruptured diseased supraspinatus tendons before and after treatment using tendon matched healthy control tissues. Moreover, there are little data regarding the potential roles of S100A9 and HMGB1 in tendinopathy.

endon disease.19 20 To our knowledge, no study has previously investigated alarmin protein expression in non-ruptured diseased supraspinatus tendons before and after treatment using tendon matched healthy control tissues. Moreover, there are little data regarding the potential roles of S100A9 and HMGB1 in tendinopathy. The purpose of this study was to investigate expression of three alarmins in human supraspinatus tendons that have been implicated in sustaining chronic inflammation in diseased musculoskeletal tissues: 1) S100A9 2) high-mobility group box 1 (HMGB1) and 3) interleukin-33 (IL-33).18 Although there is a growing list of alarmins in the literature, these are the most thoroughly studied and implicated in other musculoskeletal diseases yet have not been investigated in tendinopathy. Additionally, we investigated expression of hypoxia-inducible factor-1α (HIF-1α), a subunit of HIF-1, an oxygen-dependent heterodimeric transcription factor implicated in pro-inflammatory and profibrotic diseases.21 22 HIF-1 has been implicated as a critical regulator of early tendinopathy but has not yet been investigated in non-ruptured, diseased tendons.20 We investigated protein expression of these mediators in three cohorts of human supraspinatus tissues: 1) healthy ‘control’ tendons, 2) painful diseased tendons (not torn) and 3) tendons from patients who became asymptomatic after subacromial decompression (SAD) treatment. An additional objective was to identify the cell types expressing alarmins in diseased tendon tissues. We hypothesised that expression of S100A9, HMGB, IL-33 and HIF-1α would be increased in painful diseased compared with healthy tendons. We further hypothesised that macrophages and resident tendon cells in diseased tissues would express these proteins.

tify the cell types expressing alarmins in diseased tendon tissues. We hypothesised that expression of S100A9, HMGB, IL-33 and HIF-1α would be increased in painful diseased compared with healthy tendons. We further hypothesised that macrophages and resident tendon cells in diseased tissues would express these proteins. Methods Overview Ethical approval for this study was granted by the local research ethics committee, Oxfordshire REC B references 10/H0402/24, 09/H0605/111 and South Central Oxford B reference 14/SC/0222. Full informed consent according to the Declaration of Helsinki was obtained from all patients. Tissue was obtained from three well-phenotyped patient groups: 1) 6 healthy supraspinatus tendons from patients undergoing shoulder stabilisation, 2) 13 diseased supraspinatus tendons (intact and not torn) from patients with shoulder pain undergoing SAD and 3) 5 pain-free post-treatment supraspinatus tendons from patients who previously had SAD surgery for shoulder impingement (tissue collected 1–3 years post-SAD). Samples were collected by a consultant orthopaedic surgeon during surgery (healthy and diseased) or under local anaesthetic (treated group). All patients included in the study completed the Oxford Shoulder Score (OSS), a validated pain and functional outcome measure prior to tissue collection and/or surgery.23 All tissue samples were processed, embedded, sectioned and stained using immunohistochemistry (IHC) to quantify protein expression of S100A9, HMGB1, IL-33 and HIF-1α. Statistical analysis and sample size justification were determined from previous studies that were sufficiently powered using similar IHC protocols.11 13 Fluorescent immunostaining was also performed on diseased supraspinatus tissues to identify whether macrophages (CD68+ cells) previously identified in diseased tendons13 also expressed HIF-1α, S100A9, IL-33 and HMGB1.

were determined from previous studies that were sufficiently powered using similar IHC protocols.11 13 Fluorescent immunostaining was also performed on diseased supraspinatus tissues to identify whether macrophages (CD68+ cells) previously identified in diseased tendons13 also expressed HIF-1α, S100A9, IL-33 and HMGB1. Patient cohort and tissue collection All patients were recruited from orthopaedic referral clinics where the structural integrity of the supraspinatus tendon was determined ultrasonographically. Samples of healthy supraspinatus tendon tissues were collected from patients recruited to the Nuffield Orthopaedic Centre (NOC) from orthopaedic referral clinics for shoulder instability (table 1). During the surgical procedure a tendon biopsy was collected. Diseased supraspinatus tendon tissues were collected from patients presenting to orthopaedic referral clinics for shoulder pain. Patients presenting to the shoulder clinic had failed non-operative treatments and had experienced pain for a minimum of 6 months. Tendon tissues were collected via ultrasound-guided biopsy prior to surgical subacromial decompression. The biopsy was taken using a Tru-Cut needle 5 to 10 mm posterior to the anterior edge of the supraspinatus tendon. This validated biopsy technique is described in detail elsewhere.24 Post-treatment pain-free supraspinatus tendon tissues were collected from patients who underwent the referral process for chronic shoulder pain/supraspinatus tendinopathy, had a SAD procedure and who were clinically asymptomatic for at least 1 year after surgery. Patients in the pain-free group had significant pain before SAD that resolved post-treatment, evidenced by a post-treatment median OSS of 48 (range, 45–48). Pain-free post-treatment samples were collected 1–3 years after treatment using a percutaneous ultrasound-guided biopsy technique under local anaesthetic. Exclusion criteria for any patient included significant problems in the other shoulder, significant neck problems, RA, systemic inflammatory disease, osteoarthritis, previous shoulder surgery and dual shoulder pathological lesions.

eatment using a percutaneous ultrasound-guided biopsy technique under local anaesthetic. Exclusion criteria for any patient included significant problems in the other shoulder, significant neck problems, RA, systemic inflammatory disease, osteoarthritis, previous shoulder surgery and dual shoulder pathological lesions. Table 1 Patient cohort details for tendon tissue samples used in the current study Tissue group n Median age (years) History Prior treatment Median OSS Healthy 6 24 (range: 19–26) Shoulder instability N/A 33 (range: 11–44) Diseased 13 45 (range: 36–62) Painful tendinopathy N/A 30 (range: 16–36) Pain-free post-treatment 5 57 (range: 43–72) Painful tendinopathy SAD 48 (range: 45–48) N/A, not available; OSS, Oxford Shoulder Score, SAD, subacromial decompression. Tissue processing Samples of tendons were immersed in 10% buffered formalin for 0.5 mm/hour. After formalin fixation, tendon samples were processed in a Leica ASP300S tissue processor and subsequently embedded in paraffin wax; 4 µm tissue sections were cut using a RM2135 microtome (Leica Microsystems) and baked onto adhesive glass slides at 60°C for 30 min and 37°C for 60 min.

uffered formalin for 0.5 mm/hour. After formalin fixation, tendon samples were processed in a Leica ASP300S tissue processor and subsequently embedded in paraffin wax; 4 µm tissue sections were cut using a RM2135 microtome (Leica Microsystems) and baked onto adhesive glass slides at 60°C for 30 min and 37°C for 60 min. Immunohistochemistry Prior to antibody staining and antigen retrieval, slides were baked at 60°C for 60 min. Deparaffinisation and antigen retrieval was performed using an automated PT link (Dako, heat-mediated antigen retrieval at high pH). An Autostainer Link 48 (Dako) was used to perform antibody staining using an EnVision FLEX visualisation system. Details of antibodies and their working dilutions are shown in supplementary table S1. Using protocols provided by the manufacturer, antibody binding was visualised using a FLEX 3,3′-diaminobenzidine (DAB) substrate working solution and haematoxylin counterstain (Dako). After staining, slides were taken through graded industrial methylated spirit and xylene and mounted in Pertex mounting medium (Histolab). Isotype control staining was performed to confirm specificity of staining, whereby the primary antibody was substituted for universal isotype control antibodies. The murine universal isotype control used was a cocktail of mouse IgG1, IgG2a, IgG2b, IgG3 and IgM (Dako IR750). The universal isotype control for rabbits was an immunoglobulin fraction of serum from non-immunised rabbits, solid-phase absorbed (Dako IR600). 10.1136/bmjsem-2017-000225.supp1Supplementary table 1

Immunohistochemistry Prior to antibody staining and antigen retrieval, slides were baked at 60°C for 60 min. Deparaffinisation and antigen retrieval was performed using an automated PT link (Dako, heat-mediated antigen retrieval at high pH). An Autostainer Link 48 (Dako) was used to perform antibody staining using an EnVision FLEX visualisation system. Details of antibodies and their working dilutions are shown in supplementary table S1. Using protocols provided by the manufacturer, antibody binding was visualised using a FLEX 3,3′-diaminobenzidine (DAB) substrate working solution and haematoxylin counterstain (Dako). After staining, slides were taken through graded industrial methylated spirit and xylene and mounted in Pertex mounting medium (Histolab). Isotype control staining was performed to confirm specificity of staining, whereby the primary antibody was substituted for universal isotype control antibodies. The murine universal isotype control used was a cocktail of mouse IgG1, IgG2a, IgG2b, IgG3 and IgM (Dako IR750). The universal isotype control for rabbits was an immunoglobulin fraction of serum from non-immunised rabbits, solid-phase absorbed (Dako IR600). 10.1136/bmjsem-2017-000225.supp1Supplementary table 1 Image acquisition and quantitative analysis for IHC All images were acquired on an inverted microscope using Axiovision software (Zeiss). Twenty images of each stained section were acquired in a systematic manner at ×100 magnification with oil immersion by a single-blinded investigator. If the tissue section was not large enough to capture 20 images, images were taken until the available tissue was exhausted. ImageJ (National Institutes of Health, Bethesda, Maryland, USA) was used to analyse the acquired images. Previously validated algorithms that quantify DAB staining by a colour deconvolution methods were used in analysis.25 26 For each antibody, the colour deconvolution threshold was manually adjusted to best represent immunopositive staining. Depending on expression pattern (cytoplasmic/nuclear or extracellular), specific algorithms were used to quantify immunostaining. For antibodies that showed nuclear and/or cytoplasmic staining, a cell ratio was calculated: positively stained cells/total cells (S100A9, HMGB1 and IL-33). Once positive cells and total cells were counted for each image, the results were summed for each sample to give total positive cells/total cells. For antibodies that showed staining of the extracellular matrix, a ratio of the stained area was calculated: stained area ratio = total positively stained pixels in the image/total pixels of the image (HIF-1α). Once positive pixels and total pixels were quantified for each image, the results were summed to give the stained area ratio: total number of stained pixels/total pixels of all images. For each sample, immunopositive staining was normalised to the number of haematoxylin-counterstained nuclei within the field of view to account for cellularity of the tissue sample.

ntified for each image, the results were summed to give the stained area ratio: total number of stained pixels/total pixels of all images. For each sample, immunopositive staining was normalised to the number of haematoxylin-counterstained nuclei within the field of view to account for cellularity of the tissue sample. Immunofluorescence Staining protocols are adapted from a previously validated protocol.13 Prior to antibody staining and antigen retrieval, slides were baked at 60°C for 60 min. Slides were taken through deparaffinisation and target retrieval steps (high pH heat-mediated antigen retrieval) using an automated PT Link FLEX TRS system (Dako). Slides were placed in a humid chamber throughout staining to prevent drying. Prior to antibody staining, tissues were blocked in 5% normal goat serum (Sigma G9023) in phosphate buffered saline (PBS) for 45 min at room temperature to prevent non-specific antibody binding. Slides were then incubated with the selected primary antibody combination and diluted in 5% normal goat serum in PBS at room temperature for 2 hours (see online supplementary table S1). Isotype controls were completed to confirm staining seen in tendon tissues was specific; the primary antibody was substituted for universal isotype control antibodies available from the manufacturer (Dako). The universal isotype control for mice was a cocktail of mouse IgG1, IgG2a, IgG2b, IgG3 and IgM (Dako IR750). The universal isotype control for rabbits was an immunoglobulin fraction of serum from non-immunised rabbits, solid-phase absorbed (Dako IR600).

sotype control antibodies available from the manufacturer (Dako). The universal isotype control for mice was a cocktail of mouse IgG1, IgG2a, IgG2b, IgG3 and IgM (Dako IR750). The universal isotype control for rabbits was an immunoglobulin fraction of serum from non-immunised rabbits, solid-phase absorbed (Dako IR600). After primary antibody staining, slides were washed three times in phosphate buffered saline with Tween 20 (PBST) solution for 5 min each time. Slides were then incubated with three secondary antibodies: 1) goat antimouse IgG1 (Southern Biotech), 2) Alexa Fluor 568 goat antimouse IgG2a and 3) Alexa Fluor 633 antirabbit IgG (Life Technologies). Secondary antibodies were incubated at 1:200 dilution in PBS containing 5% normal equine serum (Sigma) for 2 hours at room temperature. After secondary antibody staining, slides were washed three times in PBST. POPO-1 nuclear counterstain (Life Technologies) was then applied to all slides diluted in PBS containing 5% Saponin (Sigma) for 20 min at room temperature. Slides were then washed with PBST and incubated in a solution of 0.1% Sudan Black B (Applichem A1407) in 70% ethanol for 5 min to quench tissue autofluorescence. After staining was complete, slides were mounted in VectaShield fluorescent mounting medium (H1000) and a coverslip secured. After mounting, slides were stored in a humid box at 4°C until imaging.

incubated in a solution of 0.1% Sudan Black B (Applichem A1407) in 70% ethanol for 5 min to quench tissue autofluorescence. After staining was complete, slides were mounted in VectaShield fluorescent mounting medium (H1000) and a coverslip secured. After mounting, slides were stored in a humid box at 4°C until imaging. Immunofluorescence image acquisition and analysis Immunofluorescence images were acquired on a Zeiss LSM 710 confocal microscope using a ×40 oil immersion objective following a previously described protocol.13 Two-dimensional reconstructions were created using ZEN 2009 (Zeiss). Statistical analyses Statistical analyses were performed using GraphPad Prism V.7 (GraphPad Software). Normality was tested using Shaprio-Wilk test for normality. Kruskal-Wallis with pairwise post hoc Mann-Whitney U tests were used to test for differences in immunopositive staining between healthy, diseased and post-treatment pain-free sample groups. These tests were performed on DAB-stained immunohistochemical images. Statistical significance was set at p<0.05.

t for normality. Kruskal-Wallis with pairwise post hoc Mann-Whitney U tests were used to test for differences in immunopositive staining between healthy, diseased and post-treatment pain-free sample groups. These tests were performed on DAB-stained immunohistochemical images. Statistical significance was set at p<0.05. Results HIF-1α and S100A9 proteins are increased in diseased supraspinatus tendons HIF-1α staining was increased in diseased compared with healthy supraspinatus tendons (p=0.0002, 6.2-fold increase; figure 1A, B and D). Healthy tendons showed low-level extracellular HIF-1α immunostaining. HIF-1α staining was reduced in post-SAD treatment patients whose pain resolved compared with diseased tissue from symptomatic patients (p=0.03, 3.2-fold decrease (figure 1C)). There was no significant difference in HIF-1α staining between healthy and treated tissue (p=0.97). Diseased tendons showed high-level HIF-1α staining throughout the samples, although nuclear staining was not observed. Pain-free treated tissues showed moderate HIF-1α staining that was also strongly present in the nucleus.

re was no significant difference in HIF-1α staining between healthy and treated tissue (p=0.97). Diseased tendons showed high-level HIF-1α staining throughout the samples, although nuclear staining was not observed. Pain-free treated tissues showed moderate HIF-1α staining that was also strongly present in the nucleus. Figure 1 Representative images of 3,3′-diaminobenzidine immunostaining (brown) for hypoxia-inducible factor 1α (HIF-1α) in (A) healthy, (B) diseased and (C) postsubacromial decompression pain-free human supraspinatus tendons. Nuclear counterstain is haematoxylin (blue). Scale bar=20 µm. (D) Quantitative analysis of immunopositive staining for HIF-1α. Bars represent median value with 95% CIs. Data were analysed using Kruskal-Wallis test with pairwise post hoc Mann-Whitney U test; *p<0.05, **p<0.01, ***p<0.001. S100A9 immunostaining was increased in diseased compared with healthy tendons (p=0.0002, 14.2-fold increase; figure 2A, B and D). Immunopositive staining for S100A9 showed low-level expression among tissue cells in healthy samples. S100A9 staining was reduced in tendons from post-SAD treatment patients whose pain resolved compared with diseased tissue (p=0.03, 4.5-fold decrease (figure 2C). There was no significant difference between healthy and pain-free treated tissue (p=0.97). Treated tissues showed similar characteristics to healthy samples.

S100A9 staining was reduced in tendons from post-SAD treatment patients whose pain resolved compared with diseased tissue (p=0.03, 4.5-fold decrease (figure 2C). There was no significant difference between healthy and pain-free treated tissue (p=0.97). Treated tissues showed similar characteristics to healthy samples. Figure 2 Representative images of 3,3′-diaminobenzidine immunostaining (brown) for S100A9 in (A) healthy, (B) diseased and (C) postsubacromial decompression pain-free human supraspinatus tendons. Nuclear counterstain is haematoxylin (blue). Scale bar=20 µm. (D) Quantitative analysis of immunopositive staining for S100A9. Bars represent median value with 95% CI. Data were analysed using Kruskal-Wallis test with pairwise post hoc Mann-Whitney U test; *p<0.05, **p<0.01, ***p<0.001. IL-33 protein expression is reduced in diseased tendons Healthy and treated tendon tissues showed similar IL-33 protein expression profiles (p>0.99; figure 3A, C and D). IL-33 staining was significantly increased in healthy compared with diseased tendons (p=0.0006, 2.8-fold increase) (figure 3B). IL-33 staining was also significantly increased in pain-free post-treatment compared with diseased tendons (p=0.01, 2.6-fold increase) (figure 2C). Immunopositive staining of IL-33 showed consistent high-level nuclear staining in healthy tendon samples. Diseased tendons showed a significant reduction in nuclear IL-33 staining compared with healthy or treated tissue; there was some cytoplasmic and extracellular staining. Treated tendon samples showed high-level nuclear IL-33 staining.

ing of IL-33 showed consistent high-level nuclear staining in healthy tendon samples. Diseased tendons showed a significant reduction in nuclear IL-33 staining compared with healthy or treated tissue; there was some cytoplasmic and extracellular staining. Treated tendon samples showed high-level nuclear IL-33 staining. Figure 3 Representative images of 3,3′-diaminobenzidine immunostaining (brown) for interleukin-33 (IL-33) in (A) healthy, (B) diseased and (C) postsubacromial decompression pain-free human supraspinatus tendons. Nuclear counterstain is haematoxylin (blue). Scale bar=20 µm. (D) Quantitative analysis of immunopositive staining for IL-33. Bars represent median value with 95% CIs. Data were analysed using Kruskal-Wallis test with pairwise post hoc Mann-Whitney U test; *p<0.05, **p<0.01, ***p<0.001. HMGB1 protein is increased in pain-free post-treatment tendons HMGB1 expression was increased in post-treatment tendon tissues compared with healthy and diseased tendons (figure 4A-D). There was no significant difference in HMGB1 staining between healthy and diseased tendon tissues (p>0.99). Nuclear staining of HMGB1 was increased in pain-free post-treatment tissues compared with both healthy and painful diseased tissues (p=0.006 for both comparisons, 6.6-fold increase from healthy tissue, 5.3-fold increase from diseased tissue). HMGB1 immunostaining was primarily localised to the nuclear and perinuclear region in the tendon tissues studied.

sed in pain-free post-treatment tissues compared with both healthy and painful diseased tissues (p=0.006 for both comparisons, 6.6-fold increase from healthy tissue, 5.3-fold increase from diseased tissue). HMGB1 immunostaining was primarily localised to the nuclear and perinuclear region in the tendon tissues studied. Figure 4 Representative images of 3,3′-diaminobenzidine immunostaining (brown) for high-mobility group box 1 (HMGB1) in (A) healthy, (B) diseased and (C) postsubacromial decompression pain-free human supraspinatus tendons. Nuclear counterstain is haematoxylin (blue). Scale bar=20 µm. (D) Quantitative analysis of immunopositive staining for HMGB1. Bars represent median value with 95% CIs. Data were analysed using Kruskal-Wallis test with pairwise post hoc Mann-Whitney U test; *p<0.05, **p<0.01, ***p<0.001. Macrophages and tendon cells express alarmins Given that macrophages have previously been identified in samples of diseased human supraspinatus tendons, we sought to identify if these CD68+ cells expressed alarmins and HIF-1α. Confocal images illustrating staining for HIF-1α, IL-33, S100A9, HMGB1 and CD68 are shown in figures 5 and 6. Costaining revealed that HIF-1α, S100A9 and IL-33 were expressed by CD68+ cells (macrophages) and CD68− cells, likely tendon stromal cells. HMGB1 was predominantly expressed by CD68− cells. The intensity of HMGB1 staining appeared to be more marked in perivascular regions (figure 6).

CD68 are shown in figures 5 and 6. Costaining revealed that HIF-1α, S100A9 and IL-33 were expressed by CD68+ cells (macrophages) and CD68− cells, likely tendon stromal cells. HMGB1 was predominantly expressed by CD68− cells. The intensity of HMGB1 staining appeared to be more marked in perivascular regions (figure 6). Figure 5 Representative confocal immunofluorescent images showing antibody labelling for macrophages (CD68, green), hypoxia-inducible factor 1α (HIF-1α) (red) and interleukin-33 (IL-33) (purple) in diseased human supraspinatus tendon tissue. Nuclear counterstain was POPO-1 (cyan). Scale bar=20 µm. Figure 6 Representative confocal immunofluorescent images showing antibody labelling for macrophages (CD68, green), high-mobility group box 1 (HMGB1) (red) and S100A9 (purple) in diseased human supraspinatus tendon tissue. Nuclear counterstain was POPO-1 (cyan). Blood vessels are circled red. Scale bar=20 µm.

Figure 5 Representative confocal immunofluorescent images showing antibody labelling for macrophages (CD68, green), hypoxia-inducible factor 1α (HIF-1α) (red) and interleukin-33 (IL-33) (purple) in diseased human supraspinatus tendon tissue. Nuclear counterstain was POPO-1 (cyan). Scale bar=20 µm. Figure 6 Representative confocal immunofluorescent images showing antibody labelling for macrophages (CD68, green), high-mobility group box 1 (HMGB1) (red) and S100A9 (purple) in diseased human supraspinatus tendon tissue. Nuclear counterstain was POPO-1 (cyan). Blood vessels are circled red. Scale bar=20 µm. Discussion This study investigates protein expression in diseased tendon tissues before and after treatment and reveals pro-inflammatory roles for HIF-1α and S100A9 in tendinopathy. Additionally, IL-33 has a potentially protective role in healthy tendons that is lost with the onset of tendon disease. These findings support known mechanisms of chronic inflammation and fibrosis identified in other musculoskeletal diseases where abnormal S100A9, IL-33 and HMGB1 expression has been implicated in the initiation and propagation of various chronic inflammatory and autoimmune disorders.18 27–29 HMGB1 immunostaining was increased in post-treatment pain-free tendon tissues, suggesting HMGB1 may play a potential role in tissue healing.

diseases where abnormal S100A9, IL-33 and HMGB1 expression has been implicated in the initiation and propagation of various chronic inflammatory and autoimmune disorders.18 27–29 HMGB1 immunostaining was increased in post-treatment pain-free tendon tissues, suggesting HMGB1 may play a potential role in tissue healing. Previous studies have illustrated that inflammation can become dysregulated over time resulting in fibrotic repair.13 27 30 31 There is increasing evidence that activation of the immune system after tissue injury is, in part, due to the interactions of alarmins released from necrotic cells with their associated toll-like receptors (TLRs).32 Alarmins have known intracellular and extracellular roles in several inflammatory signalling pathways and have been implicated in affecting resident cell phenotype.18 In this study, we identified S100A9, HIF-1α and IL-33 protein expression differed between healthy and diseased supraspinatus tendon tissues. Our results provide some support for the concept that dysregulation of these mediators may prime tissues to increasingly react to future inflammatory stimuli.18 27–29 33

notype.18 In this study, we identified S100A9, HIF-1α and IL-33 protein expression differed between healthy and diseased supraspinatus tendon tissues. Our results provide some support for the concept that dysregulation of these mediators may prime tissues to increasingly react to future inflammatory stimuli.18 27–29 33 HIF-1α immunostaining was significantly increased in diseased compared with healthy and treated tendon tissues. Healthy and treated tendons showed similar low-level immunostaining, although post-treatment tendons showed moderately increased levels of immunostaining than healthy tendons. These results are consistent with previous findings of HIF-1α in chronic inflammatory/autoimmune disorders and previous studies in tendon tissue.13 20 Stabilisation of HIF-1α with subsequent upregulation of HIF-1 is known to stimulate the release of alarmins S100A9, HMGB1 and IL-33 in inflammatory disorders.34 Upregulation of HIF-1 is also known to stimulate expression of vascular endothelial growth factor, which has shown to form a positive feedback loop by further stabilising HIF-1α in all cells. Extracellular HIF-1α has numerous known pro-inflammatory effects in addition to facilitating alarmin secretion: it protects myeloid cells against apoptosis and it drives macrophages towards an M1 phenotype.35 36 HIF-1 activation also leads to increased expression of profibrotic mediators.28 HIF-1α upregulation likely plays similar roles in tendon tissue: our results show that S100A9 and IL-33 dysregulation occurred alongside increased HIF-1α staining.

ls against apoptosis and it drives macrophages towards an M1 phenotype.35 36 HIF-1 activation also leads to increased expression of profibrotic mediators.28 HIF-1α upregulation likely plays similar roles in tendon tissue: our results show that S100A9 and IL-33 dysregulation occurred alongside increased HIF-1α staining. S100A9 staining was significantly increased in diseased tendon tissues. Previous studies have shown that extracellular S100A9 induces pro-inflammatory responses, including facilitating cell recruitment and promoting monocyte differentiation.37 38 S100A9 is implicated in promoting NF-κβ expression, resulting in increased secretion of tumour-necrosis factor-α (TNF-α) and interleukin-1β (IL-1β).38 39 Overexpression of S100A9 in ‘stressed’ tissues has been shown to result in a positive feedback mechanism resulting in downstream expression of TNF-α.38 40 Continued insults to tissues can induce cell phenotype changes over time and predispose tissues to failed healing: S100A9 from myeloid cells promotes further leucocyte recruitment, priming myeloid cells for more intense inflammatory responses following injuries.38 40 S100A9 has not previously been investigated in tendon tissues. Increased S100A9 protein in diseased tendons seen in this study is analogous to chronic inflammatory and autoimmune disorders. These findings potentially implicate S100A9 as an important initiator and propagator of chronic inflammation in tendon diseases.

0 S100A9 has not previously been investigated in tendon tissues. Increased S100A9 protein in diseased tendons seen in this study is analogous to chronic inflammatory and autoimmune disorders. These findings potentially implicate S100A9 as an important initiator and propagator of chronic inflammation in tendon diseases. IL-33 immunostaining was significantly reduced in diseased compared with healthy and treated tendons. These results are consistent with a proposed protective role of nuclear IL-33 against pro-inflammatory stimuli.41 Physical interaction of nuclear IL-33 with NF-κβ sequesters NF-κβ and curbs any incoming pro-inflammatory or fibrotic signals from other receptors.41 42 Extracellular IL-33 can stimulate interferon-γ (INF-γ) production and lead to upregulation of NF-κβ signalling pathways in infiltrating cells.41 However, the pro-inflammatory ‘alarmin’ role of extracellular IL-33 does not appear to be mutually exclusive with its protective nuclear function in resident cells.41 Our results suggest that the protective nuclear IL-33 is reduced in the resident cells of diseased tissue.

β signalling pathways in infiltrating cells.41 However, the pro-inflammatory ‘alarmin’ role of extracellular IL-33 does not appear to be mutually exclusive with its protective nuclear function in resident cells.41 Our results suggest that the protective nuclear IL-33 is reduced in the resident cells of diseased tissue. HMGB1 has become categorised as an alarmin due to its involvement in mobilisation and activation of immune cells; its overexpression has been implicated in chronic inflammatory disorders.18 43 Stimuli for the secretion of HMGB1 from monocytes, macrophages and dendritic cells include cellular stress, pathogen-associated molecular patterns and cytokines, including TNF-α, IL-1 and INF-γ.17 The primary receptors for extracellular HMGB1 are receptor for advanced glycation end products (RAGE), TLR2 and TLR4. Activation of RAGE by HMGB1 promotes chemotaxis and cytokine production through the upregulation NF-κβ pathways.17 18 43 HMGB1 has also been suggested to have regenerative potential. HMGB1 can induce migration of stem cells towards inflamed regions, promoting repair and regeneration.44

d products (RAGE), TLR2 and TLR4. Activation of RAGE by HMGB1 promotes chemotaxis and cytokine production through the upregulation NF-κβ pathways.17 18 43 HMGB1 has also been suggested to have regenerative potential. HMGB1 can induce migration of stem cells towards inflamed regions, promoting repair and regeneration.44 HMGB1 has not been previously investigated in diseased human tendon tissues. We hypothesised that HMGB1 staining would be increased in diseased tissues due to its pro-inflammatory and pathogenic role in some musculoskeletal disorders. Contrary to our hypothesis, HMGB1 staining was reduced in diseased tendons compared with both healthy and pain-free post-treatment tendons. These findings suggest HMGB1 may have potential role in tendon repair and regeneration, analogous to reports of HMGB1 function in diseased cardiac muscle.44–46 A limitation of this study was the relatively small number of patients studied in healthy and post-treatment groups. Also age-related changes may also contribute to differences in expression between healthy and diseased tendon tissues. Our healthy tendon samples were collected from younger patients and this may have influenced the findings from this study. A significant advantage of this study is the use of supraspinatus tendons in both the diseased and healthy control groups, avoiding any biological variations seen in tendons with different structure and function. Improving understanding of the role of alarmins in tendon disease will facilitate identification of potential therapeutic targets for patients with tendinopathy.

praspinatus tendons in both the diseased and healthy control groups, avoiding any biological variations seen in tendons with different structure and function. Improving understanding of the role of alarmins in tendon disease will facilitate identification of potential therapeutic targets for patients with tendinopathy. Conclusions In summary, we found differential protein expression of alarmins in healthy and diseased human supraspinatus tendons, before and after treatment. Specifically, S100A9 and HIF-1α may have pro-inflammatory effects in tendon disease, nuclear IL-33 may be protective against pro-inflammatory stimuli and HMGB1 may have a potential role in tendon healing. The cell types expressing S100A9, IL-33 and HIF-1α included macrophages and tendon stromal cells. Pathological activation of tendon cells through alarmin involvement may sustain chronic inflammation in tendinopathy. We propose that tendinopathy is an alarmin-mediated pathology initiated and propagated in part by increased expression of HIF-1α. Alarmin activity may facilitate recruitment of inflammatory cells in the early stages of tendon disease. Contributors: Study concept and design: MJM, SGD, AJC, SJBJ. Acquisition of data: MJM, SGD. Analysis and interpretation of data: MJM, SGD, SJBS, AJC. Drafting of the manuscript: all authors. Critical revision of the manuscript and approval of final version: all authors. Statistical analysis: MJM, SGD. Obtained funding: AJC. Administrative, technical or material support: KW, BW.

Contributors: Study concept and design: MJM, SGD, AJC, SJBJ. Acquisition of data: MJM, SGD. Analysis and interpretation of data: MJM, SGD, SJBS, AJC. Drafting of the manuscript: all authors. Critical revision of the manuscript and approval of final version: all authors. Statistical analysis: MJM, SGD. Obtained funding: AJC. Administrative, technical or material support: KW, BW. Funding: MJM acknowledges the Harvard University Francis H Burr, Class of 1909 Prize Fund and the National Collegiate Athletic Association for financial support. SGD and SJBS are funded by Arthritis Research UK grants 20 506 and 20 087, respectively. We also acknowledge the Oxford NIHR Biomedical Research Unit. Competing interests: None declared. Ethics approval: Ethical approval for this study was granted by the local research ethics committee, Oxfordshire REC B references 10/H0402/24, 09/H0605/111, and South Central Oxford B reference 14/SC/0222. Provenance and peer review: Not commissioned; externally peer reviewed

What are the new findings? An acute bout of interval exercise increases the ability of circulating angiogenic cells to form colonies, which may increase the capacity for vascular repair in postmenopausal women. The government-recommended guidelines of 30 min of moderate-intensity continuous exercise did not acutely impact endothelial function or factors associated with repair in postmenopausal women. How might it impact on clinical practice in the near future? The findings provide support for individualised exercise prescription for postmenopausal women. Healthcare professionals may advise postmenopausal women to exercise in short bursts of activity followed by active rest periods for vascular benefits.

, hyperthermia or energy imbalance likely induce cytokine production during exercise through catecholamines, endotoxin, alarmins, ATP and proinflammatory cytokines themselves.57 58 SET is likely to contribute to a reduction of large cytokine response in EIMD and DOMS independent of physical performance/training status. Conclusion SET administered orally 72 hours before and 72 hours post exhaustive eccentric exercise resulted in higher maximal concentric strength and lower PIP in subjects who were less experienced in resistance training and in significant favourable effects on anti-inflammatory and other biomarkers in all subjects. The potential role of SET in managing EIMD and DOMS across different training groups needs further investigation. Acknowledgements: The authors would like to thank the coaches of the Sportschule Puch Fuerstenfeldbruck, and in particular Lorenz Westner for their assistance in conducting this trial. Contributors: TM: study conception and design, aquisition of data, analysis and interpretation of data. GL: analysis and interpretation of data, drafting of manuscript. CR: study conception and design, analysis and interpretation of data, drafting of manuscript. SR: study conception and design, analysis and interpretation of data. JCV: statistical analysis and interpretation of data. EP: acquisition of data. HP: main investigator, study conception and design, aquisition of data, analysis and interpretation of data. Every contributing author has reviewed the paper and had the final authority over the content.

What are the new findings? This study confirms a substantial and significant effect of systemic enzyme therapy (SET) on fatigue, muscle soreness and damage, as well as immunological and metabolic biomarkers, in male sportsmen with medium performance level (mostly runners and general athletes). Muscle soreness and maximal strength were not improved in those subjects with a higher level of strength training at baseline. Use of SET showed a significant reduction in inflammatory biomarkers in sportsmen across all training levels, indicating an application for SET in supporting normal inflammatory processes for muscle recovery. Clinicians may recommend the use of SET for mediating muscle fatigue, reducing soreness and attenuating potential muscle damage in endurance athletes. Introduction Exercise-induced muscle damage (EIMD) and its most common symptom, delayed onset of muscle soreness (DOMS), impact an athlete’s training frequency and performance. Strenuous exercise, acute or postsurgical trauma and certain diseases can all be sources of skeletal muscle injury. Regardless of the type of injury, the general injury and repair mechanism are similar1 2 and have been well characterised in EIMD.3–11

eness (DOMS), impact an athlete’s training frequency and performance. Strenuous exercise, acute or postsurgical trauma and certain diseases can all be sources of skeletal muscle injury. Regardless of the type of injury, the general injury and repair mechanism are similar1 2 and have been well characterised in EIMD.3–11 Inflammation contributes to fibrosis12 and causes pain that may impair skeletal muscle function.11 Therefore, it has been common practice to reduce inflammation with drugs, such as COX-2 inhibitors. The problem with this approach is that while inflammation causes further injury to muscles,1 2 13 14 preventing inflammation may hinder recovery.1 15–17 As a result, current treatment options for inflammation are not necessarily effective and, in some cases, they may be unsafe.

ion with drugs, such as COX-2 inhibitors. The problem with this approach is that while inflammation causes further injury to muscles,1 2 13 14 preventing inflammation may hinder recovery.1 15–17 As a result, current treatment options for inflammation are not necessarily effective and, in some cases, they may be unsafe. Systemic enzyme therapy (SET) i allows inflammatory processes to progress naturally and this overcomes the problem of preventing inflammation in a manner that may hinder recovery. The antioxidant rutin reduces oxidative stress during inflammation.18 19 Orally administered proteolytic enzymes, also called proteinases, are mainly absorbed in the small intestine and are active in body fluids and tissues as free and bound proteinases, despite their low concentrations (pmol–nmol).20 Trypsin and bromelain share two main biological activities with other proteinases: (i) they degrade proteins by their proteolytic activity which cleaves peptide bonds at specific sites, both in digestion and as markers of cell destruction and inflammation, and (ii) they bind to specific (eg, α2-antitrypsin) or unspecific (eg, α2-macroglobulin) antiproteinases to prevent uncontrolled protein degradation.

they degrade proteins by their proteolytic activity which cleaves peptide bonds at specific sites, both in digestion and as markers of cell destruction and inflammation, and (ii) they bind to specific (eg, α2-antitrypsin) or unspecific (eg, α2-macroglobulin) antiproteinases to prevent uncontrolled protein degradation. Accumulating evidence points to a role of protease-activated receptor 2, expressed on T cells, eosinophils, neutrophils and mast cells, in the regulation of inflammation and immune function.20–23 Further, both trypsin and bromelain form complexes with α2-macroglobulin leading to a conformational change that exposes receptor recognition sites in each of its four subunits.20 24 These complexes are recognised by low-density lipoprotein receptor-related protein and cell surface glucose-regulated protein (GRP78) receptors24 on blood and immune cell surfaces resulting in modification of cellular activities25 and rapid elimination by hepatocytes.20 26 During inflammation the complex of protease and antiprotease is subject to further modifications.20 Oxidation of the proteinase antiproteinase complex may serve as a switch mechanism that downregulates the progression of acute inflammation by sequestering TNF-α, interleukin (IL) 2 and IL-6, while upregulating the development of tissue repair processes by releasing bFGF, b-NGF, PDGF and TGF-β.20 Thus, SET may affect EIMD and DOMS by balancing the inflammatory response to injury.

s a switch mechanism that downregulates the progression of acute inflammation by sequestering TNF-α, interleukin (IL) 2 and IL-6, while upregulating the development of tissue repair processes by releasing bFGF, b-NGF, PDGF and TGF-β.20 Thus, SET may affect EIMD and DOMS by balancing the inflammatory response to injury. The influence of oral proteinases on EIMD was investigated in several clinical trials focusing on pain and/or muscle function and strength.27–29 Bromelain and other proteinases may reduce muscle inflammation after EIMD,1 but controversy persists.27 28 The aim of the current trial was to investigate the effects of SET before and after exhaustive eccentric exercise on functional and biochemical parameters of EIMD and DOMS in male sportsmen who had a medium level of performance.

ther proteinases may reduce muscle inflammation after EIMD,1 but controversy persists.27 28 The aim of the current trial was to investigate the effects of SET before and after exhaustive eccentric exercise on functional and biochemical parameters of EIMD and DOMS in male sportsmen who had a medium level of performance. Methods Trial design This was designed as a prospective, randomised, double-blinded, placebo-controlled, two-stage trial. Stage I was a crossover with a washout phase of 21 days between phases; stage II was continued as parallel group comparison. Subjects were randomised to either SET or placebo (stage I: n=2×14, crossover; stage II: n=2×22, parallel group) and involved parties were blinded, except for the person responsible for interim (stage I) and final confirmatory analyses. Administration of medication began 72 hours before an exhaustive eccentric exercise day and continued for 72 hours following the muscle damaging exercise. Each subject completed an activity diary and a food frequency questionnaire. The clinical trial complied with the Declaration of Helsinki and was approved by the local health authority and ethical committee. It was registered at ClinicalTrials.gov (NCT01845558).

nued for 72 hours following the muscle damaging exercise. Each subject completed an activity diary and a food frequency questionnaire. The clinical trial complied with the Declaration of Helsinki and was approved by the local health authority and ethical committee. It was registered at ClinicalTrials.gov (NCT01845558). Subjects Subjects were aged 20–50 years, had a body mass index between ≥20 kg/m2 and ≤32 kg/m2 and had a moderate performance level and moderate strength ability defined by a medium concentric strength ability of 150–300 Nm (newton metre) peak torque maximum (PTM). Nutritional status was assessed via questionnaire at screening; subjects were asked not to change nutritional habits; and standardised meals were offered during the exercise day. Subjects were not to practise any physical activity (including driving to work by bicycle) during the study. All physical activities were documented. Subjects were also instructed to avoid activities after the exercise such as massages or hot bathing or showering. Relevant exclusion criteria were history or presence of any medical disorder, intake of anti-inflammatory medication, food supplements or use of other procedures directly affecting muscle function or performance within 4 weeks prior to or during trial. Intake of analgesic medication or consumption of alcohol was not allowed 24 hours prior and until 72 hours after exercise.

any medical disorder, intake of anti-inflammatory medication, food supplements or use of other procedures directly affecting muscle function or performance within 4 weeks prior to or during trial. Intake of analgesic medication or consumption of alcohol was not allowed 24 hours prior and until 72 hours after exercise. Sample size Non-parametric sample size calculation within the framework of a multiple outcome approach30–33 was performed applying the validated software Nnpar V.1.0 (IDV, Gauting, Germany). With the sample size of 30 subjects, the power calculation for stage I indicated a 64% chance (power of 90%) to detect a ‘medium-sized’ group difference with respect to the multidimensional test. If neither success nor futility was formally determined after stage I, a subsequent stage II could be planned based on the results of stage I (sample size reassessment with adaptive design features).34 35 Recalculated total sample size for stage II was 2×22 subjects (one phase, no crossover design). Interventions Subjects were recruited through a telephone questionnaire and invited to the screening visit 2–4 weeks before exercise to give informed consent and confirm eligibility. Demographic data, medical history, physical activity (Freiburger Activity Questionnaire), nutritional performance status, vital signs and blood samples for routine parameters were collected.

hone questionnaire and invited to the screening visit 2–4 weeks before exercise to give informed consent and confirm eligibility. Demographic data, medical history, physical activity (Freiburger Activity Questionnaire), nutritional performance status, vital signs and blood samples for routine parameters were collected. After a 5 min warm-up period on an ergometer, the maximal concentric strength of the quadriceps femoris muscle of the stronger leg, determined at screening, was measured on a desmodrom (Fa. Schnell). Three repetitions of 20 maximal concentric contractions were performed (60 s at 20 oscillations per minute) with 1 min passive recovery between sets. The highest value was used for analysis. For each measurement, peak torque and angle of peak torque were documented. On day 4 of the trial (visit 1), EIMD was induced by exhaustive eccentric exercise according to McLeay.36 Maximal eccentric, isokinetic loading of the right and left quadriceps femoris was determined by using a desmodrom (Fa. Schnell) as before, but with the difference of eccentric, not concentric, exercise. The same observer motivated all subjects following a standardised protocol to avoid interobserver bias.

to McLeay.36 Maximal eccentric, isokinetic loading of the right and left quadriceps femoris was determined by using a desmodrom (Fa. Schnell) as before, but with the difference of eccentric, not concentric, exercise. The same observer motivated all subjects following a standardised protocol to avoid interobserver bias. Trial medication All subjects received four tablets of SET or placebo of identical shape and colour three times a day starting 72 hours before exhaustive eccentric exercise day and for 72 hours post exercise. Trial medication was taken on an empty stomach 30 min before meals with 250 mL water. Date and time of intake and time of subsequent meals were documented in a diary. Residual tablets were counted to control for compliance. Outcome measures The maximal concentric strength of the M. Quadriceps femoris of both legs was assessed at screening (2–4 weeks before visit 1), pre visit 1 (day 0, before start of trial medication), at visit 1 (day 4 of administration of trial medication), immediately before (pre) and immediately after exhaustive eccentric exercise (0 hour), as well as after 3 hours and 6 hours, and at day 5 (24 hours), day 6 (48 hours) and day 7 (72 hours). PTM and angle of peak torque were measured as parameters of maximal concentric strength. Pressure-induced pain (PIP) was assessed by the same observer for all subjects using a 1 cm2 metal disk against the middle of the muscle belly. Pressure was constantly increased until it became unpleasant. PIP was assessed at the indicated time points and the mean of three tests was used for analysis.

Outcome measures The maximal concentric strength of the M. Quadriceps femoris of both legs was assessed at screening (2–4 weeks before visit 1), pre visit 1 (day 0, before start of trial medication), at visit 1 (day 4 of administration of trial medication), immediately before (pre) and immediately after exhaustive eccentric exercise (0 hour), as well as after 3 hours and 6 hours, and at day 5 (24 hours), day 6 (48 hours) and day 7 (72 hours). PTM and angle of peak torque were measured as parameters of maximal concentric strength. Pressure-induced pain (PIP) was assessed by the same observer for all subjects using a 1 cm2 metal disk against the middle of the muscle belly. Pressure was constantly increased until it became unpleasant. PIP was assessed at the indicated time points and the mean of three tests was used for analysis. Biomarkers Biomarkers of muscle metabolism and damage, inflammatory and immune response, and redox status were determined. Samples were taken at individual time points, stored frozen below −80°C, except for lactate and natural killer (NK) cell activity, and analysed batchwise for each subject after trial completion according to manufacturer’s instructions, if not otherwise indicated. Creatine kinase and lactate dehydrogenase as markers of muscle damage were determined at Medizinisches Versorgungszentrum Leinfelden for samples taken pre-exercise, and at 0, 3, 6, 24, 48 and 72 hours after exercise. Lactate was determined pre-exercise, and 10 and 30 min after exercise taken by capillary blood samples by trial site. Inflammatory response was investigated as IL-6 (pre-exercise, and 0, 3 and 24 hours after exercise) (Quantikine High Sensitivity ELISA; R&D Systems) and prostaglandin E derived from cyclooxygenase 2 (pre-exercise, and 3, 6 and 24 hours) (Prostaglandin E Metabolite EIA kit; Cayman Chemical) at BioTeSys, Esslingen. Immune function was investigated by IL-2 inducible NK cell activity (pre, 3 hours, 24 hours) at the Institut für Medizinische Diagnostik, Berlin.37 38 Total antioxidative status (TAS) and total oxidative status (TOS) (pre, 3 hours, 6 hours, 24 hours) were assessed using a photometric test assay (ImAnOx [TAS/TAC] Kit; perOS [TOS/TOC] Kit; Immundiagnostik AG, Bensheim, Germany).

activity (pre, 3 hours, 24 hours) at the Institut für Medizinische Diagnostik, Berlin.37 38 Total antioxidative status (TAS) and total oxidative status (TOS) (pre, 3 hours, 6 hours, 24 hours) were assessed using a photometric test assay (ImAnOx [TAS/TAC] Kit; perOS [TOS/TOC] Kit; Immundiagnostik AG, Bensheim, Germany). Safety assessments Blood samples for routine laboratory tests, including haemogram, liver enzymes, lipids, glucose, uric acid and creatinine, were taken at screening, before, and 24 and 72 hours after exercise and analysed by Synlab Medizinisches Versorgungszentrum Leinfelden. Vital signs were determined at all visits. Concomitant medication and adverse events were documented. Statistical analysis As no single measure captures the multidimensional nature of recovery from EIMD, the combination of single efficacy endpoints reduction of maximal strength (PTM) and PIP was chosen to be evaluated by a multivariate, directional test approach in stage I of the trial.39

Safety assessments Blood samples for routine laboratory tests, including haemogram, liver enzymes, lipids, glucose, uric acid and creatinine, were taken at screening, before, and 24 and 72 hours after exercise and analysed by Synlab Medizinisches Versorgungszentrum Leinfelden. Vital signs were determined at all visits. Concomitant medication and adverse events were documented. Statistical analysis As no single measure captures the multidimensional nature of recovery from EIMD, the combination of single efficacy endpoints reduction of maximal strength (PTM) and PIP was chosen to be evaluated by a multivariate, directional test approach in stage I of the trial.39 The multiple level alpha of the trial (multiple level of significance) was defined as α=0.025 (one sided). The confirmatory analyses were performed with the intention-to-treat population. Missing values were replaced by last observation carried forward technique. Results are given as p values and effect size measures with their CIs (Mann-Whitney (MW) statistic as corresponding effect size measure of the Wilcoxon-Mann-Whitney test). The traditional benchmarks for the MW effect size measure are as follows: 0.29 = large inferiority, 0.36 = medium inferiority, 0.44 = small inferiority, 0.50 = equality, 0.56 = small superiority, 0.64 = medium superiority and 0.71 = large superiority.

orresponding effect size measure of the Wilcoxon-Mann-Whitney test). The traditional benchmarks for the MW effect size measure are as follows: 0.29 = large inferiority, 0.36 = medium inferiority, 0.44 = small inferiority, 0.50 = equality, 0.56 = small superiority, 0.64 = medium superiority and 0.71 = large superiority. Multiple a priori ordered hypothesis testing was performed as a two-stage procedure30 40 41 including the possibility to stop the trial after stage I due to success (proof of efficacy) (p1≤α1=0.0102) and futility (p1≥α0=0.5) or to plan stage II of the trial based on results of stage I, including reassessment of sample size and finalisation of the trial design. This procedure based on Fisher’s combination test shows only negligible loss in test power compared with fixed sample size trials42 and uses the adjusted value α1=0.0102 as a critical value for the test of stage I.34 Recalculated total sample size for stage II was 2×22 subjects with no crossover design. Results Trial population In stage I, 30 subjects were enrolled between 20 April 20 and 1 August 2013, and 28 subjects (safety population) were treated in a crossover design. Per protocol (PP) analysis consisted of 26 subjects (figure 1). Figure 1 Disposition of subjects to stages I and II and different analysis populations. AE, adverse event; ITT, intent-to-treat population; n, number; PP, per protocol population; SAF, safety population; SET, systemic enzyme therapy (Wobenzym).

