Information provision for stroke patients and their caregivers

Anne Forster; Lesley Brown; Jane Smith; Allan House; Peter Knapp; John J Wright; John Young

doi:10.1002/14651858.CD001919.pub3

. 2012 Nov 14;2012(11):CD001919. doi: 10.1002/14651858.CD001919.pub3

Information provision for stroke patients and their caregivers

Anne Forster ^1,^✉, Lesley Brown ², Jane Smith ², Allan House ³, Peter Knapp ⁴, John J Wright ⁵, John Young ⁶

Editor: Cochrane Stroke Group

PMCID: PMC6544775 PMID: 23152210

Abstract

Background

Research shows that stroke patients and their families are dissatisfied with the information provided and have a poor understanding of stroke and associated issues.

Objectives

To assess the effectiveness of information provision strategies in improving the outcome for stroke patients or their identified caregivers, or both.

Search methods

For this update we searched the Cochrane Stroke Group Trials Register (June 2012), the Cochrane Central Register of Controlled trials (CENTRAL), the Cochrane Database of Systematic Reviews (CDSR), the Database of Abstracts of Reviews of Effects (DARE), the NHS Economic Evaluation Database (EED), and the Health Technology Assessment (HTA) Database (The Cochrane Library June, 2012), MEDLINE (1966 to June 2012), EMBASE (1980 to June 2012), CINAHL (1982 to June 2012) and PsycINFO (1974 to June 2012). We also searched ongoing trials registers, scanned bibliographies of relevant articles and books and contacted researchers.

Selection criteria

Randomised trials involving patients or carers of patients with a clinical diagnosis of stroke or transient ischaemic attack (TIA) where an information intervention was compared with standard care, or where information and another therapy were compared with the other therapy alone.

Data collection and analysis

Two review authors independently assessed trial eligibility and methodological quality and extracted data. Primary outcomes were knowledge about stroke and stroke services, and impact on mood.

Main results

We have added four new trials to this update. This review now includes 21 trials involving 2289 patient and 1290 carer participants. Nine trials evaluated a passive and 12 trials an active information intervention. Meta‐analyses showed a significant effect in favour of the intervention on patient knowledge (standardised mean difference (SMD) 0.29, 95% confidence interval (CI) 0.12 to 0.46, P < 0.001), carer knowledge (SMD 0.74, 95% CI 0.06 to 1.43, P = 0.03), one aspect of patient satisfaction (odds ratio (OR) 2.07, 95% CI 1.33 to 3.23, P = 0.001), and patient depression scores (mean difference (MD) ‐0.52, 95% CI ‐0.93 to ‐0.10, P = 0.01). There was no significant effect (P > 0.05) on number of cases of anxiety or depression in patients, carer mood or satisfaction, or death. Qualitative analyses found no strong evidence of an effect on other outcomes. Post‐hoc subgroup analyses showed that active information had a significantly greater effect than passive information on patient mood but not on other outcomes.

Authors' conclusions

There is evidence that information improves patient and carer knowledge of stroke, aspects of patient satisfaction, and reduces patient depression scores. However, the reduction in depression scores was small and may not be clinically significant. Although the best way to provide information is still unclear there is some evidence that strategies that actively involve patients and carers and include planned follow‐up for clarification and reinforcement have a greater effect on patient mood.

Plain language summary

Information provision for stroke patients and their caregivers

Studies have shown that stroke survivors and their carers often report they have not been given enough information about stroke and feel unprepared for life after discharge from hospital. However, the best way to provide information after stroke is unclear. The authors of this review looked at the evidence for the effectiveness of providing information to patients, or carers of patients, who have had a stroke or transient ischaemic attack (TIA), sometimes called a mini‐stroke. They examined randomised trials (studies) in which one group of stroke patients or carers who were given the intervention being tested (such as a course of lectures) was compared with a group of stroke patients or carers who received standard care. Twenty‐one studies, involving 2289 patients and 1290 carers, are now included in this updated review. Overall, the studies showed that providing information to patients and carers improved their knowledge of stroke and increased patient satisfaction with some, but not all, of the information they received about stroke. There was also an effect on reducing patient depression, although the reduction was small and may not be enough to seem meaningful to patients. When information was provided in a way that more actively involved patients and carers, for example by offering repeated opportunities to ask questions, it had more effect on patient mood than information which was given on one occasion only. There is not much evidence that providing information had effects on other aspects of patient or carer stroke recovery such as independence or social activities.

Background

Every year approximately 110,000 people in England have a stroke (National Audit Office 2005) and at any one time over 300,000 people are living with moderate to severe disability as a result of a stroke (Adamson 2004). The provision of appropriate, accurate, timely information and advice about stroke has been recommended as a key component of service provision (Canadian Stroke Network 2006; RCP 2008; National Stroke Foundation Australia 2010). Information, combined with the right support, is the key to better care, better outcomes and reduced costs. Patients should have information and data on all aspects of health care, to enable them to share in decisions about their care and access appropriate services (Department of Health 2010). The information needs of people who have had a stroke and their carers are diverse and change over time. Information should be tailored to an individual's requirements and provided in a variety of formats (Department of Health 2007; Eames 2011), taking into account their stroke‐specific impairment and personal situation (RCP 2008).

There is a wide range of nationally and locally produced leaflets, booklets, videos and audio tapes available for patients and carers. However, despite this emphasis on giving information, research suggests that patients' understanding of stroke, its consequences and the support available, remains poor. A recent systematic review identified multiple and diverse unmet educational needs by stroke patients and their caregivers (Hafsteinsdottir 2011). In a survey of community dwelling adults who suffered a stroke at least one and up to five years previously, over half reported wanting more information about their stroke (McKevitt 2011).

In a UK study, carers of stroke patients reported that whilst leaflets were available, they were not always appropriate to the situation (Mackenzie 2007). A survey by primary care trusts in England, of the information provided to patients after stroke, reported that the majority provided good information. However, only 40% contained information relevant to local services. Furthermore, the size, content and organisation of the information varied extensively (Care Quality Commission 2011). Inadequate provision and receipt of appropriate information has important consequences for compliance with secondary prevention and the longer‐term psycho‐social outcome for patients and carers (O'Mahoney 1997). Enhanced knowledge of stroke care by carers may improve the quality of discharge home from hospital for stroke patients (Evans 1991). Despite the perceived and expressed need for information, successful strategies have not as yet been identified. In order to fully explore available evidence we have undertaken a systematic review of information provision for patients and their carers after stroke.

Description of the condition

A stroke is defined by the World Health Organization as: "Rapidly developing clinical signs of focal (or global) disturbance of cerebral function, lasting more than 24 hours or leading to death, with no apparent cause other than vascular origin" (Aho 1980). A transient ischaemic attack (TIA) is a brief reversible episode of focal, non‐convulsive ischaemic dysfunction of the brain with a duration of less than 24 hours (Adams 1998). Stroke can lead to death or physical and cognitive impairment (McKevitt 2011; Mukherjee 2011) and can have long lasting psychological and social implications (Knapp 2000).

Description of the intervention

The intervention is the provision of information for stroke survivors or their informal caregivers, or both, following a stroke or TIA. The intervention may be provided in a variety of formats such as leaflets, workbooks, or verbal communication including lectures or teaching sessions. Whilst the content of the intervention may vary, it is likely to contain at least one of the following components: information about the causes and nature of stroke; management and recovery from the effects of stroke; prevention or reducing the risk of future strokes; information on resources or services. Whilst the provision of information should be incorporated as standard practice following stroke, evidence suggests it is lacking or inconsistent (Mackenzie 2007; Care Quality Commission 2011).

How the intervention might work

If stroke survivors and carers are to be active in their decision making and management of the long‐term effects of stroke, appropriate information delivered in a timely and effective format is necessary. Information is considered necessary to recognise and act upon symptoms, manage disease exacerbation and to access effective treatments and medicines and produce better outcomes (Department of Health 2001; Department of Health 2010). Furthermore, inadequate provision of information has implications for compliance with secondary prevention and psycho‐social outcomes for stroke patients and carers (O'Mahoney 1997). Evidence from non‐stroke populations suggests providing written information improves adherence to hospital after‐care regimens (Gibbs 1989; Firth 1991) and may assist with self‐care (Coulter 1998), which may indirectly produce beneficial outcomes.

Why it is important to do this review

It has been proposed that information, combined with the right support, is the key to better care, better outcomes and reduced costs (Department of Health 2010). The information derived from this review has the potential to lead to the development of more effective information provision strategies and highlight which outcomes might be affected by such interventions.

Objectives

The objective of this review was to examine the effectiveness of information strategies provided with the intention of improving the outcome for stroke patients or their identified caregivers or both.

Methods

Criteria for considering studies for this review

Types of studies

We included unconfounded randomised trials where an information intervention was compared with standard care or where information and another therapy was compared with the other therapy alone.

Types of participants

Patients with a clinical diagnosis of stroke or TIA and their identified caregivers or both.

Types of interventions

Information provided with the intention of improving the outcome of patients or their caregivers or both. We excluded trials in which information‐giving was only one component of a more complex rehabilitation intervention, for example family support worker trials (Forster 1996; Dennis 1997; Mant 2000; Lincoln 2003; Ellis 2005), which are the subject of a separate Cochrane review (Ellis 2010).

Types of outcome measures

We considered that information provision would impact most directly on knowledge and patients' or carers' mood state (anxiety and depression) or both. Therefore, we used the following primary and secondary outcome measures to assess the effectiveness of information provision.

Primary outcomes

Patient or carer knowledge about stroke and stroke services or both.
Patient or carer impact on mood (e.g. Hospital Anxiety and Depression Scale).

Secondary outcomes

Activities of daily living (e.g. Barthel Index).
Participation (e.g. London Handicap Scale).
Social activities (e.g. Frenchay Activities Index).
Perceived health status (e.g. Short‐form 36, Nottingham Health Profile).
Quality of life (e.g. Dartmouth Coop Chart).
Satisfaction with information.
Hospital admissions, service contacts or health professional contacts.
Compliance with treatment/rehabilitation (e.g. Miller's Health Behaviour Scale).
Death or institutionalisation or both.

Resource outcomes

Cost to health and social services.

Search methods for identification of studies

See the 'Specialized register' section in the Cochrane Stroke Group module. We searched for trials in all languages and arranged translation of papers published in languages other than English.

Electronic searches

For this update we searched the Cochrane Stroke Group Trials Register (last searched in June 2012). In addition, we searched the following electronic bibliographic databases and trials registers:

Cochrane Central Register of Controlled trials (CENTRAL) (The Cochrane Library 2012, Issue 4) (Appendix 1);
Cochrane Database of Systematic Reviews (CDSR) (The Cochrane Library 2012, Issue 4) (Appendix 1);
Database of Abstracts of Reviews of Effects (DARE) (The Cochrane Library 2012, Issue 4) (Appendix 1);
NHS Economic Evaluation Database (EED) (The Cochrane Library 2012, Issue 4) (Appendix 1);
Health Technology Assessment (HTA) Database (The Cochrane Library 2012, Issue 4) (Appendix 1);
MEDLINE (1966 to June 2012) (Appendix 2);
EMBASE (1980 to June 2012) (Appendix 3);
CINAHL (1982 to June 2012) (Appendix 4);
PsycINFO (1974 to June 2012) (Appendix 5);
Current Controlled Trials (http://www.controlled‐trials.com/) (June 2012);
National Rehabilitation Information Center (www.naric.com) (June 2012);
RePORT Expenditures and Results (RePORTER) query tool (http://projectreporter.nih.gov/reporter.cfm) (December 2012);
Internet Stroke Center stroke trials registry (www.strokecenter.org) (June 2012).

Searching other resources

In an effort to identify further published, unpublished and ongoing trials, we searched bibliographies of relevant articles and books and contacted authors of relevant research and previous articles on information provision.

For the previous version of the review we searched:

Science Citation Index and Social Science Citation Index (1981 to March 2007);
ASSIA (Applied Social Science Index and Abstracts) (1987 to March 2007);
Index to UK theses (1970 to March 2007);
Dissertation Abstracts (1961 to March 2007);
National Research Register (www.nrr.nhs.uk) (to September 2007);
Journal of Advanced Nursing (1996 to March 2007).

Data collection and analysis

Selection of studies

Two review authors independently assessed the titles and abstracts of records from the electronic searches and excluded obviously irrelevant studies. We obtained the full text of the remaining studies and at least two review authors assessed these against the review inclusion criteria to determine which trials would be eligible for inclusion. The review authors resolved disagreements by discussion with other members of the review team.

Data extraction and management

At least two review authors scrutinised all the eligible trials to grade methodological quality, patient selection, the intervention, outcome measures used, and length of follow‐up. We allocated studies to one of two categories ‐ passive information or active information ‐ according to the nature of the intervention. An intervention was classified as passive if the information was provided on a single occasion and there was no subsequent systematic follow‐up or reinforcement procedure. An intervention was classified as active if, following the provision of the information, there was a purposeful attempt to allow the participant to assimilate the information and a subsequent agreed plan for clarification and consolidation or reinforcement. We made this classification because it would inform future research and be helpful for service planners in terms of committing resources. Two review authors extracted data independently using piloted data extraction forms, and measured agreement. They resolved disagreement through group consensus. Where necessary, we contacted study authors for additional information and data.

Assessment of risk of bias in included studies

We assessed the methodological quality of selected studies using the tool for assessing risk of bias as described in section 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We scored each of the following domains as 'high risk of bias', 'low risk of bias', or 'unclear risk of bias' and reported them in the 'Risk of bias' tables.

Random sequence generation (selection bias).
Allocation concealment (selection bias).
Blinding of participants and personnel (performance bias).
Blinding of outcome assessment (detection bias).
Incomplete outcome data (attrition bias).
Selective reporting (reporting bias).
Other possible bias.

Measures of treatment effect

We compared studies based on end‐of‐study results. We used the mean difference (MD) or standardised mean difference (SMD) for continuous outcomes. We treated ordinal data as continuous data and combined them using the MD. We combined dichotomous data using the Peto odds ratio (OR).

Dealing with missing data

If data were missing, we performed an available case analysis. The proportion of participants in each study arm who did not provide data is shown in the 'Data and analyses' section.

Assessment of heterogeneity

We tested for the presence of heterogeneity between the trials using the I² statistic. We used a fixed‐effect model if we detected no substantial heterogeneity (I² < 50%). Where there was substantial heterogeneity (I² ≥ 50%) we used a random‐effects model.

Assessment of reporting biases

We were able to reduce reporting bias by undertaking comprehensive searches of multiple databases and trials registers, and contacting authors. There were no restrictions based on language and translations were undertaken if required. It was not possible to detect reporting bias by the method of assessment of funnel plots as there were insufficient studies included in the meta‐analyses.

Data synthesis

We compared studies based on the end‐of‐study results. Meta‐analyses have been undertaken for the domains of knowledge, emotional outcome, death and for selected satisfaction questions. For the domain of knowledge, we combined data using the SMD as all the trials had used different knowledge questionnaires. We combined dichotomous data (domains of mood, satisfaction, death) using the Peto OR. For the domain of mood, we dichotomised the Hospital Anxiety and Depression Scale scores, the Geriatric Depression Scale scores and the General Health Questionnaire scores using the recommended cut‐off points (Zigmond 1983; Sheikh 1986; Goldberg 1988). We treated ordinal data (domain of patient mood) as continuous data and combined them using the MD. For a number of studies both ordinal and dichotomised patient Hospital Anxiety and Depression Scale data were available. If this was the case, we extracted and analysed both forms of data. For other outcomes, we present a narrative summary stratified by subgroup and a summary of the data is provided in the Data and analyses section.

Subgroup analysis and investigation of heterogeneity

We undertook post‐hoc subgroup analyses for the type of intervention (passive and active). We used the method described by Deeks et al (Deeks 2001) to compare the magnitude of treatment effect of the two subgroups.

