Systematic review of respite care in the frail elderly

The ageing population

In 2002 4,464,000 people in the UK were aged 75 years and over1 and it is projected that the number of people over pensionable age will increase to over 15 million by 2040. 2 This will impact on health-care systems as age-related conditions become more common. In 2001 46% of people over the age of 75 years reported having a limiting long-standing illness. 3 The most frequently reported chronic conditions in people aged 65 years and over in 2001 were heart and circulatory diseases and musculoskeletal ailments. 4 Dementia is a particularly debilitating problem associated with ageing, with around one in 20 people aged 65 years and over having the condition, rising to around one in five people over 80 years of age. 5 Stroke is also one of the most prevalent causes of morbidity in older people. In the UK around 110,000 people per year experience a first stroke and a further 30,000 have recurrent strokes. 6

Provision of care for people with disabilities

Many older people with chronic conditions are cared for in the community, with their main source of support coming from informal carers. Such informal carers of the frail and elderly are frequently in mid- to later life themselves, being spouses or adult children of the care recipient. In 2001 almost 2.8 million people in England and Wales aged 50 years and over provided unpaid care for family members, friends or neighbours. In total, 24% of carers in the 50- to 60-year age group spend 50 hours per week or more on caring activities. 3,4 Although people from white British or white Irish backgrounds were more likely to be carers than other ethnic groups, this probably reflects the older age structure of the white UK-born population. However, Bangladeshi and Pakistani carers were just as likely to spend 50 hours a week or more on caring activities as their white UK counterparts, and numbers of ethnic minority carers will increase in the future as these populations age.

According to the General Household Survey3 women were more likely to be carers than men and have a heavy caring commitment of over 20 hours per week. About one-third of carers were the only means of support for the care recipient. In total, 21% of carers had been in a caring role for at least 10 years and 45% for 5 years or more; 62% of carers were looking after someone with a physical disability only, 6% with a mental disability only and 18% with both physical and mental disabilities; 14% reported caring for a person simply because they were ‘old’.

The types of help given by informal carers consisted mainly of practical help with activities of daily living (ADL) such as meal preparation, shopping and household tasks. A total of 60% reported that they ‘kept an eye’ on the person they cared for and 55% reported providing company; 26% gave more personal care such as personal hygiene and 35% reported helping with mobility.

Impact of caring on carers’ health and well-being

Caring can have a direct effect on health, such as physical strain and musculoskeletal problems, as well as causing emotional strain. It can also have an indirect effect on health status through lower earnings or income or increased costs when the recipient of care takes up residence with the carer. 7 As a result, carers tend to report poorer health than their peers who are not carers. Health is particularly poor among those who devote at least 20 hours a week to caring, with around half reporting a long-standing illness. 4 In many cases poor health is directly attributed to the caring role. In total, 39% of carers report that their physical or mental health has been impaired as a result of caregiving. Other complaints include tiredness, depression, loss of appetite, disturbed sleep, stress and short temper. Such complaints were higher in those caring for someone who lived in the same household than in those caring for someone living elsewhere, probably reflecting the number of hours spent caring and the level of care needed. 3,8,9

The impact of caring on mental health was explored in a survey carried out by the Office for National Statistics (ONS) for the Department of Health. 10 Neurotic symptoms were measured using the revised version of the Clinical Interview Schedule (CIS-R). Psychiatric morbidity was related to hours spent caring, with only 13% of those caring for less than 20 hours a week having a CIS-R score of 12 or more, compared with 27% of those spending 20–34 hours caring. Sole carers were more likely to have mental health problems than those not having the main responsibility for the care recipient. Mental health was also related to the ability to take a break from caring, with 36% of carers who had not had a break experiencing high levels of neurotic symptoms, compared with 17% of those able to take a break. Over half of the carers reported worrying about their caring responsibilities and one-third felt that caring had made them depressed. Relationships and social life were also adversely affected in around one-third of carers, which again was related to high levels of neurotic symptoms.

To capture the caring impacts on these different aspects of health and well-being, research studies have focused on the concept of ‘carer burden’. This is an all-encompassing term that refers to the financial, physical and emotional impact of caring. It may include factors such as restrictions to social activity of the carer, lack of privacy, impaired sleep, feelings of stress, satisfaction with the caregiving relationship, effects on family/job, etc. Carer outcome measures often include a general health measure and/or a standard measure of anxiety and depression. But it should also be noted that not all outcomes of caring are negative. Qualitative studies have reported positive feelings related to caring such as pride, gratification and a sense of closeness to the person being cared for. 8