Results Trial population In stage I, 30 subjects were enrolled between 20 April 20 and 1 August 2013, and 28 subjects (safety population) were treated in a crossover design. Per protocol (PP) analysis consisted of 26 subjects (figure 1). Figure 1 Disposition of subjects to stages I and II and different analysis populations. AE, adverse event; ITT, intent-to-treat population; n, number; PP, per protocol population; SAF, safety population; SET, systemic enzyme therapy (Wobenzym). In stage II, a total of 44 subjects were enrolled between 15 May and 27 July 2014. The safety population was comprised of all enrolled and treated subjects (n=44). The PP population (41 subjects; SET: 20; placebo: 21) resulted from exclusion of two subjects due to protocol violation (timing of PTM >20% and severe violation of screening PTM) (figure 1). Stage I The PTM and PIP results for both treatment groups (all subjects pooled) compared with mean baseline values are provided in figure 2. Figure 2 Stage I results for peak torque maximum and pressure-induced pain in response to exhaustive eccentric exercise in all subjects. Mean values of peak torque maximum (Nm) (A) and pressure-induced pain (kg/cm2) (B) and 95% CIs for all subjects of the total blinded review population of stage I (red circles and lines) compared with the mean baseline peak torque maximum of all subjects (blue line).

nse to exhaustive eccentric exercise in all subjects. Mean values of peak torque maximum (Nm) (A) and pressure-induced pain (kg/cm2) (B) and 95% CIs for all subjects of the total blinded review population of stage I (red circles and lines) compared with the mean baseline peak torque maximum of all subjects (blue line). As expected, PTM is reduced in the ‘acute’ phase at 3 hours (3.4%; 201 Nm) and 6 hours (4.8%; 198 Nm) after exhaustive eccentric exercise compared with baseline (100%; 208 Nm) (figure 2A). However, during'recovery' phase (24–48 hours), the average PTM returns to 205 Nm at 24 hours and to 209 Nm at 48 hours, the level before exercise. Thus, the potential to discover differences between two treatments is substantially reduced beyond 24 hours. PIP measured by algometry (both treatment groups combined) resulted in a reduction of pain threshold after exhaustive eccentric exercise in both phases (figure 2B). The PTM results in stage I (phase 1) with SET or placebo are shown in figure 3. Exhaustive eccentric exercise led to a reduction in PTM in the SET group of 2.8% at 3 hours and 1.5% at 6 hours compared with placebo of 6.2%, 10% and 5.7% at 3, 6 and 24 hours, respectively. Physical performance returned to baseline at 24 hours with SET but not until 48 hours with placebo. The associated MW effect size indicated more than ‘small’ superiority of the SET group (MW=0.6153, p=0.0332) compared with placebo.

analysis and interpretation of data. JCV: statistical analysis and interpretation of data. EP: acquisition of data. HP: main investigator, study conception and design, aquisition of data, analysis and interpretation of data. Every contributing author has reviewed the paper and had the final authority over the content. Competing interests: GL works among several others as consultant for Mucos Pharma GmbH & Co. KG. Mucos Pharma GmbH & Co. KG has contributed the product needed for intervention and funded the trial with SR as company representative. Provenance and peer review: Not commissioned; externally peer reviewed. i Wobenzym plus contains 67, 5–76, 5 mg bromelain (standardised to 450 FlP units); 32–48 mg trypsin (standardised to 24 μkat) and 100 mg rutoside 3 H2O, and is manufactured by Mucos Pharma GmbH & Co. KG, Berlin.

6 hours compared with placebo of 6.2%, 10% and 5.7% at 3, 6 and 24 hours, respectively. Physical performance returned to baseline at 24 hours with SET but not until 48 hours with placebo. The associated MW effect size indicated more than ‘small’ superiority of the SET group (MW=0.6153, p=0.0332) compared with placebo. Figure 3 Peak torque maximum in response to exhaustive eccentric exercise in subjects administered SET or placebo. Mean changes from baseline of peak torque maximum [Nm] and 95% CIs for the intent-to-treat population are shown for SET (red circles and lines) and for placebo (blue circles and lines). ITT, intent-to-treat population; SET, systemic enzyme therapy. Stage II In contrast to stage I, confirmatory analysis of both hypotheses of stage II resulted in no evidence for corresponding treatment effects (hypothesis 1: p=0.8596, hypothesis 2: p=0.8783, both one sided). As shown in figure 4, there is ‘severe’ heterogeneity between both stages for both hypotheses (hypothesis 1: MW=0.6153 vs MW=0.4379; I2=0.7692, p=0.0374; hypothesis 2: MW=0.5917 vs MW=0.4267; I2=0.6778, p=0.0781). As both I2 values are above 0.5, indicating ‘large’ heterogeneity, results are to be interpreted separately for each stage. Figure 4 Statistical analysis of multidimensional ensemble of peak torque maximum and pressure-induced pain at 3 hours and 6 hours (hypothesis 1) and at 24 hours and 48 hours (hypothesis 2).

Stage II In contrast to stage I, confirmatory analysis of both hypotheses of stage II resulted in no evidence for corresponding treatment effects (hypothesis 1: p=0.8596, hypothesis 2: p=0.8783, both one sided). As shown in figure 4, there is ‘severe’ heterogeneity between both stages for both hypotheses (hypothesis 1: MW=0.6153 vs MW=0.4379; I2=0.7692, p=0.0374; hypothesis 2: MW=0.5917 vs MW=0.4267; I2=0.6778, p=0.0781). As both I2 values are above 0.5, indicating ‘large’ heterogeneity, results are to be interpreted separately for each stage. Figure 4 Statistical analysis of multidimensional ensemble of peak torque maximum and pressure-induced pain at 3 hours and 6 hours (hypothesis 1) and at 24 hours and 48 hours (hypothesis 2). Biomarkers Pooled biomarker analysis at 3 hours after exercise demonstrated significant advantages for SET compared with placebo (stage I: p=0.0011; stage II: p=0.0114) (figure 5) and no heterogeneity (I2=0.0) between stages I and II. Stage I and II combined biomarker effect size shows significant and ‘more than small’ superiority of the SET group (MW=0.5847, p=0.0001) compared with the placebo group (figure 5).

ges for SET compared with placebo (stage I: p=0.0011; stage II: p=0.0114) (figure 5) and no heterogeneity (I2=0.0) between stages I and II. Stage I and II combined biomarker effect size shows significant and ‘more than small’ superiority of the SET group (MW=0.5847, p=0.0001) compared with the placebo group (figure 5). Figure 5 Superiority of SET compared with placebo in improving inflammatory, immune and metabolic biomarkers in response to exhaustive eccentric exercise. Effect sizes (changes from baseline) at 3 hours after exhaustive eccentric exercise were pooled for stage I, stage II and both stages combined. Pooled biomarkers include creatine kinase, lactate dehydrogenase, lactate, interleukin 6, prostaglandin E2, total oxidative status, total antioxidative status and natural killer cells, and were analysed for the direction of superiority. df, degrees of freedom; MW, Mann-Whitney; N, number; n. def., not defined; p, probability; SET, systemic enzyme therapy. Safety In stage I, out of 16 single adverse events, only one was characterised as ‘possibly’ related (diarrhoea, SET). In stage II, two single adverse events, one possibly related to SET (acne-like rush at chin and mild pruritus), were reported. No adverse event was ‘serious.’ Both events related to SET were characterised as mild and resolved. In both stages, there were no clinically relevant findings in any of the treatment groups. SET is well tolerated at the dosage of 3×4 tablets per day.

ted to SET (acne-like rush at chin and mild pruritus), were reported. No adverse event was ‘serious.’ Both events related to SET were characterised as mild and resolved. In both stages, there were no clinically relevant findings in any of the treatment groups. SET is well tolerated at the dosage of 3×4 tablets per day. Discussion We demonstrated the superiority of SET compared with placebo in maintaining strength and reducing pain in response to exhaustive eccentric exercise (acute phase, stage I). Stage II data were evaluated separately due to significant heterogeneity (I2>0.5) and, in contrast to stage I, did not result in differences of SET or placebo on EIMD. Biomarkers responded to SET across both study stages. Strengths and limitations The trial was carefully planned using a multidimensional approach for outcome assessment and classification, taking into account the multidimensional aspect of recovery from EIMD. Additionally, a two-stage approach with interim analysis and continuation with refined hypotheses and readjusted sample size offered all opportunities to finalise the trial successfully.

imensional approach for outcome assessment and classification, taking into account the multidimensional aspect of recovery from EIMD. Additionally, a two-stage approach with interim analysis and continuation with refined hypotheses and readjusted sample size offered all opportunities to finalise the trial successfully. Surprisingly, the preplanned test for carryover effects was statistically significant on the defined level α1 (p=0.0092, one sided, Wei-Lachin procedure, Bauer-Köhne α1=0.0102) in stage I. Thus, according to statistical plan, only phase 1 data from stage I could be used for non-parametric confirmatory analysis, and the crossover approach had to be abandoned in stage II in favour of a parallel group comparison. One possible explanation for carryover effects is that impairment of muscle function could have been influenced by recent, that is, within some months,43 high-force eccentric work using the same muscle(s). This effect is commonly referred to as the ‘repeated-bout effect.’44 However, as EIMD during the current trial was only mild, the washout phase during the eccentric exercise sessions should have been sufficient to avoid a repeated bout effect. A carryover effect from SET to placebo cannot be excluded as irreversible changes of cell surface receptors of blood and immune cells have been reported.20 21 23

during the current trial was only mild, the washout phase during the eccentric exercise sessions should have been sufficient to avoid a repeated bout effect. A carryover effect from SET to placebo cannot be excluded as irreversible changes of cell surface receptors of blood and immune cells have been reported.20 21 23 Stages I and II revealed ‘severe’ heterogeneity even though inclusion criteria were unchanged at screening for both stages at PTM 150–300 Nm and a variability of below 20%. Large interindividual variation in response to eccentric exercise is commonly reported.45 EIMD should even be higher in this age range as the magnitude of muscle damage is increased from preadolescent, adolescent to postadolescent men.46 Flexibility,47 48 eccentric peak and end-range torque49 as well as angle of peak torque (cf. the joint angle–torque relationship) are further factors contributing generally to a variety of EIMD responses.50 Additionally, extended training of the same muscle(s) comparable to exhaustive eccentric exercise may influence regeneration mechanisms and individual autonomous pain threshold.45 Large variations in individual responses to eccentric exercise are also evident from the literature, and gross muscle damage does not occur in all individuals.45 51–54 Study model may be limited in ability to observe a functional response in more strength-trained athletes A closer look at single subject data at baseline revealed differences between stages in the number of main disciplines and training duration per week (table 1).

Stages I and II revealed ‘severe’ heterogeneity even though inclusion criteria were unchanged at screening for both stages at PTM 150–300 Nm and a variability of below 20%. Large interindividual variation in response to eccentric exercise is commonly reported.45 EIMD should even be higher in this age range as the magnitude of muscle damage is increased from preadolescent, adolescent to postadolescent men.46 Flexibility,47 48 eccentric peak and end-range torque49 as well as angle of peak torque (cf. the joint angle–torque relationship) are further factors contributing generally to a variety of EIMD responses.50 Additionally, extended training of the same muscle(s) comparable to exhaustive eccentric exercise may influence regeneration mechanisms and individual autonomous pain threshold.45 Large variations in individual responses to eccentric exercise are also evident from the literature, and gross muscle damage does not occur in all individuals.45 51–54 Study model may be limited in ability to observe a functional response in more strength-trained athletes A closer look at single subject data at baseline revealed differences between stages in the number of main disciplines and training duration per week (table 1). Table 1 Summary of anthropometric data and baseline characteristics

Study model may be limited in ability to observe a functional response in more strength-trained athletes A closer look at single subject data at baseline revealed differences between stages in the number of main disciplines and training duration per week (table 1). Table 1 Summary of anthropometric data and baseline characteristics Parameter Stage I Stage II Total (n=27) SET (n=15) Placebo (n=12) Total (n=42) SET (n=21) Placebo (n=21) Age (years)±SD 31.6±9.3 29.2±8.8 34.6±9.3 29.7±8.7 30.7±7.6 28.7±9.8 Height (m)±SD 1.81±0.06 1.79±0.06 1.83±0.07 1.83±0.05 1.84±0.06 1.82±0.04 Weight (kg)±SD 80.0±10,0 76.5±10.2 84.3±8.0 81.2±7.4 81.0±8.6 81.3±61 BMI (kg/m2)±SD 24.4±2.1 23.8±2.0 25.2±1.9 24.3±1.8 24.0±1.8 24.6±1.8 Sport activity (min/week)±SD 369±206 415±197 312±211 482±433 483±502 481±363 CK (U/L)±SD 214±157 227±148 199±122 244±165 225±160 264±172 PTM/BW (Nm/kg)±SD 2.76±0.44 2.76±0.51 2.75±0.34 2.82±0.51 2.75±0,53 2.89±0.48 Anthropometric data, sporting activity and creatine kinase were assessed at screening; peak torque maximum/body weight was measured the day before start of supplementation. BMI, body mass index; CK, creatine kinase; Nm, newton metre; PTM/BW, peak torque maximum/body weight; SET, systemic enzyme therapy.

Parameter Stage I Stage II Total (n=27) SET (n=15) Placebo (n=12) Total (n=42) SET (n=21) Placebo (n=21) Age (years)±SD 31.6±9.3 29.2±8.8 34.6±9.3 29.7±8.7 30.7±7.6 28.7±9.8 Height (m)±SD 1.81±0.06 1.79±0.06 1.83±0.07 1.83±0.05 1.84±0.06 1.82±0.04 Weight (kg)±SD 80.0±10,0 76.5±10.2 84.3±8.0 81.2±7.4 81.0±8.6 81.3±61 BMI (kg/m2)±SD 24.4±2.1 23.8±2.0 25.2±1.9 24.3±1.8 24.0±1.8 24.6±1.8 Sport activity (min/week)±SD 369±206 415±197 312±211 482±433 483±502 481±363 CK (U/L)±SD 214±157 227±148 199±122 244±165 225±160 264±172 PTM/BW (Nm/kg)±SD 2.76±0.44 2.76±0.51 2.75±0.34 2.82±0.51 2.75±0,53 2.89±0.48 Anthropometric data, sporting activity and creatine kinase were assessed at screening; peak torque maximum/body weight was measured the day before start of supplementation. BMI, body mass index; CK, creatine kinase; Nm, newton metre; PTM/BW, peak torque maximum/body weight; SET, systemic enzyme therapy. Participants in stage I were primarily runners/joggers (endurance training) whereas participants in stage II primarily focused on strength (resistance) training. The duration of weekly strength training was reported at nearly three times for stage II compared with stage I (179 vs 61 min). Further, overall training duration across all disciplines was higher in stage II than stage I (482 vs 369 min). Therefore, stage II participants were more well trained than stage I participants, particularly related to strength (resistance) exercises. This may increase PTM by higher physical performance and by increase in pain threshold or the ability to increase PTM despite pain. Insensitivity of the study model to detect advantages of SET might have been caused by these differences in baseline resistance training. It has been suggested to report individual data and to classify subjects, for example, low, medium/moderate or high responders, to allow for a better presentation and interpretation of the data.49 51

the study model to detect advantages of SET might have been caused by these differences in baseline resistance training. It has been suggested to report individual data and to classify subjects, for example, low, medium/moderate or high responders, to allow for a better presentation and interpretation of the data.49 51 Biomarker data confirm anti-inflammatory effects of SET In contrast to functional parameters PTM and PIP, biomarker analysis resulted in a significant and ‘more than small’ superiority of the SET group compared with placebo group in single stages and both stages combined. Cytokines may play a relatively minor role in regulating the health benefits of low-intensity exercise, such as brisk walking.55 In contrast, marathon running induces high physiological stress and a large cytokine response56 as a more generalised response to internal and/or external stress. Factors such as oxidative or nitrosative stress, damaged or unfolded proteins, hyperthermia or energy imbalance likely induce cytokine production during exercise through catecholamines, endotoxin, alarmins, ATP and proinflammatory cytokines themselves.57 58 SET is likely to contribute to a reduction of large cytokine response in EIMD and DOMS independent of physical performance/training status.

The government-recommended guidelines of 30 min of moderate-intensity continuous exercise did not acutely impact endothelial function or factors associated with repair in postmenopausal women. How might it impact on clinical practice in the near future? The findings provide support for individualised exercise prescription for postmenopausal women. Healthcare professionals may advise postmenopausal women to exercise in short bursts of activity followed by active rest periods for vascular benefits. Introduction The risk of cardiovascular disease (CVD) in women increases during perimenopause and postmenopause primarily due to the loss of oestrogen. This hormonal change, in addition to increasing age, reduces endothelial function1 and the number and function of circulating angiogenic cells (CACs; 15). CACs migrate and aggregate to areas of vascular damage and aid in endothelial repair through the secretion of proangiogenic cytokines and growth factors.2 Low levels of CACs in circulation and reduced function following culture (ie, impaired colony-forming capacity) are associated with an increased risk of developing CVD.3 Fewer numbers of colony-forming units (CFUs) are associated with poorer brachial artery endothelial function and increased severity of coronary artery disease.4 5 Decreased oestrogen levels significantly impact the capacity for endothelial repair.6 Thus, the reduction in endothelial function and inability of CACs to aid in vascular repair following the menopausal transition indicates an imbalance in vascular homeostasis and plays a part in CVD risk.

f coronary artery disease.4 5 Decreased oestrogen levels significantly impact the capacity for endothelial repair.6 Thus, the reduction in endothelial function and inability of CACs to aid in vascular repair following the menopausal transition indicates an imbalance in vascular homeostasis and plays a part in CVD risk. Lifestyle interventions such as exercise are recommended for postmenopausal women as a preventative strategy for CVD.7 Aerobic exercise training studies in postmenopausal women have demonstrated improved endothelial function8 and reduced oxidative stress.9 However, the effect of exercise on CAC number and function in postmenopausal women has not been studied. Recent evidence suggests that interval exercise is more or equally effective as UK government guidelines for moderate-intensity continuous exercise in improving cardiorespiratory fitness, endothelial function and arterial stiffness.10 11 We have suggested that interval exercise performed at a heavy intensity may induce an increase in CAC number compared with moderate-intensity exercise, likely due to greater metabolic stress.2 12 However, a comparison between the effects of interval and continuous exercise on endothelial function, CAC number and function of these cells has not been examined in postmenopausal women.

tensity may induce an increase in CAC number compared with moderate-intensity exercise, likely due to greater metabolic stress.2 12 However, a comparison between the effects of interval and continuous exercise on endothelial function, CAC number and function of these cells has not been examined in postmenopausal women. The aims of this study were to (1) compare the acute effects of moderate-intensity continuous and interval exercise, on endothelial function and CAC number/function in postmenopausal women, and (2) compare these effects with an interval exercise session performed in the heavy-intensity domain. Endothelial function measured by brachial artery flow-mediated dilation (FMD) was the primary outcome. Secondary outcomes were brachial artery diameter, shear rate and reactive hyperaemic variables and CAC number and function. We did not include a session of heavy-intensity continuous exercise as reports suggest that postmenopausal women dislike exercise of this nature.13 Investigating the acute effects of exercise on vascular health allows identification of (1) exercise that has an immediate impact and (2) the type and intensity of exercise that may yield the greatest improvements if undertaken chronically. We hypothesised that interval exercise would be tolerable and would improve markers of vascular health and repair to a greater extent than continuous exercise due to the brief excursions to higher work rates.

and (2) the type and intensity of exercise that may yield the greatest improvements if undertaken chronically. We hypothesised that interval exercise would be tolerable and would improve markers of vascular health and repair to a greater extent than continuous exercise due to the brief excursions to higher work rates. Methods Participants Fifteen healthy postmenopausal women (age: 63±4 years) volunteered for the study and provided written informed consent. Postmenopausal status was defined as absence of a menstrual cycle for >2 years and confirmed through follicle stimulating hormone >30 iU/L. Exclusion criteria included smoking, known cardiovascular, pulmonary and metabolic disease, musculoskeletal impairments, cancer, contradictions to exercise, medication use (eg, hormone replacement therapy) and if participants had given blood in the previous three months. Participants were not currently exercising more than twice per week. Ethical approval was provided by the University of Leeds Faculty of Biological Sciences Ethics Committee and all procedures conformed to the Declaration of Helsinki.

What this study adds? Following a transient ischaemic attack (TIA), people are at risk of further cardiovascular events and this is a key time to initiate secondary cardiovascular prevention. Previous studies have focused on pharmacological secondary prevention following a TIA. However, there is a clear need to initiate non-pharmacological secondary cardiovascular measures, including physical activity promotion, following a TIA. One way to promote both pharmacological and non-pharmacological secondary cardiovascular prevention following a TIA is to adapt a home-based cardiac rehabilitation programme, ‘The Healthy Brain Rehabilitation Manual’, for this population following the Medical Research Council guidelines for developing complex health service interventions, which will be done within the Stroke Prevention Rehabilitation Intervention Trial of Exercise study. Background Definition of transient ischaemic attack Transient ischaemic attack (TIA) is defined as ‘a transient episode of neurological dysfunction caused by focal brain, spinal cord or retinal ischaemia, without acute infarction’1 and is diagnosed by the patients’ history, a neurological examination and/or neuroimaging (typically a CT head scan). Typical symptoms of TIA include the rapid onset of speech disturbance, unilateral weakness or sensory loss, monocular blindness, visual field defect or ataxia.

eplacement therapy) and if participants had given blood in the previous three months. Participants were not currently exercising more than twice per week. Ethical approval was provided by the University of Leeds Faculty of Biological Sciences Ethics Committee and all procedures conformed to the Declaration of Helsinki. Experimental protocol For the assessment of endothelial function and calculation of work rate for the exercise bouts, participants attended the laboratory on two occasions (separated by 1 week). Participants were instructed to refrain from exercise participation and consuming alcohol and caffeine for 12 hours prior to each visit. At the first visit, a fasted blood sample (~50 mL) for the assessment of CVD risk blood markers and CAC number and function was completed. At the second visit, a cardiorespiratory fitness test was completed to determine peak aerobic capacity (V˙O2peak) and lactate threshold. On three further occasions, each separated by ≥1 week (to allow exercise-induced changes in CACs to return to baseline levels), participants completed 30 min of either moderate-intensity continuous (CON), moderate-intensity interval (MOD-INT) or heavy-intensity interval (HEAVY-INT) exercise bouts. Only nine participants completed the heavy-intensity bout. To assess the acute effects of each exercise bout on endothelial function, assessments were completed before and 15 min after exercise. For CAC number and function, blood samples were acquired 30 min post exercise and results compared with that of visit 1.

uts. Only nine participants completed the heavy-intensity bout. To assess the acute effects of each exercise bout on endothelial function, assessments were completed before and 15 min after exercise. For CAC number and function, blood samples were acquired 30 min post exercise and results compared with that of visit 1. Variables assessed before and after exercise sessions Cardiorespiratory fitness A seated ramp-incremental exercise test (10 W/min) was performed for the assessment of V˙O2peak and the lactate threshold, and for calculation of the work rates achieved at these points. These values were used to establish work rates for the subsequent exercise sessions. Participants were seated on an electronically braked cycle ergometer (Excalibur Sport V.2.0; Lode BV, Groningen, The Netherlands) and a mouthpiece and nose clip were fitted for breath by breath analysis of pulmonary gas exchange. The protocol has been described in detail previously.12 Heart rate, blood pressure and rating of perceived exertion (Borg's scale of rating of perceived exertion) were measured every 2 min during the test using a 12-lead ECG, sphygmomanometer and a visual scale of exertion (6-20), respectively. The work rate at the end of the ramp-incremental test was calculated as test duration × ramp rate (10)+20W. Breath-by-breath data were exported and analysed using OriginPro software (OriginPro 8, OriginLab, Northampton, Massachusetts, USA). Breaths were eliminated if V˙O2peak values fell outside 4 SD from the local mean. As previously described,24 a 12-breath moving average was calculated and the highest value was defined as V˙O2peak. The estimated lactate threshold was determined using the V-slope method25 and confirmed by a rise in end-tidal O2 and plateau in end-tidal CO2.

ed if V˙O2peak values fell outside 4 SD from the local mean. As previously described,24 a 12-breath moving average was calculated and the highest value was defined as V˙O2peak. The estimated lactate threshold was determined using the V-slope method25 and confirmed by a rise in end-tidal O2 and plateau in end-tidal CO2. Endothelial function Endothelial function was assessed in the morning in a temperature-controlled laboratory (20–24°C) after 20 min supine rest. The protocols for our lab have been described in detail elsewhere.2 12 Briefly, endothelial function was assessed by brachial artery FMD using ultrasound imaging following a 5 min period of forearm occlusion. Images were recorded at end-diastole using vascular imaging software (Vascular Imager, Medical Imaging Applications, Coralville, Iowa, USA) and analysed using semiautomated edge-detection software (Brachial Tools V.5, Medical Imaging Applications) to determine brachial artery diameter. Peak reactive hyperaemia, peak shear rate, area under the shear rate curve from cuff release to peak dilation (AUCpeak) and to 60 s (AUC60) and 90 s (AUC90) post cuff release, and their corresponding velocity time integrals (VTIs) were also calculated. FMD was not normalised to shear rate/AUC as not all assumptions for the use of ratios were met.26 During recording of blood velocity, the insonation angle for each participant before and after exercise and between different exercise visits was within 2°. Day-to-day coefficient of variation for FMD in our lab is 15%.2

. FMD was not normalised to shear rate/AUC as not all assumptions for the use of ratios were met.26 During recording of blood velocity, the insonation angle for each participant before and after exercise and between different exercise visits was within 2°. Day-to-day coefficient of variation for FMD in our lab is 15%.2 Blood markers and CAC number and function Visit 1 blood samples were analysed by local hospital pathology services for serum follicle stimulating hormone levels, lipoproteins, insulin, glucose and HbA1c. CACs were enumerated from 22 mL of blood via flow cytometry using a commercially available kit (Miltenyi Biotec) as previously described.2 CACs were defined as CD34+, double positive (CD34+KDR+) or triple positive (CD34+KDR+CD133+). To assess the in vitro function of CACs, a CFU assay was performed, according to the manufacturer's instructions. Briefly, peripheral blood mononuclear cells were separated by Ficoll density-gradient centrifugation of whole blood (Ficoll Paque PLUS, GE Healthcare, Buckinghamshire, UK), and 5×106 cells were suspended in 2 mL of EndoCult growth medium (StemCell Technologies, Vancouver, Canada) and cultured in one well of a fibronectin-coated six-well plate for 48 hours at 37°C in 5% CO2. After 48 hours, the non-adherent cells were collected from the well and seeded in duplicate on a 24-well fibronectin-coated plate at a density of 1×106 cells/well. Following three further days of culture, the non-adherent cells were removed and the number of CFUs per well counted and an average calculated. CFUs were defined as clusters of >100 round cells with spindle-shaped cells surrounding the core.

licate on a 24-well fibronectin-coated plate at a density of 1×106 cells/well. Following three further days of culture, the non-adherent cells were removed and the number of CFUs per well counted and an average calculated. CFUs were defined as clusters of >100 round cells with spindle-shaped cells surrounding the core. Exercise session protocols Participants completed 30 min CON, MOD-INT and HEAVY-INT exercise bouts on a cycle ergometer (Excalibur Sport V.2.0) on separate days. Duration of 30 min was chosen as it reflects the current UK Department of Health guidelines of 30 min of moderate-intensity exercise, 5 days/week (ref. 27; p. 34). CON exercise involved cycling at 80% of the work rate achieved at the lactate threshold (moderate-intensity domain as V˙O2peak remained below the lactate threshold28). The interval exercise bouts were based on work by Turner et al.29 Thus, our participants cycled at 90% of the work rate achieved at V˙O2peak for duty cycles of 10:20 s for MOD-INT and 30:60 s for HEAVY-INT exercise. Active recovery was conducted at 10 W and confirmation of intensity domain was confirmed by stabilisation of oxygen uptake above or below the lactate threshold accordingly. By design the MOD-INT and HEAVY-INT exercise bouts were matched for average work rate and work completed.

D-INT and 30:60 s for HEAVY-INT exercise. Active recovery was conducted at 10 W and confirmation of intensity domain was confirmed by stabilisation of oxygen uptake above or below the lactate threshold accordingly. By design the MOD-INT and HEAVY-INT exercise bouts were matched for average work rate and work completed. Statistical analysis All analyses was completed using SPSS V.22. Data were assessed for normal distribution via Kolmogorov-Smirnov. A non-parametric Friedman's analysis of variance (ANOVA) was conducted on CAC number only as data were not normally distributed and could not be transformed. The remaining data were examined by a mixed-design ANOVA with time (pre vs post exercise) as the within-subject factor and exercise bout (CON, MOD-INT and HEAVY-INT) as the between-subject factor. Pearson's correlations were performed between V˙O2peak, blood pressure, full lipid profile and the change in CFUs post exercise. Using previously reported acute increases of 4.6% in FMD following 45 min of treadmill exercise in postmenopausal women,17 and SD of 3% calculated from our lab,12 a minimum of nine participants in total were required to obtain 80% power (α=0.05) in a crossover study. Data are presented as mean±SD and percentage change with accompanying 95% CI.

acute increases of 4.6% in FMD following 45 min of treadmill exercise in postmenopausal women,17 and SD of 3% calculated from our lab,12 a minimum of nine participants in total were required to obtain 80% power (α=0.05) in a crossover study. Data are presented as mean±SD and percentage change with accompanying 95% CI. Results Participant and exercise session characteristics Participant characteristics are displayed in table 1. Total cholesterol and low-density lipoprotein levels were higher than the desirable healthy range (>5.2 and >3.4 mmol/L, respectively) in nine and seven participants, respectively. By design, work completed (54±3 kJ) during MOD-INT and HEAVY-INT exercise sessions was equal. The average work rate of the peaks during MOD-INT and HEAVY-INT (70±5 W) were significantly higher than the average work rate of CON exercise (35±6 W). Table 1 Participant characteristics of postmenopausal women at visit 1

Results Participant and exercise session characteristics Participant characteristics are displayed in table 1. Total cholesterol and low-density lipoprotein levels were higher than the desirable healthy range (>5.2 and >3.4 mmol/L, respectively) in nine and seven participants, respectively. By design, work completed (54±3 kJ) during MOD-INT and HEAVY-INT exercise sessions was equal. The average work rate of the peaks during MOD-INT and HEAVY-INT (70±5 W) were significantly higher than the average work rate of CON exercise (35±6 W). Table 1 Participant characteristics of postmenopausal women at visit 1 n Mean (±SD) Age (years) 15 63±4 BMI (kg/m)  15 25±3 Brachial artery SBP (mm Hg) 15 137±15 Brachial artery DBP (mm Hg) 15 84±5 Brachial artery MAP (mm Hg) 15 102±8 Absolute V˙O2peak (L/min) 15 1.40±0.29 Relative V˙O2peak (mL/kg/min) 15 21.6±5.4 Plasma glucose (mmol/L) 13 4.8±0.5 HbA1c (mmol/mol/HB) 14 39±2 Total cholesterol (mmol/L) 14 5.9±1.0 HDL (mmol/L) 14 1.9±0.5 LDL (mmol/L) 14 3.5±0.8 Cholesterol:HDL ratio 14 3.1±0.7 Triglycerides (mmol/L) 14 1.0±0.4 FSH (iU/L) 13 69.9±31.0 Insulin (mU/L) 14 6.5±3.1 BMI, body mass index; DBP, diastolic blood pressure; FSH, follicle-stimulating hormone; HbA1c, haemoglobin A1c; HDL, high-density lipoprotein; LDL, low-density lipoprotein; MAP, mean arterial pressure; SBP, systolic blood pressure.

±0.7 Triglycerides (mmol/L) 14 1.0±0.4 FSH (iU/L) 13 69.9±31.0 Insulin (mU/L) 14 6.5±3.1 BMI, body mass index; DBP, diastolic blood pressure; FSH, follicle-stimulating hormone; HbA1c, haemoglobin A1c; HDL, high-density lipoprotein; LDL, low-density lipoprotein; MAP, mean arterial pressure; SBP, systolic blood pressure. Brachial artery endothelial function Brachial artery endothelial function was unaltered by acute CON, MOD-INT and HEAVY-INT exercise (p>0.05, table 2). There were no significant post-exercise changes in brachial artery resting diameter (p=0.72) or the associated shear rate and reactive hyperaemic variables (p>0.05). Additionally, there were no time by exercise interactions (p>0.05). Table 2 Brachial artery endothelial function (mean±SD) pre and post an acute 30 min bout of moderate-intensity continuous (CON, n=15), moderate-intensity interval (MOD-INT, n=14) and heavy-intensity interval (HEAVY-INT) exercise (n=9)

Brachial artery endothelial function Brachial artery endothelial function was unaltered by acute CON, MOD-INT and HEAVY-INT exercise (p>0.05, table 2). There were no significant post-exercise changes in brachial artery resting diameter (p=0.72) or the associated shear rate and reactive hyperaemic variables (p>0.05). Additionally, there were no time by exercise interactions (p>0.05). Table 2 Brachial artery endothelial function (mean±SD) pre and post an acute 30 min bout of moderate-intensity continuous (CON, n=15), moderate-intensity interval (MOD-INT, n=14) and heavy-intensity interval (HEAVY-INT) exercise (n=9) CON MOD-INT HEAVY-INT Pre Post Pre Post Pre Post Resting diameter (mm) 3.4±0.5 3.4±0.4 3.5±0.4 3.4±0.4 3.2±0.5 3.2±0.5 Time from cuff release to peak diameter (s) 45±17 49±23 53±16 51±19 69±20 75±27 Insonation angle (°) 68±1 68±1 68±1 68±1 68±1 68±1 Relative FMD (%) 6.1±2.5 6.0±4.2 5.3±2.5 5.3±3.2 4.9±1.7 4.3±1.7 VTIpeak (cm) 1442±602 1620±775 1681±597 1719±709 2198±526 2213±835 VTI60 (cm) 1634±5493 1797±565 1806±435 1861±410 2106±472 1909±455 VTI90 (cm) 2111±648 2261±765 2281±564 2364±579 2550±650 2368±592 Peak reactive hyperaemia (cm/s) 71±28 74±32 88±33 87±25 110±14 104±23 Peak shear rate (/s–1) 1635±659 1746±813 2072±892 2096±723 2830±557 2640±573 AUCpeak (a.u.) 33 204±15 071 37 500±16 200 39 477±15 399 40 027±14 399 55 508±13 056 55 647±19 664 AUC60 (a.u.) 37 825±12 899 42 027±12 830 42 292±11 723 44 053±9443 53 132±11 319 48 403±12 482 AUC90 (a.u.) 48 785±16 752 52 657±16 587 53 211±14 127 55 796±12 521 64 237±15 371 59 953±16 039 No significant effects of exercise or group by time interactions were revealed (p>0.05).

027±14 399 55 508±13 056 55 647±19 664 AUC60 (a.u.) 37 825±12 899 42 027±12 830 42 292±11 723 44 053±9443 53 132±11 319 48 403±12 482 AUC90 (a.u.) 48 785±16 752 52 657±16 587 53 211±14 127 55 796±12 521 64 237±15 371 59 953±16 039 No significant effects of exercise or group by time interactions were revealed (p>0.05). AUC, shear rate area under the curve; FMD, flow-mediated dilation; VTI, velocity–time integral. CAC number Neither CON nor MOD-INT exercise altered concentrations of CD34+ (p=0.28), CD34+KDR+ (p=0.57) or CD34+KDR+CD133+ cells (p=0.74). In addition, there was no change in any CAC population following HEAVY-INT (p>0.05, table 3). No time by exercise bout interactions were observed (p>0.05). Table 3 Circulating angiogenic cell number (mean±SD) at visit 1 and post an acute 30 min bout of moderate-intensity continuous (CON) and interval (MOD-INT) exercise and heavy-intensity interval (HEAVY-INT) exercise Visit 1 CON MOD-INT HEAVY-INT CD34+ cells /mL blood 2.7±1.3×105 2.2±1.2×105 2.6±1.1×105 2.3±0.9×105 CD34+KDR+ cells / mL blood 251±176 108±119 244±188 144±201 CD34+KDR+CD133+ cells /mL blood 68±102 11±13 40±85 14±26 No significant effects of exercise or group by time interactions were revealed (p>0.05).

Table 3 Circulating angiogenic cell number (mean±SD) at visit 1 and post an acute 30 min bout of moderate-intensity continuous (CON) and interval (MOD-INT) exercise and heavy-intensity interval (HEAVY-INT) exercise Visit 1 CON MOD-INT HEAVY-INT CD34+ cells /mL blood 2.7±1.3×105 2.2±1.2×105 2.6±1.1×105 2.3±0.9×105 CD34+KDR+ cells / mL blood 251±176 108±119 244±188 144±201 CD34+KDR+CD133+ cells /mL blood 68±102 11±13 40±85 14±26 No significant effects of exercise or group by time interactions were revealed (p>0.05). CFU numbers A significant time effect (p=0.0001) and time by exercise bout interaction (p=0.017) for CFU number was revealed. The analysis of the means and 95% CI for pre exercise and post exercise (figure 1) illustrates that CFU number increased following MOD-INT (pre: 12.2 CFUs, 95% CI 6.4 to 17.9; post: 32 CFUs, 95% CI 19.4 to 44.7) and HEAVY-INT (pre: 11.2 CFUs, 95% CI 4.1 to 18.3; post: 37.8 CFUs, 95% CI 22 to 53.6) but remained unchanged following CON (pre: 12.2 CFUs, 95% CI 6.4 to 17.9; post: 12.8 CFUs, 95% CI 0.2 to 25.5). There was no correlation between the change in CFUs and pre-exercise  V˙O2peak, blood pressure or cholesterol levels (p>0.05). Figure 1 The number (mean with 95% CI) of colony-forming units (CFUs) at visit 1 and following a 30 min moderate-intensity continuous (CON), moderate-intensity interval (MOD-INT) and heavy-intensity interval (HEAVY-INT) exercise bout. CFUs increased following the MOD-INT (n=14) and the HEAVY-INT (n=9) exercise bouts, but not following CON (n=14).

th 95% CI) of colony-forming units (CFUs) at visit 1 and following a 30 min moderate-intensity continuous (CON), moderate-intensity interval (MOD-INT) and heavy-intensity interval (HEAVY-INT) exercise bout. CFUs increased following the MOD-INT (n=14) and the HEAVY-INT (n=9) exercise bouts, but not following CON (n=14). Discussion The present study was the first to compare the acute effects of continuous and interval exercise on endothelial function and CAC number and function in postmenopausal women. The population of postmenopausal women studied were healthy but presented risk factors for CVD. The main findings were that moderate-intensity continuous exercise had no immediate effect on endothelial function or CAC concentrations or function. Conversely, MOD-INT and HEAVY-INT exercise acutely increased the ability of cultured CACs to form colonies in vitro. Presentation of CVD risk factors did not influence this effect.

indings were that moderate-intensity continuous exercise had no immediate effect on endothelial function or CAC concentrations or function. Conversely, MOD-INT and HEAVY-INT exercise acutely increased the ability of cultured CACs to form colonies in vitro. Presentation of CVD risk factors did not influence this effect. Continuous and interval exercise did not acutely effect endothelial function Brachial artery endothelial function did not change following any of the exercise bouts. In response to lower-limb exercise, measurement of upper limb vascular function is used to reflect systemic endothelial function. Our findings suggest that for increases in systemic endothelial function in postmenopausal women to occur, a greater stimulus or repeated acute exercise bouts are required. Indeed, an absolute increase of ~5% in brachial artery FMD was observed by Harvey et al. following 45 min of continuous treadmill exercise at 60% V˙O2max, in postmenopausal women with similar FMD values as in the present study (~5%; 17). Greater endothelial function following exercise is mediated by shear stress-induced increases in nitric oxide bioavailability which induces vasodilation.21 The exercise duration was greater in Harvey's study and a different definition of intensity adopted compared with the present study. It is likely that a greater volume/magnitude of shear was thus induced.

xercise is mediated by shear stress-induced increases in nitric oxide bioavailability which induces vasodilation.21 The exercise duration was greater in Harvey's study and a different definition of intensity adopted compared with the present study. It is likely that a greater volume/magnitude of shear was thus induced. HEAVY-INT exercise may well have been expected to induce higher levels of shear stress; however, compared with the women studied by Harvey et al., women in the present study were older (64±4 years vs 54±2 years) and had greater blood pressure (137/84 mm Hg vs 108/64 mm Hg). Hypertension and older age (>60 years) have an additive effect on reducing nitric oxide bioavailability and increasing oxidative stress.30 Additionally, the sensitivity of the endothelium to detect shear stress and trigger nitric oxide synthesis may be reduced. Aged endothelial cells in vitro exhibit impaired eNOS protein upregulation in response to shear stress compared with young cells cultured under identical conditions.19 Thus, a greater shear stress stimulus may be required to induce increases in nitric oxide in the present cohort.

nitric oxide synthesis may be reduced. Aged endothelial cells in vitro exhibit impaired eNOS protein upregulation in response to shear stress compared with young cells cultured under identical conditions.19 Thus, a greater shear stress stimulus may be required to induce increases in nitric oxide in the present cohort. Circulating angiogenic cells were not mobilised following exercise The present study is the first to measure CAC mobilisation following exercise in postmenopausal women and use novel comparisons between interval and continuous exercise sessions. Although some studies have observed acute increases in CAC concentrations following maximal and submaximal exercise in healthy and diseased populations,14 31 32 others report no change.33 34 We observed no increase in CACs following continuous and interval exercise at moderate and heavy intensities. These discrepancies might be explained by differing populations, definitions of CACs, exercise stimuli and the techniques used to numerate cells. However, acute exercise-induced CAC mobilisation is mechanistically driven by an increase in shear stress-induced nitric oxide.14 Thus, if endothelial sensitivity is impaired in postmenopausal women, it may well be that a larger magnitude and/or volume of shear stress are required to mobilise CACs. This could be achieved by exercising at greater work rates or by repeated acute bouts in a training programme.

in shear stress-induced nitric oxide.14 Thus, if endothelial sensitivity is impaired in postmenopausal women, it may well be that a larger magnitude and/or volume of shear stress are required to mobilise CACs. This could be achieved by exercising at greater work rates or by repeated acute bouts in a training programme. Interval exercise acutely increases circulating angiogenic cell functional capacity A low capacity of cultured CACs to form colonies is associated with a greater risk of CVD.4 The increase in CFUs following interval exercise regardless of intensity, but not following continuous exercise, suggests that interval exercise is more effective at increasing the capacity for endothelial repair by CACs among postmenopausal women. However, the mechanisms involved are unclear.

h a greater risk of CVD.4 The increase in CFUs following interval exercise regardless of intensity, but not following continuous exercise, suggests that interval exercise is more effective at increasing the capacity for endothelial repair by CACs among postmenopausal women. However, the mechanisms involved are unclear. In vitro characterisation studies have demonstrated that CFUs are a heterogeneous population of aggregated monocytes and T cells35 and the assay reflects the intercellular communication of these cell populations.18 The ability to form CFUs depends on angiogenic T cells (located in the centre of the CFU colonies) that express the platelet endothelial cell adhesion molecule (CD31) and the stromal-derived factor-1 receptor (CXCR4).20 36 Higher levels of circulating angiogenic T cells are associated with more CFUs in vitro, even when equal numbers of peripheral blood mononuclear cells are cultured,20 while lower levels are associated with older age and increased CVD risk.20 36 Angiogenic T cells are mobilised immediately post exercise, and thus, it may be that interval exercise mobilised circulating angiogenic T cells to a greater extent than continuous exercise, thus, enabling more CFUs to form in vitro. Future studies should measure whether more angiogenic T cells (CD3+CD31+CXCR4+) are present post exercise.

are mobilised immediately post exercise, and thus, it may be that interval exercise mobilised circulating angiogenic T cells to a greater extent than continuous exercise, thus, enabling more CFUs to form in vitro. Future studies should measure whether more angiogenic T cells (CD3+CD31+CXCR4+) are present post exercise. Angiogenic T cells in culture may also enable CACs to form colonies through the secretion of proangiogenic cytokines.20 Messenger RNA expression of genes involved in immune and T cell function (eg, interleukin (IL)-1β and IL-8) are upregulated following acute cycling exercise.37 It is plausible that interval but not continuous exercise induced a change in this gene expression. Future studies should characterise the phenotype and gene expression of CFUs post exercise.

es involved in immune and T cell function (eg, interleukin (IL)-1β and IL-8) are upregulated following acute cycling exercise.37 It is plausible that interval but not continuous exercise induced a change in this gene expression. Future studies should characterise the phenotype and gene expression of CFUs post exercise. The mechanisms for increased CFUs post-interval exercise but not continuous exercise remain to be fully elucidated. Interval exercise induces greater peak heart rates and fluctuations in vascular shear rate profiles compared with moderate-intensity continuous exercise.11 12 38 Given that differentiation of peripheral blood mononuclear cells into different T cell subtypes is dependent on cytokine and catecholamine (ie, epinephrine, norepinephrine and cortisol) concentrations,23 37 39 the repeated fluctuations to higher work rates during interval exercise may act as a greater stimulus for angiogenic T cell mobilisation and activation essential for CFUs in vitro. Indeed, a recent study demonstrated that highly differentiated T cells were increased to a greater extent following interval exercise compared with continuous exercise, and regulatory T cells were only mobilised following interval exercise, likely due to increased plasma epinephrine.22

tion essential for CFUs in vitro. Indeed, a recent study demonstrated that highly differentiated T cells were increased to a greater extent following interval exercise compared with continuous exercise, and regulatory T cells were only mobilised following interval exercise, likely due to increased plasma epinephrine.22 Strengths and limitations A strength of this study is the cross-over design which reduces between-participant variability as each participant acts as their own control. We recognise that the inclusion of a non-exercise control group would have further supported the conclusion that acute interval exercise increases CFU number. However, given that 30 min of continuous exercise did not alter CFU number we are confident using this as our reference group. Additionally, the mean insonation angle of 68° may overestimate blood velocity26; however, we reduced this impact by maintaining the same angle before and after exercise for each participant. It is important to note that this study only investigated the effects of single acute bouts of exercise on endothelial function and CAC number/function in postmenopausal women; different effects might be observed following repeated exercise sessions over longer time periods.

ore and after exercise for each participant. It is important to note that this study only investigated the effects of single acute bouts of exercise on endothelial function and CAC number/function in postmenopausal women; different effects might be observed following repeated exercise sessions over longer time periods. Conclusions and future work The UK government recommended guidelines of 30 min of moderate-intensity continuous exercise do not have an immediate impact on endothelial function and CAC number and function in postmenopausal women. In contrast, while MOD-INT or HEAVY-INT exercise did not increase the mobilisation of CACs, it did increase the colony-forming ability of peripheral blood mononuclear cells potentially involved in the repair of vascular damage. The potential impact of interval exercise for vascular health and repair in postmenopausal women and other populations that are at risk of developing CVD is significant. Studies manipulating the magnitudes and fluctuations in shear are now imperative. The authors thank all the study volunteers for taking part in this project. This article includes some text that is similar to that from the first author's PhD thesis(9). Contributors: KMB and MR designed the study. MR and EH performed data collection. All authors contributed to data analysis and interpretation and drafting of the manuscript. Funding: This study was funded by a British Heart Foundation Project Grant 378 (PG/08/060/25340). Competing interests: None declared. Ethics approval: University of Leeds Faculty of Biological Sciences Ethics Committee.

Contributors: KMB and MR designed the study. MR and EH performed data collection. All authors contributed to data analysis and interpretation and drafting of the manuscript. Funding: This study was funded by a British Heart Foundation Project Grant 378 (PG/08/060/25340). Competing interests: None declared. Ethics approval: University of Leeds Faculty of Biological Sciences Ethics Committee. Provenance and peer review: Not commissioned; externally peer reviewed.

What are the new findings? People with mild to moderate chronic low back pain (CLBP) presented with similar standing postural control to asymptomatic individuals. During gait, spatiotemporal parameters were similar in people with and without CLBP. During gait, hip kinetics and kinematics were similar in people with and without CLBP. Treatments directed at influencing postural stability or specific parameters of gait may be an unnecessary addition to a treatment programme for people with CLBP. Introduction Differences in postural control1–4 and gait5–10 have been identified between people with and without chronic low back pain (CLBP). During more challenging standing conditions people with CLBP have demonstrated increased centre of pressure (CoP) displacements and velocities,1–4 indicative of poorer postural stability.11 12 A systematic review investigating difference in standing postural sway between those with and without CLBP reports inconsistent findings.13 Although the majority of studies reported an increased postural sway in people with LBP, evidence from fewer studies, many with larger sample sizes and more robust methodologies, demonstrated no difference between groups.13 Hence, whether a true difference exists remains unclear.

t CLBP reports inconsistent findings.13 Although the majority of studies reported an increased postural sway in people with LBP, evidence from fewer studies, many with larger sample sizes and more robust methodologies, demonstrated no difference between groups.13 Hence, whether a true difference exists remains unclear. During gait, people with CLBP have demonstrated reduced self-selected walking speed,5–8 stride time,9 10 stride length5 6 and range of hip movement9 compared with people without back pain. Due to the proposed decrease in stride length, walking speed and hip range of movement, hip joint moments are also likely to be decreased in people with CLBP compared with people without.14 Researchers have proposed that such gait changes may be an attempt by the individual to reduce pain by reducing: ground reaction forces at heel strike15; excessive muscle activity; or joint movement.16 Alternatively, differences may be a result of altered proprioceptive feedback17 or psychological factors associated with CLBP, such as anxiety, fear avoidance and catastrophising.18 Psychological factors may lead to adaptation of normal physical activities, such as fast walking, due to the fear of increasing pain. Although gait alterations may initially be protective, such alterations may induce mechanical problems in the long term, for example, a slower walking produces longer periods of loading on the lumbar spine during gait,19 which may be detrimental to spinal structures in the long term, whereas shorter periods of loading, thought to be less detrimental, occur during faster walking.19

erations may induce mechanical problems in the long term, for example, a slower walking produces longer periods of loading on the lumbar spine during gait,19 which may be detrimental to spinal structures in the long term, whereas shorter periods of loading, thought to be less detrimental, occur during faster walking.19 These differences in postural control1–4 and gait5–10 have been proposed as contributing factors to the presence and recurrent nature of CLBP.1 4 15 However, previous studies have used: small sample sizes2 3 (possibly introducing a type 2 error); methodological design likely to result in low reliability of data, for example, analysing data from one trial instead of multiple trials1 5 8 9; outcomes that have demonstrated poor reliability; or provide results not representative of the general population (eg, all or mainly male participants8 9; or walking on a treadmill as opposed to on normal ground7 9 10). This study aimed to add to current research by using a more reliable and valid methodology to determine whether participants with CLBP have similar or different barefoot standing postural control, and gait parameters, when compared with age-matched and gender-matched asymptomatic participants. This methodology has previously been published as part of a randomsied trial.20 The following hypotheses were investigated: H1: The CLBP group will demonstrate greater postural instability when compared with the asymptomatic group during more challenging standing conditions.