Results

Description of studies

Results of the search

For this update we reviewed 28,110 titles; 134 papers were reviewed of which duplicate papers were identified for 14 studies, four new studies are included in the review (Johnston 2007; Chiu 2008; O'Connell 2009; Chinchai 2010). Of the 134 papers reviewed for this update, five were commentaries, reviews or meta‐analyses, seven trials included non‐stroke participants and six studies did not investigate the effectiveness of information provision after stroke; 43 studies have been added to the excluded studies section (details reported in Characteristics of excluded studies); 14 studies are currently pending assessment and 10 studies are currently ongoing.

Included studies

The current analysis includes 21 completed trials with 2289 patient and 1290 carer participants (Lomer 1987; Evans 1988; Pain 1990; Downes 1993; Banet 1997; Mant 1998; Rodgers 1999; Frank 2000; Johnson 2000; Kalra 2004; Smith 2004; Ellis 2005; Larson 2005; Draper 2007; Hoffmann 2007; Johnston 2007; Lowe 2007; Maasland 2007; Chiu 2008; O'Connell 2009; Chinchai 2010).

Setting

Three of the included trials were conducted in the USA (Evans 1988; Banet 1997; Johnson 2000), 11 in the UK (Lomer 1987; Pain 1990; Downes 1993; Mant 1998; Rodgers 1999; Frank 2000; Kalra 2004; Smith 2004; Ellis 2005; Johnston 2007; Lowe 2007), three in Australia (Draper 2007; Hoffmann 2007; O'Connell 2009), one in Sweden (Larson 2005), one in the Netherlands (Maasland 2007), one in Taiwan (Chiu 2008) and one in Thailand (Chinchai 2010).

Participants

In 19 trials the majority of patients were at least 60 years old (Evans 1988; Pain 1990; Downes 1993; Mant 1998; Rodgers 1999; Frank 2000; Johnson 2000; Kalra 2004; Smith 2004; Ellis 2005; Larson 2005; Draper 2007; Hoffmann 2007; Johnston 2007; Lowe 2007; Maasland 2007; Chiu 2008; O'Connell 2009; Chinchai 2010). Two trials did not report age (Lomer 1987; Banet 1997). Six trials reported carer age (Evans 1988; Downes 1993; Rodgers 1999; Smith 2004; Larson 2005; Draper 2007). Carers were younger than the patients, notably so in Evans 1988 where the mean age of the carers was under 50 years old. This study was carried out at a Veterans Administration Medical Centre and was also exceptional in that 94% of the stroke patients were male. In Larson 2005 the majority of spouses were female. Ten trials were concerned with the patient only (Pain 1990; Banet 1997; Frank 2000; Johnson 2000; Ellis 2005; Hoffmann 2007; Lowe 2007; Maasland 2007; Chiu 2008; O'Connell 2009) and in four trials the intervention involved the carer or spouse only (Evans 1988; Kalra 2004; Larson 2005; Draper 2007). In the remaining trials the focus of the intervention was the patient and carer (Lomer 1987; Downes 1993; Mant 1998; Rodgers 1999; Smith 2004; Johnston 2007; Chinchai 2010).

Interventions

Two of the included trials evaluated two interventions (Evans 1988; Downes 1993): one evaluated education and counselling (Evans 1988) and the other evaluated information provision plus counselling (Downes 1993). Only the data from the information/education group and the control groups have been analysed in this review.

Content and administration

Nine trials evaluated a passive intervention (Lomer 1987; Pain 1990; Downes 1993; Banet 1997; Mant 1998; Hoffmann 2007; Lowe 2007; Maasland 2007; O'Connell 2009). In three trials (Lomer 1987; Downes 1993; Mant 1998) this comprised written generic information about stroke in the form of booklets and leaflets. In five studies (Pain 1990; Banet 1997; Hoffmann 2007; Lowe 2007; Maasland 2007) the information was tailored to be of relevance to the individual. In Pain 1990 and Hoffmann 2007 participants were provided with individualised booklets. In Banet 1997 the intervention group were given a copy of their medical history, clinical résumés, test results and leaflets. In Maasland 2007 information was delivered via an individualised multimedia computer programme. In Lowe 2007 the intervention comprised personalised information presented by a research registrar who explained its contents and addressed any additional concerns. In O'Connell 2009, the intervention group were given a patient‐held record that included telephone numbers, generic stroke information and fact sheets relevant to the patient's specific stroke problems.

Twelve trials evaluated an active intervention (Evans 1988; Rodgers 1999; Frank 2000; Johnson 2000; Kalra 2004; Smith 2004; Ellis 2005; Larson 2005; Draper 2007; Johnston 2007; Chiu 2008; Chinchai 2010). In four trials (Evans 1988; Rodgers 1999; Johnson 2000; Larson 2005) the intervention consisted of a programme of lectures providing information about stroke and services available and an opportunity to ask questions. In addition to this, the four‐week course evaluated by Johnson 2000 emphasised the importance of self‐esteem and coping strategies. Participants in the trial by Larson 2005 were also able to contact the stroke specialist nurse between sessions for extra information and support. Five studies (Frank 2000; Kalra 2004; Smith 2004; Ellis 2005; Draper 2007) evaluated a multi‐component intervention. Carers in Kalra 2004 received instruction on a range of topics plus hands‐on training. In Frank 2000 the intervention consisted of a recovery plan, an interactive workbook and a weekly phone call from the researcher. In Draper 2007 the programme for carers of aphasic patients included communication strategies, relaxation and stress management. Patients and carers in Smith 2004 were provided with an information manual supported by fortnightly pre‐arranged review meetings with their multidisciplinary team. In Ellis 2005, the intervention group patients received a monthly review by a stroke nurse specialist, specially selected relevant written information, and personalised records detailing their individual risk factors and recommended risk factor targets. In Johnston 2007, participants received a workbook which provided information about stroke, task material such as goal setting and an audio relaxation tape. In Chiu 2008, the intervention consisted of information delivered by a pharmacist over a course of six sessions. In Chinchai 2010, the intervention consisted of lectures delivered to carers with weekly follow‐up reinforcement at home by health service volunteers. Further details of the interventions are provided in the Characteristics of included studies table.

Timing

The intervention was implemented prior to discharge from hospital in nine trials (Lomer 1987; Evans 1988; Banet 1997; Rodgers 1999; Kalra 2004; Smith 2004; Hoffmann 2007; Lowe 2007; O'Connell 2009); within three weeks of discharge (Johnston 2007) and one month after stroke, or at discharge, which ever was sooner in one trial (Mant 1998). In the remaining trials the intervention was implemented at varying times post‐discharge: soon after discharge (Pain 1990; Downes 1993); within three months of stroke (Ellis 2005; Maasland 2007 ); within 12 months of stroke (Draper 2007); after 12 months since stroke (Chiu 2008); six months to three years after stroke (Johnson 2000); within 18 months of stroke (Chinchai 2010); within two years of stroke (Frank 2000), and a mean of 76 days after stroke onset (Larson 2005).

Outcomes measured

The studies measured a range of outcomes. Details of these are provided in the 'Characteristics of included studies' table.

Assessment of knowledge

Ten trials evaluated patient or carer knowledge or both. All used different questionnaires, the majority of which had been specifically developed for the study. The questionnaire used by Evans 1988 had been validated (Stroke Care Information Test, range 0 to 36) (Evans 1985). The 26‐item knowledge of stroke scale used by Rodgers 1999 and the 17‐item knowledge of stroke and services questionnaire used by Smith 2004 were based on instruments used in other studies (Wellwood 1994; Drummond 1996; Mant 1998), and the content of the specific educational programme under evaluation. In the study by Hoffmann 2007, the 25‐item knowledge of stroke questionnaire developed for the study was based partly on a previously validated measure (Sullivan 2004). The content validity and test‐retest reliability of this instrument were assessed prior to the commencement of the study. The questionnaire used by Lowe 2007 was developed from professionals' ideas of what patients should be aware of concerning secondary prevention of stroke and was piloted with 58 stroke patients. In Maasland 2007 the questionnaire was developed and validated in 42 partners of patients with TIA. None of the questionnaires in the remaining three studies (Lomer 1987; Pain 1990; Mant 1998) had been validated.

Risk of bias in included studies

Method of analyses

Twelve studies reported that an intention‐to‐treat analysis had been conducted (Pain 1990; Banet 1997; Mant 1998; Rodgers 1999; Johnson 2000; Kalra 2004; Smith 2004; Ellis 2005; Hoffmann 2007; Johnston 2007; Lowe 2007; Maasland 2007).

Allocation

Allocation was concealed in nine trials (Mant 1998; Rodgers 1999; Kalra 2004; Smith 2004; Ellis 2005; Hoffmann 2007; Lowe 2007; Maasland 2007; O'Connell 2009). Larson 2005 reported that the sequence could not be predicted but the method of allocation concealment was not reported. Allocation by random number sequence was reported in one study (Downes 1993). However, the method was not described. Johnson 2000 used a matched pair design with one member of each pair randomly assigned (unconcealed randomisation) to either the treatment or control group. The method of random sequence generation was unclear or not reported in 11 trials (Lomer 1987; Pain 1990; Banet 1997; Frank 2000; Smith 2004; Larson 2005; Draper 2007; Johnston 2007; Lowe 2007; Chiu 2008; Chinchai 2010). One study reported the use of minimisation (Evans 1988). Ellis 2005 reported the use of a computer‐generated random sequence procedure. However, they reported that three patients were entered twice in to the treatment group in error. Further details of allocation concealment and methods of randomisation are provided in the 'Risk of bias' tables.

Blinding

Blinding of both participants and personnel was not a feature in any of the trials or blinding was unclear. Blinding of outcome assessors was reported in 14 trials (Evans 1988; Pain 1990; Downes 1993; Mant 1998; Rodgers 1999; Kalra 2004; Smith 2004; Ellis 2005; Hoffmann 2007; Johnston 2007; Lowe 2007; Maasland 2007; O'Connell 2009; Chinchai 2010), not described in five (Lomer 1987; Banet 1997; Larson 2005; Draper 2007; Chiu 2008) and not undertaken in two (Frank 2000; Johnson 2000). Further details of blinding are provided in the 'Risk of bias' tables.

Incomplete outcome data

The studies ranged in sample size from 36 (Pain 1990) to 300 (Kalra 2004). Losses to follow‐up ranged from zero (Lomer 1987; Johnson 2000; Chinchai 2010) to more than 20% (Downes 1993; Mant 1998; Rodgers 1999; Smith 2004; O'Connell 2009). A sample size calculation was reported for nine trials (Downes 1993; Rodgers 1999; Kalra 2004; Smith 2004; Ellis 2005; Draper 2007; Hoffmann 2007; Maasland 2007; O'Connell 2009). Of these, Downes 1993 reported final follow‐up results for 62 couples rather than the estimated 165, and Draper 2007 recruited only 39 of the 60 caregivers required. Rodgers 1999 recruited the required number of patients and carers in each group but a larger than anticipated number of patients were unable to complete the main outcome measure (Short Form‐36) leading to a short fall in the number of patients required to meet the original power calculation (73%, 117 patients of 160). There was also a short fall in the number of carers at final follow‐up (106 of 216). In the O'Connell 2009 trial, a combination of recruitment and retention problems and non‐use of the intervention resulted in the trial being terminated prior to completion. Sample size was small in a number of trials, particularly: Pain 1990 (N = 36); Banet 1997 (N = 52); Frank 2000 (N = 41); Johnson 2000 (N = 41); and Draper 2007 (N = 39). Further details of attrition bias are provided in the 'Risk of bias' tables.

Selective reporting

Study protocols were not obtained for any of the studies. As a result, it is unclear if selective reporting contributed to bias in the majority of studies. Further details of selective reporting bias are provided in the 'Risk of bias' tables.

Other potential sources of bias

In the trial undertaken by Rodgers 1999, only 51 patients (42%) of those randomised attended three or more out of the six outpatient sessions provided. In Draper 2007, collection of baseline data occurred after randomisation (although participants were still blinded at that point). In O'Connell 2009, the trial was terminated early as it was reported that numerous participants could not remember receiving the information (a sample size of 240 was the initial target, however, the trial was stopped when 66 participants were recruited).

Effects of interventions

Results are reported separately for patients and carers. Resource outcomes are also presented separately. Meta‐analyses have been undertaken for the domains of knowledge, emotional outcome, death, and for selected satisfaction questions. For other outcomes we have presented a narrative summary stratified by subgroup, and a summary of the data are provided in the 'Data and analyses' section.

Patient outcomes

Knowledge

Seven trials (Lomer 1987; Mant 1998; Rodgers 1999; Smith 2004; Hoffmann 2007; Lowe 2007; Maasland 2007) evaluated the effect of a passive or active intervention on knowledge. All had used different questionnaires.

Patient knowledge

Data were available for 536 of 770 participants from six trials (Mant 1998; Rodgers 1999; Smith 2004; Hoffmann 2007; Lowe 2007; Maasland 2007). There was a statistically significant effect on patient knowledge in favour of the intervention (SMD 0.29, 95% CI 0.12 to 0.46, P < 0.001) (Analysis 1.1).

1.1 — Comparison 1 Passive or active information versus control, Outcome 1 Patient knowledge.

Suitable data were not available for one trial (Lomer 1987). Patients followed up at one week knew significantly more about the aetiology of stroke and the treatment they were receiving than the controls (P < 0.05) but not about the specific prognosis or help and benefits available. This was a small trial, methods of randomisation and outcome assessment are unclear, and comparability of treatment groups is not reported. The results may therefore be subject to bias.

Subgroup analysis

Data were available from four passive information and two active information trials. There was no significant difference in the magnitude of effect between passive and active information (passive: SMD 0.26, 95% CI 0.04 to 0.48, active: SMD 0. 34, 95% CI 0.07 to 0.61, test for subgroup differences P = 0.65) (Analysis 1.1).

Emotional outcomes

We performed meta‐analyses for the outcomes of patient anxiety and patient depression using both dichotomous and continuous data. For each outcome we report the results of both meta‐analyses. A narrative summary is presented for other outcomes.

Anxiety

The majority of trials that evaluated patient anxiety used the anxiety subscale of the Hospital Anxiety and Depression Scale. We converted scale data to dichotomised data using an anxiety subscale cut‐off score of 10/11(Zigmond 1983). Johnston 2007 reported data from the anxiety sub‐scale of the Hospital Anxiety and Depression Scale at baseline only and a total anxiety and depression score post‐intervention.

Patient emotional outcome: anxiety (dichotomised data)

Dichotomous data were available for 681 of 975 participants from six trials (Downes 1993; Mant 1998; Rodgers 1999; Kalra 2004; Smith 2004; Hoffmann 2007). The pooled result for all trials showed no significant difference in the number of cases of anxiety between the intervention and control groups (OR 0.89, 95% CI 0.57 to 1.38, P = 0.60) (Analysis 1.2).

1.2 — Comparison 1 Passive or active information versus control, Outcome 2 Patient emotional outcome: anxiety (dichotomised data).

Patient emotional outcome: anxiety (continuous data)

Continuous data were available for 720 of 1016 participants from seven trials (Downes 1993; Mant 1998; Rodgers 1999; Frank 2000; Kalra 2004; Smith 2004; Hoffmann 2007). The pooled result for all trials showed no significant difference in anxiety scores between the intervention and the control groups (MD ‐0.34, 95% CI ‐1.17 to 0.50, P = 0.43) (Analysis 1.3). Johnston 2007 was not included in the meta‐analysis as we were unable to obtain suitable data. However, they reported no significant difference between intervention and control at baseline (P > 0.05) and no significant effects post‐intervention (data and P value not reported).

1.3 — Comparison 1 Passive or active information versus control, Outcome 3 Patient emotional outcome: anxiety (continuous data).

Subgroup analysis

As there was no significant overall effect on anxiety, the following subgroup analyses may be unreliable.