The concept of carer burden is complex and is mediated by many factors. It is not necessarily the case that the carers of the most impaired patients experience the greatest stress and burden. 11 Factors such as age, gender and ethnicity play a role. Female carers experience greater burden than male carers. White carers have been reported to experience greater burden than African American carers. However, relationship may be a confounding factor in this context as white carers are more likely to be spouses and African Americans tend to be adult children of the care recipient, and it has been reported that spouses experience greater burden. 12 Other factors include carer support, carer health status, coping abilities and quality of the previous relationship with the care recipient. The type of problems displayed by the care recipient are also an important factor as it is suggested that carers of dementia patients experience greater stress than carers of individuals with physical disabilities, and it is specifically behaviour problems rather than cognitive impairment that cause most stress. 13

Definition of respite care

Respite care is traditionally defined as the provision of a temporary break in caregiving activities for the informal carer to reduce carer distress and promote well-being. 14 Respite care can be provided in a number of different ways. These include care as an inpatient of a care home or hospice, typically for 1 or 2 weeks, or adult day care (ADC) or in-home or sitting services. There are also some night-sitting services available. Care may be provided by a variety of bodies including voluntary services, social services or the NHS. However, operationalising a definition of respite care in a review is not straightforward. There are a number of situations when the carer may be physically separated from the caring role and the care recipient but the aim is not to achieve respite. For example, if the care recipient is admitted to hospital for medical treatment this may provide ‘respite’ for the carer; however, the aim of the intervention is to deal with a health event of the care recipient. The intervention will be focused on changing the health state of the care recipient and not the carer. The health and well-being of the carer may also be improved but it is difficult to determine to what extent this is due to a temporary relief in the caring responsibility or to an improvement in the care recipient’s health, functional abilities or dependence.

In an attempt to identify the specific effects of respite itself rather than interventions aimed at changing the state of the care recipient, this review takes a fairly restricted definition of respite care. The view is taken that respite is aimed at changing the well-being of the carer and so focuses on studies that explicitly state that the intervention is designed to provide respite for the carer and that assess carer outcomes. This also includes studies which evaluate interventions that have the potential to provide respite (such as day care or in-home service provision) without explicitly expressing the aim as being respite, but which focus on carer outcomes. It excludes studies that provide interventions whose primary purpose is to change the health state of the care recipient (e.g. rehabilitation interventions or highly medicalised interventions as in some palliative care contexts), as in this case it is more difficult to distinguish the effects of confounding factors. The aim was to include studies in which the normal care carried out by the informal carer is taken over for a set period of time by another person to allow the carer a break. However, it does not require the care recipient to be physically removed from the informal care context; for example, in-home care may provide respite without the carer actually leaving the home.

Definition of frail elderly

Frail elderly was defined as anyone over the age of 65 years in receipt of informal care from a relative or friend. In defining the older care recipient a cut-off of 65 years is common and most likely to be identified in studies of respite care. Frailty is not a concept that is consistently reported or defined in the relevant literature and so in this instance, with the focus on carers, it is assumed that anyone over the age of 65 years identified as having an informal carer can be defined as frail. The need for informal care suggests a certain level of disability whether it be cognitive or physical.

Inclusion criteria for quantitative studies

Inclusion criteria for quantitative studies were as follows:

• study mentions an intervention designed to provide the carer with a break from caring

• care recipient population is aged 65 years or over (or analyses carried out on a subsample of population aged 65 years or over)

• carer outcomes are measured

• respite intervention is compared with either no respite or another intervention (this included regression analyses in which respite was used as a predictor of carer outcome such as carer burden, and within-group longitudinal comparisons that reported carer outcomes before and after the delivery of a respite intervention)

• articles written in any language.

Two additional criteria were used to identify any quantitative papers including cost data:

• include all papers costing informal care, respite, carer outcomes or service usage (even if respite component not specifically costed)

• include above only if costs are directly measured.

Exclusion criteria for quantitative studies

• Exclude studies in which the intervention is designed to change the state of the care recipient (e.g. stroke rehabilitation).

• Palliative care/hospital-at-home interventions to be excluded unless stated aim is to provide respite for carer and carer outcomes are measured.

• Exclude if care recipient population is under 65 years, age of care recipient population is not discernible or outcome data cannot be identified for those in the care recipient sample who are aged 65 years and over.

• Exclude if only care recipient outcomes are measured.

• Exclude qualitative studies and observational studies having no comparison group, e.g. surveys providing descriptive data only.

Inclusion criteria for qualitative studies

A broader set of inclusion criteria were devised for assessing qualitative studies as it was felt important to assess both care recipient and carer views of their needs and preferences for respite care even if they were not actually in receipt of respite. Inclusion criteria for qualitative studies were as follows:

• study employs qualitative methods (face-to-face semistructured/in-depth interviews; focus groups; open questions in questionnaires)

• care recipients have a mean age of 65 years or over (or analyses carried out on a subsample aged 65 or over)

• study reports views of carers and/or recipients

and either:

• study reports views of respite care or study reports respite as a theme in relation to other types of care, e.g. care aimed to change the state of the care recipient

or:

• views of respite include:

  • respite care service provision/satisfaction with services

  • impact of respite on the carer and/or care recipient

  • unmet needs/perceived needs for respite care

  • reasons for utilising or not utilising respite care.