This study aimed to add to current research by using a more reliable and valid methodology to determine whether participants with CLBP have similar or different barefoot standing postural control, and gait parameters, when compared with age-matched and gender-matched asymptomatic participants. This methodology has previously been published as part of a randomsied trial.20 The following hypotheses were investigated: H1: The CLBP group will demonstrate greater postural instability when compared with the asymptomatic group during more challenging standing conditions. H2: Reduced self-selected walking speed, cadence and step length will be observed in people with CLBP compared with asymptomatic individuals. H3: During gait, people with CLBP will present with reduced peak hip extensor moments during stance phase and reduced hip range of movement compared with asymptomatic individuals. Methods This cross-sectional case–control study compared barefoot standing balance and gait data from CLBP participants with that from age-matched and gender-matched asymptomatic participants. This methodology has previously been published as part of a randomised trial.[new ref]

H3: During gait, people with CLBP will present with reduced peak hip extensor moments during stance phase and reduced hip range of movement compared with asymptomatic individuals. Methods This cross-sectional case–control study compared barefoot standing balance and gait data from CLBP participants with that from age-matched and gender-matched asymptomatic participants. This methodology has previously been published as part of a randomised trial.[new ref] Participant recruitment A convenience sample of asymptomatic adults was recruited from acquaintances and colleagues of the investigators. Participants with CLBP were recruited from four Physiotherapy Departments in London (UK) (three National Health Service hospitals, one private physiotherapy practice) following clinical referral from general practitioners and consultants as part of a previously reported randomised controlled trial (RCT).21 During the second half of the recruitment period of the RCT, 55 participants were asked to participate in the current study, 38 of which showed interest. Eighteen participants could not attend the session in the main due to work commitments. Of the remaining 20 participants, only 16 could be matched by age and gender to our asymptomatic group. Inclusion criteria for symptomatic individuals were: aged 18–65 years, with a 3-month or greater history of LBP. Exclusion criteria were constant non-mechanical LBP, lumbar radiculopathy, known spondylolisthesis, spinal stenosis or inflammatory back pain, specific spinal diagnosis inappropriate for physiotherapy interventions (eg, spinal fracture or infection); any condition inappropriate for exercise physiotherapy (eg, severe cardiovascular or metabolic disease) or for wearing rocker sole footwear (eg, Morton’s neuroma, peripheral neuropathy). Potential asymptomatic participants were contacted via email including the Participant Information Sheet, and were asked to contact CSM if they wished to partake in the study. Asymptomatic participants reported no history of LBP in the last year, were required to meet all other inclusion and exclusion criteria presented above. As increasing age is a contributing factor to poorer postural stability22 and gender may influence postural control,23 hence potential confounding factors, asymptomatic participants were matched by age and gender to symptomatic participants. An age range of 2 years above or below the age of the ‘matched’ CLBP participant was classed as acceptable. Sixteen asymptomatic participants were consented into the study.

ence postural control,23 hence potential confounding factors, asymptomatic participants were matched by age and gender to symptomatic participants. An age range of 2 years above or below the age of the ‘matched’ CLBP participant was classed as acceptable. Sixteen asymptomatic participants were consented into the study. Data collection Data collection occurred at the ‘One Small Step Gait Laboratory’, Guys’ Hospital, London. Demographic and pain scores (numerical rating scale) representing their level of back pain on the day of assessment were recorded from all participants. Biomechanical assessment Participants were assessed wearing short trousers and vest or no top. Participants’ anthropometric measurements (pelvic width; leg length; knee width; ankle width; height; and weight) were recorded to inform the mechanical model formulated for each participant in Vicon’s Nexus (V.1.8.1) motion capture software (Vicon Motions Systems, Oxford, UK). The motion analysis system consisted of seven cameras, capturing retroreflective markers in three-dimensional space at a rate of 120 Hz. Seventeen infrared reflective markers (14 mm diameter) were positioned on each participant by an experienced researcher (AS).24–26 The modified Helen Hayes marker set was implemented27 with additional markers on bilateral iliac crests, and posterior calcanei (figure 1). Figure 1 Participant with infrared reflective markers in situ standing on foam cushions overlying force plates.

Seventeen infrared reflective markers (14 mm diameter) were positioned on each participant by an experienced researcher (AS).24–26 The modified Helen Hayes marker set was implemented27 with additional markers on bilateral iliac crests, and posterior calcanei (figure 1). Figure 1 Participant with infrared reflective markers in situ standing on foam cushions overlying force plates. Postural stability in standing Participants were assessed barefoot, feet approximately pelvis width apart and on adjacent force plates (FP5000, AMTI, Massachusetts, USA), during four posture challenging standing conditions involving manipulation of visual input and support surface: (1) firm surface, eyes open; (2) firm surface, eyes closed; (3) compliant surface, eyes open; (4) compliant surface, eyes closed. Compliant surface was achieved by placing an AirexTM cushion (48.5×40.0×6.4 cm, 0.7 kg, density 38.6 kg/m−3, closed-cell foam) (l-group, St Louis, MO) over each force plate (figure 1). Participants were instructed to keep their eyes focused on a red sticker at eye height on a tripod 3 m in front of them.28 Participants were assessed for three 40 s trials (shown to produce acceptable reliability29) for each standing condition. The middle 30 s of each trial was analysed to avoid possible initial sway errors, effects of participant fatigue or anticipation of a trial ending. Each participant received the same instructions at the start of each trial:

d for three 40 s trials (shown to produce acceptable reliability29) for each standing condition. The middle 30 s of each trial was analysed to avoid possible initial sway errors, effects of participant fatigue or anticipation of a trial ending. Each participant received the same instructions at the start of each trial: When I say ‘Go’ I want you to stand and maintain your balance until you hear the instruction to rest. Each trial will last for 40 s. Focus on the red sticker on the tripod ahead of you. Keep your arms relaxed by your sides. A rest period of 20 s occurred between each trial. Sufficient trials were performed to provide three valid sets of data. A test was invalidated if the participant moved their foot position during the test, changed their arm starting position, or opened their eyes during an eyes-closed task. Assessment of gait Participants were asked to walk barefoot, at a pace that felt comfortable to them, from one end of the laboratory to the other, in a line which passed over three force plates. Each participant received the same instructions: When I say go I want you to walk in a straight line to the marker at the other end of the room. Walk at a pace that feels comfortable to you. Participants continued walking the length of the laboratory until CSM had observed three clear force plate strikes (heel strike and toe-off occurring with the foot making contact with one plate only, without contacting the plate with the contralateral foot) for each foot. The biomechanical assessment lasted approximately 30 min.

ts continued walking the length of the laboratory until CSM had observed three clear force plate strikes (heel strike and toe-off occurring with the foot making contact with one plate only, without contacting the plate with the contralateral foot) for each foot. The biomechanical assessment lasted approximately 30 min. Outcome measures The following postural stability primary outcomes were assessed during standing: (1) root mean squared error and (2) velocity of the CoP in the anteroposterior direction (CoPRMSEAP and CoPVELAP, respectively, online supplementary appendix 1). CoP is a term that refers to the mean position of the forces acting under the feet at any instant in time. The RMSE (or SD) of the CoP position reflects the spread of these measurements over a particular time interval (in this case 30 s). The CoPVELAP refers to the mean displacement of the CoP in the anterior-posterior direction, divided by the sample time (1/1080 s) over the course of the 30 s trial. Reliability of CoPVEL has been reported as excellent (intraclass correlation (ICC) 0.8–0.95) and CoPRMSE reported as fair to good (ICC 0.32–0.58) for studies employing similar number of trials and trial durations as the current study.12 10.1136/bmjsem-2017-000286.supp1Supplementary file 1 The following outcome measures were assessed during gait: self-selected walking speed, stride length, cadence, maximum, minimum and total hip range of movement, peak hip flexor and extensor moments.

Outcome measures The following postural stability primary outcomes were assessed during standing: (1) root mean squared error and (2) velocity of the CoP in the anteroposterior direction (CoPRMSEAP and CoPVELAP, respectively, online supplementary appendix 1). CoP is a term that refers to the mean position of the forces acting under the feet at any instant in time. The RMSE (or SD) of the CoP position reflects the spread of these measurements over a particular time interval (in this case 30 s). The CoPVELAP refers to the mean displacement of the CoP in the anterior-posterior direction, divided by the sample time (1/1080 s) over the course of the 30 s trial. Reliability of CoPVEL has been reported as excellent (intraclass correlation (ICC) 0.8–0.95) and CoPRMSE reported as fair to good (ICC 0.32–0.58) for studies employing similar number of trials and trial durations as the current study.12 10.1136/bmjsem-2017-000286.supp1Supplementary file 1 The following outcome measures were assessed during gait: self-selected walking speed, stride length, cadence, maximum, minimum and total hip range of movement, peak hip flexor and extensor moments. Data extraction Force plate data (forces and moments) captured at 1080 Hz and filtered with a low pass Woltering filter (mean SE 10 mm2) were exported into Vicon’s Nexus software (V.1.8.1) to calculate biomechanical outcome measures.

The following outcome measures were assessed during gait: self-selected walking speed, stride length, cadence, maximum, minimum and total hip range of movement, peak hip flexor and extensor moments. Data extraction Force plate data (forces and moments) captured at 1080 Hz and filtered with a low pass Woltering filter (mean SE 10 mm2) were exported into Vicon’s Nexus software (V.1.8.1) to calculate biomechanical outcome measures. Industry-standard motion capture files (.c3d) containing force data were extracted. Force plate data were filtered with a low pass (10 Hz) Butterworth filter. CoP parameters were calculated using a proprietary programme written in Visual Basic for Applications (Microsoft Excel, Reading, UK). Sample size A sample size calculation was not conducted due to the lack of reported data of minimal clinically important difference (MCID) for the primary outcome measures (CoP parameters). This study aimed to recruit 20 asymptomatic participants age-matched and gender-matched to symptomatic participants recruited by the authors in a previous RCT.21

lation was not conducted due to the lack of reported data of minimal clinically important difference (MCID) for the primary outcome measures (CoP parameters). This study aimed to recruit 20 asymptomatic participants age-matched and gender-matched to symptomatic participants recruited by the authors in a previous RCT.21 Data analysis Independent t-tests for parametric, or Mann-Whitney U tests for non-parametric data, were applied to determine differences between groups for demographic data and gait outcomes. A mixed-repeated measures analysis of variance with two within-subject factors each with two levels—vision (eyes open and eyes closed) and support surface (firm and compliant)—determined possible significant main effects and interactions of the two groups for CoP variables. The alpha level for determining statistical significance was set at 0.05. Data were analysed using IBM SPSS V.20.0.0 (IBM). Results are presented as means (SD) unless otherwise stated. Results Recruitment and retention During the recruitment period (June 2010 to November 2011), 16 asymptomatic participants were age-matched and gender-matched with 16 CLBP participants. The recruitment of matched asymptomatic participants, over the age of 50 years, who had not experienced LBP over the past 12 months proved difficult. This prevented recruitment of the planned sample size of 20 participants per group. There was 100% retention with all 32 participants completing the data collection process.

recruitment of matched asymptomatic participants, over the age of 50 years, who had not experienced LBP over the past 12 months proved difficult. This prevented recruitment of the planned sample size of 20 participants per group. There was 100% retention with all 32 participants completing the data collection process. Baseline characteristics of participants Demographic characteristics of CLBP and asymptomatic individuals are presented in table 1. No differences were observed between groups other than self-reported pain scores. Participants with CLBP reported mild to moderate pain with a numerical rating score range of 3–8, and a mean duration of symptoms of 6.17 (SD 7.59, range 0.25–31) years. Table 1 Demographic data for chronic low back pain and asymptomatic participants Asymptomatic participants (n=16) Low back pain participants (n=16) P value Gender Male 8 (50.0%)* 8 (50.0%)* 1.00† Female 8 (50.0%)* 8 (50.0%)* Age (years) 37.3 (11.1) 36.8 (10.1) 0.90 Weight (kg) 76.3 (13.6) 73.4 (10.6) 0.52 Height (cm) 173.4 (9.3) 173.4 (8.9) 1.00 Numerical rating score for pain (0–10; 0=best) 0.0 (0.0) 5.9 (1.5) 0.00 Summary measures represent means (SD) or numbers (percentages). *Numbers (percentages). †Χ2 test, otherwise independent t-test. CoP parameters during standing

Asymptomatic participants (n=16) Low back pain participants (n=16) P value Gender Male 8 (50.0%)* 8 (50.0%)* 1.00† Female 8 (50.0%)* 8 (50.0%)* Age (years) 37.3 (11.1) 36.8 (10.1) 0.90 Weight (kg) 76.3 (13.6) 73.4 (10.6) 0.52 Height (cm) 173.4 (9.3) 173.4 (8.9) 1.00 Numerical rating score for pain (0–10; 0=best) 0.0 (0.0) 5.9 (1.5) 0.00 Summary measures represent means (SD) or numbers (percentages). *Numbers (percentages). †Χ2 test, otherwise independent t-test. CoP parameters during standing Table 2 presents data for the anteroposterior CoP parameter data for CLBP and asymptomatic participants during different standing conditions. There were no differences between the groups in CoPRMSEAP, or CoPVELAP for any of the four standing conditions (F(2.35, 70.38)=1.39, P=0.26, η2=0.04; F(1.76, 52.87)=0.47, P=0.60, η2=0.02, respectively). Table 2 Anteroposterior centre of pressure parameters for chronic low back pain and asymptomatic participants during different standing conditions CoPRMSEAP (mm) CoPVELAP (mm/s) Eyes open, firm surface Asymptomatic 3.76 (0.84) 6.57 (1.09) Chronic low back pain 4.21 (1.88) 7.14 (1.52) Eyes closed, firm surface Asymptomatic 3.93 (1.47) 7.14 (1.10) Chronic low back pain 4.23 (1.38) 7.39 (1.24) Eyes open, compliant surface Asymptomatic 8.29 (1.70) 10.97 (1.78) Chronic low back pain 9.10 (2.95) 12.57 (3.96) Eyes closed, complaint surface Asymptomatic 8.93 (1.45) 17.15 (4.29) Chronic low back pain 10.56 (2.85) 17.98 (4.38) Summary measures represent means (SD). AP, anteroposterior; CoP, centre of pressure; RMSE, root mean squared error; VEL, velocity.

CoPRMSEAP (mm) CoPVELAP (mm/s) Eyes open, firm surface Asymptomatic 3.76 (0.84) 6.57 (1.09) Chronic low back pain 4.21 (1.88) 7.14 (1.52) Eyes closed, firm surface Asymptomatic 3.93 (1.47) 7.14 (1.10) Chronic low back pain 4.23 (1.38) 7.39 (1.24) Eyes open, compliant surface Asymptomatic 8.29 (1.70) 10.97 (1.78) Chronic low back pain 9.10 (2.95) 12.57 (3.96) Eyes closed, complaint surface Asymptomatic 8.93 (1.45) 17.15 (4.29) Chronic low back pain 10.56 (2.85) 17.98 (4.38) Summary measures represent means (SD). AP, anteroposterior; CoP, centre of pressure; RMSE, root mean squared error; VEL, velocity. Spatiotemporal parameters of gait No differences were observed between groups for any of the spatiotemporal gait parameters assessed (table 3). Table 3 Spatiotemporal parameters of gait in chronic low back pain and asymptomatic individuals Asymptomatic group Chronic low back pain group P value Walking speed (m/s) 1.32 (0.13) 1.25 (0.20) 0.26 Cadence (steps per minute) 115.14 (6.59) 112.43 (11.81) 0.42 Stride length (m) 1.38 (0.12) 1.33 (0.13) 0.33 Summary measures represent means (SD). Analysis by independent t-test. Hip moments and range of movement during gait No differences were detected between groups for maximum, minimum and total ranges of movement at the hip in the sagittal plane during gait (table 4). No differences were observed between groups for peak hip flexor or extensor moments during gait (table 4). Table 4 Sagittal plane hip range of movement and peak hip joint moments during gait in people with chronic low back pain and asymptomatic individuals

Hip moments and range of movement during gait No differences were detected between groups for maximum, minimum and total ranges of movement at the hip in the sagittal plane during gait (table 4). No differences were observed between groups for peak hip flexor or extensor moments during gait (table 4). Table 4 Sagittal plane hip range of movement and peak hip joint moments during gait in people with chronic low back pain and asymptomatic individuals Asymptomatic Chronic low back pain P value Left maximum hip flexion (°) 34.35 (5.55) 33.70 (8.55) 0.78 Right maximum hip flexion (°) 34.46 (4.51) 33.82 (9.17) 0.79 Left maximum hip extension (°) −9.71 (7.39) −10.44 (9.02) 0.80 Right maximum hip extension (°) −9.40 (6.67) −9.12 (8.74) 0.92 Left hip range of movement (°) 44.07 (3.94) 44.14 (4.79) 0.97 Left hip extensor moment (Nmm/kg) 1029.30 (329.38) 955.80 (429.78) 0.58 Right hip extensor moment (Nmm/kg) 960.99 (235.24) 1029.57 (460.62) 0.94* Left hip flexor moment (Nmm/kg) −990.76 (184.25) −1098.07 (231.85) 0.14 Right hip flexor moment (Nmm/kg) −1041.87 (174.80) −977.77 (194.64) 0.31 Summary measures represent means (SD). *Represents Mann-Whitney U test for non-parametric data, otherwise independent t-test conducted. Nmm/kg, Newton-millimetre/kilogram.

Asymptomatic Chronic low back pain P value Left maximum hip flexion (°) 34.35 (5.55) 33.70 (8.55) 0.78 Right maximum hip flexion (°) 34.46 (4.51) 33.82 (9.17) 0.79 Left maximum hip extension (°) −9.71 (7.39) −10.44 (9.02) 0.80 Right maximum hip extension (°) −9.40 (6.67) −9.12 (8.74) 0.92 Left hip range of movement (°) 44.07 (3.94) 44.14 (4.79) 0.97 Left hip extensor moment (Nmm/kg) 1029.30 (329.38) 955.80 (429.78) 0.58 Right hip extensor moment (Nmm/kg) 960.99 (235.24) 1029.57 (460.62) 0.94* Left hip flexor moment (Nmm/kg) −990.76 (184.25) −1098.07 (231.85) 0.14 Right hip flexor moment (Nmm/kg) −1041.87 (174.80) −977.77 (194.64) 0.31 Summary measures represent means (SD). *Represents Mann-Whitney U test for non-parametric data, otherwise independent t-test conducted. Nmm/kg, Newton-millimetre/kilogram. Discussion In contrast to much other research, the current findings suggest that postural control during standing, and the kinetics, kinematics and spatiotemporal parameters of gait do not differ between people with CLBP of a mild to moderate intensity and asymptomatic individuals. There were no differences between people with and without CLBP in postural stability during all standing conditions assessed. During barefoot gait, both groups presented with similar peak hip moments and ranges of movement, and spatiotemporal parameters of gait. Hence, all stated hypotheses are rejected.

c individuals. There were no differences between people with and without CLBP in postural stability during all standing conditions assessed. During barefoot gait, both groups presented with similar peak hip moments and ranges of movement, and spatiotemporal parameters of gait. Hence, all stated hypotheses are rejected. CoP parameters There was no difference in postural stability between CLBP and asymptomatic individuals during stable and more challenging standing conditions. These findings differ from previous research1–4 possibly due to methodological variation. della Volpe et al 2 assessed a smaller sample (n=12 per group) with an ‘instrumented platform system’, constructed of a moveable support surface and moveable visual surround likely to present participants with a greater postural challenge. This may contribute to the reduced postural stability observed in the CLBP group in their study.2 Brumagne et al 1 assessed a larger sample size than the current study (n=45); however, trials were only repeated once—the current study averaged three trials per standing condition, likely to increase reliability of data.11 Although Brumagne et al 1 reported reduced postural stability in the CLBP group during more challenging standing conditions, the between-group difference in CoPRMSEAP was 1.8 mm, and the P value, 0.046–bordering on non-significance. In the current study, the non-significant difference in CoPRMSEAP between the symptomatic and asymptomatic groups during the most challenging postural condition was 1.76 mm. Although Brumagne et al 1 demonstrated statistical significance, based on the very similar yet non-significant between-group difference in CoP displacement found in the current study (and in the absence of knowledge regarding cause or effect), it seems unlikely that such a minimal difference in CoPRMSEAP is responsible for the clinical differences in pain and disability observed between the two groups. Mientjes and Frank3 assessed a small sample (n=8 per group) and although reported significant differences between CLBP and asymptomatic groups during challenged standing conditions, these differences were small (less than 2 mm) and similar to those of both the current study and Brumagne et al’s study.1 Furthermore, Mientjes and Frank3 report a mean pain score of 0.5 in the ‘asymptomatic’ group raising concerns that the asymptomatic data may not be a true representation of a pain-free population.

, these differences were small (less than 2 mm) and similar to those of both the current study and Brumagne et al’s study.1 Furthermore, Mientjes and Frank3 report a mean pain score of 0.5 in the ‘asymptomatic’ group raising concerns that the asymptomatic data may not be a true representation of a pain-free population. The CoP parameters assessed in a research study may influence the reliability of results. CoP velocity consistently demonstrates the best overall reproducibility of all CoP parameters in the short and long terms,12 30 hence, findings from this parameter are likely to provide more reliable conclusions to those gained from CoPRMSEAP data or other CoP parameters. The current study demonstrated similar CoPVELAP in people with and without CLBP, whereas previous research has demonstrated reduced4 31 (n=24 and 22 per group, respectively) and increased2 32 33 (n=12, 12 and 10 per group, respectively) CoP velocities. These mixed results suggest it is likely that research demonstrating no difference between groups has been conducted, however, due to publication bias may not have gained acceptance for publication. Interestingly, the studies conducted with the greater sample size demonstrate poorer postural control in the asymptomatic groups, not the CLBP groups. Furthermore, findings from previous research31 34 highlight that the small differences observed between groups in this study may be due to random error associated with the reliability of the measurement technique and not clinical change.

monstrate poorer postural control in the asymptomatic groups, not the CLBP groups. Furthermore, findings from previous research31 34 highlight that the small differences observed between groups in this study may be due to random error associated with the reliability of the measurement technique and not clinical change. Differences in participant demographics (eg, age,2 31 gender33 or disability4) and methodological design (eg, trial duration and repetitions4 30 32) make it difficult to directly compare study findings. Due to the numerous factors which may contribute to the variation in CoP outcomes reported, comparison of one study data with another is likely to reveal potential differences; however, choice of outcome measures and the number and duration of trials conducted in the current investigation improves the likelihood that data collected are reliable.

which may contribute to the variation in CoP outcomes reported, comparison of one study data with another is likely to reveal potential differences; however, choice of outcome measures and the number and duration of trials conducted in the current investigation improves the likelihood that data collected are reliable. Gait No differences were detected in spatiotemporal parameters between groups. In support of the current study findings, Al-Obaidi et al 5 and Simmonds and Claveau35 demonstrated no difference in cadence and self-selected walking speed, respectively, between people with and without CLBP (with a similar age and gender to those in the current study). However, research investigating participants with similar self-reported pain (mild to moderate) to the current study demonstrated reduced walking speed,5–8 stride time9 10 and stride length5 6 in people with LBP. The current study averaged data from three trials for each participant, aiming to improve reliability,12 whereas other studies analysed data from only one walking trial,5 8 9 possibly reducing data reliability. In addition, where other studies investigated predominantly6 or all male participants,8 9 the current study assessed male and female participants, enabling findings to be more representative of a general population. Furthermore, the current study assessed participants walking on normal ground, as opposed to on a treadmill,7 9 10 hence, the current study findings are likely to be more representative of a natural walking pattern. These factors increase confidence that the current results are a more reliable and valid representation of gait in CLBP than that reported in previous research.5–10

on normal ground, as opposed to on a treadmill,7 9 10 hence, the current study findings are likely to be more representative of a natural walking pattern. These factors increase confidence that the current results are a more reliable and valid representation of gait in CLBP than that reported in previous research.5–10 In contrast to the current study, previous research has reported reduced hip range of movement in people with LBP during gait compared with asymptomatic individuals.9 This may be due to co-contraction of muscles crossing the hip and pelvic region36 limiting hip movement, or from participants reducing step length, and hence hip range, in an attempt to reduce potentially detrimental ground reaction forces at heel strike.15 37 Reduced hip range demonstrated by Vogt et al 9 occurred during treadmill gait, hence may not be representative of natural gait.38 Furthermore, Vogt et al 9 assessed hip range by attaching an electrical goniometer to the greater trochanter. This method of assessment provides less reliable data than the retroreflective marker system used in the current study39 40; again increasing confidence that the current results are likely a more valid representation of gait in people with CLBP. In the current study, due to the lack of difference in stride length between CLBP and asymptomatic individuals, the similar range of hip movement between the two groups was an expected finding.

9 40; again increasing confidence that the current results are likely a more valid representation of gait in people with CLBP. In the current study, due to the lack of difference in stride length between CLBP and asymptomatic individuals, the similar range of hip movement between the two groups was an expected finding. Strengths and limitations The authors did not conduct a formal sample size calculation using MCID data due to the absence of reported MCID data within the literature. However, SEs of measurements from repeatability studies for similar sample populations are reported in the literature for the more reliable postural stability outcome measure of CoPVELAP.41 If minimal detectable change (MDC) is substituted for MCID in a sample size calculation (where alpha=0.05, beta=0.8, MDC for CoPVELAP=5.4 mm, SD of groups=1.09 and 1.52 where groups contain equal number of participants) this suggests that six participants would need to be recruited. The authors note the convenience sample recruited in this study for the asymptomatic participants may not be representative of the general population; however, potential asymptomatic participants were required to meet inclusion and exclusion criteria with a view to reducing this potential source of sampling bias. Sampling bias may have been reduced in the symptomatic sample as recruitment of participants occurred more broadly from a population with CLBP in multiple recruitment sites.

l asymptomatic participants were required to meet inclusion and exclusion criteria with a view to reducing this potential source of sampling bias. Sampling bias may have been reduced in the symptomatic sample as recruitment of participants occurred more broadly from a population with CLBP in multiple recruitment sites. Although participants were matched for age and gender, the authors note that unaccounted confounders, such as anthropometric factors, level of physical activity, or kinesiophobia, may have influenced study results. Given the small sample size in the current study, multivariate modelling was deemed inappropriate. Research investigating the influence of anthropometric factors (including body height, limb and trunk length, and body mass) on postural balance concluded postural balance assessed with eyes open and closed is only slightly42 and moderately43 influenced by these anthropometric variables; the variables that most influenced postural balance being height and body mass index. The similarity of height and weight between groups in the current study (table 1) is therefore reassuring.

al balance assessed with eyes open and closed is only slightly42 and moderately43 influenced by these anthropometric variables; the variables that most influenced postural balance being height and body mass index. The similarity of height and weight between groups in the current study (table 1) is therefore reassuring. The current study recruited CLBP participants from clinical populations,21 who had sought medical opinion regarding their symptoms, hence, represented a typical population treated within physiotherapy departments. Previous research has recruited participants from alternative sources such as university populations,4 which may not be representative of the subgroup of CLBP individuals who seek medical guidance; hence caution should be taken if relating findings from such studies to a person with CLBP who is attending for treatment. Further research The velocity of the CoP is reported as the most reliable CoP parameter; however, it is unclear if this measure is the most appropriate to detect difference in postural stability. Hence, a difference in postural control between the symptomatic and asymptomatic groups may have been present, but not detected. Alternative balance measures could be investigated, such as the forward reach test to determine whether more functional or challenging outcomes possess the necessary discriminatory value to detect differences in balance in people with and without CLBP and assist in confirming whether such differences exist.

t detected. Alternative balance measures could be investigated, such as the forward reach test to determine whether more functional or challenging outcomes possess the necessary discriminatory value to detect differences in balance in people with and without CLBP and assist in confirming whether such differences exist. Clinical implications Based on the findings of this study, clinicians can be informed that standing postural stability, kinetic, kinematic and spatiotemporal parameters of gait in people with and without mild to moderate CLBP may not differ, and that treatments directed at influencing postural stability (eg, standing on a wobble board) or specific parameters of gait may be an unnecessary addition to a treatment programme. Conclusions In contrast to previous research, this study suggests that people with mild to moderate CLBP may present with similar standing postural control, hip moments and range of movement, and spatiotemporal parameters of gait to asymptomatic individuals. The authors thank all the participants for their contributions to this study. The authors also thank Tanya Forster, Andrew Lewis and Jonathan Noble for their assistance during the data collection and analysis process. The authors thank all the physiotherapy departments that participated in this study, namely: Balance Performance Physiotherapy, Clapham, London, UK, SW4; Chelsea and Westminster Hospital, Chelsea, London, UK, SW10; Queen Mary’s Hospital, Roehampton, UK, SW14; and St George’s Hospital, Tooting, UK, SW18.

alysis process. The authors thank all the physiotherapy departments that participated in this study, namely: Balance Performance Physiotherapy, Clapham, London, UK, SW4; Chelsea and Westminster Hospital, Chelsea, London, UK, SW10; Queen Mary’s Hospital, Roehampton, UK, SW14; and St George’s Hospital, Tooting, UK, SW18. Contributors: CSM, the primary investigator, was involved in all aspects of the study, including methodology, data collection, analysis and interpretation of data, and was the primary author of the article. All authors contributed to methodology, data interpretation and editing of the manuscript for publication. All authors approved the final revision of the submitted manuscript. In addition, JSL received grant funding for the study, AS contributed to data collection. The corresponding author had full access to all data in the study and had final responsibility for the decision to submit for publication. Funding: The clinical study from which CLBP participants in the current study were recruited was funded by a Masai GB project grant. Masai GB had no role in the writing of the manuscript, data collection, analysis or interpretation; study design; patient recruitment; or the decision to submit for publication. Competing interests: None declared. Ethics approval: Ethical approval for the recruitment of symptomatic participants (Outer North London Research Ethics Committee (REC: 10/H0724/7)) and asymptomatic participants (King’s College London Research Ethics Subcommittee (BDM/10/11–7)) was gained. Provenance and peer review: Not commissioned; externally peer reviewed.

What are the new findings? Appropriately trained Sport and Exercise Medicine consultants have excellent inter-rater and intra-rater reliability using power Doppler enabled ultrasound to assess neovascularity in chronic Achilles and patellar tendinopathy. Previously, such results have only been demonstrated among sonographers and radiologists. Neovascularity is present in two-thirds of chronic Achilles and patellar tendinopathies.

What are the new findings? Appropriately trained Sport and Exercise Medicine consultants have excellent inter-rater and intra-rater reliability using power Doppler enabled ultrasound to assess neovascularity in chronic Achilles and patellar tendinopathy. Previously, such results have only been demonstrated among sonographers and radiologists. Neovascularity is present in two-thirds of chronic Achilles and patellar tendinopathies. Introduction Incidence and prevalence of Achilles and patellar tendinopathies are increased in physically active populations.1–3 During the progression of tendinopathic disease, the formation and ingrowth of neovascular structures into the tendon is a widely recognised pathological hallmark.4–7 The role of neovascularisation within tendinopathy is the source of significant interest regarding its contribution to the pathophysiology of the injured and healing tendon.6 8 It may also be associated with neoneuralisation9 exacerbated by nociceptive stimuli such as glutamate10 and substance P.11 This provides an explanation as to why the point of ingress of vessels into the tendon is often associated with the site of maximal pain on palpation.12 Neovascularisation is a normal physiological response in early repair, and its continued presence is likely to be a result of prolonged overload that delays the healing timeline.13 Association between neovascularity and pain and function has been shown in small interventional case and longitudinal studies14–17 yet refuted in Randomised Controlled Trial (RCT)18 and only weakly correlated in a larger prospective cohort.7 This supports the theory that neovascularity may be more a sign of chronicity than severity. Polidocanol14 and high volume image-guided injections (HVIGI)19 target the presence of vessels in order to address this proposed pain stimulus.14 19 To that end, the ability to reliably locate and quantify vascularity is clinically relevant.

the theory that neovascularity may be more a sign of chronicity than severity. Polidocanol14 and high volume image-guided injections (HVIGI)19 target the presence of vessels in order to address this proposed pain stimulus.14 19 To that end, the ability to reliably locate and quantify vascularity is clinically relevant. The hypoechoic areas seen in tendinopathy on gray-scale ultrasound (US) offer limited information which can be enhanced by the use of power Doppler (PD) to identify hyperaemic neovasculature.7 10 14 20 21 In clinical practice, there have been 3 methods of assessing the degree of neovascularisation. The first is a simple statement of whether vessels are either absent or present in the tendon structure.8 Depending on the significance of degree of vascularity as discussed above, the utility of such binary grading is likely to be limited. A second system involves determining the PD signal surface area by pixel count. Although a seemingly robust method of capturing a quantifiable measure of neovascularity, these authors recognise that interpretation of findings in clinical practice will be both machine and operator setting dependent.22 23 A third method was pioneered by Ohberg and Alfredson14 who undertook PD US examination of symptomatic Achilles tendons to understand the extent of neovascularity and locate target areas for sclerosing polidocanol injections. In this study, an experienced sonographer assessed neovascularisation as 0, 1+, 2+, 3+ or 4+. In the absence of vessels, the tendon scored a 0. A score of 1+ was ascribed if the tendon showed 1–2 small vessels primarily around the anterior surface of the tendon. Scores of 2+, 3+ or 4+ were given if the vessel contained 2, 3 or more than 4 vessels respectively throughout its structure. They later adapted this score for use in patellar tendinopathy with the score of 1+ referring to 1–2 vessels mostly in the posterior portion of that tendon.16

the anterior surface of the tendon. Scores of 2+, 3+ or 4+ were given if the vessel contained 2, 3 or more than 4 vessels respectively throughout its structure. They later adapted this score for use in patellar tendinopathy with the score of 1+ referring to 1–2 vessels mostly in the posterior portion of that tendon.16 Sengkerij et al21 undertook an inter-rater reliability study in which a cohort of radiologists undertook PD US examinations of symptomatic and asymptomatic Achilles tendons and rated the neovascularisation referring to the system described above as the Modified Ohberg Score (MOS). Of the 50 tendons examined in the study, each underwent scans by 2 of 8 radiologists. Excellent reliability was found with absolute agreement of 62% and an intraclass correlation coefficient (ICC) of 0.85. Sunding et al24 used a more qualitative 4-point MOS for neovascularity: 0, None; 1, Mild; 2, Moderate; 3, Severe. They therefore used Kappa in their analysis of 27 Achilles and 26 patella tendons finding good inter-rater agreements of 0.63 and 0.70 respectively using 2 sonographers. US scanning is a common skill used frequently in clinical practice, yet there are no data to support its reliability among Sport and Exercise Medicine (SEM) consultants. This study aims to contribute to future research by establishing reliability among SEM consultants when assessing tendon neovascularity using the 5-point MOS.

ing is a common skill used frequently in clinical practice, yet there are no data to support its reliability among Sport and Exercise Medicine (SEM) consultants. This study aims to contribute to future research by establishing reliability among SEM consultants when assessing tendon neovascularity using the 5-point MOS. Method Using a power of 80% (β-level=0.8) and an α-level of 0.05 anticipating good to excellent reliability (P0>0.7 to P1=0.9), the calculations of Walter et al25 were used to estimate the requirement for at least 5 raters (n) of 10 participants (k). Participants All participants gave written informed consent. Participants self-identified following advertisement for volunteers with Achilles or patella tendon pain. A subjective and objective assessment from an experienced physiotherapist (JW) against inclusion/exclusion criteria (table 1) was conducted consecutively. Participants were requested to avoid any heavy sporting or loading activity 24 hours prior to US examination. Table 1 Inclusion and exclusion criteria Inclusion criteria Exclusion criteria 18–60 years of age Clinical findings of Achilles or patellar tendinopathy: Localised pain to tendon Pain/stiffness on waking Pain on commencing loaded activity Pain after activity Pain on mobilising after rest

Participants All participants gave written informed consent. Participants self-identified following advertisement for volunteers with Achilles or patella tendon pain. A subjective and objective assessment from an experienced physiotherapist (JW) against inclusion/exclusion criteria (table 1) was conducted consecutively. Participants were requested to avoid any heavy sporting or loading activity 24 hours prior to US examination. Table 1 Inclusion and exclusion criteria Inclusion criteria Exclusion criteria 18–60 years of age Clinical findings of Achilles or patellar tendinopathy: Localised pain to tendon Pain/stiffness on waking Pain on commencing loaded activity Pain after activity Pain on mobilising after rest Symptoms within last 12 months lasting for period in excess of 2 months Systemic condition causing inflammatory tendinopathy Six SEM consultant raters were recruited to the study having completed professional training for a mean of 8 years (range 1–23 years). Four had undergone additional postgraduate training in musculoskeletal US, while the remainder had attained significant clinical experience in scanning both Achilles and patella tendons. Data collection Prior to testing, the consultant raters received an instructional brief on the examination protocol which included standardised positioning of participants, surface pressure produced through the probe at the contact point, labelling of recorded scans and clarity as to the descriptions of the MOS (table 2). Participants were subjected to repeat PD US examination by all 6 raters. Table 2 Description of the 5-point Modified Ohberg Score

Data collection Prior to testing, the consultant raters received an instructional brief on the examination protocol which included standardised positioning of participants, surface pressure produced through the probe at the contact point, labelling of recorded scans and clarity as to the descriptions of the MOS (table 2). Participants were subjected to repeat PD US examination by all 6 raters. Table 2 Description of the 5-point Modified Ohberg Score Modified Ohberg Score Description 0 No vessels 1+ One vessel anterior to the Achilles One vessel posterior to the patella tendon 2+ One to two vessels throughout the tendon 3+ Three vessels throughout the tendon 4+ Four or more vessels throughout the tendon Scans were performed using a Logic 9 US Scanner (GE Medical Systems, Chalfont St Giles, UK) with a high frequency linear ML 6-15 MHz transducer (GE Medical Systems). Consultants oriented themselves to the tendon using standard gray-scale US to identify areas of the tendon to focus subsequent PD scanning. Once PD mode was engaged, the first consultant  when scanning a participant’s tendon set and standardised the PD gain used for the subsequent consultants. The first consultant to scan was varied for each patient to control any bias in setting of PD gain. Where possible, participants were scanned in series by all 6 consultants within the same session. To control for changes in tendon loading that may affect vessel signal, participants were instructed to avoid heavy sporting or loading activity 24 hours prior to repeat examinations.

ontrol any bias in setting of PD gain. Where possible, participants were scanned in series by all 6 consultants within the same session. To control for changes in tendon loading that may affect vessel signal, participants were instructed to avoid heavy sporting or loading activity 24 hours prior to repeat examinations. To prevent vessel compression, positioning of the tendon was strictly controlled. Achilles tendons were scanned in the prone position with the foot in a relaxed non-dorsiflexed position over the end of the treatment couch (figure 1). Patella tendons were scanned in a standardised position of approximately 30° of knee flexion (figure 2A). Consultants were also permitted to scan in full knee extension (figure 2B) to remove any tension from the tendon. Figure 1 Standardised position for ultrasound imaging of the Achilles tendon. Figure 2 (A)Standardised position for US imaging of the patella tendon (position 1). (B) Standardised position for US imaging of the patella tendon (position 2). US, ultrasound. Data storage Participants were assigned an anonymised unique identifier used to save images on the Logic 9 Scanner. Scans were saved once each consultant determined the MOS. Consultants were requested to save any scans that contributed to them formulating an overall MOS for the tendon. All MOS scores were recorded on paper retained out of sight of other raters. Three weeks after the initial examination, raters rescored the saved scans. Clinical information including previous MOS ratings was not available to consultant raters at any stage.

contributed to them formulating an overall MOS for the tendon. All MOS scores were recorded on paper retained out of sight of other raters. Three weeks after the initial examination, raters rescored the saved scans. Clinical information including previous MOS ratings was not available to consultant raters at any stage. Statistical analysis Inter-rater and intra-rater reliability across the cohort of consultants was calculated using ICC and Kappa in IBM SPSS Statistics V.23 (IBM, Amonk, New York, USA) with a one-way random, single measures model. Fleiss defines an excellent ICC as higher than 0.75,26 a good ICC as 0.4–0.75 and poor if less than 0.4. For clinical investigations, an ICC of 0.60 has been described as the minimally acceptable level.27 Fleiss Kappa statistic was used to measure the level of agreement between raters, while a weighted Kappa was used to allow marginal agreement between individual raters’ baseline and follow-up scores.28 These were calculated using SPSS extension bundles. Both ICC and Kappa were used to provide comparability across the literature. Results Eleven participants (7 males and 4 females with a mean age of 36.2 years (SD 6.25)) were recruited to the study and provided 16 symptomatic tendons. The mean duration of symptoms was 20.1 months (range 6–96 months). Presence of neovascularisation Eight symptomatic Achilles and 8 patella tendons (n=16) were assessed by 6 consultant raters (K=6) resulting in 96 examinations. Neovascularisation was reported in 65.6% of scans and the MOS distribution is shown in table 3.

Results Eleven participants (7 males and 4 females with a mean age of 36.2 years (SD 6.25)) were recruited to the study and provided 16 symptomatic tendons. The mean duration of symptoms was 20.1 months (range 6–96 months). Presence of neovascularisation Eight symptomatic Achilles and 8 patella tendons (n=16) were assessed by 6 consultant raters (K=6) resulting in 96 examinations. Neovascularisation was reported in 65.6% of scans and the MOS distribution is shown in table 3. Table 3 Distribution of the Modified Ohberg Scores across all 96 examinations Modified Ohberg Score Number of scores recorded Percentage of scores recorded 0 33 34.4 1+ 23 23.9 2+ 16 16.7 3+ 5 5.2 4+ 19 19.8 Inter-rater reliability In 6 tendons (37.5%), there was absolute scoring agreement across all raters, and a further 6 tendons had agreement from at least 4 of the 6 raters. This represents an agreement of 4 or more consultant raters in 75% of the participants. The ICC for inter-rater reliability of the MOS was 0.86 (95% CI, 0.76 to 0.94), illustrating excellent inter-rater reliability. The Fleiss Kappa agreement for inter-rater reliability was 0.52 (95% CI, 0.45 to 0.59). It should be noted the more modest result of Fleiss Kappa is not weighted and based on absolute agreement. Intra-rater reliability Comparison of the agreement between baseline and follow-up scores can be found in table 4. Across raters absolute agreement between initial and follow-up scans occurred in 84 of 95 scans (one scan did not save), representing 88% of scans undertaken. Table 4 Agreement between baseline and follow-up scoring

Intra-rater reliability Comparison of the agreement between baseline and follow-up scores can be found in table 4. Across raters absolute agreement between initial and follow-up scans occurred in 84 of 95 scans (one scan did not save), representing 88% of scans undertaken. Table 4 Agreement between baseline and follow-up scoring Follow-up scan Baseline scan 0 1+ 2+ 3+ 4+ 0 29 3 0 0 0 1+ 1 19 2 0 1 2+ 0 2 14 0 1 3+ 0 0 0 4 1 4+ 0 0 0 0 18 The ICC for intra-rater reliability within the cohort was 0.95 (95% CI, 0.92 to 0.97) illustrating excellent reliability of the MOS when reviewing scans. The weighted Kappa agreement score was 0.91 (95% CI 0.85 to 0.97) demonstrating excellent reliability.

3 0 0 0 1+ 1 19 2 0 1 2+ 0 2 14 0 1 3+ 0 0 0 4 1 4+ 0 0 0 0 18 The ICC for intra-rater reliability within the cohort was 0.95 (95% CI, 0.92 to 0.97) illustrating excellent reliability of the MOS when reviewing scans. The weighted Kappa agreement score was 0.91 (95% CI 0.85 to 0.97) demonstrating excellent reliability. Discussion This study has demonstrated excellent inter-rater reliability (ICC 0.86) of a cohort of SEM consultants when using the MOS to rate neovascularisation in Achilles and patella tendons. Furthermore, excellent intra-rater reliability (ICC 0.95) was demonstrated via review of saved scans. Neovascularisation was reported in 65.6% of the original scans in keeping with previous studies of Achilles and patellar tendinopathy.21 29–31 These results suggest that diagnostic US scoring of tendinopathic neovascularisation can be reliably employed in the SEM setting. The most important variable that will allow results to be extrapolated to a wider audience is the degree of clinical exposure to tendon assessment and ultrasonography clinicians have undertaken. All clinicians involved in this study described themselves as highly competent in the ultrasonography of Achilles and patella tendons, although this was not corroborated independently the results are comparable with those reported using radiologist or sonographer raters.21 24

ultrasonography clinicians have undertaken. All clinicians involved in this study described themselves as highly competent in the ultrasonography of Achilles and patella tendons, although this was not corroborated independently the results are comparable with those reported using radiologist or sonographer raters.21 24 In terms of distribution of MOS, it can be noted (table 3) that there was a lower number of Grade 3+ scores attributed to any of the primary scans undertaken. This is in keeping with the findings of Sengkerij and colleagues21 who found that only 9% of their scans received a 3+ score. An explanation for this is possibly that the raters find it easy to identify 1–2 vessels within the tendon structure, but once vascularity is more prevalent, it may be difficult to differentiate the existence of additional single vessels, the result being a lean towards a higher score of 4+. As the results of this study agree with those of Sengkerij et al,21 it may provide evidence to suggest that the 5-point scale is reduced to a 4-point scale, with the final score (grade 3) representing 3 or more vessels.

te the existence of additional single vessels, the result being a lean towards a higher score of 4+. As the results of this study agree with those of Sengkerij et al,21 it may provide evidence to suggest that the 5-point scale is reduced to a 4-point scale, with the final score (grade 3) representing 3 or more vessels. This study builds on the work of Sengkerij et al21 who did not attempt to control any of the operator variables once the PD setting was enabled. Of particular importance in this respect is the gain setting, which regulates the sensitivity of the system to flow.32 Essentially, too high a gain and the amount of random noise artefact in the form of colour foci in the image will increase, potentially affecting the clarity of colour that indicates vascular flow. In this study, the first consultant to scan a given participant was asked to select and set what they considered to be an appropriate gain, which other consultants would then use and this stayed consistent for that participant. As a result, the sensitivity of the system was standardised for each participant. Sengkerij et al21 used 2 radiologists each scanning session of a larger cohort (n=8). In such an uncrossed design, ICC does not readily deal with missing data. Using default SPSS settings with a one-way random ICC model cases that are not scanned by all raters would be excluded from the analysis. One solution to this would be to arrange data in 2 columns giving the appearance of a full data set from 2 raters exclusively. Collapsing data across empty columns and unequal proportions of scans undertaken by raters could overestimate the inter-rater ICC they reported of 0.85, though such limitations can only be assumed.

solution to this would be to arrange data in 2 columns giving the appearance of a full data set from 2 raters exclusively. Collapsing data across empty columns and unequal proportions of scans undertaken by raters could overestimate the inter-rater ICC they reported of 0.85, though such limitations can only be assumed. Arguably, the MOS is only a semiquantitative scale; therefore, this study has also used Fleiss and weighted Kappa. These favourable results compare well with Sunding et al who reported Kappas of 0.63 and 0.70 in Achilles and patellar tendinopathy, respectively using 2 independent sonographers.24 Weighted Kappa, which, like ICC, recognises marginal agreement and does not extend to comparison of more than 2 data sets. This presents a problem for Kappa analysis of more than 2 raters. The commonly used Fleiss Kappa result in this study should be interpreted differently from the other results because it can only reward absolute agreement. Light’s Kappa, which is the mean of all rating pairs,33 is a different solution yet not often reported. Therefore, the Fleiss Kappa has been presented in this study alongside an excellent inter-rater ICC, which has been able to provide context.

ently from the other results because it can only reward absolute agreement. Light’s Kappa, which is the mean of all rating pairs,33 is a different solution yet not often reported. Therefore, the Fleiss Kappa has been presented in this study alongside an excellent inter-rater ICC, which has been able to provide context. A strength of this study was the ability to recruit an appropriately powered number of consultant raters and where possible have them available for sequential scanning of all participants. Shoukri et al27 noted the difficulty in recruiting a number of suitably qualified specialists within the Hospital setting. While a large cohort of SEM consultants undertook ratings, short notice changes in availability were a limitation. The initial design aim for each participant to be scanned by 6 consultants sequentially in one session was achieved for 9 of 16 tendons. The remainder had to undertake subsequent scanning sessions (within a mean of 7.8 days). Boesen et al22 reported an increase in vessel appearance as an acute response to exercise, although postulate that this may be short-lived. As a result, those participants that were required to return for subsequent scans were requested to refrain from any activity that would either aggravate their symptoms or potentially make an acute change to the vessel appearance. This variable would have a potential to reduce consultant agreement on their primary scans of each patient, but despite this, excellent reliability was still demonstrated.

were requested to refrain from any activity that would either aggravate their symptoms or potentially make an acute change to the vessel appearance. This variable would have a potential to reduce consultant agreement on their primary scans of each patient, but despite this, excellent reliability was still demonstrated. Further research The findings of this study have direct implications for future research and clinical practice. Primarily, a subcohort of the consultants in this study are also involved in a large RCT34 to investigate the effects of HVIGI (with and without steroid) on chronic Achilles and patella tendon pain. The neovessels found on scanning of the participants will be targeted by the injections delivered, and therefore this reliability study serves to provide assurance that the consultants are reliable when scoring neovasculature presence and extent. Furthermore, it can be extrapolated that consultants of similar experience to those in this study will provide reliable PD US assessments for future studies using the MOS, as either an outcome measure or target for intervention. In the clinical setting, this study provides evidence to support the use of diagnostic US in SEM clinical practice by proving reliability both in initial scanning and in the test-retest situation, such as follow-up scanning postintervention.

tudies using the MOS, as either an outcome measure or target for intervention. In the clinical setting, this study provides evidence to support the use of diagnostic US in SEM clinical practice by proving reliability both in initial scanning and in the test-retest situation, such as follow-up scanning postintervention. Conclusion This study demonstrated excellent inter-rater and intra-rater reliability of the scoring of neovasculature within tendon structure among SEM consultants. While this contributes to the clinical assessment of patients with tendon pain, it does not explain the role of neovasculature in the pathological process or as a source of pain in chronic tendinopathy. Well-controlled research is required to establish whether procedures targeting neovascularisation can effect improvements in validated pain and functional outcome measures. Contributors: JW and RMB-D contributed equally in study design, ethical approval, study implementation, clinical protocols, drafting and editing of the manuscript. ANB is responsible for study design, clinical protocols, ethical approval, clinical and scientific supervision of manuscript drafting, editing and review. DTPF, ML and CR are responsible for study design, scientific protocols, clinical and scientific supervision, manuscript editing and review. PCW is responsible for study design, clinical protocols, scientific protocols, clinical and scientific supervision, manuscript editing and review. All authors read and approved the final manuscript.

e responsible for study design, scientific protocols, clinical and scientific supervision, manuscript editing and review. PCW is responsible for study design, clinical protocols, scientific protocols, clinical and scientific supervision, manuscript editing and review. All authors read and approved the final manuscript. Funding: This study has attracted funding through Loughborough University from the Higher Education Funding Council for England and a PhD Studentship. Competing interests: None declared. Ethics approval: Ethical approval was obtained via both the Ministry of Defence Research and Ethics Committee (684/MODREC/15) and the Cardiff Metropolitan University School of Sport Research Ethics Committee. Provenance and peer review: Not commissioned; externally peer reviewed.

