Browse the corpus

at home, and the service was used to help facilitate this. This potential confounding by indication would result in increased uptake of the service by those who might have died at home anyway. In view of the limitations of the study design, we have reported associations but can make no claims for causal relationships. Less than a quarter of all potential patients accessed Delivering Choice services. Moreover, about two-thirds of Delivering Choice service users had cancer while only about 30% of the total eligible population died from this condition. It is possible that by including a high proportion of patients with cancer, the DCP achieved more home deaths. Indeed, in the years 2008–2010, which preceded the introduction of Delivering Choice, 26% of patients with cancer in North Somerset and 28% of patients with cancer in Somerset died at home compared with 18% and 20% of all deaths being at home.6 However, among all patients with DCP, the proportions of deaths at home were 41% and 54%. Even among patients with cancer this represented a substantial increase in proportions of people dying at home.

Introduction Since the publication of the End of Life Care Strategy in the UK in 2008, there has been increasing emphasis on supporting people to die in their own homes.1 From 2008 to 2011, Marie Curie Cancer Care worked with local professionals to develop end-of-life services in several parts of the UK, including North Somerset (North Somerset) and Somerset, two counties in southwest England.2 All palliative patients, regardless of condition, were eligible for the Delivering Choice Programme (DCP), which included. An out-of-hours (OOH) advice and response line manned by specialist palliative care nurses from 17:00 to 1:00, weekends and bank holidays, who responded to calls from professionals, family carers and patients (Somerset). Two ‘front of house’ hospital-based end-of-life care discharge-in-reach nurses who identified patients who wanted a non-hospital death and facilitated fast discharges accordingly (Somerset). Two end-of-life care coordination centres (one in each county) that took referrals from community, hospital and hospice staff to organise packages of care including equipment, night nurses and personal carers. In both counties, personal care was supplied from local agencies, but North Somerset also had its own in-house generic support team. These services were supported by an electronic end-of-life care register (EOL register) to record advanced care wishes.

Two end-of-life care coordination centres (one in each county) that took referrals from community, hospital and hospice staff to organise packages of care including equipment, night nurses and personal carers. In both counties, personal care was supplied from local agencies, but North Somerset also had its own in-house generic support team. These services were supported by an electronic end-of-life care register (EOL register) to record advanced care wishes. In addition, in North Somerset, two end-of-life care facilitators were employed. These staff provided specific training, primarily to care homes and community teams, on issues such as advance care planning, end-of-life care pathways, use of the electronic end-of-life care register and medication including syringe drivers. No patient-level outcome data were linked to this intervention. The aim of this study was to investigate the impact of Delivering Choice on place of death and hospital admissions and emergency department (ED) use. A parallel qualitative evaluation has been published elsewhere.3

In addition, in North Somerset, two end-of-life care facilitators were employed. These staff provided specific training, primarily to care homes and community teams, on issues such as advance care planning, end-of-life care pathways, use of the electronic end-of-life care register and medication including syringe drivers. No patient-level outcome data were linked to this intervention. The aim of this study was to investigate the impact of Delivering Choice on place of death and hospital admissions and emergency department (ED) use. A parallel qualitative evaluation has been published elsewhere.3 Methods The sample was a retrospective cohort of people who died between 1 September 2011 and 29 February 2012 in North Somerset and Somerset and whose deaths were expected and potentially eligible for end-of-life care according to criteria derived by the UK National End of Life Care Intelligence Network.4 Quantitative data on use of Delivering Choice services were collected and linked via NHS number, age and postcode by National Health Service analysts to the Primary Care Mortality Database, in-patient hospital data and ED data obtained from NHS Connecting for Health. All patient data were pseudonymised during linkage. Differences between Delivering Choice and non-Delivering Choice users for place of death, emergency hospital admissions and ED visits in the last 30 and 7 days of life were explored. In addition, length of time between date of first contact with a Delivering Choice service and date of death was calculated, as was a Charlson Comorbidity Index for those people where data on comorbidities were available.5 Univariable analyses and multivariable logistic regression were undertaken using STATA V.12. The regression models adjusted for a priori confounding factors, including gender, age, deprivation, cause of death and other DCP services received.

omorbidity Index for those people where data on comorbidities were available.5 Univariable analyses and multivariable logistic regression were undertaken using STATA V.12. The regression models adjusted for a priori confounding factors, including gender, age, deprivation, cause of death and other DCP services received. The study was approved by local NHS Research and Development managers as a service evaluation, therefore NHS Research Ethics Committee and Research Governance approvals were not required. However, the study was reviewed by the ethics committee of the Faculty of Medicine at the University of Bristol and relevant Caldicott permissions were obtained. Results Participant characteristics From 1 September 2011 to 29 February 2012, 1022 people potentially eligible for end-of-life care died in North Somerset and 2572 in Somerset (table 1). The overall mean age at death was 81 (North Somerset) and 82 (Somerset) with comparable proportions of women (54% and 55%) and men (46% and 45%). Commonest causes of death were cancer (North Somerset 28%, Somerset 29%), heart disease (18%, 18%), respiratory disease (15%, 13%) and dementia (15%, 13%). Table 1 Characteristics of end-of-life populations, place of death and hospital use

Results Participant characteristics From 1 September 2011 to 29 February 2012, 1022 people potentially eligible for end-of-life care died in North Somerset and 2572 in Somerset (table 1). The overall mean age at death was 81 (North Somerset) and 82 (Somerset) with comparable proportions of women (54% and 55%) and men (46% and 45%). Commonest causes of death were cancer (North Somerset 28%, Somerset 29%), heart disease (18%, 18%), respiratory disease (15%, 13%) and dementia (15%, 13%). Table 1 Characteristics of end-of-life populations, place of death and hospital use North Somerset (n=1022) (%)* Somerset (n=2572) (%)* Delivering choice Yes No Yes No N 213 809 616 1956 Age at death (mean SD) 79.4 (10.7) 81.2 (12.8) 77.3 (12.5) 82.4 (11.4) Gender male 109 (51) 361 (45) 301 (49) 864 (44) Deprivation (index of multiple deprivation) Least deprived 67 (31) 242 (30) 94 (15) 259 (13) Below average deprivation 63 (30) 229 (28) 213 (35) 631 (32) Average deprivation 33 (15) 119 (15) 178 (29) 629 (32) Above average deprivation 10 (5) 65 (8) 100 (16) 345 (18) Most deprived 37 (17) 139 (17) 29 (5) 80 (4) Not available 3 (1) 15 (2) 2 (0) 12 (1) Charlson score† Number of individuals 168 (79) 506 (63) 513 (83) 1321 (68) Weighted score 4.6 3.0 5.3 2.9 Cause of death Cancer 145 (68) 142 (18) 440 (71) 314 (16) Heart disease 15 (7) 170 (21) 44 (7) 430 (22) Cerebrovascular 9 (4) 78 (10) 19 (3) 218 (11) Respiratory 14 (7) 142 (18) 32 (5) 308 (16) Dementia 12 (6) 145 (18) 30 (5) 316 (16) Other 18 (8) 132 (16) 51 (8) 370 (19) Place of death Acute hospital 40 (19) 347 (43) 84 (14) 836 (43) Home‡ 88 (41) 225 (28) 337 (54) 779 (40) Care home (not usual place of residence) 34 (16) 116 (14) 58 (9) 173 (9) Hospice 34 (16) 30 (4) 98 (16) 55 (3) Community hospital N/A N/A 31 (5) 31 (5) Elsewhere 17 (8) 91 (11) 8 (1) 12 (1) Hospital utilisation Patients with one or more emergency admissions <30 days (SD) 61 (29) 335 (41) 233 (38) 875 (45) Mean emergency admissions per patient <30 days (SD) [median, min, max] 0.31 (0.52) [0,0,2] 0.47 (0.60) [0,0,3] 0.53 (0.69) [0,0,3] 0.45 (0.64) [0,0,7] Patients with one or more emergency admission <7 days (SD) 13 (6) 179 (22) 60 (10) 467 (24) Mean emergency admissions per patient <7 days (SD) [median, min, max] 0.07 (0.27) [0,0,2] 0.23 (0.43) [0,0,2] 0.11 (0.33) [0,0,2] 0.25 (0.46) [0,0,2] Patients with one or more ED attendance <30 days (SD) 54 (25) 363 (36) 159 (26) 712 (36) Mean ED attendance per patient <30 days (SD) [median, min, max] 0.27 (0.50) [0,0,4] 0.52 (0.64) [0,0,2] 0.39 (0.51) [0,0,3] 0.41 (0.60) [0,0,5] Patients with one or more ED attendance <7 days 13 (6) 213 (26) 43 (7) 432 (22) Mean ED attendance per patient <7 days (SD) [median, min, max] 0.07 (0.29) [0,0,2] 0.27 (0.47) [0,0,2] 0.07 (0.27) [0,0,2] 0.26 (

0 days (SD) [median, min, max] 0.27 (0.50) [0,0,4] 0.52 (0.64) [0,0,2] 0.39 (0.51) [0,0,3] 0.41 (0.60) [0,0,5] Patients with one or more ED attendance <7 days 13 (6) 213 (26) 43 (7) 432 (22) Mean ED attendance per patient <7 days (SD) [median, min, max] 0.07 (0.29) [0,0,2] 0.27 (0.47) [0,0,2] 0.07 (0.27) [0,0,2] 0.26 ( 0.43) [0,0,3] *Figure in parentheses refers to percentage unless otherwise indicated. †Charlson score was calculated for those patients who had a hospital admission. ‡Home includes a care home where this was recorded as usual place of residence. ED, emergency department; N/A, not applicable. Using Delivering Choice Of the 1022 individuals in North Somerset who were eligible, 213 (21%) used one or more Delivering Choice intervention, most frequently the care coordination centre (153, 15%), followed by entry on the EOL register (93, 9%) (table 2). Less than 5% of people had a key worker listed on the EOL register (34, 4%) or accessed a generic support worker (GSW) (25, 2%). Data on who used GSW services were only available for 101 of the potential 181 days of the study. Table 2 Delivering choice service use Service North Somerset N=1022 (%) Somerset N=2572 (%) Care coordination centre 153 (15) 294 (11) EOL register 93 (9) 331 (13) Key worker on EOL register 34 (4) 156 (6) Generic support worker* 25 (2) N/A Out of hours advice line N/A 243 (9) Discharge in reach nurse N/A 144 (6) Total people using DCP 213 (21) 616 (24) *Data were only available for 101 days. DCP, Delivering Choice Programme.

Service North Somerset N=1022 (%) Somerset N=2572 (%) Care coordination centre 153 (15) 294 (11) EOL register 93 (9) 331 (13) Key worker on EOL register 34 (4) 156 (6) Generic support worker* 25 (2) N/A Out of hours advice line N/A 243 (9) Discharge in reach nurse N/A 144 (6) Total people using DCP 213 (21) 616 (24) *Data were only available for 101 days. DCP, Delivering Choice Programme. Of the 2572 people in Somerset who were eligible, 616 (24%) used one or more DCP intervention (table 2). The most frequently used were entry on the EOL register (331, 13%), the care coordination centre (294, 11%) and the OOH advice line (243, 9%). Fewer people had a key worker listed on the EOL register (156, 6%) or were seen by a discharge in-reach nurse (144, 6%). A much higher proportion of DCP users died from cancer than non-DCP users (table 1). The distribution of deprivation scores relative to England as a whole was similar for DCP users and non-users in both counties. Data on comorbidities were only available for those who had a hospital admission during the 6 months before death. Among these people, those who accessed DCP services had higher levels of comorbidity than those who did not access DCP (table 1). Duration of use of services In North Somerset, median time for accessing the care coordination centre was 8 days before death (IQR 4–21). In Somerset, median time for accessing the care coordination centre was 9.5 days before death (IQR 4–20), while median time to death after first contact for the discharge in-reach service was 6 days (IQR 2–23) and for the OOH line 10 days (IQR 2–31).

sing the care coordination centre was 8 days before death (IQR 4–21). In Somerset, median time for accessing the care coordination centre was 9.5 days before death (IQR 4–20), while median time to death after first contact for the discharge in-reach service was 6 days (IQR 2–23) and for the OOH line 10 days (IQR 2–31). Place of death Those receiving Delivering Choice were less likely to die in hospital (table 1). In North Somerset, the OR for dying in hospital was 0.33 (95% CI 0.21 to 0.50, p≤0.001) after adjusting for gender, age, deprivation and cause of death (table 3). In Somerset, the OR was 0.20 (95% CI 0.17 to 0.27, p<0.001). Looking at individual interventions, in North Somerset those who used the care coordination centre were less likely to die in hospital (OR 0.42, 95% CI 0.25 to 0.69, p<0.001) as were those patients entered on the EOL register (OR 0.30, 95% CI 0.13 to 0.69, p=0.005). In Somerset, patients accessing the care coordination centre were less likely to die in hospital (OR 0.11, 95% CI 0.06 to 0.22, p<0.001) as were those who used the OOH advice line (OR 0.34, 95% CI 0.20 to 0.57, p<0.001) or were entered on the EOL register (OR 0.22, 95% CI 0.12 to 0.40, p<0.001). Table 3 Impact of Delivering Choice services versus non-Delivering Choice services on deaths in hospital, emergency admissions and A&E use*

Place of death Those receiving Delivering Choice were less likely to die in hospital (table 1). In North Somerset, the OR for dying in hospital was 0.33 (95% CI 0.21 to 0.50, p≤0.001) after adjusting for gender, age, deprivation and cause of death (table 3). In Somerset, the OR was 0.20 (95% CI 0.17 to 0.27, p<0.001). Looking at individual interventions, in North Somerset those who used the care coordination centre were less likely to die in hospital (OR 0.42, 95% CI 0.25 to 0.69, p<0.001) as were those patients entered on the EOL register (OR 0.30, 95% CI 0.13 to 0.69, p=0.005). In Somerset, patients accessing the care coordination centre were less likely to die in hospital (OR 0.11, 95% CI 0.06 to 0.22, p<0.001) as were those who used the OOH advice line (OR 0.34, 95% CI 0.20 to 0.57, p<0.001) or were entered on the EOL register (OR 0.22, 95% CI 0.12 to 0.40, p<0.001). Table 3 Impact of Delivering Choice services versus non-Delivering Choice services on deaths in hospital, emergency admissions and A&E use* North Somerset Somerset OR 95% CI p Value OR 95% CI p Value Death in hospital overall 0.33 0.21 to 0.50 <0.001 0.20 0.17 to 0.27 <0.001 Key worker 0.74 0.18 to 3.05 0.679 0.73 0.27 to 1.95 0.524 Care coordination centre 0.42 0.25 to 0.69 0.001 0.11 0.06 to 0.22 <0.001 Generic support workers 0.34 0.04 to 2.64 0.30 N/A – – EOL register 0.30 0.13 to 0.69 0.005 0.22 0.12 to 0.40 <0.001 Out-of-hours advice line N/A – – 0.34 0.20 to 0.57 <0.001 Discharge in reach nurses N/A – – 1.6 0.98 to 2.60 0.06 Emergency admission <30 days 0.49 0.33 to 0.74 0.001 0.61 0.48 to 0.76 <0.001 Key worker 0.60 0.21 to 1.69 0.33 1.13 0.68 to 1.87 0.642 Care coordination centre 0.55 0.34 to 0.90 0.016 0.58 0.42 to 0.80 0.001 Generic support workers 0.35 0.08 to 1.59 0.175 N/A – – EOL register 0.65 0.33 to 1.30 0.225 0.41 0.28 to 0.60 <0.001 Out-of-hours advice line N/A – – 0.78 0.56 to 1.10 0.159 Discharge in reach nurses N/A – – 4.15 2.68 to 6.43 <0.001 Emergency admission <7 days 0.22 0.12 to 0.44 <0.001 0.32 0.23 to 0.45 <0.001 Key worker 1.13 0.18 to 7.03 0.898 1.04 0.45 to 2.40 0.934 Care coordination centre 0.09 0.02 to 0.39 0.001 0.26 0.15 to 0.46 <0.001 Generic support workers 6.26 0.81 to 48.65 0.079 N/A – – EOL register 0.39 0.11 to 1.34 0.136 0.57 0.33 to 0.98 0.043 Out-of-hours advice line N/A – – 0.44 0.25 to 0.78 0.005 Discharge in reach nurses N/A – – 1.54 0.95 to 2.50 0.081 A&E attendance <30 days 0.41 0.28 to 0.62 <0.001 0.66 0.51 to 0.85 0.001 Key worker 0.71 0.25 to 2.02 0.515 1.23 0.71 to 2.13 0.454 Care coordination centre 0.46 0.29 to 0.76 0.002 0.58 0.40 to 0.82 0.002 Generic support workers 0.45 0.10 to 2.01 0.295 N/A – – EOL register 0.57 0.29 to 1.11 0.097 0.61 0.40 to 0.92 0.018 Out-of-hours advice line N/A – – 0.60 0.41 to 0.87 0.007 Discharge in reach nurses N/A – – 3.29 2.23 to 4.87 <0.001 A&E attendance <7 days 0.22 0.11 to 0.42 <0.001 0.32 0.22 to 0.47 <0.001 Key worker 1.16 0.17 to 7.84 0.878 0.85 0.32 to 2.20 0.730 Care coordination centre 0.15 0.05 to 0.43 <0.001 0.24 0.12 to 0.48 <0.001 Generic support workers 2.39 0.27 to 21.61 0.437 N/A – – EOL register 0.38 0.12 to 1.27 0.116 0.79 0.44 to 1.41 0.427 Out-of-hours advice line N/A – – 0.34 0.17 to 0.70

22 to 0.47 <0.001 Key worker 1.16 0.17 to 7.84 0.878 0.85 0.32 to 2.20 0.730 Care coordination centre 0.15 0.05 to 0.43 <0.001 0.24 0.12 to 0.48 <0.001 Generic support workers 2.39 0.27 to 21.61 0.437 N/A – – EOL register 0.38 0.12 to 1.27 0.116 0.79 0.44 to 1.41 0.427 Out-of-hours advice line N/A – – 0.34 0.17 to 0.70 0.003 Discharge in reach nurses N/A – – 1.25 0.74 to 2.11 0.427 *Adjusted for gender, age, deprivation and condition.

22 to 0.47 <0.001 Key worker 1.16 0.17 to 7.84 0.878 0.85 0.32 to 2.20 0.730 Care coordination centre 0.15 0.05 to 0.43 <0.001 0.24 0.12 to 0.48 <0.001 Generic support workers 2.39 0.27 to 21.61 0.437 N/A – – EOL register 0.38 0.12 to 1.27 0.116 0.79 0.44 to 1.41 0.427 Out-of-hours advice line N/A – – 0.34 0.17 to 0.70 0.003 Discharge in reach nurses N/A – – 1.25 0.74 to 2.11 0.427 *Adjusted for gender, age, deprivation and condition. Hospital admissions In the last 30 days of life, emergency hospital admissions were proportionally lower in both counties among those receiving a Delivering Choice intervention compared with non-users (table 1). For both 30-day (North Somerset: OR 0.49, 95% CI 0.33 to 0.74, p<0.001; and Somerset: OR 0.61, 95% CI 0.48 to 0.76, p<0.001) and 7-day (North Somerset: OR, 0.22 95% CI 0.12 to 0.44, p<0.001; Somerset: OR 0.32, 95% CI 0.23 to 0.45, p<0.001) results, these findings persist after adjusting for gender, age, deprivation and cause of death (table 3). The two coordination centres appeared to be the most effective components of the interventions at both 30 days (North Somerset: OR 0.55, 95% CI 0.34 to 0.90, p=0.016; and Somerset: OR 0.58, 95% CI 0.42 to 0.80, p<0.001) and 7 days (North Somerset: OR 0.09, 95% CI 0.02 to 0.39, p<0.001; and Somerset: OR, 0.26 95% CI 0.15 to 0.46, p<0.001). The OOH advice line was associated with lower risk of admission in the last week of life only (OR 0.44, 95% CI 0.25 to 0.78, p=0.005). Entry on the EOL register in Somerset was associated with lower risk of admission (30 days: OR 0.41, 95% CI 0.28 to 0.60; 7 days: OR 0.57, 95% CI 0.33 to 0.98, p=0.043) but not in North Somerset. Readmissions after accessing the discharge in-reach service were low at 6%.

fe only (OR 0.44, 95% CI 0.25 to 0.78, p=0.005). Entry on the EOL register in Somerset was associated with lower risk of admission (30 days: OR 0.41, 95% CI 0.28 to 0.60; 7 days: OR 0.57, 95% CI 0.33 to 0.98, p=0.043) but not in North Somerset. Readmissions after accessing the discharge in-reach service were low at 6%. ED visits In both counties, attendance at ED in the last 30 days of life was proportionately lower among DCP users (table 1). For both 30-day (North Somerset: OR 0.41, 95% CI 0.28 to 0.62, p<0.001; and Somerset: OR 0.66, 95% CI 0.51 to 0.85, p<0.001) and 7-day (North Somerset: OR 0.22, 95% CI 0.11 to 0.42, p<0.001; and Somerset: OR 0.32, 95% CI 0.22 to 0.47, p<0.001) results, these findings persist after adjusting for confounders (table 3). Again, the Care Coordination Centres appeared to be the most effective component at 30 days (North Somerset: OR 0.46, 95% CI 0.29 to 0.76, p=0.002; and Somerset: OR 0.58, 95% CI 0.40 to 0.82, p=0.002) and 7 days (North Somerset: OR 0.15, 95% CI 0.05 to 0.43, p<0.001; and Somerset: OR 0.24, 95% CI 0.12 to 0.48, p<0.001). The OOH advice line was associated with lower risk of ED attendance in the last week only (OR 0.34, 95% CI 0.17 to 0.70, p=0.003). Discussion Patients who used Delivering Choice services were less likely to die in hospital, have emergency admissions or visit ED within 30 or 7 days before death than those who did not use Delivering Choice. The two coordination centres appeared to have the biggest impact.

ED visits In both counties, attendance at ED in the last 30 days of life was proportionately lower among DCP users (table 1). For both 30-day (North Somerset: OR 0.41, 95% CI 0.28 to 0.62, p<0.001; and Somerset: OR 0.66, 95% CI 0.51 to 0.85, p<0.001) and 7-day (North Somerset: OR 0.22, 95% CI 0.11 to 0.42, p<0.001; and Somerset: OR 0.32, 95% CI 0.22 to 0.47, p<0.001) results, these findings persist after adjusting for confounders (table 3). Again, the Care Coordination Centres appeared to be the most effective component at 30 days (North Somerset: OR 0.46, 95% CI 0.29 to 0.76, p=0.002; and Somerset: OR 0.58, 95% CI 0.40 to 0.82, p=0.002) and 7 days (North Somerset: OR 0.15, 95% CI 0.05 to 0.43, p<0.001; and Somerset: OR 0.24, 95% CI 0.12 to 0.48, p<0.001). The OOH advice line was associated with lower risk of ED attendance in the last week only (OR 0.34, 95% CI 0.17 to 0.70, p=0.003). Discussion Patients who used Delivering Choice services were less likely to die in hospital, have emergency admissions or visit ED within 30 or 7 days before death than those who did not use Delivering Choice. The two coordination centres appeared to have the biggest impact. There are a number of limitations to this study. Randomisation of patients was not practical in this service development. We addressed potential confounders such as age, gender, deprivation and cause of death. However, there are other ways in which the patient groups may have differed. For example, we were unable to include data on comorbidities in the logistic regression analyses. It is also possible that patients referred to the Delivering Choice services were people that a district nurse or GP considered were suitable to die at home, or who possibly wished to die at home, and the service was used to help facilitate this. This potential confounding by indication would result in increased uptake of the service by those who might have died at home anyway. In view of the limitations of the study design, we have reported associations but can make no claims for causal relationships.

erset and 28% of patients with cancer in Somerset died at home compared with 18% and 20% of all deaths being at home.6 However, among all patients with DCP, the proportions of deaths at home were 41% and 54%. Even among patients with cancer this represented a substantial increase in proportions of people dying at home. The most effective components of the DCPs in both counties appear to be the care coordination centres. A qualitative evaluation of the components of Delivering Choice identified a number of factors that contribute to further understanding of the effectiveness of the individual interventions.2 The care coordination centres in the two counties operated in different ways. The North Somerset centre had an in-house model that included a fast-track coordinator, nurse assessors and its own team of personal care workers (GSWs). This maximised their flexibility to respond to patient and family needs, and this was particularly valued by families.3 Co-location with social service staff meant that the centre was well placed to identify potential patients early in the referral process for end-of-life care. Although the Somerset Care Coordination Centre had the same key function of organising care packages as the North Somerset Care Coordination Centre, the model was different. The Somerset Care Coordination Centre was led by a nurse and staffed exclusively by administrators, without any in-house care staff, additional nurses or fast-track coordinator. Thus to make this model work, the Somerset centre was heavily reliant on high-quality management and good external relationships, particularly with community and palliative care nurses and private care agencies. However, both services were valued for providing fast and efficient access to care packages and equipment. Previous systematic reviews have provided conflicting evidence with suggestions that specialist palliative care input at home can reduce general healthcare use and increase family and patient satisfaction with care but not all the evidence is supportive of this.7 8

t and efficient access to care packages and equipment. Previous systematic reviews have provided conflicting evidence with suggestions that specialist palliative care input at home can reduce general healthcare use and increase family and patient satisfaction with care but not all the evidence is supportive of this.7 8 As only a third of patients accessing Delivering Choice had non-cancer diagnoses, this would suggest that primary healthcare teams should be encouraged to identify more patients for a palliative care approach and service providers should also redesign services that are manifestly appropriate for non-malignant patients. The study also highlights the difficulties of indentifying palliative patients in a timely way as half of Delivering Choice patients came into contact with the services just 6–20 days before death. Previous research in Scotland has identified that encouraging GPs to identify palliative care patients can be beneficial as 60% of patients on the GP palliative care register died at home, whereas 60% of people who were not on the palliative care register died in hospital.9 10 However, GPs find early identification of non-cancer palliative care patients to be particularly challenging.11 Therefore, other interventions to improve the identification of patients for palliative care need to be put in place, in addition to providing services such as Delivering Choice for those already identified to be at the end of life. If services can identify and care for patients earlier, there is a possibility that hospital use at the end of life could be reduced further in line with national policy.1

ive care need to be put in place, in addition to providing services such as Delivering Choice for those already identified to be at the end of life. If services can identify and care for patients earlier, there is a possibility that hospital use at the end of life could be reduced further in line with national policy.1 In future studies, further evaluation of the components of the DCP that seemed to be most beneficial would be useful.2 Although randomised controlled trials of complex interventions are costly and difficult to undertake, the addition of trial evidence to the current literature would provide more robust evidence for those who commission and deliver services. Conclusion The results from this evaluation suggest that patients using or supported by Delivering Choice were less likely to die in hospital or call on hospital services. Those considering new ways of providing end-of-life care could explore the possibility of adopting similar services and evaluating the outcomes from patient, carer and system perspectives. We are grateful to the staff from the DCP services, the local Marie Curie Cancer Care team and the PCT staff who linked the data. Contributors: All authors made substantial contributions to the conception or design of the work; the acquisition, analysis or interpretation of data for the work; drafting the work and revising it critically for important intellectual content and approved the version to be published. SP is the guarantor.

We are grateful to the staff from the DCP services, the local Marie Curie Cancer Care team and the PCT staff who linked the data. Contributors: All authors made substantial contributions to the conception or design of the work; the acquisition, analysis or interpretation of data for the work; drafting the work and revising it critically for important intellectual content and approved the version to be published. SP is the guarantor. Funding: This evaluation was commissioned and funded by Marie Curie Cancer Care. SP was funded by a Medical Research Council Clinician Scientist Fellowship, number G0501936. Competing interests: SP, LW, GL and TG received grant funding support from Marie Curie Cancer Care to conduct this study. SP was funded by the Medical Research Council. Ethics approval: University of Bristol Faculty of Medicine and Dentistry Ethics Committee. Provenance and peer review: Not commissioned; externally peer reviewed. Data sharing statement: All data relating to this study and a parallel qualitative evaluation are available at: http://www.bristol.ac.uk/primaryhealthcare/docs/reports/deliveringchoice.pdf

Introduction Polyneuropathy, organomegaly, endocrinopathy, M-protein and skin changes (POEMS) syndrome is a rare paraneoplastic disorder associated with an underlying plasma cell dyscrasia and multiorgan failure.1 The pathogenesis of POEMS syndrome is likely caused by overproduction of vascular endothelial growth factor (VEGF).2 POEMS syndrome is potentially fatal and adversely affects quality of life. Oedema is common with many patients affected by pleural effusions and ascites.1 2 There is limited evidence to determine the association between hydration and symptoms in advanced cancer.3 Physical examination has a low sensitivity and specificity for identifying fluid deficit and, biochemical methods show little association with hydration status.3 4 There is no routine hydration assessment method for patients with advanced cancer. The evidence for the efficacy of clinically assisted hydration (CAH) in advanced cancer is limited, conflicting and inconclusive. The subject of hydration is extremely important to patients and caregivers; there is concern about the risk of harm to patients through the use or non-use of CAH.3 Bioelectrical impedance vector analysis (BIVA) is a non-invasive, validated body composition assessment method, which may be useful in the assessment of hydration.5 To date, there are no published reports about the utility of BIVA in POEMS syndrome.

There is limited evidence to determine the association between hydration and symptoms in advanced cancer.3 Physical examination has a low sensitivity and specificity for identifying fluid deficit and, biochemical methods show little association with hydration status.3 4 There is no routine hydration assessment method for patients with advanced cancer. The evidence for the efficacy of clinically assisted hydration (CAH) in advanced cancer is limited, conflicting and inconclusive. The subject of hydration is extremely important to patients and caregivers; there is concern about the risk of harm to patients through the use or non-use of CAH.3 Bioelectrical impedance vector analysis (BIVA) is a non-invasive, validated body composition assessment method, which may be useful in the assessment of hydration.5 To date, there are no published reports about the utility of BIVA in POEMS syndrome. Case history This case describes the use of BIVA in a woman aged 52 years with POEMS syndrome. She had a history of two autologous stem cell transplants, renal impairment and recurrent lower limb and abdominal oedema. Oedema was a cause of great discomfort and had adversely affected her mobility. She was referred to the specialist palliative care team for symptom management and fluid assessment using BIVA. Following the baseline assessment, she received 40 mg of oral furosemide daily in combination with advice about fluid restriction. Two further BIVA assessments were conducted at weekly intervals following the commencement of diuretic therapy to assess the response to diuretic therapy. A clinical evaluation of peripheral oedema (upper and lower limbs) was also conducted during these assessments.

y in combination with advice about fluid restriction. Two further BIVA assessments were conducted at weekly intervals following the commencement of diuretic therapy to assess the response to diuretic therapy. A clinical evaluation of peripheral oedema (upper and lower limbs) was also conducted during these assessments. Bioelectrical impedance vector analysis (BIVA) Bioimpedance analysis involves a tetrapolar technique to deliver a single-frequency electrical current of 50 kHz. The technique works on the principle that fluid and cellular structures will provide different levels of resistance to an electrical current as it passes through the human body. Bioimpedance provides the following direct measurements: resistance (R) assessing cellular hydration, reactance (Xc) assessing tissue integrity and phase angle (PA), which is reported to be a useful indicator of health and prognosis.5 Bioimpedance analysis was conducted using the EFG-3 ElectroFluidGraph Vector Impedance Analyser (Akern) in line with methods and recommendations described elsewhere.5 Regression equations of the manufacturer (Akern BodyGram Pro 3.0) were used to calculate total body water (TBW), intracellular water (ICW) and extracellular water (ECW).6 These validated equations were derived from previous research.7

dance Analyser (Akern) in line with methods and recommendations described elsewhere.5 Regression equations of the manufacturer (Akern BodyGram Pro 3.0) were used to calculate total body water (TBW), intracellular water (ICW) and extracellular water (ECW).6 These validated equations were derived from previous research.7 BIVA enables interpretation of bioimpedance data, which is independent of regression equations and body weight. To establish BIVA, the direct impedance measurements (R and Xc) were plotted as a point (bivariate random vector) on a probability graph (RXc graph); this represented the sex-specific and race-specific tolerance intervals of a non-cancer reference population used for the analysis (figure 1).8 Figure 1 Longitudinal change of hydration represented by the BIVA RXc graph. The RXc graph method allows statistical analysis of bivariate distributions of successive impedance vectors of an individual relative to the 50%, 75% and 95% tolerance ellipses of a non-cancer reference population. The black dots represent the study assessments. The black arrow demonstrates the linear trend of the assessments. BIVA, bioelectrical impedance vector analysis; R, resistance; Xc, reactance.

sive impedance vectors of an individual relative to the 50%, 75% and 95% tolerance ellipses of a non-cancer reference population. The black dots represent the study assessments. The black arrow demonstrates the linear trend of the assessments. BIVA, bioelectrical impedance vector analysis; R, resistance; Xc, reactance. Results At baseline, BIVA demonstrated the participant's overall body composition was just outside the normal ellipse 50th centile and did not suggest fluid overload (figure 1). Following diuretic therapy, the subsequent assessments demonstrated a reduction in hydration volume. This corresponded with weight loss and a reduction in clinically detectable oedema. Throughout the assessments, TBW was low relative to body weight (table 1) and ECW was high relative to TBW. Table 1 Results of bioelectrical impedance analysis

Results At baseline, BIVA demonstrated the participant's overall body composition was just outside the normal ellipse 50th centile and did not suggest fluid overload (figure 1). Following diuretic therapy, the subsequent assessments demonstrated a reduction in hydration volume. This corresponded with weight loss and a reduction in clinically detectable oedema. Throughout the assessments, TBW was low relative to body weight (table 1) and ECW was high relative to TBW. Table 1 Results of bioelectrical impedance analysis Baseline Follow-up 1 Follow-up 2 Bioimpedance variables N (ohms) R (Ω) 679 715 696 Xc (Ω) 57 57 63 PA (Ω) 4.8 4.6 5.2 Weight (kg) 69.5 67.7 66.4 Height (cm) 157.9 157.9 157.9 BMI (kg/m2) 27.9 27.1 27.0 Hydration variables N (L) Normal values in an adult woman TBW 29.6 (43% of body weight) 28.6 (42% of body weight) 28.9 (44% of body weight) 45–50% of body weight ICW 14.2 (48% of TBW) 13.3 (47% of TBW) 13.3 (47% of TBW) 55% of TBW ECW 15.4 (52% of TBW) 15.3 (53% of TBW) 15.3 (53% of TBW) 45% of TBW Clinical examination of oedema Pitting oedema evident around ankles and shin Small non-pitting oedema to ankle No oedema detectable BMI, body mass index; ECW, extracellular water; ICW, intracellular water; PA, phase angle; R, resistance; TBW, total body water; Xc, reactance. Discussion Main findings BIVA demonstrated a reduction in hydration volume following the intervention (diuretic therapy and fluid restriction). This corresponded clinically with weight loss and a reduction in clinically detectable oedema.

Baseline Follow-up 1 Follow-up 2 Bioimpedance variables N (ohms) R (Ω) 679 715 696 Xc (Ω) 57 57 63 PA (Ω) 4.8 4.6 5.2 Weight (kg) 69.5 67.7 66.4 Height (cm) 157.9 157.9 157.9 BMI (kg/m2) 27.9 27.1 27.0 Hydration variables N (L) Normal values in an adult woman TBW 29.6 (43% of body weight) 28.6 (42% of body weight) 28.9 (44% of body weight) 45–50% of body weight ICW 14.2 (48% of TBW) 13.3 (47% of TBW) 13.3 (47% of TBW) 55% of TBW ECW 15.4 (52% of TBW) 15.3 (53% of TBW) 15.3 (53% of TBW) 45% of TBW Clinical examination of oedema Pitting oedema evident around ankles and shin Small non-pitting oedema to ankle No oedema detectable BMI, body mass index; ECW, extracellular water; ICW, intracellular water; PA, phase angle; R, resistance; TBW, total body water; Xc, reactance. Discussion Main findings BIVA demonstrated a reduction in hydration volume following the intervention (diuretic therapy and fluid restriction). This corresponded clinically with weight loss and a reduction in clinically detectable oedema. Strengths To our knowledge, this is the first study to use BIVA to evaluate hydration in POEMS syndrome. Furthermore, the evaluation of longitudinal change in hydration in POEMS syndrome using BIVA, following intervention with diuretics, is novel. The advantage of BIVA is that it allows information to be obtained simultaneously about changes in tissue hydration or soft tissue mass, independent of regression equations, or body weight. This allows for accurate interpretation of BIVA readings even if patients are at extremes of weight or volume distribution.

novel. The advantage of BIVA is that it allows information to be obtained simultaneously about changes in tissue hydration or soft tissue mass, independent of regression equations, or body weight. This allows for accurate interpretation of BIVA readings even if patients are at extremes of weight or volume distribution. Significance of the findings and comparison with previous studies The correlations between hydration change, body weight and observable oedema are consistent with non-cancer studies using BIVA to measure hydration.9 The participant commented on the simplicity of the BIVA method. Her experience is consistent with the findings of our previous research using BIVA to evaluate hydration in advanced cancer, which adds to the evidence that the BIVA method is not burdensome in palliative care.10 TBW was low relative to body weight; however, ECW was high compared with TBW. This suggests an imbalance of ECW:ICW ratio, which may be the potential cause of the oedema. Although there was a clinical reduction in weight and oedema, this was not reflected by change in the ECW volume. The reason for this is not clear; however, a potential explanation may be that the bioimpedance regression equations lack the sensitivity to detect small changes in fluid volume.

ential cause of the oedema. Although there was a clinical reduction in weight and oedema, this was not reflected by change in the ECW volume. The reason for this is not clear; however, a potential explanation may be that the bioimpedance regression equations lack the sensitivity to detect small changes in fluid volume. Limitations Bioimpedance regression equations are limited in cancer owing to their reliance on physiological assumptions (which include requirements about body shape, the relationship between trunk and leg lengths, soft tissue hydration level and fat fraction).5 Although the equations have been validated against reference methods with good accuracy, they may lack accuracy in situations where basic assumptions of the bioimpedance method are not met (eg, in cancer or at extremes of body mass and hydration).5 The limitations of regression equations are overcome by BIVA, which is independent of the regression analysis and body weight. We acknowledge that we are unable to account for potential confounding factors, which may have influenced hydration between assessments. Furthermore, the relative under-representation of the trunk by whole-body impedance limits our ability to quantify or localise the oedema.5 Consequently, BIVA should be interpreted with acknowledgement of the clinical presentation of the individual.

ding factors, which may have influenced hydration between assessments. Furthermore, the relative under-representation of the trunk by whole-body impedance limits our ability to quantify or localise the oedema.5 Consequently, BIVA should be interpreted with acknowledgement of the clinical presentation of the individual. Implications for practice The properties of BIVA (non-invasive, safe, accurate and painless) highlight its potential to assess hydration in palliative care. BIVA could potentially be used to measure hydration in specific clinical scenarios (eg, bowel obstruction, vomiting and diarrhoea) or to monitor the response to interventions (eg, CAH).9 In cancer, BIVA has advantages over regression equations; however, the results should be interpreted carefully with knowledge of the clinical presentation of the patient. Research opportunities In research, BIVA can be used to monitor hydration in studies that evaluate the efficacy of CAH. Furthermore, longitudinal hydration assessments could be conducted at specific phases of illness (eg, to examine the relationship between symptomatic burden and clinical hydration states in the last hours and days of life). Conclusion This case report highlights the potential to use BIVA to monitor hydration states over time in response to interventions. BIVA overcomes the limitations of bioimpedance regression equations. More research is needed to determine the potential of BIVA to improve the evaluation and management of hydration states in advanced cancer. Twitter: Follow Amara Nwosu at @amaranwosu

Conclusion This case report highlights the potential to use BIVA to monitor hydration states over time in response to interventions. BIVA overcomes the limitations of bioimpedance regression equations. More research is needed to determine the potential of BIVA to improve the evaluation and management of hydration states in advanced cancer. Twitter: Follow Amara Nwosu at @amaranwosu Contributors: ACN designed the study, conducted the analysis, interpreted the results and wrote the paper. LM, CM, SM, AP and JE assisted with the data analysis and provided critical review of the final manuscript. Competing interests: None declared. Patient consent: Obtained. Provenance and peer review: Not commissioned; externally peer reviewed. Data sharing statement: The results of this analysis will be available upon request.

Introduction The shortcomings of war-related metaphors for cancer have been discussed in previous research and in the media.1–3 Recent UK policy documents, such as the 2007 NHS Cancer Reform Strategy, have avoided these metaphors in favour of the notion of cancer as a ‘journey’.4 Little is known, however, about how and to what extent patients use these different metaphors for their illness experiences. This paper reports the results of a computer-assisted quantitative and qualitative study of the Violence and Journey metaphors used by patients with cancer in a 500 134-word ‘corpus’ (data set) of contributions to a UK-based online forum. We show how both types of metaphors are used significantly more frequently in this corpus than in a 253 168-word corpus of online writing by health professionals. We then focus on how patients use Violence and Journey metaphors, and show that both types can have empowering as well as disempowering functions. A metaphor involves talking and potentially thinking about one thing in terms of another, on the basis of a perceived similarity between the two. A patient's description of herself as ‘fast becoming a chemo veteran’, for example, suggests the perception of a similarity between the experience of being treated with chemotherapy and the experience of fighting in a war.5

ing about one thing in terms of another, on the basis of a perceived similarity between the two. A patient's description of herself as ‘fast becoming a chemo veteran’, for example, suggests the perception of a similarity between the experience of being treated with chemotherapy and the experience of fighting in a war.5 Metaphors have been found to occur between 3 and 18 times per 100 words.6–8 They are used to talk about abstract, complex, subjective and sensitive experiences in terms of more concrete, simpler, less subjective and less sensitive ones.9 Illness, death and the emotions around them are among the sensitive experiences that are often talked about metaphorically.10 Different metaphors ‘frame’ a topic in different ways, highlighting some aspects and backgrounding others.9 Expressions such as dying after a ‘long battle with cancer’ have become controversial precisely because of the framing that they may impose on the patient's experience: they have associations of violence and threat; they cast the patient in the aggressive role of a fighter; they suggest the presence of an enemy—the disease itself—inside the patient's body; and they associate not recovering with defeat.

precisely because of the framing that they may impose on the patient's experience: they have associations of violence and threat; they cast the patient in the aggressive role of a fighter; they suggest the presence of an enemy—the disease itself—inside the patient's body; and they associate not recovering with defeat. The notion of cancer as a metaphorical journey is arguably the most prominent current alternative to Violence metaphors. The UK's 2007 Cancer Reform Strategy4 contains no instances of ‘battle’ or ‘war’, but includes repeated references to the patient's cancer ‘journey’, with different ‘pathways’ delineated as models of care. Journey metaphors frame the illness experience differently: they potentially cast the illness as a companion to live and travel with; and they do not involve the implication that not recovering amounts to personal failure. While metaphors for cancer have been explored in a variety of contexts,11–13 patients’ spontaneous language use has not been studied systematically. We conducted a large-scale investigation of the Violence and Journey metaphors that patients with cancer use when writing online; we compared their frequencies with those we found in online writing by health professionals; and we considered their implications for empowering or disempowering the patients themselves. We show that Violence metaphors are not necessarily negative and Journey metaphors are not necessarily positive conceptualisations of cancer. The key issue, rather, is the function of a particular framing in a given context.

; and we considered their implications for empowering or disempowering the patients themselves. We show that Violence metaphors are not necessarily negative and Journey metaphors are not necessarily positive conceptualisations of cancer. The key issue, rather, is the function of a particular framing in a given context. Methods Design and setting As part of the project ‘Metaphor in End-of-Life Care’,14 we studied the use of Violence and Journey metaphors in the following two data sets, or ‘corpora’: 500 134 words of online forum contributions by patients on a UK-based website dedicated to cancer; 253 168 words of online forum contributions, blog entries and comments by health professionals on a UK-based website. All material was publicly accessible, but only registered members could contribute to the online fora. For our purposes, the spontaneity of expression allowed by anonymous online writing outweighs the lack of systematic demographic information for contributors.15 16 Sampling and data collection Fifty-six contributors to the patient online forum were included in our corpus according to the following criteria: They described themselves as having received a terminal cancer diagnosis, or discussed palliative or end-of-life care; and They posted at least 50 contributions to the forum between 2007 and 2012.

Sampling and data collection Fifty-six contributors to the patient online forum were included in our corpus according to the following criteria: They described themselves as having received a terminal cancer diagnosis, or discussed palliative or end-of-life care; and They posted at least 50 contributions to the forum between 2007 and 2012. To reduce the data sample to the desired size of approximately 500 000 words, we scaled the patients’ contributions down proportionally according to the total size of each user's contributions, taken from the point at which the patients began discussing end-of-life care (but with a minimum of 1000 words per patient). For comparison purposes, 253 168 words were downloaded from a UK-based website for health professionals, consisting of online forum contributions, blog entries and comments about terminal illness, palliative care and end-of-life care. This data set spanned the period 2008–2013, was primarily concerned with cancer and included contributions from 307 health professionals, most of whom identified themselves as physicians. The data were mass-downloaded using a bespoke computer application, and stored in a format appropriate for exploration by the software tools developed within Corpus Linguistics.17

For comparison purposes, 253 168 words were downloaded from a UK-based website for health professionals, consisting of online forum contributions, blog entries and comments about terminal illness, palliative care and end-of-life care. This data set spanned the period 2008–2013, was primarily concerned with cancer and included contributions from 307 health professionals, most of whom identified themselves as physicians. The data were mass-downloaded using a bespoke computer application, and stored in a format appropriate for exploration by the software tools developed within Corpus Linguistics.17 Analysis A 15 000-word sample from each corpus was analysed manually in order to (1) identify metaphorical expressions according to a well-established procedure,18 and (2) allocate each expression to a ‘semantic field’ corresponding to its literal meaning (eg, ‘veteran’ in ‘a chemo veteran’ was identified as a metaphor and allocated to the semantic field ‘War’). This phase of the analysis was carried out by three team members: the main analyst's codings were independently verified by two other team members to ensure accuracy and consistency.

nding to its literal meaning (eg, ‘veteran’ in ‘a chemo veteran’ was identified as a metaphor and allocated to the semantic field ‘War’). This phase of the analysis was carried out by three team members: the main analyst's codings were independently verified by two other team members to ensure accuracy and consistency. This sample analysis resulted in a list of linguistic expressions and semantic fields to be investigated in the two complete corpora. The online software Wmatrix was used for this purpose.19 20 This tool allowed us to search for (A) all instances of words that we identified as potentially relevant metaphors in the sample analysis (eg, ‘weapon’); and (B) all instances of words that the in-built lexicon categorised under particular semantic fields (eg, the semantic field ‘Warfare’). The latter function is distinctive to our approach to large-scale metaphor analysis.21 The resulting lists of expressions were exported into spreadsheets and coded for metaphoricity by a team member. To ensure accuracy and consistency, three further team members independently verified and agreed on the codings in each spreadsheet.

unction is distinctive to our approach to large-scale metaphor analysis.21 The resulting lists of expressions were exported into spreadsheets and coded for metaphoricity by a team member. To ensure accuracy and consistency, three further team members independently verified and agreed on the codings in each spreadsheet. In this study, we consider the metaphorical expressions in our two complete corpora that relate to the experience of cancer, and that have literal meanings that can be subsumed under the semantic fields of Violence and Journey. The search for potential Violence metaphors was conducted by searching for relevant semantic fields in the data (see (B) above). The search for potential Journey metaphors was conducted by means of a combination of word-level searches (see (A) above) and semantic-field searches (see (B) above), due to the high frequency of movement-related vocabulary in English (eg, ‘through’, ‘going to’).

or relevant semantic fields in the data (see (B) above). The search for potential Journey metaphors was conducted by means of a combination of word-level searches (see (A) above) and semantic-field searches (see (B) above), due to the high frequency of movement-related vocabulary in English (eg, ‘through’, ‘going to’). We first calculated and compared the frequencies of Violence and Journey metaphors in the two data sets. We then investigated in detail the ways in which these metaphors are used by patients to express their experiences. In particular, we considered: (1) what aspects of the patients’ experience are described in terms of Violence or Journey; and (2) the framings provided by different uses of metaphor, particularly in terms of the patient's empowerment and disempowerment in the context of the illness, and the associated emotions. We define empowerment and disempowerment as an increase or decrease in the degree of agency that the patient has, or perceives him/herself to have, as manifest in the metaphors and their co-text. This involves the (perceived) ability to control or react to events for one's own benefit, where this ability is desired by the patient and not externally imposed. Results Quantitative findings Figure 1 provides normalised frequencies per 1000 words of occurrences of relevant Violence metaphors in the Patient corpus and the Health Professional corpus. The higher frequency in the Patient corpus is statistically significant at p<0.0001. Figure 1 Frequencies per 1000 words of Violence metaphors in the Patient corpus and the Health Professional corpus.

Results Quantitative findings Figure 1 provides normalised frequencies per 1000 words of occurrences of relevant Violence metaphors in the Patient corpus and the Health Professional corpus. The higher frequency in the Patient corpus is statistically significant at p<0.0001. Figure 1 Frequencies per 1000 words of Violence metaphors in the Patient corpus and the Health Professional corpus. Figure 2 provides normalised frequencies per 1000 words of occurrences of relevant Journey metaphors in the Patient corpus and the Health Professional corpus. The higher frequency in the Patient corpus is statistically significant at p<0.0001. Figure 2 Frequencies per 1000 words of Journey metaphors in the Patient corpus and the Health Professional corpus. The fact that patients use both types of metaphors significantly more frequently than health professionals raises further questions. This is especially noteworthy considering that healthcare professionals use other types of metaphors as frequently as patients do (eg, Machine metaphors), or even more frequently (eg, Openness metaphors). Qualitative findings As summarised in table 1, patients can use both Violence and Journey metaphors in empowering and disempowering ways. Table 1 Summary of empowering and disempowering Violence and Journey metaphors

The fact that patients use both types of metaphors significantly more frequently than health professionals raises further questions. This is especially noteworthy considering that healthcare professionals use other types of metaphors as frequently as patients do (eg, Machine metaphors), or even more frequently (eg, Openness metaphors). Qualitative findings As summarised in table 1, patients can use both Violence and Journey metaphors in empowering and disempowering ways. Table 1 Summary of empowering and disempowering Violence and Journey metaphors Violence scenarios Journey scenarios Patient as disempowered Disease fighting the patient Patient unsuccessfully fighting the disease Treatment fighting the patient Patient as a traveller on a difficult journey Patient travelling without control over their journey Patient as empowered Patient successfully fighting the disease Patient successfully fighting health professionals Mutual encouragement and solidarity Patient as a traveller in charge of the journey Patients as travelling companions Qualitative findings: Violence metaphors Patients’ Violence metaphors can be used in both empowering and disempowering ways. Violence metaphors and disempowerment Patients’ Violence metaphors suggest several kinds of scenarios in which the patient appears in a disempowered position.

Violence scenarios Journey scenarios Patient as disempowered Disease fighting the patient Patient unsuccessfully fighting the disease Treatment fighting the patient Patient as a traveller on a difficult journey Patient travelling without control over their journey Patient as empowered Patient successfully fighting the disease Patient successfully fighting health professionals Mutual encouragement and solidarity Patient as a traveller in charge of the journey Patients as travelling companions Qualitative findings: Violence metaphors Patients’ Violence metaphors can be used in both empowering and disempowering ways. Violence metaphors and disempowerment Patients’ Violence metaphors suggest several kinds of scenarios in which the patient appears in a disempowered position. The disease fighting the patient When Violence metaphors express patients’ perception of their illness, cancer can be described as ‘attacking from the inside’ and ‘invading’ the body. A patient describes her breast cancer as a ‘killer’ that ‘strangles and shocks your soul’. In such cases, the disease is presented as an aggressive opponent, while the patient is in a passive position. A particularly strong sense of vulnerability is expressed by patients who describe themselves as ‘time bombs’ while in remission.

escribes her breast cancer as a ‘killer’ that ‘strangles and shocks your soul’. In such cases, the disease is presented as an aggressive opponent, while the patient is in a passive position. A particularly strong sense of vulnerability is expressed by patients who describe themselves as ‘time bombs’ while in remission. The patient unsuccessfully fighting the disease Violence metaphors are also used to express patients’ attempts to recover or to prolong their lives. When their condition does not improve, patients describe themselves as unsuccessful fighters. In some cases, this is because they are not prescribed the most effective treatment, as in: “it must be dispiriting when you are battling as hard as you can, not to be given the armour to fight in”. In other cases, lack of recovery is described in terms of defeat, as in: “I feel such a failure that I am not winning this battle”. This patient blames herself for the failure of her treatment, thus adding feelings of guilt to the emotional consequences of facing the terminal phase of the disease. The treatment fighting the patient Patients also use Violence metaphors to express their perception of the effects of cancer treatment. Chemotherapy is described as giving the patient's body ‘a hammering’ or ‘a battering’. One patient talks about her ‘normal little cells’ being ‘hit by a sledgehammer of both toxic chemicals and radiation’. These metaphors suggest a perception of vulnerability and passivity in relation to cancer treatment that is similar to what is expressed for the disease itself.

t's body ‘a hammering’ or ‘a battering’. One patient talks about her ‘normal little cells’ being ‘hit by a sledgehammer of both toxic chemicals and radiation’. These metaphors suggest a perception of vulnerability and passivity in relation to cancer treatment that is similar to what is expressed for the disease itself. Violence metaphors and empowerment Violence metaphors are also used by patients to express a perception of themselves as engaged and effective agents within the illness experience. The patient successfully fighting the disease Some patients describe themselves as ‘fighters’ in ways that suggest agency and pride, as in “I am such a fighter” and “my consultants recognised that I was a born fighter”. One patient attributes her ‘desire to fight and win’ to the fact that she is young and has a family. Another thanks other forum contributors for restoring her ‘fighting spirit’, so that she is ‘ready to kick some cancer butt’. The patient successfully fighting health professionals Some patients employ Violence metaphors to describe their interactions with health professionals. A patient describes a successful outcome in a consultation as ‘winning that battle’, while another uses the expression ‘twin attack’ to refer to how two family members managed to obtain a medical appointment. After expressing dissatisfaction with the stitching of her wound, another patient comments that now she has ‘another thing to beat my surgeon up about’.

in a consultation as ‘winning that battle’, while another uses the expression ‘twin attack’ to refer to how two family members managed to obtain a medical appointment. After expressing dissatisfaction with the stitching of her wound, another patient comments that now she has ‘another thing to beat my surgeon up about’. Mutual encouragement and solidarity Violence metaphors are sometimes used by forum contributors to encourage and motivate one another. Some patients praise others for being ‘fighters’ and for ‘winning the battle’ against cancer. Expressions such as ‘Soldier on everybody’ occur at the end of some posts. Contributors to one thread (including both women and men) jokingly use army titles such as ‘Captain’ for one another; one particular patient says that she would ‘promote’ another if he had not already ‘reached top rank’. Although this part of our study was qualitative in nature, 42 out of 100 randomly selected Violence metaphors were found to be used in a potentially empowering way, and 38 in a potentially disempowering way (the remaining 20 instances did not clearly fall under either category). Qualitative findings: Journey metaphors Like Violence metaphors, Journey metaphors are used by patients in our data in both empowering and disempowering ways. Journey metaphors and empowerment Journey metaphors work in potentially empowering ways when they are used to convey a sense of purpose, control and companionship.

Although this part of our study was qualitative in nature, 42 out of 100 randomly selected Violence metaphors were found to be used in a potentially empowering way, and 38 in a potentially disempowering way (the remaining 20 instances did not clearly fall under either category). Qualitative findings: Journey metaphors Like Violence metaphors, Journey metaphors are used by patients in our data in both empowering and disempowering ways. Journey metaphors and empowerment Journey metaphors work in potentially empowering ways when they are used to convey a sense of purpose, control and companionship. The patient as a traveller in charge of the journey Some patients use Journey metaphors to express a sense of control over their cancer experience, and occasionally to point out some positive aspects of being ill, as in: “My journey may not be smooth but it certainly makes me look up and take notice of the scenery!” Another patient points out that, even when he thinks he has “gone as far as” he can, he has a “happy surprise by being able to push myself that little bit extra”.

ally to point out some positive aspects of being ill, as in: “My journey may not be smooth but it certainly makes me look up and take notice of the scenery!” Another patient points out that, even when he thinks he has “gone as far as” he can, he has a “happy surprise by being able to push myself that little bit extra”. Patients as travelling companions Patients often use Journey metaphors to express a sense of group solidarity among forum users, and to encourage one another. This involves Journey scenarios in which patients travel together, or where patients with earlier diagnoses ‘guide’ those with more recent diagnoses. One patient comments that “rocks in our paths are easier to handle when we're all in it together”, and that “the best people to help you are the ones who've been there before or are heading there with you”. Journey metaphors are also used as mutually encouraging greetings, as in ‘Safe journey’. Journey metaphors and disempowerment Journey metaphors are also used by patients in our data to express a sense of lack of acceptance of or control over their situation.

Patients as travelling companions Patients often use Journey metaphors to express a sense of group solidarity among forum users, and to encourage one another. This involves Journey scenarios in which patients travel together, or where patients with earlier diagnoses ‘guide’ those with more recent diagnoses. One patient comments that “rocks in our paths are easier to handle when we're all in it together”, and that “the best people to help you are the ones who've been there before or are heading there with you”. Journey metaphors are also used as mutually encouraging greetings, as in ‘Safe journey’. Journey metaphors and disempowerment Journey metaphors are also used by patients in our data to express a sense of lack of acceptance of or control over their situation. The patient as a traveller on a difficult journey Some patients use Journey metaphors to emphasise the overwhelming difficulties they face as cancer sufferers. One patient comments that the journey has “many twists and turns that means that no two people go exactly the same route”, and adds that having cancer “is like trying to drive a coach and horses uphill with no back wheels on the coach”. Another comments that she has ‘not done so well’ on her ‘‘cancer journey’ through the wilderness’ of her local hospital.

has “many twists and turns that means that no two people go exactly the same route”, and adds that having cancer “is like trying to drive a coach and horses uphill with no back wheels on the coach”. Another comments that she has ‘not done so well’ on her ‘‘cancer journey’ through the wilderness’ of her local hospital. The patient travelling without control over their journey Many patients express the problems they have accepting their situation by describing themselves as travelling against their will. One patient talks about a ‘reluctant journey’, while another wonders how she can ‘navigate this road’ that she does ‘not even want to be on’. Some patients describe themselves as being on a journey that they cannot control. One forum user talks about other patients as “passengers” who are ‘nearing the end of their journey’ or have ‘finished their journey’. In a sample of 100 randomly selected Journey metaphors, 26 were used in a potentially empowering way, while 39 suggested that the patient was in a disempowered position (the remaining 35 instances did not clearly fall under either category). Discussion To the best of our knowledge, this is the first large-scale study of the use of Violence and Journey metaphors by patients with cancer, and the first systematic comparison of differences in frequency of use between patients and health professionals.

In a sample of 100 randomly selected Journey metaphors, 26 were used in a potentially empowering way, while 39 suggested that the patient was in a disempowered position (the remaining 35 instances did not clearly fall under either category). Discussion To the best of our knowledge, this is the first large-scale study of the use of Violence and Journey metaphors by patients with cancer, and the first systematic comparison of differences in frequency of use between patients and health professionals. The higher frequency of both types of metaphor in the patient data suggests that they are helpful in expressing different aspects of the patients’ experience. The fact that health professionals use Violence metaphors less frequently may reflect an awareness of their potential shortcomings, and, more generally, of the importance of using appropriate and sensitive metaphors in relation to illness. As for function, our study confirms that Violence and Journey metaphors facilitate different ways of framing the patients’ experiences. Violence metaphors present the experience as an antagonistic one, in which the patient faces an opponent, whether this be the illness, the treatment, health professionals, etc. Violence metaphors may therefore both reflect and reinforce an adversarial approach to the cancer experience, as well as feelings of vulnerability, passivity, impending threat and, most negatively, personal failure if the disease is found to be incurable. In contrast, Journey metaphors can present the experience of illness as a process that is shared by others with similar diagnoses, or with family and friends. The patient can therefore take on a role that is active without being oppositional, while setbacks are not necessarily as catastrophic and irreversible as military defeats or bomb explosions.

can present the experience of illness as a process that is shared by others with similar diagnoses, or with family and friends. The patient can therefore take on a role that is active without being oppositional, while setbacks are not necessarily as catastrophic and irreversible as military defeats or bomb explosions. However, our data also show that the specific framings provided by both Violence and Journey metaphors vary depending on who uses them and how. This variation relates particularly to: the degree of agency that the patient feels they have to act in their own interest; the extent to which this degree of agency is welcome; and the associated emotions. Some uses of Violence metaphors suggest that some patients find meaning, purpose and a sense of pride and identity in approaching the illness experience as a fight. Conversely, Journey metaphors can be used to convey feelings of passivity, lack of acceptance and pessimism. Overall, therefore, Violence metaphors are not always negative, while Journey metaphors are not always positive. Metaphor use should be evaluated on the basis of its empowering or disempowering function, and associated emotions, in particular contexts.

However, our data also show that the specific framings provided by both Violence and Journey metaphors vary depending on who uses them and how. This variation relates particularly to: the degree of agency that the patient feels they have to act in their own interest; the extent to which this degree of agency is welcome; and the associated emotions. Some uses of Violence metaphors suggest that some patients find meaning, purpose and a sense of pride and identity in approaching the illness experience as a fight. Conversely, Journey metaphors can be used to convey feelings of passivity, lack of acceptance and pessimism. Overall, therefore, Violence metaphors are not always negative, while Journey metaphors are not always positive. Metaphor use should be evaluated on the basis of its empowering or disempowering function, and associated emotions, in particular contexts. Strengths and weaknesses of the study The use of online forum data allowed us to investigate the language that patients use in an informal and supportive environment, in which they are able to control how much information they disclose about themselves. By combining manual and computer-aided analysis, we identified a larger number of Violence and Journey metaphors than is possible by more traditional approaches.

e language that patients use in an informal and supportive environment, in which they are able to control how much information they disclose about themselves. By combining manual and computer-aided analysis, we identified a larger number of Violence and Journey metaphors than is possible by more traditional approaches. However, our data only represent people who are comfortable with writing about their experiences online. Moreover, the lack of demographic information for forum contributors limits our ability to generalise our findings. Finally, our suggestions about the extent to which different uses of metaphor are empowering or disempowering are based on textual analysis alone, and cannot be confirmed by interviewing the patients or by considering the stage and seriousness of their condition. Our reliance on patients’ own self-descriptions also means that we cannot verify their diagnoses or disease status.

t uses of metaphor are empowering or disempowering are based on textual analysis alone, and cannot be confirmed by interviewing the patients or by considering the stage and seriousness of their condition. Our reliance on patients’ own self-descriptions also means that we cannot verify their diagnoses or disease status. Meaning of the study: implications for health professionals and policymakers Our study provides enough evidence of the disadvantages and potential disempowering effects of Violence metaphors to support their exclusion from policy documents, and to discourage physicians from introducing them in interactions with patients. It can be particularly harmful for patients to have the role of ‘fighter’ imposed on them by external pressures, whether from relatives, health professionals, charity campaigns or a more general sense that refusing to ‘fight’ suggests a lack of determination and moral fibre. Our study also confirms that Journey metaphors can be employed to suggest a positive, empowering approach to the cancer experience, in which the patient feels a sense of companionship with others and can choose the degree of control he or she wishes to have in the decisions and processes that affect them. In this sense, the adoption of the metaphor of a ‘cancer journey’ can be appropriate and effective.

empowering approach to the cancer experience, in which the patient feels a sense of companionship with others and can choose the degree of control he or she wishes to have in the decisions and processes that affect them. In this sense, the adoption of the metaphor of a ‘cancer journey’ can be appropriate and effective. However, our study also shows that Violence metaphors are not by default negative and Journey metaphors are not by default a positive means of talking and thinking about cancer. Patients frequently use Violence metaphors in ways that seem to empower and motivate them, while their use of Journey metaphors can sometimes indicate a sense of disempowerment. Furthermore, patients also use Violence metaphors to describe their perception of difficulties and problems in the healthcare system, which may need to be addressed in the provision of healthcare. Hence, a blanket rejection of Violence metaphors would deprive some patients of the positive functions that these metaphors can have, while an uncritical promotion of Journey metaphors overlooks the negative ways in which they can be used.

ealthcare system, which may need to be addressed in the provision of healthcare. Hence, a blanket rejection of Violence metaphors would deprive some patients of the positive functions that these metaphors can have, while an uncritical promotion of Journey metaphors overlooks the negative ways in which they can be used. We argue therefore that metaphor use should not be assessed only on the basis of type (Violence or Journey), but on the basis of its function (empowering or disempowering, and emotional associations). This has implications for training and practice in healthcare communication. By developing the skills of noticing and attending to patients’ metaphors, health professionals can be in a better position to question metaphors that seem to have negative, disempowering effects, and to accept or even adopt metaphors that seem to work in positive, empowering ways. Unanswered questions and future research Further research is needed to identify variation in metaphor use and appropriateness for different groups of patients, depending on age, gender, cultural background, illness, stage of disease and so on. This would require the use of a data collection method that includes detailed demographic information. Future work is also needed to gather patients’ own perceptions of the metaphors they use and are exposed to during their illness experience.

ng on age, gender, cultural background, illness, stage of disease and so on. This would require the use of a data collection method that includes detailed demographic information. Future work is also needed to gather patients’ own perceptions of the metaphors they use and are exposed to during their illness experience. In conclusion, patients with cancer use both Violence and Journey metaphors, and do so more frequently than health professionals. Both Violence and Journey metaphors can suggest empowerment as well as disempowerment, depending on who uses them and how. It follows that communication in cancer care could benefit from a focus on the function (empowering or disempowering) of individual metaphor uses rather than on their type (Violence or Journey). Twitter: Follow Elena Semino at @elenasemino Contributors: ES co-planned and co-conducted the research, and took charge of writing the paper. She is responsible for the content as guarantor. ZD and JD co-conducted the research and contributed to the writing of the paper. VK co-planned and co-conducted the research, and contributed to the writing of the paper. SP co-planned the research and contributed to the writing of the paper. AH and PR co-planned the research, conducted the data collection and commented on the paper. Funding: Economic and Social Research Council; grant numbers: ES/J007927/1; ES/K002155/1). Competing interests: None.

Contributors: ES co-planned and co-conducted the research, and took charge of writing the paper. She is responsible for the content as guarantor. ZD and JD co-conducted the research and contributed to the writing of the paper. VK co-planned and co-conducted the research, and contributed to the writing of the paper. SP co-planned the research and contributed to the writing of the paper. AH and PR co-planned the research, conducted the data collection and commented on the paper. Funding: Economic and Social Research Council; grant numbers: ES/J007927/1; ES/K002155/1). Competing interests: None. Ethics approval: All data analysed in the study were obtained from publicly accessible online fora. The organisation in charge of the patient online forum was informed of our data collection process. Research ethics approval for the study was obtained from Lancaster University, Lancaster, UK. Provenance and peer review: Not commissioned; externally peer reviewed.

Background Technology is increasingly being integrated into medicine to support new opportunities for the delivery of clinical practice, education and research.1 Podcasts are episodic digital audio recordings that are downloaded through web syndication or streamed online.2 Research demonstrates that podcast listenership is increasing.3–5 The percentage of Americans who have listened to a podcast has increased from 9% to 17% between 2008 and 2015.6 Podcasts are increasingly being used to support medical education.7–10 Palliative care podcasts are available;11 these include ‘Get Palliative Care’ (by the Center to Advance Palliative Care—CAPC),12 the ‘CAPC Palliative Care Podcast’13 and the ‘Hospice of the Bluegrass Podcast’.14 However, there are no published studies about the use of podcasts in palliative care. Podcasts can potentially be used to facilitate communication about palliative care with researchers, policymakers and the public.1 Further study is needed to evaluate the utility of podcasts to facilitate knowledge-transfer about subjects related to palliative care. The aims of this article are to: Describe the development of a palliative care podcast according to international recommendations for podcast quality. To analyse the listenership of the podcast over a 14-month period. Methods The development of the podcast involved defining the scope and focus of the podcast; developing an infrastructure; identifying quality indicators of podcast quality; designing content; coordinating dissemination and analysing data.