Exclusion criteria for qualitative studies

• Quantitative data reported as part of a qualitative or quantitative study, e.g. descriptive statistics.

• Data not reporting themes or concepts related to views of respite care, respite needs, use of respite or impact of respite care on carer and/or care recipient, e.g. data reporting general experiences of caring.

• Studies using direct observation methods, e.g. participant observation.

• Care recipients are under 65 years of age or data relating to those over 65 years are not discernible in the study findings.

• Non-English language papers.

In the qualitative synthesis all foreign language papers were excluded as the issue of translation and interpretation is of greater significance and would have a potential impact on the findings. It was felt that these difficulties outweighed the potential limitation of excluding relevant studies. The impact of these exclusions would depend on the similarity of the different health-care systems and any cultural differences. Although many of the good-quality European studies are published in English journals there is the possibility of relevant studies being published in the language of origin.

Qualitative studies involving direct observation were excluded to maintain comparability of the type of data included in the synthesis,15 i.e. self-reported views rather than inferences made from observation. However, no observational studies were identified in the searches.

Year of publication

The year of publication was defined by the databases searched. All years were searched for each database.

Search strategy

The remit of the current review is very broad: respite care might feasibly occur within both community and institutional settings and across many different conditions (e.g. dementia, palliative care, stroke, etc.). In addition, interventions or services designed to give carers a break from their caring role may not be explicitly labelled as respite care. Therefore, an inclusive and broad search strategy was felt most appropriate to capture all potentially relevant literature and specificity was sacrificed to some degree to maximise sensitivity.

The final search strategy was developed iteratively following discussion with the review management group (all investigators, listed as authors of this review) and carer representatives (Carers Wales). This was based on the most appropriate definitions of respite care, the target population (i.e. frail elderly) and possible respite settings. The search strategy comprised three distinct blocks: the first set of terms were designed to capture all studies reporting respite care; the remaining two sets were included to limit results to studies carried out within elderly populations and those specifically citing carers or the caring role respectively. Words and phrases within each set were combined using the Boolean OR operator; the three sets were then combined using the AND operator. Search terms were trialled initially on MEDLINE, mapping words and phrases to MeSH headings (using the .mp operator). Keywords using the .mp operator were used either in addition to MeSH headings or in place of them when they produced the same or additional hits. Input on the appropriateness and comprehensiveness of the search terms was sought from an information specialist, all members of the review group and user representatives from two UK charities. The final MEDLINE search terms are shown in Figure 1.

FIGURE 1.

MEDLINE search terms.

Data sources

The terms provided in Figure 1 were adapted as appropriate to search an additional 24 electronic databases from the earliest possible date to the end of September 2005. Searches were rerun until the end of 2005, resulting in an additional 332 references not previously identified. All databases searched (and number of hits retrieved from each) are shown in Table 1.

Once duplicates were removed a total of 12,992 unique references were identified. Overlap between databases was substantial. Taking four of the major electronic data sources as an example, 64% of citations were identified on MEDLINE and original citations on EMBASE accounted for a further 24%, PsychInfo for 11% and the British Nursing Index for just 1%.

Hand searching of the following journals was also undertaken from the earliest possible date to the end of 2005: Gerontologist, Journal of Gerontology, Age and Ageing, International Psychogeriatrics, Journal of Palliative Medicine and Stroke.

Study selection

A preliminary title sift of all 12,992 references was undertaken by two reviewers. Obviously irrelevant titles such as those relating to respite for carers of children or pharmacological interventions were excluded at this stage. For any titles on which disagreements occurred the abstracts were assessed, along with all of the remaining abstracts, by two reviewers. When disagreements occurred papers were selected for full retrieval. Inter-rater agreement ranged from fair to moderate (kappa coefficient range 38–52). 16 At the full paper stage all exclusions were checked by a second reviewer selected from within the project management team. Inclusions of papers in the meta-analysis were checked by group discussion of the statistical team (RM, KH, KA), and inclusions of papers in the narrative syntheses (longitudinal and cross-sectional observational studies and qualitative studies) were checked by CS. The number of studies included at each of these stages is shown in Table 2.

Before full paper retrieval all quantitative, qualitative and cost papers were grouped together. Following full paper retrieval papers were categorised according to their content (i.e. quantitative/qualitative). All full papers were assessed against the inclusion criteria by a single reviewer, with any excluded papers checked by a second reviewer. The number of papers categorised into each component of the review on full paper retrieval and second-stage inclusion assessment are shown in Table 3.

A much greater number of qualitative studies was retrieved than originally anticipated. This is due in part to the wider remit and less stringent inclusion criteria adopted for qualitative studies and in part to reliance on qualitative methods in an area in which it is difficult to carry out controlled trials for ethical reasons.