inal ischaemia, without acute infarction’1 and is diagnosed by the patients’ history, a neurological examination and/or neuroimaging (typically a CT head scan). Typical symptoms of TIA include the rapid onset of speech disturbance, unilateral weakness or sensory loss, monocular blindness, visual field defect or ataxia. Aetiology of TIAs: TOAST classification system The underlying aetiology of each TIA event can be classified as per the TOAST classification system.2 The TOAST classification denotes the underlying cause of the TIA event and this paper focuses on secondary prevention following a TIA due to atherosclerosis or small vessel occlusion, two TOAST subtypes. TIA risk factors TIAs are most commonly caused by the embolic or thrombotic consequences of atherothrombotic disease,3 which is similar to the underlying pathological mechanism for cardiovascular disease.4–6 In addition to sharing a similar underlying pathological mechanism, cerebrovascular and cardiovascular disease share common underlying risk factors.5 7 The modifiable risk factors for all vascular diseases include smoking, excessive alcohol intake, physical inactivity, dietary factors, hypertension, dyslipidaemia, diabetes and obesity8 as well as low VO2max. 9 Thus, there are several lifestyle modifications that might contribute to a substantial reduction in the risk of vascular events post-TIA and there is evidence that the earlier these interventions can be introduced, the better the outcome.10 11

The modifiable risk factors for all vascular diseases include smoking, excessive alcohol intake, physical inactivity, dietary factors, hypertension, dyslipidaemia, diabetes and obesity8 as well as low VO2max. 9 Thus, there are several lifestyle modifications that might contribute to a substantial reduction in the risk of vascular events post-TIA and there is evidence that the earlier these interventions can be introduced, the better the outcome.10 11 Pharmacological secondary prevention of stroke National UK guidelines for the pharmacological treatment of TIA/stroke have been established by the National Institute for Health and Care Excellence (NICE)12 and are supplemented by guidelines on tackling individual risk factors. Following the acute diagnosis of TIA, a prophylactic daily dose of 75 mg of aspirin should be initiated and other agents, for example, dipyridamole and clopidogrel, may be added. These agents reduce blood clotting and therefore reduce the chances of a future clot forming within the circulation. Statins should be initiated to lower cholesterol levels13 and appropriate antihypertensive medications are used for blood pressure control14 as per national management guidelines. However, evidence is growing regarding the contribution of change in modifiable risk factors to reductions in deaths15 and there is a need to consider how to promote non-pharmacological measures within secondary prevention.16

nsive medications are used for blood pressure control14 as per national management guidelines. However, evidence is growing regarding the contribution of change in modifiable risk factors to reductions in deaths15 and there is a need to consider how to promote non-pharmacological measures within secondary prevention.16 Non-pharmacological/ lifestyle risk factors Physical activity Physical activity promotion and participation must be one of the key goals for modern-day health systems. Indeed, the WHO in 201017 identified physical inactivity as the fourth leading risk factor for global mortality and this equates to 6% of global deaths. Following a TIA, patients should be encouraged to achieve at least the minimum recommended levels of physical activity as established by the chief medical officers and the departments of health.18 Cardiorespiratory fitness Related to physical activity is cardiorespiratory fitness, which can be measured by VO2max, and is inversely correlated with mortality,19–22 the progression of carotid atherosclerosis23 and the risk of stroke.24 Myers et al25 found that in male subjects with and without cardiovascular disease, peak exercise capacity after adjustment for age was the strongest predictor of the risk of death and each one metabolic equivalent increase in exercise capacity conferred a 12% improvement in survival. Exercise, a form of physical activity, can increase VO2max in sedentary persons26 and in subacute stroke survivors.27

Cardiorespiratory fitness Related to physical activity is cardiorespiratory fitness, which can be measured by VO2max, and is inversely correlated with mortality,19–22 the progression of carotid atherosclerosis23 and the risk of stroke.24 Myers et al25 found that in male subjects with and without cardiovascular disease, peak exercise capacity after adjustment for age was the strongest predictor of the risk of death and each one metabolic equivalent increase in exercise capacity conferred a 12% improvement in survival. Exercise, a form of physical activity, can increase VO2max in sedentary persons26 and in subacute stroke survivors.27 Smoking Smoking is a well-recognised vascular risk factor. The landmark prospective observational study by Doll et al28 found that British male doctors born between 1900 and 1930 who continued to smoke had a life expectancy 10 years less than that of lifelong non-smokers. Smoking as a vascular risk factor has been continually supported by other studies,15 29 30 and patients with TIA should be advised about smoking cessation.31

l et al28 found that British male doctors born between 1900 and 1930 who continued to smoke had a life expectancy 10 years less than that of lifelong non-smokers. Smoking as a vascular risk factor has been continually supported by other studies,15 29 30 and patients with TIA should be advised about smoking cessation.31 Diet With regards to diet, a recent meta-analysis has shown that dietary fibre is inversely correlated with the risk of stroke,32 with fish oils also being protective.33 Indeed the ‘Mediterranean diet’ has shown favourable effects on cardiovascular risk factors.34 Moreover, hypercholesterolaemia, of which dietary intake may be a source, is a modifiable risk factor for cardiovascular and cerebrovascular diseases.35 Cholesterol levels were found to be positively associated with the risk of non-haemorrhagic stroke,36 and dyslipidaemia was also a significant risk factor for ischaemic stroke in the INTERSTROKE study.37 Patients with TIA should, therefore, be advised accordingly about their dietary habits.

cerebrovascular diseases.35 Cholesterol levels were found to be positively associated with the risk of non-haemorrhagic stroke,36 and dyslipidaemia was also a significant risk factor for ischaemic stroke in the INTERSTROKE study.37 Patients with TIA should, therefore, be advised accordingly about their dietary habits. Stress Psychological distress is a well-known risk factor for TIAs. In the observational study by Everson-Rose et al,386749 adults free of vascular disease at baseline in the USA, aged 45–84 years old, were followed up for a median of 8.5 years as part of the Multi-Ethnic Study of Atherosclerosis. The authors found that higher levels of stress and depressive symptoms were associated with increased TIA risk, independent of other known vascular risk factors. Moreover, the diagnosis of TIA often leaves survivors with stress, anxiety and depressive symptoms. Indeed, a recent systematic review39 has highlighted the prevalence of these often forgotten symptoms following a TIA and/or stroke diagnosis. Patients should therefore be educated about the signs and symptoms to be aware of and signposted appropriately for further management, with the general practitioner often as their first contact.

systematic review39 has highlighted the prevalence of these often forgotten symptoms following a TIA and/or stroke diagnosis. Patients should therefore be educated about the signs and symptoms to be aware of and signposted appropriately for further management, with the general practitioner often as their first contact. Alcohol Alcohol excess is a well-known modifiable vascular risk factor, including for TIAs. Gill et al40 report the ‘J-shaped’ association between alcohol and risk of stroke in a case–control study of approximately 1200 patients, that is, low alcohol consumption may have a protective effect for cerebrovascular events, whereas heavy consumption predisposes to TIAs. Safe alcohol consumption levels should therefore be promoted to patients with TIA to reduce the risk of future vascular events.41 Discussion The ‘evidence gap’ from research to practice Despite the knowledge surrounding vascular risk factors and the recognition that TIAs are often the precursors of disabling strokes, more needs to be done in reducing stroke as the leading cause of adult disability.42 Indeed the WHO, as part of their 2013 Global Action Plan For the Prevention and Control of Non-Communicable Diseases, is trying to target a 25% relative risk reduction in overall mortality from cardiovascular diseases, including TIAs.43

ore needs to be done in reducing stroke as the leading cause of adult disability.42 Indeed the WHO, as part of their 2013 Global Action Plan For the Prevention and Control of Non-Communicable Diseases, is trying to target a 25% relative risk reduction in overall mortality from cardiovascular diseases, including TIAs.43 One novel way to tackle vascular risk factors and promote secondary prevention in patients with TIA could be to adapt a cardiac rehabilitation programme for these patients. Indeed cardiovascular and cere­brovascular diseases share common underlying pathological mechanisms and risk factors. Moreover, cardiac rehabilitation after a myocardial infarction results in a statistically significant reduction in reinfarction (OR 0.53), cardiac mortality (OR 0.64) and all-cause mortality (OR 0.74),44 and these findings concur with those of a recent Cochrane review.45 Although Heran et al45 report that the studies included in the review mainly comprise middle-aged men, who are generally at low cardiovascular risk and this should be considered when developing future studies in this area. A Cochrane review46 also demonstrated that hospital- and home-based cardiac rehabilitation programmes result in similar health gains, with home-based programmes improving adherence to the programme47 and promoting longer-term sustainability of health benefits.48

e considered when developing future studies in this area. A Cochrane review46 also demonstrated that hospital- and home-based cardiac rehabilitation programmes result in similar health gains, with home-based programmes improving adherence to the programme47 and promoting longer-term sustainability of health benefits.48 The Stroke Prevention Rehabilitation Intervention Trial of Exercise (SPRITE) is a feasibility and pilot study (ClinicalTrials.gov Identifier: NCT02712385) funded by National Institute for Health Research (NIHR), which is attempting to adapt a home-based cardiac rehabilitation programme for use in the acute period following a TIA. The WHO has defined cardiac rehabilitation as the, ‘sum of activity and interventions required to ensure the best possible physical, mental and social conditions so that patients with chronic or post-acute cardiovascular disease may, by their own efforts, preserve or resume their proper place in society and lead an active life'.49

defined cardiac rehabilitation as the, ‘sum of activity and interventions required to ensure the best possible physical, mental and social conditions so that patients with chronic or post-acute cardiovascular disease may, by their own efforts, preserve or resume their proper place in society and lead an active life'.49 NICE have recommended that the components of cardiac rehabilitation should include exercise, health education and stress management,50 helping to tackle the known vascular risk factors as previously documented. Health education would include addressing the known modifiable vascular risk factors as well as advice regarding work, mental health and sexual activity.50 These components will all be addressed within our adapted home-based cardiac rehabilitation programme, ‘The Healthy Brain Rehabilitation Manual’. Such research has immediate clinical significance and the potential to change guidelines for the management of TIAs, as well as the potential to reduce morbidity and mortality resulting from TIAs, with clear benefit to patients.

ted home-based cardiac rehabilitation programme, ‘The Healthy Brain Rehabilitation Manual’. Such research has immediate clinical significance and the potential to change guidelines for the management of TIAs, as well as the potential to reduce morbidity and mortality resulting from TIAs, with clear benefit to patients. Summary One novel way to tackle vascular risk factors and promote secondary prevention in patients with TIA could be to adapt a cardiac rehabilitation programme for these patients. SPRITE, a feasibility and pilot study (ClinicalTrials.gov Identifier: NCT02712385), is attempting to adapt a home-based cardiac rehabilitation programme, ‘The Healthy Brain Rehabilitation Manual’, for use in the acute period following a TIA. The use of cardiac rehabilitation programmes post-TIA requires further research, particularly within the primary care setting. I acknowledge my study supervisors (Professor Frank Kee, Professor Michael Donnelly, Professor Jonathan Mant and Professor Margaret Cupples) who were helping me to develop the SPRITE study. Contributors: NH conceived the study and drafted and revised the manuscript. Funding: NIHR funded this study. Disclaimer: The funding source had no role in the study design, writing the manuscript, in the decision to submit the manuscript for publication and in data interpretation, collection and analysis. Competing interests: NH is currently undertaking an NIHR clinical fellowship. Provenance and peer review: Not commissioned; externally peer reviewed.

What are the new findings? The YPAQ demonstrates reasonable criterion validity metrics in 12–13 year old Scottish adolescents. Although confirming previous validity coefficients in English adolescents, this is the first YPAQ validity study of Scottish adolescents. Bland Altman plots demonstrate poor agreement and a tendency for the YPAQ to under- report moderate to vigorous physical activity at lower levels of activity and over-report at higher levels of activity. A greater level of measurement error is introduced as activity levels increase. YPAQ should not be used interchangeably with accelerometry, and the employment of this questionnaire in a population setting needs careful consideration.

Bland Altman plots demonstrate poor agreement and a tendency for the YPAQ to under- report moderate to vigorous physical activity at lower levels of activity and over-report at higher levels of activity. A greater level of measurement error is introduced as activity levels increase. YPAQ should not be used interchangeably with accelerometry, and the employment of this questionnaire in a population setting needs careful consideration. Background The development of accurate methods to measure health behaviours forms an integral component in behavioural epidemiology.1 Within physical activity (PA) research, high quality measures are crucial in all stages of the research process, including population surveillance. Accelerometry and movement sensors have become a widely used objective method for quantifying PA levels through their ability to derive information relating to frequency, duration and intensity of PA from actual body movement/acceleration. Although successfully integrated into large-scale studies,2 only a few population level datasets exist using this particular method. Self- or proxy-reported questionnaires remain popular, despite a number of limitations:3 4 questionnaire responses depend on perception, encoding, storage and retrieval of information;5 and concerns exist over the accuracy of questionnaire data from children under 10 years due to their cognitive underdevelopment.6 These concerns translate to poor validity coefficients,7 where a tendency exists for questionnaires to over-report PA levels compared with directly measured PA.8 However, within population surveillance research, a questionnaire approach requires less technical knowledge and expertise and is considered less burdensome than accelerometry. Although cheaper and more practical to administer,9 questionnaires can be adapted more readily to different delivery methods (ie, postal or face-to face administration), have been developed to suit different population groups (eg, children, adolescents),10 and, finally, they have the ability to extract information on ‘type’ (eg, sporting activities, play or active living) and ‘domain’ (eg, school, home, commute) of activity. For these reasons, it is important that they accurately measure the activity being studied. The Youth Physical Activity Questionnaire (YPAQ)11 is based on the Children’s Leisure Activities Study Survey (CLASS);12 and measures frequency, duration, intensity and mode, over the past 7 days, of both PA and sedentary activities throughout all domains.

ortant that they accurately measure the activity being studied. The Youth Physical Activity Questionnaire (YPAQ)11 is based on the Children’s Leisure Activities Study Survey (CLASS);12 and measures frequency, duration, intensity and mode, over the past 7 days, of both PA and sedentary activities throughout all domains. Original validation work was conducted in England with 12–13-year-olds and demonstrated acceptable levels of validity compared with accelerometry (rs=0.42, p=0.04). With reasonable measurement properties, and with the ability to capture the multiple components of PA, it was decided to employ the YPAQ for this study, specifically within a Scottish population. Prior to commencing a large-scale, country-wide data collection (Studying Physical Activity in Children's Environments across Scotland (SPACES)), employing both objective and self-reported measures of PA to estimate the prevalence of children meeting the UK Chief Medical Officer's PA guidelines,13 we set out to examine the ability of the YPAQ to accurately capture the main outcome variable used to assess guideline adherence, namely moderate-to-vigorous physical activity (MVPA). Specifically, we examined the individual level criterion validity and measurement agreement between YPAQ-derived MVPA and accelerometry-derived MVPA (ActiGraph GT3X+; ActiGraph LLC, Pensacola, FL, USA) to assess the suitability of YPAQ to measure this outcome variable for inclusion in the SPACES study.

activity (MVPA). Specifically, we examined the individual level criterion validity and measurement agreement between YPAQ-derived MVPA and accelerometry-derived MVPA (ActiGraph GT3X+; ActiGraph LLC, Pensacola, FL, USA) to assess the suitability of YPAQ to measure this outcome variable for inclusion in the SPACES study. Methods Participants A convenience sample of 90 adolescents (12–13 years old) from two schools in Central/West Scotland were invited to take part. Participants were automatically enrolled (following participant assent) in the study unless parents withdrew consent (opt out consent). Ethical approval for the study was granted by the University of Glasgow’s College of Social Sciences, the participating school’s local educational authorities and the head teachers of both schools. The study fieldwork was conducted in May 2013 and included three school days, two weekend days and 2 days which fell on public holidays. Measures Objective measurement: accelerometer PA was measured using an accelerometer (ActiGraph GT3X+) worn on a belt around the waist for seven consecutive days. The GT3X+ is a small (4.6×3.3×1.5 cm), lightweight (19 g), tri-axial device that records and stores raw acceleration signals in three axes, at a user-specified sample rate (between 30 and 100 Hz). It has a dynamic range of ±6 G and memory capacity of 512MB. ActiGraph devices are used extensively, and internationally, in children’s PA research;2 14 15 the GT3X+ has been validated against indirect calorimetry in children’s energy expenditure research.16

xes, at a user-specified sample rate (between 30 and 100 Hz). It has a dynamic range of ±6 G and memory capacity of 512MB. ActiGraph devices are used extensively, and internationally, in children’s PA research;2 14 15 the GT3X+ has been validated against indirect calorimetry in children’s energy expenditure research.16 Following data collection, ActiGraph data were uploaded to a computer for post-processing using ActiGraph’s proprietary software (ActiLife, v6.7.1). PA files were trimmed to include only the measurement period. The software aggregated the raw acceleration data (100 Hz) into 30-second epochs. Periods of 60 consecutive zeros, allowing for ‘spikes’ of 2 min of activity (less than 100 counts/min), were classified as non-wear and subsequently removed in any PA outcome measure. Participants had to wear the device for 500 min for it to be classified as a valid day,9 and a minimum of three valid days were required for inclusion in the analyses.17 18 MVPA per valid day per participant was extracted using the Evenson threshold (counts per minute >2295) cut points.19 Mean MVPA was calculated per participant (as a function of number of valid days per participant), and then across the full sample. Self-reported questionnaire: YPAQ The YPAQ contains 47 different activities and requests participants to report the frequency and duration of each activity for both weekdays and weekend days over the past 7 days. The YPAQ is broken into contextual settings/domains: sporting, leisure, school and free-time activities.11

Following data collection, ActiGraph data were uploaded to a computer for post-processing using ActiGraph’s proprietary software (ActiLife, v6.7.1). PA files were trimmed to include only the measurement period. The software aggregated the raw acceleration data (100 Hz) into 30-second epochs. Periods of 60 consecutive zeros, allowing for ‘spikes’ of 2 min of activity (less than 100 counts/min), were classified as non-wear and subsequently removed in any PA outcome measure. Participants had to wear the device for 500 min for it to be classified as a valid day,9 and a minimum of three valid days were required for inclusion in the analyses.17 18 MVPA per valid day per participant was extracted using the Evenson threshold (counts per minute >2295) cut points.19 Mean MVPA was calculated per participant (as a function of number of valid days per participant), and then across the full sample. Self-reported questionnaire: YPAQ The YPAQ contains 47 different activities and requests participants to report the frequency and duration of each activity for both weekdays and weekend days over the past 7 days. The YPAQ is broken into contextual settings/domains: sporting, leisure, school and free-time activities.11 On completion of the accelerometer protocol (on day 8), participants attended a large classroom, where trained fieldworkers assisted with the completion of the YPAQ over an allocated school period (55 min). The fieldworkers read the instructions, showed an example of how a question should be filled out and allowed the pupils to ask questions before starting. Upon completion, fieldworkers were instructed to check for errors or omissions (eg, missed questions, illegible/ambiguous answers).

allocated school period (55 min). The fieldworkers read the instructions, showed an example of how a question should be filled out and allowed the pupils to ask questions before starting. Upon completion, fieldworkers were instructed to check for errors or omissions (eg, missed questions, illegible/ambiguous answers). Scoring Each activity in the questionnaire was assigned a metabolic equivalent (MET) value according to previously published values.20 For the purposes of this study, activities with values above 4 METs were considered to be at least moderate and included in the analysis.21 The activities included cricket, dancing, football, gymnastics, martial arts and rugby. Mean time per day in MVPA was calculated per participant (derived from the total MET minutes divided by seven) and then across the group.

th values above 4 METs were considered to be at least moderate and included in the analysis.21 The activities included cricket, dancing, football, gymnastics, martial arts and rugby. Mean time per day in MVPA was calculated per participant (derived from the total MET minutes divided by seven) and then across the group. Statistical analyses The null hypothesis that no bias exists between measurement methods (YPAQ vs accelerometer) was initially tested using a paired t-test. The strength of the association between both measures was tested using Pearson's correlation and Spearman's rank correlation. A Bland-Altman plot,22 showing mean bias and 95% limits of agreement was used to assess the degree of absolute agreement between methods, and differences between measurements were calculated for each participant (YPAQ−accelerometer) and plotted against the mean of each method ((YPAQ+accelerometer)/2). The relationship between these differences (YPAQ−accelerometer) and the mean was tested using a Pearson correlation. This provided an indication of the dependency of the differences on the underlying measurement range. Considering accelerometry to be the criterion method, the values representing the differences (YPAQ−accelerometer) were plotted against the accelerometer (figure 3A). Potential heteroscedasticity across the range of MVPA (accelerometer) was assessed by conducting a Breusch-Pagan/Cook-Weisberg test:23 visually represented by plotting the residuals versus predicted values (figure 3B).

representing the differences (YPAQ−accelerometer) were plotted against the accelerometer (figure 3A). Potential heteroscedasticity across the range of MVPA (accelerometer) was assessed by conducting a Breusch-Pagan/Cook-Weisberg test:23 visually represented by plotting the residuals versus predicted values (figure 3B). All statistical analyses were performed using Stata version 13 (Stata Corp, College Station, TX, USA). Results Of the original 90 participants invited, 7 opted out prior to the study commencing. A further six withdrew their consent during the study, four were absent during data collection and two accelerometers were lost during the monitoring period, leaving 71 participants who took part in the full data collection period. Forty-four participants (61% girls) provided at least three valid days of PA and were included in the agreement analyses. The mean age was 12.7 years. Mean PA levels On average, children spent 58.2±20.3 min per day in MVPA according to accelerometry, with boys spending approximately 1.3 more minutes per day in MVPA than girls. Self-reported time spent in MVPA was much higher than that recorded by accelerometer, with an average time of 99.8±56.2 min per day, with boys reporting on average 29.3 more minutes in MVPA than girls (table 1). Table 1 Baseline characteristics and time spent in MVPA (both methods)

Mean PA levels On average, children spent 58.2±20.3 min per day in MVPA according to accelerometry, with boys spending approximately 1.3 more minutes per day in MVPA than girls. Self-reported time spent in MVPA was much higher than that recorded by accelerometer, with an average time of 99.8±56.2 min per day, with boys reporting on average 29.3 more minutes in MVPA than girls (table 1). Table 1 Baseline characteristics and time spent in MVPA (both methods) MVPA (min) Girls (n=27) Boys (n=17) Overall (n=44) Age (years) 12.7 12.8 12.7 Accelerometer 57.68±22.07 58.96±17.71 58.18±20.28 YPAQ 88.44±55.08 117.76±54.83 99.77±56.23 MVPA, moderate-to-vigorous physical activity; YPAQ, Youth Physical Activity Questionnaire. Validity coefficients Pearson's and Spearman’s correlations between YPAQ and accelerometer were r=0.47 and rs=0.39 (p<0.01), respectively, indicating a statistically significant medium monotonic relationship between the two methods. The mean difference in minutes spent in MVPA between YPAQ and accelerometer was 25.6±50.2 min (95% CI 10.4 to 40.9; p<0.001); the mean difference between methods was more pronounced among boys (table 2). Table 2 Measurement differences in time spent in MVPA: YPAQ−accelerometer Average difference (min) 95% CI p Value* Girls 13.75±45.32 −4.2 to 31.7 0.20 Boys 44.5±53.0 17.3 to 71.8 <0.001 Overall 25.6±50.2 10.4 to 40.9 <0.001 *Wilcoxon signed-ranked test. MVPA, moderate-to-vigorous physical activity; YPAQ, Youth Physical Activity Questionnaire. Agreement between methods

Table 2 Measurement differences in time spent in MVPA: YPAQ−accelerometer Average difference (min) 95% CI p Value* Girls 13.75±45.32 −4.2 to 31.7 0.20 Boys 44.5±53.0 17.3 to 71.8 <0.001 Overall 25.6±50.2 10.4 to 40.9 <0.001 *Wilcoxon signed-ranked test. MVPA, moderate-to-vigorous physical activity; YPAQ, Youth Physical Activity Questionnaire. Agreement between methods Figure 1A demonstrates that data points do not fall on the line of equality (perfect agreement) across levels of measurement. This initial plot illustrates that YPAQ scores tend to be greater than accelerometer-derived MVPA, with a slight trend in the bias: being negative (YPAQ scores lower) for lower levels of accelerometer-derived MVPA and positive for high levels of accelerometer-derived MVPA. Figure 1 (A) Individual participant data points plotting YPAQ-derived MVPA versus accelerometer-derived MVPA. (B) Bland-Altman plot with mean bias and 95% limits of agreement (noted by shaded section and upper and lower dashed lines). MVPA, moderate-to-vigorous physical activity; YPAQ, Youth Physical Activity Questionnaire.

e 1 (A) Individual participant data points plotting YPAQ-derived MVPA versus accelerometer-derived MVPA. (B) Bland-Altman plot with mean bias and 95% limits of agreement (noted by shaded section and upper and lower dashed lines). MVPA, moderate-to-vigorous physical activity; YPAQ, Youth Physical Activity Questionnaire. The Bland-Altman plot (figure 1B) identified a mean bias between the methods of 25.65 min of MVPA, with 95% limits of agreement of −72.69 and +123.99 min (YPAQ−accelerometer). There is evidence of both under- and over-reporting, dependent on the mean level of MVPA. The differences tended to be negative when mean MVPA was low and positive when mean MVPA was high. Pearson's correlation between the difference and the mean was 0.81, indicating a significant positive linear relationship between these two variables. Where there are instances of a relationship of this magnitude, Bland and Altman24 suggest a regression approach for non-uniform differences (figure 2). Using this approach, the limits are slightly narrower at lower levels of MVPA and widen as MVPA increases. Figure 2 Mean bias and 95% limits of agreement using a regression approach for non-uniform differences.

The Bland-Altman plot (figure 1B) identified a mean bias between the methods of 25.65 min of MVPA, with 95% limits of agreement of −72.69 and +123.99 min (YPAQ−accelerometer). There is evidence of both under- and over-reporting, dependent on the mean level of MVPA. The differences tended to be negative when mean MVPA was low and positive when mean MVPA was high. Pearson's correlation between the difference and the mean was 0.81, indicating a significant positive linear relationship between these two variables. Where there are instances of a relationship of this magnitude, Bland and Altman24 suggest a regression approach for non-uniform differences (figure 2). Using this approach, the limits are slightly narrower at lower levels of MVPA and widen as MVPA increases. Figure 2 Mean bias and 95% limits of agreement using a regression approach for non-uniform differences. The Breusch-Pagan/Cook-Weisberg test (figure 3A and B) was conducted to test for constant variance of residuals across predicted values (of differences between measurement methods). This led us to accept the null hypothesis that all error variances were equal (p=0.2899). However, once influential cases and outliers (figure 3B, circled data points) were identified and removed, there was evidence of heteroscedasticity, as shown in figure 3B (the variance increases as the values increase).

methods). This led us to accept the null hypothesis that all error variances were equal (p=0.2899). However, once influential cases and outliers (figure 3B, circled data points) were identified and removed, there was evidence of heteroscedasticity, as shown in figure 3B (the variance increases as the values increase). Figure 3 (A) Scatter plot representing the difference between YPAQ-derived MVPA and accelerometer-derived MVPA plotted against the criterion method (accelerometer); the regression line applied indicates that residuals increase as MVPA increases. (B) Fitted values from figure 3A versus residuals; this plot visually confirms a tendency for greater measurement error at higher levels of MVPA. MVPA, moderate-to-vigorous physical activity; YPAQ, Youth Physical Activity Questionnaire. Conclusions Interpretation of findings The main purpose of the analysis was to investigate the validity metrics of the YPAQ as a self-reported measure for extracting time spent in MVPA in young adolescents as compared with accelerometry. In the event of its acceptability, the measure could be translated to testing in the population setting where its purpose would be to estimate the population prevalence of children meeting the PA guidelines. The results demonstrated that a moderate linear correlation existed between methods (Pearson's r=0.48; Spearman’s rs=0.39), although results from the Bland-Altman analysis demonstrated a poor level of agreement, with error between measures dependent on the underlying PA level (r=0.81).

ildren meeting the PA guidelines. The results demonstrated that a moderate linear correlation existed between methods (Pearson's r=0.48; Spearman’s rs=0.39), although results from the Bland-Altman analysis demonstrated a poor level of agreement, with error between measures dependent on the underlying PA level (r=0.81). We were interested in determining whether the YPAQ would be a valid proxy for accelerometry, given that questionnaires could be considered more practical for population surveillance than activity monitors. When used to assess the prevalence of children meeting PA guidelines, Sallis and Saelens25 have stated the importance of measuring absolute levels of validity. As can be seen from our findings, the agreement between the two methods becomes less evident as MVPA increases (error and overestimation increases), effectively widening the limits of agreement. The YPAQ, although demonstrating acceptable validity through correlational metrics, including the ability to rank individuals' PA, shows systematic bias through the measurement range as demonstrated by the Bland-Altman analyses. We would therefore advise caution if it is used to extract accurate levels of MVPA to be used in population prevalence estimates.

table validity through correlational metrics, including the ability to rank individuals' PA, shows systematic bias through the measurement range as demonstrated by the Bland-Altman analyses. We would therefore advise caution if it is used to extract accurate levels of MVPA to be used in population prevalence estimates. Comparisons with the original validation work The initial validation work undertaken by Corder and colleagues11 was conducted using a population group (12–13 year olds; n=25) similar to that of the present study (12–13 year olds; n=44). Compared with participants in our study, those in the Corder study recorded 14 min/day more in MVPA (72 vs 58 min) as measured by accelerometry; median MVPA by YPAQ in the Corder sample was 92 min/day compared with 100 min/day in our sample. The differences in accelerometry can be explained, to some degree at least, by the particular cut point used in each method (>1952 vs >2295 counts/min), although the use of different ActiGraph models, epoch length from which the MVPA was calculated and processing options, such as non-wear time and total valid time per day, will have also contributed to these differences. YPAQ scores were similar in both studies, indicating some consistency in the measure conducted across different samples and in different years (2005–2006 and 2013–2014). Furthermore, we found a similar relationship between questionnaire and accelerometer when ranking the data (Spearman’s rs of 0.42 in the Corder study vs 0.39 in the current study). This finding provides some support for the ability of the YPAQ to rank PA levels at an individual level.

t years (2005–2006 and 2013–2014). Furthermore, we found a similar relationship between questionnaire and accelerometer when ranking the data (Spearman’s rs of 0.42 in the Corder study vs 0.39 in the current study). This finding provides some support for the ability of the YPAQ to rank PA levels at an individual level. Both studies demonstrated a general over-reporting of MVPA by the YPAQ although a stronger, and significant, bias was found in the present study; 22.4 min MVPA/week (95% CI −155.6 to 200.4) in the Corder sample compared with 25.6 min/day (95% CI 10.4 to 40.9) in the present study. Only our study found that the degree of questionnaire error was dependent on activity level (Pearson correlation of 0.81 vs 0.02), with the complicated pattern observed suggesting under-reporting at lower levels of activity and over-reporting at higher levels. The dependence of error across the measurement range is seldom reported in the literature,7 but it can be seen from the present study that this finding may be considered problematic if it is used in population surveillance studies where guideline prevalence is of key importance.

ity and over-reporting at higher levels. The dependence of error across the measurement range is seldom reported in the literature,7 but it can be seen from the present study that this finding may be considered problematic if it is used in population surveillance studies where guideline prevalence is of key importance. Comparisons with other literature The literature supports the premise that self-reported/indirect measures of PA may over-report activity levels. In a systematic review conducted by Adamo and colleagues,8 it was found that 72% of the reviewed indirect measures overestimated the directly measured values. Within the same review, correlations ranged from −0.56 to 0.89 highlighting both negative and positive relationships between the measures. In contrast to our findings, the review by Adamo and colleagues reported that girls were more likely to overestimate direct values of PA than boys(by 584% vs 114%, girls and boys, respectively). Why we have observed the opposite pattern is unclear, although one potential explanation may be that football and running were more commonly recorded among boys—often with large and extreme values.

that girls were more likely to overestimate direct values of PA than boys(by 584% vs 114%, girls and boys, respectively). Why we have observed the opposite pattern is unclear, although one potential explanation may be that football and running were more commonly recorded among boys—often with large and extreme values. The ability of the YPAQ to successfully rank MVPA is supported by a number of recent reviews on self-reported measures.7 26 However, having the ability to rank PA is different to its ability to be used accurately as a surrogate for PA prevalence. Helmerhorst and colleagues in their recent review suggested that ‘despite considerable effort, accurate and precise self-report physical activity instruments are still scarce’.7 The reduction of a complex multidimensional construct (PA) into a single metric, potentially misleading understandings of what criterion methods are and, importantly, a lack of a comprehensive measurement framework have been cited as potential reasons for the inconsistencies seen in the literature.27 We have to consider the participants themselves when we discuss inconsistency. Feedback from our fieldworkers suggested that many participants struggled with the concept of frequency and duration of activities. Cognitive immaturity, including memory recall, and the comprehension of questionnaire content can be problematic in youth.25 Future work will collect data from 10-year-old children, and will be self-administered rather than interviewer/fieldwork administered. Our experience in this study—with older children—suggests that issues may arise over comprehension of the YPAQ, and consequently affect data quality. A recent review28 assessed 89 PA measures for their applicability to population surveillance and identified a small group of measures that received scientific and expert support. One of these measures, the Physical Activity Questionnaire for Children (PAQ-C)10 may address the memory and comprehension issues faced within this study by assessing general levels of PA rather than trying to ascertain all facets of the behaviour.

a small group of measures that received scientific and expert support. One of these measures, the Physical Activity Questionnaire for Children (PAQ-C)10 may address the memory and comprehension issues faced within this study by assessing general levels of PA rather than trying to ascertain all facets of the behaviour. Strengths and limitations This study endeavoured to replicate the original design conducted by Corder and colleagues,11 including similar participant ages and statistical approach. A strong scientific approach, particularly in measurement studies, is one where previous work is replicated, challenged, supported or refuted. This is especially true when employing different population groups in different settings. By doing so, greater confidence in the YPAQ's validation properties can be expected. As advocated in the literature,29 we have been clear with regard to the measurement purpose, derivation of our outcome and analysis. Furthermore, this study tried to improve data quality by allocating significant time and resources to the process, including the employment of fieldworkers to actively supervise questionnaire completion.

the literature,29 we have been clear with regard to the measurement purpose, derivation of our outcome and analysis. Furthermore, this study tried to improve data quality by allocating significant time and resources to the process, including the employment of fieldworkers to actively supervise questionnaire completion. One limitation of our study is that our sample size lacked the power to detect subgroup differences (eg, by gender), as 52% (n=46) of the participants did not meet the accelerometry inclusion criteria. Additionally, our decision to include children with 3 days of valid PA may not have been sufficient enough to provide an accurate representation of daily PA.30 Moreover, these valid days may not have included a weekend day, which often involve less wear time, and PA levels, than weekdays.31 Therefore, compared with the YPAQ, which asks participants to recall based on the ‘previous 7 days’, we may have introduced error into the analyses and inflated the level of MVPA as measured by accelerometry. Even so, there remained a significant overestimation of the YPAQ against the accelerometer. Furthermore, the accelerometers were worn during the waking hours and removed only for water-based activities and contact activities. As such, some activities (eg, swimming or rugby) may not have been recorded. Additionally, some activities (eg, cycling) may have been misclassified due to the placement of the accelerometer. This information could be included in future studies through the addition of self-reported cycling time; doing so may reduce the size of the overestimation.

, swimming or rugby) may not have been recorded. Additionally, some activities (eg, cycling) may have been misclassified due to the placement of the accelerometer. This information could be included in future studies through the addition of self-reported cycling time; doing so may reduce the size of the overestimation. In summary, although moderately correlated, these two methods should not be used interchangeably as agreement was poor, with error in the measurement highly dependent on activity level. From a practical perspective, the face-to-face administration of the YPAQ highlighted a number of concerns, and its employment in population surveillance (where a face-to face delivery may not be possible) to extract individual level MVPA should be considered carefully. Conversely, if a suitable standardised error was identified and adjusted for, then the YPAQ could be a cheaper, more practical way to measure PA if methods were employed to improve in situ participant comprehension. Contributors: PRWM was responsible for the original idea, with all authors contributing to the analytical approach. AP was the dedicated statistician on the paper, conducting all analyses. PRWM was responsible for drafting the paper and AP assisted with the Methods and Conclusionssections. All authors reviewed, commented and signed off the final draft. Funding: All authors are supported by the UK Medical Research Council Neighbourhoods and Communities Programme (MC_UU_12017/10). Competing interests: None declared.

What are the new findings? Majority of sedentary Kenyans aged above 50 years have metabolic health risks. Moderate-intensity exercise offered in <10 min bouts has similar effects to longer bouts in modifying cardio-metabolic markers in sedentary adults. Regardless of the sex, accumulated short bouts of moderate-intensity exercise that last less than 10 min have beneficial health effects. Introduction The current recommendation for weekly duration of moderate-intensity PA for health benefit in adults is a minimum of 150 min, traditionally achieved in 3–5 sessions of 30–60 min per week, or accumulated through bouts that last more than 10 min.1 In some developed economies, 51%–79% of adults do not meet this weekly PA recommendation.1 2 In Eldoret, Kenya, where the current study was performed, more than 82% of elderly adults do not achieve recommended levels of exercise.3 A lack of time due to other responsibilities partly contributes to lack of adherence to this PA recommendation.1 4–6 To improve adherence, it has been recommended that more experimental studies on the intensity, frequency and duration of exercise be undertaken and that would have a more global appeal on the benefits of exercise1 7–11 and, in particular, studies that lead to more appealing exercise regimes for older adults.

tion.1 4–6 To improve adherence, it has been recommended that more experimental studies on the intensity, frequency and duration of exercise be undertaken and that would have a more global appeal on the benefits of exercise1 7–11 and, in particular, studies that lead to more appealing exercise regimes for older adults. In the 1990s, suggestions were made that short exercise bouts were as effective as single continuous sessions of similar intensities provided that the cumulative weekly exercise time is equal.12 However, evidence for this suggestion remains inconclusive as existing guidelines still focus on either single continuous sessions or bouts that last at the very least 10 min. This has thus failed to mitigate the poor exercise adherence observed among the older populace.13 Studies on the efficacy of even shorter exercise regimes that last less than 10 min remain scarce, leading the WHO to highlight the need for further studies on effectiveness of short-duration exercise.1

very least 10 min. This has thus failed to mitigate the poor exercise adherence observed among the older populace.13 Studies on the efficacy of even shorter exercise regimes that last less than 10 min remain scarce, leading the WHO to highlight the need for further studies on effectiveness of short-duration exercise.1 The cardiometabolic health benefits received from exercise for the elderly is clear, with a strong linkage to improved cardiorespiratory fitness (CRF),14–18 but the specific type of exercise regime that shows higher adherence yet yields similar cardiometabolic benefits needs to be determined. Several studies12 19 20 have reported benefits of accumulated shorter exercise bouts. The lack of longer term follow-up in these studies has raised the question of the long-term benefits21–23 the short bouts may accrue. Therefore, in this study, cardiometabolic markers were compared for 24 weeks between sedentary elderly adults undertaking long-duration exercise bouts lasting ≥30 min and those performing short-duration exercise bouts that lasted <10 min.

as raised the question of the long-term benefits21–23 the short bouts may accrue. Therefore, in this study, cardiometabolic markers were compared for 24 weeks between sedentary elderly adults undertaking long-duration exercise bouts lasting ≥30 min and those performing short-duration exercise bouts that lasted <10 min. Materials and methods Participants Healthy sedentary males (n=27) and females (n=26) aged ≥50 years were recruited from Eldoret, Kenya, following a local print advertisement. Sedentariness was defined as exercise/activity amounting to <600 metabolic equivalent minutes, using the WHO Global Physical Activity Questionnaire (GPAQ). A physical examination to rule out existing health problems was performed on all volunteers on their verbal declaration they were in good health. Individuals with cardiorespiratory disease or other physical ailments/injuries were excluded. Also excluded were participants on β-blockers and participants with diabetes on therapy. Signed informed consent was obtained from all participants.

ormed on all volunteers on their verbal declaration they were in good health. Individuals with cardiorespiratory disease or other physical ailments/injuries were excluded. Also excluded were participants on β-blockers and participants with diabetes on therapy. Signed informed consent was obtained from all participants. Protocol A biodemographics questionnaire on exercise history was completed for each participant following which they were randomly allocated into four groups1: male short-duration bouts of exercise (MS),2 female short-duration bouts of exercise (FS),3 male long-duration bouts of exercise (ML) and female long-duration bouts of exercise (FL). In the short-duration bouts of exercise groups, the participants engaged in three sessions of 5–10 min of moderate-intensity jogging daily. In the long-duration bouts of exercise groups, participants engaged in 30–60 min jogging sessions for 3–5 days each week. Data from the participants were collected and analysed at the recruitment stage (September 2016) and then at 8-week intervals for a period of 24 weeks after the start of the exercise intervention to culminate in May 2017. Participants kept an exercise log, and objective verification of adherence to the exercise regime was done weekly using Polar Wearlink ActITrainer Accelerometers (Actigraph, Pensacola, Florida, USA), and activity monitors were fastened onto participants on select days. The data, metabolic equivalent minutes achieved from the activity monitors, were analysed weekly to ascertain that participants in the different exercise regimes had comparable cumulative exercise time. Since this was only available for select days and used to primarily verify adherence, exercise intensity was monitored using WHO GPAQ. For clinical analysis, 4 mL of venous blood was collected from an average of two participants daily following an overnight 12 hours’ fast, usually between 06:00 and 08:00. Two drops of fleshly drawn blood were immediately used to measure blood glucose using Freestyle Optium Glucometer (Abbott, Oxfordshire, UK). For lipid profiles, centrifugation of blood was done within an hour of collection following which the serum was frozen and kept at −12°C, awaiting to run accumulated weekly samples together.

f fleshly drawn blood were immediately used to measure blood glucose using Freestyle Optium Glucometer (Abbott, Oxfordshire, UK). For lipid profiles, centrifugation of blood was done within an hour of collection following which the serum was frozen and kept at −12°C, awaiting to run accumulated weekly samples together. This analysis into total cholesterols (TCs), high-density lipoproteins (HDLs), low-density lipoproteins (LDLs) and triglycerides (TGs) was done using Cobas Integra 400 Plus (Roche, Germany) at the Academic Model Providing Access to Healthcare Reference Laboratories, a certified clinical laboratory run by Indiana University/Moi University partnership. Analysis Data analysis was done at univariate, bivariate and multivariate levels based on sex of participant and their exercise regime. Paired t-tests were used for intragender differences of metabolic markers following the different exercise regimes. The outcome variables were further analysed at the multivariate level by performing multiple analysis of variance for the dependent variables of lipid profiles and blood glucose and, further, linear regressions controlling for sexgroup performed. For comparison across the four phases of the experiment, repeated measures analysis of variance was conducted for equality of means of these blood chemistry results. Analysis was done using STATA V.13; p value was set at ≤0.05.

files and blood glucose and, further, linear regressions controlling for sexgroup performed. For comparison across the four phases of the experiment, repeated measures analysis of variance was conducted for equality of means of these blood chemistry results. Analysis was done using STATA V.13; p value was set at ≤0.05. Results The mean age of the males (n=27) was 55.5±3.0 years and 53.9±3.0 years for the females (n=26). The proportion of participants who had completed tertiary education was 88.4% among the male and 70.4% among the female. Ninety-one per cent of participants were in white-collar jobs. All (100%) male and female participants in the MS (n=14) and FS (n=13) group adhered to their 24-week exercise regime, compared with 61.5% of the ML (n=14) and 76.9% of FL (n=13). The weekly cumulative exercise minutes was similar for MS and ML (161.8±7.2 vs 162.56±6.1, respectively) and also for FS and FL (158.3±3.6 vs 156.05±2.7, respectively). Specific baseline characteristics for participants from the four groups are presented in table 1. Table 1 Baseline characteristics of participants in the four groups

Results The mean age of the males (n=27) was 55.5±3.0 years and 53.9±3.0 years for the females (n=26). The proportion of participants who had completed tertiary education was 88.4% among the male and 70.4% among the female. Ninety-one per cent of participants were in white-collar jobs. All (100%) male and female participants in the MS (n=14) and FS (n=13) group adhered to their 24-week exercise regime, compared with 61.5% of the ML (n=14) and 76.9% of FL (n=13). The weekly cumulative exercise minutes was similar for MS and ML (161.8±7.2 vs 162.56±6.1, respectively) and also for FS and FL (158.3±3.6 vs 156.05±2.7, respectively). Specific baseline characteristics for participants from the four groups are presented in table 1. Table 1 Baseline characteristics of participants in the four groups MS ML FS FL Age (years) 55.0±5.6 55.2±3.0 53.9±2.6 53.9±3.5 BMI (kg/m2) 25.8±4.0 28.6±4.8 33.3±4.8 32.0±5.4 Waist/height ratio 0.52±0.07 0.56±0.08 0.61±0.05 0.57±0.08 Waist/hip ratio 0.93±0.06 0.96±0.07 0.82±0.10 0.84±0.09 Fat % 20.7±7.3 24.9±7.9 39.8±3.6 37.4±5.1 Resting systolic pressure (mm Hg) 138.9±17.4 140.8±22.1 133.7±13.2 137.6±23.8 Resting diastolic pressure (mm Hg) 82.1±11.3 83.7±8.7 83.7±10.6 84.2±8.1 Resting heart rate (b/m) 73.9±9.5 76.8±7.7 79.8±12.0 71.8±8.4 Data presented as mean±SD. BMI (kg/m2), basal metabolic index in kilograms per meter squared; b/m, beats per minute; FL, long bouts female; FS, short bouts female; ML, long bouts male; MS, short bouts male.

MS ML FS FL Age (years) 55.0±5.6 55.2±3.0 53.9±2.6 53.9±3.5 BMI (kg/m2) 25.8±4.0 28.6±4.8 33.3±4.8 32.0±5.4 Waist/height ratio 0.52±0.07 0.56±0.08 0.61±0.05 0.57±0.08 Waist/hip ratio 0.93±0.06 0.96±0.07 0.82±0.10 0.84±0.09 Fat % 20.7±7.3 24.9±7.9 39.8±3.6 37.4±5.1 Resting systolic pressure (mm Hg) 138.9±17.4 140.8±22.1 133.7±13.2 137.6±23.8 Resting diastolic pressure (mm Hg) 82.1±11.3 83.7±8.7 83.7±10.6 84.2±8.1 Resting heart rate (b/m) 73.9±9.5 76.8±7.7 79.8±12.0 71.8±8.4 Data presented as mean±SD. BMI (kg/m2), basal metabolic index in kilograms per meter squared; b/m, beats per minute; FL, long bouts female; FS, short bouts female; ML, long bouts male; MS, short bouts male. Mean fasting blood glucose (FBG) of all participants at the start of the study was 5.92±1.4 mmol/L (males) and 6.23±0.74 mmol/L (females). The percentage of males and females with FBG >5.5 mmol/L was 40.7% (MS group), 33.3% (ML group), 46.2% (FS group) and 42.3% (FL group). In 22.2% (MS), 25.9% (ML), 30.8% (FS) and 11.5% (FL), baseline TC was above the cut-off of >5.2 mmol/L for males and >5.3 mmol/L for females. HDLs below 0.9 mmol/L was found in 18.5% (MS), 11.1% (ML), 7.7% (FS) and 15.4% (FL). TC/HDL ratio of >5.0 mmol/L and>4.5 mmol/L, the cut-offs for males and females, respectively, was found in 44.4% of the males (22.2% each in MS (mean=6.58±1.21) and ML (mean=6.63±1.29)) and 38.5% of the females (19.2% each in FS (mean=5.38±0.92) and FL (mean=5.42±0.59)). LDL/HDL ratio >3.5 was found in 14.8% MS group (mean=4.07±0.53) and in 18.5% ML group (mean=4.79±1). LDL/HDL ratio >3.0 was found in 23.1% FS group (mean=3.53±0.27) and 11.1% FL group (mean=3.32±0.43).

8±1.21) and ML (mean=6.63±1.29)) and 38.5% of the females (19.2% each in FS (mean=5.38±0.92) and FL (mean=5.42±0.59)). LDL/HDL ratio >3.5 was found in 14.8% MS group (mean=4.07±0.53) and in 18.5% ML group (mean=4.79±1). LDL/HDL ratio >3.0 was found in 23.1% FS group (mean=3.53±0.27) and 11.1% FL group (mean=3.32±0.43). At the end of the 24-week period, male participants with FBG of ≥5.5 mmol/L in both the MS and ML groups decreased by 29.6%. The percentage of female participants in FS group with FBG of ≥5.5 mmol/L dropped from 46.2% to 0, while in FL group, it decreased by 30.8%. Mean glucose values for all the groups also dropped, but there was no significant gender difference between different exercise regimes (figure 1). All MS and FS who had baseline HDL of <0.9 mmol/L reached values of ≥0.9 mmol/L, the recommended lower cut-off. There was no percentage change in the subjects in ML group with HDL <0.9 mmol/L. Seventy-five per cent of FL group reached HDL ≥0.9 mmol/L. For TC in the MS group, participants with >5.2 mmol/L fell from 22.2% to 14.8%, with mean value decreasing from 6.4±1.38 mmol/L to 5.48±0.41 mmol/L. In the ML group, the reduction was from 25.9% to 3.7%, with mean TC decreasing from 6.09±0.58 mmol/L to 5.76 mmol/L. Similarly, in FS group, participants with TC >5.3 mmol/L decreased from 30.8% to 11.5% with the mean TC decreasing from 6.75±1.46 to 5.72±0.15 mmol/L. In FL group, there was no change in number of participants with TC >5.3 mmol/L, although their mean values decreased from 7.78±1.79 mmol/L to 5.76±0.44 mmol/L. For the TC/HDL ratio, in the MS group participants with ratios >5.0 decreased from 22.2% to 7.4% (mean decreased from 6.58±1.21 to 5.59±0.36), and in FS group, those with the ratio >4.5 decreased from 19.2% to 7.7% (mean decreased from 5.38±0.92 to 5.14±0.03). In the ML group, the decrease was from 22.2% to 15.4% (mean change from 6.63±1.29 to 5.66±0.5), and in FL group, the decrease was from 19.2% to 3.8% (mean change from 5.42±0.59 to 5.18±0). Percentage of MS group with LDL/HDL >3.5 mmol/L was halved to 7.4% with mean LDL/HDL decreasing to 3.61±0.14. In the ML group, the decrease was from 18.5% to 11.1% with mean LDL/HDL decreasing to 4.65±0.67 mmol/L. In FS group, participants with LDL/HDL ratio >3.0 decreased from 23.1% to 15.4% (mean 3.49±0.55) and to 3.8% from 11.1% (mean 3.49±0) in the FL group. Mean changes for the lipid ratios are provided in figure 2.

