Updated meta-review of evidence on support for carers

Post-protocol decisions prior to data extraction

From the initial searches and selection, it was clear that there had been substantial development in the volume, content and complexity of the literature since the original meta-review was carried out in 2010. Consequently, > 100 reviews were selected for potential inclusion in the update. It appeared that the average quality of reviews identified had improved since the first review, which potentially offered the opportunity for a ‘best evidence’ approach. Given the time and resource available, we needed to find a way to focus attention on those reviews that would provide the most robust information. Therefore, we revisited a number of decisions from the original protocol to focus our work.

The following post-protocol issues were discussed and agreed.

1. Review protocols were excluded on the basis that:

   1. in their current form they were not a published systematic review and, therefore, they failed to meet our inclusion criteria

   2. CRD’s PROSPERO database had been searched to locate publications from relevant protocols (we contacted the authors of such protocols but there were no available publications)

   3. for older protocols we would expect our search strategy to have picked up relevant publications (given that not everything is registered on PROSPERO).

2. Conference abstracts were excluded on the basis that:

   1. in their current form they were not a full systematic review publication and, therefore, they failed to meet our inclusion criteria

   2. they did not provide sufficient detail to allow them to be assessed for inclusion

   3. we were confident that our robust search strategy would have identified any relevant reviews underpinning these conference abstracts.

3. We found one review15 of interventions for carers of people with delirium. After discussion, we excluded this on the basis that delirium is an acute condition that would be expected to resolve. ‘Carers’ of people with delirium might thus be so for a very short period of time, whereas the focus of our work, and of NIHR’s interest, was on people who carry caring responsibility over an extended period.

4. We also found three reviews16–18 of case or care management. Although these examined outcomes for carers, they were excluded for two reasons:

   1. Case or care management, as currently understood in the UK context, is less an intervention and more a framework within which needs are identified and assessed, care is planned, interventions and services are delivered and ongoing needs are monitored. Any type or number of specific interventions or services (or none), both for the ill or disabled person and (less frequently) the carer, might thus be delivered as a result of receiving case/care management.

   2. The focus of case/care management is the ill or disabled person, albeit that the carer’s needs might also be considered during assessment and care planning.

The growth in the literature since the first review posed other challenges; for example, there were reviews that dealt with dyad interventions, interventions directed at the ill or disabled person and/or carers, and multiple-component interventions. There was also the issue of geographic coverage, whereby it was not clear if interventions or delivery contexts were fully transferable to UK health and social care systems. We also encountered the issue, evident in the original meta-review, of reviews in which the main focus was not on outcomes for the carer, but such outcomes were reported. We discussed these issues prior to data extraction and agreed a consistent way forward on whether to include or exclude. We included reviews of interventions aimed at patient–carer dyads only when carer outcomes were reported separately. When carer outcomes were reported but were not the main focus, a judgement was made as to the usefulness of this contribution to our meta-review. In relation to geographic applicability, we focused on reviews of interventions from developed countries with similar health-care systems, regardless of any differences in payment arrangements. We included multicomponent interventions on the basis that identifying the differential effectiveness of component parts may be limited to what was reported by the review authors.

Applying the adapted quality assessment criteria and scoring system by Egan et al. 10 used in the original meta-review revealed 61 high-quality systematic reviews. This was a larger proportion than expected. However, we noted that a review could achieve high-quality status using this system on the basis of adequate reporting of research question, search strategy, inclusion criteria and study designs/numbers but with insufficient attention to quality assessment, synthesis and transparency in the review process. Reviews with such shortcomings would be scrutinised closely for overall reliability, or may fail altogether the criteria for inclusion in DARE. DARE is a database of quality-assessed systematic reviews meeting specific criteria that was produced by the CRD, University of York; included are reviews that evaluate the effects of health and social care interventions, including delivery and organisation of services. Full details of the DARE process are available. 11 Therefore, we decided to refine the scoring system in the original meta-review and introduce a second tier of criteria (using the quality threshold for DARE) to further differentiate the ‘high’-quality reviews by splitting them into ‘high’ and ‘medium’ quality.

To be classed as ‘high’ quality, reviews had to reach a minimum score of 4 points, comprising (as mandatory) 1 point each for inclusion criteria, search strategy and synthesis and, additionally, 1 point for either quality assessment or number/design of included studies.

Most of the reviews identified at this stage were about ill or disabled people with specific conditions or impairment, for example dementia, stroke or cancer. Therefore, prior to data extraction of the included high-quality reviews, we grouped them according to impairment or condition to establish any discernible patterns and weightings in the evidence base. After this, we examined the distribution of reviews, by quality and by condition/impairment. This allowed us to adopt a ‘best evidence available’ approach to each of the condition/impairment areas identified.

Twenty-seven reviews were reclassified as high quality and progressed to detailed data extraction. The remaining 25 reviews (i.e. those that were high quality using the adapted Egan et al. 10 criteria but failed to meet the threshold for inclusion on DARE) were classed as medium quality, and we proceeded to basic data extraction. Bibliographic details were provided for nine reviews of low quality (using the adapted Egan et al. 10 criteria).

Quality of the primary studies

A quality assessment of primary studies was carried out in 2519–21,23–33,35–46 of the 27 included reviews. In most cases, it was possible to determine at least the overall quality of the included studies. However, Shoenmakers et al. 30 applied quality assessment only as an inclusion criterion and did not report further on the quality of the primary studies. In Ellis et al. ,39 only selective coding was carried out, making it impossible to gauge the overall quality. In Eggenberger et al. ,21 quality assessment criteria were presented in the paper, but detailed results were not. Similarly, in Macleod et al. ,45 risk of bias was reported to have been assessed, but the results of this were not presented. Two reviews22,34 did not perform any quality assessment of primary studies.

When it was possible to determine from results reported in the reviews, the methodological quality of primary studies was variable. In reviews targeting carers of people with dementia, only one26 specifically reported that all of the included studies had a low risk of bias. 26 A majority of studies in another review20 was reported as being of high or moderate quality, and a further review presented an average quality score of 75 out of 100. 32 In other reviews, the quality of primary studies appeared to be moderate,23,25,29 variable19,21,28 or very low. 27 One review described overall quality as satisfactory;31 another reported separately on the quality of primary studies for different outcomes (overall low to moderate). 24 In reviews targeting carers of people with cancer, one33 suggested that studies were of moderate quality,33 one35 reported moderate to strong evidence35 and one36 reported fair to good-quality evidence. 36 Limitations and variable study quality were also reported in reviews of interventions for carers of people with stroke. 37,38,40,41 Reviews addressing carers of people with various conditions at the end of life indicated studies of unclear quality,42 mixed-quality studies43 or studies at serious risk of bias, particularly in those focusing on carer outcomes. 44 Low- to moderate-quality primary studies were reported in one review46 focusing on carers of people with mental health problems. 46

Approach to synthesis

In most reviews, the analysis was grouped by intervention or outcome. The multicomponent nature of many interventions meant that it was difficult to identify causal relationships. Eleven reviews undertook narrative synthesis19,21–23,28,32,35,36,43–45 and six undertook quantitative synthesis. 20,26,27,29,30,34 Ten reviews24,25,31,33,37–42,46 contributed both narrative and quantitative syntheses. Two references37,38 relate to the same review.