Quality assessment of quantitative studies

Numerous tools are available for the quality assessment of randomised controlled trials (RCTs), with at least 25 currently in use. 17 However, a MEDLINE search from 1990 to 1997 did not identify any quality checklists for assessing cohort and case–control studies. 18 A brief review of the literature by the current authors to the end of 2005 indicated that this situation has changed very little. Given the broad and inclusive nature of the current review it was important to identify a tool that could be used to assess the quality of varied quantitative designs simultaneously (i.e. RCTs and cohort and case–control studies).

Two particularly relevant quality checklists were identified from a brief review of the available literature. 18,19 Downs and Black18 developed a tool to assess the quality of both randomised and non-randomised designs. The tool comprises a 27-item checklist and an overall score pertaining to the quality of the study. Checklist items relate to the appropriateness and adequate description of the hypotheses, study design, intervention, main outcomes and methods of analysis. The checklist demonstrated good inter-rater reliability, although further development and testing of the tool was recommended. The tool devised by Kmet et al. 19 was also intended for quality assessment of both randomised and non-randomised designs and was produced following a review of the relevant literature and discussion of issues central to internal validity. The checklist provides an overall summary score, although the authors acknowledge this approach is inherently prone to bias. In addition, inter-rater reliability appeared somewhat limited (a subsample of 10 studies scored by two reviewers). The Kmet et al. 19 checklist contains 14 items relating to study design, intervention, outcome measurement and methods of analysis.

Within the context of the current effectiveness review both tools were felt to contain useful elements but each had particular drawbacks. For example, the Downs and Black18 checklist is heavily weighted towards randomised designs (likely to be small in number in the current review) and is also lengthy at 27 items. Although more concise, comprising just 14 items, the Kmet et al. 19 checklist is less detailed (e.g. adequate description of the intervention is not included). In addition, previously developed tools did not accommodate particular issues relevant to this review, such as the presence of two samples of interest (carer and care recipient). A single quality checklist was therefore created, in line with CRD recommendations,15 specifically designed for the current review (see Appendix 1) but likely to be of value in reviews of similarly complex areas encompassing few randomised trials.

The current tool was also developed within the framework recommended by the US Preventive Services Task Force. 20 The first of three strata within this framework relates to quality assessment at the individual study level. The framework does not give a quantifiable score but provides an indication of quality based on certain parameters. Study designs are first organised into a hierarchy [RCT, non-RCT, cohort, case–control, observational (i.e. multiple time series, case studies, opinion of experts)] and are then classified as ‘good’, ‘fair’ and ‘poor’ according to criteria specific to the particular study design.

The final checklist contains 18 items, with three levels of quality assessment: good (2), fair (1) or poor (0). Some items within the list are relevant only to RCTs; therefore, a ‘non-applicable’ option is provided for other study designs. Scores across items are summed to create a quality score, which is represented as a percentage to account for any non-applicable (i.e. missing) items.

Quality assessment of qualitative studies

There is some debate as to the appropriateness of formal quality assessment in qualitative research and the use of such tools is comparatively new. Qualitative research is extremely useful in addressing patient-centred views in health-care research and as such is a valuable and often expected study component. Controversy relating to the appropriateness of quantifiable quality assessment arises from the belief that this serves to stem the interpretative and creative aspects of qualitative study. 21 Nonetheless, many believe that some form of quality assessment is necessary if qualitative research is to be taken seriously within the wider research community. 22–24 Within the context of the current review, equal weight and importance are given to both the quantitative and the qualitative components and therefore a common approach to assessing study quality was needed.

To maximise consistency across these two aspects of the review, the aim was to develop a quality tool similar in structure to that of the quantitative tool previously described, keeping in mind the different aims of qualitative and quantitative research. Quantitative research seeks to eliminate bias to render results generalisable to the wider population, whereas qualitative research is context bound and seeks to expose and discuss bias. For these reasons it has been proposed that a common language may be misleading. 22 Alternatives to terms such as internal and external validity, reliability and objectivity have been proposed, for example credibility, dependability, transferability and confirmability. 25 Others, however, feel that issues of validity and relevance are appropriate to qualitative research even though the concepts supporting them may be dissimilar to those pertaining to quantitative research. 22,23

However, it has been pointed out that there is a need to qualify this by ensuring that the different paradigms within qualitative research are acknowledged. 26 Qualitative research is not a unitary activity and aims and methods will vary according to the philosophical underpinnings and requirements of the study. This could be considered similar to the varied approaches in quantitative research with the resultant difficulty of establishing a quality assessment tool that is appropriate to all types of study. However, a number of concepts are relevant across study types, if interpreted somewhat differently. To identify items for inclusion in the assessment tool, a review was undertaken of papers that presented either a quality checklist or a narrative account of quality assessment. This scoping review followed a similar approach to that used by Walsh et al. ,27 who adopted such a scoping method designed to assess commonalities between quality assessment tools and eliminate redundant items. The review by Walsh et al. 27 was based on seven existing checklists: the checklist that they produced was included in the present scope.

