Cluster Randomized Controlled Trial Clinical and Cost-Effectiveness of a System of Longer-Term Stroke Care Anne Forster, PhD; John Young, MSc; Katie Chapman, PhD; Jane Nixon, PhD; Anita Patel, PhD; Ivana Holloway, MSc; Kirste Mellish, PhD; Shamaila Anwar, PhD; Rachel Breen, PhD; Martin Knapp, PhD; Jenni Murray, PhD; Amanda Farrin, MSc Background and Purpose—We developed a new postdischarge system of care comprising a structured assessment covering longer-term problems experienced by patients with stroke and their carers, linked to evidence-based treatment algorithms and reference guides (the longer-term stroke care system of care) to address the poor longer-term recovery experienced by many patients with stroke. Methods—A pragmatic, multicentre, cluster randomized controlled trial of this system of care. Eligible patients referred to community-based Stroke Care Coordinators were randomized to receive the new system of care or usual practice. The primary outcome was improved patient psychological well-being (General Health Questionnaire-12) at 6 months; secondary outcomes included functional outcomes for patients, carer outcomes, and cost-effectiveness. Follow-up was through self-completed postal questionnaires at 6 and 12 months. Results—Thirty-two stroke services were randomized (29 participated); 800 patients (399 control; 401 intervention) and 208 carers (100 control; 108 intervention) were recruited. In intention to treat analysis, the adjusted difference in patient General Health Questionnaire-12 mean scores at 6 months was −0.6 points (95% confidence interval, −1.8 to 0.7; P=0.394) indicating no evidence of statistically significant difference between the groups. Costs of Stroke Care Coordinator inputs, total health and social care costs, and quality-adjusted life year gains at 6 months, 12 months, and over the year were similar between the groups. Conclusions—This robust trial demonstrated no benefit in clinical or cost-effectiveness outcomes associated with the new system of care compared with usual Stroke Care Coordinator practice. Clinical Trial Registration—URL: http://www.controlled-trials.com. Unique identifier: ISRCTN 67932305.    (Stroke. 2015;46:2212-2219. DOI: 10.1161/STROKEAHA.115.008585.) Key Words: cluster randomized controlled trial ◼ community health services ◼ cost-benefit analysis ◼ quality-adjusted life years ◼ rehabilitation ◼ stroke

T

he high prevalence and diversity of longer-term problems experienced by patients with stroke and their carers has long been recognized,1 but they remain poorly addressed by existing services.2 Postdischarge contact with therapy services is associated with improved outcomes.3,4 However, these interventions are generally time limited and have little effect on psychosocial outcome. National guidelines5 acknowledge that stroke should be regarded as a long-term condition, and the role of a Stroke Care Coordinator (SCC) to facilitate inputs for community-based patients with stroke and their families after initial (usually hospital-based) treatment is a

recommended policy.6,7 However, procedures and processes of this role are unevenly developed, and there has been no robust evaluation. Using the Medical Research Council framework for the development and evaluation of complex interventions,8 incorporating systematic reviews, qualitative exploration, and intervention modeling, we developed an evidence-based system of care (longer-term stroke [LoTS] care) that aimed to meet the longer-term needs of patients with stroke and their carers living at home. The system of care incorporates a structured assessment focused on patient- and carer-centered problems

Received January 21, 2015; final revision received March 19, 2015; accepted April 6, 2015. From the Academic Unit of Elderly Care and Rehabilitation, Bradford Teaching Hospitals NHS Foundation Trust (A. Forster, J.Y., K.C., K.M., R.B.), Leeds Institute of Clinical Trials Research, Clinical Trials Research Unit (J.N., I.H., S.A., A. Farrin), and Leeds Institute of Health Sciences (J.M.), University of Leeds, Leeds, United Kingdom; Institute of Psychiatry, King’s College London, London, United Kingdom (A.P.); and Personal Social Services Research Unit, London School of Economics and Political Science, London, United Kingdom (M.K.). The report of the full programme of research (including this trial) is published in the NIHR Journals Library, Volume 2, Issue 6. For further information see http://www.journalslibrary.nihr.ac.uk/pgfar/volume-2/issue-6#abstract. December 2014. Presented in part at the European Stroke Conference, London, UK, May 28–31, 2013. The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA. 115.008585/-/DC1. Correspondence to Anne Forster, PhD, Academic Unit of Elderly Care and Rehabilitation, Temple Bank House, Bradford Royal Infirmary, Bradford, BD9 6RJ, United Kingdom. E-mail [email protected] © 2015 Bradford Teaching Hospitals NHS Foundation Trust. Stroke is published on behalf of the American Heart Association, Inc., by Wolters Kluwer. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution, and reproduction in any medium, provided that the original work is properly cited. Stroke is available at http://stroke.ahajournals.org

DOI: 10.1161/STROKEAHA.115.008585

Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on November 15, 2015 2212

Forster et al   LoTS Care Stroke System of Care Trial    2213 and is linked to evidenced-based treatment algorithms and a goal and action planner based on a problem-solving approach. We report findings from a trial evaluation of the LoTS care system of care.

Methods Study Design and Participants The trial was a pragmatic cluster randomized controlled trial of the clinical and cost-effectiveness of the LoTS care system of care delivered by community-based SCCs compared with SCC usual practice. Trial procedures were informed by a survey of SCCs.9 The methods have been reported in detail elsewhere10 and undertaken after appropriate ethical approval. A SCC was eligible if the following criteria were fulfilled: a registered healthcare professional with documented experience in stroke care; undertaking a community-based coordinating role for patients with stroke; in contact with patients and coordinating a range of care inputs on behalf of the patient or carer (eg, signposting, carrying out assessments); receiving referrals from an acute stroke service that included a stroke unit fulfilling the Royal College of Physicians audit definition.11 A SCC was classified as working in a team if they participated in community multidisciplinary team meetings. Eligibility of the SCC service was confirmed before randomization by completion of a questionnaire and semistructured interview describing the service and client group (these were repeated midway through recruitment and after 12-month follow-up). This information was also used for stratification to provide context for trial implementation and monitor any potential contamination or confounding between the 2 arms of the trial. In keeping with the pragmatic trial design, patient eligibility criteria were broad. Patients were eligible if they had a confirmed primary diagnosis of a new stroke, were referred to an SCC on discharge home or within 6 weeks of stroke (if not admitted to hospital), and were still waiting for their first community SCC assessment. Patients were excluded if they had a planned permanent admission to, or were already resident in, a nursing or residential care home; or their main requirement was palliative care. Carers were eligible if a participating patient identified them as the main informal carer who provided practical support at least once per week. Written informed consent for baseline and follow-up assessments was obtained from patients (and carers if appropriate) before baseline assessment. In the event that a patient lacked capacity to consent, the patient’s family member, carer, or friend was asked to act as the consultee and provide consultee declaration.

Randomization and Masking The unit of randomization was the SCC service randomized on a 1:1 basis to either the control or the intervention group. Randomization

was stratified by the quality of the local stroke unit (National Sentinel Stroke Audit median score of 65 based on 2006 data)11; annual number of referrals to the SCC service; SCCs working alone or within a community team; and by geographical area. A method of obtaining a balanced randomization from these covariates based on a method by Carter and Hood12 was used.

Control Patients in services allocated to the control group received the SCC service in accord with the existing local policies and practices. After randomization, the control group SCCs were asked to complete time logs for all patients documenting the number and duration of contacts, and the time spent coordinating actions, note writing, and discussing the patient in multidisciplinary team meetings if these took place.

Intervention The SCC services allocated to the intervention group provided care according to the LoTS care system of care. This comprises a framework of 16 structured assessment questions (linked to evidence-based treatment algorithms and reference guides) that directly relate to longer-term stroke problems previously identified by patients with stroke and their carers13,14 and related prompts provided in a care plan. The care plan also includes a goal and action planner to be completed after each contact (patients and carers). This system of care was supported by a training program and detailed manual, underpinned by a problem-solving approach (Table 1; details are provided in the online-only Data Supplement). The number of contacts was not specified but determined by the SCCs’ usual practice and patient need. The key principles being that all assessment questions were asked and that goals and action plans were identified and reviewed with service responses being provided as appropriate within the context of the services available and the patient’s own environment.15 Recruitment was opened 4 to 6 months after the initial training meeting, providing sufficient time for the implementation of the system of care into standard practice. Compliance with the intervention was predefined by the trial team as including at least 12 (75%) of the 16 assessment areas recorded as discussed on the care plan during the first patient contact.

Procedures The trial used existing SCC referral pathways as determined during site set-up. The majority of patients were referred to an SCC service through a predischarge inpatient referral. Recruitment of trial participants was by independent research staff blinded as to whether they were recruiting within a control or an intervention service, and the SCCs were unaware which of their patients had consented to participate. This trial design reduced the potential for selection bias from

Table 1.  Components of the Longer-Term Stroke Care System of Care Care plan containing a structured assessment with a goal and action planner for each contact   Patient: 16 assessment questions (15 specific+other)   Carer: 13 assessment questions (12 specific+other) An optional checklist indicating the content of the assessment to be sent to patients beforehand A manual containing reference guides including robust evidence-based treatment algorithms linked to the assessment questions (15 patient-specific and 1 carerspecific), a frequency table of longer-term problems after stroke, service directory, and recommended assessment scales National information about services available for patients after stroke (eg, Disabled Holiday Directory, Age Concern); the SCCs were asked to develop a resource inventory of local services Training in use of the system of care   Training day 1: details of the system of care, guidance on problem solving techniques, principles of the intervention   Period of implementation of the system of care (≈1 mo)   Training day 2: review of use of the system of care and problem solving, training in specific areas (eg, pain)   Provision of a CD of the training presentations SCC indicates Stroke Care Coordinator.

Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on November 15, 2015

2214  Stroke  August 2015 differential recruitment16 inherent if the SCCs were responsible for patient identification and minimized likelihood of altering of SCCs’ clinical activity for trial participants. The research staff collected anonymous screening data (demographic data and modified Rankin Scale score17) for all patients referred to a participating SCC. For recruited patients, researchers collected baseline demographic data, including assessment of cognition (6-Item Cognitive Impairment Test),18 language ability, prestroke disability (Barthel index19), and the 6 items that allow calculation of the Edinburgh stroke case mix adjuster.20 Consistent with a patient-centered model of stroke recovery in which adjustment to activity restrictions and participation are regarded as the critical issues, the primary outcome was an assessment of psychological well-being measured by patient reported General Health Questionnaire-12 (GHQ-12)21 at 6 months after recruitment, with a secondary outcome at 12 months (for the GHQ-12 higher scores equate to poorer outcomes). This outcome measure is short and easy to complete, consistent with the high prevalence of psychological symptoms after stroke,13 and that psychological problems become more prevalent with time.22 Other patient reported secondary outcome measures were Frenchay Activities Index (extended activities of daily living),23 Barthel Index (activities of daily living),19 and European Quality of Life-5 Dimensions (health state).24,25 Patients also completed the Longer-term Unmet Needs after Stroke questionnaire (unmet stroke related needs).26 Carer-reported outcomes included GHQ-12 and Carer Burden Scale (assessment of carer burden).27 A Client Service Receipt Inventory (CSRI28,29; use of services prestroke and postdischarge) was completed to inform the economic evaluation. The research staff recorded deaths, emergency outpatient treatment, and hospital admissions at 6 and 12 months post recruitment. Patient- and carer-reported outcomes were assessed in baseline questionnaires and via postal questionnaires at 6 and 12 months post recruitment. This was supported by postal and telephone reminders if questionnaires were not returned within 2 weeks. If necessary (after postal and telephone reminders), patients were contacted by telephone to complete the primary outcome measure (GHQ-12). Process data were collected for trial patients after the end of 12 month follow-up so as not to unblind the SCC to trial patients. Time logs were collected for patients in control services. In intervention services, data on the use of the structured assessment and goal and action planner (including time taken) were collected by researchers transcribing the appropriate information from the care plans at site.

Sample Size The sample size calculations, based on the primary outcome measure GHQ-12 at 6 months, indicated that recruitment of 800 patients from 40 services would provide 90% power at 5% significance level to detect a clinically relevant difference of 2.5 GHQ-12 points (SD, 7), as reported in a previous study.30 This sample size accounted for an estimated 25% loss to follow-up and clustering: the inflation factor of 1.95 was derived from a maximum cluster size of 20 and an intracluster correlation coefficient of no >0.05.31 Losses to follow-up were anticipated to increase over time, and interpretation and credibility of results are difficult if losses exceed 30%; therefore, follow-up was limited to 12 months post recruitment and the primary outcome was defined as 6 months. We were able to identify 32 SCC services which were eligible, willing to participate and provide a principal investigator, and these were randomized. We planned that each service would recruit 25 patients for 18 months, to provide power of 88%, assuming equal cluster size and no >25% loss to follow-up. To minimize unequal recruitment to clusters, the maximum number of patients per service was capped at 45.

Statistical Methods Statistical analyses were conducted on the intention to treat population for both primary and secondary analyses. The intention to treat population was defined as all patients registered for active follow-up regardless of noncompliance with the intervention. A per-protocol

analysis was also undertaken where major protocol violators and patients not receiving care from SCC were excluded from the analysis. All statistical testing was performed at a 2-sided 5% significance level. Analyses were completed using SAS software version 9.2 (SAS Institute Inc, Cary, NC). Primary and secondary outcomes were compared between the control and intervention groups using a 2-level multilevel model, with patients nested within stroke services. The models were adjusted for the patient-level covariates (level 1): baseline Barthel Index (prestroke and poststroke), sex, age, living circumstances (living alone versus with carer), stroke severity reflected by speech and language impairment (normal/impaired), baseline 6-Item Cognitive Impairment Test score (normal/impaired cognitive function), and patient baseline score for the outcome measure; and the following stroke unit-level covariates (level 2): quality of stroke unit (National Sentinel Stroke Audit score), referral rate, and SCCs working alone versus within a community multidisciplinary team. Sensitivity analyses were undertaken to test the robustness of analysis assumptions, including patients who died by assuming worst possible GHQ-12 outcome; only including patients returning postal questionnaires at 6 months (excluding patients who provided primary outcome via telephone call); repeating the analysis without proxy responses; using data collected at 12 months for patients who did not return questionnaires at 6 months, and assuming data missing at random using multiple imputation. Details of patient deaths and hospital readmissions, unmet needs, carer deaths, any serious adverse events, and related and unexpected serious adverse events were reported for each treatment group. The relationships between adjusted primary outcome and completion of time logs in control or compliance with care plans in intervention were explored graphically.

Economic Evaluation The prospective economic evaluation was from both health/social care and societal perspectives linking costs (including SCC inputs) with the GHQ-12 and quality-adjusted life years (QALY) derived from European Quality of Life-5 Dimensions. The primary end point was 6 months, but we also examined findings at 12 months and over the year (full details are provided in the online-only Data Supplement). Unit costs (£, 2010/11 prices) were attached to individual-level resource use quantities (measured retrospectively by self-report) to calculate total costs per participant. Costs are shown in English pounds sterling (£) and can be converted to US dollars using the rate £1=$1.43, based on 2011 purchasing power parities, which equalize the purchasing power of the currencies.32 Discounting was unnecessary for the study time frame. We compared costs and QALY gains using multilevel models with baseline covariates, calculating incremental cost-effectiveness ratios only where 1 intervention showed higher costs and better outcomes. We examined the intervention’s probability of costeffectiveness by constructing cost-effectiveness acceptability curves (5000 bootstrap replications) for threshold ranges of £0 to £2000 for GHQ-12 point gains and £0 to £50 000 for QALY gains.

Results The CONSORT diagram is shown in Figure. Thirty-two stroke service clusters were randomized (75% of these involved were team-based rather than individual SCCs); 29 clusters recruited participants (14 control and 15 intervention clusters). Of the 3 services unable to participate, 1 service changed its referral process after randomization; in 1 service, the SCC was absent because of long-term sickness; and in 1 service, we were unable to identify a researcher to undertake recruitment. Between July 1, 2009 and March 31, 2011, 2488 patients were screened; 800 patients (56.3% [800/1420] of eligible patients) and 208 carers were registered into the trial (399 patients/100 carers in the control group; 401 patients/108 carers in the intervention group). Of these, 66 of 800 (8.2%)

Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on November 15, 2015

Forster et al   LoTS Care Stroke System of Care Trial    2215

Figure. Cluster allocation and patient flow by treatment arm. PI indicates principal investigator.

were by consultee declaration. The study was completed after receipt of the 12-month assessments of the final patients in May 2012. The baseline characteristics of the patients and carers were balanced across the control and intervention groups other than there were more patients with higher education and fewer with language or cognitive impairments in the control group and a difference in length of inpatient stay which was shorter in the control arm. Language and cognitive impairment were accounted for in the statistical modeling. Patient baseline characteristics are shown in Table 2, and

carer baseline characteristics are shown in the online-only Data Supplement. Of patients registered to the trial, 314 of 399 (78.7%) patients in the control group and 318 of 401 (79.3%) patients in the intervention group received the SCC service. Reasons for not receiving the service included SCC not receiving a referral; patient declining; patient death or patient not contactable. Response rates for patient reported outcomes at 6 months were 75.2% (300/399) in control and 77.3% (310/401) in intervention and at 12 months, 67.2% (268/399) in the control and

Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on November 15, 2015

2216  Stroke  August 2015 Table 2.  Patient Baseline Demographic and Clinical Details Patient

Control, n=399 Intervention, n=401

Age, mean (SD)

72.5 (12.84)

70.9 (13.18)

Sex, male, n (%)

218 (54.6)

215 (53.6)

Ethnicity, white, n (%)

389 (97.5)

388 (96.8)

Formal education, n (%)

383 (96.0)

380 (94.8)

 Left education at ≤16, n (%)

282 (73.6)

319 (83.9)

Cerebral infarction, n (%)

343 (86.0)

341 (85.0)

Left hemiparesis, n (%)

179 (44.9)

178 (44.4)

Normal language ability, n (%)

323 (81.0)

295 (73.6)

Normal cognitive function (6CIT), n (%)

267 (66.9)

229 (57.1)

In-hospital stay, d, mean (SD)*

29.5 (34.9)

38.9 (44.4)

Living alone post stroke, n (%)

138 (34.6)

118 (29.4)

6CIT indicates 6-Item Cognitive Impairment Test. *Data for patients with in-hospital stay (control, n=389; intervention, n=391).

70.1% (281/401) in the intervention group. Response rates for carer reported outcomes are provided in the online-only Data Supplement. There was no evidence of a statistically significant difference for the primary end point. The adjusted GHQ-12 mean score at 6 months was 14.9 (SE, 0.6) points for the control group and 15.5 (SE, 0.6) points for the intervention group, with a difference of −0.6 points (95% confidence interval, −1.8 to 0.7), P value of 0.394, and adjusted intracluster correlation coefficients of 0.013 in the control group and 0.025 in the intervention group. Analyses of secondary patient end points also indicated no evidence of statistically significant differences in

the Barthel Index, European Quality of Life-5 Dimensions, Frenchay Activities Index at 6 and 12 months, or the GHQ-12 at 12 months (Table 3; unadjusted patient questionnaire scores are provided in the online-only Data Supplement). The number and types of unmet needs reported (Longer-term Unmet Needs after Stroke questionnaire), deaths, hospital readmissions, institutionalization, and treatment on an emergency outpatient basis were similar for both groups (details are provided in the online-only Data Supplement). Between-group comparisons of the adjusted scores for the carer reported GHQ-12 and Carer Burden Scale at 6 and 12 months indicated no evidence of statistical differences for these outcomes (Table 3). Results of per-protocol analyses (conducted for all patient and carer end points) and sensitivity analyses (conducted on the primary end point) were consistent with results of the intention to treat analyses with no evidence of statistical differences between treatment groups (Table 4). SCCs completed care plans within the trial period for 280 of 401 (69.8%) patients in the intervention group and time logs for 207 of 399 (51.9%) patients in the control group. In the intervention group, there were 269 of 280 (96.1%) care plans that met the definition of compliance with on average 15 of 16 (93.8%) of assessment areas asked at first contact. From the care plan and time log records, the median number of SCC patient assessment contacts was 2 (range, 1–6) contacts in the intervention group and 2 (range 1–7) contacts in the control group. No linear trend was observed between percentage of compliant care plans or percentage of completed time logs and adjusted mean primary outcome for intervention and control services respectively (Figures and further information presented in the online-only Data Supplement). Review of