Overlap of primary studies

From the outset, it was clear that there was some overlap of primary studies in the reviews we included. The effect of this overlap is difficult to judge without substantial additional analysis, but it could run the risk of exaggerating effects from the undue influence of individual studies, and present difficulties arising from contradictory assessments of the same study.

Carer outcomes

Mental health

Carers’ mental health was a frequently reported outcome in the 24 included reviews. 19–25,27–35,37–43,45,46

Mental health was variably defined (when reported). The terms depression, anxiety, psychological distress and self-efficacy were commonly used. Some reviews defined the outcome more broadly as psychological well-being or carer mental health. The outcome measures were generally well reported and diverse. Frequently used measures were the Centre for Epidemiologic Studies Depression Scale (CES-D), Brief Symptom Inventory, Beck Depression Inventory, Hospital Anxiety and Depression Scale (HADS) and the GHQ-12 and GHQ-28. The following results focus primarily on the detail for positive intervention effects. Providing further detail on reviews showing no significant effects for mental health outcomes was not considered to be informative, but a brief summary of results from these reviews is provided below. All results are reported in Table 4.

Burden and stress

Carer burden and stress was reported in 21 reviews (22 papers). 19–31,33,34,37,38,40,41,43,45,46

Burden and stress were not well-defined outcomes in most reviews, but various outcome measures were reported: the Zarit Burden Scale, the Caregiver Appraisal Inventory, the Caregiver Strain Index (CSI) and others. The measurement tools used in respect of burden were fairly consistent across the reviews. All results are reported in Table 5.

Coping

Coping outcomes for carers were reported in nine reviews. 19,27,31,32,34,35,42,43,45 Definitions of coping (when reported) varied. These included coping skills; strategies (including problem-solving and reduction of ineffective coping such as avoidance and denial); caregiving competence; perceived affective and confidant support; ability to achieve previously set objectives; confidence in caregiving skills; and control of worry, loneliness and reliance on support systems. Some outcome measures were reported, including the Caregiver Burden Index, the CSI and the subscale of the Duke UC Functional Support Questionnaire. All results are reported in Table 6.

Satisfaction

Satisfaction with the intervention was reported in eight reviews (nine papers). 21,25,37–41,43,46 The outcome was rarely defined further. Several outcome measures were reported; some of them related to carer satisfaction in general. All results are reported in Table 7.

Well-being and quality of life

Twenty reviews (21 papers)19,22,24,25,27,28,31,33–46 reported on aspects related to carer well-being and quality of life. In addition to the general terms of well-being and quality of life, various others were used to describe this outcome. These included social isolation, social support needs, caregiving experience, sleep quality, marital–family relationships, social functioning and ability to perform activities of daily living. Many different outcome measures were reported. All results are reported in Table 8.

Ability and knowledge

Carers’ ability and knowledge was reported in nine reviews (10 papers). 19,21,28,34,37,38,40,43,45,46 When reported, more detailed author definitions of this outcome included the carers’ sense of competence or mastery, decision-making confidence, knowledge of the disease and caring, information needs and learning new skills. Various outcome measures were reported, including the Caregiver Demand Scale, Lawton Positivity Questionnaire and Preparedness for Caregiving Scale. All results are reported in Table 9.

Carers of people with dementia (14 reviews)

In a review of 12 studies, Boots et al. 19 concluded that internet-based support interventions may improve carer well-being and that multicomponent activities allowing carer interaction (e.g. carer discussion forums) are more effective than those offering primarily information. This tentative conclusion is based on results of mixed-quality primary studies that generally reported conflicting or non-significant effects. However, some positive effects were reported. The best evidence of effectiveness was in relation to small positive intervention effects on depression (one higher-quality study), self-efficacy (one higher-quality study) and on carer stress and strain (one moderate-quality study). Positive intervention effects were also found for carer confidence in decision-making (one good-quality study). Although there were some limitations in reporting, this was a largely well-conducted review.

Chien et al. 20 concluded that carers of people with dementia benefit from support groups and that the use of theoretical models to aid intervention design had a significant impact on the effect size for psychological well-being and depression. The overall quality of 30 primary studies included in this review was reported to be high or moderate. For depression, the effect size was small to moderate but with high statistical variation in the analysis of 17 studies. A small reduction in carer burden and stress was indicated in a further analysis of 24 studies with low statistical heterogeneity; the effect appeared to persist over time. Although the quality of primary studies was generally good, the lack of control group data, the high statistical heterogeneity for mental health outcomes and the reporting limitations in this review meant that it was difficult to be totally confident about the review authors’ conclusions.

In Eggenberger et al. ,21 the authors concluded that communication skills training had a significant impact on family carers’ communication skills, competencies and knowledge. This review of 12 studies had substantial reporting limitations and generally contained studies of variable quality. This makes it difficult to assess the reliability of the authors’ conclusions. The best evidence of effectiveness arose from one good-quality study that found reduced depression in carers after a home-care education intervention with professional support, and another good-quality study of communication skills training resulting in total carer satisfaction.

Godwin et al. 22 concluded that each of the included technology-driven multicomponent support interventions (including Caregiver’s Friend: Dealing with Dementia, ComputerLink and REACH) had some positive findings, but there was insufficient evidence to support or refute the effects. Evidence from the primary studies supports this conclusion on positive effects for all outcomes except social isolation. Although overall the review was reasonably well conducted, it comprised only eight studies with no quality assessment; control group and follow-up data were lacking.

In a review of seven studies looking at carer education focused on skills training, Jensen et al. 24 concluded that educational programmes have a moderate effect in reducing carer burden and a small effect in reducing depression; the effects were unclear for quality of life and transition to long-term care. The analysis of depression included two studies (one of high quality). The result for carer burden was based on five moderate-quality studies with some statistical heterogeneity, and that favoured interventions of shorter duration. This review appeared largely well conducted and provided additional analysis of outcomes separated by low- and high-income countries.