The scope of the quality assessment literature revealed considerable overlap and agreement between studies in terms of the relevant criteria for assessing quality. Items to be included in the tool were chosen based on their frequency of occurrence in the articles reviewed, their appropriateness to the requirements of the current review and their generalisability across different qualitative methods. However, the types of study likely to occur within the context of respite care will largely comprise thematic analysis, such as grounded theory or phenomenology (direct observational studies or data from sources other than focus groups, interviews or open-ended questions were excluded). Some quality checklists were quite broad and vague although they had the advantage of appearing shorter and more succinct than others. A more detailed and structured approach was preferred in order to give clear definitions to facilitate interpretation and increase inter-rater reliability. The rating format was based on the checklist developed by Kmet et al. 19 for quality assessment of qualitative research, in line with the format used for the quantitative tool. The tool was piloted and amended; items included in the final version are shown in Appendix 2. Three levels were assigned to each item in the tool, which were scored from 2 to 0. The scores could then be summed to produce an overall quality rating.

Data extraction for quantitative studies

A paper version of a quantitative data extraction form was circulated to all members of the review group for comment and revised appropriately. Members of the study team could use either a paper or an electronic version of the data extraction form at this stage. The electronic version comprised an Access database with identical fields to the paper version. When extraction was completed using the paper version the information extracted was entered into the Access database to enable a direct comparison of all data. The use of Access forms for data entry also aided in ensuring consistency of the extracted data by allowing only certain types of options to be entered in any one field, thus ensuring that all data were categorised in a similar way.

At the first stage of data extraction detailed information relating to study methods was extracted, which included information about the intervention (e.g. type, setting, duration and length of follow-up), carer and recipient characteristics, and types of outcomes measured (including information on the tool used to measure each outcome). The first-stage data extraction form is given in Appendix 3.

The second stage involved more detailed extraction of appropriate numerical data for all studies categorised as either randomised trials or quasi-experimental designs into an Excel spreadsheet. The same procedure was followed for all longitudinal studies in which participants served as their own control (i.e. outcomes measured before and after an intervention in a single group). Summary tables detailing the observational studies (both longitudinal and cross-sectional) were also created (see Appendices 6 and 7 respectively)

Data extraction for qualitative studies

The qualitative review followed a similar pattern with an initial meta-methods analysis in which data were extracted to a summary table (see Appendix 9). A meta-data analysis was then carried out on a subsample of papers; the findings from the qualitative papers were extracted into a software package specifically designed for qualitative analysis (NUDIST version 6). Data extracted comprised the concepts identified in the results sections of the papers but not the themes defined by the researchers or their conclusions derived from their analyses, usually presented in the discussion sections of the papers.

Quantitative data synthesis

When appropriate and possible, quantitative results from individual studies were synthesised using meta-analysis techniques, taking account of statistical, clinical and methodological heterogeneity. 28 To account for the variety of ways in which some outcomes such as carer burden and depression are measured, standardised effect sizes were used.

Initially, between-study heterogeneity was investigated within randomised and quasi-experimental studies. Separate meta-analyses were carried out for each carer outcome using the following study-level covariates when possible: length of follow-up, length of intervention (i.e. brief versus sustained) and respite setting (e.g. day versus home care). A number of studies measured outcomes at two or more follow-up periods; therefore, additional separate meta-analyses were carried out splitting studies into short- and longer-term follow-up groups.

Meta-analyses were carried out both on follow-up data only and on change scores. However, standard deviations for change scores (change SDs) were rarely provided. Change SDs were therefore estimated using two previous assumptions: first, that the correlation between baseline and follow-up scores is zero and, second, that the correlation between baseline and follow-up scores is likely to be 0.6, based on data from the Rothman et al. 29 study. This study reported correlation coefficients for a large number of outcomes from baseline to both short- and long-term follow-up. Outcomes for carers were the primary focus of the meta-analyses: however, likelihood of institutionalisation was also assessed to represent a patient outcome relative to either positive or negative effects of a respite intervention. This was felt to be an important step as a preliminary scoping of the qualitative literature indicated that a common reason for non-uptake of services is concern on the part of the carer that respite may be detrimental to the care recipient. All meta-analyses were carried out in Stata (version 9).

Studies in which outcomes were measured in a single group before and after delivery of an intervention were analysed separately.

It was intended to assess negative publication bias by funnel plots but there were few studies eventually included in each analysis and so this was not feasible. To assess any potential for publication biases the country of origin and year of publication were examined across the different types of study design.