Dichotomous data were available from three passive and three active information trials. The effect on anxiety was not significant in either subgroup (P > 0.05). However, there was a significant difference between active information and passive information on the number of cases of patient anxiety (passive: OR 1.64, 95% CI 0.80 to 3.37; active: OR 0.61 95% CI 0.35 to 1.07, test for subgroup differences P = 0.03). There was a trend towards an increase in anxiety from the passive information and a decrease from the active information.
Continuous data were available from three passive and four active information trials. The effect on anxiety was significant for the active information subgroup (P = 0.002) and not for the passive information subgroup (P = 0.21) There was a significant difference between active information and passive information on patient anxiety scores (passive: MD 0.67, 95% CI ‐0.37 to 1.71; active: MD ‐0.98 95% CI ‐1.59 to ‐0.36, test for subgroup differences P = 0.008). There was a trend towards an increase in anxiety from the passive information.

Depression

Twelve trials evaluated the effect of passive or active information on patient depression. Depression was measured using the depression subscale of the Hospital Anxiety and Depression Scale in eight trials (Downes 1993; Mant 1998; Rodgers 1999; Frank 2000; Kalra 2004; Smith 2004; Hoffmann 2007); Johnston 2007 reported the depression sub‐scale of the Hospital Anxiety and Depression Scale at baseline and a total score post intervention. Other measures included the Geriatric Depression Scale (short form) (Sheikh 1986) by (Ellis 2005); the Beck Depression Inventory (Gallagher 1982) by Johnson 2000, the Yale single question (Mahoney 1994) by Lowe 2007 and the emotions subscale of the Stroke Impact Scale (Duncan 1999) by O'Connell 2009. Scale data were converted to dichotomous data using the recommended cut‐off scores for each outcome measure (hospital anxiety and depression scale depression sub‐scale cut‐off score of 10/11 and a Geriatric Depression Scale score > 10).

Patient emotional outcome: depression (dichotomised data)

Dichotomous data were available for 956 of 1280 participants from eight trials (Downes 1993; Mant 1998; Rodgers 1999; Kalra 2004; Smith 2004; Ellis 2005; Hoffmann 2007; Lowe 2007). The pooled result for all trials showed no significant difference in the number of cases of depression between the intervention and control groups (OR 0.90, 95% CI 0.61 to 1.32, P = 0.59) (Analysis 1.4).

1.4 — Comparison 1 Passive or active information versus control, Outcome 4 Patient emotional outcome: depression (dichotomised data).

Patient emotional outcome: depression (continuous data)

Continuous data (Hospital Anxiety and Depression Scale) were available for 720 of 1016 participants from seven trials (Downes 1993; Mant 1998; Rodgers 1999; Frank 2000; Kalra 2004; Smith 2004; Hoffmann 2007). There was a significant effect on depression scores in favour of the intervention (MD ‐0.52, 95% CI ‐0.93 to ‐0.10, P = 0.01) (Analysis 1.5).

1.5 — Comparison 1 Passive or active information versus control, Outcome 5 Patient emotional outcome: depression (continuous data).

Two trials were not included in the meta‐analysis as we were unable to obtain suitable data (Johnson 2000; Johnston 2007). Johnson 2000 reported no significant difference between the two groups at baseline but when all patients were reassessed one week after completion of the intervention phase (a four‐week education course), there was a significant difference in the mean depression scores measured by the Beck Depression Inventory (possible score range 0 to 63) (baseline: treatment group 12.52, control 12.94, F = 1.36, P < 0.53; follow‐up: treatment group 8.5, control 12.61, F = 2.79, P < 0.04). Johnston 2007 reported no significant difference between the intervention and the control at baseline (P > 0.05) and no significant effect post intervention (data and P value not reported).

Subgroup analyses

The following subgroup analyses for depression (dichotomous data) may be unreliable as there was no overall net effect.

Dichotomous data were available from four passive and four active information trials. The effect on depression was not significant in either subgroup (P > 0.05). However, there was a significant difference between active information and passive information on the number of cases of patient depression (passive: OR 1.57, 95% CI 0.85 to 2.93; active: OR 0.63, 95% CI 0.38 to 1.03, test for subgroup differences P = 0.02), with a trend in favour of active information.
Continuous data were available from three passive information and four active information trials. The effect on depression was significant in the active information subgroup (P = 0.002) and not in the passive information subgroup (P = 0.44). There was a significant difference between active information and passive information on patient depression scores (passive: MD 0.39, 95% CI ‐0.61 to 1.38; active: MD ‐0.71, 95% CI ‐1.16 to ‐ 0.25, test for subgroup differences P = 0.05).

Other emotional outcomes

An active information study (Johnson 2000) evaluated hope and hopelessness (Herth Hope Scale, score range 0 to 90) (Farran 1995) and coping (Ways of Coping‐Cardiovascular Accident Scale, score range 0 to 93, specifically developed for the study). There were no differences between the two groups at baseline for either outcome. At one week after completion of the intervention phase (a four‐week education course), there was a significant difference between the two groups in the mean hope scale scores (baseline: treatment group 68.89, control 69.2, P < 0.42; follow‐up: treatment group 73.68, control 66.33, P < 0.001). There was no significant difference in the coping scores.

A further active information study (Johnston 2007) evaluated perceived control over recovery utilising the Recovery Locus of Control Scale (Partridge 1989). Scores from nine items on a five‐point scale (strongly agree to strongly disagree) are combined such that higher scores indicate greater belief in personal control. A confidence in recovery scale was also administered (Lewin 1992). This scale measured patients' confidence in recovery from 0 (not at all confident) to 10 (totally confident). There was no significant difference (P > 0.05) between the intervention and the control for perceived control over recovery. There was a significant group by time interaction effect for patients' confidence in recovery, F(1, 197) = 10.67, P = 0.001. Confidence in recovery declined over time for control group patients but remained relatively stable for patients in the intervention group.

Activities of daily living

Passive information studies

There is no evidence of an effect of passive information on activities of daily living. There were no significant differences between the intervention and control groups in any of the four trials that evaluated this outcome (Pain 1990; Banet 1997; Mant 1998; O'Connell 2009).

Active information studies

There is no evidence of an effect of active information on activities of daily living. There were no significant differences between the intervention and control groups in any of the four trials that evaluated this outcome (Kalra 2004; Smith 2004; Draper 2007; Johnston 2007).

Participation

Passive information studies

There is no evidence of an effect of passive information on participation. There were no significant differences between the intervention and control groups in any of the three trials that evaluated this outcome (Mant 1998; Lowe 2007; Maasland 2007).

Active information studies

There is no evidence of an effect of active information on participation. There were no significant differences between the intervention and control groups in any of the trials that evaluated this outcome (Rodgers 1999; Kalra 2004; Smith 2004; Draper 2007).

Social activities

Passive information studies

There is no evidence of an effect of passive information on social activities. The one trial that evaluated this outcome (Pain 1990) reported no significant difference in social activities between the intervention and control groups as measured by the Frenchay activities index (Holbrook 1983).

Active information studies

There is no evidence of an effect of active information on social activities. The two trials that evaluated this outcome (Rodgers 1999; Smith 2004) reported no significant differences in social activities between the intervention and control groups as measured by the Nottingham extended activities of daily living (Nouri 1987) or the Frenchay activities index (Holbrook 1983) respectively.

Perceived health status and quality of life

Passive information studies

There is no evidence of an effect of passive information on patient health status or quality of life (Dartmouth COOP Charts) (Rowan 1994) in the two trials that measured this outcome (Mant 1998; Hoffmann 2007).

Active information studies

Four trials (Rodgers 1999; Frank 2000; Kalra 2004; Ellis 2005) evaluated this outcome. Kalra 2004 reported significantly improved quality in life as measured by the EuroQol visual analogue scale (EuroQol Group 1990) at both three and 12 months in patients whose caregivers had received training (intervention) compared with those who had received conventional care (control) (median score (range) at three months: intervention 60 (42 to 70), control 50 (40 to 90), P = 0.019: median score (range) at 12 months: intervention 65 (55 to 80), control 60 (41 to 80), P = 0.009). Three trials (Rodgers 1999; Frank 2000; Ellis 2005) found no significant difference between the intervention and control groups as measured by the MOS 36‐item short‐form health survey (SF36) (Ware 1992), the Functional Limitations Profile (Patrick 1989) or the EuroQol (EuroQol Group 1990) respectively. Chinchai 2010 investigated quality of life with the WHO Quality of Life Measure (WHOQOL‐BRIEF THAI) (Sakthong 2007). There were significant within group differences (P < 0.05) in the intervention group for the physical, psychological and environmental categories. No significant within‐group differences (P > 0.05) resulted in the control group. Between‐group differences were reported pre‐intervention only (P > 0.05).

Satisfaction with care and information received

Eight trials (Mant 1998; Rodgers 1999; Smith 2004; Ellis 2005; Hoffmann 2007; Johnston 2007; Lowe 2007; O'Connell 2009) evaluated patient satisfaction. Of these, four trials (Mant 1998; Rodgers 1999; Smith 2004; Ellis 2005) measured patient satisfaction using the Pound scale (Pound 1994) or a modified version of that scale. Additionally, the bespoke questionnaire used in the trial by Lowe 2007 included some common items. Meta‐analysis was performed for two questions that were considered to be most relevant to the review: (1) satisfaction with information about the causes and nature of stroke; and (2) satisfaction with information about allowances and services. Three trials did not contribute data to the meta‐analysis (Hoffmann 2007; Johnston 2007; O'Connell 2009). Hoffmann 2007 used a bespoke questionnaire, Johnston 2007 assessed satisfaction with treatment and advice using a 0 to 10 scale applied in a previous study (Morrison 2000). O'Connell 2009 evaluated whether participants in the intervention group recalled receiving and reading the information and taking action as a result of the information.

Patient satisfaction with information about causes and nature of the stroke

Data were available for 541 of 772 participants from five trials (Mant 1998; Rodgers 1999; Smith 2004; Ellis 2005; Lowe 2007). There was a significant difference in favour of the intervention in satisfaction with information about the causes and nature of stroke (OR 2.07, 95% CI 1.33 to 3.23, P = 0.001) (Analysis 1.9).

1.9 — Comparison 1 Passive or active information versus control, Outcome 9 Patient satisfaction with information about causes and nature of the stroke.

Patient satisfaction with information about allowances and services

Data were available for 452 of 672 participants from four trials (Mant 1998; Rodgers 1999; Smith 2004; Ellis 2005). There was no significant difference in satisfaction with information about allowances and services (OR 1.18, 95% CI 0.76 to 1.83, P = 0.46) (Analysis 1.10).

1.10 — Comparison 1 Passive or active information versus control, Outcome 10 Patient satisfaction with information about allowances and services.

Subgroup analyses

Satisfaction with information about the causes and nature of the stroke

Data were available for two passive information and three active information trials. There was no significant difference in the magnitude of effect of passive compared to active information (passive: OR 1.86, 95% CI 0.81 to 4.27; active: OR 2.16, 95% CI 1.28 to 3.67, test for subgroup differences P > 0.2).

Satisfaction with information about allowances and services

There were insufficient data to perform a subgroup analysis.

Service use

Passive information studies

There is no evidence of an effect of passive information on service use in the one study (Mant 1998) that evaluated this outcome.

Active information studies

There is no evidence of an effect of active information on service use. The four trials that measured this outcome (Evans 1988; Rodgers 1999; Kalra 2004; Smith 2004) reported that there was no significant difference in service use between the intervention and control groups.

Modification of health‐related behaviours or risk reduction

Passive information studies

There is no evidence of an effect of passive information on the modification of health behaviours or risk reduction. Two trials (Banet 1997; Lowe 2007) evaluated this outcome. One trial (Banet 1997) reported no statistically significant difference in scores for diet or medication between the group who received their medical records and the group that received information leaflets only, although actual results were not reported. In the other study (Lowe 2007) there were no statistically significant differences in blood pressure between the intervention group and control groups. In Maasland 2007, those who regularly used tobacco or alcohol reduced these behaviours more in the intervention group, but differences were not significant. There was a decrease in systolic and diastolic blood pressure in the intervention and control group but no significant difference between the groups. Patients in neither group reduced their weight. Serum cholesterol dropped significantly in both the intervention and the control group with no differences between the groups.

Active information studies

Three trials evaluated this outcome (Rodgers 1999; Ellis 2005; Chiu 2008). There is limited evidence of an effect of active information on the modifications of health behaviours or risk reduction from one study. In Chiu 2008, there was a statistically significant difference (P < 0.001) between the intervention and the control group for satisfactory management of blood pressure. However, there was insufficient information reported to determine the effectiveness of the blinding of patients, personnel or outcomes assessment and if allocation concealment was undertaken. There was no significant difference for the management of glucose or lipids. Two trials found no significant difference between the intervention and the control group. In Ellis 2005, they reported that their initial (planned) analysis appeared to demonstrate a statistically significant reduction in systolic blood pressure in the intervention group compared with the control group (P value not reported). However, when the analysis was repeated with adjustment for baseline blood pressure the difference was not significant (P = 0.126). There were no statistically significant changes in other major modifiable risk factors: systolic and diastolic blood pressure; reported smoking rate; cholesterol; random blood glucose; or HbA1c. In the other trial (Rodgers 1999) there was no significant difference in the numbers of patients who stopped smoking after the stroke (intervention 9/25, control 3/17, P = 0.44).

Death

Mortality data were available for 1553 participants from nine trials (Evans 1988; Mant 1998; Rodgers 1999; Kalra 2004; Smith 2004; Ellis 2005; Hoffmann 2007; Lowe 2007; Johnston 2007). There was no significant difference in mortality between the intervention and control groups (OR 0.86 95% CI 0.59 to1.25, P = 0.43) (Analysis 1.13).

1.13 — Comparison 1 Passive or active information versus control, Outcome 13 Death.

Subgroup analysis

Data were available from three passive information and six active information trials. There was no significant difference in the magnitude of effect of passive information compared with active information (passive: OR 0.80, 95% CI 0.34 to 1.86; active: OR 0.88, 95% CI 0.58 to1.33, test for subgroup differences P > 0.9).

Carer outcomes

Knowledge

Six trials (Lomer 1987; Evans 1988; Pain 1990; Mant 1998; Rodgers 1999; Smith 2004) evaluated the effect of a passive information or active information intervention on carer knowledge.

Carer knowledge

Data were available for 336 of 469 participants from four trials (Evans 1988; Mant 1998; Rodgers 1999; Smith 2004). There was a significant difference in carer knowledge between the intervention and control groups in favour of the intervention (SMD 0.74, 95% CI 0.06 to 1.43, P = 0.03) (Analysis 1.14).

1.14 — Comparison 1 Passive or active information versus control, Outcome 14 Carer knowledge.

Two small trials did not contribute data to the meta‐analysis (Lomer 1987; Pain 1990); Lomer 1987 found no significant difference in carer knowledge of stroke and no difference in the level of knowledge about the specific prognosis or help and benefits available. Pain 1990 reported that individualised information enhanced the carer's knowledge of how therapists had instructed the patient, although statistical significance was not reached.

Subgroup analysis

There were insufficient data to perform a sub‐group analysis.

Emotional outcomes

We conducted a meta‐analysis for the outcome of carer stress. As a variety of outcome measures were used to measure stress we only used dichotomous data in the analysis. A narrative summary is provided for the carer emotional outcomes of psychological distress, depression and burden.

Psychological distress

Psychological distress in caregivers was measured by Downes 1993, Kalra 2004 and Johnston 2007 using the Hospital Anxiety and Depression Scale (Zigmond 1983). Rodgers 1999 and Smith 2004 used the General Health Questionnaire‐30 or the General Health Questionnaire‐28 (Goldberg 1979). we converted scale data to dichotomous data using the recommended cut‐off scores of 10/11 for the Hospital Anxiety and Depression Scale and 4/5 for the General Health Questionnaire (Goldberg 1979; Zigmond 1983).