Table 3.  Primary and Secondary Outcomes: Patient- and Carer-Adjusted Questionnaire Scores at 6 and 12 mo Questionnaire

Control Mean (SE), n

Intervention Mean (SE), n

Difference (SE)

95% CI of the Difference

Adjusted ICC P Value

Control

Intervention

Patient adjusted questionnaire scores  6 mo    GHQ-12

14.9 (0.60), 300

15.5 (0.60), 310

−0.6 (0.65)

  −1.8 to 0.7

0.394

0.013

0.025

   Barthel Index

15.8 (0.33), 296

15.3 (0.28), 307

0.5 (0.33)

−0.2 to 1.1

0.133

0.022

0

   EQ-5D

0.58 (0.025), 288

0.55 (0.022), 301

0.03 (0.025)

−0.02 to 0.08

0.252

0.014

0.059

   FAI

19.0 (0.76), 293

18.0 (0.76), 304

1.0 (0.80)

−0.6 to 2.5

0.229

0

0.014

 12 mo    GHQ-12

14.4 (0.58), 268

13.9 (0.72), 281

0.5 (0.73)

−0.9 to 2.0

0.454

0

0.063

   Barthel Index

15.6 (0.36), 266

15.4 (0.30), 282

0.2 (0.37)

−0.5 to 0.9

0.585

0.049

0.023

   EQ-5D

0.56 (0.030), 259

0.51 (0.028), 270

0.05 (0.033)

−0.02 to 0.11

0.167

0.050

0.044

   FAI

20.2 (0.78), 266

18.7 (0.84), 281

1.5 (0.87)

−0.2 to 3.2

0.078

0

0.025

Carer adjusted questionnaire scores  6 mo  GHQ-12

12.5 (0.69), 82

14.2 (0.79), 80

−1.7 (0.91)

−3.5 to 0.1

0.061

0

0

 CBS

44.5 (2.19), 81

48.4 (1.80), 80

−3.9 (2.51)

−8.8 to 1.1

0.125

0.055

0

   GHQ-12

13.5 (1.11), 71

13.9 (0.85), 73

−0.4 (1.26)

−2.9 to 2.1

0.747

0.255

0

   CBS

43.9 (2.85), 71

48.7 (2.11), 72

−4.8 (3.17)

−11.1 to 1.5

0.132

0.137

0

 12 mo

CBS indicates Carer Burden Scale; CI, confidence interval; EQ-5D, European Quality of Life-5 Dimensions; FAI, Frenchay Activities Index; GHQ-12, 12-item General Health Questionnaire; and ICC, intracluster correlation coefficient.

Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on November 15, 2015

Forster et al   LoTS Care Stroke System of Care Trial    2217 the semistructured interviews indicated that no control SCCs used a similar structured assessment tool or had access to an evidenced-based treatment manual. For the economic evaluation, 564 (70.5%) participants had the combination of complete cost and European Quality of Life-5 Dimensions data and 589 (73.6%) had complete cost and GHQ-12 data. Costs of SCC inputs were similar in both groups (mean difference, £42; 95% confidence interval, −30 to 116; Table 5; mean includes zero costs where SCC inputs were not received). There were no differences in mean total health and social care costs (Table 5). Informal care costs increased after baseline and were significantly higher in the intervention group at 6 months, 12 months, and over the year. Although informal care costs fell between the 6- and 12-month assessments in the control group, they increased over the same period in the intervention group. This is reflected in higher total societal costs in the intervention group at 6 and 12 months and over the year (mean difference at 6 months, £1163; 95% confidence interval, 56–3271; Table 5). QALY gains were similar (Table 5). Imputing missing health and social care costs and QALYs at 6 months did not alter conclusions about cost or QALY differences. Incremental cost-effectiveness ratios were unnecessary because no cost–outcome combination suggested statistically significant between-group increases in both costs and outcomes. At 6 months, the intervention had low probabilities of cost effectiveness from both perspectives and for both outcomes, remaining 750 stroke survivors in the United Kingdom 1 to 5 years after the stroke onset,34 half reported some unmet needs. This deficiency in care was identified in the specific recommendation of the National Stroke Strategy (England)7 and other national guidelines35,36 for regular poststroke reviews. The LoTS care system of care was designed to address this recommendation. Although the structured assessment was delivered to the majority of participants who received the intervention SCC service, it is more difficult to assess the extent to which a change of practice occurred in relation to the problem-solving

and goal-setting approaches. The number of patient assessment contacts was comparable with the control services, with a median of 2 contacts in both. This reflects the pretrial survey and existing SSC service models but may be insufficient contact time to meaningfully review goals and develop a problemsolving approach. Although all of the reference guides are evidence-based, the evidence points to more effective interventions for certain problems than for others, thereby potentially weakening the overall effectiveness of this complex intervention. In addition, some needs may only be successfully addressed if appropriate care and treatment are locally available for the problems identified. This may be particularly relevant in relation to psychological problems.37 This in turn impinges on our main outcome measure that assessed psychological wellbeing.

Future Research There remains a need to develop, implement, and evaluate complex interventions capable of addressing the longer-term needs of people recovering from stroke. Regular review of these needs is an essential component of a longer-term stroke service not only to improve patient care but also to highlight gaps in service provision, which require addressing. The heterogeneity of stroke survivors, who have a wide range of problems and unmet needs, indicates that targeted more bespoke interventions may be the way forward.

Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on November 15, 2015

Forster et al   LoTS Care Stroke System of Care Trial    2219

Acknowledgments We are grateful to all the patients, carers, and stroke services who participated in this trial. Details of participating services are provided in the online-only Data Supplement. The longer-term stroke (LoTS) care trial was a huge collaborative effort, immensely supported by the Stroke Research Networks in the United Kingdom. Our appreciation and gratitude goes to all who supported the development of the intervention, delivery of the training, and implementation of the trial. We thank Peter Wanklyn, Tony Rudd, and Allan House who assisted with the intervention training. We also thank the Consumer Research Advisory Group and the many colleagues in the Academic Unit of Elderly Care and Rehabilitation, Bradford Teaching Hospitals National Health Service Foundation Trust and University of Leeds, and the Clinical Trial Research Unit, University of Leeds and Kings College London, including R. Romeo, M. Heslin, A. Carter, A. Fergusson, and N. Hearfield who provided help and support in the delivery of this trial.

Sources of Funding This report presents independent research commissioned by the National Institute for Health Research (NIHR) under its Programme Grants for Applied Research Programme (grant reference number, RP-PG-0606-1128). The views and opinions expressed by the authors in this publication are those of the authors and do not necessarily reflect those of the National Health Service, the NIHR, or the Department of Health. We also acknowledge the funding support of The Stroke Association (TSA 2006/15).

Disclosures None.

References 1. Murray J, Young J, Forster A. Review of longer-term problems after a disabling stroke. Rev Clin Gerontol. 2007;17:277–292. 2. The Stroke Association. The Stroke Association Manifesto. London: The Stroke Association; 2005. 3. Langhorne P, Taylor G, Murray G, Dennis M, Anderson C, Bautz-Holter E, et al. Early supported discharge services for stroke patients: a metaanalysis of individual patients’ data. Lancet. 2005;365:501–506. doi: 10.1016/S0140-6736(05)17868-4. 4. Legg L, Langhorne P; Outpatient Service Trialists. Rehabilitation therapy services for stroke patients living at home: systematic review of randomised trials. Lancet. 2004;363:352–356. doi: 10.1016/ S0140-6736(04)15434-2. 5. Intercollegiate Stroke Working Party. National Clinical Guideline for Stroke. 4th ed. London: Royal College of Physicians; 2012. 6. Department of Health. National Service Framework for Older People. London: Department of Health; 2001. 7. Department of Health. National Stroke Strategy. London: Department of Health; 2007. 8. Craig P, Dieppe P, Macintyre S, Michie S, I Nazareth I, Petticrew M. Developing and Evaluating Complex Interventions: New Guidance. London: Medical Research Council; 2008. 9. Murray J, Forster A, Young J. Survey to investigate the role of the community stroke care coordinator. Br J Community Nurs. 2008;13:31–36. doi: 10.12968/bjcn.2008.13.1.27981. 10. Forster A, Young J, Nixon J, Chapman K, Murray J, Patel A, et al. Protocol of a cluster randomized trial evaluation of a patient and carercentered system of longer-term stroke care (LoTS care). Int J Stroke. 2015;10:259–263. doi: 10.1111/ijs.12038. 11. Royal College of Physicians of London. National Sentinel Audit for Stroke. London: Royal College of Physicians of London; 2006. 12. Carter BR, Hood K. Balance algorithm for cluster randomized trials. BMC Med Res Methodol. 2008;8:65. doi: 10.1186/1471-2288-8-65. 13. Murray J, Young J, Forster A, Ashworth R. Developing a primary carebased stroke model: the prevalence of longer-term problems experienced by patients and carers. Br J Gen Pract. 2003;53:803–807.