In a well-conducted and well-reported review by Hurley et al. 23 (containing eight studies), the authors reported tentative evidence for the effectiveness of meditation-based interventions for improving scores of depression and carer burden. This conclusion was supported by primary study evidence at the end of the intervention in five moderate-quality studies for depression, and in three low- to moderate-quality studies for carer burden. The results for both outcomes were mixed at follow-up.

A well-conducted Cochrane review by Lins et al. 25 (containing 11 studies) concluded that telephone counselling without any additional intervention can reduce depressive symptoms and also meets the important needs of carers. The conclusion on depressive symptoms was supported in the analysis of three moderate-quality studies with no evidence of statistical heterogeneity. A positive effect on depression was also found in a moderate-quality study focusing on an enhanced version of the intervention comprising telephone counselling with additional video sessions and workbook. Two moderate-quality, qualitative studies substantiated the review authors’ conclusion on carer satisfaction with the intervention. Positive control group effects for self-efficacy and satisfaction were also reported, but the quality of studies in these analyses was mixed. This was a well-conducted review, and theoretical underpinnings of the included studies were reported.

The authors in Maayan et al. 27 concluded that current evidence does not demonstrate any benefits or adverse effects from respite care in carers of people with dementia, but the results should be interpreted cautiously. The authors’ conclusion was supported by this largely well-conducted Cochrane review containing four primary studies of very low quality.

Marim et al. 26 concluded that interdisciplinary education and support programmes have a positive impact on carer burden when compared with standard care. The conclusion was supported by this well-conducted and well-reported review of seven studies containing high-quality primary studies.

In their review of 14 studies, McKechnie et al. 28 concluded that computer-mediated psychosocial interventions can benefit carers of people with dementia. The best evidence of effectiveness related to improvements in scores for depression in the analyses of four high-quality studies, for anxiety from two high-quality studies and for reductions in stress and burden from five (out of nine) medium- to high-quality studies, with remaining studies in the last analysis showing inconsistent results. Not all of the included studies had control groups, and there were potential limitations regarding the transparency of the review process.

The review by Orgeta and Miranda-Castillo29 suggested that home-based physical activity interventions were effective in reducing subjective carer burden, compared with usual care. The analysis showed a moderate intervention effect based on two studies with unclear or low risk of bias and no evidence of statistical heterogeneity. This review was well reported in terms of methods, interventions and results. However, the overall review findings are based on four clinically heterogeneous studies with largely unknown risk of bias and limited follow-up data.

The conclusion from Schoenmakers et al. 30 suggested weak evidence for interventions to support family carers of people with dementia. Although this conclusion was generally supported by the primary study evidence, it was not possible to confirm the quality of the 19 included studies, as quality assessment was used only to select studies for the review. A small to moderate statistically significant adverse effect was reported in respect of increased carer burden resulting from respite care. The review authors speculated that this effect may be a result of carers’ uneasiness with respite nursing quality and the sudden release of free time for themselves.

In the generally well-conducted review by Smith and Greenwood32 (containing three studies), the authors concluded that there is limited quantitative evidence to support volunteer mentoring for carers of people with dementia. This contrasted with qualitative evidence that carers value the opportunity to talk with others about their experiences. The best evidence of effectiveness arose from one high-quality study of a telephone befriending intervention and consequent statistically significant improvement in scores for carer anxiety and depression after 15 months. Cost-effectiveness was reported for this study but no positive impact was found. Other benefits were apparent, but it was unclear whether they applied only to carers of people with dementia or also to carers of those with stroke. Such benefits were demonstrated in one moderate-quality qualitative study in terms of coping skills and caregiving competence, reduced loneliness and reliance on social support.

The authors of the well-conducted Cochrane review by Vernooij-Dassen et al. 31 (containing 11 studies) suggested that cognitive reframing for family carers shows promise as part of an individualised, multicomponent intervention. The inclusion of cognitive reframing appeared to reduce psychological morbidity and subjective stress, but without any effects on appraisals of coping or burden. In support of this conclusion, moderate to large effects were reported for reduced depression in the analysis of six studies; small-sized effects for anxiety from the analysis of four studies; and similarly small effects for stress and distress from four studies. All primary studies had some methodological limitations, but the overall quality was considered by the review authors to be satisfactory.

Carers of people with cancer (four reviews)

The authors in Lang and Lim33 concluded that art therapy is effective in reducing anxiety, stress and negative emotions in family carers of patients with cancer. This conclusion reflects a statistically significant pooled effect in two studies for anxiety; effects for reduced stress from baseline in each of two studies; and an improvement in carer emotional balance in one study. This was a well-reported review of moderate-quality primary studies. However, the findings may be limited by the reliance on two small-sized studies, each with the same lead author.

In their well-reported review of 29 studies, Northouse et al. 34 concluded that multicomponent interventions, including psychoeducation, skills training and counselling, can have a positive effect on many important carer outcomes, including small to medium effects on burden, ability to cope, self-efficacy and quality of life. These positive effects were all substantiated in the primary studies. Additionally, small statistically significant positive effects were reported for self-care behaviours and sleep quality in six studies when measured beyond 3 months from intervention delivery; small effects were noted for distress and anxiety, and moderate effects were sustained beyond 6 months following psychoeducation in 16 studies. In an analysis of three studies, a large statistically significant effect was reported for carers’ appraisal of information needs during the first 3 months post intervention. The review reported on theoretical frameworks underpinning the intervention. Despite the large number of positive intervention effects reported, the reliability of the review findings is limited by the absence of primary study quality assessment. Additionally, although most interventions were delivered jointly to patients and carers, the authors stated that many were designed primarily to address patient care.

Regan et al. 35 included six moderate- to strong-quality primary studies out of 23 studies overall. The authors concluded that couples-based psychosocial interventions showed promise, particularly in respect of improving couple communication and relationship functioning, and reducing psychological distress. These conclusions were supported by the evidence presented. In addition, there were reductions in physical distress in one study of disease management, psychoeducation and telephone counselling; and in another study evaluating the FOCUS intervention (family coping skills and uncertainty reduction). Improvements were also noted following the FOCUS intervention for quality of life [physical and emotional functioning (two studies)].

In their review of six studies, Waldron et al. 36 concluded that psychosocial/psychoeducation interventions focusing on problem-solving and communication skills may improve quality of life in carers of people with cancer. The evidence was provided by a small effect size in the analysis of two good-quality studies. The review was well conducted and well reported.

Carers of people with stroke (four reviews)

The review authors in Cheng et al. 37,38 suggested that there was limited evidence of effect for psychosocial interventions on family functioning of carers for people with stroke. This conclusion is based on a small effect size favouring counselling over no treatment from the analysis of two moderate-quality studies. In addition, satisfaction with psychoeducation, counselling or support was higher than for usual care in two moderate-quality studies. This was a well-conducted review with small numbers of studies included in each analysis across multiple outcomes. Theoretical frameworks underpinning the interventions were reported.