Suitable data were not available for Draper 2007 or Johnston 2007. Draper 2007 reported no significant difference in carer stress scores at final follow‐up (three months). Johnston 2007 reported baseline stress data for carers utilising the anxiety sub‐scale of the Hospital Anxiety and Depression Scale. Mean (SD): Intervention 7.64 (4.89), control 7.08 (4.01) and no significant effect of group by time interaction on total Hospital Anxiety and Depression Scale score.

Carer emotional outcome: Psycholgical distress

Dichotomous data were available for 498 of 643 participants from four trials (Downes 1993; Rodgers 1999; Kalra 2004; Smith 2004). There was no significant difference in carer stress between the intervention and the control group (OR 1.13, 95% CI 0.65 to 1.97, P = 0.65) (Analysis 1.16).

1.16 — Comparison 1 Passive or active information versus control, Outcome 16 Carer emotional outcome: psychological distress.

Subgroup analysis

There were insufficient data to perform a subgroup analysis.

Depression

Passive information studies

In the Downes 1993 trial, there was no significant difference in depression as measured by the Hospital Anxiety and Depression Scale between carers in the intervention group and the control group (mean depression score at six months (SD): intervention 5.8 (5.2), control 5.1 (3.2). Johnston 2007 reported baseline data for depression of carers, measured by the depression subscale of the Hospital Anxiety and Depression Scale. Mean (SD): intervention 5.7 (4.3), control 4.8 (3.9) and reported no significant effect of group by time interaction on total Hospital Anxiety and Depression Scale score.

Active information studies

One trial (Kalra 2004) evaluated this outcome. Carers in the intervention group were significantly less depressed as measured by the Hospital Anxiety and Depression Scale than carers in the control group (median depression score at one year (IQR): intervention 2 (1 to 3), control 3 (2 to 5); P < 0.0001).

Burden

Passive information studies

In the one trial that evaluated this outcome (Mant 1998) there was no evidence of an effect of passive information on carer burden.

Active information studies

Two trials evaluated this outcome. In the study by Kalra 2004 caregiver burden was significantly reduced in carers in the intervention group compared with the control group at both three months and one year (median score at 12 months (IQR): intervention 32 (27 to 41), control 41 (36 to 50); P = 0.0001). In Draper 2007 there were no significant differences in pre to post‐treatment scores for either the intervention or wait control group.

Social Activities

Passive information studies

No trials evaluated this outcome.

Active information studies

There was no significant difference in carer social activities in the two trials (Kalra 2004; Draper 2007) that evaluated this outcome.

Perceived health status and quality of life

Passive information studies

From the one study (Mant 1998) that evaluated this outcome there is no evidence of an effect of passive information on carer perceived health and quality of life.

Active information studies

Three trials measured this outcome (Rodgers 1999; Kalra 2004; Larson 2005). The largest of these (Kalra 2004) reported that carers in the intervention group had a higher quality of life as measured by the EuroQol visual analogue scale (EuroQol Group 1990) than controls at both three months and one year (median score (IQR) at one year: intervention 80 (70 to 90), control 70 (60 to 80); P < 0.0001). In Rodgers 1999 there were no significant differences between carers in the intervention and the control groups on any of the domains of the SF36 except social functioning. This was significantly higher for carers in the control group (intervention group: mean 66.7 ± 29.8 SD; control group; mean 78.1 ± 27.4 SD; difference between means: 95% CI 11.3; 0.09 to 22.7; P = 0.04). This may be a chance finding due to multiple testing and the authors suggest that it should be interpreted with caution. In the trial by Larson 2005, there were no statistically significant differences between the groups over time.

Satisfaction

Five trials (Pain 1990; Mant 1998; Rodgers 1999; Kalra 2004; Smith 2004) evaluated carer satisfaction. Of these, three trials (Mant 1998; Rodgers 1999; Smith 2004) measured carer satisfaction using the Pound scale (Pound 1993) or a modified version of this scale. Meta‐analyses were performed for two questions considered to be of most relevance to the review: (1) satisfaction with information about recovery and rehabilitation; and (2) satisfaction with information about allowances and services. The remaining studies (Pain 1990; Kalra 2004) both evaluated aspects of carer satisfaction using a bespoke questionnaire and are not included in the meta‐analyses.

Carer satisfaction with information about recovery and rehabilitation

Data were available for 165 of 273 participants from two trials (Rodgers 1999; Smith 2004). There was no significant difference in satisfaction with information about recovery and rehabilitation (OR 1.78, 95% CI 0.88 to 3.60, P = 0.11) (Analysis 1.19).

1.19 — Comparison 1 Passive or active information versus control, Outcome 19 Carer satisfaction with information about recovery and rehabilitation.

Carer satisfaction with information about allowances and services

Data were available for 214 of 322 participants from three trials (Mant 1998; Rodgers 1999; Smith 2004) for one question only: there was no significant difference in satisfaction between the groups (OR 1.30, 95% CI 0.71 to 2.37, P = 0.39) (Analysis 1.20).

1.20 — Comparison 1 Passive or active information versus control, Outcome 20 Carer satisfaction with information about allowances and services.

Subgroup analysis

There were insufficient data to perform a subgroup analysis.

Resource outcomes

Cost to health and social services

Passive information studies

No trials evaluated this outcome.

Active information studies

Only one study (Kalra 2004) evaluated resource use. Total health and social care costs over one year for patients whose carers received training (intervention) were significantly lower (MD ‐£4043 ($7249; EUR 6072), 95% CI to ‐£1595 to £6544). The cost differences were largely due to differences in length of hospital stay.

Discussion

This review has explored the effectiveness of information provision for stroke patients and their carers as a process of care aimed at improving stroke recovery. In order to summarise effectively the available evidence on the core concept of information provision, we categorised the studies according to the nature of the intervention using two categories: passive and active. Our intention was to differentiate between interventions where participation was largely passive with no subsequent systematic follow‐up or reinforcement procedure, and those in which there was active participation with a subsequent agreed plan for clarification and reinforcement. This classification was developed, agreed, and adopted prior to results synthesis.

We performed meta‐analyses for the outcomes of knowledge, mood and death, and for selected satisfaction questions. We carried out a qualitative analysis for all other outcomes. We performed meta‐analyses for the outcomes of patient anxiety and depression using both the reported mean and standard deviation of the Hospital Anxiety and Depression Scale scores and dichotomised data. An advantage of using dichotomised data is that it may provide more clinically meaningful results as it relates to 'cases' of depression and anxiety. However, it has been argued that collapsing ordinal stroke trial data in this way can result in a loss of discrimination between groups such that significant treatment effects are missed (OAST 2007).

It is worthy of comment that we undertook extensive searches for the update of this review, we reviewed over 20,000 titles, yet only four new studies are included. This reflects a lack of precision of the search strategies but may also be a reflection on research progress. We excluded 73 studies, many of which evaluated a complex intervention of which information provision and education are components. Information provision is acknowledged as a key component of stroke service delivery, provision of leaflets is not effective and it would seem that new multi‐faceted ways of addressing the information needs of patients and their carers are being developed and evaluated.

Summary of main results

We have identified a total of 21 trials involving 2289 patients and 1290 carer participants. We found some statistically significant but clinically small benefits supporting the general concept that information provision after stroke might improve outcomes. There was evidence of benefit in relation to improved patient and carer knowledge, some aspects of patient‐reported satisfaction and for depression scores in patients. Additionally, we found some evidence that interventions using active information provision may be more effective than passive information for the clinically important outcomes of patient depression and anxiety symptoms. However, as we saw no effect with the dichotomous endpoints of anxiety or depression, effects may be small. We found no evidence that information interventions are associated with improvements in activity limitation, participation or changes in service use.

Overall completeness and applicability of evidence

All included studies were relevant to the review question. There was extensive variation in the content and delivery format of the interventions. This appears to reflect the diversity of interventions provided within clinical practice. Whilst there were sufficient data to address the primary outcomes and the majority of secondary outcomes for this review, there were limited studies to address social activities in carers or resource outcomes. Current practice on information provision after stroke varies nationally and internationally. Our review identified studies from seven countries, thus drawing conclusions on overall applicability of findings internationally is limited.

Quality of the evidence

There was considerable variation in the interventions evaluated and the 21 included trials were of variable quality. Clearly concealed randomisation was achieved in only 10 trials (Mant 1998; Rodgers 1999; Kalra 2004; Smith 2004; Ellis 2005; Larson 2005; Hoffmann 2007; Lowe 2007; Maasland 2007; O'Connell 2009). The rate of attrition was over 20% in five trials (Downes 1993; Mant 1998; Rodgers 1999; Smith 2004; O'Connell 2009). In several trials the sample size was small: less than 75 participants in eight trials (Pain 1990; Downes 1993; Banet 1997; Frank 2000; Johnson 2000; Draper 2007; Maasland 2007; Chinchai 2010).

Our evaluation of the effect on passive or active information provision on the outcome of stroke knowledge was limited by a lack of a consistently‐used measure. Knowledge of stroke was assessed in nine of the 21 studies reviewed (Lomer 1987; Evans 1988; Pain 1990; Mant 1998; Rodgers 1999; Smith 2004; Hoffmann 2007; Lowe 2007; Maasland 2007) but as each study had used a different questionnaire, combining the results in a meta‐analysis was problematic. Our initial intention was to perform a meta‐analysis using dichotomised data (knowledge improved or not improved). However, this was not feasible as in some trials knowledge was measured on one occasion only. We therefore combined the data using the SMD wherein the MDs in outcome between the groups being studied are standardised to account for differences in scoring methods. A disadvantage with this method is that interpretation of the clinical relevance of the treatment effect is difficult as estimated effect sizes serve only as a qualitative measure of the strength of evidence against the null hypothesis (de Beurs 1999). The results should therefore be treated with some caution. In addition, for the majority of the bespoke questionnaires used to measure knowledge there was limited information available about the reliability of the questions they contained.

Potential biases in the review process

Our search strategy was comprehensive and as we were able to identify a number of unpublished studies, publication bias is unlikely. Study selection, data collection and analysis were undertaken by two people, with a third person or consensus meeting used to resolve differences. As a result we are confident of limited bias in the review process for this review.

Agreements and disagreements with other studies or reviews

The positive effects of information on knowledge and depression demonstrated in this review are supported by the findings of reviews of patient education interventions in other conditions. In a meta‐analysis, Brown reported that diabetes education had a moderate to large effect on improving patient knowledge (Brown 1990). A systematic review of education for adults with rheumatoid arthritis showed a small effect on depression (Riemsma 2003). In accord with our review, an overview of systematic reviews of educational interventions for healthcare professionals reported that passive approaches were generally ineffective and unlikely to result in changes in professionals' behaviour, whereas educational approaches involving active learning were more likely to be effective (Grimshaw 2001). A systematic review of education programmes for patients with diabetic kidney disease found education programmes have beneficial effects on improving patients' knowledge of diabetes and some self‐management behavioural changes (Li 2011). A meta‐analysis of patient teaching strategies showed that the greatest effect size was associated with reinforcement, independent study, and the use of multiple strategies (Theis 1995).

Future direction

This review has demonstrated some positive effects of information provision on patient and carer knowledge, aspects of satisfaction and depression. However, the effects, although statistically significant, were clinically small and more effective information provision strategies after stroke need to be developed. The results of the review suggest that a strategy based on an active, rather than passive intervention approach should be adopted. This is perhaps unsurprising as stroke is a complex condition with wide‐ranging effects and probably requires a more profound approach to promote recovery than can be achieved by the provision of passive information alone. The specific components of the active information provision (i.e. involving recipients, planned follow‐up or reinforcement), which resulted in modest beneficial effects on some outcomes, requires further investigation. Future work should focus on the further development of a generalisable active information intervention that could be robustly evaluated in a large multicentre study.

Authors' conclusions

Implications for practice.

There is evidence to support the routine provision of information to stroke patients and their families. Providing information has been shown to improve knowledge of stroke, increase some aspects of patient satisfaction, and reduce patient depression scores. However, the reduction in depression score (as measured by the Hospital Anxiety and Depression Scale) was small and may not be clinically significant. There is currently no evidence that providing information is effective in improving other patient and carer outcomes. Although the best way to provide information is still not clear, the results of the review suggest that strategies that actively involve patients and carers and include planned follow‐up for clarification and reinforcement should be used in routine practice.

Implications for research.

Future work should focus on the further development of a generalisable intervention which could be robustly evaluated in a large multicentre study. The evaluation of interventions is currently limited by the lack of a widely recognised measure of stroke knowledge. Attention should be given to given to the design, development and evaluation of a stroke knowledge questionnaire. Consideration should be given to the most appropriate outcome domains for this type of intervention.

What's new

Date	Event	Description
18 July 2012	New citation required but conclusions have not changed	The results and the conclusions for this update are the same as for the previous version of the review.
30 December 2011	New search has been performed	This review update has added four new trials (Johnston 2007; Chiu 2008; O'Connell 2009; Chinchai 2010).The review now includes 21 trials involving 2289 patient and 1290 carer participants. Additional data have been added for analysis for the patient outcome for death.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Patient knowledge	6	536	Std. Mean Difference (IV, Fixed, 95% CI)	0.29 [0.12, 0.46]
1.1 passive information	4	327	Std. Mean Difference (IV, Fixed, 95% CI)	0.26 [0.04, 0.48]
1.2 active information	2	209	Std. Mean Difference (IV, Fixed, 95% CI)	0.34 [0.07, 0.61]
2 Patient emotional outcome: anxiety (dichotomised data)	6	681	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.89 [0.57, 1.38]
2.1 passive information	3	227	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.64 [0.80, 3.37]
2.2 active information	3	454	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.61 [0.35, 1.07]
3 Patient emotional outcome: anxiety (continuous data)	7	720	Mean Difference (IV, Random, 95% CI)	‐0.34 [‐1.17, 0.50]
3.1 passive information	3	227	Mean Difference (IV, Random, 95% CI)	0.67 [‐0.37, 1.71]
3.2 active information	4	493	Mean Difference (IV, Random, 95% CI)	‐0.98 [‐1.59, ‐0.36]
4 Patient emotional outcome: depression (dichotomised data)	8	956	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.90 [0.61, 1.32]
4.1 passive information	4	311	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.57 [0.85, 2.93]
4.2 active information	4	645	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.63 [0.38, 1.03]
5 Patient emotional outcome: depression (continuous data)	7	720	Mean Difference (IV, Fixed, 95% CI)	‐0.52 [‐0.93, ‐0.10]
5.1 passive information	3	227	Mean Difference (IV, Fixed, 95% CI)	0.39 [‐0.61, 1.38]
5.2 active information	4	493	Mean Difference (IV, Fixed, 95% CI)	‐0.71 [‐1.16, ‐0.25]
6 Patient activities of daily living and participation: summary of results			Other data	No numeric data
7 Patient social activities: summary of results			Other data	No numeric data
8 Patient perceived health status and quality of life: summary of results			Other data	No numeric data
9 Patient satisfaction with information about causes and nature of the stroke	5	541	Peto Odds Ratio (Peto, Fixed, 95% CI)	2.07 [1.33, 3.23]
9.1 passive information	2	143	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.86 [0.81, 4.27]
9.2 active information	3	398	Peto Odds Ratio (Peto, Fixed, 95% CI)	2.16 [1.28, 3.67]
10 Patient satisfaction with information about allowances and services	4	452	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.18 [0.76, 1.83]
10.1 passive information	1	57	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.51 [0.15, 1.75]
10.2 active information	3	395	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.33 [0.83, 2.12]
11 Service use: summary of results			Other data	No numeric data
12 Modification of health related behaviours: summary of results			Other data	No numeric data
13 Death	9	1553	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.86 [0.59, 1.25]
13.1 passive information	3	331	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.80 [0.34, 1.86]
13.2 active information	6	1222	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.88 [0.58, 1.33]
14 Carer knowledge	4	336	Std. Mean Difference (IV, Random, 95% CI)	0.74 [0.06, 1.43]
14.1 passive information	1	33	Std. Mean Difference (IV, Random, 95% CI)	0.28 [‐0.42, 0.97]
14.2 active information	3	303	Std. Mean Difference (IV, Random, 95% CI)	0.88 [0.05, 1.70]
15 Carer emotional outcome: summary of results			Other data	No numeric data
16 Carer emotional outcome: psychological distress	4	498	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.13 [0.65, 1.97]
16.1 passive information	1	51	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.29 [0.31, 5.38]
16.2 active information	3	447	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.11 [0.61, 2.01]
17 Carer social activities: summary of results			Other data	No numeric data
18 Carer perceived health status and quality of life: summary of results			Other data	No numeric data
19 Carer satisfaction with information about recovery and rehabilitation	2	165	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.78 [0.88, 3.60]
19.1 active information	2	165	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.78 [0.88, 3.60]
20 Carer satisfaction with information about allowances and services	3	214	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.30 [0.71, 2.37]
20.1 passive information	1	47	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.61 [0.16, 2.26]
20.2 active information	2	167	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.59 [0.81, 3.13]
21 Cost to health and social services: summary of results			Other data	No numeric data
22 Self management: summary of results			Other data	No numeric data
23 Family functioning and patient adjustment: summary of results			Other data	No numeric data