0.14. In the ML group, the decrease was from 18.5% to 11.1% with mean LDL/HDL decreasing to 4.65±0.67 mmol/L. In FS group, participants with LDL/HDL ratio >3.0 decreased from 23.1% to 15.4% (mean 3.49±0.55) and to 3.8% from 11.1% (mean 3.49±0) in the FL group. Mean changes for the lipid ratios are provided in figure 2. Figure 1 Mean change in fasting blood glucose (FBG) between week 0 and week 24. For each exercise regime, FB dropped over the 24-week period in both sexes. FBG, fasting blood glucose; FL, long bouts female; FS, short bouts female; ML, long bouts male; MS, short bouts male. Figure 2 Mean change in lipid ratios between week 0 and week 24. No differences were observed in the change in lipid ratios in 24 weeks for the different exercise regimes in each either sex. HDL, high-density lipoprotein; LDL, low-density lipoprotein; TC, total cholesterol; TG, triglycerides. In summary, except for TC in females where the short bouts appeared superior to the long bouts, there was no difference in actual value change between the two bouts for the various cardiometabolic parameters over the 24-week period as shown in table 2. Table 3 is linear regression outcomes controlling for gender in the difference of the means above. Table 2 Mean cardiometabolic values over 24 weeks

In summary, except for TC in females where the short bouts appeared superior to the long bouts, there was no difference in actual value change between the two bouts for the various cardiometabolic parameters over the 24-week period as shown in table 2. Table 3 is linear regression outcomes controlling for gender in the difference of the means above. Table 2 Mean cardiometabolic values over 24 weeks Variable Group Baseline Week 8 Week 16 Week 24 Mean Δ(week 24 to week 0) P values Male TC Short 5.4±1.3 4.8±1.2 4.8±0.9 4.7±0.8 −0.68±1.3 0.99 Long 5.0±0.8 4.7±1.2 4.8±1.0 4.3±0.7 −0.68±0.5 HDL Short 1.3±0.5 1.3±0.4 1.5±0.4 1.6±0.4 0.30±0.4 0.55 Long 1.0±0.3 1.0±0.3 1.1±0.4 1.2±0.5 0.21±0.3 LDL Short 3.1±0.9 3.1±0.8 2.9±0.9 3.1±0.7 0.04±0.7 0.81 Long 2.8±1.2 3.2±1.5 2.9±1.2 2.9±1.3 0.13±0.9 TG Short 1.8±1.3 1.6±0.9 1.7±0.8 1.6±0.5 −0.23±1.2 0.42 Long 1.9±0.9 2.4±1.8 2.3±0.9 2.1±1.1 0.16±0.8 TC/HDL Short 4.8±1.9 4.0±1.5 3.6±1.6 3.2±1.3 −1.52±1.4 0.65 Long 5.4±1.9 5.1±1.9 5.0±1.7 4.1±1.8 −1.24±1.3 LDL/HDL Short 2.8±1.2 2.6±1.0 2.3±1.3 2.2±0.9 −0.59±0.9 0.30 Long 3.0±1.5 3.2±1.8 2.9±1.4 2.9±1.7 −0.16±1.0 TG/HDL Short 1.8±2.0 1.5±1.1 1.4±1.2 1.1±0.6 −0.75±1.6 0.26 Long 2.2±1.6 3.1±3.7 2.6±2.0 2.2±1.9 −0.04±0.9 Female TC Short 5.8±1.7 4.9±0.5 5.0±0.8 4.9±0.5 −0.91±1.4 0.01 Long 4.0±1.0 4.8±0.8 4.7±0.6 4.5±1.2 −0.51±0.9 HDL Short 1.3±0.3 1.2±0.2 1.4±0.3 1.4±0.3 0.12±0.4 0.07 Long 1.0±0.3 1.2±0.2 1.3±0.2 1.4±0.3 0.41±0.3 LDL Short 3.6±0.9 3.5±0.7 3.1±0.9 3.5±0.8 −0.03±0.6 0.11 Long 2.6±0.9 3.4±0.7 2.8±1.0 3.1±1.0 0.45±0.8 TG Short 1.3±0.5 1.2±0.4 1.1±0.4 1.3±0.6 −0.04±0.4 0.21 Long 1.0±0.3 1.2±0.4 1.1±0.3 1.1±0.3 0.19±0.5 TC/HDL Short 4.5±1.0 4.1±0.8 3.7±0.8 3.6±0.9 −0.88±0.8 1 Long 4.2±0.7 4.0±0.7 3.7±0.6 3.3±0.9 −0.88±0.9 LDL/HDL Short 2.8±0.8 2.9±0.8 2.3±0.7 2.5±0.8 −0.23±0.7 0.45 Long 2.7±0.5 2.8±0.7 2.2±0.8 2.2±0.7 −0.43±0.6 TG/HDL Short 1.1±0.5 1.0±0.4 0.8±0.3 0.9±0.5 −0.13±0.2 0.81 Long 1.2±0.6 1.0±0.4 0.9±0.4 0.9±0.6 −0.18±0.7 Values are means±SD in millimoles per litre of blood. P values represent if the difference in mean change between bouts is statistically significant.

8 −0.23±0.7 0.45 Long 2.7±0.5 2.8±0.7 2.2±0.8 2.2±0.7 −0.43±0.6 TG/HDL Short 1.1±0.5 1.0±0.4 0.8±0.3 0.9±0.5 −0.13±0.2 0.81 Long 1.2±0.6 1.0±0.4 0.9±0.4 0.9±0.6 −0.18±0.7 Values are means±SD in millimoles per litre of blood. P values represent if the difference in mean change between bouts is statistically significant. Δ, change; HDL, high-density lipoproteins; LDL, low-density lipoproteins; TC, total cholesterol; TG, triglycerides. Table 3 Linear regressions for cardiometabolic variables (controlling for sex) Mean change, long Coefficient SE P>|t| 95% Cl FBG −0.36 0.25 0.15 −0.87 to 0.14 TC 0.75 0.36 0.05 0.15 to 1.48 HDL 0.11 0.11 0.33 −0.11 to 0.33 LDL 0.29 0.23 0.20 −0.17 to 0.75 TG 0.30 0.24 0.22 −0.18 to 0.78 TC/HDL 0.13 0.34 0.70 −0.56 to 0.82 LDL/HDL 0.10 0.24 0.68 −0.38 to 0.58 TG/HDL 0.31 0.32 0.33 −0.33 to 0.95 FBG, fasting blood glucose; HDL, high-density lipoproteins; LDL, low-density lipoproteins; TC, total cholesterol; TG, triglycerides.

−0.11 to 0.33 LDL 0.29 0.23 0.20 −0.17 to 0.75 TG 0.30 0.24 0.22 −0.18 to 0.78 TC/HDL 0.13 0.34 0.70 −0.56 to 0.82 LDL/HDL 0.10 0.24 0.68 −0.38 to 0.58 TG/HDL 0.31 0.32 0.33 −0.33 to 0.95 FBG, fasting blood glucose; HDL, high-density lipoproteins; LDL, low-density lipoproteins; TC, total cholesterol; TG, triglycerides. Discussion Baseline outcomes Lipid profiling is crucial in CRF assessment among healthy individuals and those with metabolic syndrome. At the start of the study, nearly half of the males and females had TC above their reference cut-offs (>5.2 and >5.3 mmol/L, respectively). A third of the males and females had unfavourable levels of both LDL and HDL. Additionally, one-fifth and one-tenth of males and females had high TG levels. For TC/HDL associated with poor CRF and cardiovascular disease,24–26 about half from both sexes had values higher than cut-off value. Similarly, a third of males and females had unfavourably high LDL/HDL, and about one-tenth had TG/HDL consistent with cardiometabolic risks. Current cut-offs used for TG/HDL25 are from non-black populations, and these may be different in other populations. Combination of various lipids and their ratios showed in the study participants a large proportion being at substantial metabolic risk at the start of the study. For FBG, participants had prediabetic to diabetic mean similar to those found in studies on other African populations of comparable ages,27 28 and mean values for females were higher than their age-matched males. Results from urban Nigerians and rural Kenyans29 30 differ from the baseline measurement found in this study, probably because we studied a relatively older population. Thus, a majority of this urban population had metabolic health risks at recruitment, a finding similar to that recently found among rural Kenyans.30

s. Results from urban Nigerians and rural Kenyans29 30 differ from the baseline measurement found in this study, probably because we studied a relatively older population. Thus, a majority of this urban population had metabolic health risks at recruitment, a finding similar to that recently found among rural Kenyans.30 Absolute lipid profile changes After 24 weeks of prescribed moderate-intensity exercise, both males and females showed improved metabolic profiles. Nearly half of the MS group with unfavourable TC at the start reached values associated with better CRF. Furthermore, mean TC in these males decreased as did the mean for those who still did not reach the recommended ranges. The finding was similar among ML group, suggesting the two exercise regimes produced comparable affect. This ML group also had a reduced percentage that remained with values above recommended cut-off and a decline in mean values for the whole group and, importantly, those who did not attain TC values <5.2 mmol/L as well. The two exercise regimes produced similar changes in the mean TC at the end of the study period. In females, except for the decrease in mean TC being higher in FS group, which suggests short-bout exercise could have better outcome compared with the currently recommended exercise regime, no other differences were observed in TC. Also, two-thirds of females with high baseline TC in FS group reached the recommended range at the end of the study; none from the FL group did. The two exercise regimes, however, had comparable decrease in the overall TC means, even among participants who were unable to achieve the reference cut-off. When baseline and end-point values were compared, mean TC change was similar in the two regimes. Thus, based on TC alone, no demonstrable difference could be shown between the exercise regimes for either sex. No study that we are aware of has described this exercise effect in individuals of comparable age and setting and followed for similar time-length as our study.

mean TC change was similar in the two regimes. Thus, based on TC alone, no demonstrable difference could be shown between the exercise regimes for either sex. No study that we are aware of has described this exercise effect in individuals of comparable age and setting and followed for similar time-length as our study. Males and females in both exercise regimes had an increase in HDL. This finding is consistent with other studies that show exercise affects HDL.31 32 Specifically, males and females in short-bout exercise regime whose baseline HDL was <0.9 mmol/L (18.5% and 7.7%, respectively) decreased to 0. In the FL group, those with HDL <0.9 mmol/L, also decreased from 15.4% to 3.8%, and there was no change in the ML group. What our study adds and that could be a beneficial interventional approach is that the short sessions appeared more beneficial in improving HDL levels when compared with the traditional regimes. For LDL and TG, the change was marginal. Males and females on long-bout exercise regime had slight rise in both LDL and TG, but there was slight improvement in the short-bout exercise participants.

ch is that the short sessions appeared more beneficial in improving HDL levels when compared with the traditional regimes. For LDL and TG, the change was marginal. Males and females on long-bout exercise regime had slight rise in both LDL and TG, but there was slight improvement in the short-bout exercise participants. Effect on lipid ratios For lipid ratios, two-thirds of MS and FS group showed reduced TC/HDL (<5.0 and <4.5, respectively). In ML, it was a quarter, and in FL, it was four-fifths. Previously, it has been suggested that intermittent exercise regimes may actually be more beneficial than current exercise regimes in regulation of attributes such as blood pressure and maximal oxygen consumption.33 What our study adds is that short exercise regimes are also beneficial in improving TC/HDL among males. Our results, however, show mixed outcomes since for the females, long-bout exercise regime was marginally superior in improving the TC/HDL ratio. This improvement in females differs from Quinn et al,33 who found that shorter exercise regimes provide better outcomes, although their study excluded a comparison of TC/HDL. Furtherore, our study period was twice as long (24 weeks vs 12 weeks), and each of our exercise sessions took half the number of minutes (7.5 min vs 15 min) adopted by Quinn et al,33 which could have contributed to the difference in our findings. Furthermore, since the long-bout exercise regime was less effective compared with the short-bout exercise regime in lowering TC in females, the difference in results could be due to the long-bout exercise raising HDL marginally higher thereby reducing the TC/HDL ratio. The decrease in mean TC/HDL in males and females in the two exercise regimes who did not achieve recommended levels was similar. With regard to TC/HDL ratio, the effect of the two exercise regimes could not be differentiated, and the apparent dissimilarity in the effect among females was not supported when controlling for sex. Thus, the effect on TC/HDL by the two exercise regimes was similar in both males and female.

mmended levels was similar. With regard to TC/HDL ratio, the effect of the two exercise regimes could not be differentiated, and the apparent dissimilarity in the effect among females was not supported when controlling for sex. Thus, the effect on TC/HDL by the two exercise regimes was similar in both males and female. LDL/HDL had lower change in rate throughout the study period. However, the overall trend was similar in both sexes from the two exercise regimes. Half of MS group with baseline values of >3.5 showed lower ratios. Slightly less than half of ML regime participants also showed LDL/HDL ≤3.5. There was no difference between the groups in overall mean change of LDL/HDL for the duration of the study. The exercises regimes also had comparable mean values in those not reaching the recommended LDL/HDL ranges. Fifty per cent of FS group that started with high values of LDL/HDL reached the recommended threshold, compared with about two-thirds in the FL group. Thus, shorter exercise regime was more favourable in LDL/HDL regulation in males, although females still attained satisfactory results since drop in LDL/HDL for traditional group was not statistically superior to that in experimental group. Based on TC/HDL and LDL/HDL, it is apparent the two exercise regimes had similar reduction of cardiovascular disease risk. This could also address the recent finding, which our baseline data also support, that majority of Kenyans aged above 50 years are at cardiovascular disease risk based on combination of various lipid ratios.30 Regardless of gender, all would thus benefit similarly, whether from short or long sessions of exercise. Overall, comparison of these ratios suggest that no difference exists in reduction of metabolic syndrome risk in individuals in either of these exercise regimes.

risk based on combination of various lipid ratios.30 Regardless of gender, all would thus benefit similarly, whether from short or long sessions of exercise. Overall, comparison of these ratios suggest that no difference exists in reduction of metabolic syndrome risk in individuals in either of these exercise regimes. Blood glucose effects FBG levels decreased for all groups. In MS group, participants with FBG >5.5 mmol/L decreased from 40.7% to 11.1%, while in the ML from 33.3% to 3.7. Absolute FBG mean change, however, was similar between the groups, and participants retaining prediabetic-to-diabetic values were less in MS and FS than in the ML and FL groups. In the FS group, all participants with prediabetic-to-diabetic baseline FBG levels (46.2%) achieved normal levels after the 24 weeks. In the FL, this was from 42.3% to 11.5%. The mean change in absolute FBG values between baseline and end-point were similar for FS and FL groups. Regression analysis controlling for gender showed no significant difference between the short-duration and long-duration bouts of exercise in FBG mean change. The view that longer exercise sessions among the males and the shorter regime for the females were superior based on percentage change (alone) therefore failed to hold further. This, coupled with the observation of no difference in mean change for FBG after 24 weeks for both sexes from the two different exercise regimes, supports our hypothesis that these exercise regimes have similar effects. Thus, shorter/intermittent exercise sessions lasting <10 min each regulate elevated FBG to the same extent as the longer traditional sessions lasting >30 min per session, as long as the cumulative exercise times are similar.

ent exercise regimes, supports our hypothesis that these exercise regimes have similar effects. Thus, shorter/intermittent exercise sessions lasting <10 min each regulate elevated FBG to the same extent as the longer traditional sessions lasting >30 min per session, as long as the cumulative exercise times are similar. Limitations The design of the protocol may have caused some limitations. Our sample was recruited through print advertisement, which self-selection may have favoured those able to read, compromising inference to the general elderly population. Furthermore, the design could not allow for blinding, depended on subjective health assessment at recruitment with only a physical examination the only form of objective verification, and did not consider lifestyle behaviours such as smoking history and diet, which could have affected health status of participants. These potential limitations may have confounded our results and affected generalisation. Conclusion An exercise regime of accumulated short bouts lasting <10 min shows improvement of cardiometabolic measurements among sedentary adults aged ≥50 years comparable with the currently advocated longer sessions that last >30 min each, as long as the intensity and cumulative exercise times are similar. Contributors: KM helped in designing protocol, data collection, analysis and writing. Both KT and NBP helped with designing of protocol and writing the manuscript.

Conclusion An exercise regime of accumulated short bouts lasting <10 min shows improvement of cardiometabolic measurements among sedentary adults aged ≥50 years comparable with the currently advocated longer sessions that last >30 min each, as long as the intensity and cumulative exercise times are similar. Contributors: KM helped in designing protocol, data collection, analysis and writing. Both KT and NBP helped with designing of protocol and writing the manuscript. Funding: This research was supported by the Consortium for Advanced Research Training in Africa (CARTA). CARTA is jointly led by the African Population and Health Research Center (APHRC) and the University of the Witwatersrand and funded by the Wellcome Trust (UK) (grant no: 087547/Z/08/Z), the Department for International Development (DfID) under the Development Partnerships in Higher Education (DelPHE), the Carnegie Corporation of New York (grant no: B 8606), the Ford Foundation (grant no: 1100–0399), Google.Org (grant no: 191994), Sida (grant no: 54100029), MacArthur Foundation (grant no:10-95915-000-INP) and British Council. Competing interests: None declared. Patient consent: Obtained. Ethics approval: The study was approved by the joint Moi Teaching and Referral Hospital and Moi University Research Ethics Committee in Eldoret, Kenya. Ethical approval number: MTRH-MU IREC 0001242. Provenance and peer review: Not commissioned; externally peer reviewed.

What is already known? Abdominal and thoracic cancers cause debilitating illness, and surgery is associated with significant decline in physical function. Exercise initiated after completion of active cancer treatment has a beneficial effect on health-related quality of life. What are the new findings? There is insufficient evidence that preoperative or postoperative resistance muscle-strengthening exercise improves or negatively affects functional outcomes in patients undergoing abdominal surgery for cancer. Large-scale, well-designed clinical trials are required to determine whether resistance muscle-strengthening exercise is beneficial for patients undergoing abdominal surgery for cancer.

What are the new findings? There is insufficient evidence that preoperative or postoperative resistance muscle-strengthening exercise improves or negatively affects functional outcomes in patients undergoing abdominal surgery for cancer. Large-scale, well-designed clinical trials are required to determine whether resistance muscle-strengthening exercise is beneficial for patients undergoing abdominal surgery for cancer. Introduction Background Abdominal and thoracic cancers affect about 12 000 people annually in the UK. Many of these patients will undergo surgery, after which there is a high risk of postoperative complications and significant decline in physical function. A systematic review of exercise for people with cancer by Stevinson et al1 found some evidence that those who exercised had better physical function compared with those who did not exercise, but there was insufficient evidence to demonstrate improvement in quality of life. In addition, they were not able to determine which type of exercise intervention was best or if any had long-term benefit. A more recent Cochrane review of exercise for people with cancer by Mishra et al2 found that exercise initiated after completion of active cancer treatment (ie, surgery, chemotherapy, radiation therapy or hormone therapy) has a beneficial effect on health-related quality of life, although no parallel improvement in self-reported physical function was found. The exercise interventions included in this review varied greatly and included strength training, yoga, walking, cycling, tai chi and qi gong. However, due to the small number of studies available, these authors were not able to evaluate the effect of different modes and intensities of exercise. Furthermore, studies of exercise in the preoperative and early postoperative stages were not included in the review. Therefore, it is not known whether exercise, when commenced before the end of active cancer treatment, would have additional benefit on physical function for those undergoing surgery.

sities of exercise. Furthermore, studies of exercise in the preoperative and early postoperative stages were not included in the review. Therefore, it is not known whether exercise, when commenced before the end of active cancer treatment, would have additional benefit on physical function for those undergoing surgery. While there is growing evidence on the beneficial effects of aerobic exercise, resistance exercise training has received much less attention.3–6 It is thought that resistance exercise training could act to aid recovery of muscle function.7 It has long been established that resistance exercise training is effective in stimulating muscle anabolic processes and increasing muscle strength.8 It may even counteract some of the metabolic pathophysiology associated with cachexia.9 Furthermore, it can be performed with very little equipment and space and while patients are bed-bound in hospital or at home. Although there have been previous systematic reviews of the effects of exercise training, there have not been any that have specifically focused on resistance training.

sociated with cachexia.9 Furthermore, it can be performed with very little equipment and space and while patients are bed-bound in hospital or at home. Although there have been previous systematic reviews of the effects of exercise training, there have not been any that have specifically focused on resistance training. Previous reviews, relating to exercise training for patients with cancer, have mostly focused on specific outcomes such as fatigue and quality of life,4 5 and most have centred on specific types of cancer.10–17 Galvão and Newton18 published a review of exercise intervention studies for all cancers and a meta-analysis of exercise training interventions. However, their review included a heterogeneous group of studies including some that were not randomised or had no control group. Quality systematic reviews require critical appraisal of the quality of the reviewed studies and share accurate descriptions of the design, delivery and interpretation of what was done in the study. In some instances detailed description of these aspects is not available.19

were not randomised or had no control group. Quality systematic reviews require critical appraisal of the quality of the reviewed studies and share accurate descriptions of the design, delivery and interpretation of what was done in the study. In some instances detailed description of these aspects is not available.19 One of the main challenges in studying the effects of a resistance exercise programme on physical function in cancer surgery patients is in identifying an appropriate outcome measure. The review by Mishra and colleagues found no significant improvement in physical function as evaluated using self-report questionnaires, but they did not measure any index of physical performance.2 Therefore, our aim was to undertake a systematic review of the literature on interventional studies investigating the effects of preoperative and postoperative resistance exercise training on recovery of physical function in patients undergoing abdominal surgery for cancer. The findings will provide clinicians and investigators a basis to choose exercise interventions for use in clinical practice or for future research. Methods The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines on systematic reviews were followed for this review.20 Figure 1 summarises the review process. Figure 1 Flow chart for systematic review of studies.

One of the main challenges in studying the effects of a resistance exercise programme on physical function in cancer surgery patients is in identifying an appropriate outcome measure. The review by Mishra and colleagues found no significant improvement in physical function as evaluated using self-report questionnaires, but they did not measure any index of physical performance.2 Therefore, our aim was to undertake a systematic review of the literature on interventional studies investigating the effects of preoperative and postoperative resistance exercise training on recovery of physical function in patients undergoing abdominal surgery for cancer. The findings will provide clinicians and investigators a basis to choose exercise interventions for use in clinical practice or for future research. Methods The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines on systematic reviews were followed for this review.20 Figure 1 summarises the review process. Figure 1 Flow chart for systematic review of studies. Search strategy The Cochrane Library, EBSCO (SPORTDiscus and Cumulative Index to Nursing and Allied Health Literature (CINAHL)), PLOS, PubMed (Medline) and Elsevier (Scopus) electronic databases were searched up to and including December 2014. The search strategy used was exercise OR training OR isometric OR static OR isotonic OR concentric OR eccentric OR resistance OR strengthening exercise OR exercise therapy OR circuit training OR rehabilitation OR physiotherapy; AND neoplasm OR abdominal cancer OR stomach cancer OR gastric cancer OR bowel cancer OR pancreatic cancer OR colorectal cancer OR colon cancer OR rectal cancer OR gastrointestinal cancer OR ovarian cancer OR endometrial cancer OR cervical cancer OR renal cancer OR kidney cancer OR bladder cancer OR uterine cancer OR gynaecological cancer OR urological cancer; AND abdominal surgery OR laparotomy OR laparoscopy OR laparoscopic OR anterior resection OR colectomy OR hemicolectomy; AND clinical trial OR random controlled trial OR quasi-randomised controlled trial OR controlled trial OR comparative trial.

der cancer OR uterine cancer OR gynaecological cancer OR urological cancer; AND abdominal surgery OR laparotomy OR laparoscopy OR laparoscopic OR anterior resection OR colectomy OR hemicolectomy; AND clinical trial OR random controlled trial OR quasi-randomised controlled trial OR controlled trial OR comparative trial. All titles and abstracts generated by the search were independently screened for inclusion by three authors (DS, FH and KC). Disagreement between authors was discussed and consensus was reached. The search was restricted to English language and were included if the following criteria were met: (1) randomised, quasi-randomised or controlled trial study design comparing a muscle-strengthening exercise intervention (ie, exercise using resistance to induce muscular contraction) ± other therapy with a comparative group; (2) included adult participants (≥18 years) who underwent abdominal surgery (ie, surgery pertaining to the contents of the abdominal cavity, its walls and orifices) for cancer; and (3) included muscle strength, physical function, self-reported functional ability, range of motion and/or performance-based test as an outcome measure.

adult participants (≥18 years) who underwent abdominal surgery (ie, surgery pertaining to the contents of the abdominal cavity, its walls and orifices) for cancer; and (3) included muscle strength, physical function, self-reported functional ability, range of motion and/or performance-based test as an outcome measure. Data extraction Participants’ age, gender, diagnosis, surgical procedure and sample size were extracted from the included studies, along with a description of the exercise intervention, including muscle group or groups exercised, contraction effort, number of repetitions and frequency, length of programme, length of follow-up, group or individual exercise programme, home or supervised exercise programme, and timing of programme (presurgery and/or postsurgery). Data synthesis and analysis The aim of this review was to evaluate the effect of resistance muscle strengthening on physical function in people undergoing abdominal surgery for cancer. For each study, means and SD of outcomes focused on physical function were extracted. Outcomes relating directly to surgery, length of stay, infection and other postsurgical complications were not considered in this review. Assessment was made of the outcome measures for physical function that were used in different studies, before progression to pooling of data for analysis of the most common outcome measure. Treatment effect of individual studies is reported as mean difference and 95% CIs, and the data summarised.

Data synthesis and analysis The aim of this review was to evaluate the effect of resistance muscle strengthening on physical function in people undergoing abdominal surgery for cancer. For each study, means and SD of outcomes focused on physical function were extracted. Outcomes relating directly to surgery, length of stay, infection and other postsurgical complications were not considered in this review. Assessment was made of the outcome measures for physical function that were used in different studies, before progression to pooling of data for analysis of the most common outcome measure. Treatment effect of individual studies is reported as mean difference and 95% CIs, and the data summarised. Risk of bias was assessed with the Physiotherapy Evidence Database (PEDro) scale.21 Items assessed included exclusion criteria, procedures for group allocation and missing data, participant, therapist and assessor blinding, and reporting of results. Studies were then graded using the Cochrane Reviews Grading of Recommendations Assessment, Development and Evaluation criteria.21 22

base (PEDro) scale.21 Items assessed included exclusion criteria, procedures for group allocation and missing data, participant, therapist and assessor blinding, and reporting of results. Studies were then graded using the Cochrane Reviews Grading of Recommendations Assessment, Development and Evaluation criteria.21 22 Results Search strategy and selection of articles The initial search strategy resulted in 588 publications. Following screening of titles and abstracts, 24 studies met the inclusion criteria and were accessed for review of the full text, of which 2 eligible studies23 24 were included in the review (see table 1 and figure 1). Full-text studies were excluded for a number of reasons: (1) the study lacked a well-defined muscle-strengthening intervention (n=18); (2) the study did not include patients undergoing abdominal surgery for cancer (n=4); and (3) the study did not use a physical function outcome measure (muscle strength, self-report questionnaires or physical performance measures). Table 1 Characteristics of included studies Methods Participants Intervention Relevant outcomes Risk of bias Dronkers et al23 Randomised study investigating the preoperative effect of an exercise programme in participants with colon cancer. Exercise group, n=22Age: 71.1±6.3Gender: 15 male, 7 femaleControl group, n=20Age: 68.8±6.4Gender: 16 male, 4 female Supervised programme 2×week for 2–4 weeks (mean 5.1±1.9) and home-based programme of walking or cycling for a minimum of 30 min per day (perceived exertion of 11–13 Borg Scale). Programme:Warm up.

cancer. Exercise group, n=22Age: 71.1±6.3Gender: 15 male, 7 femaleControl group, n=20Age: 68.8±6.4Gender: 16 male, 4 female Supervised programme 2×week for 2–4 weeks (mean 5.1±1.9) and home-based programme of walking or cycling for a minimum of 30 min per day (perceived exertion of 11–13 Borg Scale). Programme:Warm up. Resistance training of the lower limb extensors—equipment and method not stated (maximum of 1 set of 8–15 repetitions at 60%–80% of the one repetition maximum). Inspiratory muscle training (10%–60% max inspiratory pressure for 240 breathing cycles. Aerobic training—method and equipment not stated (55%–75% max HR or perceived exertion of 11–13 Borg Scale for 20–30 min). Functional activities according to patients’ capabilities and interests (Vreede et al,28 regimen—no other information provided).

Inspiratory muscle training (10%–60% max inspiratory pressure for 240 breathing cycles. Aerobic training—method and equipment not stated (55%–75% max HR or perceived exertion of 11–13 Borg Scale for 20–30 min). Functional activities according to patients’ capabilities and interests (Vreede et al,28 regimen—no other information provided). Timed Up and GoChair rise timePhysical Activity QuestionnaireAbbreviated Fatigue QuestionnaireEORTC QLQ-C30 Global Health/Functional Scale/Symptom Scale PEDro score 8/11GRADE criteria—moderate Ahn et al24 Randomised study investigating the effect of a postsurgical, inpatient exercise programme in patients with stages I–III colon cancer. Exercise group, n=17Age: 55.61±7.11Gender: 12 male, 5 femaleControl group, n=14Age: 57.43±6.12Gender: 5 male, 9 female Supervised exercise programme 2×day, 15 min/sessionSubdivided into three phases:Implemented while subjects were still unable to get out of bed: stretching (neck, shoulder, wrist, ankle and pelvis), pelvic tilt—isometric, resistance exercise (ankle dorsiflexion and plantar flexion against the hand of the therapist), unsupervised sitting or walking in the ward. Performed once subjects were able to get out of the bed, but had limited ambulation: stretching (whole body, leg and shoulder), pelvic tilt and thrust, one leg raise, crunch, resistance exercise (1 set, 10 repetitions) with 1 lb weight (chest, shoulder, arm, thigh and calf), unsupervised walking.

Timed Up and GoChair rise timePhysical Activity QuestionnaireAbbreviated Fatigue QuestionnaireEORTC QLQ-C30 Global Health/Functional Scale/Symptom Scale PEDro score 8/11GRADE criteria—moderate Ahn et al24 Randomised study investigating the effect of a postsurgical, inpatient exercise programme in patients with stages I–III colon cancer. Exercise group, n=17Age: 55.61±7.11Gender: 12 male, 5 femaleControl group, n=14Age: 57.43±6.12Gender: 5 male, 9 female Supervised exercise programme 2×day, 15 min/sessionSubdivided into three phases:Implemented while subjects were still unable to get out of bed: stretching (neck, shoulder, wrist, ankle and pelvis), pelvic tilt—isometric, resistance exercise (ankle dorsiflexion and plantar flexion against the hand of the therapist), unsupervised sitting or walking in the ward. Performed once subjects were able to get out of the bed, but had limited ambulation: stretching (whole body, leg and shoulder), pelvic tilt and thrust, one leg raise, crunch, resistance exercise (1 set, 10 repetitions) with 1 lb weight (chest, shoulder, arm, thigh and calf), unsupervised walking. Performed when subjects were able to ambulate without any discomfort; in addition to phase 2 exercises, resistance strengthening increased to 12 repetition×3 sets, supervised balance exercises—one leg standing, one leg calf raise, hip adduction, hip abduction, hip flexion with knee bent, hip extension, unsupervised walking.

rmed when subjects were able to ambulate without any discomfort; in addition to phase 2 exercises, resistance strengthening increased to 12 repetition×3 sets, supervised balance exercises—one leg standing, one leg calf raise, hip adduction, hip abduction, hip flexion with knee bent, hip extension, unsupervised walking. Timed one-leg standSit-to-stand in 30 sTecumseh step test PEDro score 8/11GRADE criteria—moderate EORTC QLQ-C30, European Organization for Research and Treatment of Cancer Quality of Life Questionnaire; GRADE, Grading for Recommendations Assessment, Development and Evaluation; HR, heart rate; PEDro, Physiotherapy Evidence Database. Description of included studies Characteristics of the participants and intervention of the two included studies are presented in table 1. Both were small (n=42 and 31) single-centre studies investigating participants undergoing abdominal surgery for excision of cancer of the colon. Dronkers et al23 investigated the effect of a preoperative exercise programme on preoperative outcomes, and Ahn et al24 investigated the effect of a postoperative exercise programme on short-term outcomes at discharge from hospital. The participants in the preoperative study were aged 10–15 years older than those in the postoperative study. In terms of gender, a higher proportion of men participated in both studies.

mes, and Ahn et al24 investigated the effect of a postoperative exercise programme on short-term outcomes at discharge from hospital. The participants in the preoperative study were aged 10–15 years older than those in the postoperative study. In terms of gender, a higher proportion of men participated in both studies. The preoperative intervention of Dronkers et al23 included a twice-weekly supervised exercise programme and a home-based programme of walking or cycling for a minimum of 30 min per day for 2–4 weeks before admission for surgery. In addition to a single set of resistance strengthening exercises of the leg (8–15 repetitions at 60%–80% of the one repetition maximum), the programme included inspiratory muscle training, aerobic training at 55%–75% max heart rate (HR) or perceived exertion of 11–13 Borg Scale for 20–30 min, and functional activities. A full description of the resistance exercise was not published. Three of the intervention groups (13.6%) did not complete the study with their data analysed as intention to treat.

aining, aerobic training at 55%–75% max heart rate (HR) or perceived exertion of 11–13 Borg Scale for 20–30 min, and functional activities. A full description of the resistance exercise was not published. Three of the intervention groups (13.6%) did not complete the study with their data analysed as intention to treat. The postoperative intervention of Ahn et al24 comprised a twice-daily 15 min supervised exercise programme performed by the participant until discharge from hospital (mean 8.87±2.28 days). In addition to resistance strengthening exercises of the chest, shoulder, arm, thigh and calf leg, the programme included stretching exercises for the neck, shoulder, wrist, ankle and pelvis, core trunk exercises and ambulation. In terms of the strengthening exercises, resistance was applied manually by the therapist initially and then using 1 lb free weights. During phase 2, one set of 10 repetitions was performed, and in phase 3, three sets of 12 repetitions were performed. Because these studies used different outcome measures, it was not possible to pool the data in order to analyse mean changes in physical function outcomes.

st initially and then using 1 lb free weights. During phase 2, one set of 10 repetitions was performed, and in phase 3, three sets of 12 repetitions were performed. Because these studies used different outcome measures, it was not possible to pool the data in order to analyse mean changes in physical function outcomes. Risk of bias of included studies The methodological quality of the two included studies was rated as moderate according to the GRADE criteria, that is, randomised studies with unclear bias or well-designed observational studies with large, consistent and precise estimates of the magnitude of an intervention effect. Difficulty in blinding trial participants and therapists to the intervention meant studies were not rated as high. Both studies scored 8 out of 11 on the PEDro scale. Block randomisation using prepared envelopes, stratified by age (60–70 and >70) by someone independent of the study, was used in the preoperative study. Randomisation, at a 1-to-1 ratio, into study groups via minimisation to balance prognostic factors between groups (age and gender) was used in the postoperative study. In the preoperative study the gender distribution was similar in the control and intervention groups; however, in the postoperative study, twice as many men were randomised to the exercise group than the control group despite the minimisation procedures to balance gender between groups. In relation to the description of the intervention, some information was lacking in terms of equipment and methodology with regard to the aerobic and functional activity components of the preoperative intervention.

ercise group than the control group despite the minimisation procedures to balance gender between groups. In relation to the description of the intervention, some information was lacking in terms of equipment and methodology with regard to the aerobic and functional activity components of the preoperative intervention. Effect of strengthening exercise Preoperative muscle strengthening The mean difference and upper and lower 95% CI between the control and intervention group in the study by Dronkers et al23 are shown in table 2. The five-session preoperative exercise programme had no significant effect on preoperative Timed Up and Go, chair rise time test, self-reported physical activity, quality of life and fatigue. Statistical power for six out of the seven measures was unacceptably low. Effect on postsurgery outcomes was not evaluated. Table 2 Summary of effect of exercise intervention

Effect of strengthening exercise Preoperative muscle strengthening The mean difference and upper and lower 95% CI between the control and intervention group in the study by Dronkers et al23 are shown in table 2. The five-session preoperative exercise programme had no significant effect on preoperative Timed Up and Go, chair rise time test, self-reported physical activity, quality of life and fatigue. Statistical power for six out of the seven measures was unacceptably low. Effect on postsurgery outcomes was not evaluated. Table 2 Summary of effect of exercise intervention Mean between-group difference Lower 95% CI Upper 95% CI Statistical power* Dronkers et al,23 preoperative intervention Timed Up and Go (s) −1.20 −2.78 0.38 31.2 Chair rise (s) −5.40 −9.24 −1.56 77.3 Physical activity (min/day) 44.00 −141.82 229.82 7.3 Abbreviated Fatigue Questionnaire −3.90 −7.41 −0.39 57.6 EORTC QLQ-C30 (Global Health) −4.00 −15.57 7.57 10.2 EORTC QLQ-C30 (Functional Scale) 12.00 −28.26 52.86 87.4 EORTC QLQ-C30 (Symptom Scale) 36.00 −31.09 103.09 17.7 Ahn et al,24  postoperative intervention Timed one-leg stand (s) −7.28 −16.25 1.69 40.0 Sit-to-stand (repetitions) −2.00 −5.78 1.78 17.7 Tecumseh step test (heart rate, beats/min) 10.29 1.63 18.95 64.8 *Probability of rejecting a false null hypothesis (where α=0.05), for a between-group comparison of means at study endpoint.

al,24  postoperative intervention Timed one-leg stand (s) −7.28 −16.25 1.69 40.0 Sit-to-stand (repetitions) −2.00 −5.78 1.78 17.7 Tecumseh step test (heart rate, beats/min) 10.29 1.63 18.95 64.8 *Probability of rejecting a false null hypothesis (where α=0.05), for a between-group comparison of means at study endpoint. Postoperative muscle strengthening The mean difference and upper and lower 95% CI between the control and intervention groups in the study by Ahn et al24 are also shown in table 2. The inpatient postoperative exercise programme had no significant effect at time of discharge from hospital on ability to balance on one leg, number of sit-to-stands in 30 s or aerobic capacity (estimated from performance of the Tecumseh step test). Statistical power was not sufficient to allow any conclusion for or against the preferential use of any of the outcome measures that were used in this trial. Effect on functional recovery postdischarge from hospital was not evaluated.

s in 30 s or aerobic capacity (estimated from performance of the Tecumseh step test). Statistical power was not sufficient to allow any conclusion for or against the preferential use of any of the outcome measures that were used in this trial. Effect on functional recovery postdischarge from hospital was not evaluated. Discussion Our aim was to systematically review the evidence on the effectiveness of preoperative and postoperative strengthening exercises on short-term and long-term recovery of physical function in patients undergoing abdominal surgery for cancer. Two studies were included, which represented 73 patients (48 men and 25 women) undergoing abdominal surgery for cancer. One exercise programme was undertaken preoperatively and the other postoperatively until discharge from hospital. This represents insufficient evidence to determine whether this type of preoperative or postoperative resistance muscle-strengthening exercise programme improves or negatively affects functional outcomes in patients undergoing abdominal surgery for cancer.

ly and the other postoperatively until discharge from hospital. This represents insufficient evidence to determine whether this type of preoperative or postoperative resistance muscle-strengthening exercise programme improves or negatively affects functional outcomes in patients undergoing abdominal surgery for cancer. The study by Dronkers et al,23 which investigated a preoperative exercise programme, was statistically underpowered with the exception of the functional measure derived from the quality of life scale. The programme included resistance strengthening of the lower limb muscle extensors and was performed for a mean of five sessions. This may not be sufficient to provide an adequate training stimulus to significantly increase muscle strength. Indeed, guidelines published by the American College of Sports Medicine recommend resistance exercise 2–3 times per week with 2–4 sets of 10–15 repetitions to improve strength in middle-aged and older persons.25

is may not be sufficient to provide an adequate training stimulus to significantly increase muscle strength. Indeed, guidelines published by the American College of Sports Medicine recommend resistance exercise 2–3 times per week with 2–4 sets of 10–15 repetitions to improve strength in middle-aged and older persons.25 In contrast, the study by Ahn et al24 investigated a postoperative exercise programme, but this was also statistically underpowered and provides inconclusive evidence in support of the intervention and the use of particular outcome measures. The intervention was different from that of Dronkers et al23 in that it used a progressive resistance programme involving the upper and lower limbs, together with stretching, functional balance strengthening and walking. Also, isometric strengthening exercises were commenced early postoperatively while the patient was still in bed and then progressed to ‘resistance-through-range’ strengthening as well as balance strengthening exercises, until discharge from hospital. The mean hospital length of stay, for the study of Dronkers et al23 was 7 days for the control group, and in the exercise group it was 8 days. Similarly, for the study by Ahn et al,24 it was 8 days of exercise, and it is likely that this will not provide an adequate training stimulus to significantly increase muscle strength and function.

ngth of stay, for the study of Dronkers et al23 was 7 days for the control group, and in the exercise group it was 8 days. Similarly, for the study by Ahn et al,24 it was 8 days of exercise, and it is likely that this will not provide an adequate training stimulus to significantly increase muscle strength and function. There are some limitations to our review. We limited our inclusion by study design, only including randomised or quasi-randomised studies where there was a clear resistance muscle strengthening component as part of an exercise programme. It is possible that other studies have included muscle-strengthening exercises or functional exercises that will have an effect on muscle strength that have not been included in this review due to our inclusion criteria, and we advocate the Consensus on Exercise Reporting Template guidelines for reporting exercise intervention studies.26 The two studies included in the review recruited almost twice as many men as women, and the results may not reflect the general population. Future studies should focus on detailed descriptions of the exercise intervention, consistent outcome measures and longer intervention and follow-up times.27

ntervention studies.26 The two studies included in the review recruited almost twice as many men as women, and the results may not reflect the general population. Future studies should focus on detailed descriptions of the exercise intervention, consistent outcome measures and longer intervention and follow-up times.27 Our systematic review suggests that the use of resistance exercise interventions for recovery of physical function in patients undergoing abdominal surgery for cancer must be considered with caution. The small number of included underpowered studies and the inability to pool the results due to the heterogeneity of outcome measures mean that there is a lack of evidence for or against the use of this type of resistance muscle-strengthening exercise programmes to improve functional outcomes in these patients. While the studies give encouraging preliminary evidence that muscle-strengthening programmes may be feasible for abdominal cancer surgery patients, further large-scale, well-designed clinical trials are required to determine whether this type of exercise intervention is beneficial for this group of patients. Contributors: DS, FH, KC performed the systematic review. DS, FH, KC, AB, MG, RPH, MH, IH, DPL, TPH, CS, IS, LT, WH and HA contributed to study design, data analysis and interpretation and preparation of the manuscript. Funding: The study was funded by an NIHR Research for Patient Benefit grant (PB-PG-0613–31107). Competing interests: None declared. Patient consent: Not required. Provenance and peer review: Not commissioned; externally peer reviewed.

What is already known? Patellar tendon related pain is relatively common in physically active children and adolescents. What are the new findings? Only three randomised controlled clinical trials have tested an intervention versus comparator in a population with a reasonable number of children and adolescents. All studies had a high risk of bias. Specifically, ‘usual care’ provided was highly variable which provided a substantial source of bias. In one study, hyperosmolar dextrose injection (combined with local anaesthetic) was superior to both local anaesthetic injection and usual care. This review found no evidence to justify surgery for Osgood-Schlatter disease. Introduction Patellar tendon related pain conditions are common in children and adolescents, appear to begin in childhood and increase in prevalence during adolescence up to age 18.1 Patellar tendon related pain is an umbrella term, which encompasses Osgood-Schlatter’s disease (OSD), Sinding-Larsen-Johansson disease and patellar tendinopathy. OSD affects 1 in 10 adolescents and as many as 1 in 5 in certain sports.2 3 There is a paucity of high-quality evidence relating to prognosis.

up to age 18.1 Patellar tendon related pain is an umbrella term, which encompasses Osgood-Schlatter’s disease (OSD), Sinding-Larsen-Johansson disease and patellar tendinopathy. OSD affects 1 in 10 adolescents and as many as 1 in 5 in certain sports.2 3 There is a paucity of high-quality evidence relating to prognosis. In one prospective cohort study in 18 adolescents with OSD, even 2 years after initial diagnosis, over 60% demonstrated persistent patellar tendon changes (evaluated by ultrasonography) and continued to display deficits in functional performance. Individuals with a history of OSD can experience persistent pain, years later, in their early 20s,4–6 associated with OSD lesions, underscoring that these pain complaints may not go away on their own if treated with a ‘wait and see’ approach. Interventions which reliably improve long-term outcomes are needed.7 While there is a large body of randomised trials and high-quality systematic reviews for managing patellar tendon related disorders in adults, there is a complete lack of systematic evaluations of treatment strategies specifically for adolescents and children.8 Considering patellar tendon related pain seem to start around the age of 10,1 there is a need for evidence and syntheses specific to this population. This is especially important considering some of these patellar tendon related pain complaints are unique to adolescent period, for example, Sinding-Larsen-Johansson syndrome and OSD.

atellar tendon related pain seem to start around the age of 10,1 there is a need for evidence and syntheses specific to this population. This is especially important considering some of these patellar tendon related pain complaints are unique to adolescent period, for example, Sinding-Larsen-Johansson syndrome and OSD. The aim of this study was to perform a systematic review (and if possible, meta-analysis) of the benefits and harms of different treatment options for patella tendon related pain in children and adolescents. Methods The systematic review was prospectively registered the review in PROSPERO (registration number 82736 link). The review was informed according to the Cochrane guidelines and is reported according to the PRISMA statement. Data sources and searches We carried out a systematic search in the following bibliographic databases: Medline via Pubmed, Embase via OVID, CINAHL via Ebsco, SportDiscus up until 24 November 2017. The search terms and strategy were developed with a research librarian and are available in online supplementary material 1. 10.1136/bmjsem-2018-000383.supp1Supplementary data

Data sources and searches We carried out a systematic search in the following bibliographic databases: Medline via Pubmed, Embase via OVID, CINAHL via Ebsco, SportDiscus up until 24 November 2017. The search terms and strategy were developed with a research librarian and are available in online supplementary material 1. 10.1136/bmjsem-2018-000383.supp1Supplementary data Inclusion/exclusion criteria Studies evaluating any intervention for any type of patellar tendon related pain in children or adolescents were eligible for inclusion, providing the design included an intervention and a comparator. Specifically, non-randomised controlled clinical trials and randomised controlled trials (RCTs) (including semi/quasi-randomised and cluster randomised trials) were eligible. Any therapeutic intervention or control treatments were included (including, but not restricted to, non-surgical interventions, injection-based interventions, exercise, surgical interventions or standard care/wait and see  and so on). Studies had to include children or adolescents (aged under 18 years), with patellar tendon related pain. We included any publications in English or Scandinavian languages. Studies were excluded if reporting on primary complaints of patellofemoral joint disorders including patellofemoral pain a, patellofemoral instability, acute traumatic causes of knee pain, inflammatory arthritis, or if they included a population age >18 years. If the study included a mixed-age population extending across the age limits in our inclusion criteria, we requested age-specific data for those <18 years from trial authors.

pain a, patellofemoral instability, acute traumatic causes of knee pain, inflammatory arthritis, or if they included a population age >18 years. If the study included a mixed-age population extending across the age limits in our inclusion criteria, we requested age-specific data for those <18 years from trial authors. Selection of studies Duplicates were removed and relevant studies identified from the search were imported into Covidence for screening. Studies were independently screened by title and abstract by two authors (TO and GC). This was followed by full text evaluation of the selected studies from the first selection step by both authors. Disagreement between the two reviewers was solved by consensus involving a third reviewer (BD). Outcomes Our primary domains of interest were pain, function and sport participation. Reports on the number and type(s) of adverse effects (harms) related to interventions were extracted and recorded. Additionally, all other reported outcomes were considered of potential interest and extracted. We did not impose any restrictions on endpoints; a priori we defined endpoints of interest as immediate effects (0–7 days after receiving treatment), short-term (1 week to 3 months), medium-term (3–6 months) and long-term (above 6 months).

all other reported outcomes were considered of potential interest and extracted. We did not impose any restrictions on endpoints; a priori we defined endpoints of interest as immediate effects (0–7 days after receiving treatment), short-term (1 week to 3 months), medium-term (3–6 months) and long-term (above 6 months). Data extraction Two reviewers (GC and BD) independently extracted data. Data were extracted using a custom data extraction sheet in Covidence. Data regarding study design and setting, sample characteristics (including diagnoses, age, sex, demographics), inclusion criteria, intervention types and characteristics, follow-up, compliance, withdrawals, outcomes and any adverse events were extracted. Inconsistencies were resolved by consensus discussion. A third review (MSR) was available for disagreements that could not be resolved by discussion. If relevant data were not available from full-text articles or trial registrations, authors were contacted to provide this information. For studies including mixed populations (both adolescents and adults), individual participant data for those 18 and under were requested. Authors were sent two subsequent reminders over 6 weeks. Studies were excluded from data synthesis if individual patients were not provided or studies included less than 10 participants under 18.

ncluding mixed populations (both adolescents and adults), individual participant data for those 18 and under were requested. Authors were sent two subsequent reminders over 6 weeks. Studies were excluded from data synthesis if individual patients were not provided or studies included less than 10 participants under 18. Risk of bias assessment Two independent raters (BD and GC) assessed risk of bias using the Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. Each trial was evaluated across seven domains of bias, including one or more items that are were appraised in two parts. Assessment followed the description in the Cochrane Handbook for Systematic Review of Interventions, V.5.1 (Part 2: 8.5.1) as follows. First, the relevant trials’ characteristics related to the item were summarised. Second, each bias domain was judged as high or low risk of bias, according to their possible effect on the results of the trial. When the possible effect was unknown or insufficient detail was reported, the item was judged as unclear. Disagreements were resolved by discussion. A third party (MSR) was available in case of persistent disagreement. Data analysis It was specified a priori, a meta-analysis would only be performed if data were available for similar time-points, outcomes and interventions. As this was not possible, we conducted a narrative synthesis of the results based on the domains of interest.