The aims of this article are to: Describe the development of a palliative care podcast according to international recommendations for podcast quality. To analyse the listenership of the podcast over a 14-month period. Methods The development of the podcast involved defining the scope and focus of the podcast; developing an infrastructure; identifying quality indicators of podcast quality; designing content; coordinating dissemination and analysing data. Scope and focus The podcast was aimed at healthcare professionals with an interest in palliative care, technology and innovation. The podcast method was chosen for its effectiveness, popularity and accessibility.7 Infrastructure development A portable audio recorder and microphone (total cost=£50) was purchased with funds from an educational grant. SoundCloud, a popular audio streaming website, was chosen to host the podcast (https://soundcloud.com/mypal). The website was accessible online and also has native applications for mobile devices (Android and iOS). An online blog was developed for the podcast (http://amaranwosu.com/amipal/) to facilitate dissemination and provide links to references presented in the podcast.

to host the podcast (https://soundcloud.com/mypal). The website was accessible online and also has native applications for mobile devices (Android and iOS). An online blog was developed for the podcast (http://amaranwosu.com/amipal/) to facilitate dissemination and provide links to references presented in the podcast. Quality indicators Quality indicators for medical education podcasts and blogs have been developed.15 These indicators were developed using a modified Delphi consensus of international healthcare professional educators. The indicators with ≥90% consensus (table 1) consist of 13 items (10 of which are relevant to podcasts) within themes that include: content, credibility, bias, transparency, academic rigour, functionality, use of resources, orientation and professionalism. These quality indicators were used to inform the podcast development. Table 1 Quality indicators for medical education podcasts and blogs as recommended by Lin et al15

Quality indicators Quality indicators for medical education podcasts and blogs have been developed.15 These indicators were developed using a modified Delphi consensus of international healthcare professional educators. The indicators with ≥90% consensus (table 1) consist of 13 items (10 of which are relevant to podcasts) within themes that include: content, credibility, bias, transparency, academic rigour, functionality, use of resources, orientation and professionalism. These quality indicators were used to inform the podcast development. Table 1 Quality indicators for medical education podcasts and blogs as recommended by Lin et al15 Per cent consensus Quality indicator Domain/subtheme How this was met Podcasts Blogs Do the authorities (eg, author, editor, publisher) that created the resource list their conflicts of interest? Credibility/bias There was no conflict of interest. 100 100 Is the information presented in the resource accurate? Credibility/academic rigour References were provided for the podcast content. 100 94 Is the identity of the resource's author clear? Credibility/transparency The blog and podcast included details of the affiliation and qualifications of ACN. 95 95 Does the resource make a clear distinction between fact and opinion? Credibility/bias The podcast and blog provided details of what constituted fact and opinion. References were provided for the podcast content. 95 95 Does the resource employ technologies that are universally available to allow learners with standard equipment and software access? Design/functionality The podcast was accessible using standard technologies (computer and mobiles devices) without the requirement of additional software or payment. 94 – Does the resource clearly differentiate between advertisement and content? Credibility/bias The podcast was freely available and was produced without commercial funding or advertising. 90 95 Is the resource transparent about who was involved in its creation? Credibility/transparency Podcast production was performed by ACN. Contributions of others were clearly acknowledged. 90 91 Is the content of this educational resource of good quality? Content The podcasts were edited to enhance audio quality. 90 91 Is the content of the resource professional? Content/professionalism Each episode was planned and researched in advance to ensure the content was accurate and professional. 90 91 Is the resource useful and relevant for its intended audience? Content/orientation The podcast format consisted of interviews, opinion pieces and education-focused activity. The podcast was aimed at palliative care professionals who were familiar with social media. 90 91 Does the resource cite its references? Credibility/use of other resources References were provided for the podcast content.

ntation The podcast format consisted of interviews, opinion pieces and education-focused activity. The podcast was aimed at palliative care professionals who were familiar with social media. 90 91 Does the resource cite its references? Credibility/use of other resources References were provided for the podcast content. – 93 Are the resources consistent with its references? Credibility/use of other resources References were provided for the podcast content. – 93 Is the author well qualified to provide information on the topic? Credibility/transparency The blog and podcast included details of the affiliation and qualifications of ACN. – 91 Content design The podcast was named AmiPal (previously MyPal), reflecting the name of the corresponding author and subject of Palliative Care. The format involved interviews, opinion pieces and education-focused content. The topics covered are presented in table 2. Podcasts were edited using Audacity (http://www.audacityteam.org), a free open-source, cross-platform audio-editing tool. Table 2 Topics covered in AmiPal podcasts since January 2015

– 93 Are the resources consistent with its references? Credibility/use of other resources References were provided for the podcast content. – 93 Is the author well qualified to provide information on the topic? Credibility/transparency The blog and podcast included details of the affiliation and qualifications of ACN. – 91 Content design The podcast was named AmiPal (previously MyPal), reflecting the name of the corresponding author and subject of Palliative Care. The format involved interviews, opinion pieces and education-focused content. The topics covered are presented in table 2. Podcasts were edited using Audacity (http://www.audacityteam.org), a free open-source, cross-platform audio-editing tool. Table 2 Topics covered in AmiPal podcasts since January 2015 Topic Focus Length Date published Introduction and welcome to the new podcast Opinion 12:02 Jan 2015 Research and innovation Opinion 17:22 Jan 2015 Integrated clinical academic training Article overview 6:13 Jan 2015 Nanotechnology to monitor cancer Opinion 9:34 Jan 2015 3D printing in clinical practice Opinion 7:15 Jan 2015 Publishing in palliative care Education 15:19 Feb 2015 Is there too much technology in healthcare Article overview 14:55 Feb 2015 Peer led learning in palliative care Article overview 5:35 Mar 2015 Palliative care day therapy Interview 21:42 Mar 2015 Undergraduate medical education in palliative care Interview 15:31 Mar 2015 Bioelectrical impedance analysis to assess hydration in advanced cancer Education 6:14 Mar 2015 Culture and palliative care Opinion 16:27 May 2015 Wearable technology in healthcare—can palliative care benefit? Opinion 14:10 Jun 2015 Five apps for clinical academics Education 16:40 Jun 2015 Social media and palliative care Article overview 4:10 Sep 2015 Technology in the delivery of healthcare: patient power in medicine Article overview 3:44 Nov 2015 What makes a good case-based discussion? Interview Education 5:37 Dec 2015 Virtual reality and palliative care Opinion 5:48 Feb 2016 Renal medicine and palliative care Interview 3:36 Feb 2016 A comparison between studies: research, audit and service evaluation Education 2:22 Feb 2016 Dissemination The podcasts were released episodically under the ‘Science and Medicine’ category on the SoundCloud website. The podcast's Rich Site Summary (RSS) feed was registered with podcast repositories, including iTunes (http://www.apple.com/itunes), Stitcher (https://www.stitcher.com), TuneIn (http://tunein.com) and Acast (https://www.acast.com). The RSS feed enabled users to access the podcast via a computer or mobile device. Each episode was promoted on social media using palliative medicine hashtags.16 Widgets (stand-alone embeddable web applications) were embedded into the blog and social media posts, which enabled the podcasts to be directly played.

.com). The RSS feed enabled users to access the podcast via a computer or mobile device. Each episode was promoted on social media using palliative medicine hashtags.16 Widgets (stand-alone embeddable web applications) were embedded into the blog and social media posts, which enabled the podcasts to be directly played. Analysis and feedback Feedback to each episode was possible using email communication and social media. Additionally, healthcare professionals (in Merseyside, UK) were contacted by email and were encouraged to provide feedback. The listenership analysis was conducted using the SoundCloud analytics tools. Data were analysed for frequency of plays and geographical location. Results Twenty podcasts were developed between January 2015 and February 2016. The cumulative total of podcast plays was 3036, an average of 217 monthly plays (table 3 and figure 1). The RSS feed was the most popular way to access the podcast (n=1937; 64%). Between January and September 2015, the podcast was most accessed via the SoundCloud website. However, from October 2015, the cumulative RSS feed plays were higher. The mean duration of each podcast was 10 min (range 3–21 min). The podcast was listened to in 68 different countries (table 4) and was most popular in English-speaking areas; specifically, the USA (n=1372, 45.2%), UK (n=661, 21.8%) and Canada (n=221, 7.3%). Table 3 Number or times the AmiPal podcast was played, via the web and RSS feed options, between January 2015 and February 2016

Results Twenty podcasts were developed between January 2015 and February 2016. The cumulative total of podcast plays was 3036, an average of 217 monthly plays (table 3 and figure 1). The RSS feed was the most popular way to access the podcast (n=1937; 64%). Between January and September 2015, the podcast was most accessed via the SoundCloud website. However, from October 2015, the cumulative RSS feed plays were higher. The mean duration of each podcast was 10 min (range 3–21 min). The podcast was listened to in 68 different countries (table 4) and was most popular in English-speaking areas; specifically, the USA (n=1372, 45.2%), UK (n=661, 21.8%) and Canada (n=221, 7.3%). Table 3 Number or times the AmiPal podcast was played, via the web and RSS feed options, between January 2015 and February 2016 Number of times AmiPal podcast was played (n) Year Month Web only Web only cumulative RSS only RSS only cumulative Monthly total (web + RSS) Total cumulative 2015 Jan 71 71 0 0 71 71 Feb 84 155 0 0 84 155 Mar 344 499 0 0 344 499 Apr 144 643 0 0 144 643 May 61 704 143 143 204 847 Jun 66 770 55 198 121 968 Jul 25 795 241 439 266 1234 Aug 34 829 107 546 141 1375 Sep 56 885 201 747 257 1632 Oct 30 915 195 942 225 1857 Nov 34 949 217 1159 251 2108 Dec 56 1005 183 1342 239 2347 2016 Jan 29 1034 197 1539 226 2573 Feb 65 1099 398 1937 463 3036 Table 4 Top 10 geographical locations for AmiPal podcast listeners

70 55 198 121 968 Jul 25 795 241 439 266 1234 Aug 34 829 107 546 141 1375 Sep 56 885 201 747 257 1632 Oct 30 915 195 942 225 1857 Nov 34 949 217 1159 251 2108 Dec 56 1005 183 1342 239 2347 2016 Jan 29 1034 197 1539 226 2573 Feb 65 1099 398 1937 463 3036 Table 4 Top 10 geographical locations for AmiPal podcast listeners Position Country Number of podcast plays (%) 1 USA 1372 (45.2) 2 UK 661 (21.8) 3 Canada 221 (7.3) 4 Australia 217 (7.1) 5 Brazil 164 (5.4) 6 New Zealand 69 (2.3) 7 Germany 38 (1.3) 8 India 26 (0.9) – The Netherlands 26 (0.9) 9 Ireland 20 (0.7) 10 Malaysia 17 (0.6) – Fifty-seven other countries 205 (6.8) Figure 1 Line chart displaying the total number of times the AmiPal podcast was listened to between January 2015 and February 2016 via the SoundCloud web and Rich Site Summary-feed options. A small amount of feedback was received (10 responses); overall, this was positive. The podcast was modified in response to the feedback with changes to the audio quality, style and format. Specifically, the podcast length shortened to <6 min (evident from the last six podcasts) and backing music was added to improve the rhythmic flow of the audio. Discussion Summary This analysis demonstrated that the AmiPal palliative care podcast had a wide geographical reach with the majority of listeners originating from Western English-speaking countries.

A small amount of feedback was received (10 responses); overall, this was positive. The podcast was modified in response to the feedback with changes to the audio quality, style and format. Specifically, the podcast length shortened to <6 min (evident from the last six podcasts) and backing music was added to improve the rhythmic flow of the audio. Discussion Summary This analysis demonstrated that the AmiPal palliative care podcast had a wide geographical reach with the majority of listeners originating from Western English-speaking countries. Strengths and uniqueness of this study This is the first study that describes the development and analysis of a palliative care podcast that was developed according to relevant quality indicators. The podcast was free and accessible across a range of computer and mobile platforms.9 The data of the geographical reach of the podcast provide evidence of the potential of this medium to facilitate international dissemination.

is of a palliative care podcast that was developed according to relevant quality indicators. The podcast was free and accessible across a range of computer and mobile platforms.9 The data of the geographical reach of the podcast provide evidence of the potential of this medium to facilitate international dissemination. Comparison with previous work Previous studies have highlighted potential to use technology to inform education and dissemination in palliative care.1 This study adds to evidence from other work, which have used podcasts in medical education.8 10 17 The podcasts were accessed and played several months after release, which may suggest that new listeners were acquired over-time, and/or the archive was used ‘on-demand’. These findings are consistent with previous work, which reports how podcasts provide a repository of information that can be continually accessed.2 18 The majority of podcasts (64%) were accessed via the RSS feed, which may suggest the use of mobile devices. This finding is consistent with the findings of USA and UK research, which demonstrates that two-thirds of podcasts are accessed on a mobile device rather than a computer.4 19 The podcast listenership was similar to the CAPC podcast, which (at the time of writing) has a total of 3831 listens from its 12 episodes over the past 24 months. In 2015, CAPC's public facing ‘Get Palliative Care’ podcast series obtained 14 318 listens from 10 podcasts about the patient journey. This highlights the potential interest for podcasts reporting the patient narrative.

h (at the time of writing) has a total of 3831 listens from its 12 episodes over the past 24 months. In 2015, CAPC's public facing ‘Get Palliative Care’ podcast series obtained 14 318 listens from 10 podcasts about the patient journey. This highlights the potential interest for podcasts reporting the patient narrative. Limitations The lack of plays from the RSS feed in the first 4 months was due to a delay in the RSS feed being available. Consequently, the potential reach of the podcast in these months was lower. It is likely that the overall proportion of RSS feed plays would have been higher, if the RSS feed was available for the entire period. It is likely that the majority of the RSS feed plays were from mobile devices; however, we cannot ascertain the exact number (as the RSS feed may have been accessed by computer). Furthermore, it is not possible to know whether users listened to the entire podcast or not. Although the podcast was available across a range of computer and mobile devices, there may be some technological challenges to accessing the podcast in some healthcare organisations and resource-poor settings (eg, old internet browsers, web-filtering issues, wireless internet coverage).

d to the entire podcast or not. Although the podcast was available across a range of computer and mobile devices, there may be some technological challenges to accessing the podcast in some healthcare organisations and resource-poor settings (eg, old internet browsers, web-filtering issues, wireless internet coverage). Very little feedback was received through the email and social media feedback options. A possible explanation, presented by experts in medical education, may be that the listeners did not place importance on interacting with the podcast host.15 Listeners may personally reflect on the podcast topics without feeling the need to communicate their reflections with the host. Consequently, it is not possible to determine if listeners found the podcasts beneficial. Furthermore, our knowledge of the listenership is relatively unknown, as listeners were not required to provide information or login to access content.

opics without feeling the need to communicate their reflections with the host. Consequently, it is not possible to determine if listeners found the podcasts beneficial. Furthermore, our knowledge of the listenership is relatively unknown, as listeners were not required to provide information or login to access content. Implications to practice It is possible to develop a palliative care podcast that has a global reach. Audio recording equipment is available for relatively low cost,20 and many mobile devices contain microphones to record audio.21 Audio hosting sites (eg, SoundCloud.com, Podomatic.com) and open-source audio editing software are freely available (eg, Audacity).20 21 Individuals and organisations planning on developing their own podcasts can use quality indicators15 22 to develop content and social media to enhance dissemination.16 20 If wide dissemination of the podcast is intended, the RSS feed should be registered with podcast databases and social media should be used for promotion.

viduals and organisations planning on developing their own podcasts can use quality indicators15 22 to develop content and social media to enhance dissemination.16 20 If wide dissemination of the podcast is intended, the RSS feed should be registered with podcast databases and social media should be used for promotion. Future opportunities and research possibilities Organisations may consider developing podcasts for specific purposes, such as education, lecture capture and research dissemination. Future studies are needed to determine whether palliative care podcasts can facilitate learning for professionals and lay people. Further work can examine the demographics of listeners (eg, using analytics software and surveys) and evaluate learning outcomes of podcasts using of pre and post assessments; this will help to plan priorities for content, quality and to evaluate the impact (eg, for learning and clinical practice) of podcasts. Developed content can be incorporated within the dissemination strategy of institutions, in order to meet learning styles of listeners. Future work can also consider the needs of individuals with hearing deficits (eg, via subtitle video). Conclusions Podcasts can be used to facilitate palliative care discussion with a global audience. Podcasts offer the potential to develop educational content and promote research dissemination. Future studies should focus on information development, quality metrics and impact analysis of educational podcasts, as this form of digital communication is likely to increase and engage wider society.

th a global audience. Podcasts offer the potential to develop educational content and promote research dissemination. Future studies should focus on information development, quality metrics and impact analysis of educational podcasts, as this form of digital communication is likely to increase and engage wider society. Ethics This project did not constitute research. Therefore, ethics committee approval was not required. Twitter: Follow Amara Nwosu at @amaranwosu Contributors: ACN designed study, conducted the analysis, interpreted the results and wrote the paper. DM, VLR and LC assisted with the data analysis and provided critical review of the final manuscript. Funding: This work was supported by an educational grant from the Friends of the University of Liverpool. Competing interests: None declared. Provenance and peer review: Not commissioned; externally peer reviewed. Data sharing statement: The data set for the study is available on request.

Background It is well established that decision making about the development of specialist palliative care (SPC) services, if it is to meet the needs, preferences and priorities of local communities, should be underpinned by accurate and current information characterising existing service provision. This includes the type of services available, their geographic range and their organisational capacity. A 2015 report by the Scottish Parliament identified a significant ‘information deficit’ concerning the provision of SPC in Scotland.1 In response, a Scottish Atlas of Palliative Care was produced in 2016, which described, at national and territorial health board levels, the provision of SPC services in Scotland.2 While international benchmarking studies3 4 have described the development and organisation of SPC services across the whole WHO European Region as well as the countries of the European Union (EU), it has not been possible to profile Scotland in these, since they include the UK as a whole, and not its constituent countries, as the unit of analysis. The data required to profile Scotland against the UK as well as other EU member states using ranking and coverage models was collected for the first time in the Scottish Atlas. We now use the primary data collected for the Scottish Atlas to establish the level of SPC provision in Scotland as compared with other EU countries and to express this as a measure of coverage for specific service types based on expert guidelines.5 6

erage models was collected for the first time in the Scottish Atlas. We now use the primary data collected for the Scottish Atlas to establish the level of SPC provision in Scotland as compared with other EU countries and to express this as a measure of coverage for specific service types based on expert guidelines.5 6 Aims The aims of this study were to describe, at national level, the provision of SPC services in Scotland, to compare these with other countries in the EU and to assess the coverage of SPC services in Scotland based on the recommendations of a European Association for Palliative Care (EAPC) ‘White Paper’ on standards and norms for hospice and palliative care in Europe.

ional level, the provision of SPC services in Scotland, to compare these with other countries in the EU and to assess the coverage of SPC services in Scotland based on the recommendations of a European Association for Palliative Care (EAPC) ‘White Paper’ on standards and norms for hospice and palliative care in Europe. Methods In Scotland, healthcare is provided free at the point of access to all citizens through the National Health Service (NHS), which is organised into 14 territorial health boards. SPC services are also provided by independent hospices, which are charitable organisations that receive up to 50% of their agreed annual operating costs from the NHS. We conducted a secondary analysis of aggregated national level data on the level of SPC service provision that had been collected in a national service mapping study as part of the Scottish Atlas of Palliative Care.2 The methods used to map SPC service provision in the Scottish Atlas have been described previously.2 In brief, the chief executive officers of independent hospices, NHS Executive Leads for Palliative and End of Life Care and, where in place, the chairperson of the Managed Clinical Network for Palliative and End of Life Care in each territorial health board were contacted between August and September 2015, informed of the nature and purpose of the service mapping study and invited to contribute information about the services they provided. Consent was implied by the willingness of service providers to nominate a key informant from their organisation to contribute data to the study. Data were collected in a telephone survey conducted by one of three trained investigators (HI, MG and AJW). Prior to the survey, respondents were sent a structured questionnaire to enable them to gather the required information in advance. All survey respondents were either directly involved in delivering or responsible for managing the SPC services on which they provided information; in some cases, it was necessary to speak with several informants from a single territorial health board to obtain a complete picture of service provision. Where it was not possible to conduct a telephone survey, despite repeated attempts, respondents were given an option to complete the structured questionnaire online (n=3). The structured questionnaire and definitions therein were derived from the European Atlas of Palliative Care 7 and adapted for local use (see appendix 1 of the Atlas).

ssible to conduct a telephone survey, despite repeated attempts, respondents were given an option to complete the structured questionnaire online (n=3). The structured questionnaire and definitions therein were derived from the European Atlas of Palliative Care 7 and adapted for local use (see appendix 1 of the Atlas). In total, 36 informants provided data from all 14 territorial health boards and all 15 fully operational independent hospices. The SPC service mapping study did not involve patients or service users. It is anomalous that information of this nature is not routinely available to inform the design and delivery of SPC services to meet the needs of local populations. The service mapping study was therefore considered as service evaluation and, as such, research ethics approval for the study was not sought. We followed the method of measuring palliative care ‘resources’ developed in previous studies.3 This involved establishing the numbers of three types of SPC service provision, inpatient units, hospital support teams and home care teams, then calculating the ratio of services per million inhabitants by service type and total number of all services. This ratio was used to compare Scotland with the top ranking countries in the EU.3 Population data were obtained from National Records Scotland.8

inpatient units, hospital support teams and home care teams, then calculating the ratio of services per million inhabitants by service type and total number of all services. This ratio was used to compare Scotland with the top ranking countries in the EU.3 Population data were obtained from National Records Scotland.8 Following an established method, we also calculated coverage of SPC provision in Scotland for each service type.4 Coverage refers to the relationship between the number of available services and the estimated number of services that would be required, based on expert opinion, to meet the palliative care needs of a given population. The EAPC White Paper identifies ‘norms’ for the structural provision of SPC, which represent a consensus statement of the levels that, if achieved, would be expected to translate into high-quality care. Recommendations based on these ‘norms’ for the number of services required per population are as follows: one home care team per 100 000 inhabitants; one inpatient unit and one hospital support team per 200 000 inhabitants.5 6 Based on these figures, the required number of the three types of services was calculated for the population of Scotland and a measure of ‘coverage’, expressed as a percentage, was established. EAPC recommendation concerning the desirable upper and lower numbers of SPC beds per million population, producing a calculation based on the number of SPC inpatient beds in Scotland, was also calculated.5

calculated for the population of Scotland and a measure of ‘coverage’, expressed as a percentage, was established. EAPC recommendation concerning the desirable upper and lower numbers of SPC beds per million population, producing a calculation based on the number of SPC inpatient beds in Scotland, was also calculated.5 Results Level of SPC service provision Scotland had a total of 23 SPC inpatient units containing 349 beds, 27 hospital support teams and 38 home care teams, which equates to 4.3 SPC inpatient units per million people, 5.0 hospital support teams per million people and 7.1 home care teams per million people. Scotland’s position for each type of service in relation to the 10 top EU countries is shown in table 1. Table 1 Comparison of types of specialist palliative care in Scotland and other European countries Inpatient units* Hospital support teams† Home care teams‡ Country Ratio/million Country Ratio/million Country Ratio/million 1 The Netherlands 12.7 Belgium 10.4 Estonia 11.3 2 Luxembourg 9.4 Ireland 8.5 Sweden 11.2 3 Germany 5.2 Slovenia 8.3 Poland 8.4 4 Denmark 5.0 UK 5.7 Ireland 7.6 5 Belgium 4.6 Luxembourg 5.7 Scotland 7.1 6 Austria 4.4 France 5.3 Hungary 7.0 7 Scotland 4.3 Scotland 5.0 UK 6.1 8 Sweden 4.0 Malta 3.6 Austria 5.8 9 Poland 3.8 Latvia 3.4 Luxembourg 5.7 10 UK 3.5 Austria 3.4 Italy 5.2 *Scotland has 23 palliative care inpatient units containing 349 beds. †Scotland has 27 hospital support teams. ‡Scotland has 38 home care teams.

Inpatient units* Hospital support teams† Home care teams‡ Country Ratio/million Country Ratio/million Country Ratio/million 1 The Netherlands 12.7 Belgium 10.4 Estonia 11.3 2 Luxembourg 9.4 Ireland 8.5 Sweden 11.2 3 Germany 5.2 Slovenia 8.3 Poland 8.4 4 Denmark 5.0 UK 5.7 Ireland 7.6 5 Belgium 4.6 Luxembourg 5.7 Scotland 7.1 6 Austria 4.4 France 5.3 Hungary 7.0 7 Scotland 4.3 Scotland 5.0 UK 6.1 8 Sweden 4.0 Malta 3.6 Austria 5.8 9 Poland 3.8 Latvia 3.4 Luxembourg 5.7 10 UK 3.5 Austria 3.4 Italy 5.2 *Scotland has 23 palliative care inpatient units containing 349 beds. †Scotland has 27 hospital support teams. ‡Scotland has 38 home care teams. Coverage Applying the EAPC norms to Scotland’s population of 5 347 600,8 the optimal level of provision for each SPC service is 27 inpatient units, 27 hospital support teams and 53 home care teams. Based on the number of services available in Scotland, we estimate coverage of SPC inpatient units, hospital support teams and home care teams to be 85%, 100% and 72%, respectively. This positions Scotland among the top 10 EU countries for coverage in all three types of service provision (table 2). Table 2 Coverage of specialist palliative care in Scotland based on EAPC White Paper recommendations (top 10 EU countries)

Coverage Applying the EAPC norms to Scotland’s population of 5 347 600,8 the optimal level of provision for each SPC service is 27 inpatient units, 27 hospital support teams and 53 home care teams. Based on the number of services available in Scotland, we estimate coverage of SPC inpatient units, hospital support teams and home care teams to be 85%, 100% and 72%, respectively. This positions Scotland among the top 10 EU countries for coverage in all three types of service provision (table 2). Table 2 Coverage of specialist palliative care in Scotland based on EAPC White Paper recommendations (top 10 EU countries) Inpatient units Hospital support teams Home care teams Country Coverage (%) Country Coverage (%) Country Coverage (%) 1 The Netherlands 254 Belgium 215 Sweden 113 2 Luxembourg 153 Ireland 170 Estonia 112 3 Germany 102 Slovenia 167 Poland 84 4 Denmark 100 UK 115 Ireland 76 5 Belgium 95 Luxembourg 100 Scotland 72 6 Austria 88 Scotland 100 Hungary 69 7 Scotland 85 France 82 UK 62 8 Sweden 80 Malta 72 Austria 58 9 Poland 76 Austria 69 Luxembourg 57 10 United Kingdom 70 Latvia 63 Italy 51 EAPC, European Association for Palliative Care.' at the bottom of Tables 2 and 3. The EAPC norms recommend a range of between 80 and 100 SPC inpatient beds per million people. Applied to Scotland’s population, this would equate to a recommended number of inpatient beds ranging from 427 to 535. Although relative to the EAPC norms, the coverage of SPC inpatient units is 85% in Scotland, the coverage of SPC inpatient beds is 65% (at 100 per million) and 82% (at 80 per million) (table 3).

eople. Applied to Scotland’s population, this would equate to a recommended number of inpatient beds ranging from 427 to 535. Although relative to the EAPC norms, the coverage of SPC inpatient units is 85% in Scotland, the coverage of SPC inpatient beds is 65% (at 100 per million) and 82% (at 80 per million) (table 3). Table 3 Comparison of number of SPC inpatient units and inpatient beds in Scotland against EAPC norms Recommended Available Coverage (%) Inpatient units* 27 23 85 At 80 beds per million† 428 349 82 At 100 beds per million† 535 349 65 *4.3 units per million inhabitants is based on 5 PC units (50 beds) per million †EAPC White Paper upgraded recommendation

Table 3 Comparison of number of SPC inpatient units and inpatient beds in Scotland against EAPC norms Recommended Available Coverage (%) Inpatient units* 27 23 85 At 80 beds per million† 428 349 82 At 100 beds per million† 535 349 65 *4.3 units per million inhabitants is based on 5 PC units (50 beds) per million †EAPC White Paper upgraded recommendation Discussion To date, published studies examining the provision of SPC services at a regional, national or international level have largely relied on information provided by one or more national experts.3 9 10 This is problematic because national experts may lack the requisite knowledge to accurately report in detail service provision at the national or subnational level and may be biased in their reporting.11 Our study analysed data gathered from 36 informants directly delivering, or responsible for managing, local SPC services in each of Scotland’s 14 territorial health boards areas, increasing the validity of our findings. Due to their local knowledge and the nature of the telephone survey, most respondents volunteered a narrative around the quantitative data on the level of service provision they provided, which was invaluable in contextualising the study findings. To our knowledge, this is the first study of its kind to use such a robust data collection strategy.

and the nature of the telephone survey, most respondents volunteered a narrative around the quantitative data on the level of service provision they provided, which was invaluable in contextualising the study findings. To our knowledge, this is the first study of its kind to use such a robust data collection strategy. The UK is generally considered to be the reference country against which the development of SPC in other countries is benchmarked.1 Until now, it has not been possible to disaggregate Scottish data from that of the rest of the UK.3 9 10 While we might expect the provision of SPC services in Scotland to be similar to the rest of the UK, this is the first study to provide confirmation. The provision of SPC services in Scotland was found to be well balanced and broadly reflective of that in the rest of the UK, with Scotland’s position being higher than the rest of the UK for inpatient and home care teams but lower for hospital support team. Analysis of coverage against EAPC norms suggests that Scotland has optimal coverage for hospital support teams. More broadly, Scotland was positioned among the top 10 EU countries in hospital and home care resources for SPC. Although the ratio of home care teams per million people (7.1) is better than the ratios for hospital support teams (5.0) and inpatient units (4.3), Scotland shows an optimal coverage of hospital support teams and a lower than recommended coverage for inpatients units (85%) and home care teams (72%). Calculated at both the upper and lower ranges, Scotland scores lower on SPC inpatient bed coverage (65%–82%) than it does on the calculation based on the number of units.

tland shows an optimal coverage of hospital support teams and a lower than recommended coverage for inpatients units (85%) and home care teams (72%). Calculated at both the upper and lower ranges, Scotland scores lower on SPC inpatient bed coverage (65%–82%) than it does on the calculation based on the number of units. For almost two decades, the strategic narrative for health and social care in Scotland has focused on the delivery of safe, effective, high-quality, patient-centred care delivered at home or in a homely setting.12 13 As a result, over the last decade, the total number of available staffed hospital beds in Scotland has been reduced by almost a fifth.14 The lower than anticipated level of inpatient SPC beds may therefore reflect successful policy implementation rather than deficiencies in the planning and provision of SPC services. Many service providers in our study, especially in remote and rural areas, identified that while they did not have dedicated SPC inpatient beds, they were providing integrated SPC to inpatients that remained under the care of their admitting specialty, through SPC hospital support teams. This flexible model of care ensures effective use of acute hospital beds and continuity of care and enables collaborative working between generalists and specialists that in turn may facilitate the ‘up-skilling’ of healthcare providers from other specialties in delivering ‘generalist’ palliative care. Indeed, in recent years, the role of SPC providers in providing education and support to generalists has been championed.15 16 Increasingly in Scotland SPC clinicians and nurse specialists are jointly appointed to work between the independent hospices, the acute hospital sector and community services to provide integrated care and facilitate early hospital discharge. This change in how and where SPC teams are based is reflected in the optimal coverage of hospital support teams.

ians and nurse specialists are jointly appointed to work between the independent hospices, the acute hospital sector and community services to provide integrated care and facilitate early hospital discharge. This change in how and where SPC teams are based is reflected in the optimal coverage of hospital support teams. The coverage of home care teams in Scotland was less than optimal. This is an important finding. If the policy of shifting the balance of care from hospitals to community settings is to be translated into practice, community services must be adequately resourced.12 There has been a growing policy recognition in Scotland of the importance of achieving preferred place of care and preferred place of death.14 Most people would prefer to die at home or in a community setting.17 It is not possible to measure how well preferences are being met in Scotland because neither preferred place of care and preferred place of death are systematically recorded. In England, an indicator for death in usual place of residence has been developed and is in use. In Scotland, such a measure is not available. A surrogate measure, the percentage of the last 6 months of life spent at home or in a community setting, is in use; in 2014/2015 among decedents in Scotland, 86% of the last 6 months of life were spent at home or in a community setting and 14% in an acute hospital.18 Moreover, people are more likely to die in hospital in Scotland (53%) than they are in England (47%).19 Adequately resourcing community-based services, across health and social care, will be important if Scotland is to realise its policy ambitions; our study provides data to advocate for further developing community-based services. Many other health and social care professionals contribute to the interdisciplinary teams that support people to live and die well in community settings, not least community district nurses, disease-specific specialist nurses and primary care physicians. In Scotland, these key individuals provide generalist palliative care and often act as gatekeepers to SPC services. The availability of high-quality generalist palliative care is likely to directly impact on the demand for and subsequently delivery of SPC services in community settings. At the same time, measuring the availability and delivery of generalist palliative care raised additional challenges.

s gatekeepers to SPC services. The availability of high-quality generalist palliative care is likely to directly impact on the demand for and subsequently delivery of SPC services in community settings. At the same time, measuring the availability and delivery of generalist palliative care raised additional challenges. It is important to note that when collecting the primary data for the Atlas, despite clear definitions, there was variation in how service providers conceptualised their services; for example, one territorial health board reported having just one home care team while another, with the same level of staffing across its home care service, reported having multiple home care teams stating that each team covered distinct geographical areas within the board. Our data on home care teams should therefore be interpreted with caution. This does raise questions about what ‘counts’ as a service in any given setting and may have led to an underestimation of the number of home care teams operating in Scotland. It also highlights the importance of common definitions and the challenges in operationalising these, both for service evaluation and for research.

is does raise questions about what ‘counts’ as a service in any given setting and may have led to an underestimation of the number of home care teams operating in Scotland. It also highlights the importance of common definitions and the challenges in operationalising these, both for service evaluation and for research. The EAPC White Paper provides norms for the number of services and resources per million people, allowing for the diversity of concepts of SPC and varying cultures across Europe. These norms are based on expert opinion and estimates of demand for services that EAPC considers, if achieved, will translate into the provision of high-quality care. This approach to establishing a range of values for benchmarking is not without limitations. It is common practice where the evidence base is lacking to rely on expert opinion. Underpinning recommendations for service provision using estimated demand for services, rather than need, is problematic; however, estimating palliative care needs is particularly challenging.20 The appropriateness of these measures is likely to change as services evolve and the evidence base grows. The measures may also have differing significance according to the context of the healthcare system within a country. For example, the coverage of hospital support teams in Belgium was over 100%; however, according to Belgian law, every hospital is obliged to have a SPC team. The extent to which ‘generalist’ palliative care principles are effectively embedded within a healthcare system may also influence the required level of SPC services and in part explain an apparent underprovision or overprovision of certain services. It is important to consider that these metrics are not sensitive enough to capture such nuance. The balance between services could also be a result of the distribution of population across different areas of the country. With the exception of the central belt of Scotland, where the population is concentrated into large conurbations, there are many dispersed and rural communities of small population. As in many other countries, the distribution and volume of SPC services in Scotland is the result of historical circumstances, local priorities and enthusiasms, as well as the ebb and flow of government interest and funding.

ed into large conurbations, there are many dispersed and rural communities of small population. As in many other countries, the distribution and volume of SPC services in Scotland is the result of historical circumstances, local priorities and enthusiasms, as well as the ebb and flow of government interest and funding. The EAPC norms provide a valuable tool for benchmarking the provision of SPC services between countries or within a country over time, which can be a valuable tool for advocacy and also, over time, for more rational planning. However, these measure require careful interpretation considering the wider sociodemographic, political and economic context within which services are being provided. Importantly, these metrics do not provide any objective measure of the quality of SPC. The difference in timing of data gathered for Scotland alone and that gathered for the 10 EU countries used as comparators is acknowledged. Data for the EU ranking and coverage studies were collected in 2013, compared with 2015 for the present study. This small time gap is not considered to be significant. Indeed, the proximity of Scottish data to that for the UK as a whole, given the common evolution of SPC provision across the UK, is reassuring.

wledged. Data for the EU ranking and coverage studies were collected in 2013, compared with 2015 for the present study. This small time gap is not considered to be significant. Indeed, the proximity of Scottish data to that for the UK as a whole, given the common evolution of SPC provision across the UK, is reassuring. The primary data used in this study were derived from a novel SPC service mapping study. Prior to this, it was not possible to describe the level of SPC provision at local or national level in Scotland; it is still not possible to measure SPC activity, outcomes or quality of care in Scotland. It is anomalous that data of this kind are not available to inform the design, strategic commissioning, delivery and continuous quality improvement of SPC services. These significant ‘information deficits’ were recognised in a 2015 Scottish Parliamentary Inquiry into Palliative Care,1 and a policy commitment has been made to address them in Scotland’s Strategic Framework for Action on Palliative and End of Life Care later that year by the Scottish Government.15 The national mapping survey was a first step towards this. The study was both labour and time intensive; however, it was necessary to provide reliable information to quantify and characterise SPC provision at local and national level in Scotland. In the study, several informants acknowledged that their service(s) were being redesigned to better meet the needs, preferences and priorities for care in their locality, as informed by local health needs assessments. This highlights the challenge of maintaining accurate and current data on SPC service provision and organisation at a national level. A more sustainable approach to collecting and collating these data is required building on models of SPC directories used elsewhere and making use of multiple providers of data.

ssments. This highlights the challenge of maintaining accurate and current data on SPC service provision and organisation at a national level. A more sustainable approach to collecting and collating these data is required building on models of SPC directories used elsewhere and making use of multiple providers of data. Finally, we acknowledge that this is an analysis of indicators for SPC. It does not address the levels and coverage of resources that may be available for the delivery of ‘generalist’ or ‘primary’ palliative care. Such a mapping and assessment is also worth considering but was beyond the remit of the present study. Conclusion This study has addressed part of the ‘information deficit’ previously identified. It provides a comparative analysis of resources and coverage of SPC in Scotland, as compared with other countries of the EU. In all the items compared - inpatient units, hospital care support teams and home care teams — Scotland is positioned in the top 10 EU countries and has a consistent level of coverage. Moreover, this is at a high level (72%–100%) in relation to the expert recommendations of the EAPC. The study provides a baseline from which to monitor the development of SPC in Scotland. The Scottish government has published a Strategic Framework for Action for Palliative and End of Life Care covering the period 2016–2021.15 The present study and the accompanying Atlas constitute an important baseline, against which some of the goals of the Framework can be assessed.

the development of SPC in Scotland. The Scottish government has published a Strategic Framework for Action for Palliative and End of Life Care covering the period 2016–2021.15 The present study and the accompanying Atlas constitute an important baseline, against which some of the goals of the Framework can be assessed. The authors would like to thank Amanda Jane Ward (AJW) and Eduardo Garralda for assistance and support. Contributors: HI, JMC and DC participated in the conception and design of the study. HI and MG participated in the data acquisition. HI and JMC participated in data analysis, interpretation and drafting of the article. JMC, MG and DC provided critical revisions and additions of important intellectual content. All authors approved the final version of the article. Funding: This work was supported by a Wellcome Trust Investigator Award (grant number 103319/Z/13/Z) and a grant from the Scottish Government. Competing interests: None declared. Provenance and peer review: Not commissioned; externally peer reviewed.

Introduction The UK’s National End of Life Care Programme aimed to ensure individuals have more choice about where they die.1 The hierarchy of preferences most commonly reported rank home death first, hospice second and hospital the least favoured.2 While achieving home death is considered a key performance indicator and a proxy for quality end-of-life care, death in hospital is generally perceived as overly aggressive and suboptimal, along with emergency presentations, hospital admissions, chemotherapy treatments and intensive care input in the last weeks/days of life.3 4 Studies examining such issues report that patients with haematological malignancies (leukaemias, lymphomas and myeloma) are more likely to receive aggressive end-of-life care than those with other cancers.5–8 This includes an increased propensity for hospital death, a situation identified across countries, regardless of healthcare systems, disease subtypes and patient characteristics,9 10 reduced chance of palliative care and hospice referral or referral very close to death11–14 and (in some countries) particularly traumatic end-of-life experiences.15

ropensity for hospital death, a situation identified across countries, regardless of healthcare systems, disease subtypes and patient characteristics,9 10 reduced chance of palliative care and hospice referral or referral very close to death11–14 and (in some countries) particularly traumatic end-of-life experiences.15 Haematological malignancies are the fifth most common cancer in the UK.16 These are complex diseases, and although they have some similarities with other cancers, they often require different management strategies. Surgical resection is not an option, for example, and while some subtypes are curable with intensive, toxic chemotherapy and long periods of hospitalisation, others are incurable from diagnosis but managed with intermittent or continuous chemotherapy.17 Taking such factors into account, we explored the experiences of clinical staff and relatives of patients who had died from these cancers, to determine reasons why hospital deaths predominate. Methods The study is set within the UK’s Haematological Malignancy Research Network (www.hmrn.org),16 an ongoing programme of work providing infrastructure for evidence-based research and aiming to improve the experiences of patients and their families. We conducted a qualitative study to elicit participants’ views, as this approach is well suited to exploring poorly understood phenomena.18

ch Network (www.hmrn.org),16 an ongoing programme of work providing infrastructure for evidence-based research and aiming to improve the experiences of patients and their families. We conducted a qualitative study to elicit participants’ views, as this approach is well suited to exploring poorly understood phenomena.18 In-depth, semistructured interviews were carried out between 2012 and 2014 with 45 clinicians involved in the delivery of end-of-life care to patients with haematological malignancies, including nine haematologists, eight haematology nurses, six palliative care/hospice doctors, seven community-based and seven hospital-based palliative care nurses and eight general practitioners (GPs). Ten relatives of deceased patients were also interviewed; four of these patients had leukaemia, four had lymphoma and two had myeloma. They were all aged between 66 and 84 years at death and four were female. Recruitment involved purposive sampling,18 which ensured inclusion of clinicians from primary and secondary care, as well as relatives of people with a range of diseases. Potential interviewees were approached via email or by post.

o had myeloma. They were all aged between 66 and 84 years at death and four were female. Recruitment involved purposive sampling,18 which ensured inclusion of clinicians from primary and secondary care, as well as relatives of people with a range of diseases. Potential interviewees were approached via email or by post. Interviews lasted 30–90 min, were conducted privately in a hospital or university setting, or in relatives’ homes, and were audiotaped and transcribed. A topic guide (see online supplementary appendix 1) was developed from existing literature and experiences of the study team, but was used flexibly to allow for unanticipated responses. Data collection continued until no new information was forthcoming.19 Transcripts were analysed for thematic content. Coding and classification of data was inductive, following the sequential steps of data familiarisation by reading/re-reading transcripts, development of a coding frame to apply to the whole data set, attribution of data to individual codes, collating codes into themes and interpretation through seeking meaning, salience and connections.20 To promote transparency, rigour and trustworthiness,20 an independent qualitative researcher checked and corroborated the coding of five interviews (around 10%). 10.1136/bmjspcare-2016-001289.supp1Supplementary Appendix 1

Interviews lasted 30–90 min, were conducted privately in a hospital or university setting, or in relatives’ homes, and were audiotaped and transcribed. A topic guide (see online supplementary appendix 1) was developed from existing literature and experiences of the study team, but was used flexibly to allow for unanticipated responses. Data collection continued until no new information was forthcoming.19 Transcripts were analysed for thematic content. Coding and classification of data was inductive, following the sequential steps of data familiarisation by reading/re-reading transcripts, development of a coding frame to apply to the whole data set, attribution of data to individual codes, collating codes into themes and interpretation through seeking meaning, salience and connections.20 To promote transparency, rigour and trustworthiness,20 an independent qualitative researcher checked and corroborated the coding of five interviews (around 10%). 10.1136/bmjspcare-2016-001289.supp1Supplementary Appendix 1 Results Based on their experiences, none of the clinical participants were surprised that patients with haematological malignancies were more likely to die in hospitals than people with other cancers. Five themes were identified: four relating to reasons for hospital deaths and one describing suggested changes to practice to facilitate non-hospital death when preferred. These themes are illustrated below using verbatim quotes.

ological malignancies were more likely to die in hospitals than people with other cancers. Five themes were identified: four relating to reasons for hospital deaths and one describing suggested changes to practice to facilitate non-hospital death when preferred. These themes are illustrated below using verbatim quotes. Theme 1: characteristics and trajectory of haematological malignancies It was generally agreed that hospital deaths were principally determined by the characteristics and trajectories of haematological malignancies. These were considered complex and uncertain, with indistinct transitions between curative/life-prolonging and palliative approaches to care, leading to difficulties determining the appropriate time to initiate discussions about end-of-life care. This uncertainty was captured in descriptions of trajectories, which were said to range from gradual decline (often punctuated by episodes of acute deterioration) to sudden and unexpected death. The propensity for the latter was recognised as greater in patients with diseases requiring urgent, intensive treatment (eg, acute leukaemia or aggressive lymphoma) but was also said to occur on a background of gradual decline, typically in indolent diseases following remitting/relapsing pathways (eg, myeloma, chronic lymphocytic leukaemia and follicular lymphoma). Even within these categorisations, however, variations were found. Transitions were considered less dichotomous than in other cancers, where patients were described as being more likely to be discharged from individual specialisms at specific pathway points, typically after surgery or failed chemotherapy.

ar lymphoma). Even within these categorisations, however, variations were found. Transitions were considered less dichotomous than in other cancers, where patients were described as being more likely to be discharged from individual specialisms at specific pathway points, typically after surgery or failed chemotherapy. ‘it could be severe neutropenic sepsis, rapid deterioration in a situation where you are trying to treat (the cancer), where somebody goes from well, to critically unwell, to dead, and you can’t plan for that, they suffer some sort of complication related to chemotherapy, so bleeding, internal haemorrhage, because they are profoundly thrombocytopenic…you can’t plan ahead for those (events)….’ (Haematologist 6) ‘some patients will deteriorate over days, some will last several months, even with acute leukaemia…’ (Haematologist 4) ‘I do think their disease trajectory makes it more difficult to, eh, actually predict when they’re going to sort of start deteriorating, compared to other cancers’ (GP 6)

‘it could be severe neutropenic sepsis, rapid deterioration in a situation where you are trying to treat (the cancer), where somebody goes from well, to critically unwell, to dead, and you can’t plan for that, they suffer some sort of complication related to chemotherapy, so bleeding, internal haemorrhage, because they are profoundly thrombocytopenic…you can’t plan ahead for those (events)….’ (Haematologist 6) ‘some patients will deteriorate over days, some will last several months, even with acute leukaemia…’ (Haematologist 4) ‘I do think their disease trajectory makes it more difficult to, eh, actually predict when they’re going to sort of start deteriorating, compared to other cancers’ (GP 6) Difficulties determining prognosis were discussed by many interviewees. This was complicated by pathways that might include sudden deaths or ‘dip up and down’ as multiple relapses and recoveries occurred. While the need for realistic, frank and early discussions of prognosis and the potential for sudden changes were considered crucial to advance planning, haematology staff described how these could sometimes only be initiated when patients had accepted that cure may not be possible. A further difficulty was identifying the point at which response to treatment (intensive/non-intensive chemotherapy or supportive care) was no longer likely to be achieved, particularly when patients may have recovered previously, even from seemingly dire situations. ‘the trouble is not knowing in advance what might be achieved and what might not…that’s quite tricky’ (Haematologist 9)

Difficulties determining prognosis were discussed by many interviewees. This was complicated by pathways that might include sudden deaths or ‘dip up and down’ as multiple relapses and recoveries occurred. While the need for realistic, frank and early discussions of prognosis and the potential for sudden changes were considered crucial to advance planning, haematology staff described how these could sometimes only be initiated when patients had accepted that cure may not be possible. A further difficulty was identifying the point at which response to treatment (intensive/non-intensive chemotherapy or supportive care) was no longer likely to be achieved, particularly when patients may have recovered previously, even from seemingly dire situations. ‘the trouble is not knowing in advance what might be achieved and what might not…that’s quite tricky’ (Haematologist 9) The frequent introduction of novel therapies for some diseases further complicated the issue of determining when to stop treatment. Treating patients aggressively in the last month and days of life was considered unavoidable, however, when this was being delivered with curative intent and sudden death occurred. In such situations, the transition to end-of-life care was described as absent, particularly among younger people, as emphasis was on saving life, often in intensive care settings. Treatment was also routinely given in the terminal phase as an intentional strategy for disease and symptom control, and to maintain quality of life for as long as possible. Non-haematology specialists considered this approach overly aggressive, however, believing that it occurred due to reluctance on the part of (some) haematologists to discuss withdrawing treatment and possible death, factors cited as impeding decision-making and advanced planning.

uality of life for as long as possible. Non-haematology specialists considered this approach overly aggressive, however, believing that it occurred due to reluctance on the part of (some) haematologists to discuss withdrawing treatment and possible death, factors cited as impeding decision-making and advanced planning. ‘some consultants treat, treat, treat’ (Palliative nurse 5) ‘I think haematology want to keep patients going as long as possible and they don’t want to break that bad news’ (Palliative nurse 1) Overtreating of certain groups of patients, such as those with acute myeloid leukaemia (AML), was acknowledged by some haematologists. ‘I personally think we treat far too many people far too aggressively, that the thrust of AML is completely wrong, if the patient is over 60 or 70 you are never going to cure them. Taking a more palliative, quality of life approach would probably be better’ (Haematologist 6) Some respondents alluded to ‘co-dependency’ between haematologists and their patients, describing haematologists’ reluctance to ‘give up’ exploring treatment options. In this context, patients contributed to the impetus for ongoing treatment and hospitalisation, being desperate to ‘try anything’ in the hope of achieving cure or postponing death. Patients were said to associate this with (possibly false) hope for continuing life and discharge from haematology to home or hospice with imminent death.

, patients contributed to the impetus for ongoing treatment and hospitalisation, being desperate to ‘try anything’ in the hope of achieving cure or postponing death. Patients were said to associate this with (possibly false) hope for continuing life and discharge from haematology to home or hospice with imminent death. ‘it might be quite hard to… be impassive… you [haematologist] think, well, we’ve pulled them back from a similar situation before, maybe we can do it again…and that has some impact on your ability to let go…in the context of withdrawing treatment…up to that point, you might have always been the guy who had the next treatment ready, the next idea, if this doesn’t work, we’ll do this…and that has on occasion worked and helped improve quality of life or prognosis’ (Haematologist 4) ‘he was a fighting spirit…he wanted to go on being treated…he wanted treatment to the end’ (Haematology nurse 7) ‘she wanted anything, if it could prolong things, if it could give her more time, then she wanted to have it’ (Relative 9, concerning deceased wife) ‘we really wanted him to go to the hospice as he was dying, fading away and he would not have it…he chose to stay on the ward…he wanted to cling on to the blood tests…the reassurance he was still being watched’ (Haematologist 1)

‘she wanted anything, if it could prolong things, if it could give her more time, then she wanted to have it’ (Relative 9, concerning deceased wife) ‘we really wanted him to go to the hospice as he was dying, fading away and he would not have it…he chose to stay on the ward…he wanted to cling on to the blood tests…the reassurance he was still being watched’ (Haematologist 1) Theme 2: mismatch between the expectations and reality of home death Patients’ and relatives’ limited knowledge and sometimes unrealistic expectations about end-of-life processes, and the sudden onset of distressing symptoms, such as bleeding and sepsis, were said to potentially precipitate a shift in preference away from home towards hospital death. ‘the reality of somebody dying at home can be very, very difficult and very different to the ‘home sweet home’ image you might have’ (Haematology nurse 4) ‘he would have continued at home if he hadn’t started coughing up blood and couldn’t get his breath’ (Relative 5—spouse died in the hospital) As a consequence, another reported cause of hospital death was the frequency with which patients ‘bounce back’ into hospital due to sudden deterioration and distressing symptoms. Emergency calls from relatives, GPs or out-of-hours doctors were cited as major reasons for readmissions, as ambulance staff (often lacking information about preferences) were said to routinely transport patients to Accident and Emergency (A&E), resulting in hospital admission.

n deterioration and distressing symptoms. Emergency calls from relatives, GPs or out-of-hours doctors were cited as major reasons for readmissions, as ambulance staff (often lacking information about preferences) were said to routinely transport patients to Accident and Emergency (A&E), resulting in hospital admission. ‘the problem is, if they’re bleeding and the relatives panicking, they call 999, obviously they end up going to A&E…then they end up with the problem of inappropriate treatments… they’re [emergency services] not aware of the fact that somebody’s got leukaemia in the terminal phase…we [primary care clinicians] need to make sure that we communicate well with A&E and the ambulance service, the paramedics, so they know exactly what they’re dealing with’ (GP 8) Theme 3: preference for hospital death The close relationship between haematology staff, patients and their relatives was often discussed and was perceived to lead to preferences to remain in hospital at the time of death. This bond was considered a positive consequence of ongoing contact between the individuals involved over prolonged time periods. It was described as arising from experiences of ‘exemplary care’, in a setting often regarded as ‘a second home’, and associated with feelings of safety and security. Consequently, situations were described where remaining in the haematology ward, in a familiar environment with well-known staff, was said to be preferable to moving to a palliative care or hospice setting where new relationships would need to be built, often over short periods of time.

feelings of safety and security. Consequently, situations were described where remaining in the haematology ward, in a familiar environment with well-known staff, was said to be preferable to moving to a palliative care or hospice setting where new relationships would need to be built, often over short periods of time. Some haematology staff shared similar sentiments, describing how transferring their patients to community staff they did not know for end-of-life care could feel like ‘abandoning’ them. These interviewees also described a certain satisfaction in being able to provide terminal care for patients they had known a long time, although this was sometimes countered by concerns about capacity (eg, staff time and availability of a private room). ‘you’ve been there for the relatives, for the carers and… you want to see all that through… the difficulty sometimes… is that by transferring somebody purely to palliative care, you almost feel as if you’re washing your hands of them’ (Haematology nurse 4) ‘I think that a lot of it… is building relationships again, and the way we work is that we, we see them at diagnosis and we do…all the chemotherapy treatments, so we get a really close bond and then they, they’re always coming back to clinic and we’re seeing them so if they relapsed, it would be us that they would see again ….so really in the end we’ve got kind of years of quite a strong bond with our patients and I think that’s the hard thing then is that when they do come towards the end I think they feel that they want to stay with us’ (Haematology nurse 7)

we’re seeing them so if they relapsed, it would be us that they would see again ….so really in the end we’ve got kind of years of quite a strong bond with our patients and I think that’s the hard thing then is that when they do come towards the end I think they feel that they want to stay with us’ (Haematology nurse 7) ‘there’s only two people on nights for 20 patients, and if you’ve got someone who’s dying… it’s awful. During the day isn’t so bad… you have five nurses on, in the morning, and you go to two at night. So you know, it’s a big difference.’ (Haematology nurse 1) Haematology doctors and nurses often questioned whether a good death could only occur at home, proposing instead that patients’ and their family’s needs for physical and emotional support could be effectively met through hospital haematology services, with input from palliative medicine as required, and that this may be entirely appropriate.

2011 and 2016. HES data on hospital admissions in the last year of life in 10 CCGs areas. Secondary data analysis and 91 surveys DiUPR and cost Qualitative Healthcare professionals, patients, carers, system leaders. 12 in-depth interviews Implementation *Data on place of death were only available for four EPaCCS sites. ACP, advance care planning; CCG, clinical commissioning group; DiUPR, death in usual place of residence;EPaCCS, Electronic Palliative Care Coordination Systems; HES, Hospital Episodes Statistics;NHS, National Health Service; ONS, Office of National Statistics; PPD, preferred place of death. Figure 1 PRISMA flow diagram if papers reporting numbers of included and excluded studies. EPaCCS, electronic palliative care coordination systems; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

nurses often questioned whether a good death could only occur at home, proposing instead that patients’ and their family’s needs for physical and emotional support could be effectively met through hospital haematology services, with input from palliative medicine as required, and that this may be entirely appropriate. ‘the current rhetoric that the only good death is a home death is absolute nonsense… it’s wherever the patient’s needs are best met… some patients get to know the nursing staff pretty well, and for them, it’s much nicer for them to go back to a ward where they know all the staff, than it is to go somewhere else or even to struggle in their home…so it may be their choice that they’re coming in to die, and I don’t think we should deny them that just because the Government has a bee in its bonnet about ‘people must die at home’… and I can see in haematology patients who have been coming in and out of a ward for years… a hospital death may be an appropriate death…and it may be done very well, actually’ (Haematologist 1) Some differences were identified between haematology practitioner perceptions and those of other clinicians and some relatives, however, with the latter being most likely to cite home as the expected end-of-life preference.

‘the current rhetoric that the only good death is a home death is absolute nonsense… it’s wherever the patient’s needs are best met… some patients get to know the nursing staff pretty well, and for them, it’s much nicer for them to go back to a ward where they know all the staff, than it is to go somewhere else or even to struggle in their home…so it may be their choice that they’re coming in to die, and I don’t think we should deny them that just because the Government has a bee in its bonnet about ‘people must die at home’… and I can see in haematology patients who have been coming in and out of a ward for years… a hospital death may be an appropriate death…and it may be done very well, actually’ (Haematologist 1) Some differences were identified between haematology practitioner perceptions and those of other clinicians and some relatives, however, with the latter being most likely to cite home as the expected end-of-life preference. Theme 4: barriers and facilitators to home or hospice death A number of practical barriers to non-hospital death that were specific to patients with haematological malignancies were identified. Some generalists expressed unease about managing haematology patients at home, either because they thought they themselves lacked knowledge about what were perceived as complex diseases, or because they considered that distressing symptoms, such as bleeding, were a barrier to death at home—a view shared by some haematology staff. Others, however, reported the relative infrequency with which such symptoms occurred, and did not consider the end-of-life care needs of haematology patients to differ much from that of people with other conditions.

ing symptoms, such as bleeding, were a barrier to death at home—a view shared by some haematology staff. Others, however, reported the relative infrequency with which such symptoms occurred, and did not consider the end-of-life care needs of haematology patients to differ much from that of people with other conditions. ‘bleeding would be the big one for patients with HM…you would have to have robust family and carers to cope with that’ (Palliative doctor 5) ‘the last patient that I tried to manage at home, she had a big, uh, GI bleed at the end of the life, so she came in, because of that, cause her husband could not cope… with that’ (Haematology nurse 5) ‘we’ve got dedicated district nurses, that, I would have every confidence, they could look after that…haematological malignancy may, may be more dramatic, but you can get just as ill with other malignancies…pain, bleeding, breathlessness, agitation, confusion, all those things you get with any cancer…and you get with the haematological malignancy patients. Interviewer: So it’s not actually that different from looking after patients with other cancers. Um, no, not in terms of providing end of life care’ (GP 3)

malignancies…pain, bleeding, breathlessness, agitation, confusion, all those things you get with any cancer…and you get with the haematological malignancy patients. Interviewer: So it’s not actually that different from looking after patients with other cancers. Um, no, not in terms of providing end of life care’ (GP 3) Due to the potential for rapid deterioration and death within hours or days, home-discharge (when preferred) could be required at short notice. While some hospitals had systems in place to organise this (eg, integrated discharge teams), others did not. In the latter case, the haematology nurses found such discharges could be time-consuming and frustrating, particularly if systems could not be put in place before the patient became too ill to be moved. Similarly, hospice beds were not always available at short notice, and inconsistencies in policies concerning the delivery of transfusions (often said to be required to maintain quality of life) represented barriers to referral. ‘when you are trying to get someone home, it takes up an incredible amount of time…phone the hospice, phone the district nurses, get stuff in place, like a bed, sometimes you don’t have time to do it…and the patient is deteriorating in front of your eyes…sometimes, a lot of the time, you have to stop and say, you are too unwell to go home now. I think that’s one of the reasons people don’t die at home’ (Haematology nurse 1) ‘the blocking point for us is that often there is no bed in the hospice, or they can’t deal with transfusion needs’ (Haematology consultant 5)

‘when you are trying to get someone home, it takes up an incredible amount of time…phone the hospice, phone the district nurses, get stuff in place, like a bed, sometimes you don’t have time to do it…and the patient is deteriorating in front of your eyes…sometimes, a lot of the time, you have to stop and say, you are too unwell to go home now. I think that’s one of the reasons people don’t die at home’ (Haematology nurse 1) ‘the blocking point for us is that often there is no bed in the hospice, or they can’t deal with transfusion needs’ (Haematology consultant 5) Inadequate communication across the primary/secondary care interface was regarded as a significant challenge to home death by GPs and community palliative care nurses. Haematological malignancies were described as cancers that were generally managed solely by haematologists in the hospital setting, with patients being encouraged to contact the haematology unit (rather than their GP) as the ‘first port of call’ for advice about their disease. Although GPs received correspondence about hospital appointments and admissions, this process was said to cause them to lose contact with their patients, and know little of their likely prognosis. ‘we can feel out of the loop….you don’t know what’s going on’ (GP 4) In such cases, GPs were sometimes not aware that fast-track hospital discharge may be required to facilitate home death, and they typically felt that they were left with little time to develop relationships with the patient or their relatives before the patient’s death.