Patient activities of daily living and participation: summary of results
Study	Follow‐up	Comparison	Outcome measure	Results	Notes
Banet 1997	6 months	Patient education packet and shared medical record versus patient education packet	Glasgow Outcome Scale   Global Outcome Scale (not validated)	Number of participants with outcome data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: 24/number unclear, control 28/number unclear   Comparison between shared record and control groups   Glasgow Outcome Scale F < 0.01, P = 0.74   Global Outcome Scale F = 1.7, P = 0.20
Frank 2000	1 month	Workbook + home visits and telephone contact versus wait control	Functional Limitations Profile	Number of participants with outcome data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 19/20, control 20/21   Mean score before intervention: intervention 69.62 (SD 17.77), control 71.73 (SD 25.41)   1 month: intervention 64.03 (SD 20.96), control 66.89 (SD 22.87)
Johnston 2007	8 weeks (post intervention) and 6 months from baseline	Post discharge workbook intervention including information and audio relaxation tape versus usual care	Barthel Index (transformed scores)	Number of participants with outcome data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 74/103, control 84/100   Mean score before intervention: intervention 1.57 (SD 0.73), control 1.50 (SD 0.63)   8 weeks: intervention 1.44 (SD 0.65), control 1.43 (SD 0.59) 6 months: intervention 1.43 (SD 0.68), control 1.39 (SD 0.61)	Transformed scores: higher score = higher disability
Kalra 2004	3 and 12 months	Education sessions plus hands on training versus conventional care	Barthel Index   Modified Rankin Scale	Number of participants with Barthel Index data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 134/151, control 134/149   Barthel Index score > 18   3 months: intervention 77/141, control 52/140   12 months: intervention 93/134, control 75/134   Number of participants Modified Rankin Score data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 133/151, control 134/149   Modified Rankin Scale score 0 to 2   3 months: intervention 80/141, control 63/140   12 months: intervention 100/134, control 87/134
Lowe 2007	3 and 6 months	CareFile (29‐page personalised information booklet) + discussion with research registrar versus usual information and follow‐up	Rankin Scale	Number of participants with outcome data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 44/50, control 40/50   Median (range)   Before intervention: intervention 3 (2 to 4), control 2 (1 to 3)   3 months: intervention 3 (2 to 3), control 3 (2 to 3)   6 months: intervention 2 (1 to 3), control 2 (2 to 3)
Maasland 2007	12 weeks	Health education plus computer programme versus health education alone	Rankin Scale	Number of participants with outcome data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 30/33, control 27/32   No significant differences between the groups
Mant 1998	6 months	Information pack versus no intervention	London Handicap Scale   Barthel Index	Number of participants with London Handicap Scale data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 34/48, control 34/45   Mean score   London Handicap Scale: intervention 0.6 (SD 0.22), control 0.5 (SD 0.18)   Number of participants with Barthel Index data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 37/48, control 34/45   Barthel Index: intervention 15.1 (SD 6.42), control 15.1 (SD 4.82)
O'Connell 2009	4 weeks and 4 months	Patient Held Record (PHR) incorporating fact sheets with patient specific stroke information versus usual discharge information	ADL sub‐scale of the Stroke Impact Scale (SIS)	Number of participants with outcome data available at the end of 4 month follow‐up/number of participants in group at outset of the trial: intervention 28/46, control 38/47   Mean score   SIS ADL sub‐scale: at 4 weeks: intervention 71.96 (SD 16.49), control 81.64 (SD 19.03)   4 months: intervention 74.91(SD 18.25), control 82.43 (SD 17.71)
Pain 1990	3 months	Individualised information booklet + advice and information versus advice and information	Barthel Index	Number of participants with outcome data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 16/21, control 13/15   Mean score   Barthel Index: at discharge: intervention 77.6 (SD 24.5), control 83.5 (SD 18.5)   3 months: intervention 75.1 (SD 26.0), control 85.0 (SD 17.6)
Rodgers 1999	6 months	Education programme versus usual care	Oxford Handicap Scale	Number of participants with outcome data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 89/121, control 64/83   Mean score   Oxford Handicap Scale: intervention 3.1 (SD 1.4), control 3.2 (SD 1.2)
Smith 2004	3 and 6 months	Stroke Recovery Programme (information manual) + fortnightly meetings with multidisciplinary team versus usual care	Barthel Index   London Handicap Scale	Number of participants with Barthel Index data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 69/84, control 64/86   Median change from baseline (range)   Barthel Index:   3 months: intervention 6 (‐5 to 17), control 5 (‐5 to 20)   6 months: intervention 7 (‐13 to 17), control 5.5 (‐9 to 19)   Median score (range)   Number of participants with London Handicap data available at the end of scheduled follow‐up/number of participants in group at outset of the trial: intervention 69/84, control 64/86   London Handicap Scale   3 months: intervention 57 (9 to 90), control 54 (25 to 95)   6 months: intervention 59 (20 to 94), control 57 (12 to 84)

Carer perceived health status and quality of life: summary of results
Study	Follow ‐up	Comparison	Outcome measure	Results	Notes
Kalra 2004	3 months and 1 year	Education sessions + hands on training versus conventional care	EuroQuol visual analogue scale   Quality adjusted life years	Number of participants with outcome data available at the end of scheduled follow up/number of participants in group at outset of the trial: intervention 112/151, control 120/149   EuroQuol visual analogue scale   Median (IQR)   Before intervention: intervention 90 (80 to 95), control 85 (80 to 90)   3 months: intervention 80 (71 to 90), control 70 (60 to 80)*   1 year: intervention 80 (70 to 90), control 70 (60 to 80)†   Quality adjusted life years mean (SD)   Pre‐intervention: intervention 0.94 (SD 0.10), control 0.94 (SD 0.14)   1 year: intervention 0.91 (SD 0.11), control 0.90 (SD 0.14)	statistically significant difference *P = 0.0001   †P = 0.0001
Larson 2005	6 months and 1 year	Education programme + support versus regular information and possibility of attending 1 open session	General quality of life visual analogue scale   Bradley's well‐being questionnaire   LISS questionnaire (life situation among spouses after stroke event)   EuroQuol visual analogue scale	Number of participants with outcome data available at the end of scheduled follow up/number of participants in group at outset of the trial: intervention 46/50, control 45/50   Mean (SD)   General quality of life visual analogue scale:   Before intervention: intervention 60.08 (22.79), control 60.22 (22.57)   6 months: intervention 63.04 (22.35), control 63.87 (20.45)   1 year: intervention 68.00 (22.89), control 66.78 (20.22)   Bradley's well‐being questionnaire   General well being:   Before intervention: intervention 25.28 (5.03), control 25.28 (5.33)   6 months: intervention 25.87 (6.57), control 24.14 (6.57)   12 months: intervention 24.57 (5.87), control 25.91 (6.20)   LISS‐questionnaire:   Life situation total score :   Before intervention: intervention 46.47 (10.29), control 47.54 (9.42)   6 months: intervention 48.43 (9.82), control 47.48 (9.26)   12 months: intervention 48.59 (10.91), control 49.11 (9.66)   EuroQol visual analogue scale:   Before intervention: intervention 72.35 (18.54), control 77.98 (20.18)   6 months: intervention 76.91 (16.39), control 76.85 (18.02)   12 months: intervention 73.63 (17.98), control 77.27 (16.77)

Methods	Patients who met all criteria and volunteered to participate were randomly assigned to treatment group; no further details given No stated blind outcome assessment 6 patients lost to follow‐up; no report of differential losses between groups 6‐month follow‐up
Participants	St Louis, Mo, USA 58 first‐ time stroke patients: number allocated to intervention or control not given No details of age Sex: women N = 28 Inclusion criteria: aged 18 years or older, first‐time stroke, medically stable, competent to give informed consent, ready for hospital discharge Exclusion criteria: aphasia or motor impairments that hindered ability to complete forms unless neurologist believed it did not interfere with giving consent and had a caregiver who could help complete forms, or could dictate answer to investigator N = 52 for final follow‐up
Interventions	Treatment: copy of medical history, clinical resumes, notes on outpatient visits, x‐ray, scan reports and pertinent laboratory results. Also received patient education packet containing leaflets on stroke care, stroke team, tests and procedures, community resources, defining terms, facts about stroke, how stroke affects behaviour and recovering from stroke Focus: patient Setting: hospital Administration: unclear who gave record Encouraged to maintain records by incorporating updated information by taking them to all appointments with physicians and all trips during the study 1 contact, length unknown Patients ready for discharge Control: given patient education packet containing leaflets on stroke care, stroke team, tests and procedures, community resources, defining terms, facts about stroke, how stroke affects behaviour and recovering from stroke
Outcomes	(1) Disability and handicap (baseline and 6 months) (2) Intention to modify health‐related behaviours and compliance (baseline and 6 months) Not included in the review: Global Outcome (baseline and 6 months)
Notes	Validity assessment: use of inappropriate statistical tests
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The procedure for generating a random sequence was not reported
Allocation concealment (selection bias)	Unclear risk	Reported that patients were randomly assigned but method not described
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No report of blinding of participants or personnel but as no intervention provided for the control group, group assignment would have been apparent
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	It was not reported if outcome assessments were blinded
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Quote: "One volunteer died, and five provided incomplete data. Thus data from 52 subjects was available for analysis." Although losses were relatively small, it was not reported which groups they were from
Selective reporting (reporting bias)	Unclear risk	Quote: "Subjects reported their intentions to modify health‐related behaviours by completing the diet, smoking, and medication sub‐scales of Miller’s Health Intention Scale." Quote: "Because so few subjects smoked, this was not included as a variable in the analysis." Comment: note that this question was typically answered (i.e. data was not generally missing), but only 7 smoked at the time of their stroke
Other bias	Low risk	No other obvious sources of bias

Methods	Cluster RCT No participants lost to follow‐up
Participants	Chiang Mai, Thailand 60 stroke patients and their primary caregivers (N = 60) Patients: intervention N = 30; control N = 30 Caregivers: intervention N = 30; control N = 30 Age range of patients intervention (years): < 40 N = 9; 40 to 59 N = 8; 60 to 69 N = 9; 70 to 79 N = 7 Age range of patients control (years): < 40 N = 4; 40 to 59 N = 8; 60 to 69 N = 5; 70 to 79 N = 13 Sex of patients male: intervention 60%; control 53% Age range of carers intervention (years): < 40 N = 2; 40 to 59 N = 8; 60 to 69 N = 11; 70 to 79 N = 9 Age range of carers control (years): < 40 N = 5; 40 to 59 N = 12; 60 to 69 N = 6; 70 to 79 N = 7 Sex of carers male: intervention 47%; control 53% Inclusion criteria patient: discharged from hospital < 18 months, physical function recovery level 2 to 4 classified by Brunnstorm; communication (verbal, non‐verbal), no complications (e.g. bedsores, pain, fever during data collection), willingness to participate in the study Inclusion criteria carer: primary caregiver (family member or relative), not previously attended the home health care and stroke rehabilitation programme, minimum 8 hours a day caring, willingness to participate in the study Exclusion criteria: not reported
Interventions	Intervention: an education programme for caregivers with follow‐up reinforcement. Included lectures and active practice of activities of daily living and written information in guidebooks. Intervention started within 18 months of patient stroke. Carers attended a 1‐day, 7‐hour education session on 3 consecutive weeks and received weekly visits for reinforcement by health service volunteers Focus: patient and caregiver Setting: primary healthcare unit Administration: occupational therapists with a minimum of 2 years experience Control: usual care information from health stations located in the community
Outcomes	Patient outcomes: Quality of Life (7 days pre‐intervention and 2 months post intervention)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The procedure for generating a random sequence was not reported
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not reported
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Although not specifically reported, control participants received usual care, therefore the (lack of) intervention could have been obvious
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Research assistants blind to group assignment performed assessments
Incomplete outcome data (attrition bias)   All outcomes	Low risk	No drop outs or exclusions
Selective reporting (reporting bias)	High risk	When describing the WHOQOL‐BREF the authors report the individual items for overall health and overall QOL, as well as a total score (the summation of all items). However these were not presented in the results
Other bias	Low risk	No other obvious sources of bias

Methods	RCT by simple random sampling. Patients were stratified by age (over 65 or not) and sex 4 patients from the control group and 2 patients from the intervention group lost to follow‐up
Participants	Kaohsiung, Taiwan 160 stroke patients (intervention N = 80, control N = 80) Mean age of patients: intervention 66 years; control 65 years Sex of patient male: 50% Inclusion criteria: stroke out‐patients who had visited clinics regularly after stroke (> 12 months) Exclusion criteria: enrolled in other studies, terminal illness, no consent
Interventions	Intervention: consultation (drug effects, lifestyle modification, benefits of therapies, importance of compliance, verification of drug interaction and reminder of adverse events). Focus: patient Setting: unclear Administration: intervention delivered by pharmacist over 6 x 1‐hour sessions over a 6‐month period Control: no information reported
Outcomes	(1) Management of hypertension (2) Management of lipids (3) Management of glucose
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as simple random sampling but method of sequence generation not reported
Allocation concealment (selection bias)	Unclear risk	No information reported
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	No information reported
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	No information reported
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Small number lost to follow‐up (2 from intervention and 4 from the control)
Selective reporting (reporting bias)	Unclear risk	Study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious signs of bias

Methods	Allocation by random number sequence; no other details given Blinded outcome assessment Number lost to follow‐up unclear; no report of differential losses between groups 6‐month follow‐up
Participants	Birmingham, UK Stroke patients and carers (couples): number initially recruited to control and information groups unknown (105 couples recruited to the 3‐group trial) Information provided for N = 18 control group, N = 22 information group who completed 6‐month assessment Age of patient > 60 years: treatment 91%; control 89% Sex of patient male: treatment 55%: control 44% Inclusion criteria: stroke survivors living at home with their informal carers, recent stroke (not necessarily first) causing increase on modified Rankin Disability Scale and post‐stroke Rankin score of 2 to 5 Exclusion criteria: none stated
Interventions	Treatment: information pack designed for study containing information about physical, cognitive and emotional effects of stroke, carer well being and local services Focus: patient and carer Setting: home Administration: single visit by nurse counsellor who demonstrated how to access relevant information and answered questions. 1 x 1‐hour visit at least 2 weeks after discharge but exact time unknown Control: usual care, no intervention
Outcomes	(1) Emotional outcome (baseline and 6 months)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "they were allocated by a random number sequence generation." However, method not described.
Allocation concealment (selection bias)	Unclear risk	No information reported
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No evidence of blinding of participants or personnel
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Assesment was carried out by a research assistant who was blind to group allocation
Incomplete outcome data (attrition bias)   All outcomes	High risk	150 couples originally recruited in to the study but only 62 completed and were in the final analysis. Unclear how many from each group
Selective reporting (reporting bias)	Unclear risk	Study protocol not available so cannot assess reporting bias
Other bias	Low risk	Appears free from other sources of bias