In the Cochrane review of individual patient data (involving 4759 patients and carers) by Ellis et al. ,39 the authors concluded that there was no evidence of effectiveness for the introduction of stroke liaison workers to help carers and patients, although some studies found improved user satisfaction with service provision. The conclusions were supported by the evidence presented in this largely well-conducted review. However, the reliability of the findings is limited by the inability to judge the overall quality of the primary studies.

Another Cochrane review by Forster et al. 40 (containing 21 studies) concluded that education-based interventions can improve carer knowledge of stroke. The conclusion is supported by evidence for active education programmes, but only in a pooled analysis of four studies containing substantial heterogeneity and at some risks of bias. The interventions in this review showed evidence of cost-effectiveness, possibly linked to benefits in length of hospital stay. The review was well conducted, but the overall findings are limited by the fact that only a small subset of the included studies related to carer outcomes.

Finally, a well-conducted Cochrane review by Legg et al. 41 (containing eight studies) concluded that interventions based on ‘teaching procedural knowledge’ (formal multidisciplinary training of caregivers in the prevention and management of common problems related to stroke), delivered to carers prior to the hospital discharge of stroke patients, appear to be the most promising. The authors’ tentative conclusion is justified on the basis of one small higher-quality study of ‘teaching procedural knowledge’. This intervention showed a small statistically significant reduction in carer depression, together with a statistically significant more substantial reduction in stress and improved quality of life.

Carers of people with various conditions at the end of life (three reviews)

The well-reported Cochrane review by Candy et al. 42 (containing 11 studies) proposed that multicomponent support interventions can help to reduce carers’ psychological distress. However, the evidence is derived from a pooled analysis of eight low-quality studies showing only a small effect size at the end of the intervention.

In the Gomes et al. 43 review of 23 studies, the authors stated that there was clear and reliable evidence that carer grief was not affected detrimentally after the receipt of home palliative care. This arises from a positive effect on carer distress favouring a reinforced version of the intervention, but it was based on one low-quality study and so the conclusion should be considered as tentative. A cost-effectiveness analysis involving six high-quality studies was inconclusive. This was a well-conducted review.

Nevis44 concluded that educational and problem-solving interventions were effective in improving carer quality of life but with no decrease in resource use. This conclusion was based on two (out of six included) studies containing serious risk of bias. No costs were reported.

Carers of people with mental health problems (two reviews)

In a large review of 68 studies focusing on multicomponent support interventions involving mental health nurses, Macleod et al. 45 concluded that, although findings were mixed, there was some evidence of effectiveness for a range of approaches. The primary evidence confirmed positive effects on carer mental health outcomes, burden and stress, coping, well-being and knowledge. The quality of primary studies (although claimed by the authors to be assessed) was not reported, so it is not possible to judge the reliability of these findings. Aside from this and some reporting discrepancies, the review was largely well conducted.

A well-conducted review by Yesufu-Udechuku et al. 46 (containing 21 studies) concluded that carer-focused interventions, including psychoeducation, support groups, problem-solving and self-management, appear to improve quality of life and reduce psychological distress, but no single specific intervention could be recommended. The best evidence of effectiveness for psychological distress is derived from statistically significant effects in one small high-quality study of psychoeducation, and from one moderate-quality study of problem-solving bibliotherapy measured up to 6 months post intervention. Quality of life (caregiving experiences, not further defined) was also improved, but the evidence was based largely on very low-quality studies with substantial heterogeneity.

Practice

This meta-review, like the earlier one, was established to explore the evidence on interventions specifically designed and intended to support carers, not to explore the ways in which generic services might also help carers. It is important to remember this, because we know from other types of research that services intended for the person being supported are often crucial to the carers’ own health and well-being. For example, if people with dementia attend day activities and are offered things to do that engage their interest and give them an opportunity to meet other people, carers may benefit both from the short break it gives them and from any positive effect the activity may have on the mood of the person with dementia. Similarly, if people who are nearing the end of their life receive good-quality palliative care, both they and their close family members may benefit.

In the original meta-review, the strongest evidence of effectiveness of any sort was in relation to education, training and information for carers. When active and targeted, such interventions increased carers’ knowledge and abilities as carers, with some indication that they might also thereby improve carers’ mental health or coping. This updated meta-review suggests that interventions with this sort of content, for carers of people with dementia, may also have a positive impact on subjective burden.

However, the updated meta-review suggests other possibilities for effective support for carers. These are as follows.

• For carers of people with dementia:

  • Opportunities to share with and learn from others (whether carers or professionals) may have a positive impact on depression and anxiety, and on subjective burden.

  • Opportunities to reframe the way they think about dementia may have a positive impact on mental health and on subjective burden and stress.

  • Learning meditation techniques may have a positive impact on depression.

  • Psychosocial interventions delivered via a computer may have a positive impact on depression and anxiety and on subjective burden and stress.

• For carers of people with cancer:

  • Psychosocial interventions may have a positive impact on carers’ physical and psychological ‘distress’ and quality of life.

  • Art therapy may have a positive impact on carers’ mental health.

  • Counselling may have a positive impact on relationship functioning.

• For carers of people with stroke:

  • Counselling may have positive impact on family functioning.

None of the above suggests that a ‘one intervention fits all’ approach would meet the needs of all carers; neither do we know which elements of multicomponent interventions have the most effect.

Research

The implications for research are much as they were previously:

1. There remains a dearth of good-quality primary research about the effectiveness of most support interventions for carers. In large part, we suspect that this is driven by the lack of underpinning theories of change or intervention logic that would allow the right questions about the right outcomes to be asked. This applies as much to recent initiatives in Britain, such as carers’ champions and health checks for carers (for which we were unable to identify any review evidence), as it does to more established interventions.

2. The issue of respite care and its effectiveness, or not, in supporting carers of people with dementia remains a paradox, given the apparent conflict between the empirical evidence and the views of carers. On the one hand, the earlier meta-review and this one found reviews suggesting that respite care might have negative effects for carers of people with dementia. On the other hand, our carer advisers and the qualitative literature suggest that respite care is essential to carers feeling that they can continue to care. A major issue with the reviews of respite care, however, is the assessment of the quality of the respite being evaluated and, in some cases, careful distinction between the different types of respite on offer. Furthermore, there have been recent developments that may provide a different sort of respite from traditional models. Examples of such new models include memory cafés and the use of volunteers to provide people with dementia with meaningful daytime occupation; however, these have not yet been fully evaluated. Primary research that explores what type of respite is better or worse, for whom, and in relation to outcomes that carers themselves think are valuable is urgently needed. This should be research that articulates clear intervention logic driven by carers’ views about outcomes, and that carefully triangulates both quantitative and qualitative findings.