Risk of bias assessment Two independent raters (BD and GC) assessed risk of bias using the Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. Each trial was evaluated across seven domains of bias, including one or more items that are were appraised in two parts. Assessment followed the description in the Cochrane Handbook for Systematic Review of Interventions, V.5.1 (Part 2: 8.5.1) as follows. First, the relevant trials’ characteristics related to the item were summarised. Second, each bias domain was judged as high or low risk of bias, according to their possible effect on the results of the trial. When the possible effect was unknown or insufficient detail was reported, the item was judged as unclear. Disagreements were resolved by discussion. A third party (MSR) was available in case of persistent disagreement. Data analysis It was specified a priori, a meta-analysis would only be performed if data were available for similar time-points, outcomes and interventions. As this was not possible, we conducted a narrative synthesis of the results based on the domains of interest. Results A total of 530 studies were identified by the search strategy. After screening, eight studies were identified as eligible for inclusion (figure 1, PRISMA flowchart). Of these, seven were RCTs, and one was a retrospective non-randomised controlled trial. Two studies specifically included an adolescent-only population (both OSD). The other six studies were on patellar tendinopathy and included a mixed sample. Study characteristics are provided in table 1 and table 2. Additional information of included studies is available in online supplementary material 1 (S1).

trial. Two studies specifically included an adolescent-only population (both OSD). The other six studies were on patellar tendinopathy and included a mixed sample. Study characteristics are provided in table 1 and table 2. Additional information of included studies is available in online supplementary material 1 (S1). 10.1136/bmjsem-2018-000383.supp2Supplementary data Figure 1 Flowchart of systematic review protocol. Table 1 Study characteristics

trial. Two studies specifically included an adolescent-only population (both OSD). The other six studies were on patellar tendinopathy and included a mixed sample. Study characteristics are provided in table 1 and table 2. Additional information of included studies is available in online supplementary material 1 (S1). 10.1136/bmjsem-2018-000383.supp2Supplementary data Figure 1 Flowchart of systematic review protocol. Table 1 Study characteristics Author Year Journal Setting Population Age Type of study Intervention/s Comparator Outcomes Time points Biernatet al 11 2014 J Strength Cond Res University physical therapy Male volleyball players with patellar tendinopathy Mixed Randomised controlled trial Specific exercises including eccentric decline squat Continued competitive activity and functional exercises Visa-P Maximal power Jump height Baseline, 12 weeks, 24 weeks Cannellet al 16 2001 Br J Sports Med Sports medicine centre Mixed athletes with jumper’s knee Mixed Randomised controlled trial Drop squat exercises Leg extension/curl exercises Pain score Quadriceps moment Hamstring moment Baseline, 4 weeks Jonsson et al 14 2005 Br J Sports Med Sports medicine unit Athletes with jumper’s knee Mixed Randomised controlled trial Eccentric exercises Concentric exercises VISA Pain scores (VAS) Baseline, 6 weeks, 12 weeks Topolet al 9 2011 Pediatrics Physical medicine Children with Osgood Schlatter’s who played sports All <18 years old Randomised controlled trial Local anaesthetic and Local anaesthetic plus dextrose Usual care NPSS (Nirschl Pain Phase Scale) Baseline, 3 months Trailet al 10 1988 J Pediatr Orthop Secondary care—Orthopaedics Children with Osgood Schlatter’s All <18 years old Retrospective non-randomised parallel group design Tibial sequestrectomy and mixture of cast/injections/physiotherapy Reduction of activity/sport avoidance and mixture of cast/injections/physiotherapy Proportion symptomatic Complication rate Variable as at final review Wanget al 12 2007 Am J Sports Med Orthopaedic surgery department Patients with chronic patellar tendinopathy Mixed Randomised controlled trial Shockwave treatment followed by period of light activity NSAIDs, physiotherapy and exercise program Pain scores VISA Knee range of movement Baseline, variable—2 to 3 years Willberget al 15 2011 Br J Sports Med Sports medicine unit Patients with patellar tendinopathy Mixed Randomised controlled trial Arthroscopic shaving Sclerosing polidocanol injections VAS at rest VAS on activity Satisfaction Baseline, variable—approximately 12 months Van Ark et al 13 2016 J Sci Med Sport Physiotherapy department Volleyball/basketball pla

d Sports medicine unit Patients with patellar tendinopathy Mixed Randomised controlled trial Arthroscopic shaving Sclerosing polidocanol injections VAS at rest VAS on activity Satisfaction Baseline, variable—approximately 12 months Van Ark et al 13 2016 J Sci Med Sport Physiotherapy department Volleyball/basketball pla yers with patellar tendinopathy Mixed Randomised controlled trial Isometric exercises Isotonic exercises Pain score Baseline, 4 weeks NSAIDs, non-steroidal anti-inflammatory drugs. Table 2 Details of study participants demographics and whether data were provided

d Sports medicine unit Patients with patellar tendinopathy Mixed Randomised controlled trial Arthroscopic shaving Sclerosing polidocanol injections VAS at rest VAS on activity Satisfaction Baseline, variable—approximately 12 months Van Ark et al 13 2016 J Sci Med Sport Physiotherapy department Volleyball/basketball pla yers with patellar tendinopathy Mixed Randomised controlled trial Isometric exercises Isotonic exercises Pain score Baseline, 4 weeks NSAIDs, non-steroidal anti-inflammatory drugs. Table 2 Details of study participants demographics and whether data were provided Author Year Journal Intervention group age Comparator group age Intervention group sex Comparator group sex Biernat et al 11 2014 J Strength Cond Res 17.2 (0.6) 16.5 (0.8) All male All male Cannell et al 16 2001 Br J Sports Med No response from author within time limit Jonsson et al 14 2005 Br J Sports Med Author confirmed age specific data not obtainable Topol, et al 9 2011 P ediatrics 13.3 (range 10–17) 51 boys and three girls Trail et al 10 1988 J Pediatr Orthop 13 years 9 months (range 11–17) 12 years 7 months (range 10–17) Not stated Not stated Wang et al 12 2007 Am J Sports Med Author confirmed no participants under age 18 took part in the study (note study did state age range lower limit was 16) Willberg et al 15 2011 Br J Sports Med No response from author within time limit Van Ark et al 13 2016 J Sci Med Sport Only one under 18 year old participant took part hence study excluded from data analysis Osgood-Schlatter Topol et al 9 compared usual care versus local anaesthetic versus local anaesthetic and dextrose in a threearmed RCT for OSD. The outcome was the Nirschl Pain Phase Scale (NPPS) at 3 months. The superiority of dextrose and local anaesthetic compared with local anaesthetic alone was shown (SMD: 0.83, 95% CI 0.20 to 1.45, p=0.006) or usual care (SMD: 1.66, 95% CI 0.96 to 2.36, p<0.0001) (see online supplementary material for Forest plots). 10.1136/bmjsem-2018-000383.supp3Supplementary data

NPPS) at 3 months. The superiority of dextrose and local anaesthetic compared with local anaesthetic alone was shown (SMD: 0.83, 95% CI 0.20 to 1.45, p=0.006) or usual care (SMD: 1.66, 95% CI 0.96 to 2.36, p<0.0001) (see online supplementary material for Forest plots). 10.1136/bmjsem-2018-000383.supp3Supplementary data 10.1136/bmjsem-2018-000383.supp4Supplementary data 10.1136/bmjsem-2018-000383.supp5Supplementary data 10.1136/bmjsem-2018-000383.supp6Supplementary data 10.1136/bmjsem-2018-000383.supp7Supplementary data 10.1136/bmjsem-2018-000383.supp8Supplementary data Trail10 compared surgery (tibial sequestrectomy) with conservative treatment in a retrospective non-randomised parallel group study of OSD. Participants were categorically rated as ‘symptomatic’ or ‘asymptomatic’ at final review approximately 5 years from initial diagnosis. The odds of being asymptomatic was not significantly different with surgery compared with conservative treatment (OR 1.48, 95% CI 0.33 to 6.65) (see online supplementary material for Forest plots). Several patients in both groups were subject to other interventions (including refraining from sport, cast treatment, physiotherapy and steroid injections).

not significantly different with surgery compared with conservative treatment (OR 1.48, 95% CI 0.33 to 6.65) (see online supplementary material for Forest plots). Several patients in both groups were subject to other interventions (including refraining from sport, cast treatment, physiotherapy and steroid injections). Patellar tendinopathy Only one11 of the six studies including mixed populations of adolescents and adults was included in the data synthesis (details of exclusion outlined below). This study by Biernat et al compared the effect of usual care to eccentric exercises in male adolescents on patient-recorded outcomes (VISA-P questionnaire). There was no significant difference between groups at either 12 or 24 weeks (SMD: 0.45, 95% CI −0.44 to 1.35, p=0.32 and SMD 0.37, 95% CI 0.52 to 1.26, p=0.42, respectively) (see online supplementary material for Forest plot) table 2. Two studies were excluded from data synthesis as they did not contain a large enough sample of adolescents (N=0 and N=1 adolescent included).12 13 One study was excluded from data synthesis because data for adolescents could not be extracted,14 and two studies could not be included in the data synthesis due to non-response to requests for data for adolescent participants.15 16

ot contain a large enough sample of adolescents (N=0 and N=1 adolescent included).12 13 One study was excluded from data synthesis because data for adolescents could not be extracted,14 and two studies could not be included in the data synthesis due to non-response to requests for data for adolescent participants.15 16 Harms Only one study reported on potential harms. Trail reported a lower overall complication rate for conservative treatment,10 compared with surgery (OR 0.11, 95% CI 0.03 to 0.38) (see online supplementary material for plot  harms OR).10 Complications were primarily presence of bony prominence (6 in conservative group and 18 in surgical group). The surgical group had additional complications (three with areas of anaesthesia lateral to scar, one infection, one wound dehiscence, one stiffness and one recurvatum). Risk of bias Overall, all studies were deemed to be at high risk of bias, particularly in the domains of reporting bias and performance bias (figure 2 and figure 3) Figure 2 Risk of bias within studies. Figure 3 Risk of bias summary graph. Studies identified and excluded at each stage are detailed in figure 1.

Harms Only one study reported on potential harms. Trail reported a lower overall complication rate for conservative treatment,10 compared with surgery (OR 0.11, 95% CI 0.03 to 0.38) (see online supplementary material for plot  harms OR).10 Complications were primarily presence of bony prominence (6 in conservative group and 18 in surgical group). The surgical group had additional complications (three with areas of anaesthesia lateral to scar, one infection, one wound dehiscence, one stiffness and one recurvatum). Risk of bias Overall, all studies were deemed to be at high risk of bias, particularly in the domains of reporting bias and performance bias (figure 2 and figure 3) Figure 2 Risk of bias within studies. Figure 3 Risk of bias summary graph. Studies identified and excluded at each stage are detailed in figure 1. Discussion The key finding of this review is the lack of studies assessing the effectiveness of interventions for patellar tendon related pain in children and adolescents. Despite commonality of these disorders in this population, only three studies have assessed any intervention. All three were at high risk of bias, especially reporting bias, blinding and had significant methodological weaknesses (with one being a retrospective parallel group case-series). This makes in nearly impossible to guide clinical practice on how to best to manage these common pain complaints in children and adolescents.

ll three were at high risk of bias, especially reporting bias, blinding and had significant methodological weaknesses (with one being a retrospective parallel group case-series). This makes in nearly impossible to guide clinical practice on how to best to manage these common pain complaints in children and adolescents. Adults and adolescents—potentially not the same The problem is that while there is evidence supporting certain interventions (eg, tendon loading programmes) for patellar tendon related pain in adults,17 the findings of these cannot be directly transferred into children and adolescents. The pathological features of tendinopathy includes increased cellularity, a loss of collagen structure and chondroid metaplasia.18 19 There is an increasing body of evidence demonstration the role of chronic inflammation as part of this degenerative process18 20 21 Tendon loading programmes have been associated with an improved collagen turnover in adults suggesting a possible mechanisms of effect. However, an important difference between adults and adolescents is the maturity of the patellar tendon insertion, the enthesis organ.22 It is a possible that overly aggressive sporting activity and tendon loading during development may actually have a negative long-term impact in terms of both tendon structure and the risk of adulthood symptoms. Notably, a recent study has shown that increased sporting activity has a detrimental impact on the developing hip, which supports this theory. As we do not yet have a clear understanding of how the pathology of patellar tendon related pain during maturation, caution needs to be taken in applying such effective interventions from adults to children and adolescents. This is particularly as adolescence is proposed to be a critical time for the formation of normal tendon attachment.23 24

lear understanding of how the pathology of patellar tendon related pain during maturation, caution needs to be taken in applying such effective interventions from adults to children and adolescents. This is particularly as adolescence is proposed to be a critical time for the formation of normal tendon attachment.23 24 Devil in the detail One of the serious concerns with the included studies is the multiple differences in treatment between control and experimental arms. For example, in Topol et al 9 the control group did not have a period of rest, which was in contrast to both injection therapy groups. This makes it difficult to elucidate if the observed effects were actually due to the intervention or to the removal of aggravating activities. Furthermore, a dextrose injection at 3 months was used as an ‘incentive’ for study participation and offered to all participants that did not reach an NPPS score of 0 (regardless of treatment allocation), to avoid dropout in the control arm. This could potentially influence participants’ reporting, if they had a favourable perception of the injection, and reported lower outcomes to receive it.

study participation and offered to all participants that did not reach an NPPS score of 0 (regardless of treatment allocation), to avoid dropout in the control arm. This could potentially influence participants’ reporting, if they had a favourable perception of the injection, and reported lower outcomes to receive it. The high frequency of patellar tendon related disorders in the sporting population is consistent with an ’overuse’ type aetiology. In this context, any advice on activity modification and the manner in which it is delivered may be crucial in influencing both short-term and long-term outcomes. Keeping this in mind, it is arguable what constitutes ‘usual care’ in daily practice and in the included studies may have been. The ‘usual care’ was highly variable and included factors such as the degree to which sporting activity was continued and the degree to which pain was experienced and/or allowed during sport or exercises. In the study by Topol et al,9 only participants in the injection arms were encouraged to engage in a sporting activity, only if the activity was not accompanied or followed by pain during the period of treatment. This contrasted with Biernat et al in which both experimental and control groups were subject to the same advice as regards continuing sporting activity.11

in the injection arms were encouraged to engage in a sporting activity, only if the activity was not accompanied or followed by pain during the period of treatment. This contrasted with Biernat et al in which both experimental and control groups were subject to the same advice as regards continuing sporting activity.11 Harms The study by Trail10 demonstrated a significantly higher complication rate with surgery versus conservative treatment, indicating that currently the use of surgery in this context should only take place in the more regulated context of a clinical trial. Notably, no studies analysed longer term outcomes into early adulthood, highlighting an area for future research.

ly higher complication rate with surgery versus conservative treatment, indicating that currently the use of surgery in this context should only take place in the more regulated context of a clinical trial. Notably, no studies analysed longer term outcomes into early adulthood, highlighting an area for future research. Lack of knowledge hampers clinical practice The absence of high-quality evidence presents a challenge to clinicians treating children and adolescents with patellar tendon related pain. Adolescence is a key developmental period, with a mature proximal patellar tendon enthesis occurring after peak height velocity.25 Given that patellar tendon structure is a predictor for pain,26 it is reasonable to argue that caution should be exercised when treating patients with an immature enthesis. Training beyond pain in order to satisfy short-term sporting objectives could have negative long term implications for the individual.27 Until further evidence arises, clinicians should be aware of the potential differences in patellar tendon related pain between adolescents and adults. These types of pain complaints are often associated with high levels of sports participation and sports specialisation. As sustained overload is thought to be a risk factor for developing, for example, OSD,28 it seems sensible that the mainstay of treatment include some form of load management. This type of treatment could include education where adolescents and parents are taught to modify training and loading (including recovery) based on symptoms but this has yet to be evaluated in a clinical trial.

, for example, OSD,28 it seems sensible that the mainstay of treatment include some form of load management. This type of treatment could include education where adolescents and parents are taught to modify training and loading (including recovery) based on symptoms but this has yet to be evaluated in a clinical trial. Future research Malliaras et al 17 concluded that high-quality studies comparing different loading programmes and evaluating clinical and mechanistic outcomes are needed in both Achilles and patellar tendinopathy rehabilitation in adults.17 The evidence relating to adolescents and children is worse, and this is concerning given the potential long-term consequences of overuse on the developing bone tendon interface. Understanding the impact of patellar tendon related pain disorders on the maturing tendon and entheses on both structural and functional outcomes in both the short and longer term, will be critical in the understanding of the natural history of patellar tendon related disorders in adolescents. It is also essential to explore the needs and preference of the young patients being treated. It is likely that these developing young individuals have specific needs which have yet to be addressed during treatment. Future studies may also wish to identify phenotypes associated with a particular poor prognosis to understand the trajectory of pain and impairments during this crucial period of life.

eing treated. It is likely that these developing young individuals have specific needs which have yet to be addressed during treatment. Future studies may also wish to identify phenotypes associated with a particular poor prognosis to understand the trajectory of pain and impairments during this crucial period of life. Despite the commonality of patella tendon related pain in children and adolescents, there is a paucity of evidence specific to this population to guide clinical practice. There is weak evidence to support the use of dextrose injections and no evidence to support the use of specific types of exercises. Further research is essential to ascertain how best to manage the many children and adolescents suffering from these pain complaints. Until further evidence arises, clinicians should include load modification and advise on a return to sport based on symptoms. Contributors: BJFD, SH, SK and MSR contributed to study conception and design, data analysis, drafting of the article and critical revision of the article. GC and TO contributed to study design, data collection and analysis and the critical revision of the article. NT contributed to study design, data collection and critical revision of the article. Funding: The TRYG Foundation is acknowledged for provided support for this project (Grant ID: 118547). Competing interests: None declared. Patient consent: Not required. Provenance and peer review: Not commissioned; externally peer reviewed.

What are the new findings? Moderate-intensity exercise bouts of <10 min have favourable effects on changing body composition in sedentary adults from either sex. There is no difference in the magnitude of body composition change from different exercise prescriptions as long as cumulative exercise durations are equal. How might it impact on clinical practice in the near future? Individuals aged above 50 years could benefit from a review of current exercise recommendations. More appealing exercise prescription options for individuals at risk of the metabolic syndrome triad will make it easier to control inactivity-related non-communicable diseases.

How might it impact on clinical practice in the near future? Individuals aged above 50 years could benefit from a review of current exercise recommendations. More appealing exercise prescription options for individuals at risk of the metabolic syndrome triad will make it easier to control inactivity-related non-communicable diseases. Introduction Sedentary lifestyles and related morbidities are on the rise, and physical inactivity currently is the number four cause of death worldwide.1 2 Globally, 51%–79% of adults are physically inactive and do not meet their exercise recommendations.2 3 This trend has also been shown in sub-Saharan Africa, and physical inactivity has been identified as a major risk factor in the now increasing burden of lifestyle diseases here.4 5 In Eldoret, Kenya, site of the current study, 82% of adults are inactive.6 The rapid urbanisation and increasing sedentary lifestyles associated with ‘the haves’ in Kenya and indeed the larger sub-Saharan Africa today are thought to be contributory.7 This further compounds the lack of exercise time associated with daily engagements among the older populace that is variously also age-compromised for sustained exercise participation. This may be associated with the observed epidemiological transition where cardiovascular disease is on the rise among adults.8–14 Unchecked, this inactivity may lead to longer-term poor health effect on individuals and undermine the healthcare systems already overwhelmed by communicable diseases.9–11 13 15–17

cipation. This may be associated with the observed epidemiological transition where cardiovascular disease is on the rise among adults.8–14 Unchecked, this inactivity may lead to longer-term poor health effect on individuals and undermine the healthcare systems already overwhelmed by communicable diseases.9–11 13 15–17 Recommendations of ways to tackle such physical inactivity are in existence. Evidence that participation in exercise or physical activity improves body composition is plenty, although data among the elderly are still scanty.2 18 19 Available exercise and physical activity guidelines are, however, inadequately followed and exercise recommendations are not being achieved in adults.2 3 6 Adherence to exercise has been poor and debate on the best exercise regime that could help achieve this has not been settled for a few decades now.2 20–26

y.2 18 19 Available exercise and physical activity guidelines are, however, inadequately followed and exercise recommendations are not being achieved in adults.2 3 6 Adherence to exercise has been poor and debate on the best exercise regime that could help achieve this has not been settled for a few decades now.2 20–26 Debate on the value and effectiveness of shorter exercise bouts on body composition is unsettled. Even where studies have focused on the benefits of accumulated short bouts of moderate-intensity exercise, benefits have not been congruent.27 28 Further, observation that 30 min bouts yield poor adherence calls for further studies on the value of the shorter (but with cumulative time equalling the existing guidelines) bouts in body composition improvement. To the best of our knowledge, there has been minimal studies on the value of exercise regimes of bouts lasting <10 min in older individuals. Where some exist, they followed their participants for a short period (8 weeks) failing to answer the question of the long-term benefits.24–26 The current study compared body composition indicators among sedentary adults following different exercise regimes performed over a 24-week period to identify any distinct differences in the regimes.

followed their participants for a short period (8 weeks) failing to answer the question of the long-term benefits.24–26 The current study compared body composition indicators among sedentary adults following different exercise regimes performed over a 24-week period to identify any distinct differences in the regimes. Materials and methods We studied healthy sedentary adults aged at least 50 years (men=27; women=26) and residents of Eldoret town, Kenya. They volunteered in response to a local print advertisement. Using the WHO Global Physical Activity Questionnaire Sedentariness, sedentary individuals were those with <600 weekly metabolic equivalent (MET)-minutes of exercise. Further, only volunteers devoid of existing physical injuries and reported cardiovascular disease or treatment were included. All measurements were performed by the same researcher throughout for consistency and reproducibility.

ary individuals were those with <600 weekly metabolic equivalent (MET)-minutes of exercise. Further, only volunteers devoid of existing physical injuries and reported cardiovascular disease or treatment were included. All measurements were performed by the same researcher throughout for consistency and reproducibility. The study was unblinded. Participants from either sex were randomly allocated into two groups per sex based on either of the two exercise prescriptions to be administered. This yielded four subgroups: (1) male short-duration bouts of exercise (MS), (2) female short-duration bouts of exercise (FS), (3) male long-duration bouts of exercise (ML) and (4) female long-duration bouts of exercise (FL). Those in the short-duration bouts of exercise, the experimental group, engaged in three daily bouts of 5–10 min of moderate-intensity jogging. Those in the long-duration bouts, the control group, involved in 30–60 min jogging bouts for 3–5 days weekly. Participants kept a record of exercise, which was objectively verified on select days using Polar Wearlink ActITrainer accelerometers (Actigraph, Pensacola, Florida, USA), monitors that detect activity and motion changes. These data were analysed weekly for confirmation that participants in the different exercise regimes had comparable MET-minutes.

exercise, which was objectively verified on select days using Polar Wearlink ActITrainer accelerometers (Actigraph, Pensacola, Florida, USA), monitors that detect activity and motion changes. These data were analysed weekly for confirmation that participants in the different exercise regimes had comparable MET-minutes. The same researcher collected data from all the participants at recruitment stage and then at 8th, 16th and 24th week of exercise involvement. For each participant, height and weight without shoes but on light clothing were measured using a stadiometer and a mechanical scale (CAMRY Mechanical scale, BR9012, Shanghai, China). Waist and hip circumferences were measured with a tape measure. Skin-fold measures were taken using callipers (Harpenden Skinfold Callipers; BATY International, England). This was done from three different body sites: chest, abdomen and thigh for men; and the triceps, suprailiac and thigh for women. From these measurements, Body Mass Index (BMI), waist:hip ratio (WHR) and waist:height ratio (WHtR) were calculated. The sum of the three skin-fold measurements were used to determine body density using published generalised equations, while the fat percentage was computed from body densities using Brozek formulae.29–32

rom these measurements, Body Mass Index (BMI), waist:hip ratio (WHR) and waist:height ratio (WHtR) were calculated. The sum of the three skin-fold measurements were used to determine body density using published generalised equations, while the fat percentage was computed from body densities using Brozek formulae.29–32 Data were analysed with STATA V.13. We used summary statistics and t-test functions, and outputs were in means and SD. Analysis was done at univariate, bivariate and multivariate levels based on sex and exercise regime adopted. Paired t-tests were conducted for exercise regime groups for each sex. Linear regressions controlling for the various predictor variables and potential confounders were further performed for the body composition outcome variables. Comparisons were evaluated at a set p value ≤0.05 for significance.

rcise regime adopted. Paired t-tests were conducted for exercise regime groups for each sex. Linear regressions controlling for the various predictor variables and potential confounders were further performed for the body composition outcome variables. Comparisons were evaluated at a set p value ≤0.05 for significance. Results At the start point of the study, male participants in the experimental arm (n=14) were aged 55.0±5.6 years while those in the control group (n=13) were aged 55.2±3.0 years. Experimental arm women (n=13) were aged 53.9±2.6 years with their control group counterparts (n=13) at 53.9±3.5 years. Sixty-four per cent (64.3%) of MS and 76.9% of ML, and 100% of FS and 76.9% of FL had acquired tertiary level of education. The rest had secondary-level training. All participants in the two experimental arms completed the 24-week exercise regime. In the control group, 61.5% of ML and 76.9% of FL completed the protocol. For BMI, 14.3% of MS, 38.5 of ML, 92.3% of FS and 69.2% of FL started as obese. Similarly, for WHtR, 64.3% of MS, 76.9% of ML, 100% of FS and 84.6.3% of FL had ratios >0.5. The MS group with baseline WHR ≥0.9 were 64.3%, with ML at 69.2%; the FS whose baseline WHR was ≥0.85 were 46.2% and their FL counterparts were 30.8%. Mean values for these variables are presented in table 1. Table 1 Baseline demographic and clinical characteristics of the participants MS ML FS FL

Results At the start point of the study, male participants in the experimental arm (n=14) were aged 55.0±5.6 years while those in the control group (n=13) were aged 55.2±3.0 years. Experimental arm women (n=13) were aged 53.9±2.6 years with their control group counterparts (n=13) at 53.9±3.5 years. Sixty-four per cent (64.3%) of MS and 76.9% of ML, and 100% of FS and 76.9% of FL had acquired tertiary level of education. The rest had secondary-level training. All participants in the two experimental arms completed the 24-week exercise regime. In the control group, 61.5% of ML and 76.9% of FL completed the protocol. For BMI, 14.3% of MS, 38.5 of ML, 92.3% of FS and 69.2% of FL started as obese. Similarly, for WHtR, 64.3% of MS, 76.9% of ML, 100% of FS and 84.6.3% of FL had ratios >0.5. The MS group with baseline WHR ≥0.9 were 64.3%, with ML at 69.2%; the FS whose baseline WHR was ≥0.85 were 46.2% and their FL counterparts were 30.8%. Mean values for these variables are presented in table 1. Table 1 Baseline demographic and clinical characteristics of the participants MS ML FS FL Age (years) 55.0±5.6 55.2±3.0 53.9±2.6 53.9±3.5 BMI (kg/m2) 25.8±4.0 28.6±4.8 33.3±4.8 32.0±5.4 WHtR 0.52±0.07 0.56±0.08 0.61±0.05 0.57±0.08 WHR 0.93±0.06 0.96±0.07 0.82±0.10 0.84±0.09 Sum of 3 skin folds (mm) 62.2±27.75 78.9±31.2 120.3±20.6 109.6±23.81 Body density 1.05±0.02 1.04±0.02 1.01±0.01 1.01±0.01 Fat % 20.7±7.3 24.9±7.9 39.8±3.6 37.4±5.1 Data presented as mean±SD.

/m2) 25.8±4.0 28.6±4.8 33.3±4.8 32.0±5.4 WHtR 0.52±0.07 0.56±0.08 0.61±0.05 0.57±0.08 WHR 0.93±0.06 0.96±0.07 0.82±0.10 0.84±0.09 Sum of 3 skin folds (mm) 62.2±27.75 78.9±31.2 120.3±20.6 109.6±23.81 Body density 1.05±0.02 1.04±0.02 1.01±0.01 1.01±0.01 Fat % 20.7±7.3 24.9±7.9 39.8±3.6 37.4±5.1 Data presented as mean±SD. BMI, Body Mass Index; FL, female long-duration bouts of exercise; FS, female short-duration bouts of exercise; ML, male long-duration bouts of exercise; MS, male short-duration bouts of exercise; WHR, waist:hip ratio; WHtR, waist:height ratio. At the end of the 24-week follow-up, MS, who had started as obese, reduced to 7.1% compared with a drop to 15.4% for ML. Similarly, obese FS dropped to 61.5% versus a drop to 30.8% for FL. The percentage MS whose WHtR was above 0.5 dropped to 42.9 compared with a drop to 30.8% for ML. For FS, it dropped to 92.3% versus a drop to 26.2% for FL. The MS group whose baseline WHR was ≥0.9 dropped to 23.1% while ML dropped to 21.4%, and for the FS participants whose baseline WHR was ≥0.85, it dropped to 15.4% with the FL counterparts dropping to 7.7%.

5 dropped to 42.9 compared with a drop to 30.8% for ML. For FS, it dropped to 92.3% versus a drop to 26.2% for FL. The MS group whose baseline WHR was ≥0.9 dropped to 23.1% while ML dropped to 21.4%, and for the FS participants whose baseline WHR was ≥0.85, it dropped to 15.4% with the FL counterparts dropping to 7.7%. Generally, all variables associated with body composition were found to decrease in measured values for both sexes. There were significant drops in body weight, BMI, WHtR, WHR and percentage body fat (all p<0.05). However, the mean change difference for each variable was not different between the two exercise regimes for both sexes as tabulated for the overall change from baseline values in table 2, and shown by the p values derived from the comparison of the mean change difference between the regimes. On controlling for sex, table 3 presents a summary of linear regressions for these variables. The change in percentage body fat for the two sexes and groups showed a similar drop as plotted in figure 1. Table 2 Body composition measurements

Generally, all variables associated with body composition were found to decrease in measured values for both sexes. There were significant drops in body weight, BMI, WHtR, WHR and percentage body fat (all p<0.05). However, the mean change difference for each variable was not different between the two exercise regimes for both sexes as tabulated for the overall change from baseline values in table 2, and shown by the p values derived from the comparison of the mean change difference between the regimes. On controlling for sex, table 3 presents a summary of linear regressions for these variables. The change in percentage body fat for the two sexes and groups showed a similar drop as plotted in figure 1. Table 2 Body composition measurements Variable Group Baseline Week 8 Week 16 Week 24 Mean change (week 24–week 0) P values Men Weight Short 76.4±14.7 75.2±14.5 74.9±14.0 72.3±14.2 −4.07±5.8 0.74 Long 82.8±16.2 82.3±15.8 81.4±15.3 79.4±15.8 −3.31±3.0 BMI Short 25.8±4.0 25.4±4.0 25.3±3.8 24.4±4.0 −1.37±2.0 0.75 Long 28.2±5.0 27.9±4.6 27.7±4.7 27.0±4.1 −1.13±1.01 WHtR Short 0.52±0.07 0.51±0.06 0.50±0.06 0.49±0.05 −0.03±0.02 0.63 Long 0.55±0.08 0.54±0.09 0.53±0.08 0.51±0.07 −0.04±0.02 WHR Short 0.93±0.06 0.92±0.06 0.90±0.05 0.87±0.05 −0.05±0.04 0.44 Long 0.96±0.08 0.94±0.08 0.91±0.08 0.89±0.07 −0.06±0.03 Fat % Short 20.7±7.3 15.6±5.9 14.2±5.5 12.4±4.8 −8.32±5.0 0.36 Long 21.6±8.5 19.3±7.5 17.3±6.3 15.2±5.8 −6.43±3.8 Women Weight Short 85.3±13.7 84.2±13.9 83.0±14.1 81.7±13.2 −3.61±1.54 0.22 Long 78.1±11.0 75.0±9.6 73.9±9.6 72.5±9.9 −5.65±5.53 BMI Short 33.3±4.8 32.9±4.8 32.4±4.9 31.9±4.6 −1.41±0.60 0.25 Long 30.5±5.1 29.3±4.6 28.9±4.9 28.4±5.1 −2.15±2.13 WHtR Short 0.605±.049 0.573±.054 0.565±.049 0.557±.050 −0.05±0.02 0.21 Long 0.55±.085 0.534±.081 0.527±.085 0.518±.082 −0.04±0.02 WHR Short 0.822±.095 0.822±.065 0.811±.066 0.801±.066 −0.02±0.05 0.84 Long 0.803±.078 0.794±.085 0.783±.081 0.785±.079 −0.02±0.04 Fat % Short 39.8±3.6 35.0±5.1 32.3±5.3 31.1±5.6 −8.69±4.41 0.67 Long 35.8±4.4 30.8±4.6 29.3±4.9 27.8±4.9 −7.96±3.34 Values are means±SD.

0.55±.085 0.534±.081 0.527±.085 0.518±.082 −0.04±0.02 WHR Short 0.822±.095 0.822±.065 0.811±.066 0.801±.066 −0.02±0.05 0.84 Long 0.803±.078 0.794±.085 0.783±.081 0.785±.079 −0.02±0.04 Fat % Short 39.8±3.6 35.0±5.1 32.3±5.3 31.1±5.6 −8.69±4.41 0.67 Long 35.8±4.4 30.8±4.6 29.3±4.9 27.8±4.9 −7.96±3.34 Values are means±SD. BMI, Body Mass Index; WHR, waist:hip ratio; WHtR, waist:height ratio. Table 3 Linear regressions for body composition (controlling for sex) Mean change, long Coefficient SE P value>|t| 95% CI Weight −0.71 1.35 0.60 −3.44 to 2.02 BMI −0.27 0.49 0.58 −1.25 to 0.71 WHtR 0.004 0.01 0.53 0.01 to 0.09 WHR −0.004 0.01 0.75 0.03 to 0.02 Fat% 1.28 1.29 0.33 1.33 to 3.89 BMI, Body Mass IndexWHR, waist:hip ratio; WHtR, waist:height ratio. Figure 1 Percentage fat change between week 0 (baseline) and week 24 (endpoint). Discussion Body composition at start point Men were overweight while women were obese (class I) at recruitment. Such BMI classes are associated with unfavourable health outcomes.33–35 The high baseline BMI of our participants suggests that they were at an increased risk for cardiovascular disease and, further, other pathologies associated with the metabolic syndrome. A few studies, however, propose that values within the 23–33 kg/m2 range have lower mortality risks among the elderly, so that the higher BMI values as observed in the current study may actually be associated with better cardiovascular disease epidemiology in this cohort.36–39 Based on this, women in the current study are especially safer.

r, propose that values within the 23–33 kg/m2 range have lower mortality risks among the elderly, so that the higher BMI values as observed in the current study may actually be associated with better cardiovascular disease epidemiology in this cohort.36–39 Based on this, women in the current study are especially safer. Recently, studies have suggested that WHtR is superior to BMI in assessing risks associated with metabolic syndrome pathologies among different races and ages.40–45 Using the proposed WHtR cut-off of 0.5, participants in the current study were at cardiovascular disease risk at recruitment.42–44 46 In fact, considering proposals this ratio is superior to BMI in evaluating health risks and outcomes among apparently healthy populations as it particularly considers fat distribution, participants in this study had a considerably high risk for metabolic syndrome.40 41 47 A comparison of the baseline WHR against WHO cut-off guidelines of ≥0.90 (men) and ≥0. 85 (women) associated with substantially increased metabolic complications showed four in every five men and two in every five women had central/abdominal obesity with mean value above their respective WHO cut-offs.48 This indicates that at the start point, men in this study had a relatively higher risk for metabolic syndrome and other health problems, and bore a higher mortality risk.35 They had twice the risk portrayed by their female counterparts based on WHR alone.

l obesity with mean value above their respective WHO cut-offs.48 This indicates that at the start point, men in this study had a relatively higher risk for metabolic syndrome and other health problems, and bore a higher mortality risk.35 They had twice the risk portrayed by their female counterparts based on WHR alone. Body composition reference data for the elderly African population with which comparison of the present study would be done are unavailable. While data abound elsewhere, these are based on non-African populations, and comparison may be affected by racial differences. Men and women in the current study had baseline percentage fat mass comparable with those found in an Indian population, although that study covered a wider age range.49 The high baseline percentage body fat for the women in the current study, making well over one-third of their body mass, and with a concurrent lower body density, is consistent with the baseline blood chemistry findings from another of our study on the same sample where the women had higher total cholesterol (TC) values.50 Based on Jackson and Pollock equations, men in this study had average baseline body fat, but their female counterparts were obese at the start point.29–31

ty, is consistent with the baseline blood chemistry findings from another of our study on the same sample where the women had higher total cholesterol (TC) values.50 Based on Jackson and Pollock equations, men in this study had average baseline body fat, but their female counterparts were obese at the start point.29–31 These baseline findings indicated a potentially poor future health outcome for this cohort as such clinical features have been shown to get worse with advancing age and adoption of sedentary lifestyles, a finding consistent with other measures associated with cardiovascular disease independently and, broadly, the conglomeration of metabolic syndrome pathologies in the East-African region.51–53 Our exercise intervention and follow-up investigation improved these health parameters for the participants. Changes in body composition Previous works have shown that 6–12 weeks of exercise is sufficient to cause body composition changes across all ages.54–56 Sustained exercise involvement is necessary to maintain any such gains, which is why the current study measurements were done 8-weekly for a follow-up period that allowed at least four measurements. Over the 24 weeks, all values on variables associated with body composition decreased in a similar manner for both men and women in the two exercise regimes. Although this drop was not to the recommended levels in some variables, it was significant nevertheless, and there was no demonstrable difference in the magnitude of change between regimes.

es on variables associated with body composition decreased in a similar manner for both men and women in the two exercise regimes. Although this drop was not to the recommended levels in some variables, it was significant nevertheless, and there was no demonstrable difference in the magnitude of change between regimes. Given that BMI drop among the elderly to normal values as provided by the international WHO classification is a huge challenge, the percentage drop for our participants with BMI >24.9 kg/m2 was a significant achievement.33 Short bouts’ regime halved the percentage of obese men with the longer-bouts’ regime dropping the same by one-third. The longer bouts for the women had a marginally higher proportion of the drop for the obese when compared with the shorter. The finding, however, still resonates with a recent suggestion that shorter-frequent exercise bouts are efficient in weight management in obese and overweight women.28 Effectively, the reduction in the BMI for both men and women in the short bouts’ regime was similar to that observed in their longer bouts’ counterparts, with the comparison of the mean change for different regimes within each sex showing no significant difference over the 24 weeks.

ent in obese and overweight women.28 Effectively, the reduction in the BMI for both men and women in the short bouts’ regime was similar to that observed in their longer bouts’ counterparts, with the comparison of the mean change for different regimes within each sex showing no significant difference over the 24 weeks. Lower values of WHtR are associated with lesser cardiovascular disease risk and indeed other pathologies related to metabolic syndrome.42 43 In this study, there was a drop in the percentage of both men and women on short bouts whose WHtR remained >0.5 cut-off at the end of the 24-week follow-up. Similarly, the mean values for those whose ratios remained >0.5 dropped. Regardless of the exercise regime and whether or not participants attained a mean value <0.5, however, all participants in the current study portrayed a drop in their WHtR. While this drop was significant from the baseline to the endpoint, a similar trend of no demonstrable difference in the rate of the change based on the exercise regime was noticed. No previous work we are aware of has compared the effect of intermittent and traditional regimes with WHtR in such a follow-up as our current work in this regard. It is safe to argue that the two exercise regimes have similar modification of central obesity and body composition at large given that no difference was noticed in this aspect for the two regimes on controlling for sex.

ermittent and traditional regimes with WHtR in such a follow-up as our current work in this regard. It is safe to argue that the two exercise regimes have similar modification of central obesity and body composition at large given that no difference was noticed in this aspect for the two regimes on controlling for sex. A similar picture was observed for WHR. There was a more than half reduction in the percentage of both men and women whose WHR remained above the WHO cut-off of ≥0.9 and ≥0.85, respectively, in both the short and the long bouts’ regimes of exercise. The difference between the two bout types was, however, indistinct.48 For participants whose values remained above the WHO threshold, short bouts were found to nonetheless decrease the absolute values in a similar manner as the long bouts. Thus, no difference was demonstrable in the manner the two exercise regimes affected WHR. The short bouts were as effective as the long ones in reducing abdominal obesity and therefore the risk associated with high WHR values.35 On controlling for sex, no differences would yet be found between the short and the long regimes in this regard, depicting their similarity in effect.

exercise regimes affected WHR. The short bouts were as effective as the long ones in reducing abdominal obesity and therefore the risk associated with high WHR values.35 On controlling for sex, no differences would yet be found between the short and the long regimes in this regard, depicting their similarity in effect. Participants in this study had a drop in the percentage body fat, regardless of the exercise regime adopted. However, only the men had a drop to within their recommended levels based on the ideal body fat percentage with widest reference.30 The women had a significant drop as well regardless of their regime of exercise but could not attain their recommended ranges.29 This could be explained by fact that at the baseline, they had a higher baseline percentage body fat so that although there was a drop, only a longer follow-up period would have helped attain the recommended ideal levels. By a direct comparison of the two exercise regimes, the shorter bouts had a slightly although statistically insignificant advantage by causing an absolute higher mean drop of the fat percentage, in both men and women. Recent works on this area have found bouts lasting <10 min to be equally effective in lowering body fat in men and women when cumulative intermittent bouts are considered.28 57 That our study found no difference between the two exercise regimes on controlling for sex in this variable further shows that the short bouts’ regime of moderate-intensity exercise is at least as good as the traditional longer bouts in the regulation of body fat across sexes.

ve intermittent bouts are considered.28 57 That our study found no difference between the two exercise regimes on controlling for sex in this variable further shows that the short bouts’ regime of moderate-intensity exercise is at least as good as the traditional longer bouts in the regulation of body fat across sexes. Limitations Generalisability of our findings may be affected by participants having been volunteers following a local print advert. Further, blinding was not done, and some lifestyle factors such as diet that would affect physical activity and cardiorespiratory fitness were not tracked given the logistical follow-up difficulties with this lengthy home-based trial, both of which may have confounded our results. That we used activity monitors on select days to verify the individually filled exercise logs may also have affected our results. Conclusions As long as the aggregated exercise time and intensity be equivalent, bouts lasting up to 10 min are as effective as the traditional longer bouts of 30–60 min in modification of body composition among sedentary Kenyan adults of 50 years and above. Contributors: KM helped in designing the protocol, data collection, analysis and writing. Both KT and NBP helped in designing the protocol and writing the manuscript.

Conclusions As long as the aggregated exercise time and intensity be equivalent, bouts lasting up to 10 min are as effective as the traditional longer bouts of 30–60 min in modification of body composition among sedentary Kenyan adults of 50 years and above. Contributors: KM helped in designing the protocol, data collection, analysis and writing. Both KT and NBP helped in designing the protocol and writing the manuscript. Funding: This research was supported by the Consortium for Advanced Research Training in Africa (CARTA). CARTA is jointly led by the African Population and Health Research Center (APHRC) and the University of the Witwatersrand and funded by the Wellcome Trust (UK) (grant no. 087547/Z/08/Z), the Department for International Development (DfID) under the Development Partnerships in Higher Education (DelPHE), the Carnegie Corporation of New York (grant no. B 8606), the Ford Foundation (grant no. 1100–0399), Google.org (grant no. 191994), Sida (grant no. 54100029), MacArthur Foundation (grant no. 10-95915-000-INP) and British Council. Competing interests: None declared. Patient consent: Obtained. Ethics approval: Joint institutional research ethics body at Moi Teaching and Referral Hospital/Moi University (approval no. IREC 0001242). Provenance and peer review: Not commissioned; externally peer reviewed.

What are the new findings? Three-dimensional multiple-object tracking (3D-MOT) processing and learning abilities are disrupted following concussion (48 hours). 3D-MOT performance is correlated to the total number of symptoms, Standardized Assessment of Concussion and Modified Balance Error Scoring System tests performance at 48 hours. 3D-MOT is useful to assess and monitor athletes’ perceptual-cognitive capacity following concussion. Perceptual-cognitive skills should be evaluated following sport-related concussion. Introduction Sport-related concussion, also referred to as mild traumatic brain injury (mTBI), is ‘induced by biomechanical forces following an impulse force transmitted to the head’ as defined by the fifth consensus statement on concussion in sport.1 Over the past two decades, a high number of sport activities were associated with an increased risk and rate of concussion.2 3 For instance, rugby is a high contact sport where the incidence of concussion varies from 0.62 to 9.05 per 1000 hours of play, with a higher risk in professional players.4–7 A recent study reported rugby with the highest incidence rate of mTBI by far in youth sports, among other contact sports such as football (7.8 times higher) or hockey (3.5 times higher).8

rt where the incidence of concussion varies from 0.62 to 9.05 per 1000 hours of play, with a higher risk in professional players.4–7 A recent study reported rugby with the highest incidence rate of mTBI by far in youth sports, among other contact sports such as football (7.8 times higher) or hockey (3.5 times higher).8 The main problem with mTBI is that patients are often considered to be ‘normal’ on neurological examinations or standard clinical testing and are discharged from follow-up programme. Even worse, <10% of sports-induced concussions result in a loss of consciousness9 which consequently results in the trauma either not being diagnosed or concealed by the player who does not wish to be sidelined. To help healthcare providers during the initial diagnosis of the concussion and management of the decision to return to play (RTP), the Concussion in Sport Group consensus statement provides updated guidelines including the Sport Concussion Assessment Tool (SCAT).1 The SCAT is a concussion evaluation tool relying on several subsets including cognition, balance and symptom evaluation assessed respectively with the Standardized Assessment of Concussion (SAC), the Modified Balance Error Scoring System (M-BESS) and a symptom checklist.10 Normative data and concussion cut-off scores have recently been published.11 12 However, the reliability and repeatability of these standard clinical tests are still debated.13–18 Therefore, more sensitive clinical tools are still required for the diagnosis and the management of sport-related concussion.

m checklist.10 Normative data and concussion cut-off scores have recently been published.11 12 However, the reliability and repeatability of these standard clinical tests are still debated.13–18 Therefore, more sensitive clinical tools are still required for the diagnosis and the management of sport-related concussion. A key performance indicator in sport is an athlete’s perceptual-cognitive capacity, which is highly solicited due to the dynamic and time constraints of their changing environment.19 This ability represents the human brain’s capacity to extract meaningful contextual information from the dynamic visual scene and is reflected on the field by anticipation and decision-making skills.20 Therefore, it appears critical to include an evaluation of ones perceptual-cognitive capacity while monitoring their return to sport activities. In this regard, the three-dimensional multiple-object tracking (3D-MOT) exercise is proposed. This context-free task involves the processing of a visual dynamic scene reflecting some of the fundamental demands required during sport (eg, keeping track of teammates, opponents and the ball) or daily life activities (eg, driving, walking in a crowd).21 22 The test requires the participant to process complex motion using selective, sustained and distributed attention as well as working memory.21 23 It has been widely used to assess the human perceptual-cognitive capacity in sport but also in other domains such as driving or flying.22 24–27 Importantly, in a cohort of 304 healthy athletic participants from 6 to 29 years, the 3D-MOT exercise has recently demonstrated shared predictive validity with other cognitive tests used for concussion assessment such as the SCAT3 and the King-Devick test.28 In addition, a preliminary study in a paediatric population with and without mTBI suggested that 3D-MOT could be beneficial for stimulating recovery and informing return to activity decisions.29 Given its relevance to perceptual-cognitive abilities in sport, its predictive power for performance in socially relevant tasks and the recent evidence for concussion assessment, 3D-MOT was tested for the evaluation and management of sport concussion in professional athletes in the context of this study. The main objective of this study was to introduce the 3D-MOT technique for concussion evaluation and return to play in professional athletes.

and the recent evidence for concussion assessment, 3D-MOT was tested for the evaluation and management of sport concussion in professional athletes in the context of this study. The main objective of this study was to introduce the 3D-MOT technique for concussion evaluation and return to play in professional athletes. Methods Participants Fifty-nine professional athletes (24.78±5.15 years, table 1) from national French leagues (Ligue Nationale de Rugby, Ligue de Football Professionnel, Ligue Nationale de Handball, Fédération Française de Judo) participated in the study. Participants were referred by the sport’s club physician following what was suggested to be a concussion on the playing field (eg, HIA 1 protocol). The concussion was confirmed by the neurologist during the first evaluation within 48 hours following the injury (table 2). Table 1 Characteristics of the professional athlete participants Athletes Number Percentage Gender Male 48 81 Female 11 19 Sports Rugby 50 85 Forwards 31 62 Backs 19 38 Judo 6 10 Soccer (Football Association) 1 2 Football 1 2 Handball 1 2 Athletic level International 39 66 National 20 34 History of injury Previous concussion 44 75 Mean number ±SEM 2.75±0.33 concussions Max/min 12/1 concussions No previous concussion 15 25 Neurologist evaluation Mean delay 48 hours—return to play 20.61±2.47 days Max/min delay 84/2 days Mean delay before return to play 26.08±2.78 days Table 2 History of the injury

20 34 History of injury Previous concussion 44 75 Mean number ±SEM 2.75±0.33 concussions Max/min 12/1 concussions No previous concussion 15 25 Neurologist evaluation Mean delay 48 hours—return to play 20.61±2.47 days Max/min delay 84/2 days Mean delay before return to play 26.08±2.78 days Table 2 History of the injury mTBI athletes Number Percentage Cause of injury Hits 54 92 Falls 5 8 Occurence of injury Game 46 78 Training 13 22 Admission Cantu grade Severe (3) 46 78 Moderate (2) 9 15 Mild (1) 2 3 Unknown 2 3 Admission AAN grade 3 14 24 2 43 73 1 2 3 LOC Yes 16 27 No 43 73 Antero-amnesia No 32 54 ≤1 min 8 14 ≥1 min 19 32 Retro-amnesia No 48 81 ≤1 min 5 8 ≥1 min 6 10 Initial symptoms Headache 47 80 Fatigue 40 68 Pressure in head 31 53 Sonophobia/photophobia 30 51 Neck pain 30 51 Difficulty concentrating/remembering 29 49 Balance problems 24 41 Drowsiness 23 39 Nausea—vomitting 22 37 Trouble falling asleep 21 36 Visual problems 19 32 Duration of symptoms Mean duration ±SEM 9.88±1.56 days Max/min duration 56 days/30 min AAN, American Academy of Neurology; LOC, Loss of consciousness; mTBI, mild traumatic brain injury.

ng/remembering 29 49 Balance problems 24 41 Drowsiness 23 39 Nausea—vomitting 22 37 Trouble falling asleep 21 36 Visual problems 19 32 Duration of symptoms Mean duration ±SEM 9.88±1.56 days Max/min duration 56 days/30 min AAN, American Academy of Neurology; LOC, Loss of consciousness; mTBI, mild traumatic brain injury. Evaluation tests Three-dimensional multiple-object tracking The commercial version of 3D-MOT called NeuroTracker (CogniSens Inc.) was used to assess the perceptual-cognitive state of the participants. The ‘CORE’ mode of the NeuroTracker was displayed on a 65’’ 3D-TV (figure 1). The exercise required participants to track four targets among eight spheres projected within a virtual cube space for 8 s (one trial), controlling for a visual angle of 45°. A detailed description of the methodology can be found in previous studies.21 30 A typical session, based on a staircase procedure, required approximately 6 min which consisted of a total of 20 trials. Three consecutive 3D-MOT sessions were completed during each evaluation for a total of 20 min. Figure 1 NeuroTracker ‘CORE’ mode: (A) presentation, (B) target identification, (C) displacement, (D) user response and (E) feedback.