Methods	Allocation by random selection of names by blinded investigator Postal outcome assessment 8 carers lost to follow‐up
Participants	Sydney, Australia 39 carers of aphasic stroke patients recruited from rehabilitation services of 3 public hospitals: treatment N = 19; control N = 20 Completed final follow‐up: N = 31 Mean age carer: treatment 64 years; control 60 years Mean age of patient: treatment 69 years; control 68 years Sex of carer: not reported Sex of patient: not reported
Interventions	Treatment: education programme covering the impact of stroke, managing the resulting life changes, communication strategies, relaxation and stress management, managing emotions, accessing community services and relapse prevention strategies. At the end of course the caregivers were encouraged to remain in contact as a self‐help group Focus: carer Setting: held in outpatient area of hospital rehabilitation department Administration: 4 x 1‐weekly group session, each session 2 hours, numbers in each group varied from 6 to 11, sessions run by a speech pathologist and social worker, clinical psychologist included for 1 session Control: usual care, wait‐list control commenced the treatment after a delay of 3 months
Outcomes	Carer Primary outcomes (1) Psychological distress (4 weeks and 3 months) (2) Caregiver burden (4 weeks and 3 months) (3) Communication (4 weeks and 3 months) Secondary outcomes (1) Attitudes towards care‐giving (4 weeks and 3 months) (2) Self ‐rated health (4 weeks and 3 months) (3) Social/recreational activities (baseline, 4 weeks and 3 months) (4) Social support (baseline, 4 weeks and 3 months) (5) Behaviour and mood disturbance Patient (1) Level of dependency in personal care (baseline, 4 weeks and 3 months)
Notes	Shortfall in recruitment: recruited 39/60 required
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Reported as random but specific method not reported
Allocation concealment (selection bias)	Unclear risk	Reported as concealed but specific method for concealment not reported
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: ''Caregivers did not know which group they were in when the baseline measures were completed, however this blinding could not be subsequently maintained.''
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	There was no external outcome assessor
Incomplete outcome data (attrition bias)   All outcomes	High risk	Control group lost 40% of participants
Selective reporting (reporting bias)	High risk	Intimate Bonds Measure not reported in results
Other bias	Unclear risk	Baseline data collected after randomisation

Methods	Random allocation using a computer‐generated random sequence concealed in sequentially numbered opaque sealed envelopes Blinded outcome assessment Stated intention‐to‐treat analysis 13 patients (6 treatment, 7 control) lost to follow‐up 5‐month follow‐up
Participants	Glasgow, UK 205 patients at stroke clinic or geriatric day hospital: treatment N = 100; control N = 105; completed final follow‐up N = 192 Mean age of patient: treatment 64 years; control 66 years Sex of patient: male: treatment 54%, control 50% Inclusion criteria: clinical diagnosis of stroke, TIA or amaurosis fugax commencing in the previous 3 months; 1 or more risk factors from raised BP, history of concurrent smoking, high cholesterol, diabetes (regardless of their risk factor control) Exclusion criteria: patients with cognitive impairment (defined as AMT < 5 on screening)
Interventions	Treatment: monthly review with Stroke Nurse Specialist for 3 months at which individual given advice on lifestyle changes, the importance of medication compliance and relevance to secondary prevention Focus: patient Setting: outpatient consultation Administration: reviewed by Stroke Nurse Specialist in consultation lasting approximately 30 minutes. Lifestyle issues including diet, exercise or increased activity and medical services discussed in depth and tailored to the patient's circumstances and functional abilities. Verbal information backed up by written information selected by Stroke Nurse Specialist as relevant to the individual patient. Personalised patient‐held records, detailing their risk factors and the recommended risk factor targets given to the patient and updated at each visit (considered a key part of intervention). Patients given opportunity to bring up participants as appropriate. If risk factor (e.g. BP) deemed to be at unacceptable level, patients encouraged to consult their General Practitioner with that information Control: usual care including generic risk factor advice from medical staff as well as the Stroke Nurse Specialist
Outcomes	Primary (1) Proportion of patients whose risk factors were 'on target ' Secondary (1) Survival (2) Perceived health status (3) Mood (4) Satisfaction with stroke services
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "eligible patients were randomly allocated to treatment or control groups using a computer‐generated random sequence concealed in sequentially numbered opaque sealed envelopes." Quote: "Three patients were entered twice in error, each time to the treatment group. These subjects were analysed on their initial data only and subsequent data were excluded from the analysis." Comment: errors in sequence generation could have subverted randomisation
Allocation concealment (selection bias)	Low risk	Randomly allocated to treatment or control groups using a computer‐generated random sequence concealed in sequentially numbered opaque sealed envelopes
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Single blinded trial with blinded assessment so presume unblinded participants or personnel
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "Outcomes were recorded at 5 months by an independent blinded assessor."
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Low numbers lost to follow‐up and similar across groups
Selective reporting (reporting bias)	Unclear risk	All outcomes specified in the methods were reported in the results. However, study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

Methods	Allocation by method of Taves (minimisation) Blinded outcome assessment 13 patients and carers (6 treatment, 7 control) lost to follow‐up 6‐month and 1‐year follow‐up
Participants	Seattle, WA, USA 140 stroke patients and carers (majority couples) recruited: treatment N = 70; control N = 70; completed final follow‐up: N = 127 Mean age of patient: treatment 63 years; control 62 years Sex of patient male: treatment 95%; control 94% Inclusion criteria: all stroke patients on inpatient wards from any referring service, hospitalised primarily for stroke, living with primary caregiver Exclusion criteria: none stated
Interventions	Treatment: 2 classes: (1) lecture and video 'Living with stroke', followed specific outline of information developed by psychiatrists, included basic information about the consequences of stroke; (2) explanation of treatment unique to the family's situation and questions Focus: carer Setting: hospital Administration: occupational therapist (class 1), social worker (class 2). 2 x 1‐hour classes during third week of stroke; second class within 3 working days of the first Control: routine care
Outcomes	(1) Knowledge of stroke (6 months and 1 year) (2) Family function (6 months and 1 year) (3) Patient adjustment (6 months and 1 year) (4) Use of social resources (6 months and 1 year)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Patients were randomly assigned to conditions after minimizing the differences for variates known to predict stroke recovery: mood, self‐care ability (Barthel Index), mental status, age, and location of the lesion. The method of Taves[14] was used."
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not reported
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No information reported. However, as no alternative intervention for control groups, blinding of participants not possible
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	No report of blinded assessment
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Small numbers lost to follow‐up with similar reasons reported
Selective reporting (reporting bias)	Unclear risk	Study protocol not available so cannot assess reporting bias
Other bias	Unclear risk	Imbalance in reported baseline conditions (marital status and number in household) may mean choice of minimisation factors was incomplete

Methods	Randomisation using independently prepared envelopes Outcome assessment not blinded 2 patients (1 treatment, 1 control) lost to follow‐up 1 month follow‐up
Participants	Fife, UK 41 stroke patients: treatment N = 20; control N = 21; completed final follow‐up: N = 39 Mean age of patient: treatment 64 years; control 64 years Sex of patient: male: treatment 53%; control 50% Inclusion criteria: stroke within 24 months of recruitment, fluent in English, not aphasic, not cognitively impaired Exclusion criteria: none stated
Interventions	Treatment: workbook designed to increase perceptions of control by giving information, enhancing coping resources and rehearsing planning and problem‐solving skills. Recovery plan developed with researcher. Weekly phone call (over 3‐week period). First part of workbook dealt largely with information about stroke, causes, management, and recovery. Additional sections of relevance to the individual available (e.g. on diet, smoking). Second part introduced methods of coping and relaxation tape and instructions for use Focus: patient Setting: patients' home Administration: workbook introduced in 2 parts: part 1 introduced following baseline assessment; patients asked to work through the sections, answering quizzes and deciding which additional sections were relevant to them; part 2 introduced 1 week later along with relaxation tape and instructions for use. Requests for additional parts of the workbook met. A recovery plan consisting of a daily task with records made as joint exercise between researcher patient, and carer. Over next 3 weeks patient and carer worked independently on workbook and received weekly telephone call from researcher to enquire about progress and give opportunity to ask questions Control: wait control group received the workbook once the study was complete
Outcomes	(1) Functional limitations (1 month) (2) Mood (1 month) (3) Perceived control (1 month)
Notes	Validity assessment: No stated intention‐to‐treat analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of randomisation not reported
Allocation concealment (selection bias)	Low risk	Used an enveloped prepared independently of the interviewer
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No mention of blinding; treatment group received a workbook and control group received nothing until end of trial – would have been obvious which group they were in
Blinding of outcome assessment (detection bias)   All outcomes	High risk	The intervention and assessment were undertaken by the same individual
Incomplete outcome data (attrition bias)   All outcomes	Low risk	2 of 41 lost to follow‐up, 1 from each group, both unavailable
Selective reporting (reporting bias)	Unclear risk	All outcome measures specified in methods were reported. However, study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

Methods	Randomisation using predetermined computer‐generated randomisation sequence Balanced block design where randomisation occurred in blocks of 4 Blinded outcome assessment Stated intention‐to‐treat analysis 5 patients (3 treatment, 2 control) lost to follow‐up 3‐month follow‐up
Participants	Brisbane, Australia 138 stroke patients: treatment N = 69; control N = 69. Completed final follow‐up N = 133 Mean age of patients: treatment 67 years; control 69 years Sex of patients male: treatment 64%; control 46% Inclusion criteria: diagnosed stroke or TIA, medically stable, reported English‐proficiency level, corrected hearing and vision and communication status adequate to participate in an interview and complete assessment tasks, no reported or observable dementia, living within 50 km of the hospital Exclusion criteria: none stated
Interventions	Treatment: computer‐generated tailored written information, customised according to patients' informational needs. 34 topics available covering such issues as: how stroke occurs, risk factors, understanding and managing the effects of stroke, reducing stroke risk, treatment and rehabilitation and managing after discharge Focus: stroke patients Setting: stroke unit Administration: within 1 day of baseline interview the research nurse completed the 'what you need to know about stroke' checklist with the patient. Further information given as needed about the scope and content in each of the available topics. Once the checklist completed, the research nurse entered topic selections, desired version of each topic (detailed, shortened) and desired font size into the database. Then generated and printed an individualised booklet and placed into a ring‐binder folder. Patient name written on booklet and given to the patient Control: Within 1 day of the baseline interview, provided by research nurse with a copy of the Stroke Association of Queensland fact sheet
Outcomes	(1) Knowledge of stroke (3 months) (2) Mood (3) Self efficacy (3 months) (4) Perceived health status (3 months) (5) Use of and satisfaction with information
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "..database randomly assigned the patient to either the intervention or control group."
Allocation concealment (selection bias)	Low risk	Quote: "One of the database tables contained a predetermined computer generated randomisation sequence, thus ensuring concealed allocation."
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	No report of blinding of participants and may not have been obvious to participants which group they were in as both received written information. However, remains unclear
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "An outcome assessor who was blind to patients’ group allocation, conducted baseline interviews while the patient was in hospital, and follow‐up interviews 3 months after discharge."
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Low number of losses to follow‐up and numbers balanced across groups, with similar reasons
Selective reporting (reporting bias)	Unclear risk	Study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

Methods	Matched pairs design based on baseline scores on outcome measures, age, sex and side of stroke. Random assignment within each pair by tossing a coin Not blind outcome assessment All participants reassessed 1 week after intervention group completed a 4‐week course No losses to follow‐up
Participants	Minneapolis, USA 41 stroke patients identified from hospital‐based register of stroke survivors Treatment N = 21; control N = 20. Completed final follow‐up N = 41 Mean age: treatment 64.2 years; control 63.9 years Sex of patient male: treatment 38%; control 50% Inclusion criteria: > 18 years of age, English speaking, community dwelling, stroke 6 months to 3 years earlier, gave informed consent
Interventions	Treatment: 8 x 2‐hour structured educational classes over a 4‐week period. Content included facts on stroke, living with disability, exploring spiritual wellness Control group offered the intervention after the end of the evaluation
Outcomes	(1) Self‐reported state of depression (2) Self‐reported sense of hope (3) Self‐reported ways of coping
Notes	Match pairs design then randomisation by toss of a coin, not concealed Unpaired participant assigned to the treatment group
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Reported that one member of each pair was randomly assigned to either the treatment or control group but method not reported
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not reported
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No report of blinding of participants and personnel. Control group received usual care (compared with structured education course) so may have been obvious they were not receiving an intervention
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	No report of blinded outcome assessment
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Unclear if outcomes were reported for all participants
Selective reporting (reporting bias)	Unclear risk	Study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

Methods	RCT. A statistician prepared 2 separate randomisations for patients with carers who also agreed to participate (carer‐patient subgroup) and for carers partnered with a patient who could not participate because of cognitive and communication impairments (carer‐only subgroup) Blinded outcome assessment Reported intention‐to‐treat analysis 45 patients (29 intervention, 16 control) and 42 carers (total across intervention and control groups) lost to follow‐up 6‐month follow‐up
Participants	Dundee, Scotland 203 acute stroke patients and 172 carers Patients: intervention N = 103; control N = 100 Carers: intervention N = 82; control N = 90 Mean age of patients: intervention 69 years; control 69 years Sex of patient male: intervention 61%; control 61% Mean age of carers: intervention 63 years; control 61 years Carer sex male: 35% (across intervention and control groups) Inclusion criteria patient: fluent in English; discharged from hospital following stroke. Carers identified by the patient as the person most involved in their care Exclusion criteria: not reported
Interventions	Intervention: post‐discharge workbook intervention delivered by a workbook implementer over a 5‐week period. The workbook provided information about stroke and recovery; guidance on coping skills; and self‐management instruction. Task materials (e.g. goal setting), diary sheets and an audio relaxation cassette tape that described simple body relaxation and breathing exercises. Intervention included 2 home visits and 2 telephone contacts. Intervention started within 3 weeks (approximately) of hospital discharge Focus: patient and carer Setting: home Administration: work book implementer Control: usual care
Outcomes	Patient outcomes (1) Disability (baseline, 8 weeks post‐intervention and 6 months after baseline) (2) Anxiety and depression (baseline, 8 weeks post‐intervention and 6 months after baseline) (3) Perceived control (completed by the patient and by the carer on behalf of the patient at baseline and 8 weeks) (4) Satisfaction with treatment and advice (8 weeks post‐intervention and 6 months after baseline) (5) Confidence in recovery (baseline) Carer outcomes (1) Anxiety and depression (baseline, 8 weeks post‐intervention and 6 months after baseline). (2) Health related quality of life (baseline)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information reported
Allocation concealment (selection bias)	Unclear risk	Insufficient information reported
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Patients asked not to disclose group allocation though potentially broken. Participants would have been aware they were receiving the intervention
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Research assistants blind to the process
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	More losses to follow‐up in the intervention group (29 out of 103) compared with the control group (16 out of 100). 42 carers lost (only reported across groups)
Selective reporting (reporting bias)	High risk	Observer Assessed Disability (OAD) not reported at baseline. Hospital and Anxiety Depression Scale (HADS) sub‐scales reported for anxiety and depression at baseline but combined post intervention
Other bias	Unclear risk	No other obvious signs of bias