3. Reviews that separate out different parts of multicomponent interventions and then model their unique impact are beginning to emerge, but much more of this type of sophisticated systematic reviewing is needed. A review that attempted to separate out and assess the effects of different elements of formal ‘psychosocial’ support, and to set this alongside evidence of the effect of informal support delivered via different modes, would be a start.

4. Such reviews would also need to model differential impact for different types of carers. By definition, carers are a very diverse group; as our carer advisers pointed out, expecting a single type of intervention to meet the needs of all carers and at all stages of caring is unrealistic. There appears to be a particular gap in the review evidence relating to young carers and carers from minority groups.

5. Heavily involved carers, by definition, have very little, if any, time to spare from their daily routines, yet these are the people most likely to need support. Finding ways of engaging meaningfully with them in research processes, without imposing additional burdens, is a challenge for all empirical and review work in this field. The use of ‘virtual’ advisory groups for evaluation research, whereby user organisations act as a conduit through which carers can engage in research projects via face-to-face meetings, telephone or e-mail, as and when they have time, is proving very useful in other NIHR-funded projects at York. 82 However, this model has inevitable resource implications, as it takes time to set up and needs to be moderated independently of the research team.

Search strategy

The search strategy below incorporates a section to restrict the search to reviews only. This part of the strategy was based on the CRD search strategy for retrieving reviews from ASSIA.

((SU.EXACT(“Informal care”) OR SU.EXACT(“Carers”) OR SU.EXACT(“Respite care”) OR TI,AB(caregiv* OR care-giv* OR carer* OR ”informal care” OR befriending OR caretak* OR “care taker” OR “care takers” OR “care taking” OR “children caring” OR “families caring” OR respite) OR TI,AB(families NEAR/2 support)) AND ((TI,AB(metaanaly* OR meta-analy*) OR SU.EXACT(“Literature reviews”) OR SU.EXACT(“Systematic reviews”) OR TI,AB,IF(“meta study” OR meta-synthes* OR meta-evaluat*) OR TI,AB,IF(synthes* NEAR/3 literature*) OR TI,AB,IF(synthes* NEAR/3 research*) OR TI,AB,IF(synthes* NEAR/3 studies) OR TI,AB,IF(synthes* NEAR/3 data) OR TI,AB,IF(synthes* NEAR/3 trials) OR TI,AB,IF(synthes* NEAR/3 findings) OR TI,AB,IF(synthes* NEAR/3 evidence) OR TI,AB,IF(quantitative-synthes*) OR TI,AB,IF(pooled-analys*) OR TI,AB,IF((data NEAR/3 pool*) AND studies)) OR (TI,AB,IF(pooling NEAR/1 studies) OR TI,AB,IF(medline OR medlars OR embase OR cinahl OR cochrane OR scisearch OR psychinfo OR psycinfo OR psychlit OR psyclit) OR TI,AB,IF((hand OR manual* OR database* OR computer* OR electronic*) NEAR/3 search*) OR TI,AB,IF((electronic* OR bibliographic*) NEAR/3 database*) OR TI,AB,IF(overview*) OR TI,AB,IF(“evaluation review*”) OR TI,AB,IF(“what works”) OR TI,AB,IF(“evaluation synthes*”) OR TI,AB,IF(review*)))) AND la.exact(“English”)

Additional limits: date – from 1 January 2000 to 21 January 2016.

Search strategy

#1 MeSH descriptor: [Caregivers] this term only (1313)

#2 MeSH descriptor: [Respite Care] this term only (33)

#3 caregiv*:ti,ab,kw (4322)

#4 care next giv*:ti,ab,kw (351)

#5 carer*:ti,ab,kw (1060)

#6 "informal care":ti,ab,kw (72)

#7 befriending:ti,ab,kw (41)

#8 caretak*:ti,ab,kw (147)

#9 care next taker*:ti,ab,kw (10)

#10 care next taking:ti,ab,kw (10)

#11 children next caring:ti,ab,kw (2)

#12 families next caring:ti,ab,kw (10)

#13 families near/2 support:ti,ab,kw (48)

#14 respite:ti,ab,kw (74)

#15 (parent or parents or mother or mothers or father or fathers) near/2 care:ti,ab,kw (460)

#16 (parent or parents or mother or mothers or father or fathers) near/2 caring:ti,ab,kw (18)

#17 (parent or parents or mother or mothers or father or fathers) near/2 support:ti,ab,kw (252)

#18 (parent or parents or mother or mothers or father or fathers) near/2 supporting:ti,ab,kw (35)

#19 (sons or daughters or friends) near/2 care:ti,ab,kw (5)

#20 (sons or daughters or friends) near/2 caring:ti,ab,kw (0)

#21 (sons or daughters or friends) near/2 support:ti,ab,kw (43)

#22 (sons or daughters or friends) near/2 supporting:ti,ab,kw (0)

#23 (husband* or wives or wife or spouse* or grandparent* or grandchild* or neighbour* or neighbor* or relatives) near/2 care:ti,ab,kw (42)

#24 (husband* or wives or wife or spouse* or grandparent* or grandchild* or neighbour* or neighbor* or relatives) near/2 caring:ti,ab,kw (11)

#25 (husband* or wives or wife or spouse* or grandparent* or grandchild* or neighbour* or neighbor* or relatives) near/2 support:ti,ab,kw (125)

#26 (husband* or wives or wife or spouse* or grandparent* or grandchild* or neighbour* or neighbor* or relatives) near/2 supporting:ti,ab,kw (3)

#27 #1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11 or #12 or #13 or #14 or #15 or #16 or #17 or #18 or #19 or #20 or #21 or #23 or #24 or #25 or #26 (6318)

Note that the result at line #27 is the total for all of the databases within The Cochrane Library.