14. Murray J, Ashworth R, Forster A, Young J. Developing a primary carebased stroke service: a review of the qualitative literature. Br J Gen Pract. 2003;53:137–142. 15. Hawe P, Shiell A, Riley T. Complex interventions: how “out of control” can a randomised controlled trial be? BMJ. 2004;328:1561–1563. doi: 10.1136/bmj.328.7455.1561. 16. Farrin A, Russell I, Torgerson D, Underwood M; UK BEAM Trial Team. Differential recruitment in a cluster randomized trial in primary care: the experience of the UK back pain, exercise, active management and manipulation (UK BEAM) feasibility study. Clin Trials. 2005;2:119–124. 17. van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van Gijn J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988;19:604–607. 18. Brooke P, Bullock R. Validation of a 6 item cognitive impairment test with a view to primary care usage. Int J Geriatr Psychiatry. 1999;14:936–940. 19. Collin C, Wade DT, Davies S, Horne V. The Barthel ADL Index: a reliability study. Int Disabil Stud. 1988;10:61–63. 20. The FOOD Trial Collaboration. Performance of a statistical model to predict stroke outcome in the context of a large, simple, randomized, controlled trial of feeding. Stroke. 2003;34:127–33. 21. Goldberg D, Williams P. A User’s Guide to the General Health Questionnaire. Windsor: NFER-Nelson; 1992. 22. Dowswell G, Lawler J, Dowswell T, Young J, Forster A, Hearn J. Investigating recovery from stroke: a qualitative study. J Clin Nurs. 2000;9:507–515. 23. Wade DT, Legh-Smith J, Langton Hewer R. Social activities after stroke: measurement and natural history using the Frenchay Activities Index. Int Rehabil Med. 1985;7:176–181. 24. Krabbe P, Weijnen T. Guidelines for Analysing and reporting EQ-5D Outcomes. In: Brooks R, Rabin R, De Charro F, eds. The Measurement and Valuation of Health Status using EQ-5D: A European Perspective. Dordrecht: Kluwer Academic Publishers; 2003:7–19. 25. Brooks R. EuroQol: the current state of play. Health Policy. 1996;37:53–72. 26. LoTS care LUNS Study Team. Validation of the longer-term unmet needs after stroke (LUNS) monitoring tool: a multicentre study. Clin Rehabil. 2013;27:1020–1028. 27. Elmståhl S, Malmberg B, Annerstedt L. Caregiver’s burden of patients 3 years after stroke assessed by a novel caregiver burden scale. Arch Phys Med Rehabil. 1996;77:177–182. 28. Patel A, Knapp M, Evans A, Perez I, Kalra L. Training care givers of stroke patients: economic evaluation. BMJ. 2004;328:1102. doi: 10.1136/bmj.328.7448.1102. 29. Patel A, Knapp M, Perez I, Evans A, Kalra L. Alternative strategies for stroke care: cost-effectiveness and cost-utility analyses from a prospective randomized controlled trial. Stroke. 2004;35:196–203. doi: 10.1161/01.STR.0000105390.20430.9F. 30. Parker CJ, Gladman JR, Drummond AE, Dewey ME, Lincoln NB, Barer D, et al. A multicentre randomized controlled trial of leisure therapy and conventional occupational therapy after stroke. TOTAL Study Group. Trial of Occupational Therapy and Leisure. Clin Rehabil. 2001;15:42–52. 31. Campbell MK, Mollison J, Grimshaw JM. Cluster trials in implementation research: estimation of intracluster correlation coefficients and sample size. Stat Med. 2001;20:391–399. 32. PPPs and exchange rates. OECD. StatsExtracts 4. http://stats.oecd.org/ Index.aspx?DataSetCode=SNA_TABLE4. Accessed December 22, 2014. 33. National Clinical Guideline Centre. Chronic Heart Failure: The Management of Chronic Heart Failure in Adults in Primary and Secondary Care. London: National Clinical Guideline Centre; 2010. 34. McKevitt C, Fudge N, Redfern J, Sheldenkar A, Crichton S, Rudd AR, et al. Self-reported long-term needs after stroke. Stroke. 2011;42:1398– 1403. doi: 10.1161/STROKEAHA.110.598839. 35. Duncan PW, Zorowitz R, Bates B, Choi JY, Glasberg JJ, Graham GD, et al. Management of Adult Stroke Rehabilitation Care: a clinical practice guideline. Stroke. 2005;36:e100–e143. doi: 10.1161/01. STR.0000180861.54180.FF. 36. Lindsay P, Bayley M, Hellings C, Hill M, Woodbury E,Phillips S. Canadian best practice recommendations for stroke care. CMAJ. 2008;179:S1–S25. 37. Care Quality Commission. Supporting Life After Stroke: A Review of Services for People Who Have Had a Stroke and Their Carers. London: Care Quality Commission; 2011.

Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on November 15, 2015

SUPPLEMENTAL MATERIAL Cluster randomised controlled trial: the clinical and cost-effectiveness of a system of longer-term stroke care

Authored by the LoTS care stroke system of care trial team, who are:

Anne Forster PhD1, John Young MSc1, Katie Chapman PhD1, Jane Nixon PhD2, Anita Patel PhD3 Ivana Holloway MSc2, Kirste Mellish PhD1, Shamaila Anwar PhD2, Rachel Breen PhD1, Martin Knapp PhD5, Jenni Murray PhD4, Amanda Farrin MSc2

1 2 3 4 5

Academic Unit of Elderly Care and Rehabilitation, Bradford Teaching Hospitals NHS Foundation Trust and University of Leeds, UK Leeds Institute of Clinical Trials Research, Clinical Trials Research Unit, University of Leeds, UK Institute of Psychiatry, King’s College London, UK Leeds Institute of Health Sciences, University of Leeds, UK Personal Social Services Research Unit, London School of Economics and Political Science, UK

Correspondence to: Professor Anne Forster, Academic Unit of Elderly Care and Rehabilitation, Temple Bank House, Bradford Royal Infirmary, Bradford, United Kingdom, BD9 6RJ. Email: [email protected]. Tel: +44 1274 383406

Cover title: LoTS care stroke system of care trial

1 

LoTS care stroke system of care trial Contributions The major contributors provided the following contributions to this paper: AFo, the conception and design of the study, acquisition and interpretation of data and drafting of this paper and had final responsibility for the decision to submit for publication. JY, the conception and design of the study, interpretation of data and commenting on the draft of this paper. AP, the design of the study, analysis and interpretation of data and drafting of this paper. JN, the conception and design of the study, interpretation of data and commenting on the draft of this paper. MK, the design of the study and commenting on the draft of this paper. IH, analysis and interpretation of data and drafting of this paper. RB, the design of the study and commenting on the draft of this paper. JM, the conception and design of the study and commenting on the draft of this paper. KM, acquisition and interpretation of data and drafting of this paper, KC, acquisition and interpretation of data and commenting on a draft of this paper, SA, data acquisition and commenting on a draft of this paper. AFa, the conception and design of the study, statistical guarantor and acquisition, analysis and interpretation of data and drafting of this paper. LoTS care Trial Collaboration Trial Steering Committee. H Rodgers, P Langhorne, A Drummond, J Mant, J Firth, K Hood This project was funded by a National Institute for Health Research (NIHR) Programme Grant (Grant Reference Number RP-PG-0606-1128). The report of the full programme of research (including this trial) is published in the NIHR Journals Library Volume 2 issue 6.

For further information see: http://www.journalslibrary.nihr.ac.uk/pgfar/volume2/issue-6#abstract December 2014. The applicants of this grant were: Anne Forster, John Young, Bipin Bhakta, Terry Brady, Jenni Murray, Allan House, Jenny Hewison, Anita Patel, Martin Knapp, Amanda Farrin. Participating SCCs Alison Allerton, Susan Altham, Cate Benfold, Michelle Black, Claire Butterworth, Sarah Butterworth, Liz Brown, Yvonne Brown, Bronagh Byrne, Steve Carpenter, Carmen Cartmell, Michael Chivhunga, Mary Downing, Jo Featherstone, Frank Foreman, Debbie Gibbons, Sheila Grimes, Sandra Griffiths, Heather Hazzard, Jenny Hewitson, Grace Hogan, Barbara Holley, Jules Jeffreys, Sandra Leech, Kirstie McCue, Lisa Mewton, Vicki Moore, Nicola Reynolds, Carole Saunders, Susan Targett, Hilary Thompson, Anita Tunstall, Raj Vaithilingam, Susan Wilson, Linda Woollatt Participating hospitals and PCTs (number of patients recruited) and research teams Airedale General Hospital (20), Paula Sharratt; Ashford Hospital (16), Emma Young; Blackpool Victoria Hospital (27), J McIlmoyle, Hannah Goddard, Joanne Howard, Carol Jeffs; Cardiff Royal Infirmary (27), Catherine Watkins; Chesterfield Royal Hospital (33), S Punnoose, J Cooke, Maggie Ball; Chorley and South Ribble Hospital (12), Julie Brown, Alison O’Keeffe, Angela Alty, Jackie Purcell, Stephen Duberley; Christchurch Hospital (6), Anna Orpen; Craigavon Hospital (36), Jane Greene, Diane Crooks, Mary McParland; Crediton Hospital (1), Leigh Barron, Nicola Wedge; Daisy Hill Hospital (27), Jane Greene, Mary McParland; Darent Valley Hospital (27), P Aghoram, S Hussein, Tracey Daniel, Emily Jay; Derbyshire County PCT (2), Maggie Ball; Derriford Hospital (20), Ben Hyams, Nicola 2 

Persad, Claire Brown; Erne Hospital (20), Celia Diver-Hall; Fairfield Hospital (3), Nessa Thomas; Gloucestershire Royal Hospital (13), Fiona Davis, Pauline Brown; Inverclyde Royal Hospital (2), Marlene McKinney, Donald Fraser; Ipswich Hospital (18), A Dunthorne, Nicola Rands; Lymington New Forest Hospital (17), G Durward, Vanessa Pressly; Manchester Royal Infirmary (1), Stephen Duberley; Mount Gould Hospital (6), Ben Hyams, Nicola Persad, Claire Brown; New Cross Hospital (15), D Morgan, Karla Preece; Norfolk and Norwich University Hospital (17), Ingrid Watmough, Esther Thomas, Niki Wyatt; North Manchester General Hospital (29), H Finneran, Stephen Duberley; Norwich Community Hospital (9), Ingrid Watmough, Esther Thomas, Niki Wyatt; Pendle Community Hospital (15), M Goorah, A Sangster; Royal Blackburn Hospital (8), M Goorah, A Sangster; Royal Bournemouth Hospital (9), Anna Orpen, Catherine Ovington; Royal Cornwall Hospital (19), F Harrington, G Courtauld, Christine Schofield, Ali James; Royal Devon and Exeter Hospital (18), Leigh Barron, Nicola Wedge; Royal Preston Hospital (15), Lorna Fazackerley, Stephen Duberley, Christina Gilmour, Bindu Gregary; Singleton Hospital (45), Helen Goldring; Southampton General Hospital (13), G Durward, Vanessa Pressly; Southlands Hospital (8), Rajen Patel, Carol Simmons, Angela Dunne; Southport and Formby District General Hospital (14), Helen Terrett; Stoke Mandeville Hospital (15), Susan Buttfield; St Peters Hospital (29), Emma Young; Sunderland Royal Hospital (2), Janice O’Connell; Tyrone County Hospital (5), Celia Diver-Hall; Walsall Manor Hospital (45), Jo McCormack, Karla Preece, Steve Hurdowar; Warrington Hospital (9), O Otaiku, Lynne Connell; West Park Rehabilitation Hospital (14), D Morgan, Karla Preece; Whiston Hospital (20), Sharon Dealing; Williton Hospital (18), Libby Caudwell, Sarah Edwards; Worthing Hospital (6), Rajen Patel, Carol Simmons, Angela Dunne; University Hospital, Coventry (18), C Randall, Lucy Aldridge, Martine Pritchard; University Hospital of Hartlepool (34), David Bruce, Susan Crawford; University Hospital of North Durham (0), Cath Huntley; University Hospital of North Tees (17), Helen Skinner, Susan Crawford Location of services – categorised by UK Strategic Health Authority (geographical areas).    