‘we can feel out of the loop….you don’t know what’s going on’ (GP 4) In such cases, GPs were sometimes not aware that fast-track hospital discharge may be required to facilitate home death, and they typically felt that they were left with little time to develop relationships with the patient or their relatives before the patient’s death. ‘it’s usually two faxes, one for us (GP) and one for the district nurses, to say this patient is coming home…we’ve done this, we’ve done that, there’s no more active treatment, they will need…and there’s a list of stuff, and that’s sometimes the first we know of it…’ (GP 1) Other important barriers to home death were reported that are likely to be common across other cancers, including gaps in community nursing services (particularly overnight), patients living alone and/or without lay carer’s support and the ability of lay carers to manage at home, particularly in circumstances involving young children, or lay carers who were themselves frail or unwell.

y to be common across other cancers, including gaps in community nursing services (particularly overnight), patients living alone and/or without lay carer’s support and the ability of lay carers to manage at home, particularly in circumstances involving young children, or lay carers who were themselves frail or unwell. Analysis of data from relatives’ interviews revealed that the patients who had died at home (table 1), in line with their wishes, held strong preferences about place of care and death, which were in accord with relatives’ preferences.  They also had a good level of support from family members (who were said to have actively intervened to promote their wishes and ensure these were achieved), had symptoms that could be managed at home, considered nursing services’ support to be adequate and/or had the financial means to ‘buy in’ private carers to plug gaps, and had a GP who was closely involved in their care. Again, many of these factors are likely to be common across patients with other cancers. Table 1 Patients’ (reported) and relatives’ (expressed) PPD, changes over time and actual place of death

Analysis of data from relatives’ interviews revealed that the patients who had died at home (table 1), in line with their wishes, held strong preferences about place of care and death, which were in accord with relatives’ preferences.  They also had a good level of support from family members (who were said to have actively intervened to promote their wishes and ensure these were achieved), had symptoms that could be managed at home, considered nursing services’ support to be adequate and/or had the financial means to ‘buy in’ private carers to plug gaps, and had a GP who was closely involved in their care. Again, many of these factors are likely to be common across patients with other cancers. Table 1 Patients’ (reported) and relatives’ (expressed) PPD, changes over time and actual place of death Study ID Patient’s PPD Relative’s PPD Patient’s change of preference Relative’s change of preference Actual place of death 1 Home Hospice He ‘couldn’t have cared’ (after sudden deterioration) Hospital death was ‘peaceful’ and care was good Hospital 2 Hospice Home Home Unchanged Home 3 Not stated Not stated Not stated Not stated Home 4 Not stated Hospital Not stated Not stated Hospital 5 Hospice Home Not stated ‘with hindsight, hospital was the right place’ (after sudden deterioration) Hospital 6 Not stated Home Not stated Unchanged Home 7 Home Home Unchanged Unchanged Home 8 Home Home Unchanged Unchanged Home 9 Home Home Unchanged Unchanged Home 10 Home Unsure Not stated Hospital was the ‘right place’ Hospital PPD, preferred place of death.

and longer-term follow-up and review may be needed” “Discharging patients after a short intervention, particularly if the service has made a difference to patient and families” “Discharge at the end of the intervention may be difficult as the patient has benefited from the intervention and may be reluctant to stop it. This may be particularly difficult also for carers…” “Ending the contact after specified time as I think we are generally not very good in SPC at discharging patients/terminating our involvement…” “Time constraints - therapy intervention cannot always be addressed and completed within 6–8 weeks and require longer-term follow-up…” Resources “Workforce concerns for current team delivering SIPC” “If successful, there would be funding/capacity issues if the service was to continue” “Time, money” “Time pressures especially if majority want home visits” “Time and resources” Clinician awareness and acceptance “Lack of understanding what can be done. Not appreciating the importance for regular review” “Recognition of need for palliative care” “Making sure all teams aware how to access service & aware of its role” “I think there is a mismatch between what neurology think palliative care can offer and what we think we can offer…” Patient perceptions and acceptance “Patient and carer perceptions of palliative care and the role of specialist palliative care teams” “Reluctance to attend due to preconceptions of hospice being only for people to die in rather than seeing us as a team to help manage symptoms and discharge back to community care - with ongoing support as necessary” “Patients who feel they have already tried everything and are fairly rigid in their approach to trying different ways of doing things, for many versed and valid reasons” MS, multiple sclerosis; SIPC, short-term integrated palliative care; SPC specialist palliative care.

hospital was the right place’ (after sudden deterioration) Hospital 6 Not stated Home Not stated Unchanged Home 7 Home Home Unchanged Unchanged Home 8 Home Home Unchanged Unchanged Home 9 Home Home Unchanged Unchanged Home 10 Home Unsure Not stated Hospital was the ‘right place’ Hospital PPD, preferred place of death. Theme 5: suggested changes to practice to support home or hospice death Interviewees suggested a number of changes that they considered could facilitate home/hospice death. Early initiation of frank conversations about prognosis (where possible), alongside timely integration of haematology and palliative care specialists (including at diagnosis or while active treatment was still ongoing), was considered important by all interviewees. While some discussed difficulties identifying the appropriate time to involve palliative care clinicians, others considered mutual understanding of their roles as underpinning timely referrals. Cited methods of promoting joint working included co-location of facilities, shared clinics and integration of palliative care staff into haematology multidisciplinary team meetings. Interviewees described many such systems as already being in place, and most reported increasing collaboration over time. ‘we’re now in the same building…at the moment on the (palliative care) inpatient unit we’ve got two people with haematological malignancies here now and I’ve got one or two on the books…in the past my experience was that we got very few referred…now we’re all in one unit…we’re getting referrals through’ (Palliative doctor 5)

now in the same building…at the moment on the (palliative care) inpatient unit we’ve got two people with haematological malignancies here now and I’ve got one or two on the books…in the past my experience was that we got very few referred…now we’re all in one unit…we’re getting referrals through’ (Palliative doctor 5) ‘there is merit I think to look at scoping the need or benefits of having a joint clinic between palliative care and haematology to help identify those patients…who need palliative input’ (Palliative nurse 5) ‘over the time I’ve worked in this area, over 9 years, I think that it’s changing, haematologists are more readily involving palliative care at an earlier stage’ (Palliative nurse 3) Another common suggestion was for consistent, clear and detailed documentation of end-of-life discussions, and electronic transfer of this across care interfaces. Primary care practitioners discussed the Gold Standards Framework21 (a system that many of the study hospital-based clinicians were unaware of) and how use of the surprise question (‘would you be surprised if this patient were to die in the next few months, weeks, days?’), may help to identify patients nearing end of life.22 It was largely agreed that such a system would facilitate prognostic estimations in secondary care, and if communicated to primary care, may enable fast-track discharge. It was considered limited, however, in the context of sudden death.

t few months, weeks, days?’), may help to identify patients nearing end of life.22 It was largely agreed that such a system would facilitate prognostic estimations in secondary care, and if communicated to primary care, may enable fast-track discharge. It was considered limited, however, in the context of sudden death. ‘communication between the teams, secondary and primary care, and also of course from the haematology team as well, is essential…we need a clear indication—this is where we are and what we are or not doing, and this is what we might or might not be able to do with this patient—is a key, a key thing I think’ (Palliative doctor 4) ‘oncologists, when they’re writing to the GPs with clinic letters… there is now a drive for them to say, it would be appropriate for you to put this patient on your Gold Standards Framework register…um, that’s actually happening, I’ve seen it’ (Palliative doctor 5) Other suggestions include early mobilisation of community-based services (district, Marie Curie and Macmillan nurses and ‘hospice at home’ services), and if preference is for home care and death, the setting up of mechanisms for anticipatory prescribing and arranging the installation of equipment (eg, hospital beds, syringe drivers).

ns include early mobilisation of community-based services (district, Marie Curie and Macmillan nurses and ‘hospice at home’ services), and if preference is for home care and death, the setting up of mechanisms for anticipatory prescribing and arranging the installation of equipment (eg, hospital beds, syringe drivers). Discussion Hospital deaths among patients with haematological malignancies were largely determined by the characteristics and pathways of these diseases, which resulted in uncertain trajectories, indistinct transitions and difficulties predicting prognosis and identifying if or when to withdraw treatment, factors which complicated advance planning. Delivery of care defined as aggressive in other studies,5–8 such as treatment close to the time of death, was often considered unavoidable and appropriate in the context of blood cancers, for both disease and symptom control. This suggests that the criteria currently used to assess end-of-life care may not be appropriate for haematological malignancies, a notion alluded to by others, but not always agreed on.23–25

f death, was often considered unavoidable and appropriate in the context of blood cancers, for both disease and symptom control. This suggests that the criteria currently used to assess end-of-life care may not be appropriate for haematological malignancies, a notion alluded to by others, but not always agreed on.23–25 Much literature has emerged in recent years, both promoting home death and questioning whether this is always preferable to hospital death.26–28 We found patient/relative choice to be commonly cited as the reason for hospital death and this explains to some extent the large proportions of deaths previously reported in this setting across disease subtypes, and not only among those who die suddenly.10 The desire among haematology doctors and nurses to provide terminal care for patients they have known and treated for many years is understandable, particularly among experienced staff who are skilled in managing end-of-life issues, and in situations where specialist palliative care support is also available.

ddenly.10 The desire among haematology doctors and nurses to provide terminal care for patients they have known and treated for many years is understandable, particularly among experienced staff who are skilled in managing end-of-life issues, and in situations where specialist palliative care support is also available. The close relationship between haematologists and their patients undoubtedly underpins the concepts of ‘co-dependency’ and ‘the co-production of optimism’, which stem from the emotional and relational ties that develop within medical contexts.29 30 This is particularly important in blood cancers, where the most common barriers to quality end-of-life care have been described as unrealistic patient and clinician expectations, and clinician concerns about taking away hope.25 Complex situations have been described where haematologists must attempt to maintain hope while supporting patients to aim for cure, yet concurrently prepare for death.31

s to quality end-of-life care have been described as unrealistic patient and clinician expectations, and clinician concerns about taking away hope.25 Complex situations have been described where haematologists must attempt to maintain hope while supporting patients to aim for cure, yet concurrently prepare for death.31 Identifying the optimum time to discuss prognosis and end-of-life preferences with haematology patients and their families is crucial, yet often difficult to achieve due to the propensity for sudden deterioration and unexpected death. More accurate prognostic data, frank early discussions (possibly near to the time of diagnosis) about potential outcomes (good and bad) and the appropriateness of further treatment may generate more realistic expectations. Such discussions are, however, likely to be difficult for some people, regardless of timing and the communication skills of the clinician; and may be impossible if patients die unexpectedly. Although haematologists have been criticised in the past for lacking integration with palliative care services,17 we were informed of significant and increasing co-working between these specialities, often from diagnosis. Death in the hospice setting posed particular difficulties, however, largely due to a lack of available beds when needed and inconsistencies in service provision (ie, availability of blood product transfusions and antibiotics).

d of significant and increasing co-working between these specialities, often from diagnosis. Death in the hospice setting posed particular difficulties, however, largely due to a lack of available beds when needed and inconsistencies in service provision (ie, availability of blood product transfusions and antibiotics). Fast-track hospital discharges were particularly challenging for primary care practitioners to manage. Our findings highlight the importance of sharing information across interfaces as a prerequisite for informed decision-making by GPs, nurses, out-of-hours doctors and paramedics and the prevention of unwanted hospital (re)admissions. Communication of broad life expectancy estimates (eg, identification of patients who might be expected to die within 6 to 12 months) might also enable primary care practitioners to facilitate discussions about end-of-life choices, and prepare for the possibility of home death, should this be preferred. Decision-making could also be supported by the widespread roll out across care settings of electronic systems that facilitate the recording of preferences. Electronic Palliative Care Coordinating systems32 have been introduced since our data collection took place, which may address this need, although views on their utility are mixed.33

uld also be supported by the widespread roll out across care settings of electronic systems that facilitate the recording of preferences. Electronic Palliative Care Coordinating systems32 have been introduced since our data collection took place, which may address this need, although views on their utility are mixed.33 Strengths and weaknesses As far as we are aware, this is the first UK study to explore determinants of hospital deaths in patients with blood cancers. Qualitative data were gathered from a range of key informants (healthcare practitioners and relatives of deceased patients) and a broad range of views and experiences was obtained, facilitating cross-comparison of perspectives between groups of participants and the identification of challenges across healthcare settings. We acknowledge that including bereaved relatives may provide a limited understanding of patient preferences, and that non-inclusion of relatives from minority ethnicity groups, whose perspectives may differ due to cultural factors, could limit generalisability. Additionally, our study drew on information provided by the relatives of patients aged 66–84; the experiences of relatives of younger or older patients may differ. Due to the relatively under-researched nature of the topic, however, these issues may be considered less salient than that of ‘sensitising’ readers to novel information.20

ur study drew on information provided by the relatives of patients aged 66–84; the experiences of relatives of younger or older patients may differ. Due to the relatively under-researched nature of the topic, however, these issues may be considered less salient than that of ‘sensitising’ readers to novel information.20 Comparison to other studies Although we did not identify any other studies specifically examining determinants of hospital deaths across such a wide range of interviewees, our findings are similar to those of other UK and non-UK studies examining end-of-life care more generally.23 34 35 These studies also highlighted differences between haematological malignancies and other cancers, often caused by disease factors (eg, complex transitions and difficulties estimating prognosis), and resulting in late end-of-life discussions and hospital death. They also note the close relationships that develop between haematologists and their patients, and the variable impact this may have on discussing and determining approaches to care at the end of life, including reluctance to dispel hope and feelings of abandonment if patients are handed over to palliative care services. While other studies also reported patient preferences to remain in the hospital to die,36 the notion that it was possible to provide terminal care for haematology patients at home was also supported,37 38 including the possibility of successfully managing issues such as bleeding in the home setting.

tive care services. While other studies also reported patient preferences to remain in the hospital to die,36 the notion that it was possible to provide terminal care for haematology patients at home was also supported,37 38 including the possibility of successfully managing issues such as bleeding in the home setting. Recent studies from the USA report some haematology practitioners viewing palliative care involvement with distrust, perceiving it to be antithetical to cancer care, concerned only with end-of-life care, and contraindicative to ongoing treatment; the name of the service was also considered a barrier to referral.39 40 With the exception of limitations to the supportive treatment offered in hospice settings (eg, transfusions), we mainly found evidence of increasing collaboration and co-working. This is supported in a UK study that identified ‘universally favourable’ reports among haematologists about palliative care services.35 Possible reasons for such variations include differing perceptions and practices between countries.

transfusions), we mainly found evidence of increasing collaboration and co-working. This is supported in a UK study that identified ‘universally favourable’ reports among haematologists about palliative care services.35 Possible reasons for such variations include differing perceptions and practices between countries. Conclusion Hospital deaths were largely determined by the characteristics of haematological malignancies, which included uncertain trajectories and indistinct transitions, and difficulties predicting prognosis and identifying if or when to withdraw treatment. Although end-of-life care for patients with these diseases is often considered aggressive compared with other cancers, the reasons for this may be appropriate and unavoidable. Nonetheless, there is scope to improve practice, including: better information about prognosis (where possible), earlier, frank discussions about likely outcomes and improved communication between care settings. We wish to thank all clinicians and relatives for giving their time to take part in the study, Ruth Hart for checking transcript coding and the study steering committee, particularly David Brown. Contributors: DH, ER, AS, AG, RP and MH designed the study. DH managed the study and conducted the interviews. DM analysed the data. MH, AG, RP and MJ commented on the clinical aspects of the study. DM, DH and ER wrote the first draft of the paper. All authors commented on and approved the final draft. Funding: This work was supported by the Marie Curie Research Grants Scheme, grantC38860/A12554; HMRN is funded by Bloodwise, grant 10042.

Contributors: DH, ER, AS, AG, RP and MH designed the study. DH managed the study and conducted the interviews. DM analysed the data. MH, AG, RP and MJ commented on the clinical aspects of the study. DM, DH and ER wrote the first draft of the paper. All authors commented on and approved the final draft. Funding: This work was supported by the Marie Curie Research Grants Scheme, grantC38860/A12554; HMRN is funded by Bloodwise, grant 10042. Competing interests: None declared. Ethics approval: This study has NHS ethical approval (Yorkshire and the Humber REC: 11/YH/0306). All participants were given verbal and written information about the study aims and their involvement. Informed written consent was sought from individuals prior to interview and assurances given concerning confidentiality and anonymity. Participants were told that they could withdraw from the study at anytime. Provenance and peer review: Not commissioned; externally peer reviewed. Data sharing statement: None declared. Correction notice: This article has been corrected since it published Online First. A funding statement has been added. This paper has been amended since it was published Online First. Owing to a scripting error, some of the publisher names in the references were replaced with ‘BMJ Publishing Group’. This only affected the full text version, not the PDF. We have since corrected these errors and the correct publishers have been inserted into the references.

Background Palliative care has been proposed to help meet the needs of patients with progressive non-cancer conditions such as long-term neurological conditions (LTNCs).1 2 However, there is little or no evidence on the best ways of providing palliative care for these patients. Should it be at the ‘end of life’ or earlier such as at the point of diagnosis? Our own phase II randomised controlled trial (RCT) in patients with multiple sclerosis (MS) showed a reduction in symptoms and caregiving burden, following short-term integrated palliative care (SIPC) compared with standard care.3 More recently, a pilot RCT in Italy evaluating the impact of a new specialist palliative care service for patients with a range of LTNCs found significant improvements in quality of life and physical symptoms.4 Whether more people living with different LTNCs can benefit from SIPC and whether it can be routinely used in practice to improve care quality are of interest.

impact of a new specialist palliative care service for patients with a range of LTNCs found significant improvements in quality of life and physical symptoms.4 Whether more people living with different LTNCs can benefit from SIPC and whether it can be routinely used in practice to improve care quality are of interest. OPTCARE Neuro is a multicentre RCT evaluating the effectiveness of SIPC for progressive LTNCs (ISRCTN18337380). The SIPC service being trialled is defined as three palliative care visits over 6–8 weeks. This is a phase III RCT in patients with a range of LTNCs including: MS, motor neuron disease (MND), idiopathic Parkinson’s disease, progressive supranuclear palsy and multiple system atrophy. The overall aim of OPTCARE Neuro is to evaluate the clinical and cost-effectiveness of SIPC to optimise care for people with LTNCs. In addition to understanding the effectiveness of this service, it is also important to understand and be aware of current service provisions and the views of professionals involved in providing care for this patient group. The complexity of delivering and evaluating palliative care services requires the accumulation of knowledge from multiple sources and will depend on interprofessional behaviours.5 6 It is therefore valuable to explore clinicians’ views and opinions when shaping emerging services and informing future requirements. With that in mind, we conducted a short online survey with neurology and palliative care professionals. The main aims of the survey were to:understand what current levels of collaboration exist between the two specialties;

clinicians’ views and opinions when shaping emerging services and informing future requirements. With that in mind, we conducted a short online survey with neurology and palliative care professionals. The main aims of the survey were to:understand what current levels of collaboration exist between the two specialties; explore the expectations and views of clinicians towards the SIPC service being trialled. Methods Research teams at six UK trial centres (London, Nottingham, Liverpool, Cardiff, Brighton and Chertsey) identified local neurology and palliative care professionals who were then approached via email by the central trial team. Professionals were informed that by completing the survey, they provided informed consent for use of their anonymised data. The surveys consisted of multiple choice or open comment questions, 13 (for neurology) or 10 (for palliative care) with responses collected using online forms. The survey was launched in July 2015 and closed April 2016. The study was approved by the National Research Ethics Service Committee London South East (REC number: 14/LO/1765).

isted of multiple choice or open comment questions, 13 (for neurology) or 10 (for palliative care) with responses collected using online forms. The survey was launched in July 2015 and closed April 2016. The study was approved by the National Research Ethics Service Committee London South East (REC number: 14/LO/1765). Results The survey received responses from 33 neurology and 26 palliative care professionals (20% response rate). Two-thirds of respondents in both groups had over 10 years of experience in their respective fields. Current levels of collaboration between the two specialties were reported as being ‘good/excellent’ by 36% of neurology professionals and by 58% of palliative care professionals. However, nearly half (45%) of neurology compared with only 12% of palliative care professionals rated current levels as being ‘poor/none’ (see figure 1). When asked if there were any particular disease areas where links were better, both groups reported stronger links for MND. In addition, both professional groups felt that the new SIPC service being trialled would influence future collaborations for the better (65%–70% in both groups). Figure 1 Neurology and palliative care professionals’ ratings of their current levels of collaboration with the other specialty.

Results The survey received responses from 33 neurology and 26 palliative care professionals (20% response rate). Two-thirds of respondents in both groups had over 10 years of experience in their respective fields. Current levels of collaboration between the two specialties were reported as being ‘good/excellent’ by 36% of neurology professionals and by 58% of palliative care professionals. However, nearly half (45%) of neurology compared with only 12% of palliative care professionals rated current levels as being ‘poor/none’ (see figure 1). When asked if there were any particular disease areas where links were better, both groups reported stronger links for MND. In addition, both professional groups felt that the new SIPC service being trialled would influence future collaborations for the better (65%–70% in both groups). Figure 1 Neurology and palliative care professionals’ ratings of their current levels of collaboration with the other specialty. Participants were also asked what they thought would be the main barriers for the new SIPC service (see table 1). The most commonly identified barriers by neurologists were resources, clinician awareness of services offered, continuing collaborations and communication between teams beyond the trial and geographical limitations. Similarly, palliative care professionals also identified resources and clinician awareness (and, importantly, the appropriateness of referrals they may receive) as barriers. However, the key barrier they identified was that there may be a possible need for longer-term care beyond that offered by the SIPC service. They also drew attention to patient perceptions of palliative care as a potential barrier.

eness (and, importantly, the appropriateness of referrals they may receive) as barriers. However, the key barrier they identified was that there may be a possible need for longer-term care beyond that offered by the SIPC service. They also drew attention to patient perceptions of palliative care as a potential barrier. Table 1 Barriers identified by professionals

eness (and, importantly, the appropriateness of referrals they may receive) as barriers. However, the key barrier they identified was that there may be a possible need for longer-term care beyond that offered by the SIPC service. They also drew attention to patient perceptions of palliative care as a potential barrier. Table 1 Barriers identified by professionals Neurology professionals Resources “Workload volume” “Increasing number of referrals will put the service to the same problem as in other healthcare foci - waiting times and availability may fall behind which happens to all services sooner or later. Sadly, the better services the sooner you will use your resources” “Resources and how best to communicate with varied services/provision” “Social care funding. Pent up demand” “Cost” Clinician awareness and acceptance “Knowing what services are available in localities” “Awareness, especially in primary care” “Clinician awareness” “Senior medics previous practices” “Getting themselves trusted by the consultants” Continuing collaborations and communication “Creating good links between the MS team and the palliative care team” “Communication issues between different care providers” “Ongoing joined-up work to ensure care continuity and no repetition of service provision” “Establishing a robust and efficient process for communicating with all disciplines involved in the patient’s care …” Geography “Geographical limitations - a lot of our patients are not local” “Geographic’s” “Many of our patients live a long way from the centre to allow active engagement” Patient perceptions and acceptance “Patient resistance” “The term hospice which often patients and families feel has a strong association with immanency of dying. People often express fear of contact with a hospice if they do not feel that they are close to death” Palliative care professionals Longer-term service provision “Recognising the need but not being able to back it up with longer-term intervention” “Short-term intervention and longer-term follow-up and review may be needed” “Discharging patients after a short intervention, particularly if the service has made a difference to patient and families” “Discharge at the end of the intervention may be difficult as the patient has benefited from the intervention and may be reluctant to stop it.

mmunity care - with ongoing support as necessary” “Patients who feel they have already tried everything and are fairly rigid in their approach to trying different ways of doing things, for many versed and valid reasons” MS, multiple sclerosis; SIPC, short-term integrated palliative care; SPC specialist palliative care. Discussion Our results demonstrate that collaborations can be improved and both specialties are positive about the impact the new SIPC service will make. However, the barriers identified highlight areas for consideration and further exploration. Patient perceptions of palliative care was identified as a potential barrier to the successful integration of neurology and palliative care services. It is equally important for neurology professionals to have the right understanding of palliative care and to recognise the potential benefit of palliative care for their patients. Indeed, previous studies have demonstrated that the topic of palliative care can still often lead to anxiety in patients, caregivers as well as healthcare professionals.7 8 There is an emphasis on the need for integrated working along with improved education and awareness in order to make palliative care more recognised and more accessible for non-cancer conditions such as LTNCs.9 10 As reported by both specialties, resources must be carefully considered and systems developed for calling on palliative care specialists when truly necessary. The small number of respondents highlights the challenges of conducting research among busy health professionals; however, the geographical variation is an advantage of the survey.

specialties, resources must be carefully considered and systems developed for calling on palliative care specialists when truly necessary. The small number of respondents highlights the challenges of conducting research among busy health professionals; however, the geographical variation is an advantage of the survey. Conclusions Our results demonstrate the opportunity to increase collaboration between neurology and palliative care services for people with progressive neurological conditions, and the acceptability of SIPC as a model to support this. This survey will be repeated at the end of the trial to understand how collaborations and views have changed, whether the SIPC service has affected the care process and to identify areas for improvement. These survey results will be integrated with the qualitative trial findings to provide a wider context about the effects of SIPC on the processes of care, and the ways in which it might be working effectively. The authors thank the research teams in the six UK sites for facilitating this work and the neurology and palliative care professionals who took the time to complete the survey. The authors also thank the OPTCARE Neuro PPI committee, in particular Dr Cynthia Benz, for their comments and support in the preparation of this report.

s thank the research teams in the six UK sites for facilitating this work and the neurology and palliative care professionals who took the time to complete the survey. The authors also thank the OPTCARE Neuro PPI committee, in particular Dr Cynthia Benz, for their comments and support in the preparation of this report. Contributors: NH was responsible for the data collection, data analysis and for the initial draft and subsequent versions of this manuscript. WG, CJE, DJ, LMW, AB, VC, KEG, FL and IJH all contributed to the conception and design of the work, commented on the initial draft and approved the final version of the manuscript. Funding: WG, CJE, DJ, AB, VC and IJH had financial support from the National Institute for Health Research (NIHR), Health Services & Delivery Research programme (HS & DR) (12/130/47) for the submitted work. WG and IJH also had financial support from the NIHR Collaboration for Leadership in Applied Health Research & Care (CLAHRC), South London. In addition, CJE is funded by an Health Education England/NIHR Senior Clinical Lectureship and IJH holds an NIHR Emeritus Senior Investigator award. Disclaimer: The views expressed are those of the authors and not necessarily those of the National Health Service, NIHR or Department of Health. Competing interests: None declared. Ethics approval: National Research Ethics Service Committee London South East. Provenance and peer review: Not commissioned; externally peer reviewed.

The development of sensor-based technologies has led to the creation of wearable health monitoring systems to provide real-time feedback information to support health.1 This technology has the potential to assist the monitoring and management of symptoms in advanced disease, such as pain management.2 Within our university, we have developed a partnership between Computer Science and Palliative Medicine to examine the potential to use technology to support care for people living with advanced illness. One outcome of this collaboration is the development of a prototype smartwatch app to monitor pain experienced by patients. Two school students, completing a summer placement at the University, programmed this software. The app facilitates pain assessment (in numerical rating and descriptive scale formats) and syncs the data with a mobile device to enable storage and retrieval of clinically relevant health data over time. The King’s Fund has highlighted the potential of new emerging technologies to personalise care to patients according to individual circumstances.3 4 For patients living with advanced illness, there is potential to use collected sensor data (for variables such as movement, heart rate and activities of daily living) from wearable devices and living environments, to determine how symptoms (such as pain) affect function and quality of life.5

ndividual circumstances.3 4 For patients living with advanced illness, there is potential to use collected sensor data (for variables such as movement, heart rate and activities of daily living) from wearable devices and living environments, to determine how symptoms (such as pain) affect function and quality of life.5 Undoubtedly, the future of healthcare is one in which wearable devices and associated applications will assist the management of health and disease. The challenge for clinicians and computer scientists is to forge collaborative partnerships to understand the needs and opportunities available, and develop acceptable and effective technology. Contributors: ACN and TRP conceived the project. CQ and JS initiated the project. ACN and TRP provided supervision. ACN, CQ, JS, SM and TRP contributed to refinement of the project, wrote the paper and approved the final manuscript. Competing interests: None declared. Provenance and peer review: Not commissioned; internally peer reviewed. Data sharing statement: Further information about the project is available on request by contacting the corresponding author at anwosu@liverpool.ac.uk.

Introduction Advance care planning for the end of life is a responsibility of a growing number of healthcare professionals and an expanding niche for clinical and consumer IT products. Of the 10 leading causes of death worldwide, only one is a type of cancer, the historical priority of palliative care services.1 Non-communicable diseases accounted for 68% (38 million) of the world's deaths in 2012.2 End of life care is thus falling increasingly within the remit of chronic disease specialists. The specialised and often fragmented nature of healthcare in the developed world and the complexity of end of life needs already demand multiprofessional involvement: 36.1% of US Medicare beneficiaries were treated by 10 or more doctors in the last 6 months of life (2007 data).3 Generalist care, including out of hours, is also key: in a UK study of over 21 500 patients, general practitioners (GPs) were estimated to have recorded data on 7.3 days in the final 90 days of a patient's life.4 Online consumer products, such as My Directives (https://mydirectives.com), Best Endings (http://www.bestendings.com) and My Living Will (https://www.mylivingwill.org.uk), are also becoming popular. We discuss England's pre-eminent initiative in enabling advance care planning and improved communication and coordination at the end of life—the Electronic Palliative Care Coordination Systems. EPaCCS are also a proof of concept for other conditions requiring multiprofessional and multisetting involvement, communication and coordination around expressed patients' wishes and a range of sources of patient data.

ommunication and coordination at the end of life—the Electronic Palliative Care Coordination Systems. EPaCCS are also a proof of concept for other conditions requiring multiprofessional and multisetting involvement, communication and coordination around expressed patients' wishes and a range of sources of patient data. EPaCCS take various forms, which are still insufficiently documented—of web-based electronic registers, systems based on sharing care summaries and plans alongside patients' electronic records, patient portals, real-time extractions from the records of participating organisations, etc. They aim to provide up-to-date key information about patients believed to be in the last year of their life in GP practices, emergency telephone services (111 and 999), GP out of hours services, accident and emergency departments, ambulance services, hospitals, community nursing teams, specialist palliative care services, hospices and care homes. In the UK, each of these settings has their own (electronic or paper) patient record. EPaCCS aim to improve communication and coordination and ensure that all those involved in a patient's care are aware of their wishes, preferences and advance care plan. They are expected, and to an extent have been demonstrated, to enable more patients to die at their preferred place and reduce unnecessary hospital admissions and ambulance journeys, inappropriate interventions, use of unscheduled care and repeated ‘difficult conversations’.5 Provided that they have well-developed reporting functions, EPaCCS also supply detailed outcome metrics and enable continuous quality improvement in local end of life care services.

hospital admissions and ambulance journeys, inappropriate interventions, use of unscheduled care and repeated ‘difficult conversations’.5 Provided that they have well-developed reporting functions, EPaCCS also supply detailed outcome metrics and enable continuous quality improvement in local end of life care services. Striking outcomes have been reported around EPaCCS. For instance, 82.4% of the patients of ‘Coordinate My Care (CMC)’ have died outside of hospital; 77.8% died in their preferred place, with 71.8% of them dying in the place of their first preference (May 2016 data, since inception). Levels of hospital death in patients without cancer on and off the South West EPaCCS were found to be 8.3% vs 49.4%, respectively.6 Yet EPaCCS also face immense challenges, including ones of implementation, sustainability, cost-effectiveness, equality and service capacity.

t preference (May 2016 data, since inception). Levels of hospital death in patients without cancer on and off the South West EPaCCS were found to be 8.3% vs 49.4%, respectively.6 Yet EPaCCS also face immense challenges, including ones of implementation, sustainability, cost-effectiveness, equality and service capacity. EPaCCS take the idea of Locality Registers, proposed by the 2008 End of Life Care Strategy,7 into the age of digital communication. Evidence remains sparse. There are two evaluations of early implementers (2011 and 2013);6 8 a recent national evaluation offering tentative conclusions, reflecting the substantial challenges of EPaCCS data collection and comparison;9 descriptive studies of CMC incorporating audit data;10–13 internal project reports,14 15 only a small part of which are publicly available, and conference abstracts.16–20 The only completed peer-reviewed evaluation has substantial methodological limitations.21 Most evidence on EPaCCS—often anecdotal, at times overinterpreted by virtue of commitment to the initiative, and with limited information on methods and contexts—is solely accessible through policy documents and internal reports.

The only completed peer-reviewed evaluation has substantial methodological limitations.21 Most evidence on EPaCCS—often anecdotal, at times overinterpreted by virtue of commitment to the initiative, and with limited information on methods and contexts—is solely accessible through policy documents and internal reports. In terms of their multifunctionality and ambition, EPaCCS appear specific to the UK. They aim to coordinate care across all potential providers of care at the end of life; store a dynamic record of a patient's condition, treatment, wishes and preferences; enable advance care planning and accumulate data for service evaluation, quality improvement and research. Other countries have developed tools addressing some of these aims. Most notably, many maintain registers or databases on end of life care provision and outcomes aimed at quality assessment and improvement and research rather than care coordination. Examples include the Swedish National Quality Register in End of Life Care (The Swedish Register of Palliative Care, SRPC), which, as of 2015, contained data from two-thirds of all deaths in Sweden,22 23 the Danish Palliative Database24 25 and the PCOC (Palliative Care Outcomes Collaboration) of Australia (http://ahsri.uow.edu.au/pcoc/about/index.html). Transnational initiatives have also been tested, as in the EURO SENTIMELC study.26

RPC), which, as of 2015, contained data from two-thirds of all deaths in Sweden,22 23 the Danish Palliative Database24 25 and the PCOC (Palliative Care Outcomes Collaboration) of Australia (http://ahsri.uow.edu.au/pcoc/about/index.html). Transnational initiatives have also been tested, as in the EURO SENTIMELC study.26 In the USA, there is a growing use of electronic registries of Physician Orders for Life-Sustaining Treatment (POLST).27–29 POLST registries appear to focus entirely (Utah, Oregon)27 30 or primarily (New York State)31 on patient preferences and wishes concerning treatment options such as resuscitation efforts or use of feeding tubes or intravenous fluids. In addition to providing a record of a patient's wishes and preferences, EPaCCS also contain a summary of their condition and care, medication information and details of carers and services involved, among others. POLST are legally binding, whereas most wishes and preferences recorded on EPaCCS are not. POLST registries thus tend to contain more legal fields, for instance concerning witness requirements, concurrent physician opinion and process of reaching a mental capacity decision.31

carers and services involved, among others. POLST are legally binding, whereas most wishes and preferences recorded on EPaCCS are not. POLST registries thus tend to contain more legal fields, for instance concerning witness requirements, concurrent physician opinion and process of reaching a mental capacity decision.31 The Scottish electronic Palliative Care Summary (ePCS) is a central system updated two times per day from GP records, introduced in 2008 and fully rolled out by 2010.32 33 An audit in Lothian found that 75% (year 2012) and 71% (year 2013) of specialist palliative care patients had electronic information available to out of hours services: a significant improvement relative to 49% in 2008.32 Ali et al 34 reported that not having an ePCS was associated with a higher risk of hospital admission (OR=2.43). ePCS have been regarded as a success and led to the development of a new Electronic Key Information Summary for patients with long-term conditions.35 England's EPaCCS and Scotland's ePCS are similar in the information covered and services involved, but also differ in substantial ways. For instance, ePCS is a centralised system, whereas EPaCCS are numerous and varied. A comparison of the two approaches is outstanding.

Summary for patients with long-term conditions.35 England's EPaCCS and Scotland's ePCS are similar in the information covered and services involved, but also differ in substantial ways. For instance, ePCS is a centralised system, whereas EPaCCS are numerous and varied. A comparison of the two approaches is outstanding. Two further developments are worth mentioning in contextualising EPaCCS: large-scale disease/treatment registers and health information exchange (HIE) initiatives. Many countries have well-established cancer registers. There is also a growing number of large-scale registers of other life-threatening conditions or treatments of these (eg, Switzerland's AMIS Plus national registry for patients with acute coronary syndrome and the joint Australia and New Zealand Dialysis and Transplant (ANZDATA) Registry).36 37 Such registers perform some of the functions of EPaCCS, including patient identification and the generation of data for service evaluation, quality improvement and research. Finally, EPaCCS are a HIE initiative, as much as they have been discussed primarily in an end of life care context rather than healthcare IT. The number of HIE initiatives is rapidly growing. In the USA for instance, a 2012 survey found that 1398 hospitals (30%) and 23 341 ambulatory practices (10%) were participating in 119 operational HIE efforts, relative to 14% of hospitals, 3% of ambulatory practices and 75 operational HIE efforts 2 years earlier.38 A conceptual systematic review is needed to contextualise EPaCCS in the global availability of palliative and end of life registers, registers of life-threatening conditions and treatments for these and the broader HIE literature.

f hospitals, 3% of ambulatory practices and 75 operational HIE efforts 2 years earlier.38 A conceptual systematic review is needed to contextualise EPaCCS in the global availability of palliative and end of life registers, registers of life-threatening conditions and treatments for these and the broader HIE literature. Sources of data and discussion points We present a critical analysis of EPaCCS against a background of scarce research, limited public awareness and insufficient openness in policy documents and the official discourse. We also offer an extensive structured framework for comparing the features, contexts and key outcomes of existing EPaCCS for use in future evaluations, benchmarking and funding decisions (see online supplementary tables: table S1 includes >60 parameters, table S2 lists the 51 items from the National Information Standard and 12 further fields). Finally, we make suggestions for the way forward, some of which represent a radical shift from what might be a natural trajectory of development. 10.1136/bmjspcare-2015-001059.supp1Supplementary tables

Sources of data and discussion points We present a critical analysis of EPaCCS against a background of scarce research, limited public awareness and insufficient openness in policy documents and the official discourse. We also offer an extensive structured framework for comparing the features, contexts and key outcomes of existing EPaCCS for use in future evaluations, benchmarking and funding decisions (see online supplementary tables: table S1 includes >60 parameters, table S2 lists the 51 items from the National Information Standard and 12 further fields). Finally, we make suggestions for the way forward, some of which represent a radical shift from what might be a natural trajectory of development. 10.1136/bmjspcare-2015-001059.supp1Supplementary tables The paper draws primarily on evidence and lessons from three projects: London's ‘Coordinate My Care (CMC)’, the South West EPaCCS and the Cambridgeshire & Peterborough Project for Data Sharing in End of Life Care (C&P Project). Online supplementary tables S1 and S2 summarise their characteristics. CMC is the largest EPaCCS in England, available to a population of over 9 million people. So far, 29 083 CMC urgent care plans have been created (May 2016 data, since inception, August 2010). It is also the system with the most extensive support infrastructure of all EPaCCS, including user support and training, information materials, clinical and information governance (IG) infrastructure, reporting and analysis work and high profile publicity. The South West EPaCCS is one of the earliest EPaCCS projects, initiated in 2008 before the national pilot (2009–2011)8 and later part of it. It also works in close collaboration with the National End of Life Care Intelligence Network (NEoLCIN) (http://www.endoflifecare-intelligence.org.uk/home). For a proportion of its users, the C&P Project is based on data sharing integrated with routine record keeping—the vision for the ‘ideal EPaCCS’, provided such integration can be achieved for all users.

ith the National End of Life Care Intelligence Network (NEoLCIN) (http://www.endoflifecare-intelligence.org.uk/home). For a proportion of its users, the C&P Project is based on data sharing integrated with routine record keeping—the vision for the ‘ideal EPaCCS’, provided such integration can be achieved for all users. Sources of data and discussion points include: Audit and evaluation data from internal monthly reports and public documents (CMC), existing evaluations (South West) and in-progress evaluations (C&P Project). References are given for published data; none come from peer-reviewed papers. Key costs data from the CMC and C&P Project teams. Data on features of the geographical areas covered: end of life care statistics from the NEoLCIN;39 data on population sizes and numbers of GP practices from NHS England;40 data on clinical services and teams obtained through Freedom of Information requests (9 for the South West, 1 for London, January–August 2015), key informants and team exercises in local service mapping. Descriptive EPaCCS data from the project leads and coauthors, organised around a framework developed for this paper (see online supplementary tables S1 and S2b) and the National Information Standard for EPaCCS (see online supplementary table S2a).

Data on features of the geographical areas covered: end of life care statistics from the NEoLCIN;39 data on population sizes and numbers of GP practices from NHS England;40 data on clinical services and teams obtained through Freedom of Information requests (9 for the South West, 1 for London, January–August 2015), key informants and team exercises in local service mapping. Descriptive EPaCCS data from the project leads and coauthors, organised around a framework developed for this paper (see online supplementary tables S1 and S2b) and the National Information Standard for EPaCCS (see online supplementary table S2a). Anecdotal evidence and discussion points concerning the drivers and challenges of the projects and the way forward from the project leads and coauthors. These were obtained through structured and unstructured email exchanges, interviews, meetings and telephone calls, the majority of which conducted between June and September 2014. The national EPaCCS Conferences in July 2015 and March 201641 42 and a critical review of policy documents were additional sources of data and considerations. These data and discussion points were brought together by the first author in an extended internal report (available from the corresponding author), later distilled for this paper.

The national EPaCCS Conferences in July 2015 and March 201641 42 and a critical review of policy documents were additional sources of data and considerations. These data and discussion points were brought together by the first author in an extended internal report (available from the corresponding author), later distilled for this paper. Stage of development, outcomes and costs of EPaCCS A 2013 survey suggested that work was underway on at least 82 EPaCCS—33 developed in partnerships involving 139 of the 211 Clinical Commissioning Groups (CCGs) in England and 49 independently.43 In November 2014, 91 (43%) CCGs reported that they had a functioning EPaCCS (or similar system), 53 (25%) had ‘plans’ and 2 (1%) had no plans for an EPaCCS. Progress was unknown in 65 (31%) CCGs.44 It is unclear how commonly used descriptors of ‘functioning’ and ‘operational’ relate to the usability and use of EPaCCS. A total of 26 249 patients were on the 18 systems (49 CCGs) for which data were provided in the 2013 survey.43 More up-to-date national data are not yet available although data collection for an EPaCCS Baseline Review 2015/2016 has been completed (to be published by NHS England in 2016/2017).9 Even established systems achieve far from optimal coverage: CMC, for instance, covers 16.6% of its estimated end of life population (December 2014).

ate national data are not yet available although data collection for an EPaCCS Baseline Review 2015/2016 has been completed (to be published by NHS England in 2016/2017).9 Even established systems achieve far from optimal coverage: CMC, for instance, covers 16.6% of its estimated end of life population (December 2014). Striking outcomes have been observed for EPaCCS patients, yet high quality evidence is lacking. Data from the South West on 3012 EPaCCS patients and over 67 000 total deaths demonstrated differences in hospital deaths of 9.8% vs 33.9% for patients with cancer and 8.3% vs 49.4% for patients without cancer, respectively, on and off EPaCCS.6 Of the CMC patients who died (12 362) between August 2010 and May 2016, 7614 had a preferred and actual place of death documented and 77.8% of these died in their preferred place (first preference achieved for 71.8%). An independent evaluation suggested average savings of £2100 per person who dies with a CMC care plan, through reductions in hospital attendance and length of stay aligned with patient wishes.45 My Care Choices, the Essex EPaCCS, reported that 43% of all deaths in their 38 participating practices were preidentified by the register and 86% of those patients were able to die in their preferred place.20 Estimates from the Nottinghamshire EPaCCS, concerning a caseload averaging 3345 patients, suggested a 45% lower rate of hospital admissions for EPaCCS patients (2.5% vs 4.5%).14 Data from the Bedfordshire EPaCCS show that 69.7% (235/337) of patients have not been conveyed to hospital after a contact between the ambulance service and the service hosting the EPaCCS.46 Those are impressive outcomes, but ones concerning potentially biased patient samples, affected by limited opportunities to control confounders reliably, representing evidence of association, not causation, and coming from a small range of successful projects with good reporting capabilities. The latest national evaluation9 was not able to identify a statistically significant difference between EPaCCS and non-EPaCCS sites in terms of improvements in death in usual place of residence, hospital admissions and resource use. An economic assessment could not be completed, largely due to limitations of the data that could be collected. Rigorous evaluation and research is urgently needed.

nificant difference between EPaCCS and non-EPaCCS sites in terms of improvements in death in usual place of residence, hospital admissions and resource use. An economic assessment could not be completed, largely due to limitations of the data that could be collected. Rigorous evaluation and research is urgently needed. There is a concern that EPaCCS projects may be under-resourced. The 2013 Economic Evaluation uses as its default assumption costs of £21 104 for set-up and £8235 for annual maintenance, per 200 000 population per annum, with expectations of wider end of life care investment.6 While IT costs may be low with simpler solutions, those projects demand resources for securing stakeholder support, providing training and information, negotiating workflow and system changes, upgrading the EPaCCS, auditing and reporting, etc. Between August 2010 and April 2014, CMC cost ∼£1.5 m. Using the Economic Evaluation reference points, adjusting for a population size of 9.27 m and assuming 1.75 years have been dedicated to set-up and 2 years to annual maintenance, the projected cost of CMC is over £1.7 m for this time period. The C&P Project cost £245 000 between July 2012 and March 2016, whereas its estimated cost is £165 109 (adjusting for a population size of 0.86 m, and accounting for 20 months for set-up and 2.1 years for maintenance). While substantial progress has been made in local data sharing-in end of life care and more broadly-much further work remains to be undertaken.

2012 and March 2016, whereas its estimated cost is £165 109 (adjusting for a population size of 0.86 m, and accounting for 20 months for set-up and 2.1 years for maintenance). While substantial progress has been made in local data sharing-in end of life care and more broadly-much further work remains to be undertaken. Key challenges Projects need to involve nine key service types and hundreds of individual settings An EPaCCS project needs to involve, at an advanced though still imperfect level of uptake, GP practices, emergency telephone lines, ambulance services, GP out of hours services, hospitals, community nursing teams, specialist palliative care services, hospices and care homes. This translates into huge numbers of settings and teams, for example >330 for C&P, >1640 for the South West EPaCCS and >4600 for CMC (see online supplementary table S1). Most of these settings and teams have significant autonomy. There are no established levers for joint action of such a wide scope within a local health economy covered by an EPaCCS. An added challenge is that services with a broad coverage (eg, ambulance trusts) and/or close to geographical and health system boundaries will need to work with multiple EPaCCS.

gnificant autonomy. There are no established levers for joint action of such a wide scope within a local health economy covered by an EPaCCS. An added challenge is that services with a broad coverage (eg, ambulance trusts) and/or close to geographical and health system boundaries will need to work with multiple EPaCCS. EPaCCS teams start a register project and find themselves transforming systems and culture The nature and scope of EPaCCS projects is not apparent from existing guidance. When EPaCCS cover large areas and populations and have ambitious goals, they cannot be simply ‘end of life register’ projects. They become complex, long term, resource hungry initiatives that need to uncover and optimise existing care pathways, change workflows, patterns of collaboration and culture, educate health professionals and break new ground in data sharing. The realities of healthcare IT are far from the expectations we have from our daily IT lives Clinical information systems in the UK are not interoperable, in spite of the enticing visions posing as round-the-corner realities in political speeches and in the marketing and advertising materials of IT companies. While mobile working is integral to our daily lives, paper records persist in many care settings.

ives Clinical information systems in the UK are not interoperable, in spite of the enticing visions posing as round-the-corner realities in political speeches and in the marketing and advertising materials of IT companies. While mobile working is integral to our daily lives, paper records persist in many care settings. Most EPaCCS developers have chosen one of two suboptimal solutions while working on or awaiting a new level of record integration: web-based solutions external to the record keeping system of any of the settings involved or ones internal to a locally dominant system. The former allow breadth of access and equality but require separate log-ins and double data entry, with the concomitant disruption of workflows, risks of error, out of date information and opportunity costs. The latter enable continuity with routine record keeping for users of the dominant system, but disadvantage users of other systems and, through this, the patients cared for by them. New generation solutions appear tantalisingly near—for example, the soon-to-be-launched My Right Care (http://www.myrightcare.co.uk) promises automatic population of a joint care plan from ‘all relevant services’ and an external, yet seamlessly integrated, interface but is still to pass the go-live challenges.

by them. New generation solutions appear tantalisingly near—for example, the soon-to-be-launched My Right Care (http://www.myrightcare.co.uk) promises automatic population of a joint care plan from ‘all relevant services’ and an external, yet seamlessly integrated, interface but is still to pass the go-live challenges. IG-related decision-making for EPaCCS projects is not backed by a clear framework The IG documents in box 1 add up to 629 pages, yet they do not address many of the IG issues associated with an EPaCCS. Some IG specialists concede that IG rules need to be ‘bent’ or controversially interpreted so that EPaCCS projects are not obstructed or abandoned, while the lawfulness of decisions is ensured within broader legal and governance frameworks. Box 1 Basic IG documents potentially relevant to EPaCCS developers IG in EPaCCS Electronic Palliative Care Co-ordination Systems: Information Governance Guidance, January 2014, 20 pp. http://systems.hscic.gov.uk/qipp/library/epaccsig.pdf Key documents from the broader IG framework Data Protection Act, 1998, 133 pp. http://www.legislation.gov.uk/ukpga/1998/29/data.pdf The Information Governance Review (Caldicott 2 Review), March 2013, 142 pp. https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/192572/2900774_InfoGovernance_accv2.pdf Royal College of General Practitioners. Patient Online: The Road Map, March 2013, 63 pp. http://elearning.rcgp.org.uk/mod/page/view.php?id=4709 Law Commission. Data Sharing between Public Bodies. A Scoping Report, July 2014, 179 pp.

https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/192572/2900774_InfoGovernance_accv2.pdf Royal College of General Practitioners. Patient Online: The Road Map, March 2013, 63 pp. http://elearning.rcgp.org.uk/mod/page/view.php?id=4709 Law Commission. Data Sharing between Public Bodies. A Scoping Report, July 2014, 179 pp. https://www.gov.uk/government/publications/data-sharing-between-public-bodies-a-scoping-report Information Commissioner's Office. Data Sharing Code of Practice, May 2011, 58 pp. https://ico.org.uk/media/for-organisations/documents/1068/data_sharing_code_of_practice.pdf Key public facing documents The Care Record Guarantee, 15 pp. http://systems.hscic.gov.uk/rasmartcards/documents/crg.pdf Keeping your online health and social care records safe and secure (British Computer Society booklet for the general public), 19 pp. http://www.nhs.uk/nhsengland/thenhs/records/healthrecords/documents/patientguidancebooklet.pdf

Key public facing documents The Care Record Guarantee, 15 pp. http://systems.hscic.gov.uk/rasmartcards/documents/crg.pdf Keeping your online health and social care records safe and secure (British Computer Society booklet for the general public), 19 pp. http://www.nhs.uk/nhsengland/thenhs/records/healthrecords/documents/patientguidancebooklet.pdf End of life care is an emotive and uncertain domain There are perceptions of EPaCCS as ‘death registers’. This is a significant obstacle to their uptake. It will be a death sentence if the association becomes widespread. A further challenge is the difficulty of prognosis in end of life care. For three of the four EPaCCS considered in the 2013 Economic Evaluation, the percentage of patients who had not died within a year was close to or higher than 50% (64.3%, 47.3%, 59.0%, estimates based on data presented in Table 3 of the Economic Evaluation).6 While these high percentages will partly be due to incomplete recording of deaths and inefficient deactivation of records, difficulties of prognosis in end of life care are likely to be a significant factor. Importantly, a substantial proportion of health professionals continue to experience difficulties in discussing death and having advance care planning conversations and/or have limited skills in identifying and managing patients at the end of their life.47–50 Box 2 outlines a number of further challenges faced by EPaCCS projects. Box 2 Further challenges to EPaCCS projects Reporting and auditing difficulties: for EPaCCS which are not hosted by the team that developed them—complex processes and (perceptions of) IG constraints in having data released;

end of their life.47–50 Box 2 outlines a number of further challenges faced by EPaCCS projects. Box 2 Further challenges to EPaCCS projects Reporting and auditing difficulties: for EPaCCS which are not hosted by the team that developed them—complex processes and (perceptions of) IG constraints in having data released; for EPaCCS using a record sharing model—as there is no centralised register, a separate reporting solution is needed. Limited research and evaluation evidence. Challenges of effective education in a healthcare setting—with the limited availability of training time and difficulties of evaluating impact on outcomes. The skills set needed to lead and progress EPaCCS projects comes from too many and too different worlds; difficulties of communication between IT experts, end of life care specialists and managers. Broad context of initiative fatigue, and time and financial constraints. Key drivers Robust concept The benefits of more and up-to-date information about dying patients can be grasped immediately. The concept has a fundamental solidity about it. Striking outcomes for patients on EPaCCS Some of the outcomes for EPaCCS patients are impressive, as cited above.

Broad context of initiative fatigue, and time and financial constraints. Key drivers Robust concept The benefits of more and up-to-date information about dying patients can be grasped immediately. The concept has a fundamental solidity about it. Striking outcomes for patients on EPaCCS Some of the outcomes for EPaCCS patients are impressive, as cited above. National support and strong clinical leadership As locality registers, EPaCCS were proposed in the End of Life Care Strategy for England.7 The National End of Life Care Programme supported pilot work8 and led on their spread.5 Limited but reliable support has been available since the closure of the Programme from the NEoLCIN, NHS Improving Quality and national leads in end of life care. Recently, there has been a reinvigorated policy level interest in EPaCCS, as demonstrated by the commissioning of the national evaluation9 and still to be published Baseline Review. Comparing the achievements of individual projects suggests that strong clinical leadership is another decisive factor for their success. Clinician commitment to end of life care The recognition that we will all be facing end of life care issues, personally and in accompanying loved ones, and that we may have only ‘one chance to get it right’51 motivates many clinicians to provide exceptional service for their dying patients and those who care for them. Human kindness combined with high professionalism is the unshakeable foundation that sustains EPaCCS.

issues, personally and in accompanying loved ones, and that we may have only ‘one chance to get it right’51 motivates many clinicians to provide exceptional service for their dying patients and those who care for them. Human kindness combined with high professionalism is the unshakeable foundation that sustains EPaCCS. End of life care education Without attendant end of life care skills, limited uptake and superficial use of EPaCCS are likely. Established projects place a strong emphasis on education in end of life care: CMC has trained almost 13 000 and the C&P Project ∼600 users. Funding EPaCCS use has been boosted by incentives like CQUINs (Commissioning for Quality and Innovation), LES and DES (Local Enhanced Services and Directed Enhanced Services) and PDMA (Practice Delivery and Membership Agreement). Established EPaCCS projects have benefited from such payments. They do, however, prioritise other means of fostering engagement to avoid superficial involvement and unsustainable peaks of activity (box 3). Box 3 Further drivers used by EPaCCS projects Tapping into congruent projects, initiatives and infrastructure—for example, 2% DES (Directed Enhanced Service) for avoiding unplanned hospital admissions or existing end of life care educational initiatives. Representing projects by appealing to the core values of users—improved patient care rather than cost savings; a clinical and patient care project, that is IT-facilitated rather than an IT project. The use of sensitive language and broad user involvement in developing patient and public facing informational materials.

Funding EPaCCS use has been boosted by incentives like CQUINs (Commissioning for Quality and Innovation), LES and DES (Local Enhanced Services and Directed Enhanced Services) and PDMA (Practice Delivery and Membership Agreement). Established EPaCCS projects have benefited from such payments. They do, however, prioritise other means of fostering engagement to avoid superficial involvement and unsustainable peaks of activity (box 3). Box 3 Further drivers used by EPaCCS projects Tapping into congruent projects, initiatives and infrastructure—for example, 2% DES (Directed Enhanced Service) for avoiding unplanned hospital admissions or existing end of life care educational initiatives. Representing projects by appealing to the core values of users—improved patient care rather than cost savings; a clinical and patient care project, that is IT-facilitated rather than an IT project. The use of sensitive language and broad user involvement in developing patient and public facing informational materials. Expansion of the concept—for example, towards ‘Urgent Care Plans’ (CMC) for all patients with complex needs, which increases the scope of the service and reduces the ‘death register’ associations. Reliable feedback loops, including investigation of incidents. Quality improvement projects using evidence from EPaCCS as baseline and follow-up data. Extensive information provision, marketing and awareness raising activities.

Expansion of the concept—for example, towards ‘Urgent Care Plans’ (CMC) for all patients with complex needs, which increases the scope of the service and reduces the ‘death register’ associations. Reliable feedback loops, including investigation of incidents. Quality improvement projects using evidence from EPaCCS as baseline and follow-up data. Extensive information provision, marketing and awareness raising activities. Where next? Link to other work streams and reframe Until recently, EPaCCS were discussed almost exclusively within end of life care contexts. Yet they are only one of numerous initiatives which aim to improve coordination and communication across care settings prioritise patient choice require improved interoperability between information systems use electronic records generate Big Data and seek to shift more care into the community. EPaCCS need to be better integrated into work streams beyond end of life care. While this may redefine their nature and be experienced as disruptive on both sides, it will make EPaCCS more sustainable, cost-effective and capable of contributing 8 years of development and implementation experience. Many EPaCCS projects are expanding their scope—for example, towards urgent care plans for frail and elderly patients. This may become a general trend and increase uptake for end of life patients by bypassing anxieties and taboos. The benefits need to be balanced, however, against the danger of diluting the urgency and special needs around the end of life.

eir scope—for example, towards urgent care plans for frail and elderly patients. This may become a general trend and increase uptake for end of life patients by bypassing anxieties and taboos. The benefits need to be balanced, however, against the danger of diluting the urgency and special needs around the end of life. Provide an overview of current EPaCCS and lessons learnt Detailed descriptions and evaluations of EPaCCS and rigorous broader research are urgently needed. This will enable crucial knowledge exchange and stocktaking. It may also prompt rethinking of the number of EPaCCS needed across a geographical area. Priorities include bringing together outcomes data from currently operational projects; systematically comparing project features and relating them to markers of success, growth and sustainability (see online supplementary table S1, for an initial framework for comparing projects); benchmarking for adequate levels of use; rich descriptions of the contexts in which EPaCCS projects are developed and implemented; analysis of the types of patients on and off EPaCCS and the length of time they are on them; analysis of ‘failed’ projects and bottlenecks for successful projects. Such background work is needed to inform robust study designs, able to detect causal associations and underpin more reliable economic evaluations of EPaCCS.

analysis of the types of patients on and off EPaCCS and the length of time they are on them; analysis of ‘failed’ projects and bottlenecks for successful projects. Such background work is needed to inform robust study designs, able to detect causal associations and underpin more reliable economic evaluations of EPaCCS. Continue work on information standards An information standard for EPaCCS—SCCI 1580 (previously ISB 1580)—has been available since 2012. Continued work in this direction will facilitate further information sharing at the end of life. Such standards, however, should not be overly prescriptive and allow teams to respond to local needs and culture. In addition, the benefits of increased data capture should be continuously balanced against the (opportunity) costs of increased recording time by staff, in many cases busy clinical staff. Rethink national funding While national funding may seem unlikely in the current climate, decisions need to consider the large number of EPaCCS projects working in isolation, funded separately and to very different degrees. Funding for joint working and knowledge exchange may reduce spending by enabling uptake of existing EPaCCS and sharing hard won lessons. It is important to consider economies of scale. While a more robust comparison of costs is needed, the one made here suggests that the much larger CMC project has come out, proportionately, significantly less costly than the smaller C&P Project.

ending by enabling uptake of existing EPaCCS and sharing hard won lessons. It is important to consider economies of scale. While a more robust comparison of costs is needed, the one made here suggests that the much larger CMC project has come out, proportionately, significantly less costly than the smaller C&P Project. Rethink individual and community involvement Currently, patients can have their EPaCCS record printed out. Expectations and solutions for greater individual and community involvement can only grow. For instance, EPaCCS apps are already under development. As mentioned, there is a growing number of online consumer tools for recording care wishes and preferences. Community development approaches to advance care planning are being experimented with, where people who have been carers act as guides for those undergoing the experience of terminal illness.52 In these initiatives, health professionals help identify individuals suitable for such roles and ratify decisions made. The time will come when most individuals will be initiating their own record of end of life care wishes and preferences, informal carers will be updating care summaries and increased community involvement will be the only way to deal with limitations of service capacity. A radical shift in this direction may be needed now.

time will come when most individuals will be initiating their own record of end of life care wishes and preferences, informal carers will be updating care summaries and increased community involvement will be the only way to deal with limitations of service capacity. A radical shift in this direction may be needed now. Consider the applicability of England's EPaCCS to other national and global contexts The UK's National Health Service is a tax funded, free at the point of care, developed world healthcare service. The UK has also been ranked first, out of 80 countries, for the quality of its palliative care provision.53 The applicability of England's EPaCCS experience to other national and local contexts is therefore a fascinating and challenging question. Rigorous evidence and conceptual and theoretical work on the mechanisms of effecting change is needed before judgements of potential transferability of approach can be made. Numerous factors will need to be articulated and linked in developing a conceptual model of the effectiveness of EPaCCS in the context of England's healthcare system, IT landscape, palliative and end of life care services and skills, professional and public attitudes, etc. In turn, key drivers and challenges will need to be represented at a level of generality that allows comparison with other countries, even if, on the surface, variables differ widely. Paradoxically, England may not be the best place for its own electronic innovation in palliative and end of life care. Countries with a greater maturity of IT systems and HIE tools, more consistently distributed staff IT skills, lighter regulatory frameworks of IG and more centralised decision-making in a local health economy may be better placed to institute electronic coordination systems in palliative and end of life care.

ife care. Countries with a greater maturity of IT systems and HIE tools, more consistently distributed staff IT skills, lighter regulatory frameworks of IG and more centralised decision-making in a local health economy may be better placed to institute electronic coordination systems in palliative and end of life care. Conclusions According to an already dated estimate, there have been 48 reports on palliative and end of life care in the UK since the national End of Life Care Strategy of 2008, demonstrating striking levels of concordance of analysis, agreement on main issues and lack of subsequent progress.54 With regard to patient data sharing, a YouGov survey found that 30% of respondents were ‘shocked’, 40% ‘annoyed’ and 61% ‘worried’ that their GP records were not available to accident and emergency departments (8% were ‘not bothered’ and 4% thought that was how it should be).55 There is energy and expectation for change in end of life care and data sharing in the UK. Learning from the successes and failures of EPaCCS projects cannot but help connect better the dots, data and people, leading to better care for those approaching the end of their lives.

d 4% thought that was how it should be).55 There is energy and expectation for change in end of life care and data sharing in the UK. Learning from the successes and failures of EPaCCS projects cannot but help connect better the dots, data and people, leading to better care for those approaching the end of their lives. The authors thank Anne Dowling and Mandy Shaw from the ‘Coordinate My Care’ team for providing performance data and background information; Ian Merrick, Project Manager of the Cambridgeshire & Peterborough Project for Data Sharing in End of Life Care and Joan Field (who sadly passed away recently), Karen Arnold, Margaret Johnson, Nick Roberts, Pauline Dorward and Roberta Lovick, members of the ‘Prepared to Share?’ study User Group, for feedback on the text; Sean Leathen, Dr Karen Chumbley, Stephen Crawley, Louise Price and Dr Rishabh Prasad for sharing data and updates from the EPaCCS they are working on; Angela Harper for her support for the Cambridgeshire & Peterborough team and the CCG and Freedom of Information teams in the South West for providing information on services in their areas. They also thank the journal reviewers for their extensive and helpful comments. Contributors: MP conceived of the study, summarised the data and drafted the manuscript. JR, JA and SB contributed data, discussion points and parts of the text, provided feedback on and edited the manuscript. MP is the guarantor.

The authors thank Anne Dowling and Mandy Shaw from the ‘Coordinate My Care’ team for providing performance data and background information; Ian Merrick, Project Manager of the Cambridgeshire & Peterborough Project for Data Sharing in End of Life Care and Joan Field (who sadly passed away recently), Karen Arnold, Margaret Johnson, Nick Roberts, Pauline Dorward and Roberta Lovick, members of the ‘Prepared to Share?’ study User Group, for feedback on the text; Sean Leathen, Dr Karen Chumbley, Stephen Crawley, Louise Price and Dr Rishabh Prasad for sharing data and updates from the EPaCCS they are working on; Angela Harper for her support for the Cambridgeshire & Peterborough team and the CCG and Freedom of Information teams in the South West for providing information on services in their areas. They also thank the journal reviewers for their extensive and helpful comments. Contributors: MP conceived of the study, summarised the data and drafted the manuscript. JR, JA and SB contributed data, discussion points and parts of the text, provided feedback on and edited the manuscript. MP is the guarantor. Funding: This paper presents independent research funded by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research & Care (CLAHRC) East of England, at Cambridgeshire and Peterborough NHS Foundation Trust, as well as the Health Innovation and Education Cluster (HIEC) hosted by Cambridge University Health Partners (CUHP).

nded by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research & Care (CLAHRC) East of England, at Cambridgeshire and Peterborough NHS Foundation Trust, as well as the Health Innovation and Education Cluster (HIEC) hosted by Cambridge University Health Partners (CUHP). Disclaimer: The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. Competing interests: All authors have completed the ICMJE uniform disclosure form at http://www.icmje.org/coi_disclosure.pdf and declare that all are involved in Electronic Palliative Care Coordination Systems projects and have their posts partly funded for work on such projects. MP is a research associate partly funded by the Cambridgeshire & Peterborough Project for Data Sharing in End of Life Care, JR is the Clinical Lead for ‘Coordinate My Care’, JA is the Lead for the South West EPaCCS and SB is the Clinical Lead for the Cambridgeshire & Peterborough Project for Data Sharing in End of Life Care. JA also received royalties from Adastra, which hosts the South West EPaCCS. Provenance and peer review: Not commissioned; externally peer reviewed.

Introduction Most people prefer to die in their usual place of residence (own home or care home) if provided with the right support.1 2 In England, the proportion of deaths in usual place of residence is a key quality indicator for end of life care,3 and has increased from 38% in 2008 to 45% in 2015.4 Advance care planning is a voluntary process of discussion and review concerning future care and treatment. It is considered a means for helping people die in their preferred place and is emphasised in the end of life care strategy for England.5 In England, advance statements set out general preferences to inform best-interest decisions, while advance decisions to refuse treatment are legally binding and regulated by the Mental Capacity Act (2005). In the USA, the Patient Self-Determination Act (PSDA, 1990) requires Medicare-registered and Medicaid-registered healthcare agencies to provide information about advance directives (advance decisions) and incorporate them into medical records. We undertook secondary analyses of data from the National Survey of Bereaved People, 2013, to explore whether, in England, advance care planning, indicated by an expressed preference for place of death that is recorded by healthcare staff, is associated with where people die.6 We also explored a range of secondary outcomes; sufficient support to care for a dying person at home, overall quality of care and effectiveness of pain management.

advance care planning, indicated by an expressed preference for place of death that is recorded by healthcare staff, is associated with where people die.6 We also explored a range of secondary outcomes; sufficient support to care for a dying person at home, overall quality of care and effectiveness of pain management. Methods Study population and data sources The National Survey of Bereaved People, first conducted in 2011, is commissioned by National Health Service (NHS) England and administered by the Office for National Statistics (ONS). For the 2013 release, a stratified random probability sample of 49 607 people was selected from 150 111 people registering a death between 1 January 2013 and 30 April 2013.6 Stratification was according to cause of death, place of death and geographic spread (NHS Area Team). Respondents, usually family members, were surveyed between 4 and 11 months after the registered death. Responses were received from 22 661 individuals (response rate, 45.7%). Non-response was associated with the decedent being male, younger, dying at home or in hospital (rather than care home or hospice), being from a black, Asian or minority ethnic (BAME) group and living in an area of greater deprivation. Sampling weights were used to account for bias in the probability of selection and non-response. Further information on survey methodology is available from the ONS.7

r in hospital (rather than care home or hospice), being from a black, Asian or minority ethnic (BAME) group and living in an area of greater deprivation. Sampling weights were used to account for bias in the probability of selection and non-response. Further information on survey methodology is available from the ONS.7 The advance care planning measure Previous research has used a range of different advance care planning indicators, including completion of advance directives, completion of treatment-limiting advance directives, the occurrence of end of life care discussions with family members and/or healthcare professionals, and, in a recent UK study,8 having a preference for place of death recorded in medical records. The current study uses the indicator of an expressed preference for place of death that is recorded by healthcare staff. Survey respondents were asked (Q41) Did s/he ever say where s/he would like to die? Those answering ‘yes’ (rather than ‘no’ or ‘not sure’) were then asked (Q42) Where did s/he say that s/he would like to die? and provided with a list of possible locations. They were then asked (Q43) Did the healthcare staff have a record of this? Those considered to have engaged in advance care planning were those answering ‘yes’ to Q43. Everyone else was considered, for the purposes of this research, not to have engaged in advance care planning. Primary outcome measures

The advance care planning measure Previous research has used a range of different advance care planning indicators, including completion of advance directives, completion of treatment-limiting advance directives, the occurrence of end of life care discussions with family members and/or healthcare professionals, and, in a recent UK study,8 having a preference for place of death recorded in medical records. The current study uses the indicator of an expressed preference for place of death that is recorded by healthcare staff. Survey respondents were asked (Q41) Did s/he ever say where s/he would like to die? Those answering ‘yes’ (rather than ‘no’ or ‘not sure’) were then asked (Q42) Where did s/he say that s/he would like to die? and provided with a list of possible locations. They were then asked (Q43) Did the healthcare staff have a record of this? Those considered to have engaged in advance care planning were those answering ‘yes’ to Q43. Everyone else was considered, for the purposes of this research, not to have engaged in advance care planning. Primary outcome measures Place of death: Place of death information was taken from death certificates, with the response options used in our analyses being hospital, home and care home. From these, we constructed two parallel measures for place of death; dying at home (rather than hospital) and dying in a care home (rather than hospital). Secondary outcome measures

Place of death: Place of death information was taken from death certificates, with the response options used in our analyses being hospital, home and care home. From these, we constructed two parallel measures for place of death; dying at home (rather than hospital) and dying in a care home (rather than hospital). Secondary outcome measures Sufficient support to care for the dying person at home: The survey asked (Q2) Did s/he [the decedent] spend any time at home during the last three months of life? Where the answer was yes, respondents were asked (Q5) Overall, do you feel that you and your family got as much help and support from health and social services as you needed when caring for him/her? Response categories were: Yes, we got as much help and support as we wanted; Yes, we got some support but not as much as we wanted; No, although we tried to get more support; No, but we did not ask for more help; We did not need help. A binary measure was constructed; ‘yes, we got as much help and support as we wanted’ compared with all other responses (except for ‘we did not need help’; respondents giving this answer were excluded from analysis). Overall quality of care: The survey asked (Q51) Overall, and taking all services into account, how would you rate his/her [the decedent's] care in the last three months of life? The response categories for this question were: Outstanding, Excellent, Good, Fair, Poor, Don't know. The binary measure used in analyses was overall care that was ‘outstanding’ or ‘excellent’ compared with all other responses (except for ‘don't know’, which were excluded from analysis).

Overall quality of care: The survey asked (Q51) Overall, and taking all services into account, how would you rate his/her [the decedent's] care in the last three months of life? The response categories for this question were: Outstanding, Excellent, Good, Fair, Poor, Don't know. The binary measure used in analyses was overall care that was ‘outstanding’ or ‘excellent’ compared with all other responses (except for ‘don't know’, which were excluded from analysis). Effectiveness of pain management: The survey asked (Q6) During the last three months of his/her [the decedent's] life, while s/he was at home, how well was his/her pain relieved? For those who spent time in a care home, it asked (Q22) During the last three months of his/her life, while s/he was in the care home, how well was his/her pain relieved? Finally, for those who spent time in hospital, it asked (Q26) During his/her last hospital admission, how well was his/her pain relieved? The response categories for each of these were: Does not apply—s/he did not have any pain; Completely, all of the time; Completely, some of the time; Partially; Not at all; Don't know. The binary outcome measure in each case was pain relieved ‘completely, all of the time’ compared with all other responses (except ‘does not apply—s/he did not have any pain’ and ‘don't know’, which were excluded from analysis).

Effectiveness of pain management: The survey asked (Q6) During the last three months of his/her [the decedent's] life, while s/he was at home, how well was his/her pain relieved? For those who spent time in a care home, it asked (Q22) During the last three months of his/her life, while s/he was in the care home, how well was his/her pain relieved? Finally, for those who spent time in hospital, it asked (Q26) During his/her last hospital admission, how well was his/her pain relieved? The response categories for each of these were: Does not apply—s/he did not have any pain; Completely, all of the time; Completely, some of the time; Partially; Not at all; Don't know. The binary outcome measure in each case was pain relieved ‘completely, all of the time’ compared with all other responses (except ‘does not apply—s/he did not have any pain’ and ‘don't know’, which were excluded from analysis). Statistical analysis Multivariate logistic regression analysis was conducted for each outcome in turn. Advance care planning was the outcome measure for one model, designed to identify the decedent characteristics independently associated with having a preferred place of death recorded by healthcare staff. The remaining logistic regression models included advance care planning as an independent variable. Other covariates were: Age (18–64 years/65–79 years/80+ years); Sex (male/female); Cause of death (haematological cancer/non-haematological cancer/respiratory illness/neurological conditions including dementia/heart and circulatory/renal failure/other conditions). These data were taken from death certificate information;

Statistical analysis Multivariate logistic regression analysis was conducted for each outcome in turn. Advance care planning was the outcome measure for one model, designed to identify the decedent characteristics independently associated with having a preferred place of death recorded by healthcare staff. The remaining logistic regression models included advance care planning as an independent variable. Other covariates were: Age (18–64 years/65–79 years/80+ years); Sex (male/female); Cause of death (haematological cancer/non-haematological cancer/respiratory illness/neurological conditions including dementia/heart and circulatory/renal failure/other conditions). These data were taken from death certificate information; Area deprivation (Index of Multiple Deprivation (IMD) quintiles); Surviving spouse/partner, using proxy of whether survey respondent identifies themselves as the spouse or partner of the decedent (yes/ no); Ethnic background (white/BAME). Each regression model included only those cases with complete data on the outcome and independent variables. The sample size for each model is provided in tables 2 –7. All statistical analyses were conducted using the STATA statistical package—V.12.1 (STATA, 2011). A significance level of 0.01 was selected because of the large sample size (increasing the potential for type 1 errors) and large number of tests conducted. The fit of each model was assessed using the proportion of cases for which the model predicted the same outcome as observed in the data.

kage—V.12.1 (STATA, 2011). A significance level of 0.01 was selected because of the large sample size (increasing the potential for type 1 errors) and large number of tests conducted. The fit of each model was assessed using the proportion of cases for which the model predicted the same outcome as observed in the data. Results The characteristics of the complete sample of decedents (n=22 661) and survey respondents are summarised in table 1. Women comprise the majority of decedents and survey respondents. People from BAME groups are under-represented in both groups. Most respondents (59.8%) are sons or daughters of the decedent, while a quarter (24.5%) are spouses or partners. Table 1 Demographic data for decedents Characteristic Number* Percentage Sex of decedent Female 12 701 56.0 Male 9960 44.0 Age at death of decedent 18–59 1146 5.1 60–69 2208 9.7 70–79 4363 19.3 80–89 8726 38.5 90 and over 6218 27.4 Ethnic background of decedent White 20 719 97.6 Mixed 37 0.2 Asian/Asian British 281 1.3 Black/African/Caribbean/black British 144 0.7 Other 38 0.2 Place of death Home 4523 20.0 Hospital 10 851 47.9 Care home 6013 26.5 Hospice 1274 5.6 Area deprivation of place of residence (IMD 2010) 1 (most deprived) 3488 15.4 2 4310 19.0 3 4947 21.8 4 5038 22.2 5 (least deprived) 4878 21.5 Source: ONS 2014. *Demographic data for decedents is complete (n=22 661), with missing data identified from death certificates. IMD, Index of Multiple Deprivation; ONS, Office for National Statistics.