Search strategy

The search strategy below incorporates a section to restrict the search to reviews only. This part of the strategy was based on the CRD search strategy for retrieving reviews from CINAHL. 83

Search strategy

The search strategy for EMBASE below incorporated the Hedges best optimisation of sensitivity and specificity filter for retrieval of systematic reviews in EMBASE. 84

1. Caregivers/ (28,457)

2. Caregiver support/ (1813)

3. Respite Care/ (874)

4. caregiv$.ti,ab. (54,102)

5. care giv$.ti,ab. (6712)

6. carer$.ti,ab. (13,826)

7. informal care.ti,ab. (1329)

8. befriending.ti,ab. (136)

9. caretak$.ti,ab. (4320)

10. care taker$.ti,ab. (245)

11. care taking.ti,ab. (322)

12. children caring.ti,ab. (42)

13. families caring.ti,ab. (257)

14. (families adj2 support).ti,ab. (1603)

15. respite.ti,ab. (1705)

16. or/1-15 (87,059)

17. meta-analys$.mp. (160,679)

18. search$.tw. (356,113)

19. review.pt. (2,121,136)

20. 17 or 18 or 19 (2,456,581)

21. 16 and 20 (12,459)

22. limit 21 to yr = “2000 – Current” (10,497)

23. limit 22 to english language (9443)

24. exp Animal/ (21,018,527)

25. exp animal-experiment/ (1,902,970)

26. nonhuman/ (4,667,343)

27. (rat or rats or mouse or mice or hamster or hamsters or animal or animals or dog or dogs or cat or cats or bovine or sheep).ti,ab,sh. (5,214,665)

28. 24 or 25 or 26 or 27 (22,616,514)

29. exp human/ (16,646,006)

30. exp human-experiment/ (346,372)

31. 29 or 30 (16,647,452)

32. 28 and 31 (16,646,482)

33. 28 not 32 (5,970,032)

34. 23 not 33 (9397)

35. (2009$ or 2010$ or 2011$ or 2012$ or 2013$ or 2014$ or 2015$ or 2016$).em. (9,456,931)

36. 34 and 35 (4869)

Search strategy

The search strategy for HMIC incorporated a strategy for finding reviews that was translated from the CRD search strategy for retrieving reviews from ASSIA.

1. Carers/ (4005)

2. Informal Care/ (393)

3. exp Respite Care/ (448)

4. caregiv$.ti,ab. (1161)

5. care giv$.ti,ab. (743)

6. carer$.ti,ab. (6824)

7. informal care.ti,ab. (424)

8. befriending.ti,ab. (82)

9. caretak$.ti,ab. (66)

10. care taker$.ti,ab. (1)

11. care taking.ti,ab. (28)

12. children caring.ti,ab. (14)

13. families caring.ti,ab. (60)

14. (families adj2 support).ti,ab. (220)

15. respite.ti,ab. (611)

16. or/1-15 (10,299)

17. exp LITERATURE REVIEWS/ (5537)

18. meta analysis/ (726)

19. (metaanaly$ or meta-analy$).ti,ab. (1605)

20. (meta study or meta synthes$ or meta evaluat$).ti,ab. (40)

21. (synthes$ adj3 (literature$ or research$ or studies or data or trials or findings or evidence)).ti. (104)

22. quantitative synthes$.ti,ab. (22)

23. pooled analys$.ti,ab. (98)

24. ((data adj3 pool$) and studies).ti,ab. (93)

25. (pooling adj2 studies).ti,ab. (3)

26. (medline or medlars or embase or cinahl or cochrane or scisearch or psychinfo or psycinfo or psychlit or psyclit).ti,ab. (2342)

27. ((hand or manual$ or database$ or computer$ or electronic$) adj3 search$).ti,ab. (1424)

28. ((electronic$ or bibliographic$) adj3 database$).ti,ab. (972)

29. review$.ti,ab. (35,957)

30. overview$.ti,ab. (4451)

31. evaluation synthes$.ti,ab. (0)

32. evaluation review$.ti,ab. (6)

33. what works.ti,ab. (274)

34. or/17-33 (41,527)

35. 16 and 34 (1523)

36. limit 35 to yr=”2000 –Current” (909)

Search strategy

The search strategy for MEDLINE below incorporated the Hedges optimised sensitivity and specificity balanced search filter for retrieval of systematic reviews in MEDLINE. 85

1. Caregivers/ (24,035)

2. Respite Care/ (914)

3. caregiv$.ti,ab. (35,082)

4. care giv$.ti,ab. (4089)

5. carer$.ti,ab. (8206)

6. informal care.ti,ab. (930)

7. befriending.ti,ab. (93)

8. caretak$.ti,ab. (3226)

9. care taker$.ti,ab. (132)

10. care taking.ti,ab. (215)

11. children caring.ti,ab. (31)

12. families caring.ti,ab. (211)

13. (families adj2 support).ti,ab. (1061)

14. respite.ti,ab. (1196)

15. or/1-14 (57,920)

16. meta-analysis.mp,pt. (86,024)

17. review.pt. (1,996,933)

18. search$.tw. (244,702)

19. 16 or 17 or 18 (2,177,630)

20. 15 and 19 (8440)

21. exp animals/ not humans/ (4,173,052)

22. 20 not 21 (8396)

23. (2009$ or 2010$ or 2011$ or 2012$ or 2013$ or 2014$ or 2015$ or 2016$).ed. (5,415,956)

24. 22 and 23 (3442)

25. limit 24 to english language (3109)

Search strategy

The search strategy below incorporated a strategy for finding reviews that was translated from the CRD search strategy for retrieving reviews from MEDLINE. 83

1. caregiv$.ti,ab. (4875)

2. care giv$.ti,ab. (439)

3. carer$.ti,ab. (1055)

4. informal care.ti,ab. (137)

5. befriending.ti,ab. (22)

6. caretak$.ti,ab. (283)

7. care taker$.ti,ab. (26)

8. care taking.ti,ab. (32)

9. children caring.ti,ab. (5)

10. families caring.ti,ab. (15)

11. (families adj2 support).ti,ab. (141)

12. respite.ti,ab. (95)

13. or/1-12 (6675)

14. systematic$ review$.ti,ab. (14,947)

15. meta-analytic$.ti,ab. (564)

16. meta-analysis.ti,ab. (12,106)

17. metanalysis.ti,ab. (13)

18. metaanalysis.ti,ab. (101)

19. meta analysis.ti,ab. (12,106)

20. meta-synthesis.ti,ab. (90)

21. metasynthesis.ti,ab. (31)

22. meta synthesis.ti,ab. (90)

23. meta-regression.ti,ab. (562)

24. metaregression.ti,ab. (44)

25. meta regression.ti,ab. (562)

26. (synthes$ adj3 literature).ti,ab. (287)

27. (synthes$ adj3 evidence).ti,ab. (736)

28. integrative review.ti,ab. (244)

29. data synthesis.ti,ab. (628)

30. (research synthesis or narrative synthesis).ti,ab. (252)

31. (systematic study or systematic studies).ti,ab. (1782)

32. (systematic comparison$ or systematic overview$).ti,ab. (384)

33. evidence based review.ti,ab. (243)

34. comprehensive review.ti,ab. (1391)

35. critical review.ti,ab. (1391)