Control (n=16)  Intervention (n=16) Total (n=32)

England: 

 

North East 

1  

2 

3 

North West 

2  

4 

6 

East Midlands 

1  

0 

1 

West Midlands 

2  

1 

3 

South East Coast 

2  

1 

3 

South Central 

1  

0 

1 

South West 

3  

3 

6 

East  

1  

1 

2 

 

 

Scotland 

1  

1 

2 

Northern Ireland 

2  

1 

3 

Wales 

0  

2 

2 

 

 

 

3 

The LoTS care system of care (intervention) The LoTS care system of care was developed (2000-2006) in keeping with the MRC Framework for the development and evaluation of complex interventions1 and was based on systematic reviews and synthesis of qualitative and quantitative literature, national surveys, interviews with patients and carers and feasibility testing. Through these procedures a structured and systematic approach to patient assessment which was based on the problems that patients and carers themselves had identified as of central importance was developed.   In order to identify treatment algorithms, in collaboration with information specialists at the University of Leeds, a hierarchical comprehensive structured protocol for identifying evidence in each of the 15 problem areas was developed. This protocol included identifying relevant stroke specific and problem specific guidelines, metaanalyses and systematic reviews, and if necessary individual randomised controlled trials. Two researchers independently reviewed all outputs. Guidelines, reviews and papers identified for inclusion were assessed for quality using standard tools. Drafted treatment algorithms were peer reviewed by external experts, prior to compilation in the manual developed to support our newly developed system of care. The resulting LoTS care stroke system of care manual comprises: patient and carer structured assessment representing the identified problem areas linked to a reference guide and treatment algorithm; a directory of service information; and a selection of validated assessment scales for specific areas such as depression and cognitive impairment are included as appendices in the manual. A Care Plan was developed to be completed for each patient (and if relevant, carer) following each contact which included patient details, the structured assessment and related prompt questions, patient and carer goal and action plans. An optional checklist detailing the content of the assessment to be given to patients beforehand was also developed (see below). Thus a manualised structured system for longer-term stroke care that is comprehensive (encompasses all areas of potential concern to patient and carer) but individualised (patient specific action plans constructed) was created. 1. Craig P, Dieppe P, Macintyre S, Michie S, I Nazareth I, Petticrew M, Developing and evaluating complex interventions: new guidance. London: Medical Research Council; 2008.

4 

CLIENT CHECKLIST These are the main areas that will be addressed during your assessment. If you wish, you can make a note of any concerns or queries on this sheet as a reminder. Communication - Information needs about your stroke and who to contact.

Medicine – Any problems relating to your medication.

Pain – Any problems with pain and current treatment.

5 

Mobility/falls – Getting around inside and outside your home and fear of falling.

Personal hygiene – Washing/bathing, cutting toenails, dressing, dental treatment.

Shopping/meal preparation – Getting to the shops, getting a balanced diet, making meals.

House & home – Doing housework, gardening, general repairs or finding a new home.

6 

Memory – Difficulty remembering things or concentrating.

Transport – Getting back to driving or using private / public transport.

Finances & benefits – Organising bills payments, entitlements, council rebates, working.

Continence – Need for pads etc., further assessments, advice.

Sexual relations – Need for advice on changes to your intimate life.

7 

Mood (patient) – Feeling down, anxious or worried.

Social life (patient) - Day-time interests and hobbies, getting to see others, religious practice, relationship problems, having a holiday.

Mood (partner/relative/friend) – Feeling down, anxious, worried, coping with your caring role in any of the above areas e.g. paying bills, house and home, patient’s personal hygiene.

Social (partner/relative/friend) – Having a break, arranging a holiday, relationship problems, working.

Any other problems?

 

8 

Patients structured assessment (as represented in the Care Plan see below) Transfer of Care ‐ Reference Guide 1   Has a discharge plan been prepared by the hospital?   Have you had a home visit? What was the outcome?  (Discuss issues around knowledge/understanding of care plan, involvement, falls  management, information sources)  Communication & Information ‐ Reference Guide 2   Have you been given clear information about your condition, treatment and services  availaable?    Medicines & General Health ‐ Reference Guide 3  Do you have any problems with your medication (side‐effects, drug cocktail, non‐prescription  drugs)?  Do you have any problems taking your medication?    Do you have any other health problems?    Are these being reviewed?   Pain ‐ Reference Guide 4  Do you have any pain?  Is the pain being  treated?  Is the treatment helping?  Mobility/Falls ‐ Reference Guide 5  Can you get around indoors?  Can you get out of your house?    How confident are you about carrying out various daily activities without falling?    How do you feel about your recovery so far?   (Discuss patterns of physical and emotional recovery)  Personal Hygiene & Dressing ‐ Reference Guide 6  Are you having any difficulty with personal care like washing, cutting your nails, oral hygiene  or dressing?  Do you ever need help?  Shopping & Meal Preparation ‐ Reference Guide 7  Can you prepare your own meals?    Can you go shopping?  Are you getting the right food? 

9   

House & Home ‐ Reference Guide 8  Can you do your housework?    Does anyone help?     Do you have any problems with your accommodation?   Cognition ‐ Reference Guide 9  Do you often have a problem remembering things that happened recently?  Do you often forget where you have put things?  Do you find it difficult to concentrate?  Driving & General Transport ‐ Reference Guide 10  Do you have trouble with transport?  Finances & Benefits ‐ Reference Guide 11   Do you have any difficulty managing your money?    Are you able to pay your bills?    Are you receiving all the benefits you are entitled to? (Use a benefits checklist)  Continence ‐ Reference Guide 12  Do you ever have accidents with your bladder/bowel? How is this being managed?  Sexual Functioning ‐ Reference Guide 13  Some people after stroke experience sexual problems such as lack of interest or practical  difficulties. Have you experienced any changes sexually since your stroke?  Patient Mood ‐ Reference Guide 14  Have you recently felt very sad or fed up?    Have you felt anxious, frightened or worried?  Do you have a partner, relative or friend you feel close to?  Do you get on well?  Can you talk  about your worries or problems with them?  Patient Social Needs ‐ Reference Guide 15  How do you spend your day?    Do you have enough to do?    Are you happy with your social life?  Do you wish you had more social contact with others?   (Discuss work as appropriate, changes in role, self image, and relationships.)  Other   Are you having any other problems relating to your stroke? 

The manual is available from the corresponding author: [email protected] 10   

Sample (contact 1) of the Care Plan

11   

12   

13   

14   

Training in the LoTS care system of care Training in the delivery of the system of care was provided for each of the SCC services randomised to the intervention through two centrally based Royal College of Nursing (England) accredited training days approximately one month apart. Because of the nature of a cluster trial the SCCs randomised to the intervention were unaware of what the new service model consisted of prior to these training days. The system of care was well received when presented by the trial team demonstrating ‘face validity’ for the intervention. In the first training day the principles of the intervention were presented as described below:  Patient-centred (comprehensive coverage of problems identified by patients and carers)  Provide assessment areas (checklist) before assessment whenever possible  Ask all assessment questions  Keep accurate records  Problem solving approach with collaborative goal setting  Follow-up on actions  Review goals  Non prescriptive – individual creativity  According to local services / resources  Within patient’s own environment wherever possible  Timing / duration of intervention (according to national recommendations)  Flexible approach to carer assessment Practical issues of implementation were addressed, specifically, design of the care plans and manual. A workshop was led by Professor Allan House on assessment and problem solving techniques. The content of the second training day was informed by the needs of the SCCs. Thus a lecture on pain management was provided (by Professor Tony Rudd or Dr Peter Wanklyn) and discussion on benefits was led by a colleague from the Department for Work and Pensions. The format and layout of the Care Plan was further reviewed and refined following the SCCs’ feedback. The intent was to ensure that the paperwork captured all information that they might require to deliver their service, and therefore replace the current documentation. The training was supported by a CD of the training day. After the second training day, the SCCs were asked to use the Care Plan for all patient contacts (and carers if carer assessments were performed).The SCCs implemented the new system of care in their service over four to six months providing sufficient time for the implementation of the system of care into standard practice, when they were comfortable with implementation (demonstrated through review of completed care plans by the trial team) trial recruitment began. Full details of the development of the LoTS care system of care is available in: Forster A, Mellish K, Farrin A, Bhakta B, Murray J, Patel A, House A, Hewison J, Knapp M, Breen R, Chapman K, Holloway I, Hawkins R, Shannon R, Nixon J, Jowett A, Horton M, Alvarado N, Anwar S, Tennant A, Godfrey M, Young J. Development and evaluation of interventions and tools to improve patient and carer centred outcomes in longer-term stroke care (LoTS care) and exploration of adjustment post stroke. NIHR Journals Library. For further information see: http://www.journalslibrary.nihr.ac.uk/pgfar/volume-2/issue6#abstract December 2014. Access date April 2015.