Characteristic Number* Percentage Sex of decedent Female 12 701 56.0 Male 9960 44.0 Age at death of decedent 18–59 1146 5.1 60–69 2208 9.7 70–79 4363 19.3 80–89 8726 38.5 90 and over 6218 27.4 Ethnic background of decedent White 20 719 97.6 Mixed 37 0.2 Asian/Asian British 281 1.3 Black/African/Caribbean/black British 144 0.7 Other 38 0.2 Place of death Home 4523 20.0 Hospital 10 851 47.9 Care home 6013 26.5 Hospice 1274 5.6 Area deprivation of place of residence (IMD 2010) 1 (most deprived) 3488 15.4 2 4310 19.0 3 4947 21.8 4 5038 22.2 5 (least deprived) 4878 21.5 Source: ONS 2014. *Demographic data for decedents is complete (n=22 661), with missing data identified from death certificates. IMD, Index of Multiple Deprivation; ONS, Office for National Statistics. Characteristics of people with a recorded preference for place of death People with non-cancer diagnoses had significantly lower odds than those with (non-haematological) cancer of having a recorded preference for place of death. This includes those with cardiovascular disease, OR 0.16 (99% CI 0.14 to 0.19); neurological conditions including dementia, OR 0.20 (99% CI 0.17 to 0.24); respiratory illness, OR 0.29 (99% CI 0.24 to 0.35); and ‘other’ conditions, OR 0.21 (99% CI 0.18 to 0.25). Compared with those living in the least deprived IMD quintile, people in other IMD quintiles had lower odds of having a recorded preference for place of death. This includes people in the most deprived IMD quintile (OR 0.73; 99% CI 0.62 to 0.87), the second most deprived IMD quintile (OR 0.80; 99% CI 0.68 to 0.94) and the second least deprived IMD quintile (OR 0.84; 99% CI 0.72 to 0.98). The OR for those in the middle IMD quintile was similar (OR 0.87); however, this did not quite achieve statistical significance (99% CI 0.74 to 1.01). People without a spouse or partner also had lower odds of having a recorded preference for place of death (OR 0.78; 99% CI 0.69 to 0.89), as did those from a BAME group, compared with people of white ethnicity (OR 0.64; 99% CI 0.44 to 0.94). Full results are given in table 2.

istical significance (99% CI 0.74 to 1.01). People without a spouse or partner also had lower odds of having a recorded preference for place of death (OR 0.78; 99% CI 0.69 to 0.89), as did those from a BAME group, compared with people of white ethnicity (OR 0.64; 99% CI 0.44 to 0.94). Full results are given in table 2. Table 2 Logistic regression of factors associated with advance care planning†; N=21 032 Effect OR 99% CI Age at death 18–64 0.94 0.79 to 1.12 65–79 1.07 0.95 to 1.21 80 or above 1.00 Sex Female 1.04 0.94 to 1.16 Male 1.00 Cause of death Cardiovascular disease 0.16* 0.14 to 0.19 Haematological cancer 0.82 0.63 to 1.07 Neurological condition 0.20* 0.17 to 0.24 Respiratory illness 0.29* 0.24 to 0.35 ‘Other’ causes‡ 0.21* 0.18 to 0.25 Non-haematological cancer 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.73* 0.62 to 0.87 Second most deprived 0.80* 0.68 to 0.94 Third most deprived 0.87 0.74 to 1.01 Fourth most deprived 0.84* 0.72 to 0.98 Least deprived 1.00 Relationship of respondent Child/friend/other 0.78* 0.69 to 0.89 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.64* 0.44 to 0.94 White 1.00 Per cent correctly classified (predicted vs observed) 69.7% *p Value <0.01. †Indicated by having expressed a preference for place of death that was recorded by healthcare staff. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory. IMD, Index of Multiple Deprivation.

Effect OR 99% CI Age at death 18–64 0.94 0.79 to 1.12 65–79 1.07 0.95 to 1.21 80 or above 1.00 Sex Female 1.04 0.94 to 1.16 Male 1.00 Cause of death Cardiovascular disease 0.16* 0.14 to 0.19 Haematological cancer 0.82 0.63 to 1.07 Neurological condition 0.20* 0.17 to 0.24 Respiratory illness 0.29* 0.24 to 0.35 ‘Other’ causes‡ 0.21* 0.18 to 0.25 Non-haematological cancer 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.73* 0.62 to 0.87 Second most deprived 0.80* 0.68 to 0.94 Third most deprived 0.87 0.74 to 1.01 Fourth most deprived 0.84* 0.72 to 0.98 Least deprived 1.00 Relationship of respondent Child/friend/other 0.78* 0.69 to 0.89 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.64* 0.44 to 0.94 White 1.00 Per cent correctly classified (predicted vs observed) 69.7% *p Value <0.01. †Indicated by having expressed a preference for place of death that was recorded by healthcare staff. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory. IMD, Index of Multiple Deprivation. Primary outcome measures: place of death Dying at home (rather than hospital) Of our total sample, 20% (n=4523) died at home and 48% (n=10 851) died in hospital. Of these, 15% had a preferred place of death recorded by healthcare staff. Those with a recorded preference had significantly greater odds of dying at home rather than in hospital (OR 6.25; 99% CI 5.56 to 7.14). This factor had the largest effect. Other significant factors were being younger than 80, being male, having (non-haematological) cancer, living in the least deprived areas and having a spouse or partner. Full results are given in table 3.

er odds of dying at home rather than in hospital (OR 6.25; 99% CI 5.56 to 7.14). This factor had the largest effect. Other significant factors were being younger than 80, being male, having (non-haematological) cancer, living in the least deprived areas and having a spouse or partner. Full results are given in table 3. Table 3 Logistic regression of factors associated with death at home†; N=14 242 Effect OR 99% CI Age at death 18–64 1.46* 1.20 to 1.78 65–79 1.35* 1.19 to 1.54 80 or above 1.00 Sex Female 0.89* 0.80 to 0.99 Male 1.00 Cause of death Cardiovascular disease 0.72* 0.66 to 0.78 Haematological cancer 0.45* 0.33 to 0.63 Neurological condition 0.49* 0.40 to 0.59 Respiratory illness 0.38* 0.33 to 0.45 ‘Other’ causes‡ 0.32* 0.28 to 0.37 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.69* 0.58 to 0.82 Second most deprived 0.83* 0.70 to 0.98 Third most deprived 0.88 0.75 to 1.03 Fourth most deprived 0.88 0.75 to 1.02 Least deprived 1.00 Relationship of respondent Child/friend/other 0.87* 0.77 to 1.00 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.91 0.65 to 1.27 White 1.00 Advance care plan§ Yes 6.25* 5.56 to 7.14 No 1.00 Per cent correctly classified (predicted vs observed) 73.6% *p Value <0.01. †Dependent variable ‘place of death’ coded as 0—death in a hospital and 1—death at home. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff.

Effect OR 99% CI Age at death 18–64 1.46* 1.20 to 1.78 65–79 1.35* 1.19 to 1.54 80 or above 1.00 Sex Female 0.89* 0.80 to 0.99 Male 1.00 Cause of death Cardiovascular disease 0.72* 0.66 to 0.78 Haematological cancer 0.45* 0.33 to 0.63 Neurological condition 0.49* 0.40 to 0.59 Respiratory illness 0.38* 0.33 to 0.45 ‘Other’ causes‡ 0.32* 0.28 to 0.37 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.69* 0.58 to 0.82 Second most deprived 0.83* 0.70 to 0.98 Third most deprived 0.88 0.75 to 1.03 Fourth most deprived 0.88 0.75 to 1.02 Least deprived 1.00 Relationship of respondent Child/friend/other 0.87* 0.77 to 1.00 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.91 0.65 to 1.27 White 1.00 Advance care plan§ Yes 6.25* 5.56 to 7.14 No 1.00 Per cent correctly classified (predicted vs observed) 73.6% *p Value <0.01. †Dependent variable ‘place of death’ coded as 0—death in a hospital and 1—death at home. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff. IMD, Index of Multiple Deprivation.

†Dependent variable ‘place of death’ coded as 0—death in a hospital and 1—death at home. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff. IMD, Index of Multiple Deprivation. Dying in a care home (rather than hospital) Thirty-two per cent of people in our sample (n=7287) died in a care home. Of people who died in a care home or in hospital, 10% had a preferred place of death recorded by healthcare staff. These individuals had significantly greater odds of dying in a care home rather than hospital (OR 2.70; 99% CI 2.33 to 3.13). Other significant factors were being aged 80 or over, being male, having a neurological condition (including dementia), living in the least deprived areas, being of white ethnicity and not having a spouse or partner. Full results are presented in table 4. Table 4 Logistic regression of factors associated with death in a care home†; N=16 845

Dying in a care home (rather than hospital) Thirty-two per cent of people in our sample (n=7287) died in a care home. Of people who died in a care home or in hospital, 10% had a preferred place of death recorded by healthcare staff. These individuals had significantly greater odds of dying in a care home rather than hospital (OR 2.70; 99% CI 2.33 to 3.13). Other significant factors were being aged 80 or over, being male, having a neurological condition (including dementia), living in the least deprived areas, being of white ethnicity and not having a spouse or partner. Full results are presented in table 4. Table 4 Logistic regression of factors associated with death in a care home†; N=16 845 Effect OR 99% CI Age at death Ages 18–64 0.46* 0.38 to 0.55 Ages 65–79 0.51* 0.45 to 0.57 80 or above 1.00 Sex Female 0.66* 0.60 to 0.72 Male 1.00 Cause of death Cardiovascular disease 0.31* 0.29 to 0.33 Haematological cancer 0.47* 0.34 to 0.65 Neurological condition 1.76* 1.56 to 1.97 Respiratory illness 0.34* 0.30 to 0.39 ‘Other’ causes‡ 0.30* 0.27 to 0.33 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.61* 0.53 to 0.71 Second most deprived 0.79* 0.69 to 0.91 Third most deprived 0.95 0.83 to 1.09 Fourth most deprived 0.98 0.86 to 1.12 Least deprived 1.00 Relationship of respondent Child/friend/other 1.43* 1.26 to 1.62 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.59* 0.42 to 0.83 White 1.00 Advance care plan§ Yes 2.70* 2.33 to 3.13 No 1.00 Per cent correctly classified (predicted vs observed) 72.3% *p Value <0.01. †Dependent variable ‘place of death’ coded as 0—death in a hospital and 1—death in a care home.

Summary of included studies Study design Fourteen of 26 articles comprised ‘expert opinion’ (editorials, commentaries and discussion papers).5 15–27 Of the remaining 12 articles, nine were quantitative studies,6 7 28–34 two were qualitative35 36 and one was a mixed-method study though predominantly qualitative.37 All nine quantitative studies had an observational design: one was a service evaluation,33 two were descriptive cross-sectional studies,31 32 four were retrospective cohort studies without control groups6 7 29 30 and two were retrospective cohort studies with control groups28 34 (figure 2). All studies using qualitative methods used semistructured interviews,35–37 and one used analysis of documentation and service observations in addition to interviews.36 Table 1 reports characteristics of included articles. Figure 2 Type of studies included according to the hierarchy of evidence based on Murad et al.12 Of four reports, one was a cross-sectional survey,38 one used a qualitative approach through interviews and documentation analysis,39 and two used mixed-methods approaches combining both quantitative and qualitative data analysis.40 41 Table 2 reports characteristics of included reports.

1.62 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.59* 0.42 to 0.83 White 1.00 Advance care plan§ Yes 2.70* 2.33 to 3.13 No 1.00 Per cent correctly classified (predicted vs observed) 72.3% *p Value <0.01. †Dependent variable ‘place of death’ coded as 0—death in a hospital and 1—death in a care home. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff. IMD, Index of Multiple Deprivation. Secondary outcomes Having sufficient support to care for a dying person at home Fifty-five per cent of respondents (n=12 348) said they needed help to look after the decedent at home. Of these, half (n=6127) felt they had all the help and support from health and social services they wanted and 22% (n=2718) cared for someone with a preference for place of death recorded by healthcare staff. Those who cared for someone with a recorded preference had significantly greater odds of receiving sufficient support (OR 2.02; 99% CI 1.77 to 2.29). These, along with being a spouse or partner, were the most important factors. Other significant factors were the decedent being aged 80 or over, female and having other factors with significantly (non-haematological) cancer or a neurological condition (including dementia), as well as living in the least deprived areas. Full results are given in table 5.

partner, were the most important factors. Other significant factors were the decedent being aged 80 or over, female and having other factors with significantly (non-haematological) cancer or a neurological condition (including dementia), as well as living in the least deprived areas. Full results are given in table 5. Table 5 Logistic regression of factors associated with sufficient help and support from health and social services to care for the decedent at home; N=11 458 Effect OR 99% CI Age at death 18–64 0.66* 0.55 to 0.80 65–79 0.79* 0.70 to 0.90 80 or above 1.00 Sex Female 1.32* 1.19 to 1.46 Male 1.00 Cause of death Cardiovascular disease 0.82* 0.71 to 0.95 Haematological cancer 0.72* 0.53 to 0.97 Neurological condition 0.83 0.68 to 1.02 Respiratory illness 0.84* 0.70 to 1.00 ‘Other’ causes‡ 0.78* 0.67 to 0.91 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.77* 0.65 to 0.91 Second most deprived 0.80* 0.68 to 0.93 Third most deprived 0.92 0.79 to 1.07 Fourth most deprived 0.93 0.80 to 1.08 Least deprived 1.00 Relationship of respondent Child/friend/other 0.49* 0.43 to 0.55 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.83 0.61 to 1.13 White 1.00 Advance care plan§ Yes 2.02* 1.77 to 2.29 No 1.00 Per cent correctly classified (predicted vs observed) 62.3% *p Value <0.01. †Dependent variable: ‘insufficient support’ coded as 0; ‘sufficient support’ coded as 1. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness.

Effect OR 99% CI Age at death 18–64 0.66* 0.55 to 0.80 65–79 0.79* 0.70 to 0.90 80 or above 1.00 Sex Female 1.32* 1.19 to 1.46 Male 1.00 Cause of death Cardiovascular disease 0.82* 0.71 to 0.95 Haematological cancer 0.72* 0.53 to 0.97 Neurological condition 0.83 0.68 to 1.02 Respiratory illness 0.84* 0.70 to 1.00 ‘Other’ causes‡ 0.78* 0.67 to 0.91 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.77* 0.65 to 0.91 Second most deprived 0.80* 0.68 to 0.93 Third most deprived 0.92 0.79 to 1.07 Fourth most deprived 0.93 0.80 to 1.08 Least deprived 1.00 Relationship of respondent Child/friend/other 0.49* 0.43 to 0.55 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.83 0.61 to 1.13 White 1.00 Advance care plan§ Yes 2.02* 1.77 to 2.29 No 1.00 Per cent correctly classified (predicted vs observed) 62.3% *p Value <0.01. †Dependent variable: ‘insufficient support’ coded as 0; ‘sufficient support’ coded as 1. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff. IMD, Index of Multiple Deprivation.

Effect OR 99% CI Age at death 18–64 0.66* 0.55 to 0.80 65–79 0.79* 0.70 to 0.90 80 or above 1.00 Sex Female 1.32* 1.19 to 1.46 Male 1.00 Cause of death Cardiovascular disease 0.82* 0.71 to 0.95 Haematological cancer 0.72* 0.53 to 0.97 Neurological condition 0.83 0.68 to 1.02 Respiratory illness 0.84* 0.70 to 1.00 ‘Other’ causes‡ 0.78* 0.67 to 0.91 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.77* 0.65 to 0.91 Second most deprived 0.80* 0.68 to 0.93 Third most deprived 0.92 0.79 to 1.07 Fourth most deprived 0.93 0.80 to 1.08 Least deprived 1.00 Relationship of respondent Child/friend/other 0.49* 0.43 to 0.55 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.83 0.61 to 1.13 White 1.00 Advance care plan§ Yes 2.02* 1.77 to 2.29 No 1.00 Per cent correctly classified (predicted vs observed) 62.3% *p Value <0.01. †Dependent variable: ‘insufficient support’ coded as 0; ‘sufficient support’ coded as 1. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff. IMD, Index of Multiple Deprivation. Overall quality of care Forty per cent of respondents (n=9268) thought that overall quality of care in the last 3 months of life was ‘outstanding’ or ‘excellent’, while 53% (n=12 065) rated it ‘good’, ‘fair’ or ‘poor’ and 7% gave no opinion. In cases where an opinion was given, 16% had a preference for place of death recorded by healthcare staff, with these decedents having significantly greater odds of receiving care considered to be ‘outstanding’ or ‘excellent’ (OR 2.27; 99% CI 2.04 to 2.53). This factor had the largest effect. Other significant factors were being over 80 (compared with age 65–79), being female, having a cancer diagnosis (when compared with having a cardiovascular, respiratory or ‘other’ condition, but not having a neurological condition such as dementia) and living in the least deprived areas. Table 6 provides full results.

Other significant factors were being over 80 (compared with age 65–79), being female, having a cancer diagnosis (when compared with having a cardiovascular, respiratory or ‘other’ condition, but not having a neurological condition such as dementia) and living in the least deprived areas. Table 6 provides full results. Table 6 Logistic regression of factors associated with ‘excellent’ or ‘outstanding’ overall quality of care†; N=19 850 Effect OR 99% CI Age at death 18–64 0.88 0.75 to 1.02 65–79 0.85* 0.77 to 0.95 80 or above 1.00 Sex Female 1.21* 1.12 to 1.32 Male 1.00 Cause of death Cardiovascular disease 0.75* 0.67 to 0.84 Haematological cancer 1.00 0.77 to 1.29 Neurological condition 1.10 0.97 to 1.26 Respiratory illness 0.75* 0.65 to 0.87 ‘Other’ causes‡ 0.72* 0.63 to 0.81 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.83* 0.73 to 0.95 Second most deprived 0.90 0.80 to 1.02 Third most deprived 0.98 0.87 to 1.10 Fourth most deprived 1.00 0.89 to 1.12 Least deprived 1.00 Relationship of respondent Child/friend/other 0.66* 0.60 to 0.73 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.77 0.59 to 1.01 White 1.00 Advance care plan§ Yes 2.27* 2.04 to 2.53 No 1.00 Per cent correctly classified (predicted vs observed) 60.7% *p Value <0.01. †Dependent variable: overall care was outstanding or excellent; ‘no’ coded as 0; ‘yes’ coded as 1. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness.

Effect OR 99% CI Age at death 18–64 0.88 0.75 to 1.02 65–79 0.85* 0.77 to 0.95 80 or above 1.00 Sex Female 1.21* 1.12 to 1.32 Male 1.00 Cause of death Cardiovascular disease 0.75* 0.67 to 0.84 Haematological cancer 1.00 0.77 to 1.29 Neurological condition 1.10 0.97 to 1.26 Respiratory illness 0.75* 0.65 to 0.87 ‘Other’ causes‡ 0.72* 0.63 to 0.81 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.83* 0.73 to 0.95 Second most deprived 0.90 0.80 to 1.02 Third most deprived 0.98 0.87 to 1.10 Fourth most deprived 1.00 0.89 to 1.12 Least deprived 1.00 Relationship of respondent Child/friend/other 0.66* 0.60 to 0.73 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.77 0.59 to 1.01 White 1.00 Advance care plan§ Yes 2.27* 2.04 to 2.53 No 1.00 Per cent correctly classified (predicted vs observed) 60.7% *p Value <0.01. †Dependent variable: overall care was outstanding or excellent; ‘no’ coded as 0; ‘yes’ coded as 1. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff. IMD, Index of Multiple Deprivation.

Effect OR 99% CI Age at death 18–64 0.88 0.75 to 1.02 65–79 0.85* 0.77 to 0.95 80 or above 1.00 Sex Female 1.21* 1.12 to 1.32 Male 1.00 Cause of death Cardiovascular disease 0.75* 0.67 to 0.84 Haematological cancer 1.00 0.77 to 1.29 Neurological condition 1.10 0.97 to 1.26 Respiratory illness 0.75* 0.65 to 0.87 ‘Other’ causes‡ 0.72* 0.63 to 0.81 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 0.83* 0.73 to 0.95 Second most deprived 0.90 0.80 to 1.02 Third most deprived 0.98 0.87 to 1.10 Fourth most deprived 1.00 0.89 to 1.12 Least deprived 1.00 Relationship of respondent Child/friend/other 0.66* 0.60 to 0.73 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.77 0.59 to 1.01 White 1.00 Advance care plan§ Yes 2.27* 2.04 to 2.53 No 1.00 Per cent correctly classified (predicted vs observed) 60.7% *p Value <0.01. †Dependent variable: overall care was outstanding or excellent; ‘no’ coded as 0; ‘yes’ coded as 1. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff. IMD, Index of Multiple Deprivation. Pain management Forty-three per cent of the total sample (n=9716) had pain that required management while being cared for at home. Eighteen per cent (n=1731) of these had their pain relieved ‘completely, all the time’ while 82% (n=7985) continued to experience pain to some degree. The 24% that had a preferred place of death recorded by healthcare staff had significantly greater odds of having their pain relieved ‘completely, all the time’ (OR 2.32; 99% CI 1.97 to 2.73). This analysis controlled for cause of death, so this effect was independent of the symptom burden associated with different conditions. This was the factor with the largest effect. Other significant factors were being aged 80 or over (compared with 18–64), having a cancer diagnosis (compared with cardiovascular, respiratory and ‘other’ conditions, but not neurological conditions including dementia) and having a spouse or partner.

different conditions. This was the factor with the largest effect. Other significant factors were being aged 80 or over (compared with 18–64), having a cancer diagnosis (compared with cardiovascular, respiratory and ‘other’ conditions, but not neurological conditions including dementia) and having a spouse or partner. Twenty-three per cent of the total sample (n=5374) had pain that required management while being cared for in a care home. Forty-six per cent of these (n=2480) reported that pain was relieved ‘completely, all the time’, while the remainder continued to experience pain to some degree. Having a preferred place of death recorded by healthcare staff was not significantly associated with this outcome. Only being aged 80 or over (compared with age 65–79) and having a spouse or partner were significant factors. Forty-eight per cent of the total sample (n=10 974) required pain management while being cared for in hospital. Of these, 39% (n=4235) had pain that was relieved ‘completely, all the time’. The only significant factors were having a neurological condition (including dementia) or ‘other’ condition, living in the most deprived IMD quintile, having a spouse or partner and being of white ethnicity. Full results are set out in table 7. Table 7 Logistic regression of factors with decedent having their pain relieved ‘completely, all the time’ while at home, in a care home and in hospital† Effect OR 99% CI

Twenty-three per cent of the total sample (n=5374) had pain that required management while being cared for in a care home. Forty-six per cent of these (n=2480) reported that pain was relieved ‘completely, all the time’, while the remainder continued to experience pain to some degree. Having a preferred place of death recorded by healthcare staff was not significantly associated with this outcome. Only being aged 80 or over (compared with age 65–79) and having a spouse or partner were significant factors. Forty-eight per cent of the total sample (n=10 974) required pain management while being cared for in hospital. Of these, 39% (n=4235) had pain that was relieved ‘completely, all the time’. The only significant factors were having a neurological condition (including dementia) or ‘other’ condition, living in the most deprived IMD quintile, having a spouse or partner and being of white ethnicity. Full results are set out in table 7. Table 7 Logistic regression of factors with decedent having their pain relieved ‘completely, all the time’ while at home, in a care home and in hospital† Effect OR 99% CI Home (n=9019) Age at death 18–64 0.73* 0.57 to 0.94 65–79 0.89 0.74 to 1.06 80 or above 1.00 Sex Female 1.09 0.94 to 1.27 Male 1.00 Cause of death Cardiovascular disease 0.60* 0.48 to 0.76 Haematological cancer 0.72 0.48 to 1.08 Neurological condition 1.20 0.87 to 1.66 Respiratory illness 0.76* 0.58 to 1.00 ‘Other’ causes‡ 0.63* 0.49 to 0.80 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 1.13 0.88 to 1.45 Second most deprived 1.02 0.80 to 1.28 Third most deprived 1.08 0.86 to 1.35 Fourth most deprived 1.00 0.80 to 1.25 Least deprived 1.00 Relationship of respondent Child/friend/other 0.58* 0.49 to 0.69 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.80 0.49 to 1.30 White 1.00 Advance care plan§ Yes 2.32* 1.97 to 2.73 No 1.00 Per cent correctly classified (predicted vs observed) 65.3%

most deprived 1.00 0.80 to 1.25 Least deprived 1.00 Relationship of respondent Child/friend/other 0.58* 0.49 to 0.69 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.80 0.49 to 1.30 White 1.00 Advance care plan§ Yes 2.32* 1.97 to 2.73 No 1.00 Per cent correctly classified (predicted vs observed) 65.3% Care home (n=5001) Age at death 18–64 0.93 0.58 to 1.49 65–79 0.76* 0.60 to 0.97 80 or above 1.00 Sex Female 1.06 0.90 to 1.24 Male 1.00 Cause of death Cardiovascular disease 0.93 0.73 to 1.19 Haematological cancer 0.75 0.35 to 1.60 Neurological condition 1.24 0.98 to 1.56 Respiratory illness 1.02 0.76 to 1.39 ‘Other’ causes‡ 0.96 0.75 to 1.23 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 1.07 0.83 to 1.38 Second most deprived 1.01 0.80 to 1.28 Third most deprived 1.18 0.94 to 1.48 Fourth most deprived 1.03 0.82 to 1.28 Least deprived 1.00 Relationship of respondent Child/friend/other 0.64* 0.49 to 0.85 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.49 0.22 to 1.08 White 1.00 Advance care plan§ Yes 1.17 0.94 to 1.45 No 1.00 Per cent correctly classified (predicted vs observed) 54.6%

h most deprived 1.03 0.82 to 1.28 Least deprived 1.00 Relationship of respondent Child/friend/other 0.64* 0.49 to 0.85 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.49 0.22 to 1.08 White 1.00 Advance care plan§ Yes 1.17 0.94 to 1.45 No 1.00 Per cent correctly classified (predicted vs observed) 54.6% Hospital (n=10 217) Age at death 18–64 0.85 0.69 to 1.04 65–79 0.92 0.80 to 1.05 80 or above 1.00 Sex Female 0.97 0.87 to 1.09 Male 1.00 Cause of death Cardiovascular disease 1.12 0.96 to 1.31 Haematological cancer 1.10 0.81 to 1.51 Neurological condition 1.43* 1.17 to 1.75 Respiratory illness 1.05 0.86 to 1.29 ‘Other’ causes‡ 1.19* 1.01 to 1.40 Non-haematological cancers 1.00 Level of deprivation of area of residence (IMD quintile) Most deprived 1.20 * 1.00 to 1.44 Second most deprived 1.07 0.90 to 1.27 Third most deprived 1.04 0.88 to 1.23 Fourth most deprived 1.12 0.95 to 1.32 Least deprived 1.00 Relationship of respondent Child/friend/other 0.78* 0.68 to 0.90 Spouse or partner 1.00 Ethnicity Black or Asian minority ethnic 0.57* 0.40 to 0.81 White 1.00 Advance care plan§ Yes 0.89 0.77 to 1.04 No 1.00 Per cent correctly classified (predicted vs observed) 54.0% *p Value <0.01. †Dependent variable: overall care was outstanding or excellent; ‘no’ coded as 0; ‘yes’ coded as 1. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff. IMD, Index of Multiple Deprivation.

†Dependent variable: overall care was outstanding or excellent; ‘no’ coded as 0; ‘yes’ coded as 1. ‡Other than cancer, cardiovascular disease, neurological conditions, renal failure or respiratory illness. §Indicated by having expressed a preference for place of death that was recorded by healthcare staff. IMD, Index of Multiple Deprivation. Discussion Strengths and limitations The National Survey of Bereaved People is statistically well powered. Information on place of death is taken directly from death certificates and the survey includes a wide range of demographic indicators, allowing us to control for covariates. These include factors that previous studies have shown to be associated with a greater likelihood of home death, including cancer diagnosis,9–11 living in a more affluent area9 12 13 and having a spouse or partner,9 14 and with death in a care home, including being aged 80 or over.7 9 Nonetheless, we were limited to covariates in the data set and important explanatory variables may have been omitted, such as clinical need or functionality, decedents' and family members' attitudes to dying, and the availability and practices of local services.14 15 We were also not able to take account of when and in what context preferences were expressed, the extent and quality of discussion underpinning these or whether other preferences were expressed and/or met.

ity, decedents' and family members' attitudes to dying, and the availability and practices of local services.14 15 We were also not able to take account of when and in what context preferences were expressed, the extent and quality of discussion underpinning these or whether other preferences were expressed and/or met. How findings compare with previous studies Despite emphasis in policy, there is little research on the relationship between advance care planning and place of death in England, and none, prior to this study, using nationally representative, individual-level data. In a systematic review of international research on the effects of advance care planning, Brinkman-Stoppelenburg et al16 identified 12 studies with place of death as an outcome. Five used nationally representative samples. These included a study based on a postal survey of a large random sample of general practices in the UK, which found that practices using advance care planning were 2.5 times more likely to have a home death rate for patients with cancer of over 60%.17 The remaining four studies involved secondary analysis of data from the US Health and Retirement Study (HRS).18–21 Two found an association between reduced risk of hospital death and having an advance directive18 or a treatment-limiting advance directive.19 Since the review by Brinkman-Stoppelenburg et al, two further HRS-based studies have found a positive association. Bischoff et al22 found a lower risk of hospital death for those who had either completed an advance directive, had an end of life care discussion with a family member or had assigned power of attorney, while Nicholas et al23 found a lower risk of hospital death for people with a treatment-limiting advance directive, among those with severe dementia living in the community and those with normal cognition living in care homes.

e, had an end of life care discussion with a family member or had assigned power of attorney, while Nicholas et al23 found a lower risk of hospital death for people with a treatment-limiting advance directive, among those with severe dementia living in the community and those with normal cognition living in care homes. Effect sizes in our study were greater than in these HRS-based studies. The difference in the predicted probability of hospital death between those with and without an advance care plan for three of the studies were 9.8%,19 −13%18 and −17.9%,23 and in the remaining study, the relative risk of hospital death for those with an advance care plan was 0.87.22 By comparison, in our study, the difference in the predicted probability of hospital death was −21%, and the relative risk of dying in hospital with an advance care plan was 0.46.

9.8%,19 −13%18 and −17.9%,23 and in the remaining study, the relative risk of hospital death for those with an advance care plan was 0.87.22 By comparison, in our study, the difference in the predicted probability of hospital death was −21%, and the relative risk of dying in hospital with an advance care plan was 0.46. It is difficult to know how to interpret these differences because of the different data sources and measures used. While the National Survey of Bereaved People and HRS have different designs, both are large scale and nationally representative and both gather data from relatives. The National Survey of Bereaved People collects data for decedents of all ages, while HRS includes only people aged 50 and over. However, in the National Survey of Bereaved People, only 5% are aged under 60 (table 1). It could be that the larger effect sizes in our study reflect that the indicator of advance care planning used is specific to the primary outcome (ie, place of death). It also includes only preferences that are recorded by healthcare staff, which are therefore more likely to be acted on. In comparison, advance directives, the main indicator used in the HRS studies, can include preferred place of death, although this will not always be the case. Nonetheless, the vast majority will be used to limit medical treatments requiring hospitalisation in the last weeks and days of life. Preferences should also be recorded by healthcare staff since Medicare-registered and Medicaid-registered healthcare providers are required, under the PSDA (1990), to record advance directives in medical records.

rity will be used to limit medical treatments requiring hospitalisation in the last weeks and days of life. Preferences should also be recorded by healthcare staff since Medicare-registered and Medicaid-registered healthcare providers are required, under the PSDA (1990), to record advance directives in medical records. The larger effect sizes may also, therefore, reflect substantive differences between England and the USA. We know from previous research that levels of advance care planning are higher in the USA than in the UK,24 25 with evidence that some advance care planning in the USA is ‘document-led’ or ‘tick-box’, and therefore potentially less effective.26–28 In England, an expressed preference for death in usual place of residence may be, at least where recorded by healthcare staff, more considered and appropriate to the needs and circumstances of patients and their families. Advance care planning and/or the recording of end of life care preferences by healthcare staff may also be undertaken more selectively, focusing on those who are most able to be supported to die in their usual place of residence. Results from our study indicate that having a recorded preference for place of death is also associated with enhanced support to be cared for and die at home and a better experience of care. There could be various reasons for this. It may be that additional support is provided to these patients to help them achieve a home death. Alternatively, family members may give a better evaluation of the care received when, in fact, these patients are simply more readily supported to be cared for and die at home. It could also be the case that health and care systems that promote advance care planning tend to be those with higher quality community-based services.

vely, family members may give a better evaluation of the care received when, in fact, these patients are simply more readily supported to be cared for and die at home. It could also be the case that health and care systems that promote advance care planning tend to be those with higher quality community-based services. Implications for policy and future research Our study establishes for the first time, using nationally representative, individual-level data, that there is an association, in England, between advance care planning, using the indicator of an expressed preference for place of death recorded by healthcare staff, and a greater likelihood of dying in one's usual place of residence. These findings provide broad support for the emphasis on advance care planning in end of life care policy in England and, in particular, for the role of advance care planning in helping to shift care and resources for end of life care away from hospitals and towards alternative community-based provision. Having a recorded preference for place of death was also found to be associated with a range of quality outcomes. However, the mechanisms underlying these relationships are not well understood. It is unclear, for example, why some people engage (or are engaged) in advance care planning and/or have their preferences for place of death recorded by healthcare staff while others do not, and how and when this occurs. We also need to better understand how advance care planning is linked to wider service responses, including the provision of necessary support to help people to die in their usual place of residence. The larger effect sizes found in our study compared with similar research in the USA may also reflect important differences between England and the USA in how advance care planning is undertaken, when and how preferences are recorded by healthcare staff and the ways in which health and social care services respond. Research into these underlying processes will help in developing and refining advance care planning practices and, given the practical and ethical challenges of randomised designs in this area, help researchers design better controlled observational studies into the effects of advance care planning in future.29 30

ond. Research into these underlying processes will help in developing and refining advance care planning practices and, given the practical and ethical challenges of randomised designs in this area, help researchers design better controlled observational studies into the effects of advance care planning in future.29 30 Contributors: JD was responsible for the conception and overall design of the study, with advice and input from DK and MK. She specified the statistical analyses with support from DK, the LSE team statistician. The National Survey of Bereaved People, 2013 data set, at the time of writing, was not available to researchers outside of the Office for National Statistics (ONS). Consequently, the team at LSE worked closely with Helen Colvin (Senior Research Officer/VOICES Project Manager/Statistician) at the ONS. She advised on the structure of the data and conducted the analyses to a specification designed by the LSE team, under the team's direction. All authors of this paper (JD, DK and MK) had full access to all of the statistical output from these analyses (including statistical reports and tables) and can take responsibility for the integrity of the data and the accuracy of the data analysis. In particular, DK liaised closely with ONS throughout to ensure the correct derivation of variables and checked, interrogated and tabulated all statistical output. He also calculated statistics to determine the fit of the model to the observed data and to compare the results with those achieved in previous studies. JD provided ongoing direction and oversight of all analyses. Helen Colvin and her line manager, Neil Bannister, have both reviewed this paper and have confirmed in writing that the analysis conducted within the ONS has been accurately presented and interpreted. Both Helen Colvin and Neil Bannister are available to answer queries arising about the handling or analysis of data for the purposes of these analyses. Information about the data set is publicly available. Helen Colvin can be contacted at The Office for National Statistics (ONS), Cardiff Road, Newport, Wales NP10 8XG. Tel: 01633 456431, email: helen.r.colvin@ons.gsi.gov.uk. JD is responsible for the drafting of the paper, including all interpretation and contextualisation of results. All authors provided comments on several versions of the paper. In particular, MK provided strategic advice and critical review throughout and provided important feedback and edits on several versions of the final paper.

sible for the drafting of the paper, including all interpretation and contextualisation of results. All authors provided comments on several versions of the paper. In particular, MK provided strategic advice and critical review throughout and provided important feedback and edits on several versions of the final paper. Funding: The research reported in this paper was funded by the NIHR School for Social Care Research; Ref: T976/EM/LSE6. Disclaimer: This article presents independent research funded by the NIHR School for Social Care Reserach (SSCR). The views expressed in this publication are those of the authors and not necessarily those of the SSCR, National Institute for Health Research (NIHR) or Department of Health. Competing interests: None declared. Provenance and peer review: Not commissioned; externally peer reviewed.

ACP, advance care planning; CCG, clinical commissioning group; DiUPR, death in usual place of residence;EPaCCS, Electronic Palliative Care Coordination Systems; HES, Hospital Episodes Statistics;NHS, National Health Service; ONS, Office of National Statistics; PPD, preferred place of death. Figure 1 PRISMA flow diagram if papers reporting numbers of included and excluded studies. EPaCCS, electronic palliative care coordination systems; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses. Summary of included studies Study design Fourteen of 26 articles comprised ‘expert opinion’ (editorials, commentaries and discussion papers).5 15–27 Of the remaining 12 articles, nine were quantitative studies,6 7 28–34 two were qualitative35 36 and one was a mixed-method study though predominantly qualitative.37 All nine quantitative studies had an observational design: one was a service evaluation,33 two were descriptive cross-sectional studies,31 32 four were retrospective cohort studies without control groups6 7 29 30 and two were retrospective cohort studies with control groups28 34 (figure 2). All studies using qualitative methods used semistructured interviews,35–37 and one used analysis of documentation and service observations in addition to interviews.36 Table 1 reports characteristics of included articles. Figure 2 Type of studies included according to the hierarchy of evidence based on Murad et al.12

Palliative care research is essential for evidence-based models of care that improve terminal illness outcomes.1 Investment in hospice research is essential for the projected increases in palliative needs.2 Hospices are crucial to practice and policy development, but there are many hospice research barriers.3 Well-documented problems include lack of research funding, governance challenges, methodological issues and absence of research culture.1 4 Hospices require strategic infrastructure and processes to progress from ‘research aware’ to full participation and research leadership.3 Over 2 years, Marie Curie Hospice Liverpool developed a series of quality improvement initiatives to establish a UK hospice research culture and improve research opportunities.

e.1 4 Hospices require strategic infrastructure and processes to progress from ‘research aware’ to full participation and research leadership.3 Over 2 years, Marie Curie Hospice Liverpool developed a series of quality improvement initiatives to establish a UK hospice research culture and improve research opportunities. First, a governance process was established to facilitate the appropriate sponsorship and indemnity arrangements. This involved partnership with a local National Health Service Trust and University research teams. Patient and public engagement activities for research were established through our day therapy services. Patients met with researchers to provide perspectives for proposed research (from idea conception to research delivery). Second, a new morning board-round handover system provided focused clinical handover to hospice staff. As part of that, potential research participation eligibility was routinely discussed for all hospice inpatients. Third, Liverpool Clinical Commissioning Group funds provided a doctor (AMcD) and nurse (SS) with protected research time over 6 months to support National Institute for Health Research (NIHR) portfolio research studies. Initial duties involved training in hospice research processes. This included relevant training completion, data record management, ethical review facilitation and research promotion in the hospice. Our staff established contacts with the NIHR North West Clinical Research Network nurses in the local Trust. We focused on recruitment (patients and family caregivers) and data collection for the NIHR studies.

raining completion, data record management, ethical review facilitation and research promotion in the hospice. Our staff established contacts with the NIHR North West Clinical Research Network nurses in the local Trust. We focused on recruitment (patients and family caregivers) and data collection for the NIHR studies. The process has led to several positive outcomes. Notably, patients were eager to participate in studies and showed great desire to improve future care, consistent with previous work.5 The process improved research opportunities for hospice patients, increased recruitment and created new research collaborations. Both staff members (AMcD and SS) developed research skill and knowledge, with one (SS) in receipt of a national award. Challenges encountered were time-related due to research governance process delays. Ongoing issues exist with funding for further work. We have recently appointed a research lead (ACN) to further support research delivery. In summary, the process demonstrated research improvement potential through steps to develop and engage hospice staff, secure funding and establish collaborative research partnerships. This activity has directly resulted in a growing research hospice culture and has potential as a model for other hospices. Amara Nwosu, Laura Chapman and Sarah Stanley’s hospice posts are supported by Marie Curie. Amara Nwosu and Sarah Stanley's academic posts are supported by the National Institute for Health Research (NIHR) North West Coast - Clinical Research Network (CRN). Twitter: Follow Sarah Stanley at @Sarah_Stan_

The process has led to several positive outcomes. Notably, patients were eager to participate in studies and showed great desire to improve future care, consistent with previous work.5 The process improved research opportunities for hospice patients, increased recruitment and created new research collaborations. Both staff members (AMcD and SS) developed research skill and knowledge, with one (SS) in receipt of a national award. Challenges encountered were time-related due to research governance process delays. Ongoing issues exist with funding for further work. We have recently appointed a research lead (ACN) to further support research delivery. In summary, the process demonstrated research improvement potential through steps to develop and engage hospice staff, secure funding and establish collaborative research partnerships. This activity has directly resulted in a growing research hospice culture and has potential as a model for other hospices. Amara Nwosu, Laura Chapman and Sarah Stanley’s hospice posts are supported by Marie Curie. Amara Nwosu and Sarah Stanley's academic posts are supported by the National Institute for Health Research (NIHR) North West Coast - Clinical Research Network (CRN). Twitter: Follow Sarah Stanley at @Sarah_Stan_ Follow Amara Nwosu at @amaranwosu Contributors: SS, AMcD, LC and ACN all contributed to the writing of this manuscript. Funding: National Institute for Health Research (NIHR) North West Coast Clinical Research Network (CRN) (£19 086.57). Liverpool Clinical Commissioning Group (CCG) Research Capability Funding (RCF) (£22 029).

Follow Amara Nwosu at @amaranwosu Contributors: SS, AMcD, LC and ACN all contributed to the writing of this manuscript. Funding: National Institute for Health Research (NIHR) North West Coast Clinical Research Network (CRN) (£19 086.57). Liverpool Clinical Commissioning Group (CCG) Research Capability Funding (RCF) (£22 029). Competing interests: None declared. Patient consent for publication: Not required. Ethics approval: The study of hydration status and complex symptoms in advanced cancer using bioelectrical impedance vector analysis (BIVA): North West - Haydock (REC reference 17/NW/0050). Biology of dying study: North Wales Research Ethics Committee - West (REC reference 13/WA/0266). Provenance and peer review: Not commissioned; internally peer reviewed.

Introduction Advance care plans allow individuals to specify their wishes and preferences for treatment and care, and have been proposed as a quality indicator for end-of-life care.1 For advance care planning to impact maximally on patients and their caregivers, plans need to be made available to all relevant health professionals, including out-of-hours and primary care services, so that care can be delivered in line with patients’ preferences, in a coordinated manner.2 3

tor for end-of-life care.1 For advance care planning to impact maximally on patients and their caregivers, plans need to be made available to all relevant health professionals, including out-of-hours and primary care services, so that care can be delivered in line with patients’ preferences, in a coordinated manner.2 3 In the UK, there has been a policy focus on end-of-life care coordination since the Department of Health’s 2008 End of Life Care Strategy, which recommended the creation of Locality Registers for people approaching the end of their life.4 Electronic palliative care coordination systems (EPaCCS) emerged to support this. EPaCCS are electronic registers that aim to facilitate documentation of up-to-date information about patients’ preferences and plans for care, in a format that allows sharing of information among different healthcare providers including general practitioner (GP) practices, out-of-hour services, emergency departments, ambulance services, hospices and care homes, among others.5 Although many countries have developed electronic palliative care registers, the focus on coordination of care between different settings appears unique to the UK.5 EPaCCS started as eight pilots commissioned by the Department of Health in 2009–2011.6 7 By 2013, 175 (82.9%) of clinical commissioning groups (CCGs) had either implemented EPaCCS or started planning for their implementation.8 In 2015, the Department of Health’s National Commitment for End of Life Care recommended national roll-out of EPaCCS by 2020.9

mmissioned by the Department of Health in 2009–2011.6 7 By 2013, 175 (82.9%) of clinical commissioning groups (CCGs) had either implemented EPaCCS or started planning for their implementation.8 In 2015, the Department of Health’s National Commitment for End of Life Care recommended national roll-out of EPaCCS by 2020.9 A discussion paper published in 2016 highlighted challenges associated with implementation of EPaCCS and identified key drivers for their success.5 This review noted a lack of rigorous research around EPaCCs though this was not systematically synthesised. Our aim was to systematically search, evaluate and report the state of the science of EPaCCS, in order to identify gaps in the evidence and make recommendations for policy and research. Methods Study design A systematic review of the state of the science of EPaCCS. The review was conducted and reported following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.10 Eligibility criteria Study design All published articles were considered, including editorials, narrative summaries, observational, experimental and qualitative studies. Government or independent reports were included if they reported an evaluation of EPaCCS and contained data not previously published. Conference abstracts were not included. Participants All studies were included, regardless of participants’ age or disease.

Eligibility criteria Study design All published articles were considered, including editorials, narrative summaries, observational, experimental and qualitative studies. Government or independent reports were included if they reported an evaluation of EPaCCS and contained data not previously published. Conference abstracts were not included. Participants All studies were included, regardless of participants’ age or disease. Interventions We included any type of electronic system that enables professionals to share information about advance care plans or advance directives. The rationale for the intervention was the coordination and/or sharing of information about advance care plans between different healthcare providers. Studies describing electronic systems that have no relation to advance care plans or advance directives, or reporting aspects of electronic systems other than sharing or coordination of information (eg, for use by single providers), were excluded. As EPaCCS are a UK initiative, we included only articles describing UK tools.

es describing electronic systems that have no relation to advance care plans or advance directives, or reporting aspects of electronic systems other than sharing or coordination of information (eg, for use by single providers), were excluded. As EPaCCS are a UK initiative, we included only articles describing UK tools. Search strategy The electronic databases searched were: CINAHL (1981 to January 2019), MEDLINE (R) In-Process (1946 to January 2019, via OVID), Embase (1974 to January 2019, via OVID) and the Cochrane Library (up to January 2019). All searches were carried out on the 11 January 2019. Search terms including subject headings and free-text words were developed and adapted for each database (online supplementary file 1). Reference lists of relevant papers were handsearched to identify additional studies. Grey literature including government and independent reports was identified from the references lists of included papers and a simple Google search. Unpublished literature was not identified. 10.1136/bmjspcare-2018-001689.supp1Supplementary data

Search strategy The electronic databases searched were: CINAHL (1981 to January 2019), MEDLINE (R) In-Process (1946 to January 2019, via OVID), Embase (1974 to January 2019, via OVID) and the Cochrane Library (up to January 2019). All searches were carried out on the 11 January 2019. Search terms including subject headings and free-text words were developed and adapted for each database (online supplementary file 1). Reference lists of relevant papers were handsearched to identify additional studies. Grey literature including government and independent reports was identified from the references lists of included papers and a simple Google search. Unpublished literature was not identified. 10.1136/bmjspcare-2018-001689.supp1Supplementary data Selection of studies One of the authors (JL) screened all titles and abstracts retrieved from the electronic database and reference list searches and excluded all those that were not from the UK. Two authors (JL and KES) independently screened the remaining titles and abstracts, and disagreements were resolved by discussion. Full texts were screened by two authors independently (JL, KES or AW). Disagreement was resolved by discussion or consulting a third author. Because we only included UK-based studies, language restrictions were not relevant.

tly screened the remaining titles and abstracts, and disagreements were resolved by discussion. Full texts were screened by two authors independently (JL, KES or AW). Disagreement was resolved by discussion or consulting a third author. Because we only included UK-based studies, language restrictions were not relevant. Data extraction Data were extracted from all included studies using a data extraction spreadsheet. Data items included study design, aim of the study, setting and participants, the outcome evaluated and results (online supplementary file 2, table 1). Two authors (JL, AW or KES) independently extracted data on all studies and disagreements were resolved by discussion or in consultation with a third author (KES or AW). 10.1136/bmjspcare-2018-001689.supp2Supplementary data Table 1 Characteristics of included studies

Data extraction Data were extracted from all included studies using a data extraction spreadsheet. Data items included study design, aim of the study, setting and participants, the outcome evaluated and results (online supplementary file 2, table 1). Two authors (JL, AW or KES) independently extracted data on all studies and disagreements were resolved by discussion or in consultation with a third author (KES or AW). 10.1136/bmjspcare-2018-001689.supp2Supplementary data Table 1 Characteristics of included studies Author Study design Level of evidence Qual/quant Region Participants Methods of data collection Type of outcome reported Ali et al 28 Retrospective cohort 3. Controlled cohort and case control studies Quantitative Grampian Patients with diagnosis of cancer and ‘terminal care’ who contacted the GMED from January 2011 to December 2011 Secondary data analysis Admission to hospital and length of stay Allsop15 Commentary 6. Expert opinion NA NA NA   NA Allsop et al 6 Retrospective cohort 4. Qualitative and observational without control group Quantitative Leeds Patients who died from April 2014 to March 2015 with an EPaCCS record Secondary data analysis Achievement of PPD and ACP documentation Broadhurst et al 29 Retrospective cohort 4. Qualitative and observational without control group Quantitative London Patients with an EPaCC created between December 2015 and September 2016 Secondary data analysis Achievement of PPD Callender et al 30 Retrospective cohort 4. Qualitative and observational without control group Quantitative London Patients who died from March 2011 to September 2016 with an EPaCCS record Secondary data analysis Achievement of PPD Hall et al 35 Qualitative interview study 4. Qualitative and observational without control group Qualitative Grampian and Lothian Primary care and out-of-hours healthcare professionals, patients and carers users of EPaCCS 22 semistructured interviews Implementation Hamilton16 Editorial 6. Expert opinion NA NA NA   NA Henry and Hayes17 Commentary 6.Expert opinion NA NA NA   NA Hunt18 Commentary 6.Expert opinion NA NA NA   NA Johnson19 Commentary 6.Expert opinion NA NA NA   NA Jones and Whitmore20 Editorial 6. Expert opinion NA NA NA   NA Lindsey and Hayes21 Commentary 6. Expert opinion NA NA NA   NA Millares Martin31 Cross-sectional 4. Qualitative and observational without control group Quantitative UK CCGs across England 209 surveys Implementation Millington-Sanders et al 32 Descriptive 4. Qualitative and observational without control group Quantitative Richmond, London Patients with an EPaCCS record between November 2010 and August 2012 Secondary data analysis Achievement of PPD Mullick et al 22 Narrative review 6.

CCGs across England 209 surveys Implementation Millington-Sanders et al 32 Descriptive 4. Qualitative and observational without control group Quantitative Richmond, London Patients with an EPaCCS record between November 2010 and August 2012 Secondary data analysis Achievement of PPD Mullick et al 22 Narrative review 6. Expert opinion NA NA NA   NA Murphy-Jones23 Commentary 6. Expert opinion NA NA NA   NA Petrova et al 5 Discussion paper 6. Expert opinion NA NA NA   NA Pringle et al 33 Service evaluation 4. Qualitative and observational without control group Quantitative Lothian Patients known to specialist palliative care community team Secondary data analysis ACP documentation Purdy et al 34 Retrospective cohort with control group 3. Controlled cohort and case control studies Quantitative North Somerset and Somerset Patients who died from September 2011 to February 2012 and potentially eligible for EoLC Secondary data analysis Place of death and hospital admissions Riley and Madill24 Discussion paper 6. Expert opinion NA NA NA   NA Sleeman and Higginson25 Commentary 6. Expert opinion NA NA NA   NA Smith et al 7 Retrospective cohort 4. Qualitative and observational without control group Quantitative London Patients who died from August 2010 to March 2012 with an EPaCCS record Secondary data analysis Achievement of PPD Smith and Riley26 Commentary 6. Expert opinion NA NA NA   NA Wye et al 36 Realistic evaluation 4. Qualitative and observational without control group Qualitative North Somerset and Somerset Healthcare professionals, patients and carers users of EPaCCS Documentation analysis, 15 observations of services, 148 interviews Implementation Wye et al 37 Mixed-methods 4. Qualitative and observational without control group Mixed methods North Somerset and Somerset Healthcare professionals users of EPaCCS 101 interviews Implementation Wye27 Commentary 6. Expert opinion NA NA NA   NA ACP, advance care planning; CCG, clinical commissioning group;EPaCCS, electronic palliative care coordination systems; EoLC, end-of-life care; GMED, Grampian medical emergency department; NA, not applicable; PPD, preferred place of death.

CS 101 interviews Implementation Wye27 Commentary 6. Expert opinion NA NA NA   NA ACP, advance care planning; CCG, clinical commissioning group;EPaCCS, electronic palliative care coordination systems; EoLC, end-of-life care; GMED, Grampian medical emergency department; NA, not applicable; PPD, preferred place of death. Quality evaluation The quality of the included articles was assessed using the Standard Quality Assessment Criteria for evaluation of primary research papers from different fields and design.11 The checklist for assessing the quality of quantitative articles consists of 14 potential criteria scored on a 3-point scale (higher points indicate better quality). The quality assessment checklist for qualitative studies is scored in a similar way and consists of 10 essential criteria. Two authors (JL, AW or KES) independently appraised the included articles. Disagreements were resolved by discussion or in consultation with the third author. Reports were not included in the quality assessment. All eligible studies were included in the review, irrespective of quality score. Analysis We derived a hierarchy of evidence incorporating both quantitative and qualitative studies, adapted from primary work by Murad et al.12 We included editorials, commentaries, discussion papers and narrative summaries as ‘expert opinion’. Reports were not included in the evidence hierarchy but reported separately. References were managed using EndNote V.X7,13 and data were tabulated using Excel.14 A descriptive summary of results was reported.

d et al.12 We included editorials, commentaries, discussion papers and narrative summaries as ‘expert opinion’. Reports were not included in the evidence hierarchy but reported separately. References were managed using EndNote V.X7,13 and data were tabulated using Excel.14 A descriptive summary of results was reported. Results The search strategy identified 1163 articles. Thirty-one additional articles and reports were identified from handsearching. After removing duplicates, 730 titles and abstracts were screened. Six hundred and seventy-seven articles were excluded based on the title or abstract. Fifty-three relevant full texts were screened. Twenty-three were excluded because they were not about advance care planning, not about coordination systems, were conference abstracts or were reports with data reported in another article already included. Table 2 in the online supplementary file 2 reports characteristics of excluded studies. Twenty-six studies and four reports were included in the qualitative synthesis (figure 1). Table 2 Characteristics of included reports

Results The search strategy identified 1163 articles. Thirty-one additional articles and reports were identified from handsearching. After removing duplicates, 730 titles and abstracts were screened. Six hundred and seventy-seven articles were excluded based on the title or abstract. Fifty-three relevant full texts were screened. Twenty-three were excluded because they were not about advance care planning, not about coordination systems, were conference abstracts or were reports with data reported in another article already included. Table 2 in the online supplementary file 2 reports characteristics of excluded studies. Twenty-six studies and four reports were included in the qualitative synthesis (figure 1). Table 2 Characteristics of included reports Author Study design Level of evidence Qual/quant Region Participants Methods of data collection Type of outcome reported Ipsos MORI39 Pilot evaluation Report Qualitative Brighton and Hove, London, Leeds, Mid Essex, Salford, Sandwell and North Somerset Pilot leaders Document analysis, interviews Implementation NHS Improving Quality38 Cross-sectional Report Quantitative CCGs in England CCGs across England 188 surveys Implementation, ACP documentation and place of death NHS Improving Quality40 Mixed methods Report Quantitative Brighton and Hove, London, Leeds, Mid Essex, Salford, Sandwell, Medway, Bedfodshire, Birmingham, North East and North Somerset HES hospital care, local EPaCCS data and ONS data on DiUPR from January 2008 to June 2012. Members of EPaCCS team. Secondary data analysis and 55 surveys DiUPR, achievement of PPD, hospital admissions, cost and implementation* Qualitative EPaCCS team members Two focus group Implementation Whole System Partnership41 Mixed methods Report Quantitative 10 sites from London, South West England, East of England, North West and East Midlands ONS data on place of death in 139 CCGs between 2011 and 2016. HES data on hospital admissions in the last year of life in 10 CCGs areas. Secondary data analysis and 91 surveys DiUPR and cost Qualitative Healthcare professionals, patients, carers, system leaders. 12 in-depth interviews Implementation *Data on place of death were only available for four EPaCCS sites.

Figure 2 Type of studies included according to the hierarchy of evidence based on Murad et al.12 Of four reports, one was a cross-sectional survey,38 one used a qualitative approach through interviews and documentation analysis,39 and two used mixed-methods approaches combining both quantitative and qualitative data analysis.40 41 Table 2 reports characteristics of included reports. Population Of the 12 articles including primary data, all included individuals with a terminal condition. One article focused exclusively on patients with cancer,28 while the rest did not specify any particular condition. In four articles the population was individuals who had died.6 7 30 34 In five articles, the population was patients who were registered on an EPaCCS system.6 7 29 30 32 In four of the articles, participants were healthcare professionals35–37 or commissioners at the CCGs level.31 The four reports included variously EPaCCS leaders,39–41 healthcare professionals and patients,41 and commissioners.38 For four of the articles the setting was London.7 29 30 32 Three were from Grampian and Lothian,28 33 35 three from Somerset,34 36 37 one from Leeds6 and one was nation wide.31 All four reports included multiple regions in England.

Population Of the 12 articles including primary data, all included individuals with a terminal condition. One article focused exclusively on patients with cancer,28 while the rest did not specify any particular condition. In four articles the population was individuals who had died.6 7 30 34 In five articles, the population was patients who were registered on an EPaCCS system.6 7 29 30 32 In four of the articles, participants were healthcare professionals35–37 or commissioners at the CCGs level.31 The four reports included variously EPaCCS leaders,39–41 healthcare professionals and patients,41 and commissioners.38 For four of the articles the setting was London.7 29 30 32 Three were from Grampian and Lothian,28 33 35 three from Somerset,34 36 37 one from Leeds6 and one was nation wide.31 All four reports included multiple regions in England. Quality appraisal of included articles The mean quality appraisal score for quantitative studies was 84.8% (online supplementary file 2, tables 3 and 4). The most common source of poor quality for quantitative studies was the lack of an evident and appropriate study design and a poorly defined comparison group. The mean quality appraisal score for qualitative studies was 83.3%. The most common source of poor quality for qualitative studies was a lack of reflexivity of the account and description of the theoretical framework. Table 3 Characteristics of EPaCCS studies reporting place of death and preferred place of death

Quality appraisal of included articles The mean quality appraisal score for quantitative studies was 84.8% (online supplementary file 2, tables 3 and 4). The most common source of poor quality for quantitative studies was the lack of an evident and appropriate study design and a poorly defined comparison group. The mean quality appraisal score for qualitative studies was 83.3%. The most common source of poor quality for qualitative studies was a lack of reflexivity of the account and description of the theoretical framework. Table 3 Characteristics of EPaCCS studies reporting place of death and preferred place of death Author Sample % with cancer PPD home (%) PPD achieved (%) PPD achieved if PPD was home (%)/hospice (%) Likelihood of hospital death OR (95% CI)* Allsop et al 6 1229 – 55 75 65/83 – Broadhurst et al 29 6854 43 55 79 – – Callender et al 30 9027 57 65 78 72/79 – Millington-Sanders et al 32 597 – 35† – – – Smith et al 7 207 46 40 55 68/34 – Purdy et al 34 1022‡ 68 – – – 0.30 (0.13 to 0.69)* 2572§ – – – 0.22 (0.12 to 0.40)* *OR for dying at home for those with an EPaCCS record versus those without an EPaCCS record, adjusted by gender, age, deprivation and cause of death. †29.0% of the 138 individuals who died in the sample died at home. ‡North Somerset. §Somerset. EPaCCS, electronic palliative care coordination systems; EoLC, end-of-life care; PPD, preferred place of death. Table 4 Characteristics of EPaCCS studies reporting hospital admission

Author Sample % with cancer PPD home (%) PPD achieved (%) PPD achieved if PPD was home (%)/hospice (%) Likelihood of hospital death OR (95% CI)* Allsop et al 6 1229 – 55 75 65/83 – Broadhurst et al 29 6854 43 55 79 – – Callender et al 30 9027 57 65 78 72/79 – Millington-Sanders et al 32 597 – 35† – – – Smith et al 7 207 46 40 55 68/34 – Purdy et al 34 1022‡ 68 – – – 0.30 (0.13 to 0.69)* 2572§ – – – 0.22 (0.12 to 0.40)* *OR for dying at home for those with an EPaCCS record versus those without an EPaCCS record, adjusted by gender, age, deprivation and cause of death. †29.0% of the 138 individuals who died in the sample died at home. ‡North Somerset. §Somerset. EPaCCS, electronic palliative care coordination systems; EoLC, end-of-life care; PPD, preferred place of death. Table 4 Characteristics of EPaCCS studies reporting hospital admission Author Sample % with cancer Likelihood of being admitted to hospital OR (95% CI)* Likelihood of attend A&E OR (95% CI)† Ali et al 28 401 100 0.41 (0.24 to 0.71)‡ Purdy et al 34 1022§ 68 0.65 (0.33 to 1.30)¶ 0.57 (0.29 to 1.11)¶ 2572** 0.41 (0.28 to 0.60)¶ 0.61 (0.40 to 0.92)¶ *OR for being admitted to hospital for those with an EPaCCS record versus those without an EPaCCs record. †OR for having an A&E visit in the last 30 days of life for those with an EPaCCS record versus those without an EPaCCS. ‡Adjusted for reason for attending the emergency department. §North Somerset. ¶Adjusted for gender, age, deprivation and condition. **Somerset. A&E, Accident and Emergency Department; EPaCCS, electronic palliative care coordination systems.

†OR for having an A&E visit in the last 30 days of life for those with an EPaCCS record versus those without an EPaCCS. ‡Adjusted for reason for attending the emergency department. §North Somerset. ¶Adjusted for gender, age, deprivation and condition. **Somerset. A&E, Accident and Emergency Department; EPaCCS, electronic palliative care coordination systems. Description of studies Expert opinion articles Fourteen articles were classified as expert opinion. Two were editorials,16 20 nine were commentaries,15 17–19 21 23 25–27 two were discussion and analytical papers,5 24 and one was a narrative review on advance care planning.22 Five articles provided an overview of EPaCCS in the UK context, including implementation challenges and references to data from audits, internal reports and some published articles.5 17 18 21 24 Six articles were a commentary to another published study.15 19 23 25–27 In four commentary articles and one editorial, authors expressed a favourable opinion regarding the potential of EPaCCS to improve end-of-life care outcomes.17 18 20 23 26 Three commentaries raised concerns about the lack of strong evidence for EPaCCS and the difficulties and potential bias of the current available research.15 25 27 Lindsey and Hayes21 provided an overview of EPaCCS, including definitions, concepts and mandatory elements required to develop and implement them in the UK.

Three commentaries raised concerns about the lack of strong evidence for EPaCCS and the difficulties and potential bias of the current available research.15 25 27 Lindsey and Hayes21 provided an overview of EPaCCS, including definitions, concepts and mandatory elements required to develop and implement them in the UK. Riley and Madill24 discussed the experience of developing and implementing an EPaCCS in London.24 The article provided preliminary audit data, and described how EPaCCS might impact on end-of-life care outcomes, providing some experiential examples. Petrova et al 5 present a critical analysis of EPaCCS. The article discussed five key challenges for EPaCCS: the need to involve different actors and settings, the complexity of an intervention that requires a change in the culture of an organisation, the lack of interoperability of IT systems in the UK, information governance issues, and uncertainties and sensibilities regarding end-of-life issues. The authors proposed six drivers for the successful spread of EPaCCS and the next steps for improvements. Quantitative articles Of the nine quantitative articles, seven reported an aspect of patients’ management such as place of death or hospital admission,6 7 28–30 32 34 and three reported an aspect of the process of implementing EPaCCS.6 31 33 Place of death and achieving preferred place of death Six articles were focused on the place of death.6 7 29 30 32 34 These included five retrospective cohort studies,6 7 29 30 34 and one cross-sectional study.32

Quantitative articles Of the nine quantitative articles, seven reported an aspect of patients’ management such as place of death or hospital admission,6 7 28–30 32 34 and three reported an aspect of the process of implementing EPaCCS.6 31 33 Place of death and achieving preferred place of death Six articles were focused on the place of death.6 7 29 30 32 34 These included five retrospective cohort studies,6 7 29 30 34 and one cross-sectional study.32 Five articles reported the preferred place of death of people with an EPaCCS record,6 7 29 30 32 of which four reported the proportion of patients who achieved their preference.6 7 29 30 Dying at home was preferred by 35% to 65% of patients, and 55% to 79% of individuals with an EPaCCS record achieved their preference (table 3). Four articles examined factors associated with achieving preferred place of death, among people with EPaCCS records. In Callender et al, being female, having cancer as primary diagnosis, a WHO performance score of 4, a not for resuscitation status and reporting hospice or care home as preferred place of death were associated with increased odds of achieving the preferred place of death.30 Allsop et al 6 found that patients who reported hospice or care homes as preferred place of death were more likely to have achieved their preferred place of death. Smith et al 7 reported that a smaller proportion of patients that preferred to die in a hospice achieved their preferred place of death compared with patients who preferred to die at home or in a care home. Broadhurst et al 29 found that individuals with an EPaCCS who also had a ceiling of treatment plan recorded were more likely to die in their preferred place of death. None of these studies compared people with EPaCCS records with those without an EPaCCS record.

h patients who preferred to die at home or in a care home. Broadhurst et al 29 found that individuals with an EPaCCS who also had a ceiling of treatment plan recorded were more likely to die in their preferred place of death. None of these studies compared people with EPaCCS records with those without an EPaCCS record. Purdy et al 34 carried out a retrospective cohort study with control group. They identified individuals who were potentially eligible for end-of-life care and had died between 2011 and 2012, from healthcare records. They found individuals with an end-of-life electronic record had lower odds of death in hospital than individuals without an end-of-life electronic record after adjusting by gender, age, deprivation and cause of death (table 3). In this study, EPaCCS were implemented as part of a complex intervention that included improved out-of-hours provision and coordination between hospital and community settings, and only 12% of the eligible sample had an EPaCCS record. Admission to hospital Two articles focused on hospital admissions28 34; both were retrospective cohort studies with control groups.