36. quantitative review.ti,ab. (48)

37. structured review.ti,ab. (64)

38. realist review.ti,ab. (43)

39. realist synthesis.ti,ab. (34)

40. or/14-39 (28,503)

41. medline.ab. (9796)

42. pubmed.ab. (11,032)

43. cochrane.ab. (7101)

44. embase.ab. (7697)

45. cinahl.ab. (2226)

46. psyc?lit.ab. (27)

47. psyc?info.ab. (2869)

48. (literature adj3 search$).ab. (5471)

49. (database$ adj3 search$).ab. (5031)

50. (bibliographic adj3 search$).ab. (196)

51. (electronic adj3 search$).ab. (2097)

52. (electronic adj3 database$).ab. (2731)

53. (computeri?ed adj3 search$).ab. (232)

54. (internet adj3 search$).ab. (310)

55. included studies.ab. (1997)

56. (inclusion adj3 studies).ab. (1444)

57. inclusion criteria.ab. (7790)

58. selection criteria.ab. (1962)

59. predefined criteria.ab. (146)

60. predetermined criteria.ab. (55)

61. (assess$ adj3 (quality or validity)).ab. (6110)

62. (select$ adj3 (study or studies)).ab. (5529)

63. (data adj3 extract$).ab. (5037)

64. extracted data.ab. (965)

65. (data adj2 abstracted).ab. (311)

66. (data adj3 abstraction).ab. (161)

67. published intervention$.ab. (13)

68. ((study or studies) adj2 evaluat$).ab. (13,837)

69. (intervention$ adj2 evaluat$).ab. (901)

70. confidence interval$.ab. (25,616)

71. heterogeneity.ab. (11,555)

72. pooled.ab. (6160)

73. pooling.ab. (835)

74. odds ratio$.ab. (17,406)

75. (Jadad or coding).ab. (12,700)

76. or/41-75 (104,006)

77. review.ti. (40,392)

78. 77 and 76 (12,082)

79. (review$ adj4 (papers or trials or studies or evidence or intervention$ or evaluation$)).ti,ab. (16,644)

80. 40 or 78 or 79 (41,167)

81. 13 and 80 (413)

82. limit 81 to yr = “2000 –Current” (410)

83. limit 82 to english language (401)

Search strategy

The search strategy below incorporated an adapted version of the CRD search strategy for retrieving reviews from PsycINFO. 83

1. Caregivers/ (21,578)

2. Respite Care/ (405)

3. caregiv$.ti,ab. (38,420)

4. care giv$.ti,ab. (2332)

5. carer$.ti,ab. (7251)

6. informal care.ti,ab. (734)

7. befriending.ti,ab. (192)

8. caretak$.ti,ab. (4009)

9. care taker$.ti,ab. (46)

10. care taking.ti,ab. (161)

11. children caring.ti,ab. (52)

12. families caring.ti,ab. (230)

13. (families adj2 support).ti,ab. (1351)

14. respite.ti,ab. (1288)

15. or/1-14 (54,825)

16. metaanaly*.ti,sh. (68)

17. meta-analy*.ti,sh. (13,305)

18. cochrane*.ti. (155)

19. (review* or overview*).ti,ab. (481,630)

20. meta analysis/ (3771)

21. meta analysis.md. (14,073)

22. (review adj2 literature).ti. (3525)

23. “literature review”.md. (116,490)

24. “systematic review”.md. (13,184)

25. (synthes* adj3 (literature* or research or studies or data)).ti. (653)

26. pooled analys*.ti,ab. (532)

27. ((data adj2 pool*) and studies).ti,ab. (747)

28. ((hand or manual* or database* or computer* or electronic*) adj2 search*).ti,ab. (6637)

29. ((electronic* or bibliographic*) adj2 (database* or data base*)).ti,ab. (3073)

30. or/16-29 (514,573)

31. (“review software other” or “review media” or editorial or letter or “review book”).dt. (169,661)

32. (electronic collection or dissertation abstract or encyclopedia).pt. (450,365)

33. (rat or rats or mouse or mice or hamster or hamsters or animal or animals or dog or dogs or cat or cats or bovine or sheep).ti,ab,sh. (282,974)

34. 31 or 32 or 33 (844,164)

35. 30 not 34 (350,835)

36. 15 and 35 (5976)

37. limit 36 to (english language and yr=”2000 –Current”) (4431)

38. (2009$ or 2010$ or 2011$ or 2012$ or 2013$ or 2014$ or 2015$ or 2016$).up. (1,393,644)

39. 37 and 38 (2783)

Search 1

SubjectTerms: “ ‘carers’ ” including this term only

OR SubjectTerms: “ ‘young carers’ ” including this term only

OR SubjectTerms: “ ‘informal care’ ” including this term only

OR SubjectTerms: “ ‘befriending schemes’ ” including this term only

OR AllFields: ‘caregiver’

OR AllFields: ‘care-giver’

OR AllFields: ‘carer’

OR AllFields: “ ‘informal care’ ”

OR AllFields: ‘befriending’

OR AllFields: ‘caretaker’

OR AllFields: ‘care-taker’

OR AllFields: “ ‘care taking’ ”

OR AllFields: “ ‘children caring’ ”

OR AllFields: “ ‘families caring’ ”

OR AllFields: ‘respite’

AND

ContentTypes: ‘systematic review’

140 results.

Search 2

SubjectTerms: “ ‘carers’ ” including this term only

OR SubjectTerms: “ ‘young carers’ ” including this term only

OR SubjectTerms: “ ‘informal care’ ” including this term only

OR SubjectTerms: “ ‘befriending schemes’ ” including this term only

OR AllFields: ‘caregiver’

OR AllFields: ‘care-giver’

OR AllFields: ‘carer’

OR AllFields: “ ‘informal care’ ”

OR AllFields: ‘befriending’

OR AllFields: ‘caretaker’

OR AllFields: ‘care-taker’

OR AllFields: “ ‘care taking’ ”

OR AllFields: “ ‘children caring’ ”

OR AllFields: “ ‘families caring’ ”

OR AllFields: ‘respite’

AND

ContentTypes: ‘research review’

418 results.

Search 3

SubjectTerms: “ ‘carers’ ” including this term only

OR SubjectTerms: “ ‘young carers’ ” including this term only

OR SubjectTerms: “ ‘informal care’ ” including this term only

OR SubjectTerms: “ ‘befriending schemes’ ” including this term only

OR AllFields: ‘caregiver’

OR AllFields: ‘care-giver’

OR AllFields: ‘carer’

OR AllFields: “ ‘informal care’ ”

OR AllFields: ‘befriending’

OR AllFields: ‘caretaker’

OR AllFields: ‘care-taker’

OR AllFields: “ ‘care taking’ ”

OR AllFields: “ ‘children caring’ ”

OR AllFields: “ ‘families caring’ ”

OR AllFields: ‘respite’

AND

SubjectTerms: “systematic reviews” including this term only

OR SubjectTerms: “ ‘literature reviews’ ” including this term only

270 results.