15   

Summary of potential barriers and facilitators to intervention delivery The system of care was developed in 2000-2006. Although review of patients’ post-stroke needs is recommended policy, no systematic approach had previously been developed for routine monitoring, problem identification and co-ordination of services to assist stroke patients and their families as they continue to recover from their stroke and make life adjustments to its consequences. The LoTS care system of care was based on a systematic review and synthesis of the available qualitative literature reporting interviews with stroke patients and carers in which longer-term stroke-related issues were discussed.1 A complementary review of quantitative stroke surveys assessed the prevalence of these problem areas.2 To confirm content validity, emerging findings were checked and refined by stroke patients and carers in individual interviews and in focus group.3 This approach ensured that the system of care is targeted at the most common stroke-related problems of central importance to stroke patients and their carers. Intervention implementation was informed by policy recommendations and a national survey. Facilitators to intervention delivery within this trial evaluation included:  



     

All Stroke Care Co-ordinators (SCCs) randomised to deliver the intervention recognised the importance of the work and relevance of the content of the system of care (details of the intervention were not divulged until after site randomisation). The intervention was supported by a detailed manual which included 16 assessment questions (patient) and 13 questions (carer) representing the identified patient/carer problem areas, linked to reference guides containing educational text with algorithms of evidenced-based treatment options. The SCCs were also provided with a ‘Care Plan’ which included the assessment questions and a goal and action planner to be completed following each contact (patients and carers). Intervention SCCs were actively engaged in finalising the intervention materials. Specifically the Care Plan was adapted to capture all information that they might require to deliver their service, and therefore replace the current documentation. Thus the LoTScare Care Plan became embedded in routine practice rather than an additional component. The manual was recognised as a useful tool to promote multidisciplinary working The SCCS had opportunity to practice the new intervention between the two training days which enabled discussion with the trial team and their peers at the second training day. The training was delivered by eminent stroke clinicians/researchers enhancing credibility. The training was responsive to specific needs (for example, specific training was provided on pain management and benefits available). The training was supported by a CD of the training day, which could be accessed later. Participant recruitment was not opened until all parties were satisfied that the system of care was being delivered,

Potential barriers to intervention delivery: 

Some SCCs left and were replaced during the trial and in services where there were multidisciplinary teams all SCCs did not necessarily attend the training. The intent was that the training would be cascaded down by staff who attended the training days 16 

 

    

to other staff in the team. It may be that this commonly used ‘cascade’ method was not as effective as we would have wished. It may be that in this service model the notion of comprehensive holistic assessment becomes dissipated by individual allied health professionals delivering their single discipline input, for example, therapists prioritising mobility problems. Whilst the number of assessment contacts was not prescribed, a key principle of the intervention is that goals are reviewed at subsequent contacts, which will necessarily be limited if the number of contacts are small. Despite regular multidisciplinary team meetings the original Care Plan assessment had not always been re-visited to check that actions had been undertaken and goals reached. Although all of the reference guides are evidence-based, the evidence points to more effective interventions for certain problems than for others, thereby weakening the overall effectiveness of this complex intervention. The challenge in pragmatic multicentre trial evaluation in rehabilitation is to provide some guidance on the intervention to capture the main features, enhance external validity and improve generalisability whilst making it sufficiently flexible for it to be acceptable to staff and deliverable in a range of service models.4 Our approach was to manualise the intervention supported by training days with an opportunity to practice delivery prior to the start of patient recruitment. The Care Plan was well completed, but we were perhaps less successful in changing the behaviours and mind set of the SCCs. Further work should explore how to embed behaviour change techniques in both intervention training and delivery and in the intervention itself.

References 1. Murray J, Ashworth R, Forster A, Young J. Developing a primary care-based stroke

service: a review of the qualitative literature. Br J Gen Pract. 2003;53:137-42. 2. Murray J, Young J, Forster A, Ashworth R. Developing a primary care-based stroke

model: the prevalence of longer-term problems experienced by patients and carers. Br J Gen Pract 2003;53:803-7. 3. Murray J. The development of a primary care-based model for after stroke care. [PhD Thesis]. Leeds: University of Leeds; 2007. 4. Hawe P, Shiel A, Riley T. Complex interventions: how “out of control” can a randomised controlled trial be? BMJ 2004; 328: 1561–3.

Brief overview of differences between treatment arms Considerable detail was collected on the participating services before, during and after the trial. This included a survey of existing practices prior to enrolling in the trial and semistructured interviews with all participating Stroke Care Co-ordinators (SCCs) at the onset of the trial in which they described their service and client group. These interviews were repeated mid-way through recruitment and after 12 months follow-up. Through these interviews they described their service and client group (including any changes). Prior to commencement of the trial none of the SCCs used or had access to an assessment schedule that was specifically structured round the longer-term problems of stroke patients and their carers, or had access to a manual of evidenced-based treatment algorithms linked to these assessment questions.

17   

The interviews before, during and after the trial demonstrated there were clear differences between the control and intervention arms.      

Intervention SCCs were provided with specific training which included techniques of the problem solving approach (supported by a CD which they could access later) which was not provided to the Control SCCs. Intervention SCCs provided patients with the LoTS care checklist prior to assessment, whereas no control SCCs mentioned an equivalent checklist in their service. None of the Control site SCCs had an assessment schedule created around the identified problems of stroke patients and their carers. None of the Control site SCCs had a purposely designed Care Plan which included the assessment questions and a goal and action planner to be completed following each contact (patients and carers). None of the Control site SCCs used or had access to a manual of evidenced-based treatment. Although control SCCs intended to provide a fully holistic initial assessment, it is not clear from interviews whether they actually did, whereas the SCCs in the intervention arm appeared to provide a holistic assessment more consistently.

SUPPLEMENTAL TABLES Response rates for patients and carer outcomes Response rates for patient reported outcomes at six months were: 75.2% (300/399) in control and 77.3% (310/401) in intervention and at 12 months 67.2% (268/399) in control and 70.1% (281/401) in intervention. Response rates for carer reported outcomes at six months were: 88.0% (88/100) in control and 82.4% (89/108) in intervention and at 12 months 71.0% (71/100) in control and 67.6% (73/108) in intervention.

18   

Table I Carer baseline demographic details Carer baseline demographic details Carer

Control (n=100)

Intervention (n=108)

Age – Mean (SD)

61.4 (14.07)

61.0 (15.02)

Sex: Male – N (%)

32 (32.0)

35 (32.4)

Ethnicity: White – N (%)

99 (99.0)

105 (97.2)

Formal education – N (%)

97 (97.0)

103 (95.4)

If formal education: Left education at 16 or less - N (%)

68 (70.1)

76 (73.8)

Partner

67 (67.0)

70 (64.8)

Daughter / son

29 (29.0)

33 (30.6)

4 (4.0)

5 (4.6)

78 (78.0)

85 (78.7)

Carer - patient relationship – N (%)

Other Living with patient post stroke – N (%)

19   

LUNS Questionnaire outcomes The number and types of unmet needs reported (LUNS questionnaire) were similar for the control and intervention groups at six and 12 months. A median of 2.5 unmet needs was reported; the prevalence of individual unmet needs ranged from 2% to 51% of those completing the questionnaire. Table II LUNS summaries – number of unmet needs for patients at 6 months LUNS summaries Number of unmet needs for patients at 6 months Control Intervention (n=399) (n=401)

Total (n=800)

N of unmet needs at 6 months 0

57 (14.3%)

43 (10.7%) 100 (12.5%)

1

46 (11.5%)

46 (11.5%)

92 (11.5%)

2

44 (11.0%)

41 (10.2%)

85 (10.6%)

3

25 (6.3%)

39 (9.7%)

64 (8.0%)

4

28 (7.0%)

31 (7.7%)

59 (7.4%)

5

15 (3.8%)

16 (4.0%)

31 (3.9%)

6

18 (4.5%)

17 (4.2%)

35 (4.4%)

7

15 (3.8%)

18 (4.5%)

33 (4.1%)

8

13 (3.3%)

17 (4.2%)

30 (3.8%)

9

8 (2.0%)

8 (2.0%)

16 (2.0%)

10

3 (0.8%)

10 (2.5%)

13 (1.6%)

11

3 (0.8%)

5 (1.2%)

8 (1.0%)

12

7 (1.8%)

1 (0.2%)

8 (1.0%)

13

4 (1.0%)

7 (1.7%)

11 (1.4%)

14

2 (0.5%)

1 (0.2%)

3 (0.4%)

15

1 (0.3%)

1 (0.2%)

2 (0.3%)

16

0 ( 0.0%)

1 (0.2%)

1 (0.1%)

17

1 (0.3%)

0 ( 0.0%)

1 (0.1%)

18

0 ( 0.0%)

1 (0.2%)

1 (0.1%)

Missing

109 (27.3%)

98 (24.4%) 207 (25.9%)

20   

Table III LUNS summaries – number of unmet needs for patients at 12 months LUNS summaries Number of unmet needs for patients at 12 months Control (n=399)

Interventio n (n=401)

Total (n=800)

N of unmet needs at 12 months 0

62 (15.5%)

1

35 (8.8%)

36 (9.0%)

71 (8.9%)

2

34 (8.5%)

43 (10.7%)

77 (9.6%)

3

29 (7.3%)

31 (7.7%)

60 (7.5%)

4

20 (5.0%)

28 (7.0%)

48 (6.0%)

5

21 (5.3%)

11 (2.7%)

32 (4.0%)

6

11 (2.8%)

12 (3.0%)

23 (2.9%)

7

12 (3.0%)

19 (4.7%)

31 (3.9%)

8

10 (2.5%)

8 (2.0%)

18 (2.3%)

9

10 (2.5%)

7 (1.7%)

17 (2.1%)

10

6 (1.5%)

7 (1.7%)

13 (1.6%)

11

4 (1.0%)

3 (0.7%)

7 (0.9%)

12

2 (0.5%)

6 (1.5%)

8 (1.0%)

13

3 (0.8%)

2 (0.5%)

5 (0.6%)

14

1 (0.3%)

1 (0.2%)

2 (0.3%)

15

1 (0.3%)

0 ( 0.0%)

1 (0.1%)

18

1 (0.3%)