Observational studies identified by the review (both cross-sectional and longitudinal) formed the basis of a narrative synthesis, with particular reference to the rate of institutionalisation amongst those accessing respite compared with non-users and the impact on carer burden and mental health. However, much of the observational work in the area tends to focus on use of respite as a generic, dichotomous outcome and so differentiating between the effects of diverse respite settings (i.e. home care, day care) is problematic.

Qualitative data synthesis

The methods used to review the qualitative literature followed those used in carrying out primary qualitative research and were based on the methods of meta-study described by Paterson et al. 30 The synthesis aimed to be both interpretive, to provide further explanation of the research findings in the quantitative review, and aggregative, to identify the extent of the literature and gaps that need to be addressed. To do this a three-stage process was adopted by, first, carrying out a meta-method analysis, second, a meta-data analysis and, finally, meta-synthesis.

Meta-method analysis assesses both the quality of the research methods of the primary research papers and the ways in which the methodological context may have influenced the study findings. 31 Each paper was summarised into a table under the headings shown in Appendix 9. Separate tables were constructed depending on the country in which the research was conducted, and the factors influencing study findings could then be explored. For example, as well as differences in sampling procedures, variation in data collection methods might have an impact on study findings, such as data collected by face-to-face interview versus focus groups. Such tables may also provide information about the generalisability of findings if consistent results are found across samples and contexts (such as place of care) and also about the extent of the literature and any gaps for future research. The listing of sampling procedures could reveal the types of carers and care recipients whose views were not sought, giving an indication of the representativeness of the findings. These tables also give a view of the literature over time, as preferred methods have changed and developed, and how the field of research is likely to develop in the future.

Because of the large number of studies identified in the qualitative literature search, a purposive sampling technique was used in the meta-data analysis stage. At the outset of the study we had intended to sample according to type of respite provision (e.g. institutional care, day care, home care) and characteristics of the care recipients (e.g. dementia, physically impaired, palliative care). However, such categorisations were not possible as the majority of studies reported a mix of respite use and often a mix of care recipients. We therefore decided, in the first instance, to focus on the organisational context of studies and relevance to UK policy. Accordingly, all UK studies were included. Although there were a substantial number of studies carried out in the UK, studies carried out in the USA were also prominent and tended to be of higher quality, with a more direct focus on respite care issues. We considered that the concerns of carers of older people in the USA would be similar, within the Medicare system, to those of carers in the UK and so these too were included, along with all studies conducted in Canada, where the health-care system is more similar to that in the UK. Also included were studies carried out in Australia and New Zealand, where there are similarities with the UK in culture and health-care systems.

The meta-data analysis stage was carried out using similar methods to those used in primary qualitative studies,31 with each study representing a case. This was based on a grounded theory approach31 although certain aspects of the study had limitations in relation to grounded theory methodology. Because a discrete set of studies was available, theoretical sampling could not be carried out, and therefore data saturation may not necessarily have been achieved for all categories and themes. In addition, although a wide range of studies was included, representing general views of respite care, and coding aimed to focus on emergent themes rather than themes identified a priori, it is likely that the coding process was influenced by the main research question related to identifying barriers to uptake. However, there were no previous assumptions concerning the nature of barriers or other views of respite care.

The findings from each study (the concepts as reported by the authors) were extracted. Category codes were developed using a constant comparative technique. 31 Common categories were extracted from the studies by comparing for similarities and differences between the concepts expressed in each study. As concepts emerged that did not fit the coding frame a new category was instigated. The data within each category code were then compared to identify subcategories describing the range of the general properties of each main category. Characteristics of the studies were coded as base data, for example quality rating, type of data collection, characteristic of care recipient, etc. A sample of texts was coded by two reviewers to assess the reliability of the coding framework. The coding demonstrated a high level of concordance, showing good reliability. By comparing the data in the categories across groups of cases (i.e. studies) and in relation to other categories, hypotheses concerning the causal, contextual and intervening relationships between categories and subcategories were developed in a process of ‘axial coding’. 31 Data were then sought across the different studies that either supported or refuted these hypotheses.

Finally, in the meta-synthesis stage of the analysis ‘selective coding’31 was carried out, whereby a core category was identified (i.e. barriers to respite), which became the central focus of the analysis, and a theory developed concerning the causal relationships between this and the other major categories. This core category was to some extent defined by the research question as studies were selected based on their ability to answer this question. However, the category ‘barriers to respite’ did fulfil the criteria for a core category31 and related to all of the other major categories apart from three, which are reported separately (i.e. ethnicity, positive aspects of respite and palliative care). In addition to this, and for completeness, a descriptive analysis of the data occurring under the category ‘carer needs’ is reported separately although these data also related to the core category.