Search 4

SubjectTerms: “ ‘carers’ ” including this term only

OR SubjectTerms: “ ‘young carers’ ” including this term only

OR SubjectTerms: “ ‘informal care’ ” including this term only

OR SubjectTerms: “ ‘befriending schemes’ ” including this term only

OR AllFields: ‘caregiver’

OR AllFields: ‘care-giver’

OR AllFields: ‘carer’

OR AllFields: “ ‘informal care’ ”

OR AllFields: ‘befriending’

OR AllFields: ‘caretaker’

OR AllFields: ‘care-taker’

OR AllFields: “ ‘care taking’ ”

OR AllFields: “ ‘children caring’ ”

OR AllFields: “ ‘families caring’ ”

OR AllFields: ‘respite’

AND

PublicationTitle: ‘review’

466 results.

Search 5

SubjectTerms: “ ‘carers’ ” including this term only

OR SubjectTerms: “ ‘young carers’ ” including this term only

OR SubjectTerms: “ ‘informal care’ ” including this term only

OR SubjectTerms: “ ‘befriending schemes’ ” including this term only

OR AllFields: ‘caregiver’

OR AllFields: ‘care-giver’

OR AllFields: ‘carer’

OR AllFields: “ ‘informal care’ ”

OR AllFields: ‘befriending’

OR AllFields: ‘caretaker’

OR AllFields: ‘care-taker’

OR AllFields: “ ‘care taking’ ”

OR AllFields: “ ‘children caring’ ”

OR AllFields: “ ‘families caring’ ”

OR AllFields: ‘respite’

AND

PublicationTitle: ‘overview’

47 results.

Search 6

SubjectTerms: “ ‘carers’ ” including this term only

OR SubjectTerms: “ ‘young carers’ ” including this term only

OR SubjectTerms: “ ‘informal care’ ” including this term only

OR SubjectTerms: “ ‘befriending schemes’ ” including this term only

OR AllFields: ‘caregiver’

OR AllFields: ‘care-giver’

OR AllFields: ‘carer’

OR AllFields: “ ‘informal care’ ”

OR AllFields: ‘befriending’

OR AllFields: ‘caretaker’

OR AllFields: ‘care-taker’

OR AllFields: “ ‘care taking’ ”

OR AllFields: “ ‘children caring’ ”

OR AllFields: “ ‘families caring’ ”

OR AllFields: ‘respite’

AND

AllFields: ‘metaanalysis’

OR AllFields: ‘meta-analysis’

OR AllFields: “ ‘meta study’ ”

OR AllFields: ‘meta-synthesis’

OR AllFields: ‘synthesis’

OR AllFields: “ ‘pooled analysis’ ”

OR AllFields:” ‘pooling studies’ ”

OR AllFields: “ ‘what works’ ”

149 results.

Search 7

SubjectTerms: “ ‘carers’ ” including this term only

OR SubjectTerms: “ ‘young carers’ ” including this term only

OR SubjectTerms: “ ‘informal care’ ” including this term only

OR SubjectTerms: “ ‘befriending schemes’ ” including this term only

OR AllFields: ‘caregiver’

OR AllFields: ‘care-giver’

OR AllFields: ‘carer’

OR AllFields: “ ‘informal care’ ”

OR AllFields: ‘befriending’

OR AllFields: ‘caretaker’

OR AllFields: ‘care-taker’

OR AllFields: “ ‘care taking’ ”

OR AllFields: “ ‘children caring’ ”

OR AllFields: “ ‘families caring’ ”

OR AllFields: ‘respite’

AND

AllFields: ‘medline’

OR AllFields: ‘medlars’

OR AllFields: ‘embase’

OR AllFields: ‘cinahl’

OR PublicationTitle: ‘cochrane’

OR AbstractOmitNorms: ‘cochrane’

OR AllFields: ‘scisearch’

OR AllFields: ‘psychinfo’

OR AllFields: ‘psycinfo’

OR AllFields: ‘psychlit’

OR AllFields: ‘psyclit’

OR PublicationTitle: ‘search’

OR AbstractOmitNorms: ‘search’

216 results.

Search strategy

The search strategy below incorporates a section to restrict the search to reviews only. This part of the strategy was based on the CRD search strategy for retrieving reviews from the Social Science Citation Index.

Search strategy

The search strategy below incorporates a section to restrict the search to reviews only. This part of the strategy was based on the CRD search strategy for retrieving reviews from ASSIA.

(SU.EXACT(“Caregivers”) OR SU.EXACT(“Respite Care”) OR TI,AB(caregiv* OR care-giv* OR carer* OR “informal care” OR befriending OR caretak* OR “care taker” OR “care takers” OR “care taking” OR “children caring” OR “families caring” OR respite) OR TI,AB(families NEAR/2 support)) AND (TI,AB(metaanaly* OR meta-analy*) OR SU.EXACT(“Literature Reviews”) OR TI,AB,IF(“meta study” OR meta-synthes* OR meta-evaluat*) OR TI,AB,IF(synthes* NEAR/3 literature*) OR TI,AB,IF(synthes* NEAR/3 research*) OR TI,AB,IF(synthes* NEAR/3 studies) OR TI,AB,IF(synthes* NEAR/3 data) OR TI,AB,IF(synthes* NEAR/3 trials) OR TI,AB,IF(synthes* NEAR/3 findings) OR TI,AB,IF(synthes* NEAR/3 evidence) OR TI,AB,IF(quantitative-synthes*) OR TI,AB,IF(pooled-analys*) OR TI,AB,IF((data NEAR/3 pool*) AND studies) OR TI,AB,IF(pooling NEAR/1 studies) OR TI,AB,IF(medline OR medlars OR embase OR cinahl OR cochrane OR scisearch OR psychinfo OR psycinfo OR psychlit OR psyclit) OR TI,AB,IF((hand OR manual* OR database* OR computer* OR electronic*) NEAR/3 search*) OR TI,AB,IF((electronic* OR bibliographic*) NEAR/3 database*) OR TI,AB,IF(overview*) OR TI,AB,IF("evaluation review*") OR TI,AB,IF("what works") OR TI,AB,IF("evaluation synthes*") OR TI,AB,IF(review*))

Additional limits: date – from 1 January 2000 to 22 January 2016; language – English.