0 ( 0.0%)

1 (0.1%)

21

0 ( 0.0%)

1 (0.2%)

1 (0.1%)

Missing

62 (15.5%) 124 (15.5%)

137 (34.3%) 124 (30.9%) 261 (32.6%)

21   

Figure I Prevalence of individual unmet needs at 6 months and 12 months (out of those who returned completed LUNS questionnaire)

22   

Table IV Summaries of patient deaths Patients deaths Control (n=399) Intervention (n=401) Total (n=800) Patient deaths 6 months timepoint

27 (6.8%)

31 (7.7%)

58 (7.3%)

12 months timepoint

5 (1.3%)

4 (1.0%)

9 (1.1%)

After 12 months follow-up*

2 (0.5%)

2 (0.5%)

4 (0.5%)

32 (8.0%)

35 (8.7%)

67 (8.4%)

Overall deaths

*Not included in the overall deaths as reported after the follow up period

23   

Table V Treatment on an emergency outpatient basis and hospital admissions at 6 months 6 month researcher follow up summary Control (n=399)

Intervention (n=401)

Total (n=800)

395 (99.0%)

400 (99.8%)

795 (99.4%)

4 (1.0%)

1 (0.2%)

5 (0.6%)

Yes

105 (26.3%)

113 (28.2%)

218 (27.3%)

No

290 (72.7%)

261 (65.1%)

551 (68.9%)

4 (1.0%)

27 (6.7%)

31 (3.9%)

Follow-up form completed N (%) Yes Withdrawn Patient visited A&E in past six months N (%)

Missing Number of A&E visits per patient Mean (SD) Median (Range)

0.4 (0.95)

0.4 (0.83)

0.4 (0.89)

0.0 (0.0, 8.0)

0.0 (0.0, 7.0)

0.0 (0.0, 8.0)

Patient admitted to hospital overnight in past 6 months N (%) Yes

113 (28.3%)

97 (24.2%)

210 (26.3%)

No

282 (70.7%)

303 (75.6%)

585 (73.1%)

4 (1.0%)

1 (0.2%)

5 (0.6%)

Missing Number of hospital admissions Mean (SD) Median (Range)

0.4 (0.93)

0.3 (0.66)

0.4 (0.81)

0.0 (0.0, 10.0)

0.0 (0.0, 5.0)

0.0 (0.0, 10.0)

10 (2.5%)

23 (2.9%)

4 (1.0%)

12 (1.5%)

0.0 (0.20)

0.0 (0.13)

0.0 (0.17)

0.0 (0.0, 3.0)

0.0 (0.0, 2.0)

0.0 (0.0, 3.0)

0.2 (1.36)

0.0 (0.63)

0.1 (1.06)

0.0 (0.0, 20.0)

0.0 (0.0, 11.0)

0.0 (0.0, 20.0)

Number of patients that died during hospital admission 13 (3.3%) Number of patients that spent time in a specialist care unit 8 (2.0%) Number of times spent in a specialist care unit Mean (SD) Median (Range) Time spent in a specialist care unit (days) Mean (SD) Median (Range)

24   

Table VI Treatment on an emergency outpatient basis and hospital admissions at 12 months 12 month researcher follow up summary Control (n=399)

Intervention (n=401)

Total (n=800)

374 (93.7%)

374 (93.3%)

748 (93.5%)

21 (5.3%)

26 (6.5%)

47 (5.9%)

4 (1.0%)

1 (0.2%)

5 (0.6%)

Yes

71 (17.8%)

77 (19.2%)

148 (18.5%)

No

303 (75.9%)

275 (68.6%)

578 (72.3%)

25 (6.3%)

49 (12.2%)

74 (9.3%)

Follow-up form completed N (%) Yes Died Withdrawn Patient visited A&E in past six months N (%)

Missing Number of A&E visits per patient Mean (SD) Median (Range)

0.3 (1.17) 0.0 (0.0, 19.0)

0.3 (0.66)

0.3 (0.95)

0.0 (0.0, 4.0)

0.0 (0.0, 19.0)

Patient admitted to hospital overnight in past 6 months N (%) Yes

74 (18.5%)

77 (19.2%)

151 (18.9%)

No

300 (75.2%)

297 (74.1%)

597 (74.6%)

25 (6.3%)

27 (6.7%)

52 (6.5%)

Missing Number of hospital admissions Mean (SD) Median (Range)

0.3 (0.73)

0.3 (0.68)

0.3 (0.71)

0.0 (0.0, 5.0)

0.0 (0.0, 4.0)

0.0 (0.0, 5.0)

Number of patients that died during hospital admission 3 (0.8%)

5 (1.2%)

8 (1.0%)

4 (1.0%)

6 (0.8%)

Number of patients that spent time in a specialist care unit 2 (0.5%) Number of times spent in a specialist care unit Mean (SD) Median (Range)

0.0 (0.07)

0.0 (0.13)

0.0 (0.11)

0.0 (0.0, 1.0)

0.0 (0.0, 2.0)

0.0 (0.0, 2.0)

0.0 (0.29)

0.0 (0.56)

0.0 (0.45)

0.0 (0.0, 5.0)

0.0 (0.0, 10.0)

0.0 (0.0, 10.0)

Time spent in a specialist care unit (days) Mean (SD) Median (Range)

25   

Table VII: Patients' unadjusted questionnaire scores at different time points by arm Patients' unadjusted questionnaire scores at different time points by arm Baseline Questionnaire

6 months

12 months

Control

Intervention

Control

Intervention

Control

Intervention

Mean (SD) N

Mean (SD) N

Mean (SD) N

Mean (SD) N

Mean (SD) N

Mean (SD) N

Barthel Index prestroke

19.2 ( 1.98) 399

19.2 ( 2.22) 401

N/A

N/A

N/A

N/A

Barthel

15.2 ( 4.48) 398

14.4 ( 5.06) 401

16.2 ( 4.24) 296

15.5 ( 4.48) 307

16.3 ( 3.91) 266

15.7 ( 4.26) 282

GHQ12

15.2 ( 7.26) 396

15.9 ( 7.03) 397

15.3 ( 7.32) 305

16.2 ( 7.17) 318

14.5 ( 6.78) 268

14.3 ( 6.87) 281

EQ5D

0.56 (0.340) 382

0.51 (0.378) 381

0.61 (0.339) 288

0.56 (0.322) 301

0.61 (0.316) 259

0.54 (0.318) 270

FAI

28.1 ( 9.71) 398

27.8 ( 9.70) 399

20.3 ( 11.41) 293

18.4 ( 11.27) 304

21.5 ( 11.30) 266

18.8 ( 11.53) 281

Table VIII Patient adjusted questionnaire scores at 6 and 12 months – per protocol population Patient adjusted questionnaire scores at 6 and 12 months – per protocol population Adjusted ICC Difference (SE)

95% CI of the difference

p-value

Control

Intervention

-0.8 (0.76)

(-2.3, 0.7)

0.280

0

0.058

15.4 (0.34) 244

0.3 (0.33)

(-0.3,1.0)

0.310

0

0

0.57 (0.028) 223

0.56 (0.025) 228

0.01 (0.026)

(-0.04, 0.07)

0.602

0.007

0

18.9 (0.90) 225

17.9 (0.87) 467

1.0 (0.87)

(-0.7, 2.7)

0.240

0

0.011

GHQ-12

13.8 (0.70) 201

12.9 (0.85) 227

0.8 (0.85)

(-0.9, 2.5)

0.336

0

0.097

Barthel Index

15.6 (0.35) 203

15.6 (0.36) 226

0.0 (0.37)

(-0.8, 0.7)

0.933

0

0.036

0.55 (0.033) 199

0.52 (0.032) 215

0.03 (0.034)

(-0.03, 0.1)

0.328

0.034

0.046

20.0 (0.93) 201

18.2 (0.95) 225

1.7 (0.94)

(-0.1, 3.6)

0.069

0

0.016

Control Mean (SE) N

Intervention Mean (SE) N

GHQ-12

14.0 (0.69) 228

14.8 (0.78) 248

Barthel Index

15.7 (0.35) 228

Questionnaire 6 months:

EQ-5D FAI 12 months:

EQ-5D FAI

Table IX Carer adjusted questionnaire scores at 6 and 12 months – per protocol population Carer adjusted questionnaire scores at 6 and 12 months – per protocol population Adjusted ICC Control Mean (SE) N

Intervention Mean (SE) N

Difference (SE)

GHQ-12

12.6 (0.82) 58

14.0 (0.90) 67

CBS

45.5 (2.64) 57

GHQ-12 CBS

Questionnaire

95% CI of the difference

p-value

Control

Intervention

-1.4 (1.04)

(-3.5, 0.7)

0.181

0

0

49.7 (2.25) 67

-4.3 (3.02)

(-10.2, 1.7)

0.162

0.073

0.037

13.7 (1.18) 51

14.2 (0.95) 61

-0.50 (1.34)

(-3.2, 2.1)

0.707

0.145

0

45.1 (2.80) 51

50.1 (2.46) 60

-5.0 (3.22)

(-11.4, 1.3)

0.120

0.054

0.025

6 months:

12 months:

26   

Figure II: Adjusted control service differences of GHQ-12 scores in relation to adjusted control score by service completion of time logs

No linear trend was observed between a service’s adjusted mean primary outcome and mean percentage of completed time logs.