Purdy et al 34 carried out a retrospective cohort study with control group. They identified individuals who were potentially eligible for end-of-life care and had died between 2011 and 2012, from healthcare records. They found individuals with an end-of-life electronic record had lower odds of death in hospital than individuals without an end-of-life electronic record after adjusting by gender, age, deprivation and cause of death (table 3). In this study, EPaCCS were implemented as part of a complex intervention that included improved out-of-hours provision and coordination between hospital and community settings, and only 12% of the eligible sample had an EPaCCS record. Admission to hospital Two articles focused on hospital admissions28 34; both were retrospective cohort studies with control groups. Ali et al 28 report a retrospective cohort study among individuals with cancer in Grampian. Electronic records were searched for individuals with a consultation related to a cancer diagnosis and terminal care. The content of consultation was manually reviewed to confirm the cancer diagnosis. No information on type or stage of cancer was reported. The authors identified the presence or absence of an EPaCCS record. They found the likelihood of being admitted to hospital within a year for patients with a diagnosis of advanced cancer was lower in individuals registered on the EPaCCS system compared with individuals without an EPaCCS record (table 4). However, the authors reported that implementation of EPaCCS was limited by difficulties uploading summaries into the system.

d to hospital within a year for patients with a diagnosis of advanced cancer was lower in individuals registered on the EPaCCS system compared with individuals without an EPaCCS record (table 4). However, the authors reported that implementation of EPaCCS was limited by difficulties uploading summaries into the system. Purdy e t al 34 found that the chances of being admitted to hospital and having emergency department attendance in the last 30 days of life were lower in individuals with advanced life-threatening illnesses who had an EPaCCS compared with those who did not (table 4). Use of EPaCCS Three articles focused on use of EPaCCS: one retrospective cohort study,6 one cross-sectional study31 and one service evaluation.33 Two articles reported the proportion of patients with EPaCCS in the general population,6 and in a palliative care population,33 and found EPaCCS were used in 27% and 54% of patients in these groups, respectively. Millares Martin surveyed 209 CCGs in England regarding the level of EPaCCS implementation in their area. Nationally, 29% of CCGs had no EPaCCS in use, and the proportion of CCGs with a fully operative EPaCCS varied from 34% in the Midlands and East of England to 84% in London.31 Qualitative and mixed-methods articles The three qualitative (one of which was mixed-methods) articles focused on how different EPaCCS systems were implemented in practice, and the challenges of developing and implementing an EPaCCS system locally or nationally.35–37 Common challenges included difficulties with IT systems and poor uptake among primary care and in-hours staff.

ve (one of which was mixed-methods) articles focused on how different EPaCCS systems were implemented in practice, and the challenges of developing and implementing an EPaCCS system locally or nationally.35–37 Common challenges included difficulties with IT systems and poor uptake among primary care and in-hours staff. Hall et al 35 described a new EPaCCS system in Scotland and explored both perceptions and feasibility of its implementation. Views of patients, carers, primary care and out-of-hours staff were positive and thought to be beneficial to patients, while out-of-hours staff identified that it supported and facilitated their work. Key issues that arose centred on technical problems and poor uptake within the primary care setting.

ts implementation. Views of patients, carers, primary care and out-of-hours staff were positive and thought to be beneficial to patients, while out-of-hours staff identified that it supported and facilitated their work. Key issues that arose centred on technical problems and poor uptake within the primary care setting. Wye et al 36 37 presented data from a service evaluation of an innovative palliative care programme that included a new EPaCCS in Southwest England. Wye et al 36 found that use of the system was often patchy and perceived to work best for those people with cancer who had fast track funding in place and who were close to death. They also highlighted that programme’s success was in part due to their dedicated and motivated staff. Wye et al 37 identified that out-of-hours staff found EPaCCS records useful, however, in-hours staff perceived EPaCCS records to be problematic as they meant additional work with little or no benefit for them. Record access was also very poor among paramedics, which was thought to be mainly due to technical issues. This study included quantitative data on place of death and likelihood of dying in hospital, from the same cohort analysed by Purdy et al.34

ematic as they meant additional work with little or no benefit for them. Record access was also very poor among paramedics, which was thought to be mainly due to technical issues. This study included quantitative data on place of death and likelihood of dying in hospital, from the same cohort analysed by Purdy et al.34 Reports The Ipsos MORI report was an independent evaluation of the EPaCCS pilot programme. The report described variation in the level of implementation and challenges that the eight pilot sites experienced, based on views and experiences of stakeholders. Data from the pilot sites indicated EPaCCS were well received by patients and carers. Stakeholders reported benefits including improved communication, delivery of patient choices and reduction in unnecessary admissions or appointments. Concerns were expressed regarding data sharing and security, increase in GP workload, the process of patient consent, engaging in difficult end-of-life conversations, ensuring the inclusion of non-cancer patients, integrating data from multiple IT systems, the development of multiple registers systems across the UK and the timely access of all providers to the register.39

urity, increase in GP workload, the process of patient consent, engaging in difficult end-of-life conversations, ensuring the inclusion of non-cancer patients, integrating data from multiple IT systems, the development of multiple registers systems across the UK and the timely access of all providers to the register.39 National Health Service (NHS) Improving Quality performed an economic evaluation based on four early implementer sites. The report analysed aggregated data from Hospital Episode Statistics and Office of National Statistics before and after EPaCCS implementation, and found use of EPaCCS was associated with an increase in the proportion of people dying in their usual place of residence, and a small reduction in the number of unscheduled admissions and length of stay in hospital. No statistics were performed to understand whether these differences were significant or not.40 The report found 63% to 77% of individuals with an EPaCCS died in their preferred place, and 6% to 18% of individuals with an EPaCCS died in hospital. The report concluded that EPaCCS could potentially generate savings from £399 to £1480 per death avoided in hospital.40

ther these differences were significant or not.40 The report found 63% to 77% of individuals with an EPaCCS died in their preferred place, and 6% to 18% of individuals with an EPaCCS died in hospital. The report concluded that EPaCCS could potentially generate savings from £399 to £1480 per death avoided in hospital.40 The National End of Life Care Intelligence Network carried out a survey of 211 CCGs in 2013. Sixty-four CCGs (30%) reported having an operational EPaCCS in place. A total of 111 (53%) had started planning for EPaCCS implementation, and 10 (5%) had no plans. Ten different technical systems of EPaCCS were reported by CCGs and only 15 CCGs reported full compliance with all the items included in the national information standard. Only 17% of CCGs had accurate information regarding the actual place of death and death in the preferred place for individuals on EPaCCS.38

o plans. Ten different technical systems of EPaCCS were reported by CCGs and only 15 CCGs reported full compliance with all the items included in the national information standard. Only 17% of CCGs had accurate information regarding the actual place of death and death in the preferred place for individuals on EPaCCS.38 A more recent evaluation developed by the Whole Systems Partnership collected information from five sites with established EPaCCS and six control sites, and included in-depth interviews of patients, carers and healthcare professionals.41 The interviews identified challenges including identifying who should have an EPaCCS record, difficulties of having advance care planning conversations, the need for education and training in end-of-life care choices, ensuring records are updated and the time needed to complete an EPaCCS. In general, patients and healthcare professionals felt that the system worked well the majority of the time, and that EPaCCS helped patients avoid repeating information. The evaluation compared aggregate data on place of death and hospital admissions in CCGs with and without EPaCCS, and reported no difference in the proportion of people dying in their usual place of residence, or in the proportion of hospital deaths, hospital admissions or costs of care in the last year of life.41

ion. The evaluation compared aggregate data on place of death and hospital admissions in CCGs with and without EPaCCS, and reported no difference in the proportion of people dying in their usual place of residence, or in the proportion of hospital deaths, hospital admissions or costs of care in the last year of life.41 Discussion This systematic review of the state of the science of EPaCCS shows that most of the current evidence comes from expert opinion and observational studies, and that there is an absence of any studies with an experimental design evaluating the impact of EPaCCS. While observational studies focused on place of death, hospital utilisation and use of EPaCCS, qualitative studies were mainly focused on the challenges of EPaCCS design and implementation.

nd observational studies, and that there is an absence of any studies with an experimental design evaluating the impact of EPaCCS. While observational studies focused on place of death, hospital utilisation and use of EPaCCS, qualitative studies were mainly focused on the challenges of EPaCCS design and implementation. The quantitative studies identified were either retrospective or cross-sectional in design, and focused on processes of care before death: the place of death of people with an EPaCCS record, the frequency of hospital admission and EPaCCS use. The proportion of patients with an EPaCCS record who died in their preferred place varied from 55% to 79%,6 7 29 30 which was noted to be higher than average for the population. The highest level of evidence came from two retrospective cohort studies that compared people with and without EPaCCS records, and found EPaCCS use was associated with lower odds of hospital death, hospital admission and emergency department attendance.28 34 Both these studies note important limitations, including lack of information on potential confounders such as comorbidities and stage of disease. While EPaCCS might contribute to an increase in the number of people who die at home, it is also possible that individuals who are included in these records are those who are actively seeking, and therefore are more likely, to die at home. It is not possible to rule out this selection bias in the group of people who have EPaCCS records due to the lack of randomisation within the studies. Indeed, the mixed-methods study by Wye et al explored the relationship between EPaCCS and home death and found that EPaCCS were almost exclusively used by GPs and community nurses, which might explain their strong association with home death.37

ho have EPaCCS records due to the lack of randomisation within the studies. Indeed, the mixed-methods study by Wye et al explored the relationship between EPaCCS and home death and found that EPaCCS were almost exclusively used by GPs and community nurses, which might explain their strong association with home death.37 The qualitative studies identified, as well as some of the reports, suggest that EPaCCS are acceptable for patients and healthcare professionals.35 36 39 41 However, there were some discrepancies between perceptions of benefits and utility expressed by different health professionals. In-hours staff perceived EPaCCS as a potential burden due to an increased workload without perceivable benefit to them, while out-of-hours staff perceived EPaCCS to be more useful.37 Although patients’ perspectives and opinions about EPaCCS were included in two of the articles,35 36 the majority of the results and recommendations from these studies were based on healthcare professionals’ views.35–37 More insight into patient and families’ experiences of EPaCCS is needed.

EPaCCS to be more useful.37 Although patients’ perspectives and opinions about EPaCCS were included in two of the articles,35 36 the majority of the results and recommendations from these studies were based on healthcare professionals’ views.35–37 More insight into patient and families’ experiences of EPaCCS is needed. We found a dearth of information on patient outcomes or quality of life measures. None of the studies we identified explicitly explored the potential harms of using EPaCCS, though from the healthcare professional’s perspective some articles discussed the burden of inputting data, and data-sharing issues, as a potential problem.35 37 From the patient’s perspective, Wye et al 37 suggest that once created, records are infrequently updated and it is unclear if this can have an impact on patients who change their preferences as their disease progresses. There is considerable evidence that suggests people prefer to die at home.42 However, some hospital admissions near the end of life are appropriate, particularly where prognosis is uncertain. The potential for harm through avoidance of appropriate hospital care should be explored.

eferences as their disease progresses. There is considerable evidence that suggests people prefer to die at home.42 However, some hospital admissions near the end of life are appropriate, particularly where prognosis is uncertain. The potential for harm through avoidance of appropriate hospital care should be explored. Two reports explored the economic impact of EPaCCS, with conflicting results.40 41 Based on a before–after design using area-level data, the NHS Improving Quality report found implementation of EPaCCS was associated with a reduction in hospital deaths that could potentially generate savings of £399–£1480 per hospital death avoided. In contrast, the Whole Systems Partnership compared data on place of death and hospital admissions in CCGs with and without EPaCCS, and found no difference in the proportion of people dying in their usual place of residence, or in the proportion of hospital deaths, hospital admissions or costs of care in the last year of life. Both studies had limitations, including a lack of adjustment for confounders and consideration of only the direct costs involved. Further studies of the economic impact of EPaCCS are needed.

usual place of residence, or in the proportion of hospital deaths, hospital admissions or costs of care in the last year of life. Both studies had limitations, including a lack of adjustment for confounders and consideration of only the direct costs involved. Further studies of the economic impact of EPaCCS are needed. Strengths and limitations Our study was motivated by the increasing political focus on EPaCCS as a tool to improve end-of-life care in the UK, and the lack of clarity regarding the underlying evidence base for these policies. In the preliminary searches, we found that the non-UK literature described tools where coordination between healthcare settings was not the primary focus. Therefore, we included only articles describing UK EPaCCS. Research from other settings might provide relevant information to guide implementation and use of EPaCCS. While our review necessarily included only UK data, there are important implications more broadly in terms of understanding the evidence base for policy interventions. We did not involve directly an information specialist in the search design, as is suggested for systematic reviews, which could potentially have affected the sensitivity of our search strategy. We searched a limited number of databases, which might have led to exclusion of relevant articles. However, it is unlikely that any study with an experimental design could have been missed.

arch design, as is suggested for systematic reviews, which could potentially have affected the sensitivity of our search strategy. We searched a limited number of databases, which might have led to exclusion of relevant articles. However, it is unlikely that any study with an experimental design could have been missed. As our aim was to understand the state of the science of EPaCCS, we included all relevant published articles. This inclusive approach contributes to the strengths of our study, for example, by highlighting that 14 of 26 articles published are expert opinion, and a dearth of information on patient outcomes. We used the Standard Quality Assessment Criteria to appraise the articles included.11 Appraisal of both qualitative and quantitative studies using a common tool is challenging, we addressed this by using two authors to independently rate the studies and discussion with a third where there were disagreements. The inclusion of government and independent reports in this review adds to its depth, and highlights the fact that much evidence for EPaCCS is located within grey literature. Other areas of grey literature, for example, unpublished data, were not systematically identified.

n with a third where there were disagreements. The inclusion of government and independent reports in this review adds to its depth, and highlights the fact that much evidence for EPaCCS is located within grey literature. Other areas of grey literature, for example, unpublished data, were not systematically identified. Implications While innovation to improve the quality of care received by patients is highly desirable, widely promoting interventions in the absence of strong supporting evidence can be dangerous. Logical and well-intentioned policy recommendations can do more harm than good.43 The studies described in this systematic review highlight the important potential benefits of EPaCCS for improving the end-of-life care. However, observational studies can overestimate the effect of interventions,44 and the lack of strong evidence is of concern. All interventions may have benefits and harms, some of which are more predictable than others.45 More and urgent research is needed in order to fully understand the benefits and potential harms of EPaCCS. Evaluations using experimental or quasi-experimental designs are needed, and these should be carried out alongside further qualitative studies to understand what needs to be in place to maximise benefits and avoid harms.46 Given the current drive for national roll-out of EPaCCS by 2020, it is essential that rigorous evaluation of EPaCCS is prioritised. Twitter: @kesleeman

Implications While innovation to improve the quality of care received by patients is highly desirable, widely promoting interventions in the absence of strong supporting evidence can be dangerous. Logical and well-intentioned policy recommendations can do more harm than good.43 The studies described in this systematic review highlight the important potential benefits of EPaCCS for improving the end-of-life care. However, observational studies can overestimate the effect of interventions,44 and the lack of strong evidence is of concern. All interventions may have benefits and harms, some of which are more predictable than others.45 More and urgent research is needed in order to fully understand the benefits and potential harms of EPaCCS. Evaluations using experimental or quasi-experimental designs are needed, and these should be carried out alongside further qualitative studies to understand what needs to be in place to maximise benefits and avoid harms.46 Given the current drive for national roll-out of EPaCCS by 2020, it is essential that rigorous evaluation of EPaCCS is prioritised. Twitter: @kesleeman Contributors: KES and IJH had the idea for the study. KES designed the study with input from JL, AW and IJH. Data search, screening, extraction and analysis were carried out by JL, KES and AW. All authors helped interpret the data. JL and KES wrote the first draft of the paper. All authors contributed to subsequent drafts and approved the final paper.

idea for the study. KES designed the study with input from JL, AW and IJH. Data search, screening, extraction and analysis were carried out by JL, KES and AW. All authors helped interpret the data. JL and KES wrote the first draft of the paper. All authors contributed to subsequent drafts and approved the final paper. Funding: KES is funded by a National Institute for Health Research (NIHR) Clinician Scientist Fellowship (CS-2015-15-005) and has received funding from the Academy of Medical Sciences (Starter Grant for Clinical Lecturers (AMS-SGCL10-Sleeman), supported by The Wellcome Trust, Medical Research Council, British Heart Foundation, Arthritis Research UK, the Royal College of Physicians and Diabetes UK). IJH holds an NIHR Emeritus Senior Investigator Award. AW is funded jointly by RM Partners, ULCH Cancer Collaborative, South East London ACN, Imperial BRC and RMH/ICR BRC. This paper presents independent research funded by the NIHR. Disclaimer: The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care. Competing interests: None declared. Patient consent for publication: Not required. Provenance and peer review: Not commissioned; externally peer reviewed. Data availability statement: The full dataset of papers is available from the corresponding author.

a mean value of expressing preferred place of death 86 days before dying. This means that hospice staff manage to identify and complete ACP discussions on average 3 months before people die. Prediction of prognosis is notoriously difficult. It may be that the advent of ACP discussions is a marker for terminal decline. Table 1 Descriptive characteristics of cohort, comparing cancer and non-cancer diagnoses and presence or absence of ACP Female Male People Count Mean age SD Count Mean age SD Count Mean age SD Cancer ACP 223 74.9 13.0 250 75.0 11.4 473 74.9 12.2 No ACP 174 73.7 12.9 192 73.8 12.0 366 73.7 12.4 Total 397 74.3 12.9 442 74.4 11.6 839 74.4 12.3 Non-cancer ACP 36 81.8 10.1 38 73.7 9.5 74 77.6 10.5 No ACP 35 80.2 11.1 21 80.2 9.4 56 80.2 10.4 Total 71 81.0 10.5 59 76.0 9.9 130 78.7 10.5 ACP 259 75.8 12.8 288 74.8 11.2 547 75.3 12.0 No ACP 209 74.7 12.8 213 74.4 11.9 422 74.6 12.3 Grand Total 468 75.4 12.8 501 74.6 11.5 969 75.0 12.1 ACP, advance care planning. Table 2 Proportion of patients achieving their preferred place of death and percentage hospital deaths Number Achieved preferred place of death Percentage Number who died in hospital Percentage who died in hospital Care home 115 110 96 2 2 Home 302 204 68 34 11 Hospice 125 93 74 21 17 Hospital 7 6 86 6 86 Overall 549 413 75 63 11 Table 3 Proportion of choices for preferred place of death for cancer and non-cancer patients

Introduction Advance care planning (ACP) is a fundamental part of the End of Life Care Strategy published by the Department of Health in 2008.1 However, the empirical evidence of the benefits of ACP is limited. An exploratory study of the acceptability of the use of ACP conducted by Jones et al2 concluded that patients with advanced cancer found that discussions were acceptable and possible. A study by Munday et al3 observed that both general practitioners and nurses found ACP discussions difficult, particularly if they were not initiated by the patient. Empirical studies, whether prospective or retrospective, assessing the impact of ACP on place of death are limited. A systematic review by Gomes and Higginson4 of 58 studies looking at factors which influenced home deaths concluded that patient preference was important. Use of ACP, do not attempt resuscitation orders and the Liverpool Care Pathway for the Dying implemented in nursing homes in Scotland5 showed a reduction in the number of hospital deaths. However, the numbers included in these studies were small. A study of hospice notes by Holdsworth and Fisher6 found that preferred place of death was recorded in 58% of 294 patients, with congruence of preferred place of death and actual place of death being 52% for home, 86% for hospice and 50% for care home. Congruence between preferred and actual place of death varied from 30% to 91% in a review by Bell et al.7 There are no studies that report the impact of ACP on the number of days spent in hospital in the last year of life and the number of admissions to hospital in the last year.

86% for hospice and 50% for care home. Congruence between preferred and actual place of death varied from 30% to 91% in a review by Bell et al.7 There are no studies that report the impact of ACP on the number of days spent in hospital in the last year of life and the number of admissions to hospital in the last year. Figure 1 Actual place of death proportion with and without preferred place of death. ppd, preferred place of death. Figure 2 Choice of preferred place of death by proportion (by percentage) of patients completing advance care planning. We report on the outcome of the use of ACP over a 2.5-year period in a hospice in the south west of England. We conducted a retrospective cohort study of hospice patients. We looked to see if ACP could be a routine part of end of life care. We wanted to see if the use of ACP was associated with difference in place of death, time spent in hospital in the last year of life and cost of hospital care.

n the south west of England. We conducted a retrospective cohort study of hospice patients. We looked to see if ACP could be a routine part of end of life care. We wanted to see if the use of ACP was associated with difference in place of death, time spent in hospital in the last year of life and cost of hospital care. Methods The hospice has a catchment area of about 150 000 and is a mixture of rural countryside and towns. Approximately 700 patients are referred for specialist palliative care each year, which includes inpatient and outpatient services, visits from specialist palliative care community nurses at home and a day care centre. For the last 4 years, the hospice has been using a single document for ACP,8 ‘Planning Ahead’, which combines a modified version of the Preferred Priorities For Care document with a Putting Affairs In Order guide and an Advance Decision To Refuse Treatment document. ‘The Planning Ahead’ document was developed in response to requests from patients and their families to have a unified document for future care. It is part of the hospice assessment paperwork and is given out to patients and families at an appropriate time. Hospice practice has involved emphasising the use of ACP and ‘Planning Ahead’ has been accepted as the single unifying document across the local health community. The hospice uses an electronic patient record (EPR) for all clinical information, including recording the outcome of ACP discussions and actual place of death.

spice practice has involved emphasising the use of ACP and ‘Planning Ahead’ has been accepted as the single unifying document across the local health community. The hospice uses an electronic patient record (EPR) for all clinical information, including recording the outcome of ACP discussions and actual place of death. All the statistics were calculated using Stata 10.1 (Stata Corp 2009) using t tests (two-sided) to compare means. Logistic regression was done with the Logistic command in the Stata statistics package.

spice practice has involved emphasising the use of ACP and ‘Planning Ahead’ has been accepted as the single unifying document across the local health community. The hospice uses an electronic patient record (EPR) for all clinical information, including recording the outcome of ACP discussions and actual place of death. All the statistics were calculated using Stata 10.1 (Stata Corp 2009) using t tests (two-sided) to compare means. Logistic regression was done with the Logistic command in the Stata statistics package. Study design We conducted a retrospective observational cohort study of all patients who were known to the hospice who died between 01 January 2009 and 30 June 2011. All the patients had a life limiting disease and were referred to the hospice for specialist palliative care. The hospice uses an EPR for all clinical notes. Patient preferred place of death, actual place of death and reason for variance are recorded routinely. Use of ACP was defined by data entry in these fields. Although ACP is more complex than this, for the purpose of analysis, we used presence or absence of preferred place of death as a marker of whether ACP discussions had taken place. The hospice community nurses remain in contact with the patient up until death and then with the carers into bereavement. In addition, the hospice nurses inform the hospital team of any admission to hospital of a hospice patient. Likewise, deaths and discharges are passed on to the community by the hospital palliative care team. Each death is discussed at our multidisciplinary meeting, so presence or absence of ACP expressed at any time is reviewed. In addition, each relative receives a contact from our specialist palliative care community nurses to review both death and bereavement. We are therefore confident that, for the cohort of patients in this study, we were able to say whether ACP had taken place or not. Place of death was defined as being in the patient's home, in a care home (either nursing or residential), in the hospice or in hospital (acute general hospital or community hospital). Absence of ACP was defined as the patient not having expressed a preferred place of death. Reason for variance was a free text entry. The Secondary User Services database was used to match patient identifying information to find the number of days in hospital each patient spent in the last year of life. This included both day case and inpatient stays. Day case hospital stays were included as chemotherapy is often given on the oncology day unit. The cost figures were actual costs adjusted for length of stay and complexity of care, as per national agreement.

days in hospital each patient spent in the last year of life. This included both day case and inpatient stays. Day case hospital stays were included as chemotherapy is often given on the oncology day unit. The cost figures were actual costs adjusted for length of stay and complexity of care, as per national agreement. All patient data used in this study were handled and processed in accordance with National Health Service (NHS) best practice and Caldicott recommendations. Although the Data Protection Act does not specifically apply to deceased patient records, any common law duty was adhered to. Results Descriptive characteristics for the cohort are shown in table 1. The initial cohort included 971 deaths. Two were excluded as no actual place of death was recorded. When doing the analyses looking at the impact of ACP on place of death, 969 cases were used. To look at the impact of ACP on number of days in hospital and number of hospital admissions we had a truncated dataset. The following patients were excluded: 217: Non-residents who lived outside of Primary Care Trust catchment area (because admission data were not available) 85: Missing NHS number (link to hospital admissions not possible) 9: Recorded NHS number inconsistent with Cancer Register (SWPHO) and/or Primary Care Mortality Database (ONS)—(unsafe match with hospital admission data) 2: Missing place of death. The total exclusions were 313 leaving 656 for analysis.

Results Descriptive characteristics for the cohort are shown in table 1. The initial cohort included 971 deaths. Two were excluded as no actual place of death was recorded. When doing the analyses looking at the impact of ACP on place of death, 969 cases were used. To look at the impact of ACP on number of days in hospital and number of hospital admissions we had a truncated dataset. The following patients were excluded: 217: Non-residents who lived outside of Primary Care Trust catchment area (because admission data were not available) 85: Missing NHS number (link to hospital admissions not possible) 9: Recorded NHS number inconsistent with Cancer Register (SWPHO) and/or Primary Care Mortality Database (ONS)—(unsafe match with hospital admission data) 2: Missing place of death. The total exclusions were 313 leaving 656 for analysis. Main results There were 501 male subjects, average age 75.4, and 468 female subjects, average age 74.6. The mean age was 75 (range 27–105) years at date of death. A total of 839 (87%) patients had a cancer diagnosis and 130 (13%) a non-cancer diagnosis. Of the 969 patients, 309 (32%) died at home, 265 (27%) died in the hospice, 227 (23%) died in a care home and 168 (17%) died in hospital. In all, 549 (56%) of patients completed ACP, 422 (44%) did not. During the first 6 months of the study period, 42% of patients completed ACP. During the last 6 months, this had increased to 69%. This figure had steadily increased, when looked at in 6-month periods. The mean number of days between completion of ACP and death was 86 (range 0–555 days) with 40% of people having done ACP within 100 days before dying and 75% within 364 days. Overall, 55% of female subjects completed ACP and 57% of male subjects. Mean age of female subjects completing ACP was 75.8 years and non-ACP was 74.7 years. Mean age of male subjects who completed ACP was 74.8 years and non-ACP was 74.4 years. Impact of ACP on place of death is shown in figure 1.

75% within 364 days. Overall, 55% of female subjects completed ACP and 57% of male subjects. Mean age of female subjects completing ACP was 75.8 years and non-ACP was 74.7 years. Mean age of male subjects who completed ACP was 74.8 years and non-ACP was 74.4 years. Impact of ACP on place of death is shown in figure 1. The overall percentage hospital death rate for patients who had completed ACP was 11% (table 2) Choice of preferred place of death is shown in figure 2. Table 3 compares choices for cancer and non cancer diagnoses. For those without ACP, this was 26%; p value<0.001. In all, 50 (10%) of the cancer patients who did ACP died in hospital and 94 (26%) of the cancer patients without ACP died in hospital; p value<0.001. Overall, 88% of people who had completed ACP had at least one emergency admission in the last year of life and 91% of people who had not done ACP had at least one emergency admission; p value 0.71. For people who died in hospital, the mean number of days spent in hospital in the last year was 26.5 days. For people who did not die in hospital, the mean number of days in hospital in the last year was 20.5 days. The mean cost of hospital care in the last year of life for those who died in hospital was £11 298. The mean cost of hospital care for those who died outside of hospital was £7730, with a difference of £3569; p value of <0.001 (table 4). The p value for the difference in the number of days in hospital for cancer patients (20.2 days) compared with non-cancer (30.6 days) was <0.001 (table 5).

For people who died in hospital, the mean number of days spent in hospital in the last year was 26.5 days. For people who did not die in hospital, the mean number of days in hospital in the last year was 20.5 days. The mean cost of hospital care in the last year of life for those who died in hospital was £11 298. The mean cost of hospital care for those who died outside of hospital was £7730, with a difference of £3569; p value of <0.001 (table 4). The p value for the difference in the number of days in hospital for cancer patients (20.2 days) compared with non-cancer (30.6 days) was <0.001 (table 5). Discussion This study shows that it is possible to make ACP a routine part of care for hospice patients. During the first 6 months of the study period, 42% of patients had completed ACP. This had increased to 69% during the final 6 months and in the subsequent 6 months following the study had further increased to 77%. ACP has been seen as an integral part of the work of the whole hospice and effort has been made to make its use part of normal care. The proportional impact of ACP on place of death remained constant throughout the study. The use of an EPR has meant that it has been possible to monitor the use of ACP and the impact on place of death on a monthly basis. The use of the EPR has made data collection and analysis a lot more straightforward. The monthly audit of the EPR has meant that data completeness has been very high, with only two of 971 records not having place of death recorded.

possible to monitor the use of ACP and the impact on place of death on a monthly basis. The use of the EPR has made data collection and analysis a lot more straightforward. The monthly audit of the EPR has meant that data completeness has been very high, with only two of 971 records not having place of death recorded. The use of ACP was correlated with a significant impact on place of death, with 75% of patients dying in their place of choice. Hospital death rates for people who had completed ACP were 11%. Without ACP this was 26%. This latter figure is still low in comparison with national averages, where the hospital death rate for England is 54%.9 This is likely to reflect in part the impact of specialist palliative care on place of death. Hospices have historically done well enabling patients to stay out of hospital even without ACP. The use of ACP helps further over and above the impact of specialist palliative care alone.

pital death rate for England is 54%.9 This is likely to reflect in part the impact of specialist palliative care on place of death. Hospices have historically done well enabling patients to stay out of hospital even without ACP. The use of ACP helps further over and above the impact of specialist palliative care alone. In our study, this approach appears to be justified in both the cancer and non-cancer settings. Specialist palliative care has become increasingly involved in non-cancer terminal illnesses. In all, 59% of the non-cancer patients used ACP. The overall hospital death rate in this group was 18%. It seems unlikely that the non-cancer group was representative of non-cancer terminal diagnoses in the population, as the numbers were relatively small. The majority of patients with non-cancer terminal diagnoses do not receive specialist palliative care. The referred group will have had particular needs that required the use of hospice services. Nevertheless, the results suggest that use of ACP is applicable in the non-cancer setting and this has a significant impact on place of death. What was notable was that none of the non-cancer patients chose to die in hospital.

The referred group will have had particular needs that required the use of hospice services. Nevertheless, the results suggest that use of ACP is applicable in the non-cancer setting and this has a significant impact on place of death. What was notable was that none of the non-cancer patients chose to die in hospital. Gomes and colleagues conducted an extensive study using telephone interviews to ask the general population about their preferences of place of death.10 In this study, depending on region, more than 60% of people would chose to die at home (range 60%–67%), 26%–32% would chose to die in a hospice, and the remaining would chose to die in a care home, hospital or elsewhere. In our study, which had 87% cancer patients, 55% chose to die at home, 23% chose the hospice, 21% chose a care home and 1% chose hospital. These latter figures are for patients who have a terminal illness and are making a decision about where they would like to die. Although there are some similarities in these figures, the differences are likely to be due to a combination of the reality of having a terminal illness as well as our study population all being known to the hospice.

for patients who have a terminal illness and are making a decision about where they would like to die. Although there are some similarities in these figures, the differences are likely to be due to a combination of the reality of having a terminal illness as well as our study population all being known to the hospice. The hospital death rate for cardiovascular cause of death for any mention on the death certificate for England is 60.5% and for respiratory disease is 67.6%.11 These results suggest that the more widespread use of ACP in the non-cancer setting could result in a shift of deaths from hospital into the community. In this study, only a small percentage of non-cancer patients chose to die in the hospice. It is hard to know if this result is generalisable. It has meant that the hospice has not been overwhelmed by large numbers of non-cancer patients wanting to die on the inpatient unit. The mean number of days spent in hospital in the last year of life for those patients who had completed ACP was 8.3 days less than those who had not. The overall cost of hospital admissions in the last year was £3569 less for those people who died outside of hospital. Both of these figures had highly significant p values. It seems likely that this figure is an underestimate of changes that would result in ACP being used more widely. The hospital death rate in the non-ACP group was 26%. The hospice is effective at reducing hospital death rates even when ACP has not been done. The national average is 54%.

had highly significant p values. It seems likely that this figure is an underestimate of changes that would result in ACP being used more widely. The hospital death rate in the non-ACP group was 26%. The hospice is effective at reducing hospital death rates even when ACP has not been done. The national average is 54%. The mean number of admissions in the last year was 5.4 in the ACP group and 6.1 in the non-ACP group. This latter figure did not have a significant p value. The mean number of emergency admissions in the group of people who died in hospital was significantly higher than those who died outside of hospital, although the total number of admissions in ACP verses non-ACP group was not significant. This may indicate that the final admission was an emergency one and could potentially be avoided.

of emergency admissions in the group of people who died in hospital was significantly higher than those who died outside of hospital, although the total number of admissions in ACP verses non-ACP group was not significant. This may indicate that the final admission was an emergency one and could potentially be avoided. A previous paper in our locality12 showed the mean cost of the final admission to be £3093 in 2006. The mean difference in costs in the last year of life is larger than this figure and suggests that potential reinvestment and savings opportunities exist based on the results from this study. Estimation of hospital costs is complicated by whether or not there was a difference in the use of chemotherapy in the ACP group compared with the non-ACP group. Similar numbers of hospital admissions, including day case, suggest that both ACP and non-ACP patients had similar amounts of chemotherapy. The oncology and palliative care services in our locality work closely together and shared care of patients is common as their disease progresses. This may indicate that decision making about the use of chemotherapy has already been optimised to reflect end of life care. Irrespective of ACP, death in hospital increased significantly the number of days spent in hospital in the last year of life, with increased mean costs of £11 299 compared with £7730 for death outside of hospital.

that decision making about the use of chemotherapy has already been optimised to reflect end of life care. Irrespective of ACP, death in hospital increased significantly the number of days spent in hospital in the last year of life, with increased mean costs of £11 299 compared with £7730 for death outside of hospital. The impact of the use of ACP on length of life is unknown. A study of lung cancer in the USA13 showed that those who accessed specialist palliative care lived longer and used less chemotherapy than those who did not. Chemotherapy use in the UK may well be different from that in the USA. The mean cost of admissions in the last year of life was £8310. This is slightly higher than the study published by Bardsley and coworkers,14 in which the mean hospital inpatient cost in the last year of life in the three sites studied was £6231, when taking into account day case costs. The number of days spent in hospital in the last year in the same study was 30 days, which differs considerably from our result of 21.5 days. These differences are likely to reflect that the current study was mainly of cancer patients.

life in the three sites studied was £6231, when taking into account day case costs. The number of days spent in hospital in the last year in the same study was 30 days, which differs considerably from our result of 21.5 days. These differences are likely to reflect that the current study was mainly of cancer patients. Although the number of patients in this study was large in comparison with previous studies, the use of a retrospective cohort design means that we were able to look at associations rather than elucidate causative relationships. A prospective randomised study is needed to be able to determine the relationships of ACP use, place of death and cost with more certainty. The study was able to demonstrate associations among use of ACP, reduced total days spent in hospital in the last year and reduced hospital costs for those people who died outside of hospital. Use of ACP is an integral part of the End of Life Care Strategy.1 If we are going to offer choices about place of death to people with terminal diagnoses, we need to be able to identify them and have the discussion around ACP with sufficient time. In all, 75% of patients who did ACP stated their wishes within the last year of life, with a mean value of expressing preferred place of death 86 days before dying. This means that hospice staff manage to identify and complete ACP discussions on average 3 months before people die. Prediction of prognosis is notoriously difficult. It may be that the advent of ACP discussions is a marker for terminal decline.

Number Achieved preferred place of death Percentage Number who died in hospital Percentage who died in hospital Care home 115 110 96 2 2 Home 302 204 68 34 11 Hospice 125 93 74 21 17 Hospital 7 6 86 6 86 Overall 549 413 75 63 11 Table 3 Proportion of choices for preferred place of death for cancer and non-cancer patients PPD Cancer % (numbers) Non-cancer % (numbers) Home 53 (253) 67 (49) Care home 20 (97) 27 (20) Hospice 26 (123) 6 (4) Hospital 1 (7) 0 A total of 8 (11%) patients with a non-cancer diagnosis who had completed ACP died in hospital. Of those who did not complete ACP, 16 (29%) died in hospital (p value 0.14). ACP, advance care planning; PPD, preferred place of death. Table 5 Impact of ACP on mean number of days in hospital in the last year of life, number of admissions and costs of emergency admissions in the last year of life (95%CI) People Mean stay for all those with an admission Mean stay for those with or without an admission Mean number of admissions Mean number of emergency admissions Mean cost of emergency admissions (£) ACP 389 20.7 (18.4 to 23.0) 18.1 (16.0 to 20.2) 4.8 (4.2 to 5.4) 1.61 (1.4 to 1.8) 5260.4 (4586 to 5934) No ACP 275 28.9 (25.1 to 32.8) 26.4 (22.8 to 30.0) 5.5 (4.5 to 6.4) 1.75 (1.6 to 1.9) 5690.9 (4984 to 6398) Total 664 24.2 (22.0 to 26.3) 21.5 (19.6 to 23.5) 5.1 (4.5 to 5.6) 1.66 (1.5 to 1.8) 5438.8 (4948 to 5929) t Test two-sided p value <0.001 <0.001 0.2 0.3 0.4 p Value compares ACP with non-ACP.

60.4 (4586 to 5934) No ACP 275 28.9 (25.1 to 32.8) 26.4 (22.8 to 30.0) 5.5 (4.5 to 6.4) 1.75 (1.6 to 1.9) 5690.9 (4984 to 6398) Total 664 24.2 (22.0 to 26.3) 21.5 (19.6 to 23.5) 5.1 (4.5 to 5.6) 1.66 (1.5 to 1.8) 5438.8 (4948 to 5929) t Test two-sided p value <0.001 <0.001 0.2 0.3 0.4 p Value compares ACP with non-ACP. Multivariate analysis investigating factors affecting the likelihood of death in hospital suggests that ACP is a significant factor. People who have had ACP reduce the odds of dying in hospital nearly 70% lower than people who do not. Gender, age or cause of death did not have a significant effect. ACP, advance care planning. Table 4 Death in hospital compared with death elsewhere (95% CI)

Multivariate analysis investigating factors affecting the likelihood of death in hospital suggests that ACP is a significant factor. People who have had ACP reduce the odds of dying in hospital nearly 70% lower than people who do not. Gender, age or cause of death did not have a significant effect. ACP, advance care planning. Table 4 Death in hospital compared with death elsewhere (95% CI) People Mean stay Mean number of admissions, all people Mean number of emergency admissions, all people Mean cost of all admissions, all people (£) Death in hospital 108 26.5 (20.8 to 32.2) 6.4 (4.9 to 7.9) 2.2 (1.9 to 2.5) 11 299 (9161 to 13 436) Death elsewhere 556 20.6 (18.5 to 22.7) 4.7 (4.3 to 5.4) 1.6 (1.4 to 1.7) 7730 (7079 to 8381) Total 664 21.5 (19.6 to 23.5) 5.1 (4.5 to 5.6) 1.7 (1.5 to 1.8) 8310 (7659 to 8962) t Test two-sided p value 0.03 0.03 <0.001 <0.001 Conclusions The assumptions made in the End of Life Care Strategy 2008 that more people when asked would choose to die outside of hospital, and that it is possible to look after people successfully in their place of choice, are strengthened by the results from this paper, particularly for patients with cancer. It remains to be seen as to whether the same principles are applicable in the non-cancer setting. The routine use of ACP has yet to be proven, particularly as the bulk of these discussions would need to be done by generalists in palliative care. Clare-Louise Nicholls and Anne Sumerhill from North Somerset PCT helped with data collection for days in hospital and costs. Competing interests: None.

People Mean stay Mean number of admissions, all people Mean number of emergency admissions, all people Mean cost of all admissions, all people (£) Death in hospital 108 26.5 (20.8 to 32.2) 6.4 (4.9 to 7.9) 2.2 (1.9 to 2.5) 11 299 (9161 to 13 436) Death elsewhere 556 20.6 (18.5 to 22.7) 4.7 (4.3 to 5.4) 1.6 (1.4 to 1.7) 7730 (7079 to 8381) Total 664 21.5 (19.6 to 23.5) 5.1 (4.5 to 5.6) 1.7 (1.5 to 1.8) 8310 (7659 to 8962) t Test two-sided p value 0.03 0.03 <0.001 <0.001 Conclusions The assumptions made in the End of Life Care Strategy 2008 that more people when asked would choose to die outside of hospital, and that it is possible to look after people successfully in their place of choice, are strengthened by the results from this paper, particularly for patients with cancer. It remains to be seen as to whether the same principles are applicable in the non-cancer setting. The routine use of ACP has yet to be proven, particularly as the bulk of these discussions would need to be done by generalists in palliative care. Clare-Louise Nicholls and Anne Sumerhill from North Somerset PCT helped with data collection for days in hospital and costs. Competing interests: None. Provenance and peer review: Not commissioned; externally peer reviewed.

People Mean stay Mean number of admissions, all people Mean number of emergency admissions, all people Mean cost of all admissions, all people (£) Death in hospital 108 26.5 (20.8 to 32.2) 6.4 (4.9 to 7.9) 2.2 (1.9 to 2.5) 11 299 (9161 to 13 436) Death elsewhere 556 20.6 (18.5 to 22.7) 4.7 (4.3 to 5.4) 1.6 (1.4 to 1.7) 7730 (7079 to 8381) Total 664 21.5 (19.6 to 23.5) 5.1 (4.5 to 5.6) 1.7 (1.5 to 1.8) 8310 (7659 to 8962) t Test two-sided p value 0.03 0.03 <0.001 <0.001 Conclusions The assumptions made in the End of Life Care Strategy 2008 that more people when asked would choose to die outside of hospital, and that it is possible to look after people successfully in their place of choice, are strengthened by the results from this paper, particularly for patients with cancer. It remains to be seen as to whether the same principles are applicable in the non-cancer setting. The routine use of ACP has yet to be proven, particularly as the bulk of these discussions would need to be done by generalists in palliative care. Clare-Louise Nicholls and Anne Sumerhill from North Somerset PCT helped with data collection for days in hospital and costs. Competing interests: None. Provenance and peer review: Not commissioned; externally peer reviewed. Open Access: This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 3.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/3.0/

Conceptualisations of death and dying, and emotional expressions of grief and loss, are largely socially constructed. Therefore, the ability to adequately understand cultural values and practices is essential when providing supportive and palliative care.1 Recognition of an individual's needs, values and beliefs within a cultural context supports the autonomy and human rights of the persons receiving care.2 Cultural differences in attitudes toward providing comfort, truth-telling, decision-making including family roles, understanding of the meaning of the end of this physical life, and using indigenous practices and rituals reinforce the need for cultural understanding.2 Cultural understanding leads to cultural competence as healthcare professionals begin to recognise the interests, emotions and biases that are a part of their own culture,3 since culture is only meaningful when interpreted in the context of an individual's belief and value system.4 The more opportunity that care providers have to build meaningful relationships with persons approaching the end of this life, particularly with persons whose culture differs from one's own, the greater the likelihood that persons will receive comprehensive and compassionate care.

individual's belief and value system.4 The more opportunity that care providers have to build meaningful relationships with persons approaching the end of this life, particularly with persons whose culture differs from one's own, the greater the likelihood that persons will receive comprehensive and compassionate care. In this paper, Mi'kmaq understandings and recommendations are presented for consideration to enhance culturally competent supportive and palliative care. In Canada, indigenous peoples include the Inuit, Metis and status and non-status First Nations. They are collectively known as the Aboriginal peoples. In 2006, there were an estimated 24 175 Aboriginal persons living in the province of Nova Scotia, which is 2.6% of the population of the province.5 The Mi'kmaq Nation lives in Atlantic Canada as well as New England in the eastern USA. The majority of Nova Scotia's Aboriginal population belong to the Mi'kmaq Nation and reside in 13 First Nations communities. Mi'kmaq is the collective noun of the singular form and adjective Mi'kmaw. Mi'kmaq perspectives of support and healthcare as they relate to the provision of supportive and palliative care are identified and discussed both from a historical point of view as well as in the current context of care. Tribal differences withstanding, indigenous peoples report that indigenous knowledge is often not understood or is considered only through the lens of Western healthcare practice.6 Furthermore, there is now a recognition that non-indigenous society can learn much from indigenous community-based practice at the end of life.7

l differences withstanding, indigenous peoples report that indigenous knowledge is often not understood or is considered only through the lens of Western healthcare practice.6 Furthermore, there is now a recognition that non-indigenous society can learn much from indigenous community-based practice at the end of life.7 This paper provides a synthesis of traditional views of life and death of the Mi'kmaq in the context of the inter-relationship between the collective community and personal beliefs and values. Although the context is that of the Mi'kmaq, cultural understanding should be considered with all indigenous cultures and for non-indigenous peoples. This manuscript provides non-indigenous healthcare providers and researchers with a situation for reflection on cultural understanding in the provision of care. Methods This paper was developed from a review of literature and discussions of one author (SP) with four non-Mi'kmaw informants intricately involved with research and Mi'kmaq peoples and the second author (AV) who shared this paper with two Mi'kmaw consultants (Theresa Meuse, advisor and educator on Mi'kmaq ways, and Sister Veronica Matthews, Elder). This study was carried out in the context of respectful Aboriginal ways which value relationships and begins by attaining an understanding of the context and history.8

thor (AV) who shared this paper with two Mi'kmaw consultants (Theresa Meuse, advisor and educator on Mi'kmaq ways, and Sister Veronica Matthews, Elder). This study was carried out in the context of respectful Aboriginal ways which value relationships and begins by attaining an understanding of the context and history.8 The authors are not from the Mi'kmaq Nation. GJ leads research to improve access to quality care for populations at end of life. AV practiced as a nurse with the Aboriginal people and has taught northern nurse practitioners. SP was a graduate student with GJ and AV. SP worked under the direction of AV in meeting one-to-one with each of the non-Mi'kmaw informants whose research is recognised and valued by the Mi'kmaw people but who do not have the authority to speak on behalf of the Mi'kmaw people. Therefore, ensuring the authority to speak of Mi'kmaw ways was essential. AV has worked with two Mi'kmaw consultants (TM, SVM) for more than 20 years. TM and SVM are recognised spokespersons for the Mi'kmaq community particularly in relation to health issues. TM and SVM reviewed the paper prior to submission for publication, and they reviewed and approved revisions in response to journal reviewer comments. Table 1 provides a description of key Mi'kmaq terms, as well as cultural competency, and supportive and palliative care as used in Nova Scotia. Throughout the process of developing this paper, there was an ongoing need to specify and clarify the meaning of terms. Table 1 Descriptions of terms related to providing Mi'kmaq supportive and palliative care

Table 1 provides a description of key Mi'kmaq terms, as well as cultural competency, and supportive and palliative care as used in Nova Scotia. Throughout the process of developing this paper, there was an ongoing need to specify and clarify the meaning of terms. Table 1 Descriptions of terms related to providing Mi'kmaq supportive and palliative care Term Description Aboriginal peoples Indigenous people who have lived in Canada for thousands of years. Mi'kmaq First Nation people whose name roughly translates as ‘the family’. Oral tradition is important in the relational nature of the Mi'kmaq language. Apiksiktatultimk (pronounced abiksikdadldimk) A Mi'kmaq term used to describe when a person is thought to be dying, family and friends go to the bedside to partake in this act of mutually being present with each other which may include forgiveness or reconciliation. This has the intent of ensuring that the dying person will go to the spirit world without any burden while also preparing all involved for the inevitable. Nemu'ltus A commonly used Mi'kmaq saying which translates to ‘I'll see you.’ It is mostly used as a form of goodbye, but is also used when someone is dying. The implication here is that death is not final. Life and death are events. Death is understood to be a verb since if it were a noun, it would be final. Life is also a verb as it is a process of living and of being alive. Salite Mi'kmaq feast after a person leaves this world. Feasts are common for First Nations peoples for closure after events. Usually the Elders are the first to begin the meal. This is a community event with everyone bringing goods for an auction. In the past, the Salite auctioned the belongings of the person who has gone to the spirit world. Today, community members bring goods to auction to help raise funds to pay for the wake and other expenses and debts the person may have had. Salite is an important aspect of care. It acknowledges the interconnectedness of the person who has moved on to the spirit world and the community. Cultural competency As a continuum, cultural competency encompasses sensitivity, awareness, knowledge, skills and competency. Cultural competency depends on provider efforts to understand unique aspects of individual cultures and is based on the premise that knowledge of cultural characteristics equips professionals to build trusting relationships and provide holistic care.

encompasses sensitivity, awareness, knowledge, skills and competency. Cultural competency depends on provider efforts to understand unique aspects of individual cultures and is based on the premise that knowledge of cultural characteristics equips professionals to build trusting relationships and provide holistic care. Palliative care Palliative care aims to improve the quality of life and care of patients and their families facing life-threatening illness through the prevention and relief of suffering by early intervention and treatment of pain and other physical, psychosocial and spiritual problems. In Canada, palliative care and palliative medicine are typically hospital-based and provincially organised. Palliative care developed in Canada during a time period when federal-provincial agreements provided public funding for hospital and physician care, and patchwork funding at best for community-based care. The role of the voluntary sector and the federal government is minimal in the delivery of palliative care in Canada. Supportive care Supportive care relates primarily to cancer services that help patients and families with their experience during the diagnostic, treatment, follow-up and palliative phases. It includes physical and symptom support as well as information, psychological and spiritual support. Indian and Northern affairs INAC is the Canadian federal government department with a mandate to fulfil constitutional responsibilities in First Nations affairs regarding the Indian Act which was established in 1867. Since 1949, matters of healthcare and First Nations have been directed to the federal department of Health Canada. INAC's name was changed to AANDC in 23 June 2011. AANDC is responsible for economic development and social well-being. AANDC, Aboriginal Affairs and Northern Development Canada; INAC, Indian and Northern Affairs Canada.

tters of healthcare and First Nations have been directed to the federal department of Health Canada. INAC's name was changed to AANDC in 23 June 2011. AANDC is responsible for economic development and social well-being. AANDC, Aboriginal Affairs and Northern Development Canada; INAC, Indian and Northern Affairs Canada. Results The themes that emerged are interconnected and grounded in the historic Mi'kmaq context of colonialism. The role of colonialism is included in table 2 but not developed further because it has been well documented by Twohig.9 The second theme is jurisdictional issues which relate to the three systems of support for the Mi'kmaq: federal, provincial and traditional indigenous. Third, cultural understanding is needed in relation to both values and language. Table 2 Themes emerging from a review of literature and discussion with informants on culturally competent supportive and palliative care for the Mi'kmaq

Results The themes that emerged are interconnected and grounded in the historic Mi'kmaq context of colonialism. The role of colonialism is included in table 2 but not developed further because it has been well documented by Twohig.9 The second theme is jurisdictional issues which relate to the three systems of support for the Mi'kmaq: federal, provincial and traditional indigenous. Third, cultural understanding is needed in relation to both values and language. Table 2 Themes emerging from a review of literature and discussion with informants on culturally competent supportive and palliative care for the Mi'kmaq Theme Description Historic context of colonisation It is critically important to recognise the role of colonialism in shaping policy and its impact on what services are available to Canada's First Nation peoples. Unlike other parts of the country, there has been no land surrender in Nova Scotia. Jurisdictional issues Continuity of care may be lost as individuals require care outside their community and move between provincial and federal services. Cultural understanding Values: lack of alignment between values implicit in Western medicine and those of the Mi'kmaq culture can be problematic. Using a holistic health approach can act as a bridge for understanding indigenous values such as the interconnectedness of the physical, social and spiritual dimensions of self and the impact this has on health decisions. Rather than being universal and unchanging, values differ, often subtly, by family and individual as well as from community to community. Language: acknowledging culture as a part of an individual's healing process is vital to understanding health choices and preferences. The integration of spirit as a part of healing occurs through the inseparable interrelationships among self, family and community. Different world views, language and cultural contexts reflect the divide that needs to be bridged to attain cultural competency in the care of indigenous peoples at end of life. Complexity is inherent in gaining a deep understanding of language. Jurisdictional issues in providing palliative and supportive care for the Mi'kmaq The Canadian federal government has responsibility for the Aboriginal peoples. However, the delivery of health services in Canada is primarily a provincial responsibility. There are gaps and differences in services that are funded by each jurisdiction.

issues in providing palliative and supportive care for the Mi'kmaq The Canadian federal government has responsibility for the Aboriginal peoples. However, the delivery of health services in Canada is primarily a provincial responsibility. There are gaps and differences in services that are funded by each jurisdiction. Supportive10 11 and palliative care have not been core services of Indian and Northern Affairs Canada (INAC), the department which has the responsibility of meeting the Government of Canada's obligations for First Nations peoples through negotiations with them. Most Mi'kmaq communities have community health nurse(s) who are involved with families in providing care. Health Canada and INAC together developed a National Framework on Home Care for First Nations and Inuit communities using this definition: ‘a system of service delivery encompassing a range of insured, extended and uninsured health and social services for all age groups, addressing the wholistic, social and personal care needs of individuals who do not have, or have lost, some capacity for self-care.’12 The INAC adult care programme can meet some supportive and palliative care needs. INAC may pay to provide care services including laundry, meals and housekeeping. To respect client preferences, family members may provide these services. Personal care can include pain management, spiritual support and comfort measures, familial support and care planning related to disease as well as the settling of personal affairs. Personal care workers funded by INAC have some training in personal care.

t client preferences, family members may provide these services. Personal care can include pain management, spiritual support and comfort measures, familial support and care planning related to disease as well as the settling of personal affairs. Personal care workers funded by INAC have some training in personal care. The lack of extensive educational and employment opportunities for Aboriginal care givers reinforces marginalisation and results in female Aboriginal caregivers being disadvantaged socially and politically.13 Other issues include coping with the stress of physical illness, the role of the family in treatment and decision-making, financial burdens, transportation to health services and provision of support for emotional and spiritual health. A lack of quality housing is also recognised as a fundamental challenge in providing adequate home-based care.12

with the stress of physical illness, the role of the family in treatment and decision-making, financial burdens, transportation to health services and provision of support for emotional and spiritual health. A lack of quality housing is also recognised as a fundamental challenge in providing adequate home-based care.12 The Mi'kmaq population is primarily rural and small. Therefore, choices are limited. Beyond INAC and Health Canada services, some communities have a relationship with a provincially funded palliative care programme to assist in managing pain and other symptoms. Each of the nine district health authorities across Nova Scotia operates a palliative care programme. Their evolution is comparable with that in other Canadian provinces including greater gaps in rural than in urban areas.14 Their extent of services varies as does their linkage to supportive cancer care15 and to the provincial home care programme. While the Nova Scotia Department of Health and Wellness recognises the importance of cultural competency in the provision of healthcare,16 concerns remain. Jurisdictional issues are compounded by Treaty and non-Treaty status and First Nations people living on and off reserve. Issues include transportation to and from treatment areas and associated costs and financial restrictions for accessing certain services. Costs may depend on the length of time individuals are away from home accessing services. Feelings of loneliness and isolation result from unfamiliar environments and contribute to social exclusion.17

nsportation to and from treatment areas and associated costs and financial restrictions for accessing certain services. Costs may depend on the length of time individuals are away from home accessing services. Feelings of loneliness and isolation result from unfamiliar environments and contribute to social exclusion.17 A government report16 states that provincial hospital services are poorly responsive to Aboriginal healing practices. Difficulty in recruiting and retaining on-reserve physicians and poor relationships with off-reserve provincial health professionals are issues that the Mi'kmaq representatives identified. Even though it may appear that increased funding is being committed to improving Aboriginal health, the proportion per capita is still much less than what is spent per person in the rest of Canada.18 Disparities in access, quality care and health outcomes as well as provider biases, lack of continuity of care, healthcare system barriers and limited interpretation services are similar to reports for indigenous peoples elsewhere in North America.12 Jurisdictional issues experienced by the Mi'kmaq affect access, continuity and appropriateness of care. Many of these issues are connected with the impact of the historic residential school system which did not value Mi'kmaq ways, created intergenerational trauma, and led to disruption of culture, language, traditions and history. This parallels a bias that has been reported for some primary healthcare providers who appear disrespectful and have a poor knowledge or understanding of First Nations culture and misconceptions around ‘compliance’.16 Great imbalances of power exist in terms of the degree of decision-making that is enabled within the Aboriginal communities. Table 3 provides recommendations for practice at the level of the healthcare system and the individual requiring care.

rstanding of First Nations culture and misconceptions around ‘compliance’.16 Great imbalances of power exist in terms of the degree of decision-making that is enabled within the Aboriginal communities. Table 3 provides recommendations for practice at the level of the healthcare system and the individual requiring care. Table 3 Recommendations for non-indigenous care providing supportive and palliative care

rstanding of First Nations culture and misconceptions around ‘compliance’.16 Great imbalances of power exist in terms of the degree of decision-making that is enabled within the Aboriginal communities. Table 3 provides recommendations for practice at the level of the healthcare system and the individual requiring care. Table 3 Recommendations for non-indigenous care providing supportive and palliative care I. The systems: learn history of indigenous people and relevance to healthcare today (A) Be aware and adapt to the impact of the historic context of colonialism. Even though practices may differ today, pain from the past reverberates and retains a resonance that must be recognised. Colonialism shaped political relationships and resources and led to discrimination and injustices including lack of respect and unequal access to care. (B) Recognise the gaps and issues in the provision of indigenous health services. While not a main focus of this paper, addressing health system problems is essential for the provision of culturally competent supportive and palliative care. The need to lobby to improve access, quality, comprehensiveness and continuity of care is apparent. II. The individual: gain insight into the views of the person (A) Make an effort to get to know the individual. Cultural needs must be considered on an individual basis, not according to cultural assumptions or biases. Individuals are unique, even within their cultures. Dignity is individually determined. Providers should take every opportunity to build a relationship with the patient/client to enable them provide competent care. Cultural knowledge and awareness is one step in the continuum of providing quality care. (B) Treat each situation as unique and case-specific. A great opportunity for attaining cultural understanding is remembering that even specialised healthcare providers can expand their awareness of other cultures and learn from those they serve. Biomedical values are not always reconcilable with other cultural values. Sensitivity should be given to individual's previous experience with the medical system and models for achieving health. (C) Recognise the value placed on trust and respect. Sincerity is recognised as an emotional strength. Focused communication, including listening and noting responses, strengthens a relationship built on trust and respect. If trust is given to caregivers by a Mi'kmaw as a way of respecting their role, the trust should be protected, never abused or disregarded. (D) Be aware of non-verbal communication and cues for discussion.

ed communication, including listening and noting responses, strengthens a relationship built on trust and respect. If trust is given to caregivers by a Mi'kmaw as a way of respecting their role, the trust should be protected, never abused or disregarded. (D) Be aware of non-verbal communication and cues for discussion. Minimal eye contact may not indicate discomfort. Care providers can share information to encourage discussion rather than directly asking personal, intrusive questions. (E) Consider the role of religious beliefs and spirituality. Spirituality, in particular feelings toward life, death and afterlife, hold varying degrees of importance. When traditional views of going to the spirit world are identified, providers should recognise that death is viewed as a transition from one state to another and not with the same finality as other worldviews concerning death. Spiritual and religious beliefs play an important role in framing processes and choices relating to values enacted at this time. Cultural understanding in the provision of supportive and palliative care Values The lack of alignment between the values implicit in Western medicine and indigenous peoples is problematic. Cultural understanding can act as a bridge to accepting Aboriginal values such as the interconnectedness of the physical, social and spiritual dimensions of self and the impact this has on health decisions. Acknowledging culture as a part of an individual's healing process is vital to comprehending health choices and preferences. The integration of spirit as a part of healing occurs through the inseparable interrelationships among self, family and community.19

imensions of self and the impact this has on health decisions. Acknowledging culture as a part of an individual's healing process is vital to comprehending health choices and preferences. The integration of spirit as a part of healing occurs through the inseparable interrelationships among self, family and community.19 The Mi'kmaq community approaches life in the context of a worldview which may include traditional medicine and an interconnectedness of physical, mental, spiritual and environmental well-being. Ruth Whitehead described the family in pre-European contact times as ‘the centre of the universe’ and emphasised the role of alliances created by birth, marriage or adoption as being priorities in life. The collective mindset that frames kinship is still a cultural imperative for the Mi'kmaq and characterises the role of family and community in a Mi'kmaq worldview. Consequently, family and community play a predominant role in ethical decision-making and confidentiality. The preferred process for leaving the physical world is in the company of loved ones and choices are made with the family either involved or kept in mind.

s the role of family and community in a Mi'kmaq worldview. Consequently, family and community play a predominant role in ethical decision-making and confidentiality. The preferred process for leaving the physical world is in the company of loved ones and choices are made with the family either involved or kept in mind. Community preparation and participation in the journey to the spirit world is important. Wakes and funerals are commonly accepted as social gatherings where people converse openly, laugh, share stories and celebrate life which supports the tradition of strong family and community connectedness. Healing or prayer circles as well as ceremonial sweats may be held when it is anticipated that the person's journey to the spirit world is eminent. According to Angela Robinson,19 the tenet, ‘No one is born alone and no one should die alone,’ is central to Mi'kmaq understanding of not going through this journey alone. There is a community of faith that binds individuals to one another as well as to a greater collective.

on's journey to the spirit world is eminent. According to Angela Robinson,19 the tenet, ‘No one is born alone and no one should die alone,’ is central to Mi'kmaq understanding of not going through this journey alone. There is a community of faith that binds individuals to one another as well as to a greater collective. The ‘spirit road’ is sometimes referred to by the Mi'kmaq as a journey that every person must take. It is not travelled alone, but rather in the company of the ancestors, linking the individual's journey with those who came before. Drumming; praying to the four directions; lighting sage, tobacco or sweet grass; and singing or chanting sometimes are included by individuals accompanying the person in this journey. Scared fires and pipe ceremonies are also traditions that communities may partake in collectively during this transition. End of life is not so much its own period but a continuation of the realities that there are many levels of existence. Combined with the importance of family and community, the perception of spirit is linked to a strong sense of faith in both the life and death process. ‘Letting the spirit go’ refers to family, friends, and loved ones encouraging or giving the sick person opportunity to let their life spirit go to the ‘other world.’ As a result of this perspective, healthcare providers should not restrict the number of visitors, as family means more than immediate family members. Prohibiting visitors from bringing sacred objects or prohibiting ceremonies that enable an individual to travel to the spirit world is also problematic.

The ‘spirit road’ is sometimes referred to by the Mi'kmaq as a journey that every person must take. It is not travelled alone, but rather in the company of the ancestors, linking the individual's journey with those who came before. Drumming; praying to the four directions; lighting sage, tobacco or sweet grass; and singing or chanting sometimes are included by individuals accompanying the person in this journey. Scared fires and pipe ceremonies are also traditions that communities may partake in collectively during this transition. End of life is not so much its own period but a continuation of the realities that there are many levels of existence. Combined with the importance of family and community, the perception of spirit is linked to a strong sense of faith in both the life and death process. ‘Letting the spirit go’ refers to family, friends, and loved ones encouraging or giving the sick person opportunity to let their life spirit go to the ‘other world.’ As a result of this perspective, healthcare providers should not restrict the number of visitors, as family means more than immediate family members. Prohibiting visitors from bringing sacred objects or prohibiting ceremonies that enable an individual to travel to the spirit world is also problematic. Respect for handling sacred objects or personal belongings and knowing who can handle certain objects is important. A woman who is on her moon time (menses) should not touch sacred objects because women are believed to be very powerful during this time. Culturally competent advocacy on the part of healthcare providers is fundamental to providing care that is grounded in the understanding of this journey to the spirit world.

portant. A woman who is on her moon time (menses) should not touch sacred objects because women are believed to be very powerful during this time. Culturally competent advocacy on the part of healthcare providers is fundamental to providing care that is grounded in the understanding of this journey to the spirit world. Healthcare providers can play a critical role in providing an environment that positively affects the individual and those in attendance at end of life by learning the values and beliefs of the individuals and families who are willing to accept their care. Historically, coping with illness, bereavement and pain is considered acceptable when met with even-temperedness and suppression of grief rather than public displays of despair and inconsolable grief.15 Whitehead spoke of the value of patience which was often associated with the importance placed on not complaining about ailments. This was perceived as a survival mechanism or a way of being stoic in the face of disaster or illness. However, Mi'kmaq people do not view the English term stoic as being an adequate description of their cultural teaching and socialisation. Historically, Mi'kmaq suffered without showing the effect of misfortune and to do the contrary was believed to be the height of immaturity or a means of drawing attention, even causing harm. Complaining was perceived as being selfish and was considered to have the effect of making others depressed.

g and socialisation. Historically, Mi'kmaq suffered without showing the effect of misfortune and to do the contrary was believed to be the height of immaturity or a means of drawing attention, even causing harm. Complaining was perceived as being selfish and was considered to have the effect of making others depressed. Special respect and deference for elders is a fundamental given in Mi'kmaq communities. Some respected elders are known for their knowledge and wisdom of oral traditions and this may not necessarily be related to age. Elders are well known and play a significant role in the community. Healthcare providers must build relations with respected elders in the community as the relationship is seen as a partnership for addressing culturally competent care for community members. Historically, care giving was defined within the context of reciprocity, and was often experienced throughout the lifespan. Today, there are differences in the availability of supports that exist on and off reserve which has led to changes in care, particularly around the role of the community in providing assistance and the role of family in providing respite care.9 A sense of family, community and traditional views of spirituality frame the view of health among Mi'kmaq. There is a rooted belief in the strength of prayer and faith in its ability to cure as much as any intervention. More value may be placed on quality of life rather than on pursuit of a cure when life balance and wellness are a part of the holism that is ethically equated with mental, physical, emotional and spiritual health.20

a rooted belief in the strength of prayer and faith in its ability to cure as much as any intervention. More value may be placed on quality of life rather than on pursuit of a cure when life balance and wellness are a part of the holism that is ethically equated with mental, physical, emotional and spiritual health.20 Aboriginal worldviews see health as more than a lack of disease. Health is linked to the economy, the land, community and culture.21 Emphasis on biomedical treatment through technological interventions may be associated with a colonial approach to health, while at the same time diverse perspectives regarding aggressive treatment and technology abound.22 A study reported that 66% of a sample of 100 Mi'kmaw patients at a community health centre used Mi'kmaq medicine.23 Of those, 24.3% reported using it as a first-line of treatment, 92.4% had not discussed this with their physician and 31.8% said they believed Mi'kmaq medicine was better overall than Western medicine. Non-Mi'kmaq healthcare providers should not assume that someone in their care is following treatment measures prescribed by Western medicine and should openly discuss this with clients in a non-judgemental manner. It is important for non-Mi'kmaq healthcare providers to understand and appreciate the worldview, spirituality, role of family and community, and communication norms of Mi'kmaw individuals with whom they are working, and to strive for an alignment of values in the provision of care.

clients in a non-judgemental manner. It is important for non-Mi'kmaq healthcare providers to understand and appreciate the worldview, spirituality, role of family and community, and communication norms of Mi'kmaw individuals with whom they are working, and to strive for an alignment of values in the provision of care. Language The words that are used and a lack of common understanding of terms can lead to miscommunication and misinterpretation of needs and services. For example, terms such as home care, supportive care and palliative care are sometimes used interchangeably by federal and provincial service providers even though they may imply different ideas. Language barriers also affect the understanding of care plans. Even if interpretation services are available, they may not be accessed because persons are not identified as having this need since the Mi'kmaq generally have some level of understanding of English, some Mi'kmaq may not speak Mi'kmaq or because interpreters may not be readily available. Furthermore, interpretation services may not adequately handle the nuances of Mi'kmaq terms and the wording used by non-Mi'kmaw healthcare providers.

his need since the Mi'kmaq generally have some level of understanding of English, some Mi'kmaq may not speak Mi'kmaq or because interpreters may not be readily available. Furthermore, interpretation services may not adequately handle the nuances of Mi'kmaq terms and the wording used by non-Mi'kmaw healthcare providers. It is important to note that traditional Mi'kmaq do not use the term ‘end of life’ since they view the present life as a continuum into a next life. Mi'kmaq terms, which are not well understood by non-Mi'kmaq care providers, are very relevant. Three of these (apiksiktatulti, nemu'ltus and salite) are introduced in table 1. Mi'kmaq traditions associated with the journey to the spirit world are culture-specific and individually determined. Beyond access to publicly-funded health services and historic Mi'kmaq ways of caring for each other, there can be an additional dimension. Jesuit priests introduced the Mi'kmaq to Roman Catholicism which remains important to varying degrees among the Mi'kmaq in terms of both values and language; for example, heaven may be the term used for the spirit world by some Mi'kmaq.

storic Mi'kmaq ways of caring for each other, there can be an additional dimension. Jesuit priests introduced the Mi'kmaq to Roman Catholicism which remains important to varying degrees among the Mi'kmaq in terms of both values and language; for example, heaven may be the term used for the spirit world by some Mi'kmaq. Discussion By separating and reflecting on the issues of jurisdiction and cultural understanding, progress can be made toward greater cultural competency. Inclusion of indigenous ways of knowing about health is critically important for ensuring cultural competency. Non-Aboriginal healthcare providers often discuss and provide supportive and palliative care from a Eurocentric approach24 either due to lack of knowledge or lack of system support to provide culturally competent supportive and palliative care. Cultural competency transcends cultural understanding. It includes ensuring that the healthcare system is inclusive of different worldviews in the planning, implementation and evaluation of healthcare services. The authors recommend that indigenous leaders undertake a more indepth study of supportive and palliative care needs in their communities by using an integrative framework to guide this research.25 This paper demonstrates the need for understanding across conceptualisations and is in accord with the report of the Canadian Hospice Palliative Care Association, which advises building on strengths and working together to improve Aboriginal hospice palliative care.26

an integrative framework to guide this research.25 This paper demonstrates the need for understanding across conceptualisations and is in accord with the report of the Canadian Hospice Palliative Care Association, which advises building on strengths and working together to improve Aboriginal hospice palliative care.26 The authors recommend that consideration be given to having palliative care researchers work with the Mi'kmaq using research processes that respect the community. The purpose would be to explore how current perspectives on palliative and supportive care might be enhanced or reconstructed through a greater understanding of the Mi'kmaq ways. Conclusion The identification of both jurisdictional issues and cultural understanding in the historic context of the Mi'kmaq is important for enhancing the cultural competency of non-Mi'kmaq care providers. Best practice indigenous supportive and palliative care may be contrary to conventional Western healthcare assumptions and practice. Through reflection on the situation of Nova Scotia's Mi'kmaq, non-indigenous healthcare providers can assess how they might critique and increase their cultural competency. This paper will hopefully raise awareness of the need to more broadly understand and incorporate an indigenous cultural context when providing supportive and palliative care.

ituation of Nova Scotia's Mi'kmaq, non-indigenous healthcare providers can assess how they might critique and increase their cultural competency. This paper will hopefully raise awareness of the need to more broadly understand and incorporate an indigenous cultural context when providing supportive and palliative care. The authors thank and acknowledge the following for sharing their insight, understanding and guidance in the preparation of this manuscript: Theresa Meuse, Mi'kmaq educator, advisor and author; Sister Veronica Matthews, retired Aboriginal nurse and Mi'kmaq Elder; Charlotte (Loppie) Reading, Human and Social Development, University of Victoria, and formerly School of Health and Human Performance, Dalhousie University; Peter L Twohig, Canada Research Chair in Atlantic Canada Studies, Saint Mary's University; Angela Robinson, Social/Cultural Studies, Anthropology, Sir Wilfred Grenfell College, Memorial University; and Ruth Whitehead, Curator Emerita, Nova Scotia Museum.

rly School of Health and Human Performance, Dalhousie University; Peter L Twohig, Canada Research Chair in Atlantic Canada Studies, Saint Mary's University; Angela Robinson, Social/Cultural Studies, Anthropology, Sir Wilfred Grenfell College, Memorial University; and Ruth Whitehead, Curator Emerita, Nova Scotia Museum. Contributors: GJ was responsible for overall conceptualisation of the paper and guidance especially in relation to palliative and end of life care for vulnerable populations as well as critical reflection, interpretation and editing. AV advised on Mi'kmaq information sources, conceptualisation and design, and assisted with the critical review, understanding of the Mi'kmaq culture, liaison with Mi'kmaq consultants and editing of the manuscript. SP carried out the initial literature review and meetings with strategic informants, prepared the first draft of this paper and contributed to revisions. All authors contributed to the literature review and have read and approved this final manuscript. GJ and AV jointly are the guarantors for the content of this paper based on their respective areas of expertise and through ongoing consultation with two Mi'kmaq consultants. Funding: This report was carried out with support from a Canadian Institutes for Health Research Interdisciplinary Capacity Enhancement for vulnerable populations grant number FRN-80067. Competing interests: None. Provenance and peer review: Not commissioned; externally peer reviewed.

