Care home managers will be approached and provided with information about the study. They will be given a full oral explanation of the study, a leaflet describing OT and study information sheets. Care home managers will be asked to give written agreement for their home to participate in the study. Following this, residents will be recruited individually.

Care home managers will assist to identify potential participants. Participants must be resident in a care home and have a history of stroke or transient ischaemic attack (TIA). All efforts will be made to include participants with communication and cognitive impairments, because this will more accurately reflect the population characteristics. Participants will be excluded from the study if they are receiving end-of-life care (with life expectancy < 6 months). Care home members of staff will search residents’ notes to determine a diagnosis of stroke or TIA. If required, we will confirm this diagnosis with general practice records. When a potential participant is identified as being eligible for the study, they and (where appropriate) their family (outlined in the UK’s Mental Capacity Act 2005 15) will be approached by a senior member of the care home staff, a research network nurse or therapist, general practitioner or geriatrician, and will be asked if they are interested in participating in a research study. Prospective participants (and their family) will be given a full explanation of the study. This will include discussion of the treatment options in the trial and the manner of treatment allocation. Potential participants and their family will be given a participant information sheet, or a consultee information sheet and the UK Clinical Research Network’s 2007 16 publication on ‘Understanding Clinical Trials’ to read. They will be given sufficient time to decide (at least 24 hours) whether they would like to join the study. This may take a few days if relatives only visit at weekends for example. Residents will then be asked to sign the consent form. If the resident is considered to be incapacitated, according to the Mental Capacity Act 2005 guidelines 15, a family member (consultee) will be approached for consent. The participant’s general practitioner will be informed, if the patient or consultee consents, in writing of their participation in the trial. Research nurses from NHS research networks will assist in the recruitment and consent processes.

Information recorded by a member of staff from the care home about the resident will include age, gender, current medication intake and date, type and side of stroke. Furthermore, at baseline a trained assessor will administer the Sheffield Screening Test for Acquired Language Disorders, to assess receptive/expressive aphasia 23, and the mini mental state examination to assess cognitive function 24. Finally, data on resource use will be recorded through logs and collected from health and social care services to conduct a health economic evaluation for both interventions delivered during this study.

The primary outcome measure is the Barthel Activity of Daily Living Index 25, which is a commonly used measure of independent self-care in people who survived a stroke 2627. Furthermore, the Barthel index assesses specific aspects of self care targeted by the therapy. A change of two points on this scale is widely accepted as being clinically meaningful 28. The Barthel index can be completed by the resident or with assistance from a member of staff.

Secondary outcome measures will assess mobility, mood, safety, quality of life and costs. Mobility will be assessed with the Rivermead mobility index 29, which is a 15-item measurement of functional mobility. The geriatric depression scale 30 will be administered to measure mood. The full 30-item version will be used as standard, but in cases where residents are unable to self complete or follow and interview the informant version will be filled out. Quality of life is measured by the EQ-5D 31, which is a well established measure for evaluating patients preference. Costs will be estimated based on resource usage log and weighted by unit costs. This will include visits to hospital and include in- and out-patient appointments, allied health professionals time, general practitioner visits, provision of care in a care home, adaptive equipment, minor environmental adaptations and staff education. Unit costs will be obtained from the national health and social care services reference costs and personal social services research unit 32. Additionally, any adverse events will be recorded in participant logs and will include a fall or equipment failure leading to an injury requiring a visit to a hospital or general practitioner.

An overview of the assessment schedule is given in Table 1. Baseline assessments will be conducted prior to randomisation to reduce possible recruitment bias repetition 33. The primary measurement point will be at 3 months after randomisation and follow-up assessments will be conducted at 6 and 12 months prior to randomisation. Data will be collected from participants where they are currently residing. If a participant moves to another home between assessments, then the assessors will attempt to collect follow-up data.

*Full name of the test is Sheffield Screening Test for Acquired Language Disorders.

The primary outcome is the mean Barthel index score at the 3 month follow-up. A mixed model analysis will be used to compare Barthel index scores between the intervention and control groups. The primary analysis will be adjusted by care home (as a random factor), baseline Barthel index score, and stratification factors: centre and type of care home (nursing, residential). Participants that die before their follow-up date will be given a Barthel index score of zero at all subsequent follow-ups. In addition, participants will be categorised into 3 outcome groups based on an individual’s change in Barthel index score at 3 months from baseline (below 0 or death = ‘poor’, 0 to 1=’moderate’, 2 and above=’good’). A non-linear mixed effects model will be used to compare this ordinal outcome between the groups. Adjustments will be made for care home (random), centre and type of care home (fixed).

To identify any longer term effects, a repeated measures mixed model analysis of the primary outcome will be undertaken, comparing groups across all time points. The analysis will include adjustment for care home (as a random factor), baseline Barthel index score, and stratification factors: centre and type of care home. Similar analyses will be performed for mobility, mood and quality of life.

The number of falls will be compared using a Poisson or negative binomial model, with adjustments for care home, centre and type of care home as previously described. A sensitivity analysis of the outcomes will be performed to examine the potential effect of missing data. This will include best case, worst case and multiple imputation methods.

Analysis will be by intention to treat, whereby residents will be analysed according to the intervention to which they are randomised, regardless of whether they comply with the treatment. All participants irrespective of when they die, withdraw or are lost will be included in the arm to which they were randomised, including those that die after consent but prior to randomisation. Those participants that move home will be analysed by the home they were originally randomised to. Analyses will be performed using Stata (StataCorp,. Texas, USA) and SAS software (SAS Institute Inc., Cary, NC).

Outcomes from health economic evaluation analyses will establish cost effectiveness of a treatment expressed as incremental cost per quality-adjusted life-years (QALYs) gained. Generalised linear models will be used to investigate differences between interventions for QALYs and costs. Bootstrap methods will be adopted to produce incremental cost effectiveness ratio and associated confidence intervals. Cost-effectiveness acceptability curves will be generated to reflect the probability that a treatment will be cost effective given a society’s willingness to pay in terms of price per QALY gained.

Favourable ethical opinion for this study was obtained from the National Research Ethics Service, Coventry Research Ethics Committee (Reference: 09/H1210/88). The study was granted an International Standard Randomised Controlled Trial Number (00757750).

A trial steering group has been established to monitor the governance of the study consisting of the main research team, an independent chair, a geriatrician, an occupational therapist, a physiotherapist with expertise in rehabilitation research a patient representative, and a representative from the National Institute for Health Research. A data monitoring committee has been established with an independent geriatrician as chair, a statistician and an occupational therapist. During the development of this protocol, a patient representative was consulted.

Service user (and carer) involvement will be incorporated at all levels of this study. Service user (and carer) involvement will not be a ‘stand-alone’ activity but rather an integral part of all aspects of this study. Service users (and carers) will be directly involved as research ‘partners’ and not just as ‘data providers’. All support for service users (and carers) involvement will be provided by the Stroke Research Network members who have expertise in training and supporting service users (and carers) for involvement in National Health Service research, service evaluation and development.