Training carers of stroke patients: randomised controlled trial

Subjects

We undertook the study in stroke patients admitted to a stroke rehabilitation unit and their care givers. A care giver was defined as the main person (other than health, social, or voluntary care provider) helping with activities of daily living and advocating on behalf of the patient. Patients were included if they were independent in daily living activities before the stroke, medically and neurologically stable at the time of baseline assessments, and expected to return home with residual disability (defined as the need for supervision or physical assistance for core activities of daily living). Care givers were required to have no notable disability (defined as a Rankin score of 0-2) and be willing and able to provide support after discharge. We obtained informed consent from patients and care givers.

Baseline assessments and randomisation procedures

Researchers undertook baseline assessments before randomisation. Patients' assessments included demographics, stroke subtype,19Barthel index,20and estimations of premorbid function and quality of life; we used the Frenchay activity index20and the EuroQol visual analogue scale.21Caregiver assessments included details of demographics and accommodation, health profile, functional status, and quality of life assessments.

We used block randomisation procedures; each block included 10 subjects. We used computer generated random numbers to prepare the allocation schedule in advance. Allocation codes were held in a central office remote from the study environment. After baseline assessment the responsible assessor telephoned the randomisation office with patients' identification details only. A clerical worker entered these details on a computer database in strict referral order and was given patient allocation, which was relayed to the assessor.

Interventions

All patients were managed on a stroke rehabilitation unit with established multidisciplinary practice and received conventional care in accordance with existing guidelines.

Conventional care

This consisted of

• Information on stroke and its consequences, prevention, and management options

• Involvement in goal setting for rehabilitation and discharge planning

• Encouragement to attend nursing and therapy activities to learn about patients' abilities and informal instruction on facilitating transfers, mobility, and activities of daily living tasks, and

• Advice on community services, benefits, and allowances, including contact information for voluntary support services for care givers.

Care giver training

In addition to conventional support, care givers allocated to caregiver training received

• Instruction by appropriate professionals on common stroke related problems and their prevention, management of pressure areas and prevention of bed sores, continence, nutrition, positioning, gait facilitation, and advice on benefits and local services, and

• “Hands-on” training in lifting and handling techniques, facilitation of mobility and transfers, continence, assistance with personal activities of daily living and communication, tailored to the needs of individual patients.

Training started when patients' rehabilitation needs had stabilised and discharge was contemplated. Care givers received three to five sessions depending on need; each session lasted 30-45 minutes. We assessed care givers' competencies at the end of training. In addition the hospital team conducted a “follow through” session at home to adapt skills learnt to the home environment.

Assessment of outcome

An observer who did not participate in allocation or management of patients assessed outcome at three and 12 months after stroke onset. Most assessments were undertaken at home; patients and care givers were assessed separately wherever possible, and participants were encouraged to complete questionnaires themselves without prompts. Care givers were allowed to help patients who were unable to complete questionnaires without help.

Outcome measures for patients included death or institutionalisation, modified Rankin scale, Barthel index and Frenchay activities index for function, hospital anxiety and depression scale20for mood, and EuroQol visual analogue scale for quality of life. We used the Frenchay activities index for function and social activities, hospital anxiety and depression scale and caregiver burden scale22for emotional health, and EuroQol visual analogue scale for quality of life to measure outcomes for care givers. The modified Rankin score at one year was dichotomised to good outcome (score 0-2) and poor outcome (3-6). The Barthel index was similarly dichotomised; scores of > 18 were considered good. Patients and care givers were asked to complete a satisfaction questionnaire at 12 months.

Sample size calculation

The primary outcome measure for the study was the cost to health and social services during the first year of stroke. Sample size calculations were based on data indicating the cost of stroke care from the surveys of disability of the Office for Population Censuses and Surveys.23The mean weekly cost for support at home after discharge varied between £89.15 (SD £11.15) ($158.99; € 133.78) and £68.37 (SD £7.70), depending on the extent of caregiver support.23The inclusion of 300 subjects would give the study 80% power to detect a 30% reduction in cost at the 5% significance level for unpaired comparisons. This sample size could also detect differences of 1 point in the Barthel index, 1.5 points in the Frenchay activities index, 1.5 points in the hospital anxiety and depression scale, 2.5 points in the EuroQol visual analogue scale, and 1 point in the caregiver burden scale.

Data analysis and statistical methods

We analysed the data on an intention to treat basis. For patients who had died or were lost to follow up we used the last completed datasets in further analyses. We calculated means, standard deviations, medians, and interquartile ranges as appropriate. We used the Mann-Whitney test to compare differences in medians of age, length of stay, therapy input, Barthel index, Frenchay activities index, and EuroQol. We used theχ²test for discrete variables (for example, stroke subtype, mortality, institutionalisation, and the dichotomised modified Rankin scale and Barthel index). We used prospective and retrospective techniques described in the accompanying paper to collect data on use of resources from health and social services over one year.18We report all costs as mean values, and we used the ttest to undertake comparisons.

We undertook sensitivity analyses to examine the effects if all the missing values for carer burden and quality of life assessments were at the lower limit (worse outcome) and the higher limit (better outcome) of the interquartile range. Seventy five patients in each group received caregiver support consisting of a family worker who provided information on services, link with social agencies, advocacy, and emotional support for up to six months after discharge. We used regression analyses to assess the confounding effect of this and other baseline variables (such as age or extent of disability) on outcome.

Limitations of the study

The findings of the study need to be interpreted in the context of its limitations. It is not possible to blind observers in such trials fully because interactions during assessment with patients or care givers, who are aware of the training received, may disclose allocation. We kept observer bias to a minimum by using observers not involved with allocation, interventions, or patients' care. In addition we used robust, well validated, and objective outcome measures (Rankin scale, Barthel index, Frenchay activities index) and encouraged self completion of various subjective assessments (caregiver burden scale, hospital anxiety and depression scale, EuroQol) without prompts. This may have contributed to missing data on some assessments (tables 2and 3), which is another potential source of bias. Sensitivity analyses to examine the effects of missing data showed that findings remained robust for worse or better outcome assumptions.

Generalisability of results

The study was undertaken in a largely middle class suburban area, and an interaction between the acceptability of interventions and education, affluence, living accommodation, informal support, and cultural beliefs is possible. Patients and care givers who chose to participate were likely to be fit, motivated, and enthusiastic, and hence most likely to benefit from these interventions. The generalisability of these interventions to other settings where the structures of the population, health and social care, and the economy are different, needs to established.

Outlook

It is common practice to involve care givers in setting goals and planning discharge in stroke rehabilitation. The time may be ripe to use a structured programme of activities under professional supervision during inpatient rehabilitation, to empower consenting informal care givers in their future role by teaching them appropriate skills.

The CONSORT table is on bmj.com