Contributors: GJ was responsible for overall conceptualisation of the paper and guidance especially in relation to palliative and end of life care for vulnerable populations as well as critical reflection, interpretation and editing. AV advised on Mi'kmaq information sources, conceptualisation and design, and assisted with the critical review, understanding of the Mi'kmaq culture, liaison with Mi'kmaq consultants and editing of the manuscript. SP carried out the initial literature review and meetings with strategic informants, prepared the first draft of this paper and contributed to revisions. All authors contributed to the literature review and have read and approved this final manuscript. GJ and AV jointly are the guarantors for the content of this paper based on their respective areas of expertise and through ongoing consultation with two Mi'kmaq consultants. Funding: This report was carried out with support from a Canadian Institutes for Health Research Interdisciplinary Capacity Enhancement for vulnerable populations grant number FRN-80067. Competing interests: None. Provenance and peer review: Not commissioned; externally peer reviewed. Open Access: This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 3.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/3.0/

Introduction A clinician's estimate of likely survival (CES) duration has been described as a subjective process in which a ‘personal clinical judgement’ is formulated.1 Numerous reports have demonstrated that CES is most often inaccurate with a substantial tendency to overestimate.2 Despite this, CES has consistently been shown to be an independent predictor of actual survival duration3–6 with the suggestion made that CES be used as the reference standard for evaluating other methods of survival estimation.3 The European Association for Palliative Care recently published evidence-based clinical recommendations regarding overall methods and processes used in survival estimation.7 One of six key outcomes was the recommendation that CES be accepted as a valid process. Clinicians were cautioned, however, to be aware of elements impacting accuracy. With the overall aim of gaining a deeper understanding of CES formulation, the primary objective for the current study was to identify and explore the specific elements that may inform and/or impact the process of CES formulation among palliative care clinicians.

Despite this, CES has consistently been shown to be an independent predictor of actual survival duration3–6 with the suggestion made that CES be used as the reference standard for evaluating other methods of survival estimation.3 The European Association for Palliative Care recently published evidence-based clinical recommendations regarding overall methods and processes used in survival estimation.7 One of six key outcomes was the recommendation that CES be accepted as a valid process. Clinicians were cautioned, however, to be aware of elements impacting accuracy. With the overall aim of gaining a deeper understanding of CES formulation, the primary objective for the current study was to identify and explore the specific elements that may inform and/or impact the process of CES formulation among palliative care clinicians. Methods Research design and participants Individual semistructured interviews were completed for this qualitative study. Potential participants were members of a consultative palliative care team within a tertiary academic health sciences centre. Purposive sampling was used to select the physicians and advanced practice nurses who responded to a general emailed invitation and subsequently were offered formal study participation. Of note, clinicians routinely formulate a CES for every patient referred for palliative care consultation in outpatient and inpatient settings. This is done at the time of initial assessment irrespective of the reason for referral and is represented by the choice of one time-based prognostic category felt to best represent an individual patient's likely survival duration.2 Clinicians were aware the CES data was being collected purely for administrative and academic purposes and in general, individual CES formulations were neither expected to be communicated to patients nor impact clinical decision-making. In addition, at the time of the present study, feedback had not yet been given to the clinicians regarding individual rates of CES accuracy.

rely for administrative and academic purposes and in general, individual CES formulations were neither expected to be communicated to patients nor impact clinical decision-making. In addition, at the time of the present study, feedback had not yet been given to the clinicians regarding individual rates of CES accuracy. Separate interviews were completed for individual participants and informed consent and demographic information was obtained prior to the start of each. An interview guide was developed and field-tested to ensure clarity of the questions. The open-ended questions allowed clinicians to identify and explore factors perceived as possibly influencing the process of CES formulation. The interviewer used directive probes to elicit additional information and clarify responses. To capture possible attitudinal, personality or values-based factors, clinicians were asked a series of questions that included previous experiences with survival estimation, perception of general estimate accuracy and possible circumstances of a conscious choice to either overestimate or underestimate. Field notes were kept to capture key points and observations made during the interviews. Ethical considerations Ethics approval was obtained from the Research Ethics Board of the Sunnybrook Heath Sciences Centre. Data analysis Interviews were audio-recorded and transcribed verbatim. Descriptive data (ie, transcripts and the accompanying field notes) were analysed using a grounded theory approach.

Separate interviews were completed for individual participants and informed consent and demographic information was obtained prior to the start of each. An interview guide was developed and field-tested to ensure clarity of the questions. The open-ended questions allowed clinicians to identify and explore factors perceived as possibly influencing the process of CES formulation. The interviewer used directive probes to elicit additional information and clarify responses. To capture possible attitudinal, personality or values-based factors, clinicians were asked a series of questions that included previous experiences with survival estimation, perception of general estimate accuracy and possible circumstances of a conscious choice to either overestimate or underestimate. Field notes were kept to capture key points and observations made during the interviews. Ethical considerations Ethics approval was obtained from the Research Ethics Board of the Sunnybrook Heath Sciences Centre. Data analysis Interviews were audio-recorded and transcribed verbatim. Descriptive data (ie, transcripts and the accompanying field notes) were analysed using a grounded theory approach. Specifically, data was coded in a line-by-line process through which keywords, phrases and texts were identified. Concepts were subsequently organised by literal and interpretive meanings and grouped thematically into categories. Key categories and concepts were compared iteratively looking for similarities, differences and relationships, then organised into broader themes. Major themes were then organised into a theoretical framework by making first degree connections between the categories. To maximise validity, two investigators analysed the data separately and agreement for the major themes was achieved through an iterative process and comparative analyses. Minor disagreements were resolved through discussion and exploration of the choice of terminology.

first degree connections between the categories. To maximise validity, two investigators analysed the data separately and agreement for the major themes was achieved through an iterative process and comparative analyses. Minor disagreements were resolved through discussion and exploration of the choice of terminology. Results In total, nine interviews were conducted including those of seven physicians and two advanced practice nurses. The entire clinical practice for all participants was focused on palliative care with a range from 4–14 years for the physicians and 8–9 years for the nurses. Clinicians provide palliative care in inpatient and ambulatory clinic settings for patients with a range of diagnoses and actual survival duration. Five major themes were identified as being central to the process of CES formulation: Use of objective patient-specific elements All participants identified certain objective measures that they in some way use when estimating survival duration. Performance status, as an example, was cited by all participants as being critical to CES formulation:As a palliative care physician, we're looking at performance status. Additionally, patient demographic information was identified as being of particular importance:The frail elderly person…I am more likely to give a lower prognosis than a young healthy person who I think…may survive longer.

All participants identified certain objective measures that they in some way use when estimating survival duration. Performance status, as an example, was cited by all participants as being critical to CES formulation:As a palliative care physician, we're looking at performance status. Additionally, patient demographic information was identified as being of particular importance:The frail elderly person…I am more likely to give a lower prognosis than a young healthy person who I think…may survive longer. Certain biochemical markers (eg, albumin, creatinine and C reactive protein) as well as specific symptoms and signs (eg, ‘weight loss’, ‘dsypnoea’ and ‘delirium’) were felt likely to contribute to a CES formulation, particularly among actively dying patients. For patients living with cancer, the specific type, stage and grade were noted to be important factors:…if somebody's full of liver mets and brain mets, that's a pretty hard finding Most clinicians spontaneously introduced survival estimation tools or models into the discussion, conveying the general opinion that they are of limited practical value when applied to individual patients:There are references that can guide you, that will give you means and medians and standard deviations. So, that's always…in the back of your mind, the stuff that we theoretically know. But that's a statistic. The patient-clinician relationship

that they are of limited practical value when applied to individual patients:There are references that can guide you, that will give you means and medians and standard deviations. So, that's always…in the back of your mind, the stuff that we theoretically know. But that's a statistic. The patient-clinician relationship The strength of the relationship between the clinician and patient was viewed by all participants as an important element of CES formulation and had the potential to impact the process, either consciously or subconsciously. Several participants acknowledged the possibility of overestimating survival among patients with whom they had a particularly strong relationship:…if I develop a rapport or connection with a patient that's particularly strong…would that impact how I see their prognosis, and might I extend that because of my wish? That this person should live longer? And in that case I would say yes, I think it probably does. I'm human, right? So I'm going to connect with certain patients more than others. For those that I have a connection with, feel that personal drive, to want them to live. So my discussions probably are going to be biased in a way that is going to make them lean more towards being more optimistic about their prognosis. I admit that. I try and be as objective as I can be, but I can't deny that I'm going to be human and do that. I'm aware of it, and I try and be conscious of it, but I know in my mind, with patients I have better relationships with, I struggle with giving them a poor prognosis.

I'm human, right? So I'm going to connect with certain patients more than others. For those that I have a connection with, feel that personal drive, to want them to live. So my discussions probably are going to be biased in a way that is going to make them lean more towards being more optimistic about their prognosis. I admit that. I try and be as objective as I can be, but I can't deny that I'm going to be human and do that. I'm aware of it, and I try and be conscious of it, but I know in my mind, with patients I have better relationships with, I struggle with giving them a poor prognosis. The element of trust within the clinician-patient relationship, in particular its maintenance, was viewed as being important. The potential consequence of an inaccurate CES being a loss of this trust was identified as potentially impacting the CES formulation process:I don't worry that I'm going to give the patient a poor prognosis and they're going to have a different perception of our doctor/physician-patient relationship. The concern is, if I'm inaccurate, does that affect their trust in me as their physician? Purpose and context of an individual CES

the CES formulation process:I don't worry that I'm going to give the patient a poor prognosis and they're going to have a different perception of our doctor/physician-patient relationship. The concern is, if I'm inaccurate, does that affect their trust in me as their physician? Purpose and context of an individual CES The underlying purpose for any given CES and how it might be used was identified as a possible element contributing to the process of CES formulation. In particular, participants outlined the potential impact on the intended audience that is, patient, family members, the clinician themselves (ie, the individual formulating the CES), other health professionals or allocators of healthcare resources. A specific example cited by several participants was the circumstance in which the CES determines qualification for access to certain palliative care services:Maybe I might want to underestimate. And also then they get some more services in terms of palliative care services in the community. In exploring the idea that communicating any given CES to a patient may then become a ‘self-fulfilling prophecy’, nurses and physicians differed in their response. Both nurses and physicians indicated their belief in this being an important phenomenon of which clinicians should be mindful:I think for some people, if you say it's going to be, they believe it. Especially if they trust you… I've seen people really just become deflated when they hear that.

differed in their response. Both nurses and physicians indicated their belief in this being an important phenomenon of which clinicians should be mindful:I think for some people, if you say it's going to be, they believe it. Especially if they trust you… I've seen people really just become deflated when they hear that. The physician participants uniformly disagreed with the notion that communicating a CES might impact actual survival duration. All physicians however did acknowledge ‘self-fulfilling prophecies’ as being a common source of distress for family members, thus requiring insight into patient/family perspectives regarding this topic:Families are more concerned if you give them (patients) this prognosis, they'll die. Or they'll lose their will to live. I see that in, a lot, in families. Less so in patients. I get the sense that patients are in general more realistic and want to know more information than their families think. Perceived role of hope A desire to maintain a perceived sense of hope was frequently identified as possibly influential on a CES formulation. The importance of maintaining hope was balanced however with a perceived professional duty to avoid what was described as ‘giving a false sense of hope’:I'm not there to take away a patient's hope to live, and a drive to live

erceived sense of hope was frequently identified as possibly influential on a CES formulation. The importance of maintaining hope was balanced however with a perceived professional duty to avoid what was described as ‘giving a false sense of hope’:I'm not there to take away a patient's hope to live, and a drive to live I'm the type of person that likes to try and give patients hope and give them kernels of hope, but realistic hope. So I don't like giving people false hope. So I don't want to say that I'd like my estimates to be overly optimistic, I think that's giving people false hope. Likelihood of CES inaccuracy All participants rated the accuracy of their own survival estimates and those of healthcare providers in general as being poor. Frequently addressed was the notion of impossibility for any CES to be precisely accurate:It's about the same as flipping a coin I'm pretty average compared with other physicians. I wouldn't say that I'm better or worse than other physicians. I think we're all sort of in the same boat Do I ever think anything will be 100% accurate? Absolutely not. One participant indicated she endeavoured to balance the likelihood of inaccuracy by consciously underestimating:If the prognosis is just relevant to me, I would rather underestimate it…I am going to try as best I can to prepare the patient and family for what is to come, and that I have the time to do that. And that the rug isn't pulled out from under all of us by me overestimating.

d of inaccuracy by consciously underestimating:If the prognosis is just relevant to me, I would rather underestimate it…I am going to try as best I can to prepare the patient and family for what is to come, and that I have the time to do that. And that the rug isn't pulled out from under all of us by me overestimating. Participants alluded to overall limitations in human knowledge and the epistemic constraints impacting the timing of death. This was usually framed in terms of a fundamental mystery regarding death, one that scientific inquiry could not possibly resolve:So I wonder, either we just haven't found the right tool or tools, or perhaps there's something around the end-of-life that will always be…unscientific. So I think there's that thing, that kind of core to the person that we can't ultimately affect… That's the mystery of human life and human death. And so no matter how skilled we become, I don't think that we're going to ultimately be able to call everyone. At various points throughout each interview, participants’ perspective on the importance of skilful CES communication was outlined.

So I think there's that thing, that kind of core to the person that we can't ultimately affect… That's the mystery of human life and human death. And so no matter how skilled we become, I don't think that we're going to ultimately be able to call everyone. At various points throughout each interview, participants’ perspective on the importance of skilful CES communication was outlined. Discussion A recent ‘State of the Science’ meeting involving expert clinicians and researchers in prognostication was held with the aim of identifying and defining key research priorities for the field.8 Of the 40 specific topics considered, ‘what clinicians use and what they are considering when estimating survival’ was identified as the second most important priority. In seeking to understand the clinicians’ perceived elements contributing to the process of CES formulation, the results of the present study outline an initial understanding of one aspect of this key priority.

and what they are considering when estimating survival’ was identified as the second most important priority. In seeking to understand the clinicians’ perceived elements contributing to the process of CES formulation, the results of the present study outline an initial understanding of one aspect of this key priority. The main findings in this study provide a preliminary overview of several factors that may impact CES formulation for any given patient. To summarise, palliative care clinicians are aware of objective clinical factors known to correlate with actual survival duration and likely integrate this information when formulating a CES. In addition, the CES for an individual patient is formulated within a larger context comprised of a number of elements that may influence individual estimates. These include the relationship the clinician has with the patient and the patient's family, the intent and audience for the CES and the overall likelihood of inaccuracy, if a certain level of precision is the expectation. Finally, the personal belief system and set of values of a clinician, in particular any perceived need to preserve a sense of hope among patients and family members, may impact the process of CES formulation. For clinicians in this study, the importance of communication was identified and underscored as a critical component of CES. Whether or not some element of the CES communication process might influence the formulation process remains unclear.

The main findings in this study provide a preliminary overview of several factors that may impact CES formulation for any given patient. To summarise, palliative care clinicians are aware of objective clinical factors known to correlate with actual survival duration and likely integrate this information when formulating a CES. In addition, the CES for an individual patient is formulated within a larger context comprised of a number of elements that may influence individual estimates. These include the relationship the clinician has with the patient and the patient's family, the intent and audience for the CES and the overall likelihood of inaccuracy, if a certain level of precision is the expectation. Finally, the personal belief system and set of values of a clinician, in particular any perceived need to preserve a sense of hope among patients and family members, may impact the process of CES formulation. For clinicians in this study, the importance of communication was identified and underscored as a critical component of CES. Whether or not some element of the CES communication process might influence the formulation process remains unclear. Historically, survival estimation has been categorised into two general methods. In the first method, CES, has been summarised as a ‘personal clinical judgement’ that is subjective in nature with complex and poorly defined variables.1 Pontin and Jordan recently reported the findings of their study addressing views on the overall prognostication experience among a cohort of palliative care clinicians.9 The study provides an important overview of the clinicians’ experience and particularly underscores the potential impact of CES inaccuracy on clinicians themselves and the fear and uncertainty often experienced. The authors indicate clinicians ‘appear to use certain prognostic factors in prognostication’ and go on to identify that ‘further work is required to identify how and why these factors are used’. One of the strengths of the current study is the participants’ ability to reflect on the CES-related process itself that is, routine formulation for each patient upon initial assessment, without either a standard expectation of CES communication or linkage to clinical decision-making. The resulting team culture with respect to CES overall is such that clinicians are able to formulate without substantial concern for any direct impact on patients and families.

on for each patient upon initial assessment, without either a standard expectation of CES communication or linkage to clinical decision-making. The resulting team culture with respect to CES overall is such that clinicians are able to formulate without substantial concern for any direct impact on patients and families. The second method of survival estimation is comprised of objective and actuarial measures typically integrated into some form of a prognostic tool or formula, whereby an estimate is generated from the amalgamation of clinical data elements and related mortality data. Previously identified clinical indices known to correlate with actual survival duration include patient performance status, presence of anorexia and/or weight loss, presence of breathlessness and/or delirium and certain biological parameters (eg, serum albumin, white blood cell count and lymphocyte ratio).7 10–13 The finding in the present study that clinicians may consciously or subconsciously consider objective patient-specific measures when formulating a CES, outlines an objective component of the largely subjective process. The idea that during CES formulation clinicians may integrate objective measures known to correlate with actual survival duration highlights what is likely to be an overlap between the two general methods of survival estimation, previously outlined as separate subjective and objective processes.

ely subjective process. The idea that during CES formulation clinicians may integrate objective measures known to correlate with actual survival duration highlights what is likely to be an overlap between the two general methods of survival estimation, previously outlined as separate subjective and objective processes. The length of the clinician-patient relationship has previously been reported to directly correlate with CES accuracy that is, the longer the duration, the lower the accuracy.14 Although ‘length’ is somewhat different than ‘strength’ in the context of interpersonal relationships, by acknowledging the likelihood of CES overestimation for patients with whom they have a stronger relationship in comparison to others, participants in the present study qualitatively validate Christakis’ finding. Consciously acknowledging the ‘strength’ of the clinician-patient relationship and proactively factoring this element into a CES formulation may at least in part correct for its impact on the resulting CES itself. Similarly, a conscious acknowledgement could be given to each of the potentially influential elements identified by study participants, which include intended audience, implications for resource allocation and clinicians’ personal tendencies, values and belief systems.

orrect for its impact on the resulting CES itself. Similarly, a conscious acknowledgement could be given to each of the potentially influential elements identified by study participants, which include intended audience, implications for resource allocation and clinicians’ personal tendencies, values and belief systems. A number of limitations exist in the present study. First, study participants were from one institution, thus limiting generalisability. Second, since the given study participants were all full time palliative care clinicians, we did not capture the perspective of other disciplines with respect to possible elements impacting CES formulation. Another important limitation of the study relates to participants only being able to identify potentially influential elements of which they were consciously aware. As with any individual decision-making process, a range of elements have the potential to have a subconscious impact. The elements generated from this study may not be comprehensive or exhaustive and the interaction between elements not accounted for. Future research should include work that is similar, completed at different centres and among different fields of healthcare. Additionally, in having gained a better understanding of the elements comprising a CES, efforts could be focused on designing, implementing and evaluating a standard approach to CES formulation and examining the impact on patients and family members for whom the formulation is communicated.

ferent fields of healthcare. Additionally, in having gained a better understanding of the elements comprising a CES, efforts could be focused on designing, implementing and evaluating a standard approach to CES formulation and examining the impact on patients and family members for whom the formulation is communicated. Conclusion Multiple ‘calls to action’ have been made for the development and/or refinement of survival estimate tools that provide more accurate information. Results of the current study suggest parallel efforts should be made towards encouraging clinicians to develop a personalised and standardised approach to CES formulation whereby an awareness of the menu of potentially impacting elements are consciously integrated into an individual process. In addition, it may be wise to acknowledge the ultimate impossibility of absolute accuracy in survival estimation and ensure that efforts are also focused on developing and disseminating a standardised approach to CES communication. This includes identifying effective professional development strategies such that clinicians would achieve and maintain competency in communicating survival estimation in a manner that is individualised and meaningful. Contributors: Conception and design: JM and RC. Data acquisition and analysis: JM and RC. Data interpretation: JM, RC and DS. Development of manuscript draft: RC. Manuscript revision: JM, RC and DS. Final approval of submission: JM, RC and DS. Guarantor responsible for overall content: JM. Funding: None. Competing interests: None. Ethics approval: Sunnybrook Health Sciences Centre IRB.

Contributors: Conception and design: JM and RC. Data acquisition and analysis: JM and RC. Data interpretation: JM, RC and DS. Development of manuscript draft: RC. Manuscript revision: JM, RC and DS. Final approval of submission: JM, RC and DS. Guarantor responsible for overall content: JM. Funding: None. Competing interests: None. Ethics approval: Sunnybrook Health Sciences Centre IRB. Provenance and peer review: Not commissioned; externally peer reviewed.

Introduction Two thirds of women diagnosed with breast cancer today are likely to survive for at least 20 years.1 As a result more women are living with the longer-term side-effects of treatment. These can include neurocognitive changes,2 3 sleep disturbance,4 quality of life impairment and persistent fatigue.5 Of the many symptoms associated with cancer and its treatment, fatigue is probably the most under-recognised and poorly managed.6 We have previously undertaken a systematic review of the literature regarding the prevalence of fatigue in breast cancer survivors (BCS).7 Many of the early studies of fatigue in cancer populations were limited by the use of poorly validated assessment instruments and the lack of a coherent definition of cancer-related fatigue syndrome (CRFS).8 However, in 1998 Cella and colleagues9 proposed the introduction of diagnostic criteria for CRFS, which have now been successfully applied in a number of clinical studies.10–14 These criteria are based on similar principles to those used to diagnose patients with chronic fatigue syndrome (CFS).15 Most recently, our own group rigorously applied the diagnostic criteria to BCS between 3 months and 2 years after completion of treatment. We found the prevalence of CRFS to be 30%.16

l studies.10–14 These criteria are based on similar principles to those used to diagnose patients with chronic fatigue syndrome (CFS).15 Most recently, our own group rigorously applied the diagnostic criteria to BCS between 3 months and 2 years after completion of treatment. We found the prevalence of CRFS to be 30%.16 CRFS is a clinical diagnosis and as such relies on the subjective reports of patients. Broadly speaking, patients' complaints of fatigue relate to either decrements in motor performance (physical fatigue) or impaired cognitive abilities (mental fatigue). Unfortunately, the same term can be used to describe both an objective physical or mental decrement in performance and a subjective mental state. Both types of fatigue are usually found to a greater or lesser extent in the same individual. However, there is no direct correspondence between the two phenomena. It is quite possible to feel extremely subjectively fatigued but to perform relatively normally on objective tests of physical or mental functioning.

state. Both types of fatigue are usually found to a greater or lesser extent in the same individual. However, there is no direct correspondence between the two phenomena. It is quite possible to feel extremely subjectively fatigued but to perform relatively normally on objective tests of physical or mental functioning. However, patients' subjective complaints of physical fatigue relate to changes in the activities of daily living, mobility and the need for daytime naps. Unobtrusive objective monitoring is required to investigate these phenomena and studies have been performed using actigraphy (real-time activity monitoring). Overall, the data from several studies17 18 suggest that there is a strong correlation between changes in subjective fatigue and actigraphy data measures. The most consistent correlations occur between fatigue measures and disruption of sleep quality. However, the limitations of and variations in data collection mean these findings need further confirmation. In particular, these data have been recorded in patients on treatment and linked to sleep disturbances and have not adequately assessed the level of daytime fatigue. The exception to this was a study by our group16 in which, contrary to expectation, we found no significant differences in activity levels in a population of fatigued BCS compared to non-fatigued controls.

ients on treatment and linked to sleep disturbances and have not adequately assessed the level of daytime fatigue. The exception to this was a study by our group16 in which, contrary to expectation, we found no significant differences in activity levels in a population of fatigued BCS compared to non-fatigued controls. Several researchers have investigated cognitive changes in cancer survivors. Recent systematic reviews19 and a meta-analysis3 concluded that there is some evidence of objective declines in cognitive function (particularly verbal memory and executive function) following chemotherapy, but only one previous study has attempted to relate these objective findings to subjective complaints of fatigue.20 However, the authors did not use the diagnostic criteria9 and only employed a limited range of cognitive testing. The primary aim of this study was to examine differences in objective measures of cognition (specifically focusing on previously identified cognitive domains that were affected) and in physical activity in women who did or did not meet the criteria for CRFS. Materials and methods BCS were recruited from January 2009 to May 2011 from a single centre nurse-led follow-up clinic at St George's Healthcare NHS Trust. Approval was obtained from Wandsworth ethics committee prior to data collection (ref 08/H0803/182). All patients who were clinically and radiologically disease-free between 3 months and 2 years after the end of their primary treatment (of any modality) were invited to participate. Concurrent hormone use was allowed.

Materials and methods BCS were recruited from January 2009 to May 2011 from a single centre nurse-led follow-up clinic at St George's Healthcare NHS Trust. Approval was obtained from Wandsworth ethics committee prior to data collection (ref 08/H0803/182). All patients who were clinically and radiologically disease-free between 3 months and 2 years after the end of their primary treatment (of any modality) were invited to participate. Concurrent hormone use was allowed. Those patients with significant cognitive impairment, a psychiatric history or medical co-morbidities on notes review or initial discussion with the researcher (OM) were excluded from entering the study (12 women in total). This was a pragmatic decision as these women were not eligible for the complete study. Eligible women were identified from the clinic list 14 days before their appointment and received an introductory letter and an information sheet by post. Women who declined to participate in the study were asked if they would be willing to complete a short fatigue questionnaire. Assessment 1 Women who consented to participate were invited to an initial consultation in a designated research facility and completed the following interviews and actigraphy.

Eligible women were identified from the clinic list 14 days before their appointment and received an introductory letter and an information sheet by post. Women who declined to participate in the study were asked if they would be willing to complete a short fatigue questionnaire. Assessment 1 Women who consented to participate were invited to an initial consultation in a designated research facility and completed the following interviews and actigraphy. Diagnostic Interview for Cancer-Related Fatigue This interview determines whether the participant meets the four criteria for a diagnosis of CRFS9: criterion A—the presence of 2 weeks of significant fatigue in the preceding month and the presence of at least five out of nine other fatigue-related symptoms; criterion B—the fatigue has a significant effect on work or self-care; criterion C—the fatigue symptoms are a consequence of cancer or cancer therapy; and criterion D—the symptoms are not primarily a consequence of a co-morbid psychiatric disorder. The final criterion can be assessed clinically, but the most robust method is to use a contemporaneous psychiatric interview. Participants with a significant psychiatric history which was felt to be contributing to fatigue were excluded.

—the symptoms are not primarily a consequence of a co-morbid psychiatric disorder. The final criterion can be assessed clinically, but the most robust method is to use a contemporaneous psychiatric interview. Participants with a significant psychiatric history which was felt to be contributing to fatigue were excluded. Structured Clinical Interview for the DSM–IV (SCID) The Structured Clinical Interview for the DSM-IV (SCID) provides a method for obtaining Diagnostic and Statistical Manual (DSM)-IV diagnoses. The procedure has been successfully used in previous studies examining CRFS11 12 and by our group.16 All interviews were conducted by the same person (OM). The SCID can be administered by a non-psychiatrist. OM was trained in administration and supervised by a consultant liaison psychiatrist. The average administration time was 30 min. Other questionnaires All women were also asked to complete the following questionnaires: (1) Functional Assessment of Cancer Therapy (FACT-F). This is a 13-item questionnaire to measure fatigue21 in cancer and has been widely used in cancer treatment trials.22 (2) The Chalder Fatigue Scale (FS).23 This is an 11-item questionnaire to measure fatigue. It was originally developed in non-cancer patients but has been extensively used in a number of studies in cancer.24 (3) All participants were asked to give a score of average fatigue over the last week on a 0–10 visual analogue scale (VAS-F), where 0 is no fatigue and 10 is fatigue as bad as you can imagine (after National Comprehensive Cancer Network (NCCN) guidance).25

(2) The Chalder Fatigue Scale (FS).23 This is an 11-item questionnaire to measure fatigue. It was originally developed in non-cancer patients but has been extensively used in a number of studies in cancer.24 (3) All participants were asked to give a score of average fatigue over the last week on a 0–10 visual analogue scale (VAS-F), where 0 is no fatigue and 10 is fatigue as bad as you can imagine (after National Comprehensive Cancer Network (NCCN) guidance).25 (4) Hospital Anxiety and Depression Scale (HADS). This is a 14-item scale with seven items on anxiety and seven items on depression.26 It is used extensively in clinical trials and has been employed by our group previously in fatigue assessment studies.27 (5) European Organisation of Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-30). This is a 30-item instrument with five functional scales, three symptom scales and a global quality of life score.28 (6) EORTC Breast Module (BR23). This is a 23-item scale which is validated for use in breast cancer patients.29 It includes items on breast and arm symptoms and systemic treatment effects. (7) The 7-item Insomnia Severity Index (ISI).30 This scale has previously been validated in a breast cancer population.31 The seven items in this scale were combined with two additional sleep related questions to determine if insomnia diagnostic criteria were met: ▶ During the past month how many nights per week have you taken more than 30 min to fall asleep? ▶ During the past month how many nights per week were you awake for more than 30 min?

(7) The 7-item Insomnia Severity Index (ISI).30 This scale has previously been validated in a breast cancer population.31 The seven items in this scale were combined with two additional sleep related questions to determine if insomnia diagnostic criteria were met: ▶ During the past month how many nights per week have you taken more than 30 min to fall asleep? ▶ During the past month how many nights per week were you awake for more than 30 min? A diagnosis of insomnia syndrome was made if one of the above two features was present on more than three nights per week on average, there was significant distress as scored on the ISI and/or fatigue or daytime somnolence was present. It took 20 min on average to administer these questionnaires and conduct the CRFS interview. An initial appointment time of an hour was made and potential participants were made aware of this time requirement. Actigraphy After completion of the questionnaires and interview, participants were asked to wear an Actiwatch AW4 (CamNtech, Cambridge, UK) for a period of 7 days and nights. The associated software and sleep algorithm32 allows for the determination of selected measures of activity, sleep and circadian rhythm (over a 7-day average). A predetermined 12 h timeframe is used to define daytime (light) activity: 06:00 to 18:00 h. The night-time activity period (dark) is defined as 18:00 to 06:00 h. The participants used an event marker on the watch to record their own individualised sleep patterns. This was recorded as the time entering and exiting bed overnight and did not include naps.

sed to define daytime (light) activity: 06:00 to 18:00 h. The night-time activity period (dark) is defined as 18:00 to 06:00 h. The participants used an event marker on the watch to record their own individualised sleep patterns. This was recorded as the time entering and exiting bed overnight and did not include naps. Assessment 2 Participants underwent objective cognitive testing using a predetermined computerised battery approximately 1 week later (CANTABeclipse, Cambridge Cognition, Cambridge, UK). The procedure uses a touch screen computer (Slimbook P210; PaceBlade, Amersfoort, The Netherlands) and press pad. It consists of seven different tests designed to assess the cognitive domains thought to be affected by previous breast cancer treatment.19 This computerised battery has been previously used in a longitudinal study of cognitive changes after breast cancer treatment.33 The procedure takes 45 min to complete and was administered by one of the authors (OM). The tests and cognitive domains studied are listed below: ▶ Paired Associates Learning (PAL; visuospatial ability and visual memory) ▶ Rapid Visual Information Processing (RVP; sustained attention and reaction time – motor skills) ▶ Match to Sample Visual Search (MTS; visuospatial ability and motor skills) ▶ Verbal Recognition Memory (VRM; verbal memory) ▶ Delayed Matching to Sample (DMS; attention) ▶ Affective Go No-go (AGN; decision and response and information processing) ▶ Intra-Extra Dimensional Set Shift (IED, executive function/information processing).

▶ Match to Sample Visual Search (MTS; visuospatial ability and motor skills) ▶ Verbal Recognition Memory (VRM; verbal memory) ▶ Delayed Matching to Sample (DMS; attention) ▶ Affective Go No-go (AGN; decision and response and information processing) ▶ Intra-Extra Dimensional Set Shift (IED, executive function/information processing). Statistical considerations The power calculation was based on an estimated sample size of 96 women. With 48 cases and 48 controls, the study would have had 80% power to detect a 0.57 (medium) between-group effect size on questionnaire and objective monitoring data at a 5% significance level. Since this was an exploratory and hypothesis generating study, no correction was made for multiple analyses. The between-group analysis was conducted using t tests with comparisons between mean scores.

t a 0.57 (medium) between-group effect size on questionnaire and objective monitoring data at a 5% significance level. Since this was an exploratory and hypothesis generating study, no correction was made for multiple analyses. The between-group analysis was conducted using t tests with comparisons between mean scores. Results Due to time and resource constraints, recruitment of fatigued patients terminated slightly early with 45 of an expected 48 patients recruited. However, the number of control subjects was larger than required. A total of 114 women completed the study (69 controls and 45 cases of CRFS). Non-participants (NP=102) were asked to complete the Chalder fatigue questionnaire and ethics approval was obtained to collect anonymised demographic and treatment data from these women. The main reasons for non-participation were cited as work (28%) or the need for extra visits (32%), while 12% mentioned other medical problems and the remainder either had ‘had enough’ or did not want to think about their cancer (28%). The NP had a significantly lower fatigue score (10.7 vs 13.1; p=0.04) and were older (62 years vs 57 years; p=0.005) than the participants. There were no differences on treatment variables, stage at diagnosis or time since diagnosis. Cases versus controls The overall prevalence of CRFS was 39% (45/114). A between-group comparison for all data is shown in tables 1–5 (table 1: treatment variables; table 2: questionnaire data; table 3: routine laboratory parameters; table 4: actigraphy data; and table 5: cognitive testing data).

Results Due to time and resource constraints, recruitment of fatigued patients terminated slightly early with 45 of an expected 48 patients recruited. However, the number of control subjects was larger than required. A total of 114 women completed the study (69 controls and 45 cases of CRFS). Non-participants (NP=102) were asked to complete the Chalder fatigue questionnaire and ethics approval was obtained to collect anonymised demographic and treatment data from these women. The main reasons for non-participation were cited as work (28%) or the need for extra visits (32%), while 12% mentioned other medical problems and the remainder either had ‘had enough’ or did not want to think about their cancer (28%). The NP had a significantly lower fatigue score (10.7 vs 13.1; p=0.04) and were older (62 years vs 57 years; p=0.005) than the participants. There were no differences on treatment variables, stage at diagnosis or time since diagnosis. Cases versus controls The overall prevalence of CRFS was 39% (45/114). A between-group comparison for all data is shown in tables 1–5 (table 1: treatment variables; table 2: questionnaire data; table 3: routine laboratory parameters; table 4: actigraphy data; and table 5: cognitive testing data). Table 1 Treatment and pathological variables

Cases versus controls The overall prevalence of CRFS was 39% (45/114). A between-group comparison for all data is shown in tables 1–5 (table 1: treatment variables; table 2: questionnaire data; table 3: routine laboratory parameters; table 4: actigraphy data; and table 5: cognitive testing data). Table 1 Treatment and pathological variables CRFS cases Controls Treatment variable Frequency Percentage Frequency Percentage p Value Conserving surgery 31 69 40 58 0.24 Non-conserving surgery 14 31 29 42 Chemotherapy 19 42 35 51 0.37 No chemotherapy 26 58 34 49 Radiotherapy 39 87 51 74 0.10 No radiotherapy 6 13 18 26 Hormone therapy 6 13 11 16 0.79 No hormone therapy 39 87 58 84 Lymph node positive 17 38 17 25 0.13 Lymph node negative 28 62 52 75 Oestrogen receptor positive 42 93 54 78 0.03 Oestrogen receptor negative 3 7 15 22 HER2 receptor positive 6 13 8 12 0.78 HER2 Receptor negative 39 87 61 88 CRFS, cancer-related fatigue syndrome. Table 2 Questionnaire data

CRFS cases Controls Treatment variable Frequency Percentage Frequency Percentage p Value Conserving surgery 31 69 40 58 0.24 Non-conserving surgery 14 31 29 42 Chemotherapy 19 42 35 51 0.37 No chemotherapy 26 58 34 49 Radiotherapy 39 87 51 74 0.10 No radiotherapy 6 13 18 26 Hormone therapy 6 13 11 16 0.79 No hormone therapy 39 87 58 84 Lymph node positive 17 38 17 25 0.13 Lymph node negative 28 62 52 75 Oestrogen receptor positive 42 93 54 78 0.03 Oestrogen receptor negative 3 7 15 22 HER2 receptor positive 6 13 8 12 0.78 HER2 Receptor negative 39 87 61 88 CRFS, cancer-related fatigue syndrome. Table 2 Questionnaire data CRFS cases Controls Variable Mean SD Mean SD p Value FS total 18.87 5.35 9.22 6.35 <0.001 FACT-F total 26.84 8.07 14.54 6.83 <0.001 ISI total 13.96 5.99 9.01 5.98 <0.001 HADS total 14.82 6.47 7.26 5.03 <0.001 VAS-F scale 6.44 1.53 4.73 2.29 <0.001 EORTC QLQ-30 subscales Physical functioning 69.19 19.45 85.80 14.00 <0.001 Global health status 81.09 16.80 56.07 22.79 <0.001 Emotional functioning 63.70 20.73 80.80 17.98 <0.001 Cognitive functioning 56.67 23.41 79.47 19.63 <0.001 Social functioning 59.26 27.65 84.81 19.69 <0.001 Fatigue 55.80 21.52 25.44 17.68 <0.001 Nausea/vomiting 7.04 13.98 2.42 6.57 0.04 Pain 37.04 33.88 16.91 21.86 0.01 Dyspnoea 22.96 27.36 11.59 19.66 0.02 Insomnia 48.15 32.22 37.68 32.29 0.09 Appetite loss 11.85 23.74 7.25 14.98 0.25 Constipation 12.59 23.88 6.28 14.32 0.15 Diarrhoea 2.96 9.59 3.86 13.45 0.68 Financial difficulties 31.85 34.78 9.18 20.52 <0.001 EORTC BR23 subscales Breast systemic therapy side effects 30.37 16.22 16.25 11.53 <0.001 Breast symptoms 38.89 18.63 20.05 16.69 <0.001 Arm symptoms 32.59 29.24 12.91 16.25 <0.001 The VAS-F is a visual analogue fatigue scale from 0 to 10. Values indicate average fatigue over the last 7 days.

.85 34.78 9.18 20.52 <0.001 EORTC BR23 subscales Breast systemic therapy side effects 30.37 16.22 16.25 11.53 <0.001 Breast symptoms 38.89 18.63 20.05 16.69 <0.001 Arm symptoms 32.59 29.24 12.91 16.25 <0.001 The VAS-F is a visual analogue fatigue scale from 0 to 10. Values indicate average fatigue over the last 7 days. The EORTC QLQ-30 (European Organisation for Research and Treatment of Cancer 30-item Quality of Life Questionnaire) has defined subscales of functioning and symptom scores. Higher scores on symptom subscales indicate greater symptom severity, and higher scores on functional subscales indicate better functioning. The EORTC BR23 is a breast specific module with functioning and symptom scales. CRFS, cancer-related fatigue syndrome; FACT-F, Functional Assessment of Cancer Therapy – fatigue subscale; FS, Chalder Fatigue Scale; HADS, Hospital Anxiety and Depression Scale; ISI, Insomnia Severity Index. Table 3 Routine laboratory data

The EORTC QLQ-30 (European Organisation for Research and Treatment of Cancer 30-item Quality of Life Questionnaire) has defined subscales of functioning and symptom scores. Higher scores on symptom subscales indicate greater symptom severity, and higher scores on functional subscales indicate better functioning. The EORTC BR23 is a breast specific module with functioning and symptom scales. CRFS, cancer-related fatigue syndrome; FACT-F, Functional Assessment of Cancer Therapy – fatigue subscale; FS, Chalder Fatigue Scale; HADS, Hospital Anxiety and Depression Scale; ISI, Insomnia Severity Index. Table 3 Routine laboratory data CRFS cases Controls Variable Laboratory reference range Mean SD Mean SD p Value Haemoglobin (g/dl) 12–16 12.9 0.78 12.9 0.91 0.94 White cell count (×109/l) 4–11 5.89 1.51 6.12 2.1 0.26 Neutrophil (×109/l) 1.7–8.0 3.54 1.08 4.41 4.7 0.17 Lymphocyte (×109/l) 1.0–4.0 1.84 0.71 1.80 0.68 0.79 Monocyte (×109/l) 0.24–1.1 0.36 0.16 0.37 0.12 0.60 Platelet (×109/l) 150–450 274 55.5 269 53.7 0.63 Sodium (mmol/l) 135–145 139.27 2.07 139.52 2.22 0.55 Potassium (mmol/l) 3.5–4.7 4.18 0.34 4.20 0.23 0.744 Urea (mmol/l) 2.5–8.0 5.48 1.19 5.74 1.53 0.32 Creatinine (µm/l) 60–110 64.10 8.29 66.59 8.20 0.13 Uncorrected calcium (mmol/l) 2.18–2.47 2.36 0.099 2.37 0.099 0.56 Adjusted calcium (mmol/l) NA 2.38 0.085 2.40 0.085 0.31 Phosphate (mmol/l) 0.75–1.50 1.12 0.15 1.08 0.20 0.24 Alkaline phosphatase (IU/l) 30–100 74.07 21.86 67.58 19.56 0.12 Alanine transaminase (IU/l) 0–40 25.16 11.16 27.72 14.17 0.32 Bilirubin (µmol/l) 0–17 6.81 3.15 7.35 2.98 0.37 Albumin (g/l) 35–48 38.67 2.90 38.56 2.06 0.83 GGT (IU/l) 0–30 29.33 23.54 30.49 24.92 0.81 TSH (mU/l) 0.4–4.0 2.63 3.31 3.25 5.05 0.50 Free T4 (pmol/l) 12–24 13.36 4.17 13.04 3.14 0.67 Magnesium (mmol/l) 0.74–1.03 0.86 0.68 0.86 0.08 0.74 CRFS, cancer-related fatigue syndrome; GGT, gamma-glutamyl transpeptidase; TSH, thyroid stimulating hormone.

90 38.56 2.06 0.83 GGT (IU/l) 0–30 29.33 23.54 30.49 24.92 0.81 TSH (mU/l) 0.4–4.0 2.63 3.31 3.25 5.05 0.50 Free T4 (pmol/l) 12–24 13.36 4.17 13.04 3.14 0.67 Magnesium (mmol/l) 0.74–1.03 0.86 0.68 0.86 0.08 0.74 CRFS, cancer-related fatigue syndrome; GGT, gamma-glutamyl transpeptidase; TSH, thyroid stimulating hormone. Table 4 Actigraphy data

90 38.56 2.06 0.83 GGT (IU/l) 0–30 29.33 23.54 30.49 24.92 0.81 TSH (mU/l) 0.4–4.0 2.63 3.31 3.25 5.05 0.50 Free T4 (pmol/l) 12–24 13.36 4.17 13.04 3.14 0.67 Magnesium (mmol/l) 0.74–1.03 0.86 0.68 0.86 0.08 0.74 CRFS, cancer-related fatigue syndrome; GGT, gamma-glutamyl transpeptidase; TSH, thyroid stimulating hormone. Table 4 Actigraphy data CRFS cases Controls Variable Mean SD Mean SD p Value Sleep Actual sleep time (h) 06:04 00:44:41 06:07 00:45:56 0.74 Actual sleep (%) 84.40 6.41 82.49 10.28 0.27 Actual wake time (h) 01:07 00:28:29 01:11 00:30:31 0.48 Actual wake (%) 17.51 10.27 15.60 6.41 0.28 Sleep efficiency (%) 80.22 7.73 78.44 13.28 0.43 Sleep latency (min) 00:17:32 00:20:27 00:17:19 00:27:36 0.97 Average number of calculated sleep bouts 23.73 6.60 24.04 7.02 0.82 Average number of calculated wake bouts 23.77 6.68 24.10 7.02 0.81 Mean sleep bout time (min) 00:33:03 00:44:41 00:19:20 00:13:41 0.02 Activity Mean wake bout time (min) 00:02:43 00:00:42 00:03:25 00:00:28 0.19 Average value of epoch intensity (24 h period) 208.48 87.22 245.82 96.32 0.04 Difference in maximum and minimum activity 198.85 69.70 218.55 88.97 0.45 Peak intensity of epoch 11.91 5.02 11.99 4.68 0.22 Average of daytime (light) activity 299.25 131.18 342.92 119.92 0.03 Time of night time (dark) activity 117.70 55.24 148.72 90.60 0.14 Ratio of light to dark activity 2.99 1.54 2.99 1.54 0.82 Circadian rhythm Interdaily stability 0.53 0.12 0.57 0.15 0.16 Interdaily variability 0.86 0.21 0.83 0.21 0.14 Least 5 (L5) active hours 1475 1010 2348 1634 0.50 Most 10 (M10) active hours 22481 7774 24865 8722 0.50 Relative amplitude 0.88 0.07 0.86 0.14 0.22 Sleep efficiency (%) is time asleep (as per algorithm) divided by total time in bed.

ily stability 0.53 0.12 0.57 0.15 0.16 Interdaily variability 0.86 0.21 0.83 0.21 0.14 Least 5 (L5) active hours 1475 1010 2348 1634 0.50 Most 10 (M10) active hours 22481 7774 24865 8722 0.50 Relative amplitude 0.88 0.07 0.86 0.14 0.22 Sleep efficiency (%) is time asleep (as per algorithm) divided by total time in bed. Sleep latency is time in bed before algorithm records the subject being asleep. Average value of epoch intensity (24 h period): the recording is converted to and analysed as a cosine wave around a zero point. Difference in maximum and minimum in activity corresponds to the difference between the peak and trough of the cosine wave. Interdaily stability is a measure of consistency in daily activity over 7 days. Interdaily variability is a measure of fragmentation of periods of rest and activity; normal value are <1 and calculated over 7 days. L5 indicates the 5 h of calculated lowest activity over the 7-day average. M10 indicates the 10 h of calculated greatest activity over the 7-day average. Relative amplitude is a measure of the difference between L5 and M10 and thus an indication of overall level of activity over the 7-day period. CRFS, cancer-related fatigue syndrome. Table 5 Cognitive testing data

L5 indicates the 5 h of calculated lowest activity over the 7-day average. M10 indicates the 10 h of calculated greatest activity over the 7-day average. Relative amplitude is a measure of the difference between L5 and M10 and thus an indication of overall level of activity over the 7-day period. CRFS, cancer-related fatigue syndrome. Table 5 Cognitive testing data CRFS cases Controls Variable Mean SD Mean SD p Value DMS percentage correct (all delays) 81.78 11.63 84.83 11.43 0.17 DMS mean correct latency (ms) 3528 1151 3707 1416 0.48 IED stages completed 7.40 2.40 7.33 2.40 0.89 IED total errors (adjusted) 50.37 51.71 52.65 53.43 0.83 MTS mean correct reaction time 3515 1143 3156 1137 0.10 MTS per cent correct 94.32 6.10 93.38 5.78 0.41 PAL stages completed 3.40 0.78 3.50 0.86 0.76 PAL total errors (adjusted) 95.60 29.67 92.90 32.05 0.65 PAL total trials (adjusted) 20.47 6.78 20.09 7.29 0.78 RVP A' (scores between 0 and 1) 0.87 0.06 0.89 0.07 0.15 RVP total correct rejections 236.69 25.83 245.53 14.95 0.02 RVP mean latency (ms) 533 127 468 126 0.009 VRM free recall – total correct 7.00 1.77 7.75 1.75 0.03 VRM recognition – total correct 22.78 1.38 23.06 1.42 0.3 AGN mean correct latency (ms) 532 71 517 61 0.24 AGN total omissions 5.17 1.99 4.57 1.78 0.64 AGN total commissions 7.76 1.75 6.57 1.72 0.03 AGN: Affective Go No-go (information processing and reaction time). There are three outcomes: mean correct latency (time taken for a correct response), total omissions which are incorrect responses to a target word (ie, the subject not pressing the button when they should have) and total commissions which are the total number of incorrect responses to the distracter words (ie, pressing the button inappropriately).

n correct latency (time taken for a correct response), total omissions which are incorrect responses to a target word (ie, the subject not pressing the button when they should have) and total commissions which are the total number of incorrect responses to the distracter words (ie, pressing the button inappropriately). DMS: Delayed Matching to Sample (memory and reaction time). There are two outcomes: overall correct responses (percentage) and latency of response (time taken to make correct response). IED: Intra-Extra Dimensional Set Shift (rule acquisition and reversal). There are two outcomes: stages complete is the number successfully completed and total errors (adjusted for stages) is a measure of the efficiency of completing the test. MTS: Match To Sample (matching test with speed/accuracy trade off). There are two outcomes: overall correct responses (percentage) and latency of response (time taken to make correct response). PAL: Paired Associates Learning (visual memory and new learning). There are two outcomes: total errors (adjusted from stages completed) and total trials (number of attempts needed to complete task adjusted for early abortion). RVP: Rapid Visual Information Processing (sensitive measure of general performance). There are three outcomes: A' is a probability score of how good a participant is at detecting sequences, correct rejection is the number of times a false sequence is ignored between correct ones and the mean latency is the time of response between the end of a sequence and the participant pressing the button.

ormance). There are three outcomes: A' is a probability score of how good a participant is at detecting sequences, correct rejection is the number of times a false sequence is ignored between correct ones and the mean latency is the time of response between the end of a sequence and the participant pressing the button. VRM: Verbal Recognition Memory (verbal information under free recall and forced choice recognitions). There are two outcomes: free recall number of correct words and a forced choice correct total (yes/no as to whether word was in the original list). CRFS, cancer-related fatigue syndrome. There were no significant differences between groups in mean age (CRFS 56.0 years vs controls 58.6 years; p=0.17), time since treatment (CRFS 12.4 months vs controls 13.0 months; p=0.67) or body mass index (CRFS 27.1 kg/m2 vs controls 26.0 kg/m2; p=0.29). There were no significant differences in treatment variables between groups (table 1). There was a significantly higher prevalence of oestrogen receptor negative tumours in the control group. There was no difference in the percentage of non-white participants between groups. There were significant differences in the levels of fatigue, mood disturbance and quality of life. There were also a number of differences in symptom subscales on both the EORTC QLQ-30 and BR23 (table 2). The prevalence of insomnia syndrome was significantly higher in the CRFS group (44% vs 16%; p=0.001). The routine laboratory variables (table 3), measured to exclude any gross metabolic disturbance, did not show any significant differences between groups.

There were significant differences in the levels of fatigue, mood disturbance and quality of life. There were also a number of differences in symptom subscales on both the EORTC QLQ-30 and BR23 (table 2). The prevalence of insomnia syndrome was significantly higher in the CRFS group (44% vs 16%; p=0.001). The routine laboratory variables (table 3), measured to exclude any gross metabolic disturbance, did not show any significant differences between groups. The actigraphy data (table 4) which provided a measure of sleep, activity and circadian rhythm, demonstrated some differences. There was a minor difference in sleep with the control group actually having significantly shorter sleep bouts. However, there were no differences in the major sleep parameters (overall time or quality). There were significant differences in average daytime activity and duration of daytime (light) activity. The cognitive data (table 5) demonstrated subtle but important differences between the groups. Significant differences were found on the RVP and VRM tests. The AGN test, a sensitive measure of executive function, demonstrated a higher number of commissions (ie, incorrect responses) in the CRFS group. The CRFS group performed worse than the controls in all statistically significantly different tests.

the groups. Significant differences were found on the RVP and VRM tests. The AGN test, a sensitive measure of executive function, demonstrated a higher number of commissions (ie, incorrect responses) in the CRFS group. The CRFS group performed worse than the controls in all statistically significantly different tests. Discussion This study is the first to objectively characterise cognitive differences between a rigorously defined group of cases of CRFS and a group of non-fatigued BCS. We found no differences in sleep actigraphy despite a much higher prevalence of insomnia syndrome in the CRFS group. However, mean daytime activity was lower in the CRFS group. These findings raise important questions about the nature of CRFS and its relationship to objective deficits in activity and cognition. The strengths of this study are the tight categorisation of post-treatment fatigue in BCS and the extensive cognitive and activity testing. However, the sample size is small and it is only a single centre study. The multiple analyses may have led to type I errors. This was a hypothesis generating study and was underpowered to detect all the identified differences between the groups, so the discussion and conclusions should be interpreted in that context. However, we feel these are interesting data and are further validated when placed in the context of previous findings.

errors. This was a hypothesis generating study and was underpowered to detect all the identified differences between the groups, so the discussion and conclusions should be interpreted in that context. However, we feel these are interesting data and are further validated when placed in the context of previous findings. The domains in which cognitive dysfunction occurs are the same as identified in previous reviews of breast cancer patients.19 The areas of executive function, processing speed and verbal memory are important for day-to-day tasks and may explain why these are associated with the subjective sensation of mental fatigue. Similar findings were found in a comparative study between subjects with CFS and cancer fatigue34 but not in a study comparing fatigued and non-fatigued BCS.20 However, in the latter study it should be noted that the ‘fatigued’ BCS group were defined using a cut-off on a continuous fatigue scale (rather than using a case-based definition) and the cognitive testing only included measures of concentration and reaction time (rather than the more comprehensive battery of tests employed in the current study). The authors failed to find differences between the fatigued and the non-fatigued group on objective cognitive testing. The objective evidence of cognitive impairment that we identified in this study appears to be consistent with the questionnaire data. The Chalder Fatigue Scale and the QLQ-30 both assess perceived cognitive impairment, and the scores on both of these questionnaires were significantly worse in the CRFS group.

e testing. The objective evidence of cognitive impairment that we identified in this study appears to be consistent with the questionnaire data. The Chalder Fatigue Scale and the QLQ-30 both assess perceived cognitive impairment, and the scores on both of these questionnaires were significantly worse in the CRFS group. One limitation of the cognitive assessments undertaken in this study was that they were only recorded at the end of treatment and were not measured prospectively from baseline. However, we were not specifically examining cognitive change per se following treatment but comparing CRFS and matched controls. The purpose was to identify potential cognitive deficits in the CRFS group. The exclusion of psychiatric co-morbidities reinforces the hypothesis that CRFS is associated with a defined set of cognitive deficits that are unrelated to mood disturbance. The underlying mechanism of cognitive dysfunction is unclear and may be related to an increase in circulating inflammatory cytokines35 but seems unlikely to be related to a gross hypothalamic pituitary axis disturbance.36 Functional imaging studies in this population would help to further elucidate the mechanism. The association between subjective fatigue and objective cognitive deficits is also interesting in relation to the finding that psychostimulants can reduce cancer fatigue during treatment.37 However, they are unlikely to be suitable for longer term use in BCS38 because of concerns related to adverse effects and the risk of addiction.

between subjective fatigue and objective cognitive deficits is also interesting in relation to the finding that psychostimulants can reduce cancer fatigue during treatment.37 However, they are unlikely to be suitable for longer term use in BCS38 because of concerns related to adverse effects and the risk of addiction. The higher prevalence of insomnia syndrome in the CRFS group was linked to higher scores on the ISI. However, this subjective perception of difficulty with sleeping did not correlate with objective actigraphy measures of sleep. While 44% of CRFS meet the criteria for insomnia syndrome, they are clearly two separate entities and CRFS cannot be explained by sleep disturbances alone. This finding is supported by data from an intervention trial for insomnia that improved sleep but not fatigue.39

relate with objective actigraphy measures of sleep. While 44% of CRFS meet the criteria for insomnia syndrome, they are clearly two separate entities and CRFS cannot be explained by sleep disturbances alone. This finding is supported by data from an intervention trial for insomnia that improved sleep but not fatigue.39 We found significant differences between groups with respect to objective activity data. This is in keeping with a recent study40 in breast cancer but in conflict with our earlier findings.16 The reason for the disparity is unclear but may be due to the different actigraph used and associated analysis algorithm. The differences were seen in the average epoch score (a measure of movement intensity) and in the level of daytime activity. These data confirm that patients who fulfil the criteria for CRFS, as well as being subjectively more fatigued, also have evidence of decreased physical activity. The decreased activity may be a direct consequence of feeling subjectively fatigued, or the decreased activity may arise first (due to a variety of causes) and may be a causative factor in exacerbating or perpetuating the subjective fatigue that these patients experience.41 One way to disentangle these phenomena would be to undertake a longitudinal study to assess both subjective and objective manifestations of fatigue and to trace the temporal relationships between them.40 The authors of the study40 did not use diagnostic criteria to measure fatigue. The theory that, whatever the cause of the initial fatigue, it is perpetuated by physical deconditioning and decreased activity probably explains why exercise is reported to be an effective treatment for cancer related fatigue.42 It is likely, however, that the effects of exercise are relatively non-specific (exercise reduces fatigue even in healthy individuals) and this explains the relatively small effect size for exercise interventions in cancer populations.

why exercise is reported to be an effective treatment for cancer related fatigue.42 It is likely, however, that the effects of exercise are relatively non-specific (exercise reduces fatigue even in healthy individuals) and this explains the relatively small effect size for exercise interventions in cancer populations. In a previous study in BCS we found the prevalence of CRFS to be 30%16 and this general figure was supported by the results of a systematic review.7 The prevalence of 39% for CRFS that we found in this study although slightly higher than previous estimates is broadly in keeping with these figures and may be explained by the finding that the non-participant group were significantly less fatigued16 than the study sample. A prevalence of 30–39% represents a sizeable percentage of women who have successfully completed treatment for breast cancer and potentially translates into a large absolute number of BCS. This emphasises that it is important to recognise and treat fatigue. The widespread differences in questionnaire data and levels of other symptoms are in keeping with previous studies.12 We found that the CRFS group had a higher level of mood disturbance despite the exclusion of patients with specific psychiatric diagnoses. It is possible that sub-threshold mood disturbance contributes to CRFS. However fatigue and depression should be regarded as separate entities.43

in keeping with previous studies.12 We found that the CRFS group had a higher level of mood disturbance despite the exclusion of patients with specific psychiatric diagnoses. It is possible that sub-threshold mood disturbance contributes to CRFS. However fatigue and depression should be regarded as separate entities.43 There are some important negative findings in this study. There were no significant differences in any of the routine laboratory tests. While fatigue may be linked to anaemia during chemotherapy,44 this does not appear to be the case in BCS. There is also no link to thyroid disturbances. We had postulated that serum magnesium levels may be contributory but this was not found to be the case. There were also no differences in treatment or demographic variables between groups. This is consistent with a critical appraisal45 in which the authors found very limited associations between fatigue and these variables.

There are some important negative findings in this study. There were no significant differences in any of the routine laboratory tests. While fatigue may be linked to anaemia during chemotherapy,44 this does not appear to be the case in BCS. There is also no link to thyroid disturbances. We had postulated that serum magnesium levels may be contributory but this was not found to be the case. There were also no differences in treatment or demographic variables between groups. This is consistent with a critical appraisal45 in which the authors found very limited associations between fatigue and these variables. The clinical significance of these findings reinforces the importance of identifying women who have or are at risk of developing CRFS. This should be incorporated into routine practice and might include counselling on potential cognitive and sleep changes. These problems might be diminished with simple information giving at the start of treatment,46 which would in keeping with the Department of Health survivorship strategy.47 However, we are not able to recommend any particular treatment strategy at present in this group.7 In conclusion, we have found that a significant minority of women successfully treated for breast cancer meet the criteria for CRFS. This group is characterised by subjective poly-symptomology. There are important differences between the groups with objectively lower cognitive function and daytime activity in the CRFS group that need further confirmation with a larger sample size. Future work should examine this relationship longitudinally, possibly within a quality of life arm of a treatment trial. Correlation with levels of cytokines and/or functional imaging would be helpful to further characterise this group. The overall aim should be to design more targeted treatments to manage this problematic syndrome.

ould examine this relationship longitudinally, possibly within a quality of life arm of a treatment trial. Correlation with levels of cytokines and/or functional imaging would be helpful to further characterise this group. The overall aim should be to design more targeted treatments to manage this problematic syndrome. The authors would like to thank Sue Lowndes, Dr Cathy Coleman and Joe Diffley for their help in accessing the patients from the breast follow-up clinic they run. The authors would also like to thank Jo Passley and the other members of the St George's clinical research facility for their help in conducting the study. The authors also thank Dr Fiona Lofts, Mr Anup Sharma and Mr Dibeyshwar Banerjee for their help in setting up the study. Contributors: OM and PS both contributed equally to the manuscript. OM was responsible for data collection and analysis. OM and PS both obtained funding for the study. Funding: This study was funded by Cancer Research UK Grant number C31193/A10090. Competing interests: None. Ethics approval: Wandsworth research ethics committee approved this study (ref 08/H0803/182). Provenance and peer review: Not commissioned; externally peer reviewed.

Background There is consensus that people who are frail or have a specific terminal diagnosis should be provided with opportunities to discuss their future care and treatment preferences.1–3 Conversations about these matters form part of advance care planning discussions—these aim to help individuals anticipate how their condition may affect them in the future and, if they wish, to set on record their preferences, choices and advance decisions to refuse treatment.4 However, there is evidence that in practice, patients and healthcare professionals rarely discuss future difficulties and plans.56 Surveys of clinicians have found that they report lack of confidence, skills and knowledge about how to initiate such discussions, making them reticent to do so.7 Clinicians also report that these conversations can be uncomfortable and challenging,8 with difficulties including ascertaining patients’ preferences, and deciding on the ‘right time’ to raise such matters.7 These challenges are exacerbated by patients’ and their companions’ own reticence and ambivalence with regard to initiating these conversations.7 Clinicians want9 and need10 more evidence-based guidance on optimal strategies for communicating about these matters with patients and their significant others. While practitioners are encouraged to provide those who consult them with opportunities to engage in advance care planning, guidelines also stress that this should be a voluntary engagement,4 and there is evidence that some patients strongly prefer not to discuss such matters.11 Practitioners thus also need to know how to sensitively ascertain and respond to patients’ preferences in this regard.

rtunities to engage in advance care planning, guidelines also stress that this should be a voluntary engagement,4 and there is evidence that some patients strongly prefer not to discuss such matters.11 Practitioners thus also need to know how to sensitively ascertain and respond to patients’ preferences in this regard. In terms of precisely how to go about these conversations, a highly relevant body of evidence has been produced by studies that use the communication science approaches of conversation analysis12 and discourse analysis.13 These techniques involve inductively analysing audio or audiovisually recorded naturalistic interactions (rather than simulated or experimental ones). Conversation analytic studies are particularly useful because they entail analysing ‘sequences’—that is, how what is said affects what happens next, also because they allow systematic analysis not just of what is said, but also how it is said—including intonation, pauses, gaze and gestures.12 14 Doing so generates detailed understandings about communication practices and how they function at a level of detail unavailable through qualitative interview studies of communication,14 but to date findings have not been incorporated into clinical guidance and training.

uding intonation, pauses, gaze and gestures.12 14 Doing so generates detailed understandings about communication practices and how they function at a level of detail unavailable through qualitative interview studies of communication,14 but to date findings have not been incorporated into clinical guidance and training. Our aim was to gather and synthesise evidence from conversation and discourse analytic studies of how people address difficult and uncertain future matters in their healthcare-related conversations. By difficult and uncertain future matters, we mean future difficulties in one's personal life, future illness progression, loss of capacity, dying and death. To avoid cumbersome wording, we refer to these as ‘difficult future topics’; also, we refer to ‘patients’, although most of the evidence also applies to ways in which patients’ significant others communicate. The overall purpose of the review was to inform practice, education and training in clinical decision-making and communication. We designed our review so as to locate and synthesise studies conducted in a wide range of healthcare-related contexts. We did so because of the recognition that people use highly systematic and recurrent practices and conventions for communicating with one another: it is because of their shared and ubiquitous character that we can understand and communicate with people we have never met before, in situations we have never been in before. Thus, each unique communication episode—be it a telephone call with one's parents, a counselling session or a medical consultation—is built on a bedrock of common practices for conversing together in an orderly and meaningful way.15 While the specific content of our communication is unique to each interaction, the practices people use to ‘handle’ this content are recurrent and shared. Recognising this, we cast our net wide when gathering evidence about how people talk about difficult and uncertain future matters in the context of healthcare. When synthesising the evidence we found, we focused specifically on its implications for communicating about illness progression and end of life.

nt and shared. Recognising this, we cast our net wide when gathering evidence about how people talk about difficult and uncertain future matters in the context of healthcare. When synthesising the evidence we found, we focused specifically on its implications for communicating about illness progression and end of life. Methods We drew extensively on existing systematic review approaches, but revised these and developed additional procedures because the methods and findings of conversation and discourse analytic studies differ considerably from both quantitative and qualitative research approaches. A detailed report of our methodological approach has been published,16 and includes discussion of the reasoning and rationale underpinning our methods and choices. We drafted a protocol on the basis of contemporary guidance,17 18 refined it in discussion with academics, practitioners and trainers, and registered it on the PROSPERO database.19

methodological approach has been published,16 and includes discussion of the reasoning and rationale underpinning our methods and choices. We drafted a protocol on the basis of contemporary guidance,17 18 refined it in discussion with academics, practitioners and trainers, and registered it on the PROSPERO database.19 Study selection We included peer-reviewed journal articles and published book chapters reporting conversation and discourse analyses of audio or audiovisually recorded naturalistic interactions in English. Studies relying solely on coding frameworks were excluded. We included only studies where both data and analysis concerned communication about sensitive future matters. We sought studies where the talk was healthcare-related in both professional settings, for example, clinics, general practice surgeries, counselling; and informal ones, for example, family telephone calls. We included both conversation analytic and discourse analytic studies. As we have discussed elsewhere,16 discourse analysis is an umbrella term encompassing a wide range of approaches to analysing texts and talk—including qualitative content analysis. In contrast, conversation analysis is a single, defined and bounded research approach with a specific conceptual basis and method. As noted, an important distinguishing feature and advantage of conversation analysis is that the sequential analysis entailed allows empirical evidencing of the interactional consequences of particular practices. This means that conversation analytic research can show that particular ways of, for instance, asking about a patient's concerns, or making a treatment recommendation, are recurrently followed by particular responses from patients.

d allows empirical evidencing of the interactional consequences of particular practices. This means that conversation analytic research can show that particular ways of, for instance, asking about a patient's concerns, or making a treatment recommendation, are recurrently followed by particular responses from patients. One reviewer (VL) carried out the search and initial screening of titles and abstracts, excluding clearly irrelevant publications at this stage. The remaining publications were independently considered for inclusion by two reviewers (VL and RP), resolving disagreements by discussion. Information sources We searched diverse sources17 20: existing knowledge within the review team and its contacts; online discussion lists; online specialist bibliographies; and the citations and reference lists of the publications we identified through other means. The electronic databases searched were: ISI Web of Science, Amed, EMBASE, CINAHL, MEDLINE, PsycINFO, ASSIA, Sociological Abstracts CSA and Scopus. We did not restrict the date of publication. A first round of electronic database searching was completed on 25 May 2011; a second round, updating the first, was completed on 1 May 2014. The search strategy is available online.21

ce, Amed, EMBASE, CINAHL, MEDLINE, PsycINFO, ASSIA, Sociological Abstracts CSA and Scopus. We did not restrict the date of publication. A first round of electronic database searching was completed on 25 May 2011; a second round, updating the first, was completed on 1 May 2014. The search strategy is available online.21 Data collection, appraisal and synthesis We developed, piloted and refined a data extraction form. This is available online,21 as is a form used in assessing the characteristics and contribution of individual studies.21 As we have discussed elsewhere,16 the kind of research we reviewed needs to be appraised in terms of two broad dimensions: (1) the type and amount of data analysed, and (2) the detail and depth of analysis. Some studies involve examining a relatively small amount of data in great depth and provide highly detailed evidence about the structure, locations, functioning and mechanism of a particular practice. Some studies entail a less detailed analysis of structure and functioning, but nevertheless contribute evidence about how widespread and recurrent a practice is. We synthesised the extracted data using an aggregative approach, drawing together findings through structured summaries and comparing and connecting findings with one another.22 We developed and refined emerging findings by discussing them and their potential implications with clinician and educator audiences.

tice is. We synthesised the extracted data using an aggregative approach, drawing together findings through structured summaries and comparing and connecting findings with one another.22 We developed and refined emerging findings by discussing them and their potential implications with clinician and educator audiences. Results Of 2887 records obtained, 19 publications23–41 reporting findings from 13 research programmes met the inclusion criteria. (These figures differ slightly from earlier reports16 42 because our updating search in 2014 yielded two further studies23 29 for inclusion, and because we removed one publication43 initially included as it lacked a healthcare context.) The flow chart in figure 1 enumerates the publications identified, excluded and included. Ten of the 19 included papers concerned medical consultations: inpatient palliative care team consultations29; cardiology23; oncology clinics25 26 33 34; primary care27 40 and gender identity psychiatric assessment.38 39 Seven were on counselling sessions: HIV,30–32 37 genetics,35 36 and combined HIV and family therapy data sets.28 There was one study each from: family telephone conversations,24 and informed consent appointments for a randomised controlled trial.41 Figure 1 Study screening.