The final stage of the analysis described above provides a theory of the barriers to the uptake of respite over and above that described in individual studies. It is based on integration and interpretation of the data (rather than merely aggregation) and takes account of the methodological aspects of the studies reviewed by including design features, such as carer and care recipient characteristics and quality ratings, as categories within the coding frame. As such it can be considered to represent a synthesis of the data, although there are limitations concerning the contribution of meta-theory analysis (analysis of the theoretical approaches underpinning primary studies) prescribed by Paterson et al. 30 for meta-synthesis.

These findings were then integrated with the findings of the quantitative review. One important feature was to identify whether the outcomes addressed in the quantitative studies were consistent with those identified in the qualitative studies as being important for both care recipients and carers. The findings from the qualitative review were used to shed further light on findings in the quantitative review and aid interpretation.

Characteristics of RCTs/quasi-experimental studies and longitudinal before-and-after studies

The majority of randomised and quasi-experimental studies assessed day care and mixed respite care interventions, followed by in-home care and then institutional care (Table 4). Most studies were carried out in the USA or UK, with the USA having nine studies and the UK having five; none of these studies assessed institutional care. The remaining studies were carried out in Canada, Australia, New Zealand and Germany.

Similarly, the majority of longitudinal before-and-after studies were carried out in the USA and UK. These studies were more evenly spread across the different types of respite, although only one study (carried out in the UK) assessed in-home respite (Table 5).

Studies excluded from the meta-analyses

Table 6 summarises the reasons why RCTs and quasi-experimental studies and longitudinal before-and-after studies (identified for inclusion in the review) were excluded from the meta-analyses. Ten randomised and quasi-experimental studies were excluded, as well as six longitudinal before-and-after studies. A narrative summary of the studies excluded from the meta-analyses is presented first, followed by the meta-analyses according to each outcome.

Narrative review of randomised and quasi-experimental studies excluded from the meta-analyses

Narrative review of longitudinal (before-and-after) studies excluded from the meta-analyses

Six studies (reported in seven papers) were excluded from the meta-analyses of single-group longitudinal comparisons (i.e. before-and-after studies).

Adler et al. 57 found that levels of carer burden and depression were reduced during a 2-week inpatient respite intervention, but that levels returned to baseline once patients had returned home, which suggests that the effects of respite may be short-lived in some instances.

Johnson and Maguire68 examined the impact of the use of day care on a range of carer outcomes and found no difference in carer stress between baseline and follow-up (2 and 4 months).

Chi and Wong58 studied the effect of institutional respite on carer attitudes at 1 month. They found that carers were less likely to express a wish to institutionalise recipients following respite; however, perceptions of the caring role as stressful actually increased following respite.

Gilleard and colleagues^[62,63^] examined the effects of carers’ psychological well-being and self-reported strain and also the number of care recipient problems on community status after 3 and 6–7 months of day care. Institutionalisation was predicted by the number of patient problems and carer psychological distress. Day care itself was associated with reduced distress for the majority of carers; for those in whom day care did not help to alleviate psychological distress, institutionalisation had a significant positive impact in terms of this outcome.

Cox59 examined a mixed respite programme for carers of dementia sufferers, which allowed families to buy up to 164 hours of respite, consisting of in-home care, institutional care (4–5 days) or day care. Follow-up was carried out at 6 months and African American participants were compared to white participants. There was no reduction in anxiety or depression in either group but there was a decline in carer burden in both groups.

Deimling60 also examined a mixed respite programme for dementia carers, consisting of short institutional stays, day care and home health aides. Follow-up was carried out at 4–6 months and assessed depression, symptoms of health problems and relationship strain. Comparisons were made between carers of those with stable conditions and carers of those with declining conditions. Carers of stable recipients had decreases in depression, health problems and relationship strain whereas outcomes for carers of recipients with declining conditions either stabilised or deteriorated.

Meta-analysis: the effectiveness of respite care on carer well-being

The effectiveness evidence for respite care in terms of carer well-being is outlined in the following sections, presented for each outcome separately. Meta-analyses of randomised and quasi-experimental studies are given first, followed (when applicable) by the results from meta-analyses of single-group comparisons (longitudinal before-and-after studies) to examine any differences in terms of effects. All results from meta-analyses (Cohen’s method) are based on change scores and, when the change standard deviation is missing, a 0.6 correlation between baseline and follow-up is assumed. 29 Fixed models were initially fitted, except when tests for heterogeneity were statistically significant at the 5% level. In this instance a more conservative random-effects model is presented and resulting changes discussed. When significance was between the 5% and 10% levels both fixed and random models are shown. Appropriate descriptive findings are also summarised at the end of each section (outcome). Results are therefore presented in terms of a hierarchy of evidence quality; the relative quality of included studies is also discussed within each outcome.