27   

Figure III Adjusted intervention service differences of GHQ-12 scores in relation to adjusted intervention score by service compliance with intervention

No linear trend was observed between a service’s adjusted mean primary outcome and mean percentage of compliant care plans

28   

Economic evaluation methods Perspective The economic evaluation was undertaken from two perspectives: (a) health and social care and (b) societal. Health and social care costs included: nursing/residential care; hospital inpatient, outpatient, day hospital and accident and emergency services; primary care; community based health services; and social care services. Societal costs included all of these categories plus informal care costs. Time horizon In keeping with the outcome evaluation, the primary endpoint was 6 months. We further examined costs and outcomes at 12 months and over one year (one year costs and QALYs were the sum of values at 6 and 12 months). Resource use data Resource use data were collected by self-report at the individual-level using appropriately tailored versions of schedules used in previous stroke studies.1,2 These were administered alongside other measures at baseline (with reference to the previous three months), 6 and 12 months (with reference to the previous 6 months). We also prospectively measured the duration of SCC inputs (of both a contact and noncontact nature) at the individual patient level in both intervention and control groups. In the intervention group, these inputs were measured as part of the Care Plan. In the control group, staff recorded equivalent inputs on a specifically designed Time Log. Pay bands were also recorded to enable cost estimation by staff level. Data from the SCC surveys were used for service-relevant imputation values in the event of missing or partially completed Care Plans or Time Logs. Costs Individual-level resource use quantities were combined with unit costs (summarised in Table X; further details available upon request) at 2010/2011 prices to calculate a cost per participant. Costs are shown in English pounds sterling (£), and can be converted to United States dollars using the rate £1 = $1.43, based on 2011 purchasing power parities which equalise the purchasing power of the currencies.3 Discounting was not necessary since the evaluation period did not exceed one year. Table X: Unit cost summary Category Residential care home stay Nursing home stay Inpatient services Day hospital/day case services Outpatient services Primary care/community-based services Value of carer time – average wage Value of carer time – leisure time Stroke care co-ordinator Stroke multi-disciplinary meeting

Unit

Unit cost (£, 2010/11) Night 75 Night 76 Bed day Range 315 to 1213 Activity Range 230 to 1190 Visit Range 3 to 772 Contact/hour/item Range 9 to 152 Hour 15 Hour 5 Hour Range 19 to 78 Hour 284 29 

 

Quality-adjusted life years We estimated QALYs from EQ-5D4 health state measurements at baseline, 6 months and 12 months. We attached utility weights from a United Kingdom general population survey5 to health states at each time point, with appropriate adjustments for the period of time involved, and estimated QALY gains using linear interpolation to calculate the area under the QALY curve. Cost–effectiveness and cost–utility analyses Given two cost perspectives and two outcomes, we examined the following four costoutcome combinations at the primary endpoint of 6 months: (a) GHQ-12 and health and social care costs (b) GHQ-12 and societal costs (c) QALYs and health and social care costs and (d) QALYs and societal costs. Further examining costs and outcomes at 12 months and over one year led to total of 12 cost-outcome combinations to consider. We only planned to estimate incremental cost-effectiveness ratios (ICERs) for cost-outcome combinations where one group had both higher costs and better outcomes. Cost-effectiveness planes and costeffectiveness acceptability curves (CEACs) (see below) were only examined for the primary endpoint. We plotted between-group mean differences in total health and care costs, QALY gains and GHQ-12 scores at 6 months on cost-effectiveness planes. Differences were calculated using bootstrapped regressions (5000 replications; Stata version 10.1) with an adjustment for cluster and the same patient level baseline covariates as used for other group comparisons: Barthel Index (pre- and post- stroke), gender, age, living circumstances (living alone vs. with carer), stroke severity as represented by speech and language impairment (normal/ impaired) and 6CIT score (normal/ impaired), utility score and GHQ-12 score (and total cost from the relevant perspective for the comparison of costs). CEACs are an alternative to confidence intervals around ICERs. We constructed these to show the probability of the system of care being cost-effective compared to usual care for a range of values that a decision maker may be willing to pay for an additional unit gain in the GHQ-12 or QALYs. CEACs were based on the net benefit approach. Net benefits provide a single summary monetary measure of costs and outcomes for each individual (removing the need to examine ICERs which carry limitations of being based on point estimates and difficulties in estimating confidence intervals around them). Net benefits account for the value (λ) that a decision maker would be willing to pay for a greater net benefit, and are calculated as follows: net benefit = (λ x outcome) – cost. For each cost-outcome combination, we calculated a series of net benefits for a range of relevant λ values (£0 to £2000 per point improvement on the GHQ-12 and £0 to £50,000 per QALY gain). Net benefits were then compared by randomisation group using bootstrapped regressions (5000 replications) of study group upon net benefit, with an adjustment for cluster and the same patient level baseline covariates as used for other group comparisons: Barthel Index (pre- and post- stroke), gender, age, living circumstances (living alone vs. with carer), stroke severity as represented by speech and language impairment (normal/ impaired) and 6CIT score (normal/ impaired), utility score, GHQ-12 score and total cost from the relevant perspective. For each value of λ, the proportion of iterations indicating a higher net benefit for the intervention group were calculated and plotted as a CEAC.

30   

Statistical analyses All cost and QALY data are reported as mean values with standard deviations. To accommodate a cluster randomisation design, differences in costs and QALYs between groups were tested by multi level modelling using the xtreg procedure in Stata 10.1, from which we report 95% confidence intervals and p values for the differences in means. Relevant baseline patient characteristics were included as covariates for comparisons at follow-ups: Barthel Index (pre- and post- stroke), gender, age, living circumstances (living alone vs. with carer), stroke severity as represented by speech and language impairment (normal/ impaired) and 6CIT score (normal/ impaired), utility and GHQ-12 score (and total cost from the relevant cost perspective for the comparison of costs). Individuals were analysed according to the group to which they were randomised regardless of compliance with the intervention. Resource use differences were not compared statistically to avoid problems associated multiple testing. Missing data The base case evaluation was a completers’ analysis without imputation for loss of follow-up under the assumption that loss of follow-up was at random. Missing GHQ-12 and EQ-5D data were not imputed. There were inevitably some missing items on returned resource use questionnaires and to allow computation of total costs that reflected variations in resource use rather than variations in data completeness, we imputed missing cost items for returned questionnaires. We imputed a zero cost where there was no indication of a particular resource being used. Where there was indication of use but quantity was missing, we imputed using the mean cost for participants with data for that item at the same assessment point and in the same randomisation group (or the other randomisation group if there were no valid cases to impute from in the same group); in the case of hospital admissions, we used an average admission cost from NHS reference costs, rather than values from within the dataset. All such imputations were made to cost estimates, rather than to the resource use data, so descriptions of resource use data include no imputations. It was similarly necessary to impute missing data related to SCC inputs; we did this for each SCC component separately to utilise, rather than over-ride, any available data. We assumed a zero cost if a care plan/time log was not completed for the following reasons: referral not received by SCC service, patient declined SCC service, patient died prior to SCC service, SCC unable to contact patient or patient withdrew from all follow-up. For partially completed care plans/time logs, missing duration of inputs were imputed using values from the SCC survey where available (by service) or else the within-group mean cost for those with care plans/time logs and data for the relevant component. Where SCC services were received but a care plan/time log was not completed, we imputed the mean cost for those with data in the control group (the care plan was part of the intervention so providing SCC inputs without it was assumed equivalent to control services). For consenting carers in the intervention group, we assumed a zero cost if an assessment did not take place, if it was unknown whether one took place or if there was no consenting carer. If an assessment took place but duration was missing, we allocated the mean cost for those with completed assessments and data for that component. Data for inputs to non-consenting carers were found to be complete across all components. We assumed a zero cost in the absence of any identified non-consenting carer. Sensitivity analyses We examined the effect of loss of follow-up by imputing missing health and social care costs and QALYs at the primary endpoint, 6 months, and checking whether conclusions related to 31   

group differences agreed with the base case analyses (using the same covariates for estimation of mean differences). We re-computed alternative ICERs and CEACs only if different cost or outcome conclusions were suggested. For those lost to follow-up due to death, we computed QALY gains assuming a utility value of zero at 6 months and included costs of SCC inputs whilst assuming all other health and social care costs were zero. For those lost to follow-up for any other reason, we imputed total costs and QALYs using the multiple imputation procedure in Stata 10.1. Imputations were based on key baseline variables expected to predict follow-up costs and QALYs. These were the same variables used as covariates for the comparisons of costs and QALYs for those with data, plus randomisation group: at the patient level, baseline Barthel Index (pre- and post- stroke), gender, age, living circumstances (living alone vs. with carer), stroke severity as represented by speech and language impairment (normal/ impaired) and 6CIT score (normal/ impaired), utility, GHQ12 (and total cost from the relevant cost perspective for the prediction of costs); at the stroke unit level, quality of stroke unit (NSA score), referral rate and SCCs working alone vs. within a community MDT. Economic evaluation results Completion rates for resource use questionnaire and EQ-5D Completion rates for the resource use questionnaire and EQ-5D were similar at each assessment point and balanced between the two groups. Although differences were not explored statistically, baseline characteristics of patients with the necessary data at 6 months appeared similar to those of the full sample. Therefore, the completers’ analysis is likely to be representative of the full sample. Resource use Resource use was broadly comparable between the two groups at each assessment (Tables XI to XIII). For brevity, we report all inpatient services and all informal care plus other health and social care resources used by at least 10% of responders in either trial arm at the relevant assessment; full resource use data are available upon request from Anita Patel. Economic evaluation references 1. Patel A, Knapp M, Evans A, Perez I, Kalra L. Training care givers of stroke patients: economic evaluation. BMJ 2004; 328: 1102–1104. 2. Patel A, Knapp M, Perez I, Evans A, Kalra L. Alternative strategies for stroke care: cost– effectiveness and cost–utility analyses from a prospective randomized controlled trial. Stroke 2004; 35: 196–203. 3. OECD. StatsExtracts 4. PPPs and exchange rates. (Accessed 22 December 2014 at http://stats.oecd.org/Index.aspx?DataSetCode=SNA_TABLE4) 4. Dolan P, Gudex C, Kind P, Williams A. A social tariff for EuroQol: results from a UK population survey: University of York. 1995. 5. The EuroQol Group. EuroQol: a new facility for the measurement of health related quality of life. Health Policy 1990; 16: 199–208.

32   

Table XI: Resource use at baseline (for previous 3 months)c Index stroke admissionb

Unit bed days

% users 99

Intervention = 401 Meana 39

SD 44

99

Control = 398 Meana 30

SD 35

Inpatient services

bed day

7

11

24

10

7

6

Accident & emergency

visit

7

11

24

10

7

6

Outpatient services

visit

8

2

4

10

1

1

visit visit call occurrence visit contact contact contact

49 9 10 47 26 10 12 12

2 2 2 2 2 2 1 1

1 1 1 1 1 1 1