Evaluation tests Three-dimensional multiple-object tracking The commercial version of 3D-MOT called NeuroTracker (CogniSens Inc.) was used to assess the perceptual-cognitive state of the participants. The ‘CORE’ mode of the NeuroTracker was displayed on a 65’’ 3D-TV (figure 1). The exercise required participants to track four targets among eight spheres projected within a virtual cube space for 8 s (one trial), controlling for a visual angle of 45°. A detailed description of the methodology can be found in previous studies.21 30 A typical session, based on a staircase procedure, required approximately 6 min which consisted of a total of 20 trials. Three consecutive 3D-MOT sessions were completed during each evaluation for a total of 20 min. Figure 1 NeuroTracker ‘CORE’ mode: (A) presentation, (B) target identification, (C) displacement, (D) user response and (E) feedback. Standardized Assessment of Concussion The SAC (SCAT 3) was used to assess athletes’ multiple mental components such as orientation, immediate memory, concentration and delayed memory.31 The SAC test contains a series of questions which take approximately 5 min to administer and is scored out of 30 points (total score). A standard neurological screening is also included to assess LOC, retrograde and post-traumatic amnesia, deficits in strength, sensation and coordination that may result from a concussion. Three equivalent alternate forms of the test were used at baseline, 48 hours and RTP to minimise practice effects from additional administration.

rological screening is also included to assess LOC, retrograde and post-traumatic amnesia, deficits in strength, sensation and coordination that may result from a concussion. Three equivalent alternate forms of the test were used at baseline, 48 hours and RTP to minimise practice effects from additional administration. Modified Balance Error Scoring System The M-BESS is a modified version of the BESS, a standard balance assessment to evaluate static balance.15 32–34 It consists of three stances: double-leg stance (feet together), single-leg stance (standing on the non-dominant leg) and a tandem stance (non-dominant foot behind the dominant foot in a heel-to-toe fashion). The number of errors in deviations from the proper stance (eg, moving hands off of iliac crests, opening eyes, a step/stumble or fall, abduction or flexion of the hip beyond 30°, lifting forefoot or heel off testing surface) were substracted from a score of 10 for each stance. Only one error was counted when multiple errors occurred at the same time. The maximum total score for each testing condition was 30 (total score). The test was conducted on a firm surface.

flexion of the hip beyond 30°, lifting forefoot or heel off testing surface) were substracted from a score of 10 for each stance. Only one error was counted when multiple errors occurred at the same time. The maximum total score for each testing condition was 30 (total score). The test was conducted on a firm surface. Procedure Previous French rugby consensus statement on concussion in sport established that a consultation with a neurologist could help return to sports under optimal conditions to prevent recurrent concussions.35 36 Moreover, evidence has shown that consultation within 4 days following the injury could help diminish the duration of post-concussion syndrome due to a more thorough follow-up and support of the concussed athlete.35–37 In this regard, athletes who potentially sustained a concussion were sent to an independent neurologist and underwent the standard procedure developed in collaboration with the medical staff of the sport clubs that participated in the study (table 3). Two main evaluations were conducted by the same neurologist, one within 48 hours following the injury (48 hours) and the second when deciding the RTP. Note that 32 professional rugby players from the present sample were also tested on SAC and M-BESS tests in pre-season by the neurologist, and therefore, a baseline score had been established. Table 3 Study protocol for concussion management adapted from ref.36

Procedure Previous French rugby consensus statement on concussion in sport established that a consultation with a neurologist could help return to sports under optimal conditions to prevent recurrent concussions.35 36 Moreover, evidence has shown that consultation within 4 days following the injury could help diminish the duration of post-concussion syndrome due to a more thorough follow-up and support of the concussed athlete.35–37 In this regard, athletes who potentially sustained a concussion were sent to an independent neurologist and underwent the standard procedure developed in collaboration with the medical staff of the sport clubs that participated in the study (table 3). Two main evaluations were conducted by the same neurologist, one within 48 hours following the injury (48 hours) and the second when deciding the RTP. Note that 32 professional rugby players from the present sample were also tested on SAC and M-BESS tests in pre-season by the neurologist, and therefore, a baseline score had been established. Table 3 Study protocol for concussion management adapted from ref.36 Steps Concussion management protocol 1 Immediate removal from the field and total rest for the patient with concussion 2 CT scan or brain magnetic resonance imaging (optional) 3 First consultation with the neurologist within 48 hours following the injury including 3a Confirmation of the diagnosis 3b Cantu classification for concussion severity 3c symptoms’ assessment, performance on SCAT and NeuroTracker evaluation 4 Return to exercising in stages once post-concussive symptoms at rest have disappeared: 4a Walking or biking 4b Individual running 4c Training without contact 5 Second consultation with the neurologist in the absence of recurrent symptoms during exercising 5a Confirmation of the recovery based on cognitive evaluation (SCAT and NeuroTracker) 5b Clear for return to play or follow-up visit SCAT, Sport Concussion Assessment Tool.

4b Individual running 4c Training without contact 5 Second consultation with the neurologist in the absence of recurrent symptoms during exercising 5a Confirmation of the recovery based on cognitive evaluation (SCAT and NeuroTracker) 5b Clear for return to play or follow-up visit SCAT, Sport Concussion Assessment Tool. Analysis Statistical analysis was performed on the main three dependant variables: 3D-MOT speed threshold, SAC and M-BESS scores. The normality of distribution in each case was controlled using asymmetry, skewness and Shapiro-Wilk tests. In case of normal distribution, analysis of variance (ANOVA) was used including pairwise comparisons using Bonferroni corrections. Post-hoc comparisons were performed using Student’s t tests. Levene and Mauchly tests were also performed to ensure that the homogeneity of variance and the sphericity, respectively, were not violated (p>0.05). As SAC and M-BESS scores resulted in being not normally distributed, non-parametric tests were employed to analyse these variables. Eta-squared (η2) or Cohen’s d values were reported to provide information about the magnitude of the effect. Statistical analyses were performed using IBM SPSS statistics V.23.

d (p>0.05). As SAC and M-BESS scores resulted in being not normally distributed, non-parametric tests were employed to analyse these variables. Eta-squared (η2) or Cohen’s d values were reported to provide information about the magnitude of the effect. Statistical analyses were performed using IBM SPSS statistics V.23. Three-dimensional multiple-object tracking A repeated-measures ANOVA was employed to compare 3D-MOT Sessions (1, 2, 3 and 4, 5, 6) and 3D-MOT Evaluations (48 hours, RTP). An additional analysis was processed including a sample of healthy professional athletes (HP) considered as normative data.22 A repeated-measure ANOVA was conducted with the within-subject factors 3D-MOT Session (1, 2, 3 and 4, 5, 6), 3D-MOT Evaluations (48 hours, RTP) and the between-subject factor Groups (mTBI, HP). Moreover, logarithmic regression functions were fitted on speed thresholds of evaluations at 48 hours (sessions 1, 2 and 3) and RTP (sessions 4, 5 and 6) for each group. A repeated-measure ANOVA was performed with the within-subject factor Evaluations (48 hours, RTP) and the between-subject factor Groups (mTBI, HP) to compare learning rates on speed thresholds.

were fitted on speed thresholds of evaluations at 48 hours (sessions 1, 2 and 3) and RTP (sessions 4, 5 and 6) for each group. A repeated-measure ANOVA was performed with the within-subject factor Evaluations (48 hours, RTP) and the between-subject factor Groups (mTBI, HP) to compare learning rates on speed thresholds. Standardized Assessment of Concussion and Modified Balance Error Scoring System First, test scores were analysed using Wilcoxon tests with the repeated factor Evaluations (48 hours, RTP). Second, another analysis was performed on a smaller sample of concussed players (n=32) because pre-season baseline measures were available. Friedman tests were employed on SAC and M-BESS scores with the repeated factor Evaluations (baseline, 48 hours, RTP). Post-hoc comparisons were processed using Wilcoxon tests. Correlations Spearman rank correlation coefficient was used to investigate the possible associations between NeuroTracker, SAC and M-BESS scores as well as the number of total symptoms, symptoms duration, delay before returning to play or Cantu grade.

Standardized Assessment of Concussion and Modified Balance Error Scoring System First, test scores were analysed using Wilcoxon tests with the repeated factor Evaluations (48 hours, RTP). Second, another analysis was performed on a smaller sample of concussed players (n=32) because pre-season baseline measures were available. Friedman tests were employed on SAC and M-BESS scores with the repeated factor Evaluations (baseline, 48 hours, RTP). Post-hoc comparisons were processed using Wilcoxon tests. Correlations Spearman rank correlation coefficient was used to investigate the possible associations between NeuroTracker, SAC and M-BESS scores as well as the number of total symptoms, symptoms duration, delay before returning to play or Cantu grade. Results Three-dimensional multiple-object tracking The analysis demonstrated a significant main effect of Evaluations (F[1,58]=86.948, p<0.001, η2=0.600) with a moderate to strong effect size (d=0.772) indicating a better performance on 3D-MOT at RTP compared with 48 hours post injury (figure 2). There was also a significant main effect of 3D-MOT Sessions (F[2,116]=10.738, p<0.001, η2=0.156). The interaction between Evaluations and 3D-MOT Sessions was non-significant (F[2,116]=0.233, p=0.792, η2=0.004), revealing that the improvement across 3D-MOT sessions was not significantly different at RTP and 48 hours. The comparison between mTBI and healthy professional athletes resulted in a strong significant difference between Groups (F[1,117]=61.350, p<0.001, η2=0.344) corroborating the impact of the injury on 3D-MOT performance. Moreover, the significant interaction between Evaluations × Sessions × Groups (F[2,234]=4.053, p=0.019, η2=0.033) confirmed that concussed athletes had lower performance in processing and learning the 3D-MOT task compared with healthy professionals who improved better and faster on the exercise between evaluations at 48 hours and RTP. Learning rates analysis demonstrated a significant interaction between Evaluations × Groups (F[2,117]=6.717, p=0.011, η2=0.054) revealing slower learning rates on 3D-MOT in concussed compared with healthy athletes. Additional analysis demonstrated a significant difference in slopes between mTBI and healthy professionals at 48 hours (t[117]=3.867, p<0.001) but not at RTP (t[117]=0.350, p=0.727). This stems from the fact that the learning function within the 48 hours of concussion is severely affected compared with healthy pros (figure 2).

al analysis demonstrated a significant difference in slopes between mTBI and healthy professionals at 48 hours (t[117]=3.867, p<0.001) but not at RTP (t[117]=0.350, p=0.727). This stems from the fact that the learning function within the 48 hours of concussion is severely affected compared with healthy pros (figure 2). Figure 2 Comparison of the three-dimensional multiple-objecttracking scores between concussed and healthy athletes throughout a first evaluation (48 hours post injury) and a second evaluation (RTP). HP, healthy professional athletes; mTBI, mild traumatic brain injury; RTP, returnto play.

al analysis demonstrated a significant difference in slopes between mTBI and healthy professionals at 48 hours (t[117]=3.867, p<0.001) but not at RTP (t[117]=0.350, p=0.727). This stems from the fact that the learning function within the 48 hours of concussion is severely affected compared with healthy pros (figure 2). Figure 2 Comparison of the three-dimensional multiple-objecttracking scores between concussed and healthy athletes throughout a first evaluation (48 hours post injury) and a second evaluation (RTP). HP, healthy professional athletes; mTBI, mild traumatic brain injury; RTP, returnto play. Standardized Assessment of Concussion and Modified Balance Error Scoring System First results demonstrated a significant difference in performance score with moderate effect size between evaluations at 48 hours and RTP in SAC (Z=−3.982, p<0.001, d=−0.635) as well as M-BESS (Z=−3.433, p=0.001, d=−0.516) tests (figure 3). However, another repeated-measure analysis within the same group of athletes and including baseline evaluation only showed a tendency towards significance in SAC scores between evaluations at baseline, 48 hours and RTP (χ2[2]=5.766, p=0.056). Post-hoc comparisons demonstrated no significant difference in SAC scores between evaluations at baseline and 48 hours (Z=−0.87, p=0.931, d<−0.1) which cannot indicate the presence of cognitive problems caused by the concussion. Significant differences with moderate effect size were seen between evaluations at 48 hours and RTP (Z=−2.489, p=0.013, d=0.550) as well as evaluations at baseline and RTP (Z=−2.249, p=0.025, d=0.536). The analysis on M-BESS scores demonstrated no significant difference between evaluations at baseline, 48 hours and RTP (χ2[2]=2.103, p=0.349).

with moderate effect size were seen between evaluations at 48 hours and RTP (Z=−2.489, p=0.013, d=0.550) as well as evaluations at baseline and RTP (Z=−2.249, p=0.025, d=0.536). The analysis on M-BESS scores demonstrated no significant difference between evaluations at baseline, 48 hours and RTP (χ2[2]=2.103, p=0.349). Figure 3 Standardized Assessment of Concussion (SAC) and Modified Balance Error Scoring System (M-BESS) scores across evaluations before (baseline), 48 hours and at return to play (RTP) time following concussion. Correlations First, there were weak but significant positive correlations between 3D-MOT and SAC scores (rs=0.282, p=0.031), between 3D-MOT and M-BESS scores (rs=0.368, p=0.004) as well as between SAC and M-BESS scores (rs=0.301, p=0.020) at 48 hours which indicated that these three measures of evaluation shared some predictive validity related to concussion (table 4). Importantly however, only 3D-MOT scores at 48 hours correlated negatively with the number of total symptoms (rs=−0.301, p=0.020). Although the correlation was weak, it suggests an association between lower performance and higher number of symptoms and vice versa. On the other hand, there were no significant correlations between symptoms and SAC or M-BESS scores at 48 hours (p<0.05). Not surprisingly, the total number of symptoms correlated positively and moderately with symptom duration (rs=0.684, p<0.001), delay before returning to play (rs=0.542, p<0.001) and Cantu grade (rs=0.364, p=0.005). Table 4 Spearman rho correlations

l as increased subject set-up time. Accordingly, these approaches are rarely employed in settings where time and/or financial constraints exist such as preseason screening of athletes performing functional movement. Additionally, these somewhat artificial laboratory conditions can cause unknown experimental artefacts.9 Recent advances and improved access to markerless motion capture technology have made the use of low-cost motion analysis tools a possibility in the clinical setting.10 However, the validity of this technology in more complex functional movements is currently unclear. The majority of studies done so far used Kinect v1, one camera and the Software Development Kit (SDK) provided by Microsoft. Researchers have evaluated the configuration during working activities,11 functional activities,12 gait in healthy population,13 14 gait in multiple-sclerosis,15 and after cerebrovascular accident,16 and during a jump test.17 More recently, single Kinect v2 was used with Microsoft SDK to test the validity during gait10 and for balance.18 A multi-Kinect v2 configuration with Microsoft SDK was tested for its validity during gait.19 20 To our knowledge, until now, no validation of a dual-camera markerless system during dynamic, advanced movements has been done. The goals of this study were to examine the validity of a markerless motion capture system using 2 Kinect v2 cameras with custom software during functional movements commonly performed during pre-season physical screening evaluation.

Correlations First, there were weak but significant positive correlations between 3D-MOT and SAC scores (rs=0.282, p=0.031), between 3D-MOT and M-BESS scores (rs=0.368, p=0.004) as well as between SAC and M-BESS scores (rs=0.301, p=0.020) at 48 hours which indicated that these three measures of evaluation shared some predictive validity related to concussion (table 4). Importantly however, only 3D-MOT scores at 48 hours correlated negatively with the number of total symptoms (rs=−0.301, p=0.020). Although the correlation was weak, it suggests an association between lower performance and higher number of symptoms and vice versa. On the other hand, there were no significant correlations between symptoms and SAC or M-BESS scores at 48 hours (p<0.05). Not surprisingly, the total number of symptoms correlated positively and moderately with symptom duration (rs=0.684, p<0.001), delay before returning to play (rs=0.542, p<0.001) and Cantu grade (rs=0.364, p=0.005). Table 4 Spearman rho correlations MOT3D- 48 hours MOT3D -RTP SAC48 hours SACRTP M-BESS48 hours M-BESSRTP TS SD RTPdelay Cantu grade History of mTBI 3D-MOT 48 hours 1 3D-MOT RTP 0.770*** 1 SAC 48 hours 0.282* 0.187 1 SAC RTP 0.18 0.227 0.343** 1 M-BESS 48 hours 0.368** 0.174 0.301* 0.106 1 M-BESS RTP 0.184 0.146 0.121 -0.049 0.506*** 1 TS –0.301* –0.159 –0.16 –0.071 –0.139 0.109 1 SD –0.089 –0.029 –0.056 –0.13 –0.17 0.01 0.684*** 1 RTP delay –0.191 –0.148 –0.1 –0.201 –0.097 0.013 0.542*** 0.591*** 1 Cantu grade –0.062 –0.061 0.011 –0.024 0.101 0.033 0.364** 0.522*** 0.472*** 1 History of mTBI –0.036 0.036 0.271* 0.108 0.288* 0.228 0.11 0.134 0.199 0.21 1 *p<0.05, **p<0.01, ***p<0.001.

39 0.109 1 SD –0.089 –0.029 –0.056 –0.13 –0.17 0.01 0.684*** 1 RTP delay –0.191 –0.148 –0.1 –0.201 –0.097 0.013 0.542*** 0.591*** 1 Cantu grade –0.062 –0.061 0.011 –0.024 0.101 0.033 0.364** 0.522*** 0.472*** 1 History of mTBI –0.036 0.036 0.271* 0.108 0.288* 0.228 0.11 0.134 0.199 0.21 1 *p<0.05, **p<0.01, ***p<0.001. D-MOT, three-dimensional multiple-objecttracking; M-BESS, Modified Balance Error Scoring System; mTBI, mild traumatic brain injury; RTP, return to play; SAC, Standardized Assessment of Concussion; SD, symptoms duration; TS, total symptoms. Discussion This study assessed the usefulness of a perceptual-cognitive 3D-MOT exercise for the evaluation and RTP guidance of sport-related concussion. The 3D-MOT was found to be a relevant perceptual marker to help monitor sport-related concussion. Importantly, 3D-MOT performance was correlated with the total number of symptoms, SAC and M-BESS scores at 48 hours. However, results on SAC and M-BESS tests demonstrated less relevance for monitoring concussion and correlation to symptoms in the present study using professional athletes.

itor sport-related concussion. Importantly, 3D-MOT performance was correlated with the total number of symptoms, SAC and M-BESS scores at 48 hours. However, results on SAC and M-BESS tests demonstrated less relevance for monitoring concussion and correlation to symptoms in the present study using professional athletes. Perceptual-cognitive impairments following sport-related concussion The main finding revealed the negative impact of mTBI on both processing and learning the perceptual-cognitive 3D-MOT task at 48 hours compared with RTP and compared with normative scores in healthy professionals. The results were congruent with preliminary works in a general population (n=485) where three sessions of 3D-MOT were shown to be sensitive to mTBI status compared with healthy individuals or individuals with a history of mTBI.38 The 3D-MOT performance was significantly enhanced at RTP compared with 48 hours post -injury which supports the usefulness of this tool for RTP guidance.

population (n=485) where three sessions of 3D-MOT were shown to be sensitive to mTBI status compared with healthy individuals or individuals with a history of mTBI.38 The 3D-MOT performance was significantly enhanced at RTP compared with 48 hours post -injury which supports the usefulness of this tool for RTP guidance. Importantly, learning rates were strongly affected at 48 hours post injury which is again consistent with recent findings in a paediatric mTBI population where processing and learning were disrupted during the first sessions of exposure to the task compared with healthy children.29 The present result obtained in professional athletes is of particular interest knowing that this population is characteristic of better processing and learning on such a task.22 It emphasises the fact that athletes’ perceptual-cognitive capacity is very limited during the early days following the injury and reinforces the fact that one should be cautious when quickly returning to activities involving dynamic visual processing following mTBI. This result is also consistent with previous evidence showing cognitive problems following concussion.39

nitive capacity is very limited during the early days following the injury and reinforces the fact that one should be cautious when quickly returning to activities involving dynamic visual processing following mTBI. This result is also consistent with previous evidence showing cognitive problems following concussion.39 Another important finding in this study revealed that 3D-MOT performance was significantly correlated to the number of total symptoms at 48 hours. Although the correlation was weak, it suggests that speed thresholds scores were higher when the symptoms were lower and vice versa. Other correlational analysis demonstrated a link between 3D-MOT, SAC and M-BESS at 48 hours following the injury. These results support previous evidence showing predictive validity of 3D-MOT with other concussion assessment cognitive tests such as the SCAT3 and King-Devick test.28 SAC and M-BESS tests are weak predictors of the RTP The results of the present study revealed weak usefulness of the SAC and BESS tests in the RTP guidance when reviewed comparatively with the baseline scores. In fact, it was difficult to determine if the difference observed between scores at 48 hours and at RTP was due to a sensitivity to the injury or simply a test–retest effect.

of the present study revealed weak usefulness of the SAC and BESS tests in the RTP guidance when reviewed comparatively with the baseline scores. In fact, it was difficult to determine if the difference observed between scores at 48 hours and at RTP was due to a sensitivity to the injury or simply a test–retest effect. This is not surprising considering that previous studies have emphasised the limitations of the M-BESS for balance deficits assessment including insufficient repeatability, poor reliability, fatigue effects, influences from musculoskeletal injuries and learning effects.13 14 Reliability of the test can range from poor to good and has been shown to be sensitive to concussion assessment only within the first three days post injury.15 40 Importantly, poor inter-rater reliability due to the subjective nature of the scoring system and important practice effects have also been pointed out.15 Recently, the BESS test has been identified with a high rate of false positives (62.5%) when baseline scores were compared with scores obtained at a later date.41 This former study also determined a high rate of false positives with the SAC test (27.1%).

stem and important practice effects have also been pointed out.15 Recently, the BESS test has been identified with a high rate of false positives (62.5%) when baseline scores were compared with scores obtained at a later date.41 This former study also determined a high rate of false positives with the SAC test (27.1%). Similarly, the usefulness of the SAC test to evaluate cognition and memory42 while monitoring concussion in professional athletes was weak considering the present results. Previously, the literature has shown that although it has good sensitivity and specificity,31 43 scores consistently returning to baseline after 48 hours or after three trials demonstrates a practice effect that may result in misinterpretation and poor management of the rehabilitation following concussion.16 44 Another study reported small to moderate effect sizes for the SAC and BESS at 24 hours following the injury that became non-significant at day 8 (SAC) and day 15 (BESS) post concussion.17 This former study also suggested that concussion symptoms were the most sensitive component of the SCAT3 compared with the cognitive (SAC) and motor (BESS) subsets. In the present study, the SAC and M-BESS tests were not correlated with the number of total symptoms. The results confirmed that SAC and M-BESS are not helpful in monitoring concussion and suggests that reliable tools other than SAC and M-BESS are still needed to help the management of sport-related concussion.

ubsets. In the present study, the SAC and M-BESS tests were not correlated with the number of total symptoms. The results confirmed that SAC and M-BESS are not helpful in monitoring concussion and suggests that reliable tools other than SAC and M-BESS are still needed to help the management of sport-related concussion. Perceptual-cognitive assessment in sport-related concussion The present study emphasised the fact that concussion assessment tools should consider some critical components requested to process dynamical environment, such as perceptual-cognitive screening. Perceptual-cognitive skills are required while processing visual dynamic scenes (eg, walking into a crowd, avoiding collision, or anticipating and decision-making in sport). In this regard, the 3D-MOT technique relies on an attentional task requiring the tracking of multiple moving targets (ie, MOT) with increasing speed which addresses the dynamic components of the living environment. Two strategies are typically employed by the subject to process the MOT task. The first is a ‘centre-looking’ strategy that consists of grouping the targets into a single object while looking closer to the centre of the object formed by the targets.45 The other is a ‘target-looking’ strategy where the subject would saccade from target to target. Other evidence has shown that participants often engaged in both strategies by switching their gaze from the centre to the targets and so on.46 Visual search strategies involved during MOT could be closely linked to those applied by sport experts during the process of extracting visual information from their action-rich environment. Recently, more evidence has correlated mTBI with impaired attentional tracking.47 48 In addition, the virtual testing environment involves stereoscopic vision which is critical when interacting in the real world.49 The technique also integrates a large visual field of view which is similar to most sporting scenarios.50 In the future, it will be essential to compare the 3D-MOT task to other perceptual-cognitive or computerised neurocognitive tests (eg, King-Devick test) to measure the efficacy of such techniques in monitoring concussion.

hnique also integrates a large visual field of view which is similar to most sporting scenarios.50 In the future, it will be essential to compare the 3D-MOT task to other perceptual-cognitive or computerised neurocognitive tests (eg, King-Devick test) to measure the efficacy of such techniques in monitoring concussion. Limitations Despite the relevance of the perceptual-cognitive 3D-MOT technique in monitoring sport-related concussion, the study presents some limitations. For instance, there were no 3D-MOT baseline tests performed on the present cohort of athletes. Although recent evidence tends to show that there is no clear advantage of using a baseline-referenced approach over a norm-referenced approach.51 Moreover, this should have a limited impact on the results since the sample of healthy professionals closely matched the injured professionals in terms of sport, league-level (eg, Ligue Nationale de Rugby), age and experimental design. Another drawback of the present methodology is that the 3D-MOT assessment took about 20 min to complete (which included three consecutive sessions for learning assessment purposes). In the future, the predictive value of a shortened 3D-MOT assessment (eg, one session) should be evaluated and compared with other perceptual-cognitive tests. However, this modification to the procedure would not include an assessment of learning rates which has clearly shown to be an advantage over other techniques in the present and previous concussion studies.29

s practice.7 Therefore, it is an important aim of clinicians, sports professionals and researchers.8 9 Several screening tools for injuries to the ACL of the knee, hamstring, groin and ankle could be recommended for use in the field.10 One of these screening tools for injuries is the Functional Movement Screen (FMS).11 The FMS is an assessment tool which identifies the quality of movement and requires both balance and stability.11 It is popular in many fitness and rehabilitation settings.12 Seven basic exercises are scored from 0 to 3 according to the grading criteria: deep squat, inline lunge, hurdle step, shoulder mobility, active straight leg raise, trunk stability push-up and rotary stability.11 The FMS test could identify movement ability, and then suggest exercises based on the dysfunctions and limitations detected13 14 which positively influence strength and flexibility.15 The FMS has good intrarater (intraclass correlation coefficient (ICC)=0.74–0.8) and inter-rater (ICC=0.9–0.97) reliability.16 The relevance of this test is increasing due to its proposed injury-predictive ability in different sports17 18 and at different ages.19 20

ned 3D-MOT assessment (eg, one session) should be evaluated and compared with other perceptual-cognitive tests. However, this modification to the procedure would not include an assessment of learning rates which has clearly shown to be an advantage over other techniques in the present and previous concussion studies.29 Conclusion For the first time, this study presents the unique role of a perceptual-cognitive 3D-MOT exercise to monitor sport-related concussion. This technique seems to possess some of the requirements needed to appropriately respond to everyday world and sport specific demands. Future studies should compare its efficacy to other perceptual-cognitive techniques, evaluate if reducing 3D-MOT assessment time can preserve its potential towards monitoring concussion and should also test its training value for post-concussion rehabilitation. The authors thank Stephan Roux for his precious help and guidance with the NeuroTracker system. They also like to thank the rugby clubs medical staff, in particular Dr Alexis Savigny (Stade Français Rugby), Dr Sylvain Blanchard and Dr Yohann Bohu (Racing-Metro 92), Dr Regis Boxeley (PSG Handball), Dr Philippe Kuentz (AS Monaco) and Dr Jean-Marc Sène (Équipe de France de Judo). They thank Robyn Lahiji for her help in editing the manuscript. Special thanks to the athlete patients. Contributors: JFC, JF and FM participated to the study design. JFC and FM conducted the tests. TR conducted the analysis and wrote the manuscript with JF.

The authors thank Stephan Roux for his precious help and guidance with the NeuroTracker system. They also like to thank the rugby clubs medical staff, in particular Dr Alexis Savigny (Stade Français Rugby), Dr Sylvain Blanchard and Dr Yohann Bohu (Racing-Metro 92), Dr Regis Boxeley (PSG Handball), Dr Philippe Kuentz (AS Monaco) and Dr Jean-Marc Sène (Équipe de France de Judo). They thank Robyn Lahiji for her help in editing the manuscript. Special thanks to the athlete patients. Contributors: JFC, JF and FM participated to the study design. JFC and FM conducted the tests. TR conducted the analysis and wrote the manuscript with JF. Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Competing interests: JF is director of Faubert Lab at the University of Montreal and he is the Chief Science Officer of CogniSens Inc. who produces the commercial version of the NeuroTracker used in this study. In this capacity, he holds shares in the company. Patient consent: Obtained. Ethics approval: Comité d’éthique de la recherche en santé de l’université de Montréal. Provenance and peer review: Not commissioned; externally peer reviewed. Data sharing statement: Data are available by contacting the corresponding author.

What are the new findings? There has not been much recent focus on the potential importance of neck tension causing concussion. The mass of a helmet added to a head can result in increased neck tension forces in impacts primarily to the chest. Compared with direct helmet-to-helmet collisions causing concussion, these impacts primarily to the chest result in lower head accelerations and angular velocities. Neck tension or strain along the axis of the upper cervical spinal cord and brainstem is a possible mechanism of brain injury and should be considered in the design and evaluation of helmets. How might it impact on clinical practice in the near future? These findings could help identify a mechanism of concussion in sport. These data could be used by helmet manufacturers to develop protective equipment to reduce the incidence of concussion in sport and also methods to treat injured athletes. Background The 2012 consensus statement on concussion in sport included the statement that ‘concussion may be caused by a direct blow to the head, face, neck or elsewhere on the body with an impulsive force transmitted to the head.’1 There are few studies on concussion with primary impact to the chest and the study of this type of collision may shed some light on a mechanism of injury.

ded the statement that ‘concussion may be caused by a direct blow to the head, face, neck or elsewhere on the body with an impulsive force transmitted to the head.’1 There are few studies on concussion with primary impact to the chest and the study of this type of collision may shed some light on a mechanism of injury. In animal testing, Friede2 3 studied the mechanics of concussion by evaluating the signs and neuropathology in the upper spinal cord and brainstem of cats in response to a distraction load in a non-impact condition. He concluded that craniocervical distraction (tension) and flexion are the most important factors in concussion. Ommaya et al 4 produced signs of cerebral concussion, haemorrhages on and contusions over the surface of the brain and upper cervical cord by rotational displacement of the head on the neck, without direct head impact. They concluded that multiple mechanisms are involved in cerebral concussion, among them are rotational acceleration of the head, flexion-extension-tension of the neck and intracranial pressure gradients. Hodgson5 concluded that relative movement at the craniocervical junction may be an important factor in whether there is loss of consciousness in impacts resulting in inertial loading of the head.

ong them are rotational acceleration of the head, flexion-extension-tension of the neck and intracranial pressure gradients. Hodgson5 concluded that relative movement at the craniocervical junction may be an important factor in whether there is loss of consciousness in impacts resulting in inertial loading of the head. In the human, sled testing conducted by Col John P Stapp6 resulted in the loss of consciousness of one volunteer at a peak sled deceleration of 38 g with an onset rate of 1370 g/s without impact to the head. Hutchinson7 conducted a video analysis of 174 concussion-causing hits in the NHL. Twenty per cent of these injuries had a primary shoulder-to-chest contact, but less than 5% had no secondary head contact. King et al 8 used a discrete parameter model of the head and neck to study the response of the neck of pilots who ditch in the ocean and fail to eject before the jet aircraft sank. Results showed that, with the added weight of a helmet, one of the reasons for the pilots failing to eject was cord concussion due in part to upper cervical cord stretch during the combined vertical acceleration and forward deceleration of the aircraft. The computed head linear and angular accelerations were below concussive levels. Ommaya,9 Hodgson5 and Jadischke et al 10 also indicated that the mass of the helmet aggravates the potential for injury by adding bending, axial and shear loads at the craniocervical junction.

tion and forward deceleration of the aircraft. The computed head linear and angular accelerations were below concussive levels. Ommaya,9 Hodgson5 and Jadischke et al 10 also indicated that the mass of the helmet aggravates the potential for injury by adding bending, axial and shear loads at the craniocervical junction. The objective of this study was to assess the biomechanical responses from impact to the chest in American football. This study was completed in three phases. First, impact testing was conducted to the chest of a stationary anthropomorphic test device (ATD), both helmeted and unhelmeted. Second, a case study of two NFL game collisions was conducted to estimate the biomechanical forces in real-life collisions resulting in concussion. In these cases, the primary impact was to the chest, and the player experienced a concussion with a delayed return to play. Third, a finite element (FE) study was conducted using the head and neck from the Global Human Body Model Consortium (GHBMC) Average Male model to estimate the strain along the axis of the cervical spinal cord and brainstem under combined tensile and flexion loading conditions.

concussion with a delayed return to play. Third, a finite element (FE) study was conducted using the head and neck from the Global Human Body Model Consortium (GHBMC) Average Male model to estimate the strain along the axis of the cervical spinal cord and brainstem under combined tensile and flexion loading conditions. Materials and methods Test series 1: general impact testing Impact tests were conducted with head, neck and upper torso of a Hybrid III 50th percentile ATD struck at the centre of gravity of the chest. The 14 kg pelvis of the ATD was replaced with a 13 kg steel base. The ATD lumbar spine was vertical and the ATD was placed on a height-adjustable table. The tests were conducted by striking the stationary ATD with a 45 kg impactor with a 38.1 mm (1.5 inch) thick deformable vinyl nitrile end cap at impact speeds of 5–10 m/s. This end cap is used commonly in helmet-to-helmet testing to simulate a helmeted player.11 The impacts were repeated back-to-back with the ATD helmeted and unhelmeted. In the 9 and 10 m/s impacts, the facemask was removed to prevent it from striking the impactor ram. Details regarding the ATD instrumentation and filtering are found in the online supplementary figure S1. The ratio of the biomechanical responses from the ATD in the helmeted condition versus the unhelmeted condition was compared using a one-sided Student’s t-test. 10.1136/bmjsem-2018-000362.supp1Supplementary data

Materials and methods Test series 1: general impact testing Impact tests were conducted with head, neck and upper torso of a Hybrid III 50th percentile ATD struck at the centre of gravity of the chest. The 14 kg pelvis of the ATD was replaced with a 13 kg steel base. The ATD lumbar spine was vertical and the ATD was placed on a height-adjustable table. The tests were conducted by striking the stationary ATD with a 45 kg impactor with a 38.1 mm (1.5 inch) thick deformable vinyl nitrile end cap at impact speeds of 5–10 m/s. This end cap is used commonly in helmet-to-helmet testing to simulate a helmeted player.11 The impacts were repeated back-to-back with the ATD helmeted and unhelmeted. In the 9 and 10 m/s impacts, the facemask was removed to prevent it from striking the impactor ram. Details regarding the ATD instrumentation and filtering are found in the online supplementary figure S1. The ratio of the biomechanical responses from the ATD in the helmeted condition versus the unhelmeted condition was compared using a one-sided Student’s t-test. 10.1136/bmjsem-2018-000362.supp1Supplementary data Test series 2: laboratory reconstructions Game video was analysed to assess the heading angles, torso angles and closing speeds of two cases in the NFL with primary impact to the chest that resulted in concussion. The independent analyses from multiple camera views resulted in the estimated helmet location overlaying each other when plotted in three-dimensional (3D) model of the playing field. The scaled model of the playing field, distance travelled by the player’s helmet and the time between frames were used to estimate the preimpact speed and heading angle of each of the players. The players’ speeds were also checked using a two-dimensional analysis of the markings on the playing field. The helmet delta-V was calculated graphically (vector subtraction) using the average speed and 3D heading angle for 0.1 s prior to impact and 0.1 s after impact.

t speed and heading angle of each of the players. The players’ speeds were also checked using a two-dimensional analysis of the markings on the playing field. The helmet delta-V was calculated graphically (vector subtraction) using the average speed and 3D heading angle for 0.1 s prior to impact and 0.1 s after impact. In the laboratory, the upper bodies of two Hybrid III 50th percentile ATDs were used to represent the football players involved in these collisions. The ATDs consisted of the Hybrid III head, neck, upper torso, shoulders, standing lumbar spine and pelvis and were ballasted using a weight vest to represent the player’s upper body mass. A nylon stocking was placed over the Hybrid III headforms to reduce the friction at the head-helmet interface and to provide a more realistic response of the helmet on the headform. This is consistent with NFL helmet testing.12 13 A large-sized American football helmet weighing 2.15 kg was fitted onto the ATD headform representing the player struck in the chest, and a large-sized American football helmet weighing 1.85 kg was fitted onto the striking ATD headform. The brow pads were positioned 2.54 cm (1 inch) above the top of the nose. The chin strap was attached so that it fit snugly over the Hybrid III chin. Data acquisition and instrumentation for each of the ATDs were similar to that described in test series 1. Additional information is provided in the online supplementary figures S2 and S3.

positioned 2.54 cm (1 inch) above the top of the nose. The chin strap was attached so that it fit snugly over the Hybrid III chin. Data acquisition and instrumentation for each of the ATDs were similar to that described in test series 1. Additional information is provided in the online supplementary figures S2 and S3. FE modelling The head and neck were segmented from the whole GHBMC 50th percentile male model at the first thoracic vertebra along with all relevant musculature and ligaments. Validation of the head and neck was previously completed by others14 15 using cadaveric and volunteer experimental data. In the present study, the model was not used to assess tissue-level strains in the brainstem and spinal cord directly because there was no specific validation related to the brainstem and upper cervical spinal cord discussed in the literature. Rather, the kinematics of the vertebrae and skull were studied to assess the craniocervical stretch in the vertebral column in response to independently applied tensile (distraction) loading and forward flexion. These were the primary biomechanical responses of players in these impacts to the chest. The elongation of the cervical column was measured using nodes defined on the anterior, left, right and posterior sides of each cervical vertebra, and the location and orientation of the skull was monitored by tracking its centre of gravity.

the primary biomechanical responses of players in these impacts to the chest. The elongation of the cervical column was measured using nodes defined on the anterior, left, right and posterior sides of each cervical vertebra, and the location and orientation of the skull was monitored by tracking its centre of gravity. The average strain in the cervical spine was assessed at the level of C1–C5 since the literature16 17 has shown there to be caudal (downward) displacement of the spinal cord relative to the spinal column in this level and cephalad (upward) displacement of the spinal cord below this level indicating that stretch of the spinal cord (above C5) and brainstem occurs. A spinal cord coupling ratio of 0.6518 19 was applied to the vertebral column strain to estimate the strain along the axis of the spinal cord and brainstem relative to vertebral body strain. The kinematics predicted by the FE simulations were compared with existing human volunteer17 20 and cadaveric studies.21 22 Additional information is provided in the online supplementary figure S3. 10.1136/bmjsem-2018-000362.supp3Supplementary data

The average strain in the cervical spine was assessed at the level of C1–C5 since the literature16 17 has shown there to be caudal (downward) displacement of the spinal cord relative to the spinal column in this level and cephalad (upward) displacement of the spinal cord below this level indicating that stretch of the spinal cord (above C5) and brainstem occurs. A spinal cord coupling ratio of 0.6518 19 was applied to the vertebral column strain to estimate the strain along the axis of the spinal cord and brainstem relative to vertebral body strain. The kinematics predicted by the FE simulations were compared with existing human volunteer17 20 and cadaveric studies.21 22 Additional information is provided in the online supplementary figure S3. 10.1136/bmjsem-2018-000362.supp3Supplementary data Results Test series 1: general impact testing The primary ATD response to chest impact was in the sagittal plane. Table 1 illustrates the biomechanical responses for various closing velocities. There was a 40% ± 10% (t=9.84, p<0.001) increase in upper neck tensile forces when compared with unhelmeted impacts of equal severity. There was also an increase of 8% ± 3% (t=7.267, p<0.001) in head flexion angle. There was a reduction in head displacement of 18% ± 4% and a reduction of rotational velocity of 18% ± 6%. The head motion lagged behind the torso motion to a greater extent in the helmeted impacts. The helmet mass (2.15 kg) increased the effective mass of the headform (4.54 kg +2.15 kg) by 47% when compared with the unhelmeted headform (4.54 kg). This resulted in significantly greater neck forces and moments when compared with the unhelmeted impacts. High-speed video of a 10 m/s chest impact is illustrated in figure 1. Additional information is provided in the online supplementary video 1.

g +2.15 kg) by 47% when compared with the unhelmeted headform (4.54 kg). This resulted in significantly greater neck forces and moments when compared with the unhelmeted impacts. High-speed video of a 10 m/s chest impact is illustrated in figure 1. Additional information is provided in the online supplementary video 1. 10.1136/bmjsem-2018-000362.supp5Supplementary video Table 1 Summary of ATD data from test series 1

g +2.15 kg) by 47% when compared with the unhelmeted headform (4.54 kg). This resulted in significantly greater neck forces and moments when compared with the unhelmeted impacts. High-speed video of a 10 m/s chest impact is illustrated in figure 1. Additional information is provided in the online supplementary video 1. 10.1136/bmjsem-2018-000362.supp5Supplementary video Table 1 Summary of ATD data from test series 1 Test ID Helmet Impact Impact Head kinematics Chest Upper neck Translation Rotation Acceleration Forces Moment Speed Force Acceleration Velocity Displacement Velocity Rotation (m/s) (N) resultant x z x z y y res Shear Tension Flexion (g) (m/s) (m/s) (m) (m) (rad/s) (°) (g) (N) (N) (Nm) M8B No 5 4187 12.3 −6.19 3.11 −0.34 0.20 −13.9 −24.8 10.7 −326 387 15.0 M8A Yes 5 4026 11.1 −4.73 2.62 −0.27 0.20 −12.3 −25.7 10.0 −421 522 16.0 M8C No 6 4931 15.1 −7.36 3.81 −0.40 0.24 −17.8 −30.5 12.9 −384 484 17.7 M8D Yes 6 4931 13.7 −6.16 3.21 −0.32 0.24 −15.6 −33.1 12.5 −481 676 19.0 M8F No 7 5736 17.9 −8.20 4.92 −0.44 0.31 −20.9 −34.8 15.3 −448 602 19.4 M8E Yes 7 5655 16.1 −6.96 4.14 −0.36 0.30 −16.4 −38.0 14.8 −587 868 19.8 M8G No 8 6681 21.2 −9.46 6.01 −0.50 0.37 −25.2 −40.1 18.0 −533 696 22.0 M8H Yes 8 6621 19.2 −7.60 4.74 −0.39 0.37 −18.5 −42.5 17.8 −682 1094 22.3 M8K No 9 8029 25.7 −10.33 7.60 −0.56 0.47 −29.3 −44.5 21.4 −650 868 23.8 M8J Yes* 9 8130 25.2 −8.94 6.35 −0.48 0.44 −23.4 −48.6 21.0 −747 1195 26.9 M8L No 10 11 449 35.6 −11.44 9.04 −0.63 0.55 −36.0 −48.2 31.2 −816 1322 34.7 M8M Yes* 10 11 771 37.2 −10.16 7.29 −0.55 0.50 −29.3 −53.7 32.1 −955 1687 39.7 Average (helmet/unhelmeted) 0.94 0.83 0.83 0.82 0.97 0.82 1.08 0.98 1.24 1.40 1.07 SD (helmet/unhelmeted) 0.06 0.04 0.02 0.04 0.03 0.06 0.03 0.03 0.07 0.10 0.06 t 2.495 9.040 18.733 12.110 2.031 7.817 −7.177 1.657 −8.900 −9.839 −3.322 P (one tail, 0.05) 0.027 <0.001 <0.001 <0.001 0.049 <0.001 <0.001 0.079 <0.001 <0.001 <0.001 ATD, anthropomorphic test device.

3 0.82 0.97 0.82 1.08 0.98 1.24 1.40 1.07 SD (helmet/unhelmeted) 0.06 0.04 0.02 0.04 0.03 0.06 0.03 0.03 0.07 0.10 0.06 t 2.495 9.040 18.733 12.110 2.031 7.817 −7.177 1.657 −8.900 −9.839 −3.322 P (one tail, 0.05) 0.027 <0.001 <0.001 <0.001 0.049 <0.001 <0.001 0.079 <0.001 <0.001 <0.001 ATD, anthropomorphic test device. *Facemask removed to prevent contact with impactor ram. Figure 1 Comparison of a helmeted versus unhelmeted chest impact. The unhelmeted impact is overlaid onto the helmeted impact. Test series 2: laboratory reconstructions The closing velocities for case A and case B were 12.6 and 9.8 m/s, respectively. The reconstruction data from the struck ATD are summarised in table 2. A comparison of the postimpact kinematics of case A is illustrated in figure 2. A comparison of these laboratory reconstructions to the test series 1 results for the helmeted and unhelmeted ATDs is illustrated in figure 3. Figure 2 Comparison of the game impact for case A. Figure 3 Neck tension and head acceleration versus chest acceleration for test series 1 and the laboratory reconstruction of case A and case B. Table 2 Summary of ATD data representing the struck and injured player from the laboratory reconstructions Case Closing Location Kinematics Upper neck kinetics Speed Translational acceleration Translational Δvelocity Translational displacement Rotational velocity Rotation Forces Moment (m/s) x z resultant x z x z y y Shear Tension Flexion (g) (g) (g) (m/s) (m/s) (m) (m) (rad/s) ( ° ) (N) (N)

Table 2 Summary of ATD data representing the struck and injured player from the laboratory reconstructions Case Closing Location Kinematics Upper neck kinetics Speed Translational acceleration Translational Δvelocity Translational displacement Rotational velocity Rotation Forces Moment (m/s) x z resultant x z x z y y Shear Tension Flexion (g) (g) (g) (m/s) (m/s) (m) (m) (rad/s) ( ° ) (N) (N) (Nm) A 12.6 Head −38.2 49.2 50.9 −12.20 7.89 −0.73 0.55 −41.1 −51.0 −1074 2646 49.3 Chest −36.6 18.0 41.8 5.1 1.6 −0.42 0.08 8.2 10.1 B 9.8 Head −15.6 −15.0 18.7 −6.10 −3.40 −0.16 −0.14 −26.5 −46.0 −799 1342 36.0 Chest −19.2 9.7 19.0 3.8 0.9 −0.17 0.03 4.6 3.0 ATD, anthropomorphic test device. FE modelling FE modelling indicated that the strain in the vertebral column increased linearly with head flexion or tensile loading; however, it varied along the length of the cervical spine. The average strain in the vertebral column in flexion was 0.21% strain/degree of head rotation and 4.6% strain/1000 N of tensile load. The maximum strain in the vertebral column was predicted to occur in the upper cervical spine (C1–C2) and was 0.28% strain/degree of head rotation and 6.5% strain/1000 N of tensile load for flexion and tension, respectively (table 3). Table 3 Estimates of strain in the spinal canal (FE study) and CNS (FE study×0.65 spinal cord coupling ratio) for various neck tension loads and flexion angles of rotation. The estimates were applied to the laboratory reconstruction data to estimate the strain in the CNS of these injured players

FE modelling FE modelling indicated that the strain in the vertebral column increased linearly with head flexion or tensile loading; however, it varied along the length of the cervical spine. The average strain in the vertebral column in flexion was 0.21% strain/degree of head rotation and 4.6% strain/1000 N of tensile load. The maximum strain in the vertebral column was predicted to occur in the upper cervical spine (C1–C2) and was 0.28% strain/degree of head rotation and 6.5% strain/1000 N of tensile load for flexion and tension, respectively (table 3). Table 3 Estimates of strain in the spinal canal (FE study) and CNS (FE study×0.65 spinal cord coupling ratio) for various neck tension loads and flexion angles of rotation. The estimates were applied to the laboratory reconstruction data to estimate the strain in the CNS of these injured players Case Tension Flexion Sum Force Average strain C1–C5 Maximum strain C1–C2 Rotation Average strain C1–C5 Maximum strain C1–C2 Average strain C1–C5 Maximum strain C1–C2 (N) Spinal canal CNS Spinal canal CNS Spinal canal CNS Spinal canal CNS CNS CNS (%) (%) (%) (%) (°) (%) (%) (%) (%) (%) (%) FE study 500 2.4 1.6 2.1 1.3 35 7.4 4.8 9.8 6.4 – – FE study 1500 7.1 4.6 9.2 6.0 45 9.5 6.1 12.6 8.2 – – FE study 2500 10.5 6.8 16.9 11.0 55 11.6 7.5 15.4 10.0 – – Case A 2646 11.5 7.5 17.3 11.2 51 10.7 7.0 14.3 9.3 14.4 20.5 Case B 1342 5.9 3.8 8.7 5.7 46 9.7 6.3 12.9 8.4 10.1 14.1 CNS, central nervous system; FE, finite element.

2.4 1.6 2.1 1.3 35 7.4 4.8 9.8 6.4 – – FE study 1500 7.1 4.6 9.2 6.0 45 9.5 6.1 12.6 8.2 – – FE study 2500 10.5 6.8 16.9 11.0 55 11.6 7.5 15.4 10.0 – – Case A 2646 11.5 7.5 17.3 11.2 51 10.7 7.0 14.3 9.3 14.4 20.5 Case B 1342 5.9 3.8 8.7 5.7 46 9.7 6.3 12.9 8.4 10.1 14.1 CNS, central nervous system; FE, finite element. A spinal cord coupling ratio of 0.6518 19 was used to estimate the central nervous system (CNS) strain relative to vertebral body strain. A maximum strain along the axis of the spinal cord and brainstem for a flexion angle of 55° was predicted to be 7.5%–10.0%. These estimates using FE modelling were comparable to in vivo volunteer data which measured a maximum strain in the spinal cord of approximately 10.2% at a 55° flexion angle.17 The average strain along the axis of the cervical spinal cord and brainstem was predicted to be 1.6%, 4.6% and 6.8% for neck tension loads of 500, 1500 and 2500 N, respectively. The peak strains in the upper cervical spine (C1–C2) were predicted to be 1.3%, 6.0% and 11.0%, respectively.