Methods	Block randomisation procedures, each block included 10 participants. Allocation schedule prepared in advance using computer‐generated random numbers. Allocation codes held in central office remote from study environment Blinded outcome assessment Stated intention‐to‐treat analysis 32 patients and caregivers lost to follow‐up: treatment N = 17; control N = 15 12‐month follow‐up
Participants	London, UK 300 patients and caregivers: treatment N = 151; control N = 149. Completed final follow‐up: N = 268 Median age of patient: treatment N = 76 years ; control N = 76 years Sex of patient male: treatment 57%; control 50% Inclusion criteria: patient ‐ independent in daily living activities before the stroke, medically and neurologically stable at time of baseline assessments, expected to return home with residual disability; carer ‐ no notable disability (Rankin score 0 to 2), willing and able to provide support after discharge Exclusion criteria: none stated
Interventions	Treatment: conventional care plus 3 to 5 sessions of 30 to 45 minutes comprising instruction by appropriate professional on common stroke‐related problems and their prevention, management of pressure areas and prevention of bed sores, continence, nutrition, positioning, gait facilitation, advice on benefits and services. 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 the individual patients Focus: caregivers Setting: stroke rehabilitation unit Administration: training started when patients' rehabilitation needs stabilised and discharge contemplated. Caregivers competencies assessed at the end of training. Follow‐through session conducted by hospital team at home to adapt skills learnt to home environment Control: conventional care consisting 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 patient's abilities and informal instruction on facilitating mobility and activities of daily living tasks; advice on community services, benefits, and allowances including contact information for voluntary support services for caregivers
Outcomes	Patients (1) Death or institutionalisation (3 and 12 months) (2) Function (3 and 12 months) (3) Mood (3 and 12 months) (4) Quality of life (3 and 12 months) Caregiver (1) Function and social activities (3 and 12 months) (2) Emotional health (3 and 12 months) (3) Quality of life (3 and 12 months) Economic (1) Health and social care costs (12 months) (2) Informal care costs (12 months) (3) Quality adjusted life years in caregivers (over 1 year)
Notes	Validity assessment: Possibility of limited generalisability (setting was largely middle‐class suburban area)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation
Allocation concealment (selection bias)	Low risk	Quote: "Allocation codes were held in a central office remote from the study environment"
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Participants were aware of training received
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Reported that an observer who did not participate in allocation or management of patients assessed outcome
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Some missing data. However, numbers and reasons for missing data relatively balanced across groups
Selective reporting (reporting bias)	Unclear risk	Study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

Methods	Randomisation performed by authors using blocks of 20 participants, where 10 would be allocated to each arm of the trial and the sequence could not be predicted Outcome assessment by self‐rated questionnaires 9 carers(4 treatment, 5 control) lost to follow‐up 6‐month and 1‐year follow‐up
Participants	Stockholm, Sweden 100 spouses of stroke patients: treatment N = 50; control N = 50. Completed final follow‐up: N = 91 Mean age of spouse: treatment 68 years; control 67 years Sex of spouse female: treatment 76%; control 84% Inclusion criteria: spouse of stroke patient (defined as person living in the same household as the stroke patient) Exclusion criteria: not possible to obtain information from the spouse and/or if patient not able to return home after hospitalisation
Interventions	Treatment: support and education programme led by stroke specialist nurses and group discussion with issues raised by participants. Topics included: the nature of stroke, treatment and recovery, psychological and social effects, how to prevent recurrence. Participants able to call the stroke specialist nurse between sessions to get extra information or support Focus: spouse of stroke patient Setting: hospital Administration: groups of 10, attended 6 times in 6 months. Session commenced with lecture on 1 of the topics for 20 to 30 minutes, followed by group discussion Control: regular information during hospitalisation and also at discharge. Possibility of attending 1 open session of 1.5 hours by a stroke physician on the ward (only 3 control participants chose this option)
Outcomes	(1) General Quality of Life (6 and 12 months) (2) Life situation (6 and 12 months) (3) General well‐being (6 and 12 months) (4) Perceived health state (6 and 12 months)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomly assigned to intervention or control group but method of sequence generation not reported
Allocation concealment (selection bias)	Unclear risk	Reported that sequence could not be predicted but method of allocation concealment not reported
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No placebo intervention for control group
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Unclear whether an interviewer was used or whether questionnaires were self‐completed. No report of blinding
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Losses to follow‐up low (10% control and ˜2% treatment) but reasons not provided
Selective reporting (reporting bias)	Unclear risk	All outcomes described in the methods reported in the results. However, study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

Methods	Patients randomly selected to receive leaflets, no further details given No stated blind outcome assessment No reported losses to follow‐up 1‐week follow‐up
Participants	Southampton, UK Numbers unclear; report states that 73 stroke incidents were assessed No participant characteristics reported Inclusion criteria: admission to medical or geriatric wards of the 2 major teaching hospitals in Southampton, clinical diagnosis of stroke Exclusion criteria: discharge within 7 days of admission, severe illness, aphasia or dysphasia that prevents response to interview, lack of awareness that have had a stroke
Interventions	Treatment: 12‐page leaflet prepared for study personalised with name, sections on basic pathologies of stroke, predisposing factors, treatment, recovery, facilities available in the community, and financial benefits available Focus: patient and relative Setting: hospital Administration: presented to patient by a medical student with no explanation other than the leaflet may be interesting for them and their relatives to read. 1 contact, length of time unknown, between 1 and 2 weeks after admission Control: usual care, no leaflet
Outcomes	(1) Knowledge of stroke (1 week)
Notes	Validity assessment: comparability of treatment and control groups unknown as no reporting of participant characteristics
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of randomisation not reported
Allocation concealment (selection bias)	Unclear risk	Method for allocation concealment not reported
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	No placebo intervention for control group. However, participants did not appear to be informed of the study when they were provided with the leaflet and staff were not informed who received the leaflet. However, blinding may have been broken
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Unclear if blinded outcome assessment was undertaken
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	A number of exclusions and unclear at which time point in the study
Selective reporting (reporting bias)	Unclear risk	All outcomes described. However, study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

Methods	Randomised using sealed opaque envelopes in blocks of 10 and 1 to 1 ratio. Envelopes prepared by independent researcher Blinded outcome assessment Stated intention‐to‐treat analysis 16 patients (6 treatment, 10 control) lost to follow‐up 3 and 6‐month follow‐up
Participants	Liverpool, UK 100 stroke patients: treatment N = 50; control N = 50. Completed final follow‐up: N = 84 Median age of patient: treatment 68 years; control 73 years Sex of patient female: treatment 42%; control 38% Inclusion criteria: Confirmed stroke, all ages, either sex, patients who are discharged home and who can complete a questionnaire, or who have a named carer who can do so Exclusion criteria: pre‐existing cognitive impairment, discharge to institutionalised care, discharge home but unable to self‐complete questionnaire and no named carer
Interventions	Treatment: CareFile (A5 size laminated 29 page booklet). Includes general information about stroke as well as information personal to the patient, secondary prevention measures, and personal goals aimed at reducing risk of further stroke. Also contains useful telephone numbers for all stroke‐related services and local support agencies. Design allows for removal of pages not relevant to the individual. Sections included for members of the multi‐disciplinary team to complete summaries of patient's achievements and future rehabilitation goals. Also provided with advice from therapists and offered leaflets from Chest, Heart and Stroke Association Focus: patient Setting: hospital ward Administration: interview arranged between researcher and patient when patient discharge date in place. Carer also invited to attend. The CareFile and its contents explained by the research registrar and any additional concerns or issues addressed in discussion lasting approximately 15 to 20 minutes. Patients advised to take the CareFile with them to all General Practitioner and clinic appointments Control: received the usual stroke information leaflets provided by the stroke unit and follow‐up in stroke review clinic
Outcomes	Primary (1) Knowledge of stroke (3 and 6 months) Secondary (1) Utilisation of CareFile (3 and 6 months) (2) Satisfaction with information given (3 and 6 months) (3) Blood pressure (3 and 6 months) (4) Participation (3 and 6 months) (5) Screening question for depression (3 and 6 months)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Reported that eligible patients were randomised but method not reported
Allocation concealment (selection bias)	Low risk	Reported to have used sealed opaque envelopes
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No report of blinding of participants or personnel
Blinding of outcome assessment (detection bias)   All outcomes	High risk	Outcome assessors do not appear to have been blinded – "those in the intervention group were asked if they had brought the CareFile to the Review Clinic and if they found it useful"
Incomplete outcome data (attrition bias)   All outcomes	High risk	Almost twice as many lost to follow‐up in the control group (10/50) compared with the intervention group (6/50)
Selective reporting (reporting bias)	Unclear risk	Study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

Methods	Random allocation in blocks of 10 using computer‐generated random numbers. Size of blocks unknown to investigators at time of trial Blinded outcome assessment Intention‐to‐treat analysis 7 patients lost to follow‐up plus 1 withdrawn (results not reported) as breached inclusion criteria. Differential losses between the groups unclear follow‐up at 1 and 12 weeks
Participants	Rotterdam, Netherlands 65 patients at TIA/minor stroke clinic: treatment N = 33; control N = 32. Completed final follow‐up: N = 58 (results reported for 57) Mean age of patient: treatment 65 years; control 63 years Sex of patient male; treatment 57%; control 63% Inclusion criteria: 18 years or older, TIA or minor Ischaemic stroke within last 3 months, speak/write Dutch fluently, modified Rankin score < 4 Exclusion criteria: professionally engaged in cardio‐vascular health education, aphasia, dementia, visual impairment to a degree that would interfere with health education delivery
Interventions	Treatment: discussion of test results and standard education by physician plus IMCP comprising of modules containing lay information for each of 8 modifiable risks. All modules highly structured and contained combinations of slides shows, background voice and personal address Focus: patient Setting: outpatient clinic Administration: after consultation with physician shown IMCP. Given brief introduction. 1 of 2 versions used according to age and educational level: general introduction of their personal diagnosis, explanation of the used or prescribed medications, then each patient shown 4 risk factor modules, or if has less than 4 risk factors general information about frequent vascular risk factor, printed summary of the information Control: discussion of test results and standard health education
Outcomes	(1) Knowledge at 1 (primary) and 12 weeks (secondary) post intervention (2) Function (12 weeks) (3) Changes in cholesterol level, weight, cigarette, and alcohol consumption and physical activity (12 weeks)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Treatment allocation was random, and based on computer‐generated random numbers."
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not reported
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No report of blinding of participants or personnel and no placebo intervention provided
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	No report of blinded assessment
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Eight of the 65 participants were lost in total and unclear which groups they were lost from
Selective reporting (reporting bias)	High risk	Blood pressure was not a pre‐specified outcome but has been reported
Other bias	Low risk	No other obvious sources of bias

Methods	Randomisation performed by telephone in computer‐generated blocks of 10 using sequentially‐numbered opaque envelopes Blinded outcome assessment Stated intention‐to‐treat analysis 22 patients (11 treatment, 11 control) and 7 carers (4 treatment, 3 control) lost to follow‐up 6‐month follow‐up
Participants	Oxford, UK 93 stroke patients: treatment N = 48; control N = 45. Completed final follow‐up: N = 71 56 carers of these patients: treatment N = 32; control N = 24. Completed final follow‐up: N = 49 Mean age of patient: treatment 70 years; control 76 years Sex of patient male: treatment 65%; control 65% Inclusion criteria: Oxfordshire resident, admission to any Oxford hospital, stroke within past month (could be recurrent) Exclusion criteria: identified over 1 month after stroke, death within 1 month of admission or considered likely to occur prior to follow‐up, taking part in another trial involving follow‐up interview, dysphasic with no close informal carer, stroke not the major medical problem, admitted from a nursing home, subdural, subarachnoid haemorrhage when no accompanying intracerebral haemorrhage, TIA
Interventions	Treatment: a collection of 8 leaflets published by the Stroke Association assembled in an A5 folder covering what a stroke is, effects, cause, problems that might be experienced and how they might be dealt with. An introductory leaflet was specially prepared plus leaflets giving local and national contact names and addresses of support groups and services Focus: patient and closest informal carer if available Setting: home Administration: pack addressed to both patient and carer (where applicable). No contact at delivery. Sent to home address 1 week after randomisation (4 to 5 weeks after stroke). Pack left with patient and carer for 6 months Control: received nothing
Outcomes	(1) Knowledge of stroke (6 months) (2) Emotional outcome (6 months) (3) Perceived health status and quality of life (6 months) (4) Satisfaction with information and care received (6 months) (5) Disability and Handicap (6 months) (6) Service use (6 months)
Notes	Validity assessment: treatment and control groups not balanced in respect of age
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated random sequence generation
Allocation concealment (selection bias)	Low risk	Sequentially numbered opaque envelopes
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No intervention provided for the control group. As a result a high risk of bias
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Quote: "While in theory the interviewer was blinded to the treatment allocation, in practice she guessed the correct status of the patients more often than might be expected by chance." Correct guessing may indicate blinding was unsuccessful or that the outcome assessor was noticing real differences between participants
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Losses to follow‐up balanced in numbers and reasons across groups
Selective reporting (reporting bias)	Unclear risk	All measures described in the methods were reported in the results. However, study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

PERMALINK

Information provision for stroke patients and their caregivers

Anne Forster

Lesley Brown

Jane Smith

Allan House

Peter Knapp

John J Wright

John Young

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Resource outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Results

Description of studies

Results of the search

Included studies

Setting

Participants

Interventions

Category

Content and administration

Timing

Outcomes measured

Assessment of knowledge

Excluded studies

Ongoing studies

Studies awaiting assessment

Risk of bias in included studies

Method of analyses

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Patient outcomes

Knowledge

Patient knowledge

1.1. Analysis.

Subgroup analysis

Emotional outcomes

Anxiety

Patient emotional outcome: anxiety (dichotomised data)

1.2. Analysis.

Patient emotional outcome: anxiety (continuous data)

1.3. Analysis.

Methods	RCT utilising computer‐generated randomised sequence Blinded outcome assessment 27 patients (18 intervention, 9 control) lost to follow‐up
Participants	Melbourne, Australia 93 stroke patients (intervention N = 46; control N = 47) Age mean (range): 73 years (32.1 to 91.3) 33 males and 33 females completed second post intervention follow‐up (sex details not reported at baseline) Inclusion criteria: > 18 years, able to be discharged home, English proficiency, adequate communication for interview, corrected vision and hearing, no evidence of severe cognitive impairment Exclusion criteria: none reported
Interventions	Intervention: patient held‐record (PHR) which included contact details, questions for health professionals, notes on care, useful phone numbers, brochures from the national stroke foundation and fact sheets relating to specific problems associated with their stroke, level of disability and symptoms (movement and balance, swallowing difficulties, continence, driving and vision, mood changes, pain, sexuality, speech and communication). In addition, usual discharge information (health summary sheet listing medication) Focus: patient Setting: hospital prior to discharge Administration: trained health care researcher Control: usual discharge information (health summary sheet listing medication)
Outcomes	(1) Stroke Impact Scale (4 weeks and 4 months post intervention) (2) PHR evaluation questionnaire (unclear when administered)
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation procedure
Allocation concealment (selection bias)	Low risk	Computer‐generated randomisation procedure. External researcher held randomisation codes
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Insufficient information provided
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Whilst it was reported that the outcome assessor was blinded, one of the measures appeared to be given to the intervention group only which would have compromised assessor blinding (not explained how this was overcome). As a result unclear
Incomplete outcome data (attrition bias)   All outcomes	High risk	One‐third lost to follow‐up. More lost in the intervention group and reasons for losses not reported
Selective reporting (reporting bias)	High risk	Protocol unavailable so cannot determine if all outcomes have been reported. No pre‐intervention data reported
Other bias	High risk	This trial was terminated early as a number of the intervention participants were unable to recall receiving the information