Results Of 2887 records obtained, 19 publications23–41 reporting findings from 13 research programmes met the inclusion criteria. (These figures differ slightly from earlier reports16 42 because our updating search in 2014 yielded two further studies23 29 for inclusion, and because we removed one publication43 initially included as it lacked a healthcare context.) The flow chart in figure 1 enumerates the publications identified, excluded and included. Ten of the 19 included papers concerned medical consultations: inpatient palliative care team consultations29; cardiology23; oncology clinics25 26 33 34; primary care27 40 and gender identity psychiatric assessment.38 39 Seven were on counselling sessions: HIV,30–32 37 genetics,35 36 and combined HIV and family therapy data sets.28 There was one study each from: family telephone conversations,24 and informed consent appointments for a randomised controlled trial.41 Figure 1 Study screening. Online supplementary table S1 summarises each study’s characteristics, including the type, amount and range of the data on which the analysis was based, and the depth and detail of the analytic techniques applied. In all but the family phone calls study, analysis was largely focused on professionals’ communication behaviours and patients’ responses to these. Analysis in three publications33 34 40 only examined individual spoken turns, and did not systematically analyse how they were responded to. In two others, there was some limited attention to sequencing in terms of how clinicians’ practices shaped and related to how patients responded, and most of the empirical data shown in these publications comprised clinicians’ talk.23 29 In the remaining 14 publications, analysis examined in varying degrees of depth the structure, functioning and mechanism of communication practices by examining sequences of talk. In 9 publications, analysis examined only the word content, that is, what was said; in the other 10, analysis examined other aspects, that is, how it was said—including intonation, volume, hesitancy and pauses—aspects known to be important in how people convey meaning within their own communications and make sense of what others say.14 Several studies used video data, but only one26 attended to body movement (eg, gaze or touch), and even then only very briefly.

w it was said—including intonation, volume, hesitancy and pauses—aspects known to be important in how people convey meaning within their own communications and make sense of what others say.14 Several studies used video data, but only one26 attended to body movement (eg, gaze or touch), and even then only very briefly. Two papers specifically examined communication in relation to aspects of specialist palliative care29 and advanced care planning,23 settings which are particularly relevant to the overall purpose of our review. Unfortunately, some features of these studies reported meant that they were only able to make a limited contribution to our review: the data comprised audio rather than audiovisual recordings; the qualitative analytic approaches used meant that the detail about practices’ structure and functioning was less fine-grained than is possible via conversation analytic methods; and limited information was provided about the prevalence and recurrence of the identified practices.

io rather than audiovisual recordings; the qualitative analytic approaches used meant that the detail about practices’ structure and functioning was less fine-grained than is possible via conversation analytic methods; and limited information was provided about the prevalence and recurrence of the identified practices. Details of individual study findings are given in online supplementary table S2. We sorted communication practices that had been documented in two or more publications into categories. These spanned practices used in initially broaching or attempting to elicit talk on difficult future issues, those used in progressing talk on the topics once they had been raised, and those which closed the topics. While for the purposes of this review we synthesised findings into discrete sets of practices, it is worth noting that in the real world, of course, professionals improvise and apply communication practices in numerous combinations.44 The eight categories of practices are summarised in box 1; in the text, we describe them in more detail and give some illustrative examples. Box 1 Summary findings: practices for communicating about difficult future topics documented in >1 publication FISHING QUESTIONS Opening questions which seek patients’ perspectives and which may or may not hint at possible difficult topics via terms like ‘concerns’ or ‘problems’, and which do not specifically refer to difficult future topics.

Details of individual study findings are given in online supplementary table S2. We sorted communication practices that had been documented in two or more publications into categories. These spanned practices used in initially broaching or attempting to elicit talk on difficult future issues, those used in progressing talk on the topics once they had been raised, and those which closed the topics. While for the purposes of this review we synthesised findings into discrete sets of practices, it is worth noting that in the real world, of course, professionals improvise and apply communication practices in numerous combinations.44 The eight categories of practices are summarised in box 1; in the text, we describe them in more detail and give some illustrative examples. Box 1 Summary findings: practices for communicating about difficult future topics documented in >1 publication FISHING QUESTIONS Opening questions which seek patients’ perspectives and which may or may not hint at possible difficult topics via terms like ‘concerns’ or ‘problems’, and which do not specifically refer to difficult future topics. Functioning: patients rarely provide responses that bring topics such as illness progression, imminence of end of life, etc. to the surface, that is, these questions rarely ‘catch the fish’. Documented in five publications.26 28 31 32 39 INDIRECTNESS, ALLUSIVE TALK AND EUPHEMISMS

Opening questions which seek patients’ perspectives and which may or may not hint at possible difficult topics via terms like ‘concerns’ or ‘problems’, and which do not specifically refer to difficult future topics. Functioning: patients rarely provide responses that bring topics such as illness progression, imminence of end of life, etc. to the surface, that is, these questions rarely ‘catch the fish’. Documented in five publications.26 28 31 32 39 INDIRECTNESS, ALLUSIVE TALK AND EUPHEMISMS Referring to difficult future topics indirectly via practices such as: referring to outcomes of treatment rather than outcomes of illness, euphemisms, expressing emotional concern for patient, referring to ‘it’ or ‘that’, to getting ‘very, very ill’ or to ‘something serious’, referring to people in general rather than this particular patient. Patients respond by shifting away from the topic, or by displaying stoicism—with little verbal response, or by showing some understanding of the underlying difficult future topic.26 When the latter two response types occur, the professional can then move progressively towards more direct engagement with the difficult future topics.26 Functioning: can encourage engagement with the topics, and can be early moves in a trajectory towards greater openness. At the same time, they make it particularly easy for patients to deflect away from the difficult topics. For this reason, indirectness may be a useful way to gently ascertain whether a patient prefers not to discuss these matters (yet), and to abide by this preference. Documented in six publications.23 26 27 33 34 40

Functioning: can encourage engagement with the topics, and can be early moves in a trajectory towards greater openness. At the same time, they make it particularly easy for patients to deflect away from the difficult topics. For this reason, indirectness may be a useful way to gently ascertain whether a patient prefers not to discuss these matters (yet), and to abide by this preference. Documented in six publications.23 26 27 33 34 40 LINKING QUESTIONS AND PROPOSALS TO WHAT THE PATIENT HAS SAID OR NOT SAID When posing a question, or repeating or rephrasing an unanswered question, or making proposals about future plans, the professional refers back to something the patient has already raised or hinted at, or has noticeably not said. Functioning: This treats the patient as already having shown that the difficult future topics are relevant and thus something they want to and should engage with. It allows the practitioner to show that they have been paying attention to the patient, and to selectively focus in on difficult future topics. Documented in seven publications.28–32 39 41 HYPOTHETICAL QUESTIONS AND TALK Hypothetical questions entail describing a hypothetical future situation then enquiring about related feelings or plans. Hypothetical talk is phrased hesitantly and sensitively, and in ways that: (A) convey it as touched off by or connected with what the patient has already said, and (B) emphasise its hypothetical nature. Functioning: Patients usually respond by engaging with the difficult future topics. Documented in 12 publications.23 26 28 30–32 35–40

Hypothetical questions entail describing a hypothetical future situation then enquiring about related feelings or plans. Hypothetical talk is phrased hesitantly and sensitively, and in ways that: (A) convey it as touched off by or connected with what the patient has already said, and (B) emphasise its hypothetical nature. Functioning: Patients usually respond by engaging with the difficult future topics. Documented in 12 publications.23 26 28 30–32 35–40 FRAMING THE DIFFICULT ISSUE AS UNIVERSAL OR GENERAL RATHER THAN INDIVIDUAL TO THIS PATIENT The difficult future situation is framed as universal, as something anyone could face. Functioning: Its effectiveness in encouraging talk about difficult future topics is not examined in existing evidence. However, it seems probable that, like hypothetical talk, this downplays or distances the reality and seriousness of the difficult issues for this individual patient, thereby making it more likely that patients will engage with the topics. Documented in four publications.31 33 34 37 COMMUNICATION PRACTICES OTHER THAN WORDS THAT CONVEY SENSITIVITY Delays, hesitations and turbulence, for example, stuttering and repeats when broaching and talking about the difficult issues. Both professionals and patients do so. One study30 provides evidence that when doctors’ talk is turbulent, patients sustain close attention to the doctor. Functioning: This kind of talk conveys the sensitivity and seriousness of the topics. Documented in four publications.26 30 31 37

Delays, hesitations and turbulence, for example, stuttering and repeats when broaching and talking about the difficult issues. Both professionals and patients do so. One study30 provides evidence that when doctors’ talk is turbulent, patients sustain close attention to the doctor. Functioning: This kind of talk conveys the sensitivity and seriousness of the topics. Documented in four publications.26 30 31 37 Also within this category is touch, which is only mentioned in one study.26 This describes how a doctor touches a patient’s hand at the same time as indirectly talking about difficult future issues. Functioning: Touch may convey support and comfort while also conveying the seriousness of the topic(s). However, patients’ responses have not been empirically studied. COMMUNICATION PRACTICES OTHER THAN WORDS THAT ENCOURAGE PATIENTS TO TALK ONCE THE TOPIC HAS BEEN BROACHED Once a difficult future topic has been raised by a professional via some form of question, they avoid producing further talk until the patient has responded—even if this means leaving gaps of silence longer than is common in regular conversations. Also, once the patient has begun to respond, the clinician speaks only briefly and in ways that encourage the patient to continue, for example, by saying ‘Mm’. Functioning: Generally, people have a low tolerance for silences within conversations, so this encourages patients to talk on the difficult issues. Documented in two publications.37 41 Shifting to the positive

Once a difficult future topic has been raised by a professional via some form of question, they avoid producing further talk until the patient has responded—even if this means leaving gaps of silence longer than is common in regular conversations. Also, once the patient has begun to respond, the clinician speaks only briefly and in ways that encourage the patient to continue, for example, by saying ‘Mm’. Functioning: Generally, people have a low tolerance for silences within conversations, so this encourages patients to talk on the difficult issues. Documented in two publications.37 41 Shifting to the positive ‘Upbeat’ talk of various forms: referring to fighting, perseverance and hope, facing illness together, giving optimistic rather than pessimistic versions, and referring to treatment success as opposed to overall trajectory of the illness. Both patients and professionals do so. One paper25 documents a recurrent pattern wherein oncologists follow bad or uncertain news with relatively hopeful or positive news, and patients respond by focusing on the latter, positive element. Functioning: Upbeat talk conveys optimism and hope and also contributes to building and maintaining therapeutic relationships. But it can also suppress further talk on the difficult future topics, and thus limit engagement with them. Documented in three publications.24 25 27

‘Upbeat’ talk of various forms: referring to fighting, perseverance and hope, facing illness together, giving optimistic rather than pessimistic versions, and referring to treatment success as opposed to overall trajectory of the illness. Both patients and professionals do so. One paper25 documents a recurrent pattern wherein oncologists follow bad or uncertain news with relatively hopeful or positive news, and patients respond by focusing on the latter, positive element. Functioning: Upbeat talk conveys optimism and hope and also contributes to building and maintaining therapeutic relationships. But it can also suppress further talk on the difficult future topics, and thus limit engagement with them. Documented in three publications.24 25 27 Fishing questions The practice of asking fishing questions is documented in studies from counselling28 31 32 39 and oncology26 contexts. Often coming towards the start of conversations, these are open questions—in the sense that they do not give any pointers towards answers that specifically concern difficult future topics: for example, Counsellor: “Graham, are there any issues you’d like to discuss?” (Extract 8)32; “Can I just ask you what are your greatest concerns Liz?” (Extract 1).31 This question structure allows patients easily to avoid embarking on the difficult future topics by answering in terms of other matters. We term these fishing questions because of the way they fish for, but do not directly target, responses about difficult future topics. Studies examining the structure and functioning of fishing questions in the greatest detail26 32 suggest that they are usually ineffective in eliciting talk about difficult future matters. However, it is argued32 that there is nevertheless a rationale for using them—if the patient does opt to answer in terms of the difficult future topics, then this puts on record their personal commitment to talk about the topic(s), and this would not be the case were the professional to be the first to raise them.

ers. However, it is argued32 that there is nevertheless a rationale for using them—if the patient does opt to answer in terms of the difficult future topics, then this puts on record their personal commitment to talk about the topic(s), and this would not be the case were the professional to be the first to raise them. Indirectness, allusive talk and euphemisms Four publications from oncology consultations,26 33 34 40 one from primary care27 and one from cardiology23 examine indirect, allusive, euphemistic and vague forms of talk. This category encompasses various means by which both professionals and those who consult them refer to difficult future topics in an indirect manner. One of these involves doctors and patients focusing on treatment outcomes rather than disease outcomes,33–34 for example, Oncologist to a patient with terminal cancer: “I know that the surgical doctors have been pretty happy with how quickly you’ve recovered” (Extract 1A).26 Other practices include referring to topics that are associated with illness progression and end of life but not directly on-topic—for instance, hospice care26; using euphemisms for death, for example, ‘pushing up daisies’33; and using various forms of talk that allow patients to draw their own conclusions through: (A) expressing emotional concerns and hinting at difficult future developments: for example, Oncologist: “I’m concerned that, you know, there’s a possibility that things might not go so well for you” (Extract 1D)26; (B) referring to palliative rather than curative treatment; and (C) phrases known as litotes—these suggest the worse scenario by negating the better scenario, for example, Oncologist: “Do you think there’s a possibility that you might not get better?” (Extract 1C)26—perceptibly less harsh than alternatives such as “Do you think you might get worse/die?”. Another well-documented form of indirect talk used by both doctors and patients involves impersonal reference, for example, referring to the disease as ‘It’ or ‘That’26 34 rather than ‘My disease’ or ‘Your disease’; and avoiding direct reference to the individual patient, for example, Oncologist: “We can rarely cure this type of problem. What we can do, though, is shrink the mass. It will sometimes go away completely, and we can increase the life span and the quality of life, but the tumor will usually come back.”33

Your disease’; and avoiding direct reference to the individual patient, for example, Oncologist: “We can rarely cure this type of problem. What we can do, though, is shrink the mass. It will sometimes go away completely, and we can increase the life span and the quality of life, but the tumor will usually come back.”33 Only one study26 analyses the effects of indirect talk in terms of how it gets responded to, and whether and how it promotes or limits talk on the difficult future topics. (Notice that this particular study examined indirect talk used in a doctor's early moves towards difficult future topics, not occasions where doctors gave indirect responses to patients’ direct questions.) This study found that patients responded to the doctor's indirect references to difficult future matters in three alternative ways: (1) shifting away from the sensitive future topic, (2) displaying stoic unresponsiveness and (3) responding in a way that reveals an understanding of the underlying concerns—such as illness progression. When patients gave stoic responses, or ones revealing some understanding and recognition, the doctor took the opportunity to progressively ‘unpack’ the alluded to matters in a step-by-step manner so that both the patient and doctor ended up talking more directly about the difficult future matters.26

llness progression. When patients gave stoic responses, or ones revealing some understanding and recognition, the doctor took the opportunity to progressively ‘unpack’ the alluded to matters in a step-by-step manner so that both the patient and doctor ended up talking more directly about the difficult future matters.26 In sum, the available evidence indicates that indirect talk is recurrent in medical consultations and is used by both doctors and those who consult them. It allows people to very easily avoid talking about the difficult topics for the following reason: when one talks indirectly about difficult issues, one avoids putting one’s conversational partner in a position of having to voice some excuse for not taking them up. So, indirect talk gives people an easy route to deflect away from or avoid the difficult and sensitive topics that have been implied, and some patients and relatives do so.26 On the other hand, indirect talk can also function as the first step in a gradual progression towards directly talking about the difficult future issues.

rect talk gives people an easy route to deflect away from or avoid the difficult and sensitive topics that have been implied, and some patients and relatives do so.26 On the other hand, indirect talk can also function as the first step in a gradual progression towards directly talking about the difficult future issues. Overtly linking questions and proposals with what patients have already said, or have not said Studies in palliative care consultations,29 psychiatry,39 counselling settings28 30–32 and in drug trial recruitment41 report that when professionals ask questions about topics related to difficult futures, they often preface their questions by explicitly referring back to what are sometimes termed patients’ cues45: something a patient has themselves raised or hinted at. One way to do so28 30 involves the professional first issuing a summary or interpretation of what the patient has said beforehand, and then asking a related question. Professionals do this when they pose new questions, when dealing with a patient’s reluctance to answer a previous question, and when making proposals about future plans. Sometimes they link back to what a patient has mentioned immediately beforehand, sometimes to something said a good while earlier, for example, Counsellor: “What Michael mentioned initially is that you know he didn’t have any concerns but he’s been feeling funny again, is that a worry to him or is it no problem to him…” (Extract 19).32 Professionals can also make links to a matter noticeably absent from the patient’s earlier talk, for example, Counsellor: “You haven’t mentioned AIDS as a concern today. How much of a concern is it now?” (Extract 21).32 (This and some other examples are drawn from recordings made in HIV/AIDS clinics at a time when HIV constituted a terminal diagnosis.)

to a matter noticeably absent from the patient’s earlier talk, for example, Counsellor: “You haven’t mentioned AIDS as a concern today. How much of a concern is it now?” (Extract 21).32 (This and some other examples are drawn from recordings made in HIV/AIDS clinics at a time when HIV constituted a terminal diagnosis.) This linking practice has been documented to be effective in encouraging patients’ engagement in talk about future issues. It works by making it difficult for a patient to avoid answering a question or talking about a plan because the question or plan is presented as something the patient has themselves raised—and thus should engage with.32 In addition, linking allows professionals to show their appreciation of, attention to, and agreement with what patients have said,28 and to avoid going on record as unilaterally imposing the difficult future topic on the conversation.31 When linking is used in the context of talking about plans for the future, it can work to construct an atmosphere of agreement and convey the patient as the one who has the lead responsibility in terms of making plans and proposals.28

on record as unilaterally imposing the difficult future topic on the conversation.31 When linking is used in the context of talking about plans for the future, it can work to construct an atmosphere of agreement and convey the patient as the one who has the lead responsibility in terms of making plans and proposals.28 Hypothetical questions and talk Practices involving talking in hypothetical terms and posing hypothetical questions have been documented across several settings: cardiology clinics,23 counselling and psychiatry,28 30–32 35–39 and oncology.26 40 The basic structure of a hypothetical question entails the speaker describing a hypothetical future situation and following this with an enquiry about associated perspectives, feelings or plans. The evidence indicates that these questions are highly effective in encouraging patients to engage with difficult issues. Specifically, this evidence relates to hypothetical questions that are31: (1) posed part-way through conversations, usually after previously attempting to elicit talk on the difficult future topics through other practices such as fishing questions and/or indirect talk; (2) voiced carefully and hesitantly—thus conveying that what is talked about is both serious and sensitive; (3) phrased as touched off by or connected with what the patient has already said and (4) emphasised as being hypothetical in their nature. Thus, in the following example, the counsellor’s first ‘Say’ is spoken in a long, stretched way and the hypothetical and uncertain nature of the talk is very much emphasised: Counsellor: “Saaaaaay, we can’t say and you can’t say,’ Patient: ‘Yeah’ Counsellor: ‘but say you did begin to get ill (pause) or say you got so ill that you couldn’t kind of (pause) make decisions for yourself. Who would (pause) you have to make them for you? (pause) Who do you (pause) consider your…” (Extract 4)31; similar features can be seen here: Counsellor: “If you, if you- supposing- I mean this is just supposing, supposing you (pause) had got infected or were to get infected…” (Extract 8).31

ake decisions for yourself. Who would (pause) you have to make them for you? (pause) Who do you (pause) consider your…” (Extract 4)31; similar features can be seen here: Counsellor: “If you, if you- supposing- I mean this is just supposing, supposing you (pause) had got infected or were to get infected…” (Extract 8).31 The evidence indicates that patients usually respond to hypothetical questions in ways that engage with the difficult future topics. Hypothetical questions are thus a powerful and effective tool for encouraging about difficult future topics and related concerns and plans.28 This seems to relate to the way that hypothetical questions downplay the connection between this particular patient and becoming unwell, dying, etc. By emphasising the imaginary nature of the circumstances depicted, a distance is created between this particular patient and the facts of illness progression, end of life, etc—and this distancing makes matters easier to talk about (a danger may be that any associated decision-making may also not seem ‘real’—this has not been examined). Further insights into how hypothetical questions work are provided by evidence that in face-to-face communication, questions strongly oblige responses, and specifically they strongly oblige responses fitted to matters that the questioner has assumed and implied in the phrasing of their question.46 In contrast to fishing questions, hypothetical questions push hard for very specific and on-topic responses. Finally, it has been noted that hypothetical questions allow practitioners and patients to engage in talk about the future without either having to commit to accurate prognostication31 this is particularly useful in the light of the uncertain trajectories that frailty and life-limiting illnesses often take.

responses. Finally, it has been noted that hypothetical questions allow practitioners and patients to engage in talk about the future without either having to commit to accurate prognostication31 this is particularly useful in the light of the uncertain trajectories that frailty and life-limiting illnesses often take. Framing the issue as universal or general Two studies from counselling31 37 and two from oncology33 34 settings document practitioners referring to the difficult future topics as matters that anyone could face, framing them in terms of abstract rules or principles that concern or apply to everyone, for example, Counsellor: “Say, say you did die, what would be the hardest thing for your partner? Any of us can die in crashes or anything but just let’s take it out, what would be the hardest thing for your partner?” (Extract 10, adapted).31 In another example, a counsellor prefaces a question about a sensitive and potentially distressing matter as follows: “And one of the things that we know we’re sort of faced with sometimes when people get very ill and even die, is of course views about post-mortems in general and I just would like to know what Barry feels about that just in general?” (Extract 11).31 Generalising or universalising works, like hypothetical talk, to downplay the relevance of illness progression or dying to this particular patient.37 Extrapolating from this, it may be that framing issues as general or universal is, like hypothetical talk, effective in encouraging talk on the difficult future topics. However, this has not been empirically examined.

thetical talk, to downplay the relevance of illness progression or dying to this particular patient.37 Extrapolating from this, it may be that framing issues as general or universal is, like hypothetical talk, effective in encouraging talk on the difficult future topics. However, this has not been empirically examined. Communications other than words that display sensitivity Three studies in counselling and one in an oncology setting26 30 31 37 document that when people raise and pursue talk on difficult future topics, they generally do so with delays, hesitations and ‘turbulence’—that is, stuttering, cut-off words and repeats. One of the studies examined closely how patients respond to this kind of talk and found that patients sustain close attention over its course.30 The practices can be seen in the counsellor's talk within an example we have examined above; with regard to hypothetical questions, it includes slow, stretched out words and lots of pauses: “Saaaaaay, we can’t say we can’t say and you can’t say… but say you did begin to get ill (pause) or say you got so ill that you couldn’t kind of (pause) make decisions for yourself. Who would (pause) you have to make them for you? (pause) Who do you (pause) consider your…” (Extract 4).30 By talking in this way, people convey the sensitivity and also the seriousness of the matters they are talking about. This study also showed that these features have a special function when used within hypothetical questions: while talking about ‘imaginary’ matters downplays the seriousness and sensitive character of what is talked about, delivering hypothetical questions via hesitant and turbulent talk underlines the seriousness and sensitivity of the matters that are raised. Another practice within this category is touch. This has been subject to very little examination in this context; one study26 describes an episode in which a doctor touches a patient’s hand at the same time as talking indirectly about difficult future issues, and suggests that this touch works to convey support or comfort while also conveying seriousness and sensitivity. However, we found no systematic analysis of touch and its effects on patients in this context.

n which a doctor touches a patient’s hand at the same time as talking indirectly about difficult future issues, and suggests that this touch works to convey support or comfort while also conveying seriousness and sensitivity. However, we found no systematic analysis of touch and its effects on patients in this context. Communication practices other than words that encourage the patient to talk once the topic has been broached There is a small amount of evidence on this set of practices.37 41 In one practice, after a professional has raised a difficult future topic by asking a patient a question, if the patient does not respond, the professional remains silent. There is evidence that this can result in the patient eventually engaging with the topic. In their everyday conversations, people rarely leave gaps of longer than 0.3 s after questions,47 and gaps of longer than a second are rare unless some other activity (eg, writing in notes) is occurring.48 That is, people have a low tolerance for silences in conversations, and thus silences can work to encourage talk. In another documented practice, once a patient has begun to respond to a question, the clinician speaks only briefly and in ways that encourage the patient to continue talking, for instance, only saying ‘Mm’. A small amount of evidence suggests that these practices are effective in encouraging patients to talk on difficult future topics.

ed practice, once a patient has begun to respond to a question, the clinician speaks only briefly and in ways that encourage the patient to continue talking, for instance, only saying ‘Mm’. A small amount of evidence suggests that these practices are effective in encouraging patients to talk on difficult future topics. Shifting to the positive Studies of family telephone calls,24 oncology25 and primary care27 show that once a difficult future topic has been broached, it is very common for patients, family members and doctors to move rapidly towards upbeat talk of various forms. These include referring to fighting, perseverance and hope, facing illness together, optimistic rather than pessimistic versions and treatment successes. One publication25 analyses in detail the interactional consequences of doing so. It documents a pattern in which the doctors studied immediately followed bad or uncertain news with relatively hopeful or positive news.25 The study shows that patients then respond to the last thing the doctor has said—that is, the patient stays on the optimistic topic and does not engage with what was mentioned first. In doing so, patients are following a conversational convention in which second speakers generally respond to the last thing the previous speaker has said. Upbeat talk and shifting to the positive build and convey optimism and hope,24 and play an important role in building and maintaining doctor–patient relationships.27 34 However, shifting from negative topics to positive ones suppresses further talk on the difficult topics.30–44

last thing the previous speaker has said. Upbeat talk and shifting to the positive build and convey optimism and hope,24 and play an important role in building and maintaining doctor–patient relationships.27 34 However, shifting from negative topics to positive ones suppresses further talk on the difficult topics.30–44 Discussion We have examined a body of research that inductively analyses conversations and consultations via audio and audiovisual recordings. This kind of research generates explicit descriptions of communication practices and their outcomes. Such descriptions are important for evidence-based and reflective practice because they make explicit the communication practices that many clinicians use regularly, but about which their understandings reside largely in the tacit domain.

ch generates explicit descriptions of communication practices and their outcomes. Such descriptions are important for evidence-based and reflective practice because they make explicit the communication practices that many clinicians use regularly, but about which their understandings reside largely in the tacit domain. We identified and reviewed 19 publications, mostly analyses of conversations in clinical settings. A minority (7/19, see online supplementary table S1) had been written for clinical audiences. After excluding practices documented in only one publication, we collated findings into eight categories: ‘fishing questions’—open questions seeking patients’ perspectives (5/19); indirect, allusive or vague talk (6/19); linking questions and proposals to what the patient has already said or noticeably not said (7/19); hypothetical talk (12/19); framing difficult matters as universal or general (4/19); conveying sensitivity and seriousness via means other than words, for example, hesitant talk, touch (4/19); using means other than words to encourage further talk, for example, leaving silences after questions (2/19); and steering talk from negative or difficult to more optimistic aspects (3/19). Importantly, these practices differ considerably in how strongly they encourage talk on sensitive topics. Although most analyses primarily examine practitioners’ behaviours, there is some evidence that patients and their significant others use similar practices.

negative or difficult to more optimistic aspects (3/19). Importantly, these practices differ considerably in how strongly they encourage talk on sensitive topics. Although most analyses primarily examine practitioners’ behaviours, there is some evidence that patients and their significant others use similar practices. Our review was based on a carefully developed methodology16 which applied established conventions for systematic reviewing while also applying techniques specifically fitted to conversation and discourse analytic evidence. A key strength of the review is that it makes available to clinicians, policymakers and clinical educators evidence about communication that has been relatively inaccessible because of its publication in non-clinical fields, and its framing in terms of sociological and linguistic arguments and concepts. Weaknesses include the fact that we did not incorporate studies of conversations about difficult future issues in languages other than English. We did identify some publications in this category which would have been rated high in their quality and potential contribution to the review,49 but excluded them because of the possibility that different languages might entail subtly different practices for talking about the future and different consequences of these practices. Another weakness, evident in some of our illustrative examples, is the fact that some of the studies report on data that are now more than 20 years old. There is, however, good reason to assume that while contemporary clinical circumstances are different (particularly in relation to HIV/AIDS), language structures and their functioning are rather more stable matters12—this is evident in the way that the same communication practices are indeed reported in earlier and later studies. Finally, there are substantial gaps in evidence. These gaps mean that the review cannot provide a comprehensive overview of all the practices actually used in talking about difficult future issues. In particular, there are large gaps in knowledge about: practices used by patients, their companions and by healthcare professionals other than doctors and counsellors; and about the role of body movements such as touch.

comprehensive overview of all the practices actually used in talking about difficult future issues. In particular, there are large gaps in knowledge about: practices used by patients, their companions and by healthcare professionals other than doctors and counsellors; and about the role of body movements such as touch. Also, only one study to date has examined practices used by specialists in palliative care29–while this study makes a helpful contribution to understanding little-examined practices in a setting where talk about difficult future issues is a key part of the work, as we noted, its contribution to the review findings is limited because of the analytic approach applied. We strongly advocate the use of a conversation analytic approach12 in future studies, because of its capacity to yield fine-grained, systematically generated findings, and because its findings and claims are tightly tied to the details of the empirical data (what is said and how it is said), and to a cumulative body of established findings about the structure and functioning of communication practices.14 15 Most of the currently available evidence concerns those stages in consultations and conversations wherein people are initially broaching difficult future topics and those where they are closing these conversations. There is very little on how people progress conversations from raising the difficult future topic to actually engaging in associated decision-making and planning.

tages in consultations and conversations wherein people are initially broaching difficult future topics and those where they are closing these conversations. There is very little on how people progress conversations from raising the difficult future topic to actually engaging in associated decision-making and planning. The outcome of particular communication practices in terms of patients’ experience of care, and their conduct subsequent to the consultation, is another area about which little is known as yet. Recent research by Robinson and colleagues has opened up new avenues.50 Using conversation analysis-derived coding, they were able to show that levels of hopelessness of patients significantly decreased after consultations, and that this change was associated with the occurrence of two particular communication behaviours: patients asserting their treatment preferences, and doctors providing good or hopeful news. This form of research, showing links between specific and carefully operationalised communication behaviours and patient outcomes, holds great promise for advancing healthcare communication practice and training. Overall, further research is needed, in particular research that provides insights not just into healthcare professionals’ communication, but also into the contributions of patients and their significant others, and into the outcomes of specific practices. Such analyses need to attend not just to what is said and done, but how it is said and done.

needed, in particular research that provides insights not just into healthcare professionals’ communication, but also into the contributions of patients and their significant others, and into the outcomes of specific practices. Such analyses need to attend not just to what is said and done, but how it is said and done. Implications At the broadest level, the implications of our findings are that conversations about illness progression and end of life can be initiated and pursued in a variety of ways, and that these have different consequences. More specifically, fishing questions and indirect talk show sensitivity and can not only open up the topics, but also make it relatively easy for patients and others to avoid engaging with them. Thus, where a clinician judges that it is in the patient’s and their significant others’ best interests to raise the issue gently, to knock on the door rather than push at it, these practices seem optimal. Indirect talk may be useful when a clinician wishes to test whether this is ‘the right time’ to pursue these issues, although fishing questions less so because they so rarely elicit on-topic answers. Hypothetical questions more strongly oblige on-topic talk, and there is good reason to suggest that framing concerns as general rather than specific to the individual will also do so. These practices are suited to circumstances where a clinician judges it is in a patient's best interests to more strongly encourage talk on the difficult future issues, for example, when a decision about whether or not to instigate an invasive intervention is imminent. Linking questions and proposals to patients’ cues—matters they have already raised or hinted at (or even to matters a patient has noticeably not raised)—are effective in encouraging engagement with difficult future topics. Finally, shifting conversations towards more ‘optimistic’ aspects plays a role in maintaining hope, but because it also closes off further talk about difficulties, clinicians may want to delay doing so.

to matters a patient has noticeably not raised)—are effective in encouraging engagement with difficult future topics. Finally, shifting conversations towards more ‘optimistic’ aspects plays a role in maintaining hope, but because it also closes off further talk about difficulties, clinicians may want to delay doing so. Although we report findings in terms of three further categories—framing issues as general, communication other than words that conveys sensitivity and communications that encourage patients to talk once a topic has been raised—we have not proposed specific implications in relation to these because there was very little evidence about these practices’ structures and functioning. Our review makes an important contribution to the evidence base for clinical decisions about ‘what works and how’ in terms of providing patients and their significant others with opportunities to communicate about difficult issues. This information should help raise clinicians’ confidence in broaching these issues, and applying it to actual practice may increase the effectiveness of advanced care planning and end of life decision-making.

s of providing patients and their significant others with opportunities to communicate about difficult issues. This information should help raise clinicians’ confidence in broaching these issues, and applying it to actual practice may increase the effectiveness of advanced care planning and end of life decision-making. Conclusions Much research, policy and commentary advocates providing opportunities for talking about difficult uncertain future issues including illness progression and end of life with patients who have long-term and serious illnesses, and with people who are frail and elderly. There are different ways to initiate talk about feelings and plans in relation to difficult topics like end of life—in particular, posing hypothetical questions strongly encourages on-topic talk, while indirect references to the difficult issues form a more gentle way of ‘pushing at the door’ of the topics—which some patients take up and some deflect. There is, however, much we still do not know, including about communication practices used in reaching decisions about future care, and about the role of bodily communication within these conversations.

lt issues form a more gentle way of ‘pushing at the door’ of the topics—which some patients take up and some deflect. There is, however, much we still do not know, including about communication practices used in reaching decisions about future care, and about the role of bodily communication within these conversations. Supplementary Material Web table 1 Web table 2 The review advisors were Merran Toerien and Paul Drew; they discussed the review purpose and design, and suggested publications for inclusion. A number of audiences discussed our findings with us and helped shape the implications drawn; these include: the staff at LOROS Hospice Care in Leicestershire, members of the East Midlands Cancer Network, and those attending the Helsinki Collegium for Advanced Studies 2011 Symposium on Mortality and Emotions. Funding: The study was funded by the Sue Ryder Care Centre for the Study of Supportive, Palliative and End of Life Care at the School of Health Sciences, University of Nottingham (for RP's and JS's time) and by a small grant from the School of Health Sciences, University of Nottingham (for VL's time). The study sponsor was the University of Nottingham.

he Sue Ryder Care Centre for the Study of Supportive, Palliative and End of Life Care at the School of Health Sciences, University of Nottingham (for RP's and JS's time) and by a small grant from the School of Health Sciences, University of Nottingham (for VL's time). The study sponsor was the University of Nottingham. Contributors: RP conceived and instigated the review and wrote the protocol, participated in the search and selection of publications and in data extraction, and led the synthesis and drafting of this manuscript. VL collaborated on the design of the review, conducted the majority of searches for publications and the majority of data extraction, supported the synthesis, commented on drafts of this manuscript and approved this submitted version. JS acted as consultant on the design and substantive topics of this review, commented on drafts of this manuscript and approved this submitted version. Competing interests: None. Provenance and peer review: Not commissioned; externally peer reviewed.

Introduction The application of appropriate outcome measures is dependent on understanding the needs of service users.1 2 The essential contribution of patient and carer perspectives to establishing needs and measuring service provision is recognised3 as is the need for clear definitions of service goals, better measures of carer assessment and understanding of carers’ unmet needs.4

ependent on understanding the needs of service users.1 2 The essential contribution of patient and carer perspectives to establishing needs and measuring service provision is recognised3 as is the need for clear definitions of service goals, better measures of carer assessment and understanding of carers’ unmet needs.4 However, natural tensions arise as service providers and commissioners demonstrate awareness of the need to audit the service, but focus on physical or transactional aspects rather than incorporating relational care meeting psychosocial need.5 6 This underestimates the influence that contextual care has on medical outcomes, an area which remains under-researched in terms of understanding and accurate measurement.7 Contextual care is comparable with the original philosophy of holistic care in the focus on the patient as part of a wider social network and with psychosocial as well as physical needs.8 A significant factor in contextual care remains the interaction of the patient–professional relationship.9 The performance of emotional labour, whereby healthcare professionals actively manage relationships with patients, is informally recognised as a key skill facilitating the patient journey.10 Through emotional labour, the healthcare professional manages emotion in order to enable the patients to feel reassured and able to express their own feelings and concerns. This professional work contributes fundamentally to the patient experience of care but is not necessarily recognised within the formal assessment of palliative care outcomes.

he healthcare professional manages emotion in order to enable the patients to feel reassured and able to express their own feelings and concerns. This professional work contributes fundamentally to the patient experience of care but is not necessarily recognised within the formal assessment of palliative care outcomes. Although models of practice, instruments and interventions based on concepts such as dignity preservation have been implemented,11 the challenge of incorporating the perspectives of health professionals, health providers and service users into an evolving definition of palliative care is ongoing.12 A regional palliative care planning group, encompassing all aspects and settings of palliative care, commissioned a service evaluation questionnaire using a combination of closed questions and a free text option in order to base provision of excellent palliative and end-of-life care on understanding the perspectives of patients and their families.13 Methods The data set consisted of 594 completed free text responses from the data corpus of 696 questionnaires comprising nine set questions on a Likert scale and a free text box inviting comments on any aspect of palliative care received (figure 1). Figure 1 Free text box.

A regional palliative care planning group, encompassing all aspects and settings of palliative care, commissioned a service evaluation questionnaire using a combination of closed questions and a free text option in order to base provision of excellent palliative and end-of-life care on understanding the perspectives of patients and their families.13 Methods The data set consisted of 594 completed free text responses from the data corpus of 696 questionnaires comprising nine set questions on a Likert scale and a free text box inviting comments on any aspect of palliative care received (figure 1). Figure 1 Free text box. In all, 594 completed free text responses from 696 questionnaires were analysed thematically by a qualitative researcher (CS) who was not involved in the palliative care service, in commissioning or implementing the service evaluation. Cross-coding was carried out to ensure methodological rigour (AN) and COREQ guidelines for qualitative research were followed. Two-thirds of responses were from patients and a third from family members who described caring roles and are referred to as carers. Thematic analysis was interpretive and inductive, reflecting the open ended structure of the free text box. Data were managed using NVivo computer software.14 Each data item was anonymised and coded using data extracts as tree nodes and following six key stages: becoming familiar with the data, generating initial codes, searching for themes, reviewing themes, defining and naming themes and reporting the themes (figures 2 and 3).15 Figure 2 Stage 2 generating initial codes.

Thematic analysis was interpretive and inductive, reflecting the open ended structure of the free text box. Data were managed using NVivo computer software.14 Each data item was anonymised and coded using data extracts as tree nodes and following six key stages: becoming familiar with the data, generating initial codes, searching for themes, reviewing themes, defining and naming themes and reporting the themes (figures 2 and 3).15 Figure 2 Stage 2 generating initial codes. Figure 3 Reviewing themes. Interpretive analysis examined the content and the structure of the responses, as the majority of participants had structured the free text response into an outline of change that had occurred as a consequence of palliative care and how this change was accomplished. This resulted in one higher level theme relating to the overall process of care delivery and four subthemes outlining key areas in which patients and carers prioritised palliative care intervention. The local R&D office was consulted to confirm the validity and ethics of the approach taken.16

s change was accomplished. This resulted in one higher level theme relating to the overall process of care delivery and four subthemes outlining key areas in which patients and carers prioritised palliative care intervention. The local R&D office was consulted to confirm the validity and ethics of the approach taken.16 Results The emotional experience of care Thematic analysis focused on the meaning participants associated with their palliative care experience. Patients and carers constructed free text responses documenting the consequences of palliative care. Coding focused on a key emergent theme occurring across all aspects and settings of care and characterising the palliative care experience. This emotional experience of care was coded to include any reference to a change in attitude, affect or ability to cope. These ranged from major shifts in attitude from feelings of despair to feelings of renewal, and more subtle shifts where feelings of security, acceptance or reassurance were expressed. There were four subthemes associated with the emotional experience of care: Respect, Renewal, Refuge and Restorative care (figure 4). Figure 4 Stage 5 defining and naming themes.

Results The emotional experience of care Thematic analysis focused on the meaning participants associated with their palliative care experience. Patients and carers constructed free text responses documenting the consequences of palliative care. Coding focused on a key emergent theme occurring across all aspects and settings of care and characterising the palliative care experience. This emotional experience of care was coded to include any reference to a change in attitude, affect or ability to cope. These ranged from major shifts in attitude from feelings of despair to feelings of renewal, and more subtle shifts where feelings of security, acceptance or reassurance were expressed. There were four subthemes associated with the emotional experience of care: Respect, Renewal, Refuge and Restorative care (figure 4). Figure 4 Stage 5 defining and naming themes. Respect Emotional labour in healthcare is distinguished by the therapeutic relationship, based on interpersonal skills, professional expertise and multidisciplinary cooperation.17 18 References to the atmosphere of the hospice or the tenor of care within the home made the underlying palliative care philosophy visible.I have been an in-patient on three separate occasions and as someone with terminal cancer I often feel vulnerable as my health slowly deteriorates. [hospice] is the only place that I feel truly safe, and supported by the dedicated doctors and staff. I know that I can trust them to help me feel better. Hospice inpatient

Respect Emotional labour in healthcare is distinguished by the therapeutic relationship, based on interpersonal skills, professional expertise and multidisciplinary cooperation.17 18 References to the atmosphere of the hospice or the tenor of care within the home made the underlying palliative care philosophy visible.I have been an in-patient on three separate occasions and as someone with terminal cancer I often feel vulnerable as my health slowly deteriorates. [hospice] is the only place that I feel truly safe, and supported by the dedicated doctors and staff. I know that I can trust them to help me feel better. Hospice inpatient Patient and carer responses deconstructed emotional labour into situations and practices illustrating how it is accomplished and the skills it draws upon. Specific skills relating to emotional support attested to the importance of a responsive, empathic presence, where the changing needs of patients, carers and family members, whether physical or psychological, were accepted by staff.[palliative care nurse] has been the strength and support that we have needed since diagnosis and living each day with cancer. She is always on the end of the phone and no question seems silly to her and she helps us to face the future and realise that all the feelings we have are normal. Patient: Hospital palliative care outreach service

Patient and carer responses deconstructed emotional labour into situations and practices illustrating how it is accomplished and the skills it draws upon. Specific skills relating to emotional support attested to the importance of a responsive, empathic presence, where the changing needs of patients, carers and family members, whether physical or psychological, were accepted by staff.[palliative care nurse] has been the strength and support that we have needed since diagnosis and living each day with cancer. She is always on the end of the phone and no question seems silly to her and she helps us to face the future and realise that all the feelings we have are normal. Patient: Hospital palliative care outreach service Patients documented the importance of being allowed to express their feelings or the presence of a palliative care professional as they accomplished the active work of coming to terms with diagnosis and prognosis. Carers valued recognition of their role and being listened to, particularly in relation to fulfilling the wishes of their loved ones. Consequences of respectful care were often practical, such as providing respite care, but rooted in a cooperative, dynamic process with palliative care professionals.

osis and prognosis. Carers valued recognition of their role and being listened to, particularly in relation to fulfilling the wishes of their loved ones. Consequences of respectful care were often practical, such as providing respite care, but rooted in a cooperative, dynamic process with palliative care professionals. Patients prioritised being treated as a person rather than seen solely in terms of their illness. The complexity of being accepted as an individual and also incorporated into the companionship of others draws upon professional and interpersonal skills.The staff create a wonderful atmosphere. They are caring and understanding without being too overbearing. Everyone is treated and respected as an individual and yet you are made to feel like one of a large family. Hospice inpatient The mediation of these skills resulted in palliative care that is patient focused and responsive rather than prescriptive and generalised. References to the importance of a good death were coded solely from the responses of carers and family members. Palliative care provided practical but also emotional help in explaining to the patient and their family what the last stages of an illness involved. A significant aspect was the importance to carers of fulfilling the patient's wishes. There was explicit reference of the patient's wish to die at home.The care that my father received during the last few weeks from the hospice at home team, district nurses and Macmillan nurse was first class and enabled him to die peacefully at home which is what he wanted. Carer: Community palliative care service

The mediation of these skills resulted in palliative care that is patient focused and responsive rather than prescriptive and generalised. References to the importance of a good death were coded solely from the responses of carers and family members. Palliative care provided practical but also emotional help in explaining to the patient and their family what the last stages of an illness involved. A significant aspect was the importance to carers of fulfilling the patient's wishes. There was explicit reference of the patient's wish to die at home.The care that my father received during the last few weeks from the hospice at home team, district nurses and Macmillan nurse was first class and enabled him to die peacefully at home which is what he wanted. Carer: Community palliative care service However, the priorities around a good death were communication and the patient's comfort, regardless of where death occurred.My husband was in the hospice for two weeks before he passed away. The care he received was wonderful. The staff were so caring and compassionate. The aftercare I received was so helpful. It helped me through a very painful time. Carer Renewal Renewal incorporated changes of attitude, increased confidence or renewed agency. Patients specifically described coming to terms with diagnosis and prognosis.This [palliative care] has helped me to be more positive and has helped me to accept things and my problems. Patient: Day hospice

However, the priorities around a good death were communication and the patient's comfort, regardless of where death occurred.My husband was in the hospice for two weeks before he passed away. The care he received was wonderful. The staff were so caring and compassionate. The aftercare I received was so helpful. It helped me through a very painful time. Carer Renewal Renewal incorporated changes of attitude, increased confidence or renewed agency. Patients specifically described coming to terms with diagnosis and prognosis.This [palliative care] has helped me to be more positive and has helped me to accept things and my problems. Patient: Day hospice Attitude change in some cases related to the perception of palliative care itself, either by patients or their families.I would like to say that a lot of people diagnosed with cancer think that ‘Hospice’ means death. My family thought it, but when I went to my drop in clinic I was proved wrong. They were my life-line with emotional support that my family couldn't cope with. Patient: Hospice out-patient Understanding was associated with peer support and with the specialist knowledge of the palliative care service.Your response to a time of crisis in my life meant so much. Until then I despaired for the future but your help made me see that I was unable to care for my husband any longer. He is now settled into a care home—not what I wanted for us, but I have to accept my limitations. Carer: Community palliative care service

Understanding was associated with peer support and with the specialist knowledge of the palliative care service.Your response to a time of crisis in my life meant so much. Until then I despaired for the future but your help made me see that I was unable to care for my husband any longer. He is now settled into a care home—not what I wanted for us, but I have to accept my limitations. Carer: Community palliative care service Patients and carers documented having confidence in the palliative care service, but more significantly described changes to confidence brought about by the intervention of palliative care.I was glad of the support at [hospice] and they worked as a team to improve my confidence and lighten my anxiety. With their help, I recovered my mental, physical and emotional health. Hospice inpatient

Patients and carers documented having confidence in the palliative care service, but more significantly described changes to confidence brought about by the intervention of palliative care.I was glad of the support at [hospice] and they worked as a team to improve my confidence and lighten my anxiety. With their help, I recovered my mental, physical and emotional health. Hospice inpatient Aspects of care contributing to the ability to cope varied from practical help and specialist knowledge to being listened to. Multidisciplinary communication and cooperation across agencies resulted in patients and family members being enabled to approach their situation with renewed confidence.[palliative care team] efficiently dealt with hospital delays in discharging [the patient] from hospital so we could care for him at home in the last days of his life. They then dealt with all aspects of monitoring his care and co-ordinating the necessary agencies as well as supplying some of the light care and respite care for me. All the people who came to help were efficient, friendly and supportive. It was to [palliative care team] I turned in moments of panic and to understand the changes before death. Carer: Community palliative care

Aspects of care contributing to the ability to cope varied from practical help and specialist knowledge to being listened to. Multidisciplinary communication and cooperation across agencies resulted in patients and family members being enabled to approach their situation with renewed confidence.[palliative care team] efficiently dealt with hospital delays in discharging [the patient] from hospital so we could care for him at home in the last days of his life. They then dealt with all aspects of monitoring his care and co-ordinating the necessary agencies as well as supplying some of the light care and respite care for me. All the people who came to help were efficient, friendly and supportive. It was to [palliative care team] I turned in moments of panic and to understand the changes before death. Carer: Community palliative care An important aspect was the recognition by palliative care staff that terminal illness presents a crisis for the whole family.19 Being able to cope was constructed as a contrast between normality and exceptional circumstances, where the contribution of palliative care made it possible to negotiate everyday life.I found them [palliative care staff] very caring and understanding, easy to talk to and very helpful. They have helped me to cope with a long term illness and kept me sane enabling me to go out and about and relax for a few hours. Patient: Community based palliative care service

An important aspect was the recognition by palliative care staff that terminal illness presents a crisis for the whole family.19 Being able to cope was constructed as a contrast between normality and exceptional circumstances, where the contribution of palliative care made it possible to negotiate everyday life.I found them [palliative care staff] very caring and understanding, easy to talk to and very helpful. They have helped me to cope with a long term illness and kept me sane enabling me to go out and about and relax for a few hours. Patient: Community based palliative care service The implication of recognising the enabling aspect of palliative care is that good care may facilitate change in areas of life not necessarily visible or directly apparent to staff. Refuge Palliative and hospice care was repeatedly described as a place of safety, alleviating loneliness and providing a space to express fears that could not be expressed elsewhere.I was admitted as an in-patient with breathing problems and an inclination towards panic attacks. Immediately upon entering the hospice I calmed down, was able to lay my baggage down. Hospice inpatient Refuge was also constructed as a psychological aspect of care where it was safe to express both positive and negative emotions.My problems were addressed and I felt like an individual again. My problems were difficult to treat and my inability to express myself must have been frustrating but I was shown patience and respect and was made to feel safe and cared for until I was ready to move forward again. Patient: Specialist cancer hospital inpatient

Refuge was also constructed as a psychological aspect of care where it was safe to express both positive and negative emotions.My problems were addressed and I felt like an individual again. My problems were difficult to treat and my inability to express myself must have been frustrating but I was shown patience and respect and was made to feel safe and cared for until I was ready to move forward again. Patient: Specialist cancer hospital inpatient Patients and carers valued flexible palliative care that impacted on the whole family.My husband is being visited regularly by [palliative care nurse] and he has been a very great comfort and has also provided positive help to us...We know that we can contact him at any time and he will be there for us. Carer: Community palliative care Respondents distinguished the palliative care service through consistent care delivery that included volunteers as well as professionals.I have received the most amazing care from both nurses and volunteers. The unit is run smoothly and I look forward to my [weekly visit] as I know that any problems I have, medically or emotionally, will be addressed and I will receive all the care I need promptly and efficiently. They make a difficult prognosis so much easier to bear. Patient: Day hospice Good care was hallmarked by expertise and responsiveness in addition to characteristics such as warmth and sensitivity.This is a wonderful service, so caring, so efficient, from the top nurse to the volunteers. Patient: Hospice out-patient

Respondents distinguished the palliative care service through consistent care delivery that included volunteers as well as professionals.I have received the most amazing care from both nurses and volunteers. The unit is run smoothly and I look forward to my [weekly visit] as I know that any problems I have, medically or emotionally, will be addressed and I will receive all the care I need promptly and efficiently. They make a difficult prognosis so much easier to bear. Patient: Day hospice Good care was hallmarked by expertise and responsiveness in addition to characteristics such as warmth and sensitivity.This is a wonderful service, so caring, so efficient, from the top nurse to the volunteers. Patient: Hospice out-patient Refuge, whether located in the hospice or embodied in the palliative care skills of the professional in the home, emphasised the importance of managing transitions between illness stages and helping patients and carers to negotiate uncertainty.

Good care was hallmarked by expertise and responsiveness in addition to characteristics such as warmth and sensitivity.This is a wonderful service, so caring, so efficient, from the top nurse to the volunteers. Patient: Hospice out-patient Refuge, whether located in the hospice or embodied in the palliative care skills of the professional in the home, emphasised the importance of managing transitions between illness stages and helping patients and carers to negotiate uncertainty. Restorative Restorative care relates to therapeutic aspects of care leading to enjoyment or relaxation and directs attention to care of the body. Enjoyment of alternative therapies shifted focus to the context of the body as normal rather than diseased. A fundamental aspect of illness is intimate care of the body. This has been the focus of debate about the status of bodily tasks within healthcare, and the relationship for patients between social well-being, emotional health and limited agency due to illness.17 20 Patients described care and respect for the body through use of alternative therapies.The kindness and professionalism shown to me by staff is exemplary. The complimentary therapy I receive is wonderful, a peaceful haven of normality and I'm sure that without it I would find it far more difficult to cope. Patient: Day hospice

Restorative Restorative care relates to therapeutic aspects of care leading to enjoyment or relaxation and directs attention to care of the body. Enjoyment of alternative therapies shifted focus to the context of the body as normal rather than diseased. A fundamental aspect of illness is intimate care of the body. This has been the focus of debate about the status of bodily tasks within healthcare, and the relationship for patients between social well-being, emotional health and limited agency due to illness.17 20 Patients described care and respect for the body through use of alternative therapies.The kindness and professionalism shown to me by staff is exemplary. The complimentary therapy I receive is wonderful, a peaceful haven of normality and I'm sure that without it I would find it far more difficult to cope. Patient: Day hospice In addition to bodily deterioration and a terminal diagnosis, patients may experience a social death.21 The contribution of good palliative care to being acknowledged as an individual was highlighted by respondents, many of whom expressed enjoyment and satisfaction with the provision of alternative therapies.The staff and volunteers at [hospice] have without exception been wonderful. So much help and comfort—hands on therapies, explanations of treatments and drugs I'm having. Fellowship, laughter, relaxation, problem solving. You name it, they have done it... Patient: Hospice out-patient

In addition to bodily deterioration and a terminal diagnosis, patients may experience a social death.21 The contribution of good palliative care to being acknowledged as an individual was highlighted by respondents, many of whom expressed enjoyment and satisfaction with the provision of alternative therapies.The staff and volunteers at [hospice] have without exception been wonderful. So much help and comfort—hands on therapies, explanations of treatments and drugs I'm having. Fellowship, laughter, relaxation, problem solving. You name it, they have done it... Patient: Hospice out-patient Acceptance of both the ill body and the associated loss of personhood were present in alternative therapy, where the body was the site of respect rather than of pathology, and where shared experiences validated rather than negated personhood. Concerns with palliative care Concerns constituted a minor aspect of the free text coding but analysis enabled a deeper understanding of how patients and carers communicate their concerns. Questionnaire scores alone did not provide insight into the nature of concerns but qualitative analysis highlighted that participants structured their responses in two ways, each of which impacted significantly on questionnaire scoring.

abled a deeper understanding of how patients and carers communicate their concerns. Questionnaire scores alone did not provide insight into the nature of concerns but qualitative analysis highlighted that participants structured their responses in two ways, each of which impacted significantly on questionnaire scoring. Where individual care was good but the context of care was poor, with participants expressing concern about funding, facilities or provision of resources, questionnaire scores reflected satisfaction. This acknowledged good palliative care but highlighted resource concerns, thereby providing support of the service and indicating that provision of better facilities, more staff and extension of service hours were priorities.Excellent team done by a dedicated team. Treatment and counselling second to none. Could do with their own clinic, more money and resources. Patient: Hospice community service Where there was a concern with specific care this was reflected in lower questionnaire scores and related to issues around communication and consideration of individual need.The care was very good in the beginning but I did feel as though as soon as my treatment had finished I had been dropped. Although my palliative care nurse said I was able to telephone if there were any problems I felt as though I was bumbling along without any help or visits—there were still many issues I wanted help with. Patient: Community palliative care service Patients and carers clearly distinguished between their individual care and the wider context within which palliative care functions.

Where there was a concern with specific care this was reflected in lower questionnaire scores and related to issues around communication and consideration of individual need.The care was very good in the beginning but I did feel as though as soon as my treatment had finished I had been dropped. Although my palliative care nurse said I was able to telephone if there were any problems I felt as though I was bumbling along without any help or visits—there were still many issues I wanted help with. Patient: Community palliative care service Patients and carers clearly distinguished between their individual care and the wider context within which palliative care functions. Discussion Psychosocial dimensions of palliative care coexist alongside provision of pain and symptom control and are implicitly recognised by staff and patients.22 However, the challenges of collecting routine data from a population with a terminal prognosis, the difficulties of adequately defining aspects of psychosocial care and the necessity of concentrating on measurable aspects of care to justify economic considerations have led to tension between meeting more easily defined practical outcomes and evaluating contextual outcomes such as tenor of care.23 24

al prognosis, the difficulties of adequately defining aspects of psychosocial care and the necessity of concentrating on measurable aspects of care to justify economic considerations have led to tension between meeting more easily defined practical outcomes and evaluating contextual outcomes such as tenor of care.23 24 Palliative and end-of-life care requires detailed recording of communication processes and service users’ feedback in addition to expert consensus on demonstrable indicators that improve outcomes.2 25 The limitations of the scientific method in measuring humane aspects of care such as dignity and attentiveness are recognised, as is the need for these factors to be fully recognised as impacting directly on the quality of care.26 Free text analysis highlights the fundamental importance of the emotional experience of care to patients and carers. It describes the specific skills and practices that constitute good palliative care, implying a mobile set of skills united by a common philosophy and embodied in the palliative healthcare professional. Good care is distinguished by characteristics such as expertise, enablement and efficiency within the context of respectfulness. Understanding the significance of the emotional labour involved in caring is relevant to the practice of palliative care where it occurs in multiple settings across the community and not solely in institutions such as hospices or hospitals. The contribution of volunteers was significant, highlighting the importance of supporting their role.27

icance of the emotional labour involved in caring is relevant to the practice of palliative care where it occurs in multiple settings across the community and not solely in institutions such as hospices or hospitals. The contribution of volunteers was significant, highlighting the importance of supporting their role.27 The practice of emotional labour is influenced by changing expectations both organisationally and within wider society.28 29 The changing consumer orientated culture of healthcare has led to emotion work changing from an intrinsic contract between healthcare professional and patient to caring as a commodity.30 31 However, although structural changes within the caring professions may impact on the satisfaction and ability of professionals to carry out their duties, the core expectation of the therapeutic relationship remains a source of satisfaction.32 The challenge remains of acknowledging skills that have been devalued, yet which healthcare professionals see as central to their role.10 Free text analysis highlights that the emotional experience of care is highly valued by patients and carers and the attendant emotional labour is recognised and described revealing palliative care as enabling as well as provided. This is significant as lack of control and agency are key concerns for patients, in addition to the need for information, and practical and psychosocial support.33 Feelings of helplessness or lack of autonomy are also associated with greater difficulty for carers coping in stressful situations and following bereavement.34

s is significant as lack of control and agency are key concerns for patients, in addition to the need for information, and practical and psychosocial support.33 Feelings of helplessness or lack of autonomy are also associated with greater difficulty for carers coping in stressful situations and following bereavement.34 A distinction has been made between professional competencies, which can be learnt, and professional judgements, which are based in practice and experience35 and arise from the dynamic relationship between patients and professionals.36 Underevaluated aspects of palliative care include the professional and interpersonal skill of health professionals who have an intuitive understanding of the care they provide but may not recognise or define these aspects in a measurable way.37 Free text analysis contributes to an understanding of the skills that are valued by patients and carers and offers insight into the domains in which these aspects of palliative care make a significant impact. Conclusions Palliative care was founded on a holistic philosophy encompassing the physical, psychosocial and spiritual needs of patients with a terminal prognosis and their family.8 Contemporary palliative care acknowledges the contextual nature of care but there is tension between evaluating formal aspects of care delivery such as pain relief and service processes while giving equal recognition to aspects such as tenor of care.

ritual needs of patients with a terminal prognosis and their family.8 Contemporary palliative care acknowledges the contextual nature of care but there is tension between evaluating formal aspects of care delivery such as pain relief and service processes while giving equal recognition to aspects such as tenor of care. When service users were invited to complete the service evaluation free text survey box to comment on any aspect of their care, the qualitative analysis indicated overall satisfaction with the palliative care service. Service users structured their responses around their experiences of how care was delivered. This process of care delivery focused on the emotional experience of care and incorporated aspects associated with tenor of care such as agency, dignity and acceptance. The free text analysis highlights the importance of the emotional experience of care to patients and carers, providing insight into specific examples of where and how this care occurs. Specific expectations from the perspective of patients and carers are of professionalism, expertise and facilitation. The emotional experience of care encompassed practical and psychological outcomes but was characterised by the expression of renewed agency where a shift in attitude had occurred. Free text analysis indicates that the impact of palliative care may not be fully evident to health professionals or providers, and that aspects of palliative care practice that may be either taken for granted or not fully recognised are significant for patients and carers.

When service users were invited to complete the service evaluation free text survey box to comment on any aspect of their care, the qualitative analysis indicated overall satisfaction with the palliative care service. Service users structured their responses around their experiences of how care was delivered. This process of care delivery focused on the emotional experience of care and incorporated aspects associated with tenor of care such as agency, dignity and acceptance. The free text analysis highlights the importance of the emotional experience of care to patients and carers, providing insight into specific examples of where and how this care occurs. Specific expectations from the perspective of patients and carers are of professionalism, expertise and facilitation. The emotional experience of care encompassed practical and psychological outcomes but was characterised by the expression of renewed agency where a shift in attitude had occurred. Free text analysis indicates that the impact of palliative care may not be fully evident to health professionals or providers, and that aspects of palliative care practice that may be either taken for granted or not fully recognised are significant for patients and carers. The importance of contextual issues for patients and carers and the significance of the emotional experience of care indicate that they are essential components of good palliative care practice that should be included in future evaluation of the palliative care service.

Free text analysis indicates that the impact of palliative care may not be fully evident to health professionals or providers, and that aspects of palliative care practice that may be either taken for granted or not fully recognised are significant for patients and carers. The importance of contextual issues for patients and carers and the significance of the emotional experience of care indicate that they are essential components of good palliative care practice that should be included in future evaluation of the palliative care service. Funding: This work was supported by Marie Curie Cancer Care core grant funding to the Marie Curie Palliative Care Research Centre, Cardiff University. Dr. Annmarie Nelson and Dr. Anthiny Byrne's posts are supported by Marie Curie Cancer Care core grant funding. Contributors: IF conception and design of the service evaluation in Wales and acquisition of core data; CS and AN analytic input; CS article drafting; all authors article revision and approval of the final version. Competing interests: None. Ethics approval: Dr Corinne Scott consulted local ethics board and formal ethical approval not required. Provenance and peer review: Not commissioned; externally peer reviewed.

Improving end-of-life care is a key priority within current health and social care policy across European countries.1 2 A shared focus is quality of care and provision based on need rather than diagnosis. Service providers are required to ensure that, when death is inevitable, every patient receives appropriate and timely end-of-life care, focused on comfort and dignity while avoiding futile and invasive interventions that prevent a peaceful death. The importance of preparing and supporting the family and those close to the patient is also recognised as directly influencing how they cope with and adapt to bereavement, with implications for their health and social well-being.3–6 Measures towards improving the quality of care in the last days and hours of life include the introduction of multiprofessional integrated care pathways to benchmark standards and support the delivery and evaluation of clinical care. The Liverpool Care Pathway (LCP) for the Dying Patient is the most widely recognised.7 Originally developed within the UK in the mid-1990s as a vehicle to transfer the hospice model of end-of-life care for cancer patients into general care settings, the LCP came to be regarded as a model of best practice and is now used internationally.8–10

hway (LCP) for the Dying Patient is the most widely recognised.7 Originally developed within the UK in the mid-1990s as a vehicle to transfer the hospice model of end-of-life care for cancer patients into general care settings, the LCP came to be regarded as a model of best practice and is now used internationally.8–10 Identifying that the patient may be dying is, however, the crucial first step to planning and delivering effective end-of-life care, including the use of care pathways such as the LCP. Clinicians must accurately diagnose dying in order to ensure that a high standard of end-of-life care is provided for all those who need it and equally to identify when restorative treatment aims are appropriate.11 12 This process is currently viewed as part art, part science and is often regarded as a skill of individuals rather than an objective assessment of signs and symptoms.12 13 The complexity of diagnosing dying has been highlighted within recent reviews of the LCP, which report some difficulties and misinterpretations in relation to the purpose and use of this framework in practice. A focus of attention is clinical decision making surrounding end-of-life care, with key recommendations including the development of clear guidance for professionals and further research in diagnosing dying.14 15

some difficulties and misinterpretations in relation to the purpose and use of this framework in practice. A focus of attention is clinical decision making surrounding end-of-life care, with key recommendations including the development of clear guidance for professionals and further research in diagnosing dying.14 15 This integrative review was conducted to inform evidence-based practice in diagnosing dying, prompted by clinical members of the review team involved in supporting end-of-life care across hospital, hospice and community settings. The findings are also timely in relation to European policy requirements and relevant to informing the development and governance of end-of-life care internationally, reflecting priorities for improvement in other developed countries.1 2 16 Methods Aim The aim of this integrative review was to examine the nature of evidence available to identify how patients are judged by clinicians as being in the final hours or days of life. The following questions formed the basis of the review: How do clinicians diagnose dying? What factors influence clinical decision making about dying? Can any related tools, triggers or guidance for clinicians be sourced?

Methods Aim The aim of this integrative review was to examine the nature of evidence available to identify how patients are judged by clinicians as being in the final hours or days of life. The following questions formed the basis of the review: How do clinicians diagnose dying? What factors influence clinical decision making about dying? Can any related tools, triggers or guidance for clinicians be sourced? An important facet of this review was the inclusion of evidence from a range of chronic progressive conditions across the advanced disease trajectory. The outcomes of this review are presented in a conceptual map (figure 1) constructed using an inductive approach to illustrate the key concepts derived from the narrative synthesis of key literature and the relationships between these. This provides the reader with a conceptual framework that illustrates antecedents to clinical decision making in the complex process of diagnosing dying, when death can be anticipated. Figure 1 Conceptual map of findings.