10.2% at a 55° flexion angle.17 The average strain along the axis of the cervical spinal cord and brainstem was predicted to be 1.6%, 4.6% and 6.8% for neck tension loads of 500, 1500 and 2500 N, respectively. The peak strains in the upper cervical spine (C1–C2) were predicted to be 1.3%, 6.0% and 11.0%, respectively. Discussion The laboratory reconstruction data for case A and case B, as well as the FE data, were used to estimate the strain along the axis of the spinal cord and brainstem in these concussed NFL players. The estimated strain was 13.0%–18.6% in case A and 8.7%–12.2% in case B due to combined tension and forward flexion. This range represents the estimated average strain (low) to the maximum strain (high). The estimated total strain accounts for the time-varying sum of the strains due to tension and flexion. The laboratory reconstruction and FE results indicate that the axonal strain in the spinal cord and brainstem (table 3) exceeds the levels that have been documented to cause changes in functional and structural response in spinal nerve roots when stretched in tension at varying strain rates.23 The strains are similar to those documented in in vivo tests with primates which resulted in functional changes in the spinal cord as well as changes in heart rate and respiration.24

umented to cause changes in functional and structural response in spinal nerve roots when stretched in tension at varying strain rates.23 The strains are similar to those documented in in vivo tests with primates which resulted in functional changes in the spinal cord as well as changes in heart rate and respiration.24 While translational acceleration, rotational velocity and rotational acceleration of the head have been discussed as biomechanical correlates with concussion, craniocervical stretch resulting from tension and flexion in the upper cervical spine has also been reported to be an important factor in concussion.2 3 5 Neck tension and head flexion have each been shown to result in strain of the upper cervical spinal cord and the brainstem. In a study of 183 human cadavers, Breig25 found that tension generated in the spinal cord can be transmitted from the spinal cord to the brainstem. The largest elongation occurred in the medulla oblongata, and no elongation was apparent superior to the midbrain. The reticular formation of the brainstem controls heart rate, respiration and consciousness. The loss of consciousness in one of the players in this case study is consistent with injury to the brainstem.2 3 5 24 26

st elongation occurred in the medulla oblongata, and no elongation was apparent superior to the midbrain. The reticular formation of the brainstem controls heart rate, respiration and consciousness. The loss of consciousness in one of the players in this case study is consistent with injury to the brainstem.2 3 5 24 26 In case A and case B, the struck Hybrid III ATD underwent 51° and 46° of head flexion, respectively. The forward flexion of the head was combined with neck tension as a result of the inertial loading of the head and helmet. The flexion of the head is within normal range of motion of the human for quasistatic movement; however, in the human17 18 27–29 and primate,16 imaging studies have reported elongation of the cervical spinal canal and cord in flexion. The FE modelling results, combined with a coupling ratio, estimate strains in the CNS of 9.3% and 8.4% as a result of forward flexion, in cases A and B, respectively. These strains, by themselves, are within the range that has been documented for the human17 as part of the normal range of quasistatic flexion.

flexion. The FE modelling results, combined with a coupling ratio, estimate strains in the CNS of 9.3% and 8.4% as a result of forward flexion, in cases A and B, respectively. These strains, by themselves, are within the range that has been documented for the human17 as part of the normal range of quasistatic flexion. The neck tensions in this case study (case A=2646 N, case B=1342 N) are greater than the neck tensions found in volunteer studies30–32 and greater than uninjured NFL players12 (670±405 N). The neck tensions are similar to those reported by Viano et al 33 in their reconstruction of struck and injured players in the NFL (1704±432 N) and are less than the neck tensions resulting in failure of the cervical spine in musculoskeletal cadaveric studies.21 22 34 35 The tensile loads correspond to approximately 3.27 (case A) and 1.10 (case B) times the player’s body weight. This tensile load must be supported by the soft tissues of the neck. In these cases, the struck players did not appear to have the opportunity to ready themselves for the impact. From our FE study, and by applying a coupling ratio, the maximum strain in the CNS due to neck tension alone was estimated to be 11.2% and 5.7% for cases A and B, respectively.

by the soft tissues of the neck. In these cases, the struck players did not appear to have the opportunity to ready themselves for the impact. From our FE study, and by applying a coupling ratio, the maximum strain in the CNS due to neck tension alone was estimated to be 11.2% and 5.7% for cases A and B, respectively. The time-varying strain along the axis of the spinal cord and brainstem due to combined tension and flexion for case A and case B was on the order of 13.0% to 18.6% and 8.7% to 12.2%, respectively. The data presented in this case study support the mechanism of injury discussed by Friede2 3 and Hodgson and Thomas36 and Hodgson5 who have indicated that strains in the upper spinal cord and brainstem are important factors in concussion. The brainstem’s relation to concussion is further supported by the early work of Denny-Brown and Russell26 who produced concussion signs in the decerebrate animal. The addition of the helmet to the ATD headform in test series 1 resulted in an increase in neck tension and forward flexion of the head. The neck tension increased by 40% and forward flexion increased by 8% as a result of the added helmet mass and inertia. Others4 5 have indicated that the mass of the helmet added to the head can increase the strain at the craniocervical junction. If, through further research, neck tension is found to be a biomechanical predictor of concussion, helmet and equipment manufacturers could use this information to optimise helmet performance and also to develop alternative methods of protecting against concussion.

an increase the strain at the craniocervical junction. If, through further research, neck tension is found to be a biomechanical predictor of concussion, helmet and equipment manufacturers could use this information to optimise helmet performance and also to develop alternative methods of protecting against concussion. There are several limitations of this study that should be noted. The case study is limited since only two cases were reconstructed. However, the reconstruction of these two cases may help shed some light on a potential mechanism of concussion since they investigated impacts that were primarily to the chest. This case study was performed using the Hybrid III ATD in a laboratory test environment. The Hybrid III headform and neck provide a biofidelic response in the loading condition analysed; however, it is not human, therefore tissue-level strains could not be directly assessed. The data acquired were used in conjunction with FE modelling to estimate the stretch in the upper cervical spine and a coupling ratio was applied to assess the strain in the CNS under these loading conditions. There are limited data that discuss spinal cord coupling ratio. However, the relative length of the spinal cord and brainstem when compared with C1–C5 also supports a coupling ratio of approximately 0.65 (online supplementary figure S4). 10.1136/bmjsem-2018-000362.supp4Supplementary data

which positively influence strength and flexibility.15 The FMS has good intrarater (intraclass correlation coefficient (ICC)=0.74–0.8) and inter-rater (ICC=0.9–0.97) reliability.16 The relevance of this test is increasing due to its proposed injury-predictive ability in different sports17 18 and at different ages.19 20 Kiesel et al 9 were the first to note that subjects who had a total score of less than 14 out of 21 points were more likely to suffer an injury during a sports season. From this study to date, some authors have investigated in prospective studies if the FMS is associated with injuries using this cut-off.21 22 Meanwhile, other researchers performed a sensitivity and specificity analysis and then chose the cut-off for their study.23 After an analysis of all the studies in the systemic review, 14 out of 24 chose a cut-off of 14/21; hence, the use of this cut-off may still be questionable.23

There are several limitations of this study that should be noted. The case study is limited since only two cases were reconstructed. However, the reconstruction of these two cases may help shed some light on a potential mechanism of concussion since they investigated impacts that were primarily to the chest. This case study was performed using the Hybrid III ATD in a laboratory test environment. The Hybrid III headform and neck provide a biofidelic response in the loading condition analysed; however, it is not human, therefore tissue-level strains could not be directly assessed. The data acquired were used in conjunction with FE modelling to estimate the stretch in the upper cervical spine and a coupling ratio was applied to assess the strain in the CNS under these loading conditions. There are limited data that discuss spinal cord coupling ratio. However, the relative length of the spinal cord and brainstem when compared with C1–C5 also supports a coupling ratio of approximately 0.65 (online supplementary figure S4). 10.1136/bmjsem-2018-000362.supp4Supplementary data In case A, on impact, the torso’s forward motion stopped and the player’s head and helmet continued to move and flex forward. This indicates that the primary contact was to the chest of the struck player. Due to the severity of this collision, the bottom of the struck player’s facemask appears to have made contact with the top of the defending player’s helmet as his head flexed forward. This was also simulated in our laboratory reconstruction of the collision and appears to have reduced the forward flexion of the head and increased the neck tension in comparison to test series 1.

ck player’s facemask appears to have made contact with the top of the defending player’s helmet as his head flexed forward. This was also simulated in our laboratory reconstruction of the collision and appears to have reduced the forward flexion of the head and increased the neck tension in comparison to test series 1. In this study, only strain in the neck has been considered from an impact to the chest. The rate of loading indicates the strain rate effect may be a factor in concussion and deserves further attention in the future. Additional limitations are discussed in the online supplementary video 1. 10.1136/bmjsem-2018-000362.supp2Supplementary data The authors would like to thank the staff of McCarthy Engineering Inc. for their assistance in various aspects of testing and preparation of this paper. Specifically, Mr. John Herbert, who constructed test fixtures and assisted in carrying out the testing; Mr. Brian Gilbert, who conducted the video analysis for Case A and Case B; Mrs. Pam Savage for assistance in preparation of the manuscript; and Mrs. Ann Kristo for her proofreading. Thank you to Dr. King Yang, Dr. Xin Jin, and Mr. Rohit Raut, of Wayne State University, for not only providing the GHBMC finite element model for this research but also for its segmentation.

for Case A and Case B; Mrs. Pam Savage for assistance in preparation of the manuscript; and Mrs. Ann Kristo for her proofreading. Thank you to Dr. King Yang, Dr. Xin Jin, and Mr. Rohit Raut, of Wayne State University, for not only providing the GHBMC finite element model for this research but also for its segmentation. Contributors: RJ, DCV and JRM designed the test matrix for this research. RJ generated documentation for Wayne State University Institutional Review Board, conducted the laboratory impact testing, analysed and summarised the data and conducted finite element simulations. RJ, DCV and AIK reviewed the data and conducted the analysis. All authors contributed to this manuscript, its review, and approved the final version. Funding: Funding for this research has been received from McCarthy Engineering and from the Institute for Injury Research. Competing interests: None declared. Patient consent: Not required. Ethics approval: This study has been reviewed by the Wayne State University Institutional Review Board and it was determined that an IRB review was not required (HPR determination number 2017-51). Provenance and peer review: Not commissioned; externally peer reviewed. Data statement: No additional data are available.

Highlights Markerless (multiple Kinect v2) camera kinematic analysis a relatively inexpensive clinically useful tool. Excellent shin range flexion-extension in all tests. Excellent results for peak angles regarding knee flexion. Poor results for rotations. Introduction Precompetition medical assessment of athletes commonly includes assessment of movement quality while athletes perform standardised testing procedures. Depending on the particular sport’s performance requirements and injury patterns, different test batteries are employed in an effort to identify at-risk individuals to target for tailored interventions. The quantification of these movement assessment tests is typically performed with simple visual analysis and rating,1 or occasionally using video recording and later 2-dimensional analysis. Such approaches have shown limited accuracy in estimating injury likelihood, and it has been suggested that this could be attributed, in part, to the reduced objectivity of these approaches in comparison to 3-dimensional kinematic analyses. In the context of football (soccer), commonly performed functional tests include: Single Leg Squat (SLS) assessing motion in frontal plane knee motion;2–4 Single Leg Jump (SLJ)5 6 and Counter-movement Jump (CMJ) for lower limb power estimation.7 Additionally, a modification of the CMJ with the athlete landing on one leg instead of two (Modified Counter Movement Jump (MCMJ)) has been recommended as being more sport-specific.8

essing motion in frontal plane knee motion;2–4 Single Leg Jump (SLJ)5 6 and Counter-movement Jump (CMJ) for lower limb power estimation.7 Additionally, a modification of the CMJ with the athlete landing on one leg instead of two (Modified Counter Movement Jump (MCMJ)) has been recommended as being more sport-specific.8 Marker-based motion capture is currently considered the reference method for kinematic analyses. These approaches, however, require expensive equipment, significant operator training and analysis time as well as increased subject set-up time. Accordingly, these approaches are rarely employed in settings where time and/or financial constraints exist such as preseason screening of athletes performing functional movement. Additionally, these somewhat artificial laboratory conditions can cause unknown experimental artefacts.9

Recent advances and improved access to markerless motion capture technology have made the use of low-cost motion analysis tools a possibility in the clinical setting.10 However, the validity of this technology in more complex functional movements is currently unclear. The majority of studies done so far used Kinect v1, one camera and the Software Development Kit (SDK) provided by Microsoft. Researchers have evaluated the configuration during working activities,11 functional activities,12 gait in healthy population,13 14 gait in multiple-sclerosis,15 and after cerebrovascular accident,16 and during a jump test.17 More recently, single Kinect v2 was used with Microsoft SDK to test the validity during gait10 and for balance.18 A multi-Kinect v2 configuration with Microsoft SDK was tested for its validity during gait.19 20 To our knowledge, until now, no validation of a dual-camera markerless system during dynamic, advanced movements has been done. The goals of this study were to examine the validity of a markerless motion capture system using 2 Kinect v2 cameras with custom software during functional movements commonly performed during pre-season physical screening evaluation. Method Participants Thirty-four pain-free male professional football players participated in the study(table 1). All athletes had no previous lower extremity surgery and no current injury. We followed Fleiss’ recommendation21 for reliability studies after considering previous work in the area.22 23 This study was approved by the ethical review board (Institutional Review Board E2013000003) and all participants provided written informed consent as required by the Helsinki declaration.

d no current injury. We followed Fleiss’ recommendation21 for reliability studies after considering previous work in the area.22 23 This study was approved by the ethical review board (Institutional Review Board E2013000003) and all participants provided written informed consent as required by the Helsinki declaration. Table 1 Participant information Participants Mean±SD Male (n=34) Age (years) 26.63 (±4.23) Weight (kg) 73.58 (±11.44) Height (cm) 176.01 (±8.01) BMI (kg/m2) 23.62 (±2.25) BMI, body mass index. Materials Marker trajectories were measured with a 13-camera motion capture system (BTS-SMART 1000, BTS S.p.A., Italy) sampling at 250 Hz. Depth and colour image data were simultaneously recorded with 2 Kinect v2 cameras at 30 Hz (Kinect for Windows, Microsoft, Redmond, Washington, USA) and iPi Recorder (iPi Soft, Moscow, Russia). Kinect cameras were placed one in front and one to the left side of the capture area (in between the 2 Optojump sensors) at an angle of 70° between them (figure 1). Figure 1 Biomechanics lab setup with marker based motion capture system and dual Kinect V2 configuration. Kinect camera placement indicated by the two straight arrows, and athletes were tested within the area of the Opto-Jump sensors (curved arrow). The athletes were tested such that they were facing one Kinect camera, with the other one to their left.

th marker based motion capture system and dual Kinect V2 configuration. Kinect camera placement indicated by the two straight arrows, and athletes were tested within the area of the Opto-Jump sensors (curved arrow). The athletes were tested such that they were facing one Kinect camera, with the other one to their left. Data collection After warm up for a minimum of 5 min, 31 markers were placed using clusters for thigh and shin and on anatomical landmarks according to standard marker protocol (figure 2). Participants stood in the capture area and performed three repetitions each of a SLS, a SLJ and a MCMJ, in the same order. Each trial was captured from BTS and Kinect cameras simultaneously. Figure 2 Marker placement. Data analysis Kinematic data from the Kinect cameras were processed using biomechanics add-on software (iPi Soft, Moscow, Russia). Marker trajectories from the marker-based system were processed using the SMART Analyser application (BTS S.p.A., Italy). For this analysis, the trajectories were adjusted to iPi Software such that comparison of the extracted data could be made. Marker based data were filtered using Butterworth Low Pass Filter at 6 Hz and resampled at 30 Hz. Kinematic data from both systems were extracted in Euler angles (rotation sequence XYZ), in degrees, relative to the ground for thigh, shin and foot. For each trial, time synchronisation was performed manually by identifying the starting point of each trial as the moment of heel raise from the floor and the end point as the moment of heel contact to the floor.

acted in Euler angles (rotation sequence XYZ), in degrees, relative to the ground for thigh, shin and foot. For each trial, time synchronisation was performed manually by identifying the starting point of each trial as the moment of heel raise from the floor and the end point as the moment of heel contact to the floor. The range of movement at the thigh, shin and foot and peak angles at the thigh and shin were averaged across three cycles in each exercise and used for subsequent analysis. Range of movement was calculated for each joint of interest as the difference between maximum and minimum angles for each cycle. Mean subject-based values for each test were then determined. Note that these were calculated independently for both the markerless (Kinect) and marker-based (BTS) equipment.

equent analysis. Range of movement was calculated for each joint of interest as the difference between maximum and minimum angles for each cycle. Mean subject-based values for each test were then determined. Note that these were calculated independently for both the markerless (Kinect) and marker-based (BTS) equipment. Statistical analysis A two-way mixed analysis of variance (ANOVA) (absolute agreement) was performed to assess the reliability and the variability of the measurements. Between measurement agreement was assessed using intraclass correlation coefficients (ICC) (2, k; absolute agreement). Because the ICC does not allow us to fully appreciate the magnitude of within-subject variance, we also calculated the SE of measurement (SEM) and the minimal detectable change (MDC).24 SEM represents the within-subject reliability of the measure and, consequently, the reliability of the measure.24 The SEM was determined as √MSE, where MSE=mean square error from the ANOVA table. The MDC represents the threshold over which an individual change can be considered meaningful when taking into account the variability associated with both the measurement technique and the experimental sample and was calculated using the equation MDC=1.96 × √2×SEM Finally, to better understand system agreement of the peak joint angles, the 95% limits of agreement and the bias were calculated using Bland-Altman analysis.25 The bias represents the average difference in peak joint angle between the systems while the limits of agreement are the bias ±SD. Significance level was set at p<0.05. Correlation coefficients were interpreted as follows: less than 0.40 as poor, between 0.40 and 0.59 as fair, between 0.60 and 0.74 as good, between 0.75 and 1.00 as excellent.26 All analyses were performed using SPSS 23.0 (SPSS Statistics for Windows, V.23.0. Armonk, New York, USA: IBM).

icance level was set at p<0.05. Correlation coefficients were interpreted as follows: less than 0.40 as poor, between 0.40 and 0.59 as fair, between 0.60 and 0.74 as good, between 0.75 and 1.00 as excellent.26 All analyses were performed using SPSS 23.0 (SPSS Statistics for Windows, V.23.0. Armonk, New York, USA: IBM). Results Mean(±SD), absolute agreement ICC, SEM and MDC values for angles and ranges of motion are provided in tables 2–4. Table 2 Range of angles, averaged over the three cycles during the Single Leg Squat test for BTS (considered as reference standard) and IPI software-Kinect configuration

icance level was set at p<0.05. Correlation coefficients were interpreted as follows: less than 0.40 as poor, between 0.40 and 0.59 as fair, between 0.60 and 0.74 as good, between 0.75 and 1.00 as excellent.26 All analyses were performed using SPSS 23.0 (SPSS Statistics for Windows, V.23.0. Armonk, New York, USA: IBM). Results Mean(±SD), absolute agreement ICC, SEM and MDC values for angles and ranges of motion are provided in tables 2–4. Table 2 Range of angles, averaged over the three cycles during the Single Leg Squat test for BTS (considered as reference standard) and IPI software-Kinect configuration Test Segment Movement System Mean (SD) 95% CI ICC(2,k) (95% CI) P value SEM MDC (deg) Lower (deg) Upper (deg) (deg) (deg) SLS_L (n=34) THIGH Flexion/Extension BTS 42.5 (6.5) 40.2 44.7 0.532 (–0.21 to 0.84) 0.000 3.8 10.5 iPi 52.6 (7.8) 49.8 55.3 Rotation BTS 15.1 (3.2) 14.0 16.2 0.312 (–0.31 to 0.65) 0.069 3.5 9.6 iPi 13.5 (4.5) 12.0 15.1 Abduction/Adduction BTS 13.2 (4.7) 11.6 14.9 0.775 (0.55 to 0.89) 0.791 3.4 9.5 iPi 13.5 (6.4) 11.2 15.7 SHIN Flexion/Extension BTS 26.3 (5.3) 24.4 28.1 0.886 (0.73 to 0.95) 0.006 2.4 6.7 iPi 28.0 (6.2) 25.8 30.2 Rotation BTS 21.0 (4.4) 19.5 22.5 0.126 (–0.31 to 0.47) 0.000 4.7 12.9 iPi 15.5 (5.4) 13.6 17.4 Abduction/Adduction BTS 15.5 (6.7) 13.1 17.8 0.718 (0.17 to 0.88) 0.000 3.4 9.4 iPi 11.5 (5.2) 9.7 13.3 FOOT Flexion/Extension BTS 4.7 (4.0) 3.3 6.0 0.324 (–0.20 to 0.68) 0.000 3.1 8.6 iPi 12.2 (4.6) 10.6 13.8 Rotation BTS 2.4 (1.8) 1.7 3.0 0.084 (–0.09 to 0.32) 0.000 2.6 7.2 iPi 11.5 (3.8) 10.2 12.8 Abduction/Adduction BTS 11.2 (4.1) 9.8 12.7 0.867 (0.73 to 0.93) 0.711 2.1 6.0 iPi 11.4 (4.7) 9.8 13.1 SLS_R (n=34) THIGH Flexion/Extension BTS 43.0 (10.6) 39.3 46.7 0.604 (–0.13 to 0.88) 0.000 3.9 10.7 iPi 57.1 (9.6) 53.7 60.4 Rotation BTS 13.9 (4.3) 12.4 15.4 0.515 (0.01 to 0.76) 0.891 3.4 9.4 iPi 14.0 (4.0) 12.6 15.4 Abduction/Adduction BTS 16.1 (7.6) 13.4 18.7 0.758 (0.52 to 0.88) 0.134 4.1 11.4 iPi 17.6 (5.6) 15.7 19.6 SHIN Flexion/Extension BTS 26.3 (5.7) 24.3 28.3 0.854 (0.70 to 0.93) 0.051 3.1 8.5 iPi 27.8 (6.9) 25.4 30.2 Rotation BTS 20.2 (4.4) 18.7 21.8 −0.210 (–0.76 to 0.26) 0.000 4.7 13.0 iPi 14.1 (4.1) 12.6 15.5 Abduction/Adduction BTS 16.5 (9.3) 13.3 19.7 0.319 (–0.18 to 0.63) 0.000 6.2 17.1 iPi 10.1 (4.1) 8.7 11.5 FOOT Flexion/Extension BTS 5.7 (2.7) 4.7 6.6 0.079 (–0.28 to 0.41) 0.000 4.0 11.0 iPi 11.1 (5.2) 9.3 12.9 Rotation BTS 2.2 (1.9) 1.5 2.9 0.102 (–0.10 to 0.36) 0.000 1.9 5.2 iPi 8.4 (2.4) 7.6 9.2 Abduction/Adduction BTS 10.2 (2.9) 9.2 11.2 0.707 (0.42 to 0.85) 0.070 2.2 6.0 iPi 11.2 (3.6) 9.9 12.5 ICC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2.

2.9 Rotation BTS 2.2 (1.9) 1.5 2.9 0.102 (–0.10 to 0.36) 0.000 1.9 5.2 iPi 8.4 (2.4) 7.6 9.2 Abduction/Adduction BTS 10.2 (2.9) 9.2 11.2 0.707 (0.42 to 0.85) 0.070 2.2 6.0 iPi 11.2 (3.6) 9.9 12.5 ICC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLS_L, Single Leg Squat Left; SLS_R, Single Leg Squat Right. Table 3 Range of angles, averaged over the three cycles during the Single Leg Jump test for BTS (considered as reference standard) and IPI software-Kinect configuration

2.9 Rotation BTS 2.2 (1.9) 1.5 2.9 0.102 (–0.10 to 0.36) 0.000 1.9 5.2 iPi 8.4 (2.4) 7.6 9.2 Abduction/Adduction BTS 10.2 (2.9) 9.2 11.2 0.707 (0.42 to 0.85) 0.070 2.2 6.0 iPi 11.2 (3.6) 9.9 12.5 ICC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLS_L, Single Leg Squat Left; SLS_R, Single Leg Squat Right. Table 3 Range of angles, averaged over the three cycles during the Single Leg Jump test for BTS (considered as reference standard) and IPI software-Kinect configuration Test Segment Movement System Mean (SD) 95% CI ICC(2,k) (95% CI) P value SEM MDC (deg) Lower (deg) Upper (deg) (deg) (deg) SLJ_L (n=31) THIGH Flexion/Extension BTS 39.0 (9.2) 35.6 42.3 0.491 (–0.17 to 0.82) 0.000 4.4 12.3 iPi 52.5 (7.9) 49.6 55.4 Rotation BTS 21.6 (5.9) 19.5 23.8 0.622 (0.21 to 0.82) 0.870 4.7 13.0 iPi 21.8 (6.6) 19.4 24.3 Abduction/Adduction BTS 22.2 (8.6) 19.0 25.3 0.462 (–0.09 to 0.74) 0.216 5.7 15.9 iPi 20.3 (4.5) 18.7 22.0 SHIN Flexion/Extension BTS 28.9 (7.5) 26.1 31.6 0.816 (0.62 to 0.91) 0.084 3.6 9.9 iPi 27.2 (5.4) 25.2 29.2 Rotation BTS 29.0 (4.9) 27.2 30.8 −0.260 (–0.94 to 0.28) 0.000 6.0 16.5 iPi 22.2 (6.0) 20.0 24.3 Abduction/Adduction BTS 19.2 (4.4) 17.6 20.8 0.529 (–0.20 to 0.84) 0.000 2.2 6.1 iPi 13.2 (4.0) 11.8 14.7 FOOT Flexion/Extension BTS 41.6 (10.0) 38.0 45.3 0.487 (–0.13 to 0.82) 0.000 4.2 11.7 iPi 26.4 (7.9) 23.5 29.3 Rotation BTS 15.0 (4.3) 13.4 16.6 0.461 (–0.04 to 0.73) 0.010 4.3 11.9 iPi 18.0 (6.1) 15.8 20.2 Abduction/Adduction BTS 23.3 (4.5) 21.6 25.0 0.213 (–0.18 to 0.55) 0.000 3.6 10.1 iPi 15.1 (4.3) 13.5 16.7 SLJ_R (n=33) THIGH Flexion/Extension BTS 39.3 (9.1) 36.0 42.5 0.658 (–0.19 to 0.88) 0.000 4.5 12.3 iPi 47.4 (7.5) 44.7 50.1 Rotation BTS 21.2 (5.3) 19.3 23.1 0.563 (0.15 to 0.78) 0.063 3.8 10.5 iPi 19.4 (4.5) 17.8 21.0 Abduction/Adduction BTS 17.1 (5.3) 15.2 19.0 0.725 (0.44 to 0.86) 0.036 3.2 8.8 iPi 15.4 (4.6) 13.8 17.0 SHIN Flexion/Extension BTS 27.9 (8.9) 24.8 31.1 0.926 (0.84 to 0.96) 0.022 3.0 8.4 iPi 26.1 (8.3) 23.2 29.1 Rotation BTS 29.2 (5.3) 27.3 31.1 0.297 (–0.21 to 0.63) 0.000 4.1 11.3 iPi 21.6 (4.9) 19.9 23.4 Abduction/Adduction BTS 16.7 (4.6) 15.1 18.3 0.780 (–0.16 to 0.94) 0.000 1.9 5.2 iPi 12.8 (5.0) 11.0 14.6 FOOT Flexion/Extension BTS 43.1 (9.9) 39.6 46.7 0.443 (–0.10 to 0.80) 0.000 4.4 12.1 iPi 24.5 (9.7) 21.0 27.9 Rotation BTS 14.4 (4.3) 12.9 15.9 0.421 (–0.11 to 0.71) 0.000 3.3 9.2 iPi 18.0 (4.0) 16.6 19.5 Abduction/Adduction BTS 21.7 (4.9) 19.9 23.4 0.289 (–0.20 to 0.64) 0.000 3.4 9.5 iPi 13.5 (4.2) 12.0 15.0 CC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2.

tion BTS 14.4 (4.3) 12.9 15.9 0.421 (–0.11 to 0.71) 0.000 3.3 9.2 iPi 18.0 (4.0) 16.6 19.5 Abduction/Adduction BTS 21.7 (4.9) 19.9 23.4 0.289 (–0.20 to 0.64) 0.000 3.4 9.5 iPi 13.5 (4.2) 12.0 15.0 CC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLJ_L, Single Leg Jump Left; SLJ_R, Single Leg Jump Right. Table 4 Range of angles, averaged over the three cycles during the modified counter movement test for BTS (considered as reference standard) and IPI software-Kinect configuration

y a high risk of suffering from injury in healthy people. For this reason, the aim of this systematic review is to assess the association between the FMS score and subsequent injuries in healthy people by applying a cut-off of 14/21. Furthermore, it aims to perform a meta-analysis of the data from the selected studies. METHODS Design This systematic review and meta-analysis is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.27 Search strategy The electronic databases Medline, PubMed, PsycINFO, SPORTDiscus, Cumulative Index of Nursing and Allied Health Literature, Scopus, Embase, and Physiotherapy Evidence Database were searched and the last search was conducted on 1 June 2019. No hand searches were conducted. The search was limited to studies published in the last 10 years (2007–2019) because the seminal study by Kiesel et al 9 was published in 2007. The keywords searched in the titles and abstracts were ‘functional movement screen’, ‘risk’, ‘injury’ and ‘predict’. The following search strategy was developed by the authors: functional movement screen* AND (predict OR prediction OR risk OR injury).

P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLJ_L, Single Leg Jump Left; SLJ_R, Single Leg Jump Right. Table 4 Range of angles, averaged over the three cycles during the modified counter movement test for BTS (considered as reference standard) and IPI software-Kinect configuration Test Segment Movement System Mean (SD) 95% CI ICC(2,k) (95% CI) P value SEM MDC (deg) Lower (deg) Upper (deg) (deg) (deg) MCMJ_L (n=33) THIGH Flexion/Extension BTS 59.8 (10.4) 56.1 63.5 0.851 (0.07 to 0.95) 0.000 3.5 9.8 iPi 65.7 (9.3) 62.4 69.0 Rotation BTS 26.1 (7.3) 23.5 28.7 0.518 (0.00 to 0.77) 0.000 4.7 13.0 iPi 21.4 (5.1) 19.6 23.2 Abduction/Adduction BTS 26.9 (5.8) 24.8 28.9 0.644 (0.23 to 0.83) 0.002 5.6 15.5 iPi 31.6 (10.2) 28.0 35.2 SHIN Flexion/Extension BTS 35.1 (5.8) 33.1 37.2 0.801 (0.60 to 0.90) 0.578 3.4 9.5 iPi 34.7 (6.1) 32.5 36.8 Rotation BTS 35.7 (9.5) 32.3 39.1 0.571 (0.05 to 0.80) 0.000 5.8 16.1 iPi 29.7 (6.7) 27.3 32.1 Abduction/Adduction BTS 19.6 (5.2) 17.8 21.4 0.493 (−0.05 to 0.75) 0.000 3.2 8.8 iPi 16.1 (2.9) 15.1 17.1 FOOT Flexion/Extension BTS 47.8 (10.8) 44.0 51.7 0.384 (−0.13 to 0.75) 0.000 5.2 14.3 iPi 28.7 (8.1) 25.9 31.6 Rotation BTS 22.4 (10.4) 18.7 26.1 0.799 (0.59 to 0.90) 0.053 4.8 13.2 iPi 24.8 (6.0) 22.6 26.9 Abduction/Adduction BTS 23.9 (5.6) 21.9 25.9 0.550 (−0.23 to 0.83) 0.000 2.8 7.7 iPi 18.3 (3.5) 17.0 19.5 MCMJ_R (n=34) THIGH Flexion/Extension BTS 56.8 (11.2) 53.0 60.7 0.765 (−0.19 to 0.93) 0.000 3.9 10.9 iPi 66.3 (10.0) 62.8 69.8 Rotation BTS 23.3 (4.7) 21.7 25.0 −0.280 (-1.55 to 0.36) 0.253 5.8 15.9 iPi 21.7 (6.1) 19.6 23.8 Abduction/Adduction BTS 28.5 (5.8) 26.4 30.5 0.657 (0.33 to 0.83) 0.059 5.0 13.8 iPi 26.1 (8.2) 23.2 29.0 SHIN Flexion/Extension BTS 33.7 (5.2) 31.9 35.5 0.856 (0.71 to 0.93) 0.887 3.0 8.2 iPi 33.8 (6.4) 31.5 36.0 Rotation BTS 35.4 (6.3) 33.2 37.6 0.049 (−0.50 to 0.45) 0.001 6.0 16.5 iPi 30.0 (5.8) 28.0 32.1 Abduction/Adduction BTS 21.9 (7.0) 19.5 24.3 −0.030 (−0.50 to 0.36) 0.000 5.4 14.9 iPi 15.8 (2.9) 14.8 16.8 FOOT Flexion/Extension BTS 46.0 (8.6) 43.0 49.0 0.202 (−0.13 to 0.55) 0.000 5.9 16.3 iPi 26.5 (7.1) 24.0 29.0 Rotation BTS 20.9 (4.0) 19.5 22.2 0.277 (−0.22 to 0.60) 0.000 4.5 12.5 iPi 25.5 (6.1) 23.4 27.6 Abduction/Adduction BTS 25.4 (4.9) 23.7 27.1 0.238 (−0.17 to 0.58) 0.000 4.5 12.5 iPi 17.1 (3.3) 15.9 18.2 ICC(2,k), intraclass correlation coefficient (absolute agreement); MCMJ_L, Modified Counter-Movement Jump Left; MCMJ_R, Modified Counter-Movement Jump Right; MDC, minimal dete

000 4.5 12.5 iPi 25.5 (6.1) 23.4 27.6 Abduction/Adduction BTS 25.4 (4.9) 23.7 27.1 0.238 (−0.17 to 0.58) 0.000 4.5 12.5 iPi 17.1 (3.3) 15.9 18.2 ICC(2,k), intraclass correlation coefficient (absolute agreement); MCMJ_L, Modified Counter-Movement Jump Left; MCMJ_R, Modified Counter-Movement Jump Right; MDC, minimal dete ctable change calculated as SEMx1.96x√2. P<0.05; SEM, standard error of the measure calculated as the square root of the residual mean square. Our results showed excellent between system agreement for shin movement in flexion/extension in all three tests, for both legs. Additionally, during the SLS test excellent agreement was found for thigh and foot adduction/abduction motion. Results for peak angles are shown in tables 5–7. Between systems agreement was excellent for knee flexion in all tests for both legs. Table 5 Peak angles averaged over the three cycles during the Single Leg Squat test for BTS (considered as reference standard) and IPI software-Kinect configuration

Our results showed excellent between system agreement for shin movement in flexion/extension in all three tests, for both legs. Additionally, during the SLS test excellent agreement was found for thigh and foot adduction/abduction motion. Results for peak angles are shown in tables 5–7. Between systems agreement was excellent for knee flexion in all tests for both legs. Table 5 Peak angles averaged over the three cycles during the Single Leg Squat test for BTS (considered as reference standard) and IPI software-Kinect configuration Test Movement System Mean (SD) 95% CI ICC(2,k) (95% CI) P value SEM MDC (deg) Lower (deg) Upper (deg) (deg) (deg) SLS_L (n=34) Hip flexion BTS −30.3 (17.5) −36.4 −24.2 0.896 (–0.09 to 0.97) 0.000 4.1 11.3 iPi −40.6 (18.0) −46.9 −34.3 Hip adduction BTS −18.6 (5.5) −20.6 −16.7 0.749 (0.49 to 0.88) 0.029 3.3 9.2 iPi −20.5 (5.4) −22.4 −18.6 Knee flexion BTS 30.0 (8.6) 27.0 33.0 0.932 (0.71 to 0.98) 0.000 2.5 6.8 iPi 32.8 (8.5) 29.8 35.7 Knee adduction BTS −17.5 (6.1) −19.6 −15.3 0.830 (0.57 to 0.92) 0.001 2.6 7.3 iPi −15.3 (4.5) −16.8 −13.7 SLS_R (n=34) Hip flexion BTS −49.1 (14.8) −54.3 −44.0 0.767 (–0.12 to 0.94) 0.000 3.9 10.9 iPi −63.9 (14.7) −69.0 −58.7 Hip adduction BTS 22.8 (4.7) 21.1 24.4 0.684 (–0.13 to 0.89) 0.000 2.6 7.2 iPi 27.1 (5.1) 25.3 28.9 Knee flexion BTS 25.9 (9.8) 22.5 29.4 0.947 (0.83 to 0.98) 0.001 2.8 7.8 iPi 28.5 (10.7) 24.8 32.3 Knee adduction BTS 18.3 (5.8) 16.3 20.3 0.665 (0.08 to 0.86) 0.000 3.0 8.2 iPi 14.7 (3.8) 13.4 16.1 ICC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLS_L, Single Leg Squat Left; SLS_R, Single Leg Squat Right.

0.86) 0.000 3.0 8.2 iPi 14.7 (3.8) 13.4 16.1 ICC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLS_L, Single Leg Squat Left; SLS_R, Single Leg Squat Right. Table 6 Peak angles averaged over the three cycles during the Single Leg Jump test for BTS (considered the gold standard) and IPI software-Kinect configuration

0.86) 0.000 3.0 8.2 iPi 14.7 (3.8) 13.4 16.1 ICC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLS_L, Single Leg Squat Left; SLS_R, Single Leg Squat Right. Table 6 Peak angles averaged over the three cycles during the Single Leg Jump test for BTS (considered the gold standard) and IPI software-Kinect configuration Test Movement System Mean (SD) 95% CI ICC(2,k) (95% CI) P value SEM MDC (deg) Lower (deg) Upper (deg) (deg) (deg) SLJ_L (n=31) Hip flexion BTS −27.1 (16.4) −33.2 −21.1 0.890 (−0.05 to 0.97) 0.000 4.3 12.0 iPi −36.6 (16.6) −42.7 −30.5 Hip adduction BTS −19.0 (5.5) −21.0 −17.0 0.826 (0.64 to 0.92) 0.744 2.9 7.9 iPi −18.7 (4.9) −20.5 −16.9 Knee flexion BTS 31.4 (10.6) 27.5 35.3 0.951 (0.90 to 0.98) 0.694 3.0 8.3 iPi 31.7 (8.7) 28.5 34.9 Knee adduction BTS −16.5 (4.9) −18.3 −14.7 0.883 (0.32 to 0.96) 0.000 1.7 4.7 iPi −14.0 (5.0) −15.9 −12.2 SLJ_R (n=33) Hip flexion BTS −46.8 (11.8) −51.0 −42.6 0.649 (−0.19 to 0.89) 0.000 5.0 14.0 iPi −60.7 (11.6) −64.8 −56.6 Hip adduction BTS 23.1 (5.5) 21.2 25.0 0.901 (−0.04 to 0.97) 0.002 3.7 10.2 iPi 26.1 (5.9) 24.0 28.2 Knee flexion BTS 27.5 (11.2) 23.6 31.5 0.956 (0.91 to 0.98) 0.513 3.1 8.6 iPi 28.1 (9.9) 24.5 31.6 Knee adduction BTS 18.8 (4.9) 17.0 20.5 0.726 (–0.21 to 0.92) 0.000 2.0 5.5 iPi 14.3 (4.5) 12.6 15.9 ICC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLJ_L, Single Leg Jump Left; SLJ_R, Single Leg Jump Right.

o 0.92) 0.000 2.0 5.5 iPi 14.3 (4.5) 12.6 15.9 ICC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLJ_L, Single Leg Jump Left; SLJ_R, Single Leg Jump Right. Table 7 Peak angles averaged over the three cycles during the modified counter movement test for BTS (considered the gold standard) and IPI software-Kinect configuration

o 0.92) 0.000 2.0 5.5 iPi 14.3 (4.5) 12.6 15.9 ICC(2,k), intraclass correlation coefficient (absolute agreement); MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square; SLJ_L, Single Leg Jump Left; SLJ_R, Single Leg Jump Right. Table 7 Peak angles averaged over the three cycles during the modified counter movement test for BTS (considered the gold standard) and IPI software-Kinect configuration Test Movement System Mean (SD) 95% CI ICC(2,k) (95% CI) P value SEM MDC (deg) Lower (deg) Upper (deg) (deg) (deg) MCMJ_L (n=33) Hip flexion BTS −49.4 (18.7) −56.0 −42.8 0.947 (0.33 to 0.99) 0.000 3.7 10.2 iPi −56.3 (18.4) −62.8 −49.7 Hip adduction BTS −18.8 (6.0) −20.9 −16.7 0.792 (0.17 to 0.92) 0.000 2.5 7.0 iPi −15.3 (4.8) −17.1 −13.6 Knee flexion BTS 35.8 (10.1) 32.2 39.4 0.954 (0.80 to 0.98) 0.000 2.4 6.6 iPi 38.4 (9.7) 35.0 41.9 Knee adduction BTS −16.3 (6.0) −18.5 −14.2 0.873 (0.59 to 0.95) 0.000 2.4 6.5 iPi −13.9 (5.6) −15.9 −12.0 MCMJ_R (n=34) Hip flexion BTS −65.7 (14.5) −70.8 −60.7 0.846 (–0.14 to 0.96) 0.000 3.5 9.7 iPi −76.4 (13.9) −81.3 −71.6 Hip adduction BTS 24.0 (5.9) 21.9 26.0 0.713 (0.12 to 0.88) 0.000 3.1 8.5 iPi 20.1 (5.0) 18.4 21.9 Knee flexion BTS 31.0 (8.3) 28.1 33.9 0.945 (0.81 to 0.98) 0.000 2.5 6.8 iPi 33.4 (9.6) 30.1 36.8 Knee adduction BTS 16.8 (5.2) 15.0 18.6 0.742 (–0.01 to 0.91) 0.000 2.5 6.9 iPi 13.1 (4.8) 11.4 14.7 ICC(2,k), intraclass correlation coefficient (absolute agreement); MCMJ_L, Modified Counter-Movement Jump Left; MCMJ_R, Modified Counter-Movement Jump Right; MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square.

4.8) 11.4 14.7 ICC(2,k), intraclass correlation coefficient (absolute agreement); MCMJ_L, Modified Counter-Movement Jump Left; MCMJ_R, Modified Counter-Movement Jump Right; MDC, minimal detectable change calculated as SEMx1.96x√2. P<0.05; SEM, SE of the measure calculated as the square root of the residual mean square. Biases and limits of agreement (table 8) (online supplementary material 1: Bland-Altman plots) were documented. The mean differences are relatively low especially for hip adduction and knee flexion and adduction. For most of the measures examined, no systematic error is detected. For hip flexion, however, there appears to be a systematic error of approximately 10°. 10.1136/bmjsem-2018-000441.supp1Supplementary data Table 8 95% LOA and the bias of the motion capture systems

Biases and limits of agreement (table 8) (online supplementary material 1: Bland-Altman plots) were documented. The mean differences are relatively low especially for hip adduction and knee flexion and adduction. For most of the measures examined, no systematic error is detected. For hip flexion, however, there appears to be a systematic error of approximately 10°. 10.1136/bmjsem-2018-000441.supp1Supplementary data Table 8 95% LOA and the bias of the motion capture systems Test Movement Lower LOA (deg) Upper LOA (deg) Bias (deg) SLS_L Hip flexion −0.9 21.6 10.3 Hip adduction −7.4 11.1 1.8 Knee flexion −9.6 4.1 −2.8 Knee adduction −9.5 5.0 −2.2 SLS_R Hip flexion 3.8 25.7 14.7 Hip adduction −11.6 2.9 −4.4 Knee flexion −10.4 5.2 −2.6 Knee adduction −4.7 11.8 3.6 SLJ_L Hip flexion −2.5 21.4 9.5 Hip adduction −8.2 7.7 −0.2 Knee flexion −8.6 8.0 −0.3 Knee adduction −7.1 2.2 −2.5 SLJ_R Hip flexion −0.1 27.8 13.9 Hip adduction −13.2 7.2 −3.0 Knee flexion −9.1 8.1 −0.5 Knee adduction −0.9 10.0 4.5 MCMJ_L Hip flexion −3.3 17.1 6.9 Hip adduction −10.4 3.6 −3.4 Knee flexion −9.2 4.0 −2.6 Knee adduction −8.9 4.1 −2.4 MCMJ_R Hip flexion 1.0 20.4 10.7 Hip adduction −4.7 12.4 3.8 Knee flexion −9.2 4.4 −2.4 Knee adduction −3.1 10.6 3.8 LOA, limits of agreement; MCMJ_L, Modified Counter-Movement Jump Left; MCMJ_R, Modified Counter-Movement Jump Right; SLS_L, Single Leg Squat Left; SLS_R, Single Leg Squat Right. Discussion Here, we have established, for the first time, validity values for SLS, CMJ and MCMJ in a cohort of professional athletes using a 2 camera markerless motion capture system (Kinect v2).

Test Movement Lower LOA (deg) Upper LOA (deg) Bias (deg) SLS_L Hip flexion −0.9 21.6 10.3 Hip adduction −7.4 11.1 1.8 Knee flexion −9.6 4.1 −2.8 Knee adduction −9.5 5.0 −2.2 SLS_R Hip flexion 3.8 25.7 14.7 Hip adduction −11.6 2.9 −4.4 Knee flexion −10.4 5.2 −2.6 Knee adduction −4.7 11.8 3.6 SLJ_L Hip flexion −2.5 21.4 9.5 Hip adduction −8.2 7.7 −0.2 Knee flexion −8.6 8.0 −0.3 Knee adduction −7.1 2.2 −2.5 SLJ_R Hip flexion −0.1 27.8 13.9 Hip adduction −13.2 7.2 −3.0 Knee flexion −9.1 8.1 −0.5 Knee adduction −0.9 10.0 4.5 MCMJ_L Hip flexion −3.3 17.1 6.9 Hip adduction −10.4 3.6 −3.4 Knee flexion −9.2 4.0 −2.6 Knee adduction −8.9 4.1 −2.4 MCMJ_R Hip flexion 1.0 20.4 10.7 Hip adduction −4.7 12.4 3.8 Knee flexion −9.2 4.4 −2.4 Knee adduction −3.1 10.6 3.8 LOA, limits of agreement; MCMJ_L, Modified Counter-Movement Jump Left; MCMJ_R, Modified Counter-Movement Jump Right; SLS_L, Single Leg Squat Left; SLS_R, Single Leg Squat Right. Discussion Here, we have established, for the first time, validity values for SLS, CMJ and MCMJ in a cohort of professional athletes using a 2 camera markerless motion capture system (Kinect v2). Our results indicate that a dual Kinect v2 configuration is a valid tool for assessment of sagittal plane knee range and peak angles, during squat and jumping tests. Additionally, during the SLS test excellent agreement between systems was found for thigh and foot adduction/abduction motion.

Discussion Here, we have established, for the first time, validity values for SLS, CMJ and MCMJ in a cohort of professional athletes using a 2 camera markerless motion capture system (Kinect v2). Our results indicate that a dual Kinect v2 configuration is a valid tool for assessment of sagittal plane knee range and peak angles, during squat and jumping tests. Additionally, during the SLS test excellent agreement between systems was found for thigh and foot adduction/abduction motion. Although agreement improved when using two cameras configuration instead of one,27 the between system agreement varied widely, especially for movements of clinical interest like hip flexion, hip adduction and knee adduction. There was also variability in agreement for different joints and different parameters. For example, shin ab/adduction showed better reliability and validity when considering the peak values in comparison to the results from individual tests. Clinical interpretation is therefore recommended for each approach (eg, individual trials vs averaged values, vs peak values). It may be argued that, in the context of risk of an acute anterior cruciate ligament injury, the peak shin adduction is a more important metric than the average across a number of trials whereas in ‘overuse’ type injuries average values may be a more sensible estimator Also, poor agreement was found regarding all rotational movements. Regarding peak angles, we noticed slightly, but inconsistently, better results found for left side compared with right. Positioning one camera on the left side of the athletes may have influenced these results. Importantly, the estimation of the MDC allows better interpretation of the individual kinematic parameters of interest for future studies and allows for adequate planning (power analyses) of intervention trials. For example, it is suggested that knee abduction at initial contact and peak during a drop jump task is predictive of subsequent ACL injury—the between group differences being 8.4° and 7.6°, respectively, for those who were subsequently injured and those who were not.28 We suggest that this infers the amount of variability between trials for a given subject is likely so large as to exceed these suggested cut-points. In comparison, studies examining changes in hip and knee peak flexion during a landing task, after a fatigue protocol, report 5.1⁰ hip flexion and 6.7⁰ knee flexion29 and 7⁰ increase in peak knee flexion before and after a general fatigue protocol.30 In comparison to the displayed MDC values here, we suggest that both the markerless and marker-based approaches can readily detect such changes.

g a landing task, after a fatigue protocol, report 5.1⁰ hip flexion and 6.7⁰ knee flexion29 and 7⁰ increase in peak knee flexion before and after a general fatigue protocol.30 In comparison to the displayed MDC values here, we suggest that both the markerless and marker-based approaches can readily detect such changes. Further to this, it was noted that the hip flexion angle appeared to have a systematic error of approximately 10° when comparing the markerless and marker-based systems. Post processing (ie, subtracting 10° from all measures) could simply remove this artefact and result in more accurate measures. It is uncertain from where this shift arises; however, the closed nature of the processing conducted through the markerless software capture and subsequent processing likely render this a difficult problem to resolve. Recent studies using Kinect v2 multiple cameras19 20 found excellent between systems agreement when measuring spatiotemporal gait parameters. It should be noted, however, that these studies examined gait, not higher speed movements examined here. The advantages of the Kinect approach were the much shorter set-up time and much lower (financial) cost of the equipment. Processing time was approximately 7 min for each test and the results derived are for the whole body. We suggest that consideration of the accuracy presented here along with these advantages will allow clinicians to better assess if this approach would be viable for their specific situation.

ial) cost of the equipment. Processing time was approximately 7 min for each test and the results derived are for the whole body. We suggest that consideration of the accuracy presented here along with these advantages will allow clinicians to better assess if this approach would be viable for their specific situation. Some limitations should be considered in the interpretation of the results of this study. Differences in the definition of reference systems and processing between the two systems may have an important role in the extracted results. Additionally, position chosen for markerless cameras may have had a negative effect on the results obtained. Variations in footwear type and sole height may have caused variations in ankle joint centre detection, reducing measurement accuracy. Additionally, markers on the footwear may have affected results for the foot—potentially this is a source of the movement differences detected by Kinect and the BTS system. Importantly, Microsoft recently discontinued production of the Kinect v2 camera. Although the devices remain available for purchase online and in physical retail stores at the time of writing, this will change in the future.

potentially this is a source of the movement differences detected by Kinect and the BTS system. Importantly, Microsoft recently discontinued production of the Kinect v2 camera. Although the devices remain available for purchase online and in physical retail stores at the time of writing, this will change in the future. Future studies are recommended to test the clinical utility of Kinect v2–iPi software configuration using more than two cameras. Cameras set up at 45° from the frontal plane may positively influence the extracted data. Future investigations should use standardised footwear or barefoot conditions to improve ankle visualisation and improve measurement of ankle joint kinematics. In conclusion, this study supports the use of dual Kinect v2 configuration with the iPi software as a valid tool for assessment of sagittal knee kinematic parameters in athletes. Contributors: All authors contributed equally during the study design, data collection/analysis and manuscript writing/revision. Funding: The publication of this article was funded by the Qatar National Library. Competing interests: None declared. Patient consent for publication: Obtained. Ethics approval: ADLQ E2013000003. Provenance and peer review: Not commissioned; externally peer reviewed. Data sharing statement: The data can be accessed via the corresponding author.