Methods	Computer‐generated randomised allocation. Stratified for side of cerebrovascular accident Blinded outcome assessment 6 patients (2 treatment, 4 control) lost to follow‐up 3‐month follow‐up
Participants	Southampton, UK 36 stroke patients and carers (couples): treatment N = 21; control N = 15. Completed final follow‐up: N = 30 Age of patient: number < 65 years: treatment N = 8; control N = 4; number > 65 years: treatment N = 13, control N = 11 Sex of patient male: treatment N = 16; control N = 9 Inclusion criteria: admission to hospital with a CVA as defined by WHO, discharge home after a minimum period of treatment of 10 days to live with a relative or carer, agreement to participate in the study Exclusion criteria: none stated
Interventions	Treatment: individualised booklet containing information on persisting symptoms, current aims of rehabilitation, instructions concerning ADLs, description of exercises provided, pertinent photos, useful local and national addresses and contacts. Also provided with advice from therapists and offered leaflets from Chest, Heart and Stroke Association Focus: patient and carer Setting: home Administration: no contact at delivery. Sent within 7 days of discharge (> 17 days post‐stroke) by research therapist to home address Control: provided with advice from therapists and offered leaflets from Chest, Heart and Stroke Association
Outcomes	(1) Knowledge of stroke (3 months) (2) Satisfaction with information received (3 months) (3) Disability and Handicap (3 months)
Notes	Validity assessment: unequal numbers in treatment and control, participants in the treatment group had higher levels of impairment and co‐morbidity
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of random sequence generation not reported
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Therapists were not informed which patients were to receive the booklets. No report of blinding of participants, both groups were provided with advice and offered leaflets in hospital, only treatment group received booklets after discharge – may not have been obvious which group they were in
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Social services occupational therapists who were blind to the trial and control groupings undertook the interviews
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Slightly more losses to follow‐up in intervention group (4/21) than control group (2/15); reasons only given for group as a whole so cannot determine if reasons were similar between groups
Selective reporting (reporting bias)	Unclear risk	All outcomes described in the methods reported in the results. However, study protocol not available so cannot assess reporting bias
Other bias	Low risk	No other obvious sources of bias

Methods	Randomisation by a centralised telephone service. Randomised by computer initially in blocks of 8, stratified by presence of informal carer and incontinence of urine at 24 hours post stroke Blinded outcome assessment Stated intention‐to‐treat analysis 50 patients (31 treatment, 19 control) and 70 carers (42 treatment, 28 control) lost to follow‐up 6‐month follow‐up
Participants	North Tyneside, UK 204 stroke patients: treatment N = 121; control N = 83. Completed final follow‐up: N = 154 176 informal carers of these patients: treatment N = 107; control N = 69. Completed final follow‐up: N = 106 Median age of patients: treatment 74 years, control 76 years Sex of patient male: treatment 49%; control 46% Inclusion criteria: confirmed diagnosis of stroke, medically stable, normally resident in North Tyneside, not in residential home prior to admission, still in hospital within 48 hours of admission Exclusion criteria: none stated
Interventions	Treatment: 7 group sessions (1 during inpatient stay and 6 outpatient) covering the experience and nature of stroke, the role of physiotherapy and occupational therapy, psychological effects, caring, communication and swallowing problems, reducing risk. Leaflet with telephone number of stroke help line, Stroke Association, day hospital and stroke units Focus: patients and informal carer Setting: stroke unit and day hospital Administration: a rolling programme held 7 times during course of study. Presentation by speaker at each session followed by questions and discussion. Opportunity to ask questions at beginning or end of session. Inpatient session 1 x 1 hour, 6 x 1 hour outpatient sessions over 6‐week period Control: usual care. All given a basic 2‐sided leaflet about North Tyneside stroke service plus staff prompted to provide information about stroke on day of admission and at regular intervals throughout stay. Record of communication and Stroke Association literature available. Given details of telephone hotline run by the stroke service prior to discharge
Outcomes	Primary (1) Perceived health status (6 months) Secondary (1) Knowledge of stroke (6 months) (2) Emotional outcome (6 months) (3) Stress of care‐giving (6 months) (4) Satisfaction with hospital services and discharge (6 months) (5) Disability and handicap (6 months) (6) Service use (6 months)
Notes	Validity assessment: large losses to follow‐up
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated random number sequence
Allocation concealment (selection bias)	Low risk	Central telephone service used
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No placebo intervention for the control group
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "interviewed in their own homes at 6 months after stroke by a researcher who was blinded to the randomisation group."
Incomplete outcome data (attrition bias)   All outcomes	High risk	Non‐attenders and attenders included in analysis as is appropriate. Approximately 25% lost to follow‐up with relatively similar numbers and reasons across groups. However, due to dysphasia or cognitive problems the primary outcome (SF‐36) could not be completed by another 37 patients (24%) meaning almost half of these outcomes were missing. Also, approximately 40% of carers were lost to follow‐up
Selective reporting (reporting bias)	Unclear risk	All outcomes described in the methods reported. However, study protocol not available so cannot assess reporting bias
Other bias	Unclear risk	Only 51 patients (42%) of those randomised attended 3 or more of the 6 outpatient sessions

Methods	Patients randomly allocated using random length restricted permuted blocks (block lengths of 2, 4, and 6). Randomisation carried out by independent research assistant by using sealed, numbered , opaque envelopes kept in a locked separate location. Stratified by Barthel Index scores of 0 to 4, 5 to 9, 10 to 14, 15 to 19, presence of aphasia, and presence of a carer Blinded outcome assessment Stated intention‐to‐treat analysis 37 patients (15 treatment, 22 control) and 21 carers (9 treatment, 12 control) lost to follow‐up 3 and 6‐month follow‐up
Participants	Bradford, UK 170 stroke patients: treatment N = 84; control N = 86. Completed final follow‐up: N = 133 97 carers of these patients: treatment N = 49; control N = 48. Completed final follow‐up: N = 76 Median age of patients: treatment 75 years; control 74 years Sex of patients female: treatment N = 46%; control N = 52% Inclusion criteria: all patients admitted to the stroke rehabilitation unit with a confirmed diagnosis of stroke Exclusion criteria: patients with receptive aphasia, cognitive impairment or who did not understand English and did not have a carer
Interventions	Treatment: provided with the Stroke Recovery Programme, a specifically devised manual containing information about causation and consequences of stroke, recovery, financial benefits, relevant services, and a specific section for carers. Also invited to attend specifically convened meetings with members of their multidisciplinary team (doctor, nurse, physiotherapist, occupational therapist). The intention of the meeting was to provide background information about stroke, discuss patient's progress, answer specific questions, and develop shared rehabilitation goals. Focus: patient but when the patient had receptive aphasia, cognitive impairment or did not understand English the carer was the main focus Setting: stroke unit Administration: Stroke Recovery Programme given by stroke unit staff following randomisation. Meetings scheduled to last approximately 20 minutes held in the ward dayroom fortnightly for duration of stroke unit stay. Guidelines developed for use by rehabilitation teams to ensure coverage of the of the key topics included in the Stroke Recovery Programme and record of matters discussed completed following each meeting. Agreed goals recorded in the manual and retained by the patient Control: Received usual practice. A folder of information about stroke causation, consequences and recovery previously devised by ward staff and stroke association leaflets were available
Outcomes	Primary (1) Knowledge of stroke (3 and 6 months)   Secondary (1) Physical function (3 and 6 months) (2) Social function (3 and 6 months) (3) Handicap (3 and 6 months) (4) Patient mood (3 and 6 months) (5) Carer mood (3 and 6 months) (6) Patient and carer satisfaction (3 and 6 months) (7) Use of services and receipt of benefits (6 months)
Notes	Validity assessment: losses to follow‐up 22%
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of sequence generation not reported
Allocation concealment (selection bias)	Low risk	Quote: "Concealed randomisation was achieved using sealed, numbered, opaque envelopes kept in a locked separate location by an independent research assistant who carried out the randomisation and conveyed patient allocation information to the stroke unit co‐ordinator."
Blinding of participants and personnel (performance bias)   All outcomes	High risk	No report of blinding of participants and groupings probably obvious‐treatment group attended meetings, control group received usual care
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Patients and carers were followed up by a research nurse who was blind to group allocation
Incomplete outcome data (attrition bias)   All outcomes	Low risk	˜22% lost to follow‐up with similar reasons and proportions across groups.
Selective reporting (reporting bias)	Unclear risk	All outcomes described in the methods reported in the results. However, study protocol not available so cannot assess reporting bias
Other bias	High risk	Quote: "unavoidable contact and associated intervention contamination between the two groups of patients and relatives during the inpatient period"

Study	Reason for exclusion
Adie 2010	The intervention included motivational interviewing
Allen 2009	The information/education provision was part of a more complex rehabilitation intervention
Andersen 2002	The information/education provision was part of a more complex rehabilitation intervention
Ayana 2000	Non‐random
Bacchini 2011	The information/education provision was part of a more complex rehabilitation intervention
Bakas 2008	The information/education provision was part of a more complex rehabilitation intervention
Battersby 2009	The information/education provision was part of a more complex rehabilitation intervention
Boter 2004	The information/education provision was part of a more complex rehabilitation intervention
Boysen 2007	The information/education provision was part of a more complex rehabilitation intervention
Brier 2011	Non‐random
Brotons 2007	The trial included participants with conditions other than stroke and the data were not available separately
Burton 2005	The information/education provision was part of a more complex rehabilitation intervention
Byers 2010	The intervention included motivational interviewing
Chaiyawat 2009	The information/education provision was part of a more complex rehabilitation intervention
Chang 2000	The information/education provision was part of a more complex rehabilitation intervention
Chang 2011	The information/education provision was part of a more complex rehabilitation intervention
Cheng 2011	The information/education provision was part of a more complex rehabilitation intervention
Christie 1984	The information/education provision was part of a more complex rehabilitation intervention
Chumbler 2011	The information/education provision was part of a more complex rehabilitation intervention
Claiborne 2006	The information/education provision was part of a more complex rehabilitation intervention
Clark 2003	The information/education provision was part of a more complex rehabilitation intervention
Clarke 2011	The information/education provision was part of a more complex rehabilitation intervention
Dennis 1997	The information/education provision was part of a more complex rehabilitation intervention
Desrosiers 2007	The information/education provision was part of a more complex rehabilitation intervention
Dongbo 2003	Study included both stroke and non‐stroke patients and data not available separately
Ertel 2007	The information/education provision was part of a more complex rehabilitation intervention
Evans 1984	Non‐random
Folden 1993	(1) Not information/education; this was a study of goal setting (2) Non‐random
Forster 1996	The information/education provision was part of a more complex rehabilitation intervention
Friedland 1992	The information/education provision was part of a more complex rehabilitation intervention
Gillham 2010	The information/education provision was part of a more complex rehabilitation intervention
Glass 2004	The information/education provision was part of a more complex rehabilitation intervention
Goldberg 1997	The information/education provision was part of a more complex rehabilitation intervention
Grant 2002	The information/education provision was part of a more complex rehabilitation intervention
Grasel 2006	The information/education provision was part of a more complex rehabilitation intervention
Green 2006	The intervention included motivational interviewing
Habibzadeh 2007	(1) The information/education was part of a more complex rehabilitation intervention (2) Non‐random
Harari 2004	The information/education provision was part of a more complex rehabilitation intervention specifically targeted at improving bowel function
Harrington 2010	The information/education provision was part of a more complex rehabilitation intervention
Hartke 2003	The information/education provision was part of a more complex rehabilitation intervention
Harwood 2006	The information/education provision was part of a more complex rehabilitation intervention
Hochstenbach 1999	The information/education was part of a more complex rehabilitation intervention
Holzemer 2011	The information/education provision was part of a more complex rehabilitation intervention
Huijbregts 2009	(1) The information/education provision was part of a more complex rehabilitation intervention (2) Non‐random
Jones 2009	(1) The information/education provision was part of a more complex rehabilitation intervention (2) No control group
Kendall 2007	The information/education provision was part of a more complex rehabilitation intervention
Leathley 2003	The information/education provision was part of a more complex rehabilitation intervention
Lincoln 2003	Information provision was not the intervention evaluated The experimental condition was a support organiser
Linn 1979	The information/education provision was part of a more complex intervention specifically targeted at management of medication
Lorenc 1992	The study lacked a suitable control
Mackay‐Lyons 2010	The information/education provision was part of a more complex rehabilitation intervention
Mant 2000	Information provision was not the intervention evaluated. The experimental condition was a support worker
McKinney 1999	The information/education provision was part of a more complex intervention focused on providing feedback of cognitive assessment to patients' carers and members of the multidisciplinary team
Morrison 1998	Non‐random
Napolitan 1999	The information/education provision was part of a more complex rehabilitation intervention
Neubert 2011	No usual care group
Nguyen 2011	The information/education provision was part of a more complex rehabilitation intervention
Nir 2006	The information/education provision was part of a more complex rehabilitation intervention
Nour 2002	The information/education provision was part of a more complex rehabilitation intervention
Oupra 2010	Non‐random
Pierce 2007	The information/education provision was part of a more complex rehabilitation intervention
Printz‐Feddersen 1990	(1) The information/education was part of a more complex rehabilitation intervention (2) Non‐random
Redfern 2008	The information/education provision was part of a more complex rehabilitation intervention
Rimmer 2000	The information/education provision was part of a more complex rehabilitation intervention, which included classes in fitness and nutrition
Sahebalzamani 2009	The information/education provision was part of a more complex rehabilitation intervention
Sanguinetti 1987	The focus of the paper is head injury The data for stroke patients are not reported separately
Shyu 2008	The information/education provision was part of a more complex rehabilitation intervention
Sit 2007	Unacceptable randomisation procedure
Skidmore 2008	No control group
Tilling 2005	The information/education provision was part of a more complex rehabilitation intervention
Towle 1989	Information provision was not the intervention evaluated The experimental condition was a support worker
van den Heuvel 2000	(1) Non‐random (2) The information/education provision was part of a more complex rehabilitation intervention
Winkens 2009	The information/education provision was part of a more complex intervention (psycho‐education)

Methods	A randomised controlled trial
Participants	Stroke patients
Interventions	Education about memory after stroke, compensation strategies and psycho‐education.
Outcomes	Memory Self‐efficacy (MSE) and psychological quality of life, measured with the Metamemory In Adulthood questionnaire and the psychological domain of the WhoQol‐bref questionnaire
Notes

Methods	Controlled trial
Participants	Stroke patients
Interventions	The InfoCom booklet contains general information about aphasia, verbal and non‐verbal communications skills
Outcomes	Assessment of language deficiency (Montreal‐Toulouse‐1986) and communication skills (Test Lillois de Communication‐TLC and Protocole Toulousain d'Evaluation de la Communication au sein du Couple Aphasique‐ PTECCA).
Notes

Methods	Multi‐site mixed methodology pilot randomised controlled trial
Participants	Caregivers
Interventions	Stroke family support program
Outcomes	Outcome measures not reported
Notes

Methods
Participants	Primary caregivers of stroke patients
Interventions	Education classes delivered by a researcher
Outcomes	Knowledge
Notes

Methods	Single blind randomised controlled trial
Participants	Clients or carers of clients with a current admission for stroke
Interventions	The education and support package consists of a written education booklet that provides tailored information, supplemented by verbal reinforcement and repetition of the information. Verbal reinforcement will occur face‐to‐face (prior to hospital discharge) and over the telephone (after hospital discharge) for up to 3 months post‐discharge. The written education booklet contains topics including the definition, causes, warning signs, risk factors, effects, diagnosis and treatment of stroke, as well as rehabilitation, recovery, returning to activities, going home, practical management strategies and services and support available after stroke
Outcomes	Primary outcome measures: stroke‐related knowledge as determined by a stroke knowledge questionnaire Secondary outcome measures: stroke‐risk factor awareness (as assessed by an opened ended question and a checklist of stroke‐related risk factors requiring a yes/no/unsure response), self‐efficacy (using measures designed for this study), stroke risk‐related behaviour change, Anxiety and depression (Hospital Anxiety and Depression Scale), client quality of life (using the Stroke and Aphasia Quality of life Scale‐39) and carer burden (Caregiver Strain Index), satisfaction (questions regarding satisfaction and usefulness of information received)
Notes