An important facet of this review was the inclusion of evidence from a range of chronic progressive conditions across the advanced disease trajectory. The outcomes of this review are presented in a conceptual map (figure 1) constructed using an inductive approach to illustrate the key concepts derived from the narrative synthesis of key literature and the relationships between these. This provides the reader with a conceptual framework that illustrates antecedents to clinical decision making in the complex process of diagnosing dying, when death can be anticipated. Figure 1 Conceptual map of findings. Given the scope of the review questions, key aspects of the systematic review methods advocated by the Cochrane Collaboration and Scottish Intercollegiate Guideline Network (SIGN) were tailored to an integrative review design.17 18 While the systematic review of randomised controlled trials is considered to be the gold standard in research synthesis, the integrative review is increasingly recognised as appropriate to inform evidence-based practice. The integrative review synthesises findings from a diverse range of primary experimental and non-experimental research methods to provide breadth of perspectives and a more comprehensive understanding of a complex healthcare problem.19 Our review was based on the steps and processes in box 1 and informed by the PRISMA standards for reporting systematic reviews.20 Box 1 Methods of the integrative review Aims for the review are identified Criteria for including studies were identified

Given the scope of the review questions, key aspects of the systematic review methods advocated by the Cochrane Collaboration and Scottish Intercollegiate Guideline Network (SIGN) were tailored to an integrative review design.17 18 While the systematic review of randomised controlled trials is considered to be the gold standard in research synthesis, the integrative review is increasingly recognised as appropriate to inform evidence-based practice. The integrative review synthesises findings from a diverse range of primary experimental and non-experimental research methods to provide breadth of perspectives and a more comprehensive understanding of a complex healthcare problem.19 Our review was based on the steps and processes in box 1 and informed by the PRISMA standards for reporting systematic reviews.20 Box 1 Methods of the integrative review Aims for the review are identified Criteria for including studies were identified Literature was identified according to an explicit and robust search strategy that was subject to ongoing refinement Studies were selected according to inclusion/exclusion criteria Studies were summarised using an agreed format and key messages were extracted Data extraction was undertaken by two reviewers We present a summary and synthesis of relevant studies Thematic analysis was an ongoing and iterative process involving all team members.

Literature was identified according to an explicit and robust search strategy that was subject to ongoing refinement Studies were selected according to inclusion/exclusion criteria Studies were summarised using an agreed format and key messages were extracted Data extraction was undertaken by two reviewers We present a summary and synthesis of relevant studies Thematic analysis was an ongoing and iterative process involving all team members. Search strategy and results of search The search of electronic databases was comprehensive and included the Cochrane Central Register of Controlled Trials (CENTRAL) on The Cochrane Library, MEDLINE (1950–2011), EMBASE (1980–2011), PsycINFO (1980–2011), CINAHL (1982–2011), Web of Science and Google (to September 2012*). Key search terms included combining *diagnosis (MeSH Diagnos*) with Death*, dying and care. End of life* was combined with Diagnos* and key words such as Recognis* and Instrument*. The search strategy is detailed in box 2. Results from Web of Science and Google were subsequently removed as the search of other databases was discriminating enough and these two databases did not allow us to combine the MeSH and key search terms resulting in an unmanageable numbers of irrelevant hits. Box 2 Search strategy 1. (Death AND (Sign* OR symptom*)) 2. (Dying AND (Sign* OR symptom*)) 3. 1 OR 2 4. 3 AND Diagnos* 5. 3 AND Recognis* 6. 3 AND Assess* 7. 3 AND Tool 8. 3 AND Instrument* 9. 3 AND Criteria 10. 3 AND Pathway* 11. 3 AND Care 12. 4 OR 5 OR 6 OR 7 OR 8 OR 9 OR 10 OR 11 13. End of Life AND Diagnos* 14. End of Life AND Recognis* 15. End of Life AND Assess* 16. End of Life AND Tool

Box 2 Search strategy 1. (Death AND (Sign* OR symptom*)) 2. (Dying AND (Sign* OR symptom*)) 3. 1 OR 2 4. 3 AND Diagnos* 5. 3 AND Recognis* 6. 3 AND Assess* 7. 3 AND Tool 8. 3 AND Instrument* 9. 3 AND Criteria 10. 3 AND Pathway* 11. 3 AND Care 12. 4 OR 5 OR 6 OR 7 OR 8 OR 9 OR 10 OR 11 13. End of Life AND Diagnos* 14. End of Life AND Recognis* 15. End of Life AND Assess* 16. End of Life AND Tool 17. End of Life AND Instrument* 18. End of Life AND Criteria 19. End of Life AND Pathway* 20. End of Life AND Care 21. 13 OR 14 OR 15 OR 16 OR 17 OR 18 OR 19 OR 20 22. 12 AND 21 23. 22 AND Palliative 24. 22 AND Terminal* 25. 22 AND Final 26. 22 AND Cancer* 27. 22 AND Coronary 28. 23 OR 24 OR 25 OR 26 OR 27 29. Limit 28 to Human, English Language, 2001–2011, and Primary Studies, Meta-Analyses, Systematic Reviews, RCTs, Europe, North America, Australasia. We included systematic reviews, qualitative and quantitative studies that explored how patients are judged by clinicians as being in the final hours or days of life. The population investigated was adult patients with progressive life-limiting illness in hospital, hospice, community or care home settings. Our focus was to establish what factors influence clinical decision making by health professionals in determining when a patient is in the dying phase, including any tools, triggers or guidance. Papers that focused on the transition to palliative care and prognostication were excluded, as beyond our aim to extract data relevant to the last hours or days of life. Literature about the LCP was included where this considered diagnosing dying and therefore met the inclusion criteria for the review.

ers or guidance. Papers that focused on the transition to palliative care and prognostication were excluded, as beyond our aim to extract data relevant to the last hours or days of life. Literature about the LCP was included where this considered diagnosing dying and therefore met the inclusion criteria for the review. The search included literature published since 2001 in English from Europe, North America and Australasia as these were judged to be culturally relevant to a European context. A 10-year timescale was considered sufficient to ensure comprehensive coverage and currency of relevant literature given the increasing emphasis during this timeframe on planning end-of-life care. The initial search strategy generated 576 titles. In total, 245 titles were excluded due to being duplicates or not relevant to the overall aims of the review by three reviewers (CK, CG and PBY) who screened the original yields from the searches. These three reviewers independently reviewed 331 abstracts and agreed initial inclusion of papers in the review at a consensus meeting. In addition to the place and year of publication, papers had to explore diagnosing dying (the dying phase: last hours or days of life) in adults rather than the transition to palliative care where prognosis was a longer term. At this stage, 42 papers were identified and further scrutinised. Nineteen papers were excluded because they did not present data about the last hours and days of life and this rendered 23 papers for inclusion (figure 2). Searches of grey literature identified a number of conference posters and some work in progress. Authors were contacted where possible, but this yielded insufficient or no further data for inclusion.

se they did not present data about the last hours and days of life and this rendered 23 papers for inclusion (figure 2). Searches of grey literature identified a number of conference posters and some work in progress. Authors were contacted where possible, but this yielded insufficient or no further data for inclusion. Figure 2 Search results. Review process and quality assurance Once the 23 papers for inclusion were identified, the eight reviewers were paired and each pair reviewed 5–6 studies through extracting key data and findings using a preformatted table (see online supplementary table S1). This was an important part of the quality assurance process as ensuring all reviewers used a consistent approach to data extraction. Reviewers first read their allocated studies independently then agreed findings with a partner, which were presented in a summary table. This produced 23 summaries, one for each included study. We used the SIGN hierarchy of evidence to assign a quality rating to the papers in order to allow us to draw conclusions and make an overall judgement about the quality of evidence available in the field. All reviewers received detailed summaries of all the included papers and then participated in the refinement of themes and a conceptual framework initially developed by three reviewers (CK, CG and PBY). Robust methods were employed throughout the review, with analysis an ongoing and iterative process involving all team members. This included two full-team consensus meetings via teleconference, in addition to ongoing dialogue and feedback via email.

ual framework initially developed by three reviewers (CK, CG and PBY). Robust methods were employed throughout the review, with analysis an ongoing and iterative process involving all team members. This included two full-team consensus meetings via teleconference, in addition to ongoing dialogue and feedback via email. Results A central overarching theme of uncertainty in diagnosing dying emerged, alongside two subthemes, characteristics of dying and treatment orientation. What follows is the presentation of the thematic and conceptual findings from the review process. Review findings on ‘characteristics of dying’ Thirteen papers identified characteristics of dying: one systematic review,21 seven retrospective chart reviews,22–27 two qualitative studies,28 29 one structured interview study,30 one quantitative study,31 one literature review32 and one survey.33 The review by Plonk (2005) included all research relevant to death, terminal care and bereavement available on Medline between 1990 and 2004; two studies focused on older people in the nursing home setting29 30 and four studies focused on cancer (Ethunandan et al 2005):24 25 27 31 one on patients with stroke (Mazzocato et al 2010)26 , three on both cancer and other long-term conditions (Goodman et al 2002)23 28 32, one on amyotrophic lateral sclerosis (ALS)22 and one on medical decision making at the end of life.33 Findings from these studies demonstrate dying trajectories that incorporate physical, social, spiritual and psychological decline towards death and the extent to which each study dealt with these varied.

Review findings on ‘characteristics of dying’ Thirteen papers identified characteristics of dying: one systematic review,21 seven retrospective chart reviews,22–27 two qualitative studies,28 29 one structured interview study,30 one quantitative study,31 one literature review32 and one survey.33 The review by Plonk (2005) included all research relevant to death, terminal care and bereavement available on Medline between 1990 and 2004; two studies focused on older people in the nursing home setting29 30 and four studies focused on cancer (Ethunandan et al 2005):24 25 27 31 one on patients with stroke (Mazzocato et al 2010)26 , three on both cancer and other long-term conditions (Goodman et al 2002)23 28 32, one on amyotrophic lateral sclerosis (ALS)22 and one on medical decision making at the end of life.33 Findings from these studies demonstrate dying trajectories that incorporate physical, social, spiritual and psychological decline towards death and the extent to which each study dealt with these varied. Physical characteristics included symptoms such as dyspnoea, which was correlated with shorter survival across cancer and other diagnoses such as ALS and may be a marker of the final phase of the dying process.22 23 Chart reviews of 42 patients with stroke showed multiple causes of dyspnoea such as aspiration pneumonia or cardiac failure, which were a major or a contributing cause of death in 78% of these patients.26 Physical decline and symptoms such as pain, loss of appetite and increased dependency are also reported as more prevalent shortly before death in patients with cancer.31

le causes of dyspnoea such as aspiration pneumonia or cardiac failure, which were a major or a contributing cause of death in 78% of these patients.26 Physical decline and symptoms such as pain, loss of appetite and increased dependency are also reported as more prevalent shortly before death in patients with cancer.31 Glioma patients have particular physical rather than psychosocial symptoms such as decreased level of consciousness, dysphagia, neurological deficit and seizures in the last phase of life. Tailored guidelines may then be required for this group of patients to inform anticipatory care planning and treatment decisions at the end of life.25 27 A number of studies identified that for both cancer and non-cancer patients psychosocial characteristics were present but received less attention at the end of life. These included issues with coping, struggling with the meaning and purpose of life, impaired cognition and conscious level.28–31 While the relationship between physical and psychosocial symptoms such as pain and existential or psychosocial suffering remains poorly studied, there is increasing evidence of medical interventions at the end of life.21 33

g, struggling with the meaning and purpose of life, impaired cognition and conscious level.28–31 While the relationship between physical and psychosocial symptoms such as pain and existential or psychosocial suffering remains poorly studied, there is increasing evidence of medical interventions at the end of life.21 33 The relationship between physical and psychosocial symptoms also emerged in a UK study of patients with heart failure (n=24) and patients with lung cancer (n=24). Issues relating to the meaning and purpose of life may be perceived by patients as more important than physical well-being, symptoms or support near the end of life. As physical symptoms increased, there was a parallel decline in social well-being, with social withdrawal identified as an independent marker of the immanence of death. In patients with lung cancer, the physical and social care trajectories were interlinked with psychological and spiritual issues and this became more evident at times of transition, including to the final phase of the illness. Similarly, for patients with heart failure, social and psychological decline ran in parallel with physical deterioration.28 As noted for patients with head and neck cancer (N=32), insufficient recording of the psychosocial and spiritual aspects of care means that the ‘complete quality of dying’ experience cannot be assessed.24

y, for patients with heart failure, social and psychological decline ran in parallel with physical deterioration.28 As noted for patients with head and neck cancer (N=32), insufficient recording of the psychosocial and spiritual aspects of care means that the ‘complete quality of dying’ experience cannot be assessed.24 The influence of perceived suffering on medical decision making was revealed in a large survey (n=20 480) of doctors across six European countries.33 The proportion of deaths preceded by decisions made in response to suffering (as perceived by the doctors rather than an explicit request from the patient) varying between 23% (Italy) to 51% (Switzerland). Sudden and unexpected death occurred in about a third of cases in all six countries. The administration of drugs with the explicit intention of hastening death varied between countries: about 1% or less in Denmark, Italy, Sweden and Switzerland; 1.82% in Belgium and 3.40% in the Netherlands.

taly) to 51% (Switzerland). Sudden and unexpected death occurred in about a third of cases in all six countries. The administration of drugs with the explicit intention of hastening death varied between countries: about 1% or less in Denmark, Italy, Sweden and Switzerland; 1.82% in Belgium and 3.40% in the Netherlands. Residents in nursing homes experience high rates of physical and emotional symptoms at the end of life, which may not be well-managed, as in this setting the interpretation of ‘comfort care’ can vary and end-of-life care needs may not then be recognised or addressed.29 30 In the frail elderly, physical indicators were identified as anorexia, increased pain, greater frailty or weakness, weight loss, altered breathing patterns (including apnoea), mottled skin and a distinctive odour. Behavioural indicators include personality and mood changes, with increased restlessness, anxiety or agitation described as characteristics of the final decline to death. Withdrawal from conversation with staff and a new tendency to decline social interaction with other residents or family members was also identified. For people with dementia, an integrative literature review highlighted a range rather than definitive characteristics of dying and little consensus regarding the value of prognostic indicators.32

tion with staff and a new tendency to decline social interaction with other residents or family members was also identified. For people with dementia, an integrative literature review highlighted a range rather than definitive characteristics of dying and little consensus regarding the value of prognostic indicators.32 One study explored the application of the Palliative Performance Scale (PPS) to predict death/length of survival in a hospital setting for patients with cancer (n=157) and other diagnoses (n=104). Health professionals were more likely to regard cancer as a terminal illness than other groups, and while the PPS was useful in describing mortality risk across populations, it is less useful for predicting death in individual patients.23

a hospital setting for patients with cancer (n=157) and other diagnoses (n=104). Health professionals were more likely to regard cancer as a terminal illness than other groups, and while the PPS was useful in describing mortality risk across populations, it is less useful for predicting death in individual patients.23 The included studies identified a range of physical and psychosocial characteristics of dying. There was variability in recording and measurement including factors such as severity and progression rate. No definitive criteria for diagnosing dying were identified across all patient groups. Professional decision making may also be influenced by the diagnosis of the patient and care setting. A cancer diagnosis may prompt the health professional to look for confirming evidence that death may be imminent as opposed to other life-limiting conditions, particularly in frail older people who may receive comfort, rather than end-of-life care. Evidence also suggests that a focus on physical symptoms may also support the tendency towards ongoing biomedical interventions despite the imminence of death. This is explored further in the second subtheme; treatment orientation.

articularly in frail older people who may receive comfort, rather than end-of-life care. Evidence also suggests that a focus on physical symptoms may also support the tendency towards ongoing biomedical interventions despite the imminence of death. This is explored further in the second subtheme; treatment orientation. Review findings on ‘treatment orientation’ Ten papers reported findings on the factors influencing end-of-life decision making including diagnosing dying, care and symptom management. Methodologies were two case reviews,34 35 one exploratory interview study,36 two mixed methods,37 38 one quantitative study,39 one retrospective cross-sectional survey of bereaved relatives,40 one qualitative study,41 one action research study42 and one case review of patients on LCP.43

ng, care and symptom management. Methodologies were two case reviews,34 35 one exploratory interview study,36 two mixed methods,37 38 one quantitative study,39 one retrospective cross-sectional survey of bereaved relatives,40 one qualitative study,41 one action research study42 and one case review of patients on LCP.43 Findings from these papers suggest that while nurses and doctors may diagnose dying or report being able to do so, medically futile interventions may be continued in the very last phase of life.36–38 Where poor prognosis and the dying phase was recognised and documented, there was still little evidence of systematic planning for end-of-life care. Consequently, diagnostic testing and unnecessary interventions were continued in the last 24 h of life.37 A German study of medical and nursing practice in caring for dying patients (n=252) in a large hospital identified that plans for end-of-life care were implemented rarely and late. Fifty-two per cent of patient died in ICU with a focus on life-prolonging interventions evident.34 In Italy, a review of the medical and nursing records of the last 3 days of life for 370 patients reported that patients were exposed to invasive treatments despite close proximity to death and with palliative care expertise sought in only 7% of cases.38 Large variations have also been recorded regarding the extent to which decisions are discussed with patients, relatives and other caregivers and the need for much greater involvement of patients and families in end-of-life decision making acknowledged.33

and with palliative care expertise sought in only 7% of cases.38 Large variations have also been recorded regarding the extent to which decisions are discussed with patients, relatives and other caregivers and the need for much greater involvement of patients and families in end-of-life decision making acknowledged.33 The influence of diagnosis on professional responses was also identified from a survey of bereaved carers.35 In contrast to patients with cancer, those with other life-limiting illnesses were less likely to be diagnosed as dying and more likely to be moved to hospital prior to death.40 Older patients in the end stages of heart failure or dementia admitted through accident and emergency departments (n=102) have been reported as disadvantaged in relation to timely end-of-life care.35

e-limiting illnesses were less likely to be diagnosed as dying and more likely to be moved to hospital prior to death.40 Older patients in the end stages of heart failure or dementia admitted through accident and emergency departments (n=102) have been reported as disadvantaged in relation to timely end-of-life care.35 Three included papers were linked to implementation of the LCP.41–43 A retrospective review of patient data (n=168) in one hospice reported the LCP entry criteria as appropriate for a cancer population; combining multiprofessional team agreement that the patient is dying with additional criteria including bed-bound, semi-comatose and only able to take sips of fluid.43 These criteria were supported by a phenomenological study of hospice doctors and nurses (n=10), which also highlighted the influence of additional factors. These included understanding the individual patient's history, the experience of the clinician and the views of family and carers. Staff experienced anxiety over potential mistiming of the LCP and causing distress for patients and families. The authors suggest the risk that clinicians may then avoid difficult conversations with families struggling to accept the reality of dying.41 An action research study in eight UK nursing homes identified a functional rehabilitative culture and that staff lacked knowledge and skills related to diagnosing dying and end-of-life care that made implementation of the LCP challenging, requiring a cultural shift, change to care processes and facilitative education.42

n action research study in eight UK nursing homes identified a functional rehabilitative culture and that staff lacked knowledge and skills related to diagnosing dying and end-of-life care that made implementation of the LCP challenging, requiring a cultural shift, change to care processes and facilitative education.42 Implementation of the Inpatient Comfort Care Program (ICCP) (preintervention n=108; postintervention n=95) suggests it may be possible to introduce palliation in hospital units focused on curative interventions that positively impacts on end-of-life care, although the primary outcomes stated are linked to physical as opposed to psychosocial symptoms. The primary process of care, presence of an order for opioid medication at the time of death, showed a highly significant change from 57.1% to 83.2%. Presence of a do not resuscitate order at the time of death increased from 61.9% to 85.1%, with indicators that most of the dying patients were recognised and their preferences for end-of-life care documented to 90.8%. Improved documentation suggested that the intervention was successful in stimulating staff to recognise symptoms and better identify patients who are dying.39

th increased from 61.9% to 85.1%, with indicators that most of the dying patients were recognised and their preferences for end-of-life care documented to 90.8%. Improved documentation suggested that the intervention was successful in stimulating staff to recognise symptoms and better identify patients who are dying.39 It is noteworthy that seven of the papers focus on acute hospital settings, and despite the limited number of studies, decision making related to diagnosing dying in this setting may remain focused towards biomedical interventions rather than systematic planning for end-of-life care. There is some evidence, however, that while factors such as the context of care and the diagnosis of the patient may influence decision making, measures including education and supportive care processes can facilitate appropriate end-of-life care.

ical interventions rather than systematic planning for end-of-life care. There is some evidence, however, that while factors such as the context of care and the diagnosis of the patient may influence decision making, measures including education and supportive care processes can facilitate appropriate end-of-life care. Conceptual map of key findings Synthesis of the findings revealed a number of antecedents to decision making and possibilities for action. The conceptual framework (figure 1) provides a visual representation of the possible relationships between decisions, responses and strategies that are influenced by the clinical context. The evidence of active medical intervention rather than end-of-life care reported in several studies we interpreted as resulting from ‘uncertainty in diagnosing dying’ due to a number of factors that can influence decision making and responses. Care orientation in an acute hospital is more likely to result in active treatment as opposed to the nursing home, where comfort or rehabilitative care may be the focus. Our findings demonstrate that diagnosis is important, with cancer linked to more explicit recognition of dying than other life-limiting illnesses. Professionals’ expertise in end-of-life care is also important in determining their course of action.

he nursing home, where comfort or rehabilitative care may be the focus. Our findings demonstrate that diagnosis is important, with cancer linked to more explicit recognition of dying than other life-limiting illnesses. Professionals’ expertise in end-of-life care is also important in determining their course of action. Discussion The overarching theme of ‘uncertainty in diagnosing dying’ is representative of the current discourse within literature and echoes our limited understanding of what constitutes a ‘good death’.24 It also reflects a tendency towards continued unnecessary interventions and treatment into the last days of life, even when nurses and doctors recognise death is imminent.44 The complexity of identifying clinical indicators of dying across different diseases supports the drive towards care based on needs rather than diagnosis.21 37 However, expanding our understanding of the characteristics of dying for different patient groups is important to support thorough and comprehensive clinical assessments, determining whether reversible causes for decline are present and can be addressed or whether death may be imminent.45 Of consideration, however, is the need for consensus regarding terminology. A range of terms may be used for the final phase of life, with varied and often implicit rather than clearly stated definitions, which can confound synthesis of evidence and cause ambiguity in relation to care goals.46

death may be imminent.45 Of consideration, however, is the need for consensus regarding terminology. A range of terms may be used for the final phase of life, with varied and often implicit rather than clearly stated definitions, which can confound synthesis of evidence and cause ambiguity in relation to care goals.46 Of relevance to the current discourse regarding the LCP, this review has highlighted a range of contextual factors that may influence diagnosing dying reflecting calls for research in this area.14 47 The first randomised controlled trial to assess the effectiveness of the LCP reported no significant difference in the overall quality of end-of-life care for people with cancer and highlighted the influence of contextual factors.9 The authors also question whether focusing only on the last few hours or days of life may influence optimum management for those with difficult symptoms and communication with the family. This reflects earlier discourse from 2001 and further supports the need for clarification of current definitions to reflect the temporal nature of dying.43 For patients and families, planned care in place of choice, time with loved ones to say goodbye and being prepared to die are recognised as important at the end of life.31 Yet, even when death may be imminent, there remains a tendency for professionals to avoid or postpone related conversations, reflecting the reality that many care decisions are informed by role and personal world views.36

ed ones to say goodbye and being prepared to die are recognised as important at the end of life.31 Yet, even when death may be imminent, there remains a tendency for professionals to avoid or postpone related conversations, reflecting the reality that many care decisions are informed by role and personal world views.36 The importance of exposing uncertainty in diagnosing dying cannot be overly emphasised. It is an important consideration for the individual clinician and the healthcare team, but equally for patients and carer(s). Allowing the possibility of death to be acknowledged and any plan of treatment to be discussed realistically gives precious time for the patient and family to express their preferences. This facilitates an individualised plan of care, particularly where no restorative treatment is possible or further decline would herald the end of life. Paradoxically, being open about uncertainty may support planning and build trust with the patient and their carer(s), removing the need to rely on a ‘tick box’ approach to diagnosing dying.14 45

alised plan of care, particularly where no restorative treatment is possible or further decline would herald the end of life. Paradoxically, being open about uncertainty may support planning and build trust with the patient and their carer(s), removing the need to rely on a ‘tick box’ approach to diagnosing dying.14 45 Thus, strategies used to reduce uncertainty should reflect that decision making is influenced by clinical skills, professional judgement and clinical wisdom. In this context, clinical wisdom describes a depth of intuitive expertise, reflective of a strong skill base, which frames the decision-making process. Clinical wisdom is evident when symptoms are potentially reversible or when the reality of dying is in evidence and a decision on the right course of action is needed. Clinical wisdom assists in determining appropriate goals of care and interventions to ensure that the patient's wishes are met and a dignified death is achieved.48 Conclusion The findings of this review support the explicit recognition of ‘uncertainty in diagnosing dying’ and the need to work with and within this concept. Clinical decision making needs to allow for recovery where that potential exists, but equally there is the need to avoid futile interventions. This requires contextual factors to be addressed, reliable care processes and systems and education for healthcare professionals alongside raising public awareness of issues related to death and dying.

needs to allow for recovery where that potential exists, but equally there is the need to avoid futile interventions. This requires contextual factors to be addressed, reliable care processes and systems and education for healthcare professionals alongside raising public awareness of issues related to death and dying. Most of the included studies were based on retrospective case reviews or small qualitative studies and scored low/moderate quality ratings, and the findings should be viewed in this context. The search strategy and methods for synthesising a disparate collection of studies have limitations. However, we have reported our methods in detail and consistent patterns emerged to illustrate important insights into diagnosing dying. The conceptual map requires further testing and refinement but provides a basis for further exploration. It also has the potential to provide a framework for discussion in clinical teams around strategies for improvement in diagnosing dying. Supplementary Material Web supplement Contributors: CK planned the review. CBG, PB-Y and CK scrutinised the searches and agreed papers for inclusion in the review. All authors participated in the review of papers. CK constructed the initial conceptual framework, and this was subsequently refined in discussions with all team members. CK drafted the manuscript with the exception of the section on uncertainty that was drafted by PL and MC. All authors agreed on the content of the final manuscript. Competing interests: None. Provenance and peer review: Not commissioned; externally peer reviewed.

Introduction A growing body of research indicates that physicians suffer high levels of stress, depression and burnout.1–3 Related literature has found that physician stress can negatively impact patient care.4–7 Patient death is a complex stressor in physicians’ work. Experts in physician wellness have acknowledged ‘the death of patients is an experience that almost all physicians must confront’8 and asserted ‘psychological distress (among physicians) derives largely from identification with suffering, the presence of death and the spectre of failure’.9 Given the close relationships physicians develop with patients, it is reasonable to expect that patient deaths could affect physician stress and resilience. The current research grew out of earlier qualitative research conducted by David Kuhl, who supervised my research and acted as consultant to this article. In his research Kuhl explored the lived experience of patients with terminal illnesses10 11 and found evidence of â iatrogenic sufferingâ: the suffering of patients caused by the way doctors speak to them. The purpose of the current research was to use qualitative methods to understand the ‘other side’ of iatrogenic suffering: the experience of physicians dealing with patient death, and how it contributes to iatrogenic suffering for patients.

sufferingâ: the suffering of patients caused by the way doctors speak to them. The purpose of the current research was to use qualitative methods to understand the ‘other side’ of iatrogenic suffering: the experience of physicians dealing with patient death, and how it contributes to iatrogenic suffering for patients. Methods This research used a phenomenological qualitative method.12 The phenomenological approach attempts to capture the essence of a phenomenon through detailed analysis of a number of particular instances of the lived experience of that phenomenon. In this case, individual, in-depth interviews with 10 senior physicians explored (1) the daily experience of patient death, (2) particularly memorable patient deaths and (3) the impact of these experiences on their personal and professional lives. The interviews were transcribed and thematically analysed in order to identify the themes that described the essential structure of the lived experience of patient death for these physicians. Individual themes were identified for each participant and validated through a detailed member-check in a second interview. Individual themes were then revised according to the comments and suggestions of the participants and a set of common themes was created. The final set of 10 common themes was then validated by a further member-check. An independent expert in the field was also asked to review the final 10 themes, once they had been agreed to by the participants. This expert was a family physician with many years of experience as a teacher and administrator who had a keen interest in the doctor-patient relationship, physician communication practices and physician resilience. The final 10 themes were separated into three categories: the impact of the context surrounding the experience of patient death; the immediate experience of dealing with patient death; and the impact of the experience on the physicians’ personal and professional lives. Due to space constraints this article will focus on the five core themes that emerged as descriptions of the immediate experience of patient death for these physicians. The other themes will be presented in later articles.

tient death; and the impact of the experience on the physicians’ personal and professional lives. Due to space constraints this article will focus on the five core themes that emerged as descriptions of the immediate experience of patient death for these physicians. The other themes will be presented in later articles. Participants The five male and five female participants were all Caucasian. All were affiliated with a large, tertiary care hospital in Vancouver, Canada, had a minimum of 3 years work experience and self-identified as being exposed to multiple patient deaths. Three came from palliative care, two from emergency, two from hospital transplant teams, two from family practice and one from intensive care. The physicians ranged in age from 35 to 60 (average 48) and had from 6 years to over 20 years of experience (average 16.5). They represent a convenience sample identified through informal work contacts and then contacted with a formal letter of invitation. Interest in participating in this study was in part due to Kuhl's earlier research that identified ‘iatrogenic suffering’. This previous research was known to most of the participants, and the letter of invitation identified that the current study was a follow-up to that research.

ed with a formal letter of invitation. Interest in participating in this study was in part due to Kuhl's earlier research that identified ‘iatrogenic suffering’. This previous research was known to most of the participants, and the letter of invitation identified that the current study was a follow-up to that research. Results Theme (1) memories beneath the surface Most participants responded to the question concerning daily experience of patient death by saying it was a ‘normal part of the job’: ‘I think because I'm around it all the time, it's just kind of normal’ (Dr B). ‘It's become part of what I expect when I go to work’ (Dr H). However, the question of particularly memorable patient deaths brought up strong memories. Before our interview Dr G anticipated this question:I first thought…I can't remember any—and then boom … this unbelievable image of a 21-year-old guy who'd been in a car accident… This was 20 years ago, and it's still one of the most vivid experiences that stays with me.

rable patient deaths brought up strong memories. Before our interview Dr G anticipated this question:I first thought…I can't remember any—and then boom … this unbelievable image of a 21-year-old guy who'd been in a car accident… This was 20 years ago, and it's still one of the most vivid experiences that stays with me. The physicians’ recollections seemed to spring suddenly into consciousness. Their descriptions were accompanied by intense emotional reactions and vividly recalled details. Dr J described a memory from many years ago, which affected her deeply: ‘It triggered me … I felt this incredible flood of sadness,… emotion, … grief,… fear’. Most memorable patient deaths occurred early in the physicians’ careers and the consequences were often substantial regarding changes in career or personal choices. While daily experiences of patient death may have become ‘normal’, these memories carried a clarity and intensity reminiscent of traumatic memories, vividly powerful, just below the surface.

red early in the physicians’ careers and the consequences were often substantial regarding changes in career or personal choices. While daily experiences of patient death may have become ‘normal’, these memories carried a clarity and intensity reminiscent of traumatic memories, vividly powerful, just below the surface. Theme (2) expectation and responsibility The experience of patient death was consistently associated with intense feelings of responsibility and an awareness of expectations from patients, families, the medical system and society. The physicians described how responsibility over life and death was attached to their role early on. Expectations of the ‘heroic ideal’ of the physician were ingrained by their families first, and then encountered in medical school and residency. Several physicians described the expectations and responsibilities as ‘brutal’ and ‘inhuman’ (Dr C) or ‘impossible to fulfil’ (Dr H). One physician summarised these expectations: ‘We're supposed to know everything, and have the answer for everything, and somehow be larger than life’ (Dr B). Many described how the sense of intense personal responsibility over the life or death of a patient was something they were forced to come to terms with during their careers. One physician described it as ‘getting away from the expert culture of medicine’ (Dr I). Another, who described being seen as a ‘miracle worker’ by many patients and families, said he had to limit this kind of adulation: ‘The miracle wasn't me…that obviously feels good, but … nobody lives up to that’ (Dr C). These dynamics suggest that throughout a physician's career it can be difficult to balance the expectations of ‘medical responsibility’13 with the uncertain and unpredictable nature of patient care.

to limit this kind of adulation: ‘The miracle wasn't me…that obviously feels good, but … nobody lives up to that’ (Dr C). These dynamics suggest that throughout a physician's career it can be difficult to balance the expectations of ‘medical responsibility’13 with the uncertain and unpredictable nature of patient care. Theme (3) the question of competence Many memorable patient deaths caused the physicians to question whether they had done everything they could have, or should have, for the patient. In some cases they questioned whether certain decisions or actions may have caused harm, or even led to the patient's death. Mortality and morbidity rounds are intended to review difficult patient outcomes, learn from mistakes and avoid personal blame; however, they often magnify the question of competence:You're exposed, and… people are able to see you're not perfect, and you have flaws… You always want that ability to say you are competent—and that detracts from your professional identity (Dr H).

icult patient outcomes, learn from mistakes and avoid personal blame; however, they often magnify the question of competence:You're exposed, and… people are able to see you're not perfect, and you have flaws… You always want that ability to say you are competent—and that detracts from your professional identity (Dr H). Several physicians commented that in order to feel they had acted competently, they needed to feel ‘clean’ (Dr J) about the patient's death—that is, for the physician to be able to say, ‘I know that I've done everything that could conceivably have been done’ (Dr G). However, many situations include an inherent ambiguity that makes this difficult to achieve. Dr H pointed out that professional competence for physicians is not easily measured. What can be reasonably expected from a physician depends upon the situation, including what information and resources were available and what particular skills the physician had. A sense of competence depends in part upon how one compares among peers. ‘I think we all compare ourselves to other people. … That's something that probably everybody struggles with’ (Dr H). Dr J described an obstetrical emergency that created ‘terror’ regarding her competence:The terror of… how deeply responsible I felt to be competent in that moment, and how the chance of not being able to do something, or not doing it right, is so overwhelming. … I might never know whether it's that I didn't do the algorithm right, or I didn't do it quickly enough, or… whether it was just this moment of fate and there was nothing that I could have done about it. You'll never get a clear answer.

ance of not being able to do something, or not doing it right, is so overwhelming. … I might never know whether it's that I didn't do the algorithm right, or I didn't do it quickly enough, or… whether it was just this moment of fate and there was nothing that I could have done about it. You'll never get a clear answer. Theme (4) breakthrough experiences Descriptions of particularly memorable patient deaths often included unexpected details that jarred the physicians out of their normal awareness and connected them to the meaning or emotion of the event. These ‘breakthrough experiences’ would ‘hit home’ (Dr H) in a deeply personal way. One physician described a mother's comment upon seeing her 16-year-old boy dead after a car accident:‘He just got his hair cut today.’—She'd seen him earlier that day and everything was normal, and then the next thing you know… [it] makes you sit up and think about your own life (Dr H).

(Dr H) in a deeply personal way. One physician described a mother's comment upon seeing her 16-year-old boy dead after a car accident:‘He just got his hair cut today.’—She'd seen him earlier that day and everything was normal, and then the next thing you know… [it] makes you sit up and think about your own life (Dr H). These breakthrough experiences often involved a sense of identification or relationship with patients that resulted in a personal awareness of mortality or other unresolved emotional experience for the physician. ‘The ones I have trouble with are patients reminding me of my parents… that's too close to home’ (Dr D). Dr H described a patient who was close to his own age, also married with children, and previously perfectly healthy, yet he died within 24 h of entering the hospital: ‘It made me feel so vulnerable for my own mortality…Life can be ripped out of your hands so quickly’ (Dr H). The quality of these ‘breakthrough experiences’ indicated how deeply these physicians were touched by ‘the human part of the tragedy’ (Dr J).

perfectly healthy, yet he died within 24 h of entering the hospital: ‘It made me feel so vulnerable for my own mortality…Life can be ripped out of your hands so quickly’ (Dr H). The quality of these ‘breakthrough experiences’ indicated how deeply these physicians were touched by ‘the human part of the tragedy’ (Dr J). Theme (5) action versus presence Many non-palliative physicians reported that at critical moments surrounding a patient's death they entered a ‘mode’ or state of mind in which their awareness was intensely focused on the actions they needed to perform. Palliative physicians, on the other hand, reported a very different sense: being present with the patient in a way that provided calm reassurance and support. Careful review of the interview protocols revealed that at patients’ deaths many physicians experienced both of these ‘modes’ as well as a tension between them. Action mode When physicians described having to deal with a patient in crisis, many reported similar experiences of entering into ‘protocol mode’ (Dr G), ‘automatic mode’ (Dr F), ‘mechanical mode’ (Dr I) or ‘action mode’ (Dr J). The shift into this action-oriented state of mind helped the physicians focus and set aside any emotional reactions. ‘At that moment, you don't think. You just go into that automatic mode… It's almost like you turn off part of your brain’ (Dr F). Dr G described how he managed his personal reactions:[I feel] an intensified feeling in my stomach, racing thoughts, and… I have to force myself down. I slow down and go into protocol mode. I have to go back and say the ‘A, B, C's.

t go into that automatic mode… It's almost like you turn off part of your brain’ (Dr F). Dr G described how he managed his personal reactions:[I feel] an intensified feeling in my stomach, racing thoughts, and… I have to force myself down. I slow down and go into protocol mode. I have to go back and say the ‘A, B, C's. Presence mode In contrast, the palliative care physicians described ‘simply being present’ with dying patients once they no longer needed to do something for them. Addressing the patient's medical needs was important; however, after all appropriate medical interventions, the need for ‘presence’ was still important for patients, families and the physicians themselves.There are certain things that it's good to try and fix, … but … [when] it can't be fixed you have to be okay… Okay in the presence of death, knowing that some suffering you can't change, and that you'll be there with them in it (Dr B). All three palliative physicians appreciated the quality of relationship they had with dying patients. ‘It's a very close sort of relationship because people when they're dying are often very honest and open, and they need us to be honest and open as well’ (Dr E).

Presence mode In contrast, the palliative care physicians described ‘simply being present’ with dying patients once they no longer needed to do something for them. Addressing the patient's medical needs was important; however, after all appropriate medical interventions, the need for ‘presence’ was still important for patients, families and the physicians themselves.There are certain things that it's good to try and fix, … but … [when] it can't be fixed you have to be okay… Okay in the presence of death, knowing that some suffering you can't change, and that you'll be there with them in it (Dr B). All three palliative physicians appreciated the quality of relationship they had with dying patients. ‘It's a very close sort of relationship because people when they're dying are often very honest and open, and they need us to be honest and open as well’ (Dr E). The tension between these two ‘modes’ arises because physicians must either exclude or include an awareness of personal experience for themselves and their patients. Action mode requires physicians to shut down awareness of personal experience: ‘You have to go into that checklist place, which becomes quite impersonal… You absolutely have to do that’ (Dr J). Dr H described it as a ‘functional disconnect’: ‘There is a separation between your professional and your private life that has to be there … a disconnect … that has to be there for you to function’. These physicians’ descriptions of action mode indicated that it disconnected them from their own emotional reactions and from the patient as a person: ‘You've been so trained, you know—this is what has to be done at this time—that you just stop thinking of the patient as a person—as an individual’ (Dr F). The personal awareness of presence mode is the antithesis of this functional disconnect. Dr I described a poignant example of the tension between action and presence. He was called to an emergency in which the patient was someone he knew personally:I came in thinking in the mechanical mode – Then – ‘I know him, he's younger than me, and he's dying!’… and then, between ‘you do this, you get that, you do the other thing’, I would flip into ‘Oh my god, oh my god, oh my god,’ and have tears in my eyes – back and forth and back and forth.

as someone he knew personally:I came in thinking in the mechanical mode – Then – ‘I know him, he's younger than me, and he's dying!’… and then, between ‘you do this, you get that, you do the other thing’, I would flip into ‘Oh my god, oh my god, oh my god,’ and have tears in my eyes – back and forth and back and forth. Discussion The findings of this study expand our understanding of ‘the other side of iatrogenic suffering.’ They confirm other recent research indicating that while physicians do not have strong emotional reactions to every patient death, many report that some patient deaths have a strong emotional impact on them.14 15 One qualitative study reported ‘anti-emotion expectations’16 in physician training and culture, yet identified that some patient deaths result in intense emotional reactions which can remain buried for decades and still erupt suddenly in vivid memories. In recent years there has been increased awareness of the need to support physicians in training to help them manage the emotional impact of difficult work-related experiences.10 17 18 Changes have been made in some medical programmes to address this, and there has been some evidence of success.19 However, many recent studies indicate that there is still little support for medical students and residents who experience emotional reactions to patient suffering and death.14 20–22 Several authors suggest there is a ‘hidden curriculum’21 or ‘conspiracy of silence’14 23 in physician training that encourages a repression of emotional reactions. The current study indicates that physicians’ experiences of patient death are complex and at times dominated by an awareness of professional responsibilities. Emotional processing can be set aside in order to ‘functionally disconnect’ from personal emotional reactions and remain focused on the technical aspects of the physician's role. This provides a new perspective on ‘anti-emotion expectations’ in medical culture. It seems that as long as physicians deal with patient death, they will at times need to go into action mode and functionally disconnect in order to sustain their professional role. The balancing process of reconnection needs to become as integral a part of physician training and role expectation as the functional disconnect. In this way the emotional processing that is part of self-regulation and effective communication with patients will also be supported within medical training and culture.

le. The balancing process of reconnection needs to become as integral a part of physician training and role expectation as the functional disconnect. In this way the emotional processing that is part of self-regulation and effective communication with patients will also be supported within medical training and culture. The results of this study confirm that physicians’ process of grieving patient death is somewhat different from that described in traditional bereavement models. Papadatou identified that while healthcare providers (HCPs) may experience grief related to the loss of the individual patient, which could be consistent with bereavement models for the general population, there are other dimensions of loss, which can be substantially different for HCPs given the nature of their work context. The most important of these is the possible loss of confidence or recognition in the self-identity and self-efficacy expectations attached to the professional role.24 Papadatou put forward a professional grieving model for HCPs that describes a fluctuation between experiencing and repressing grief reactions. She points out that there is a similarity with the ‘dual process model’25 of bereavement in that both describe an oscillation between orienting either towards the loss experience or away from the loss towards continued engagement with life. This oscillation is similar to the dynamics between action and presence as well; however, for Papadatou and for the action/presence dynamics, emotional experience is often avoided by HCPs in order to preserve the professional role and ‘get the job done’. A point of difference from Papadatou's model is that action mode and presence mode describe two states of mind that occur in relation to patient death, but they do not yet constitute a grieving ‘process’. Physicians and other HCPs are confronted with many patient deaths over time, resulting in the possibility of repeating cycles of loss and adjustment. Further research is necessary to understand how the dynamics between action and presence influence the grieving process and psychological adjustment over time.

‘process’. Physicians and other HCPs are confronted with many patient deaths over time, resulting in the possibility of repeating cycles of loss and adjustment. Further research is necessary to understand how the dynamics between action and presence influence the grieving process and psychological adjustment over time. In recent literature on physician wellbeing there is evidence of a growing tension as physicians struggle to keep up with technical and bureaucratic demands while also being asked to improve on their ability to communicate and respond empathically to their patients.26 Understanding the balance between action and presence may help physicians negotiate these different aspects of their role. While current demands may work to emphasise action and efficiency, the consequences of a one-sided emphasis on action mode may affect physicians’ ability to respond with empathy and compassion towards patients. ‘Iatrogenic suffering’8 may occur when physicians maintain the ‘disconnect’ of action mode at times when their patients require the openness and connection of presence. Limitations This study generated descriptions of the common essential qualities of the experience of patient death for 10 physicians from widely different practice areas. The breadth of professional expertise was a strength as well as a limitation. The range of specialties helped to deepen the resulting description of the physicians’ experience of patient death, and yet it also resulted in a less detailed understanding of the experience of physicians in any one specialty.

ractice areas. The breadth of professional expertise was a strength as well as a limitation. The range of specialties helped to deepen the resulting description of the physicians’ experience of patient death, and yet it also resulted in a less detailed understanding of the experience of physicians in any one specialty. The study was also limited by the small number of physician participants, who represent a convenience sample, self-selected after hearing of the current study through word of mouth or an introductory letter. Participants consistently expressed an interest in participating in order to better understand physicians’ work-related stress or to improve doctor-patient communication, which does suggest a certain bias among this sample. Future directions Research into nurses’ experiences with patient death has identified a number of themes similar to those found by this study.27 Future research could explore whether the particular finding of action versus presence is also applicable to nurses, allied health professionals and others dealing with patient death. When considering interprofessional healthcare teams, it would be interesting for future research to investigate whether the action versus presence dynamics have an impact on emotional processing and communication among team members, particularly when dealing with end-of-life care and patient death.

with patient death. When considering interprofessional healthcare teams, it would be interesting for future research to investigate whether the action versus presence dynamics have an impact on emotional processing and communication among team members, particularly when dealing with end-of-life care and patient death. The findings of this study also have many implications for doctor-patient communication and the quality of patient care, as well as for the professional resilience of physicians. The training of the physicians who participated in this study occurred some time ago, and many training programmes have changed. Future research needs to identify and compare recent efforts to provide training and support for emotional awareness, communication and self care in medical schools and residency programmes. Future research could also determine whether specific training and emotional support using the concepts of action and presence can help young physicians understand and value the process of reconnection, and whether this aids them in learning to manage the emotional stress inherent in their work. Recent literature on physician wellness has suggested that professional resilience is enhanced when physicians engage in open communication with their dying patients, the patients’ families and the other colleagues involved in the patients’ care.28–30 This research suggests that the open communication and compassion of presence could help create ‘healing connections’29 which would allow physicians to gain strength and resilience through their interactions with dying patients. Greater understanding of the action versus presence dynamics will help physicians learn to use the functional disconnect of action mode when needed and maintain the healing connection of presence with their patients and themselves.

ysicians to gain strength and resilience through their interactions with dying patients. Greater understanding of the action versus presence dynamics will help physicians learn to use the functional disconnect of action mode when needed and maintain the healing connection of presence with their patients and themselves. Contributors: PRW is the primary contributor, who designed and conducted the research as part of his doctoral dissertation, under the guidance and supervision of DK. PRW is the sole author of the current article and guarantor of its contents. DK contributed to the article as supervisor of the research, guiding the planning of the research and acting as consultant in the interpretation and writing of the article. Funding: None. Competing interests: None. Ethical approval: In keeping with the revised (1983) Helsinki Declaration, this study was approved by the ethics board of the University of British Columbia. Provenance and peer review: Not commissioned; externally peer reviewed.

Introduction Palliative care focuses on supporting patients diagnosed with advanced, incurable disease; it is ‘family centred’, with the patient and their family (the unit of care) being core to all its functions.1 Given the significant burden associated with caring for a dying relative, WHO advocates that healthcare services focus on enhancing family members’ quality of life during caregiving and bereavement.2 Accordingly, many nations have established standards and policies for palliative care provision which focus on assessing and responding to the psychosocial needs of family caregivers.1 Although some family caregivers identify positive aspects associated with their role,3 the impact of caring can be considerable. The role is typically burdensome; family caregivers are often prone to physical and psychological morbidity, financial disadvantage and social isolation.4 Some caregivers take on the role because of a sense of obligation, perceiving they have little choice.5 Caregiving can extend for several years; it can be equivalent to a full-time job.6

burdensome; family caregivers are often prone to physical and psychological morbidity, financial disadvantage and social isolation.4 Some caregivers take on the role because of a sense of obligation, perceiving they have little choice.5 Caregiving can extend for several years; it can be equivalent to a full-time job.6 A life-threatening diagnosis can have a greater negative psychological impact on the family members than the patients.7 Depression rates of between 12% and 59%,4 and anxiety rates of between 30% and 50%4 8 in caregivers have been reported. A recent study involving caregivers (n=300) soon after referral to specialist palliative care, identified that approximately 50% met criteria for psychological distress (based on a validated screening instrument).9 Unfortunately, these high rates of psychological distress are typically under-recognised and, consequently, not addressed.10

lving caregivers (n=300) soon after referral to specialist palliative care, identified that approximately 50% met criteria for psychological distress (based on a validated screening instrument).9 Unfortunately, these high rates of psychological distress are typically under-recognised and, consequently, not addressed.10 Irrespective of a requirement for healthcare professionals to support family caregivers, the reality is that in many instances this support is less than optimal.1 Family caregivers’ psychosocial needs are frequently considered either secondary to those of the patient, or overlooked; there is neither consistency nor a systematic approach to assessing needs.11 Furthermore, caregivers are often not aware of available support, they usually have limited prior exposure to death and dying, and feel excluded from information and care planning.12 Health professionals acknowledge that providing psychosocial support to family members, as well as patients, presents an enormous challenge, and they may not be resourced or have the required skills to provide such support.13

ted prior exposure to death and dying, and feel excluded from information and care planning.12 Health professionals acknowledge that providing psychosocial support to family members, as well as patients, presents an enormous challenge, and they may not be resourced or have the required skills to provide such support.13 Systematic reviews of interventions for family caregivers reveal that effective support is in its infancy.14 15 Results from an international survey16 concluded that interventions focused on improving family caregiver psychosocial support were a priority. According to a recent review,1 interventions to improve family caregiver support are important for numerous reasons which are outlined in the Box 1. Box 1 Reasons why interventions to improve family caregiver support are required* Interventions to improve family caregiver support are required because caregivers: should receive evidence-based support from health professionals as per national and international policies and standards are receiving support which is neither systematic nor fully evidence based are prone to physical and psychological morbidity are responsible for numerous tasks, such as symptom management are financially disadvantaged become socially isolated report unmet needs (typically aligned with lack of information about their role) have needs equal to and/or greater than the patients’ needs have very limited first-hand exposure to death and dying are often excluded from information and care planning and, consequently, feel underprepared for their role

become socially isolated report unmet needs (typically aligned with lack of information about their role) have needs equal to and/or greater than the patients’ needs have very limited first-hand exposure to death and dying are often excluded from information and care planning and, consequently, feel underprepared for their role have the potential (with suitable support) to gain positive outcomes from their experience. are pivotal to achieving ‘successful’ home care (where most people prefer to die) make a substantial economic contribution to healthcare may significantly enhance the patients’ well-being when their role is well supported. *Adapted from Hudson and Payne.1 The following reflection exemplifies the experience for many caregivers who feel underprepared for their role:I just had no idea what I was in for. If I had known maybe I wouldn't have been so forthcoming about making the promise to look after him at home. It all sounds so nice when we first talked about; ‘Oh yes, it would be nice to have him at home, that's so much better, so peaceful’. And when its really happening you think, oh my God, this is horrendous. It's not at all what I expected.17

't have been so forthcoming about making the promise to look after him at home. It all sounds so nice when we first talked about; ‘Oh yes, it would be nice to have him at home, that's so much better, so peaceful’. And when its really happening you think, oh my God, this is horrendous. It's not at all what I expected.17 In summary, there are clinical, academic and policy reasons for developing a programme of intervention research focused on improving the psychosocial well-being of family caregivers; without such research, the field will continue to fall short of meeting standards of palliative care provision.1 This paper outlines the rationale, pilot work, conceptual framework for such a programme and provides an overview of several interventions and resources primarily developed and/or led by the Centre for Palliative Care (St Vincent's Hospital and Collaborative Centre of The University of Melbourne, Australia). Accordingly, this suite of strategies and resources has been labelled: the Melbourne Family Support Program (FSP). Methodological and theoretical framework and pilot work Methodological framework In order to develop the interventions tested as part of the Melbourne FSP, a theoretical framework was sourced and pilot work undertaken. This strategy is commensurate with the Medical Research Council, UK's approach for the development and evaluation of complex interventions to improve health.18

thodological framework In order to develop the interventions tested as part of the Melbourne FSP, a theoretical framework was sourced and pilot work undertaken. This strategy is commensurate with the Medical Research Council, UK's approach for the development and evaluation of complex interventions to improve health.18 Theoretical framework The theoretical framework for guiding the Melbourne FSP was based on a transactional model of stress and coping.19 Proponents of this framework argue that the diversity of responses related to end-of-life issues from patients and family caregivers can be understood from a psychological perspective based on a transactional model of coping in which caregivers make cognitive appraisals to determine the possible impact of a potentially stressful event. The more prepared and capable the caregiver is, or the greater the number of resources at their disposal to manage an event, the more likely the individual will display adaptive behaviours. Such resources include having adequate information to prepare them for their role, fewer unmet needs and focusing on positive aspects in order to minimise psychological distress. Hence, strategies that targeted these domains were core to the Melbourne FSP.19

more likely the individual will display adaptive behaviours. Such resources include having adequate information to prepare them for their role, fewer unmet needs and focusing on positive aspects in order to minimise psychological distress. Hence, strategies that targeted these domains were core to the Melbourne FSP.19 Pilot work Four approaches, summarised below, were undertaken to help inform the development of the Melbourne FSP. The first study sought insights (through focus groups) from palliative care nurses and family caregivers to inform intervention development.20 Results identified what caregivers need to prepare them for their role, the key information that should be routinely offered and the best way to present the information. Based on these data, a guidebook on preparing family caregivers for the role of supporting a relative/friend with a palliative care diagnosis was also produced.20 A critical literature review of supportive interventions for family caregivers21 confirmed the need for new supportive care strategies for families to be developed. Psychoeducational interventions evaluated via randomised controlled trials (RCT) were considered a priority.

Pilot work Four approaches, summarised below, were undertaken to help inform the development of the Melbourne FSP. The first study sought insights (through focus groups) from palliative care nurses and family caregivers to inform intervention development.20 Results identified what caregivers need to prepare them for their role, the key information that should be routinely offered and the best way to present the information. Based on these data, a guidebook on preparing family caregivers for the role of supporting a relative/friend with a palliative care diagnosis was also produced.20 A critical literature review of supportive interventions for family caregivers21 confirmed the need for new supportive care strategies for families to be developed. Psychoeducational interventions evaluated via randomised controlled trials (RCT) were considered a priority. Successful interventions need to be developed in a pragmatic way; accordingly, the challenges faced by healthcare professionals associated with providing support to family caregivers were considered.13 The supportive care challenges fell into three broad categories: family related challenges, health system barriers and communication barriers. Identifying these obstacles influenced the development of a FSP focused on these ‘coal face’ issues.

ls associated with providing support to family caregivers were considered.13 The supportive care challenges fell into three broad categories: family related challenges, health system barriers and communication barriers. Identifying these obstacles influenced the development of a FSP focused on these ‘coal face’ issues. Based on the aforementioned work, a pilot RCT; a home-based one-to-one (nurse to caregiver) psychoeducational intervention focused on preparing family caregivers (n=106) for the role of supporting a dying relative at home was undertaken.22 Data were collected at three time points: upon commencement of home-based palliative care (Time 1), 5 weeks later (Time 2), and then 8 weeks following patient death (Time 3). No intervention effects were identified with respect to preparedness to care, self-efficacy, competence and anxiety. However, participants who received the intervention reported a significantly more positive caregiver experience than those who received standard care at both Times 2 and 3. A larger follow-up trial was recommended.

tion effects were identified with respect to preparedness to care, self-efficacy, competence and anxiety. However, participants who received the intervention reported a significantly more positive caregiver experience than those who received standard care at both Times 2 and 3. A larger follow-up trial was recommended. Melbourne FSP: aim and objectives The Melbourne FSP's focus was on insights from the theoretical framework and pilot work. The Melbourne FSP's purpose was to develop evidence-based strategies (and associated resources) that focus on promoting the psychosocial well-being of primary family caregivers of patients receiving palliative care. Specific elements of the programme included developing healthcare-led psycho-educational mechanisms to: (1) increase family caregivers’ sense of preparedness for their role; (2) bolster their positive emotions; (3) reduce their unmet needs; (4) reduce their psychological distress. Particular emphasis on preparedness was considered important. Family caregivers consistently report that communication and information related to end-of-life care is inadequate; they want to be better prepared for their role.12 Inadequate family caregiver preparedness for death has been shown to be associated with depression, anxiety and complicated grief.23 24 Furthermore, the concept of preparedness fits neatly with the chosen theoretical framework.

related to end-of-life care is inadequate; they want to be better prepared for their role.12 Inadequate family caregiver preparedness for death has been shown to be associated with depression, anxiety and complicated grief.23 24 Furthermore, the concept of preparedness fits neatly with the chosen theoretical framework. Overview of Melbourne FSP strategies and resources Over a decade, five main strategies (and associated resources) have been developed and evaluated; these are outlined in table 1. A comprehensive account of the objectives, evaluation methods and outcomes of each of the strategies can be found in the publications arising from each specific study. As outlined, four of the interventions (1–4) involved healthcare professional-led, direct interaction with family caregivers in both one-to-one and group formats, conducted in either the home or inpatient/hospice setting. The interventions were primarily tested via prepost design with one examined via a RCT. All interventions showed statistically beneficial outcomes in at least one of four outcome variables: preparedness, unmet needs, positive emotions and psychological distress. The final strategy involved the development of guidelines for the evidence-based psychosocial and bereavement support of family caregivers, endorsed by several key national and international organisations. Table 1 Summary of Melbourne Family Support Program (FSP) interventions and resources

Overview of Melbourne FSP strategies and resources Over a decade, five main strategies (and associated resources) have been developed and evaluated; these are outlined in table 1. A comprehensive account of the objectives, evaluation methods and outcomes of each of the strategies can be found in the publications arising from each specific study. As outlined, four of the interventions (1–4) involved healthcare professional-led, direct interaction with family caregivers in both one-to-one and group formats, conducted in either the home or inpatient/hospice setting. The interventions were primarily tested via prepost design with one examined via a RCT. All interventions showed statistically beneficial outcomes in at least one of four outcome variables: preparedness, unmet needs, positive emotions and psychological distress. The final strategy involved the development of guidelines for the evidence-based psychosocial and bereavement support of family caregivers, endorsed by several key national and international organisations. Table 1 Summary of Melbourne Family Support Program (FSP) interventions and resources Strategy/intervention Setting Research design/methods Outcome variables # Of participants Results Resources Publications arising Melbourne FSP #1.One-to-one home-based psychoeducation programme. Delivered by nurse to primary family caregiver Family caregivers of patients enrolled in home-based/community specialist palliative care services - 4 sites, 3 states of Australia Randomised controlled trial. T1 (baseline  commencement of palliative care), T2 (2 weeks >intervention), T3 (2 mths >patient death) Preparedness, competence, rewards, unmet needs, psychological distress 300 Statistically significant improvement in preparedness and competence @ T2. Significantly less rise (worsening) in psychological distress @ T3 in the intervention group. Caregiver guidebook20  (T2)9 11 (T3)25 Melbourne FSP #2. Group home-based psychoeducation programme. Delivered by health professional to group of primary family caregivers. Face-to-face sessions x 3 @ local hospice Family caregivers of patients enrolled in home based/community specialist palliative care services - 6 sites, Victoria Australia Pretest post-test. T1 (baseline  commencement of palliative care), T2 (2 weeks > intervention Preparedness, competence, rewards, and unmet needs 156 (30 programmes) Statistically significant improvement in preparedness, competence, rewards, and needs met Caregiver guidebook20 DVD for health professionals (http://www.centreforpallcare.org) 26 27 Melbourne FSP # 3. Group hospital. Delivered by health professional to group of primary family caregivers. Face-to-face sessions x 1 @ hospital palliative care unit/hospice Family caregivers of patients admitted to hospital palliative care units/hospices - 3 states, 4 sites in Australia Pretest post-test. T1 (prior to commencement), T2 (3 days >) intervention Preparedness, competence, psychological unmet needs and psychological distress 125 (52 programmes) Statistically significant increases in preparedness, competence and needs met. Web guidebook (http://www.centreforpallcare.org) 28 29 Melbourne FSP #4.

est post-test. T1 (prior to commencement), T2 (3 days >) intervention Preparedness, competence, psychological unmet needs and psychological distress 125 (52 programmes) Statistically significant increases in preparedness, competence and needs met. Web guidebook (http://www.centreforpallcare.org) 28 29 Melbourne FSP #4. Evaluation of guidelines for conducting family meetings Family caregivers of patients admitted to hospital palliative care units/hospices in one site, Melbourne Australia T 1 (pre-FM), T 2 (post-FM); T3 (2 days later) Plus qualitative Unmet needs 22 family meetings Statistically significant increases in needs met 30 30–32 Melbourne FSP #5. Guidelines for psychosocial and bereavement support of family caregivers Designed for healthcare professionals involved in palliative care across settings Literature review. Focus groups. Semistructured interviews. Delphi survey. Endorsement by key palliative care organisations Not applicable Not applicable Officially endorsed by several institutions including: St Christopher's Hospice (UK); Palliative Care Australia; International Observatory of End of Life Care UK Complete version of guidelines (http://www.centreforpallcare.org) 33 DVD, digital video disk.

key palliative care organisations Not applicable Not applicable Officially endorsed by several institutions including: St Christopher's Hospice (UK); Palliative Care Australia; International Observatory of End of Life Care UK Complete version of guidelines (http://www.centreforpallcare.org) 33 DVD, digital video disk. Discussion The need for more research to underpin support for family caregivers related to palliative care has been acknowledged by an international assembly of senior researchers,34 and was the rationale for the inception of the International Palliative Care Family Caregiver Research Collaboration.35 The results of several published reviews1 14–16 36 37 also advocate for major investment in intervention research in this area. The Melbourne FSP has produced several interventions showing statistically significant outcomes for family caregivers, including increased sense of preparedness, competence, rewards, more favourable levels of psychological well-being and a reduction in unmet needs. Several resources have also been developed, including clinical practice guidelines endorsed by key international institutions (see table 1). Fifteen publications in international peer-reviewed journals have arisen from the Melbourne FSP.

more favourable levels of psychological well-being and a reduction in unmet needs. Several resources have also been developed, including clinical practice guidelines endorsed by key international institutions (see table 1). Fifteen publications in international peer-reviewed journals have arisen from the Melbourne FSP. Research limitations and implications Several research implications and limitations arise from this programme. A ‘pilot’ RCT formed the foundation of the research22 with an additional follow-up phase III RCT.9 The remaining interventions, however, were evaluated via pretest/post-test designs which have noteworthy limitations, including potential selection bias, and ‘regression to the mean’ issues.38 Hence, these interventions should undergo evaluation via more rigorous experimental methods. Although two of the studies involved a bereavement data-collection point, the remainder did not; future work should consider incorporating these longer-term endpoints. Other methodological challenges and strategies associated with family caregiver intervention research have been described elsewhere.1 37 38

tal methods. Although two of the studies involved a bereavement data-collection point, the remainder did not; future work should consider incorporating these longer-term endpoints. Other methodological challenges and strategies associated with family caregiver intervention research have been described elsewhere.1 37 38 Future work should also attempt to clarify primary endpoints for psychoeducational interventions of the kind undertaken via the Melbourne FSP. In theory, the primary outcome should link to the main objective of the intervention. However, although the Melbourne FSP interventions had similar aims and ingredients, it was difficult to determine, in some circumstances, which primary outcome should be measured. For example, we debated whether or not reducing caregiver psychological distress was more preferable, as a primary outcome, than reducing their unmet needs or increasing their sense of preparedness. Based on the theoretical framework underpinning the Melbourne FSP, it seemed logical to propose that if preparedness improves and family caregivers have their needs met, then it followed that their psychological well-being may be more optimal. Thus, psychological status would appear to be a justifiable primary endpoint. Intervention #1 identified that by 2 months after the patients’ deaths, the caregivers in the intervention groups had significantly less increase in psychological distress than the control group.25 Future empirical inquiry should examine this further, including clarifying the relationship between the variables outlined within the theoretical framework. It would be useful, for example, to know if preparedness and competence were highly correlated. Consideration should also be given to emerging methodological ideas surrounding examination of multiple endpoints.39

her, including clarifying the relationship between the variables outlined within the theoretical framework. It would be useful, for example, to know if preparedness and competence were highly correlated. Consideration should also be given to emerging methodological ideas surrounding examination of multiple endpoints.39 The results arising from the Melbourne FSP seem to justify the ongoing use of the transactional model of stress and coping as a means to guide interpretation of the caregiver experience and interventions to ameliorate negative outcomes.19 For example, positive aspects of the caregiver role were bolstered by one intervention, and preparedness improved in several interventions. Schumaker and colleagues40 found that negative reactions to the family caregiver experience can be buffered when caregivers are better prepared for their role. Despite these favourable findings, more empirical work is needed to determine the relationship between variables.23

paredness improved in several interventions. Schumaker and colleagues40 found that negative reactions to the family caregiver experience can be buffered when caregivers are better prepared for their role. Despite these favourable findings, more empirical work is needed to determine the relationship between variables.23 Clinical implications It is difficult at present to assess the clinical significance of the Melbourne FSP interventions and resources. Despite dissemination via peer review, international journal publications (see table 1) and conference presentations arising from each of the Melbourne FSP initiatives, and formal endorsement for the clinical guidelines from major institutions, the extent of implementation is unclear. The utility of each intervention was partially examined as the evaluation incorporated some level of exploration of applicability, affordability and accessibility (outlined in the publications arising from each individual study). Some of the pragmatic issues discussed in detail in the publications arising from each intervention include factors, such as attracting enough family caregivers to warrant resourcing a group education session; given that some family caregivers are reluctant to leave their home to attend such programmes. Issues related to optimal dose and frequency of interventions are also outlined.

ons arising from each intervention include factors, such as attracting enough family caregivers to warrant resourcing a group education session; given that some family caregivers are reluctant to leave their home to attend such programmes. Issues related to optimal dose and frequency of interventions are also outlined. The Melbourne FSP has incorporated a variety of face-to-face delivery methods, including one-to-one (health professional: family caregiver) group programmes, and conducted in the home-based and hospital palliative care setting, complemented by resources for family caregivers and health professionals. The rationale for this approach was that there is no ‘one size fits all’ for family caregiving in the context of end-of-life care. Family caregivers will most likely desire some information/guidance on how to provide support to their friend/relative. For some, a resource such as a guidebook may suffice. Others may benefit from structured face-to-face psychoeducational interaction; either one-to-one or via a group approach supported by written resources. Future work should focus on web-based initiatives. It is also important to expand these initiatives to other populations, for example, family caregivers of patients with end-stage neurological or organ failure, and to caregivers of children/young adults.1

e-to-one or via a group approach supported by written resources. Future work should focus on web-based initiatives. It is also important to expand these initiatives to other populations, for example, family caregivers of patients with end-stage neurological or organ failure, and to caregivers of children/young adults.1 The Melbourne FSP has deliberately focused on the primary family caregiver; a decision based on our pilot work which showed that the overwhelming majority of family caregivers wanted to meet a healthcare professional separately from their relative.22 Additionally, we have argued elsewhere that unless there is an increase in resources for palliative care providers, the focus of support should be the primary family caregiver in the first instance.33 Nonetheless, consideration should be given to developing and evaluating approaches that involve the primary family caregiver and their relative/friend (for those who prefer this option), and perhaps other family members/friends (where suitable resources prevail). The Melbourne FSP was based in Australia and incorporated participants from metropolitan, regional and rural areas and a variety of cultural backgrounds. It is pleasing that some of this work has been tested in other countries32; however, further exploration of its suitability in other countries/settings is warranted.

The Melbourne FSP was based in Australia and incorporated participants from metropolitan, regional and rural areas and a variety of cultural backgrounds. It is pleasing that some of this work has been tested in other countries32; however, further exploration of its suitability in other countries/settings is warranted. Future work should also focus on methods to help clinicians identify which specific interventions/strategies are most pertinent for family caregivers. Consideration should be given to systematic psychosocial screening and triaging of family caregivers4 allowing for a more targeted approach for interventions. The Melbourne FSP has focused primarily on psychoeducational initiatives. While we have justified this approach, these initiatives on their own will not provide the support that is required; some family caregivers will need comprehensive, focused interventions based on thorough assessment. Conclusion Although family caregivers are relied upon to provide much of the care to palliative care patients, they often receive inadequate preparation, information and support; thus, empirically based strategies to rectify this are required. The interventions and resources arising from the Melbourne FSP provide several evidence-based approaches that attempt to lessen the psychosocial burden of the caregiving role. Although this programme arises from comprehensive pilot work, and a well-regarded conceptual framework and beneficial outcomes have been demonstrated, in several instances more rigorous methodological testing is advocated.