Carer burden

Carer burden in randomised and quasi-experimental studies

Three studies provided sufficient data on carer burden for inclusion in the meta-analysis: two RCTs29,32 and a single quasi-experimental study. 41 Two studies29,32 assessed day care interventions and the other41 both day and home care. Care recipients were frail elders,29 elders with dementia41 and elders experiencing mixed problems. 32 Two of the studies were carried out in the USA29,41 and one in Canada. 32 Carers in one study29 were followed up at 6 and 12 months; the length of follow-up for the other two studies32,41 was 3 and 6 months and, therefore, only the 6-month follow-up period in the Rothman et al. 29 study was included in the meta-analysis. All interventions comprised day care and were delivered continuously over the respective follow-up periods. No significant effects of respite care on carer burden were observed (Figures 3 and 4) in either fixed or random models. The two RCTs can be seen to be closer to the line of no effect than the quasi-experimental study.

FIGURE 3.

Carer burden in randomised controlled trials/quasi-experimental studies (fixed model) – 6-month follow-up (sample sizes in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

FIGURE 4.

Carer burden in randomised controlled trials/quasi-experimental studies (random model) – 6-month follow-up (sample sizes in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Carer burden in longitudinal before-and-after studies

Four longitudinal studies comprising a single-group (before-and-after studies) were included in a meta-analysis of carer burden. 61,67,69,70 Two studies were carried out in Canada using a day care intervention;61,70 one in the UK on a combination of institutional and day care;67 and one in the USA on a combination of home and institutional care. 69 The two studies focusing on day care gave similar levels of respite of around 2 days per week, but the study using a combination of day and institutional care did not give any information on the amount of respite provided. Three of the studies focused on frail elders61,69,70 and one on care recipients with dementia. 67 All studies measured burden at multiple time points (3 and 6 months;61,67 2 and 6 months;70 6 and 12 months69). Warren et al. 70 also measured burden at 2 weeks post-respite; this follow-up measurement was excluded from the meta-analysis. Therefore, two meta-analyses were carried out, one at short-term follow-up (2–3 months) and one at longer-term follow-up (6 months); the 12-month follow-up69 was not included in the meta-analysis. At both short- and longer-term follow-up tests for heterogeneity fitting a fixed model were significant (p = 0.0000); therefore, results of random models are presented (Figures 5 and 6). The only individual study with a positive significant effect at either follow-up67 was focused on care recipients with dementia rather than frail elderly more generally and used a combination of institutional and day care.

FIGURE 5.

Carer burden in longitudinal before-and-after studies at 2–3 months’ follow-up (random model) (sample sizes in brackets). CI, confidence interval; ES, effect size.

FIGURE 6.

Carer burden in longitudinal before-and-after studies at 6 months’ follow-up (random model) (sample sizes in brackets). CI, confidence interval; ES, effect size.

Quality and design characteristics of studies included in the meta-analysis of carer burden

As seen in the analysis presented above, only two RCTs29,32 assessed carer burden. Neither of these studies found a significant effect of day care on carer burden. Only one of these trials was rated as high quality,32 having scored highly on all attributes on the quality assessment. This study examined a day care programme that included some functional and psychosocial activities, although these were group based, which is common in many day care facilities. The day centres did have access to a range of staff such as nurses, recreation technicians, special care counsellors and drivers, with possibly a rehabilitation technician, occupational therapist and psychosocial worker in some facilities. The intervention was fairly active but the main aim was of support rather than medical intervention. In addition, any attendees who required individual intervention were excluded from the analysis, thus excluding those who were having more treatment-focused interventions. However, there were some limitations in relation to external validity as only 34% of participants attended the facility at least once a week. The majority, therefore, had low exposure to the intervention, which may have been insufficient to exert any effects on carers. A subgroup analysis was carried out of high attendees (those attending at least once a week) and those attending less often. Carers of high attendees were substantially less burdened post-intervention whereas carers of low attendees had a slight increase in burden, although the difference was not statistically significant. Although the authors acknowledge that little weight can be placed on conclusions as there may be confounding factors in such an analysis, they suggest that future studies should aim to encourage a level of attendance that would be felt to be of consequence for carers. In addition to this, both intervention and control groups could access other services involving respite if they desired, but this was not measured or accounted for in the analysis.

The other RCT29 was of moderate quality and details had to be gleaned from a number of papers reporting different aspects of this large study. The sample in the analysis included here was not representative of the carer population generally as the trial was carried out in Veteran Administration facilities; the majority of care recipients (96%) were therefore men. The care recipient population further differed from a general community population in that 66% were in hospital at recruitment and were at high risk of nursing home placement on discharge, with the intervention being offered as an alternative to residential care. There were limited details of care recipients’ characteristics or context, but it is likely that there were more carer and care recipient dyads in crisis situation than in the population in general. There was little description of the services provided although, as in the previous study, some additional services were offered such as occupational, physical and recreational therapy. There were also some more individualised services, such as monitoring of complex medications. However, the overall aim of the intervention was focused on support, to allow people to remain at home by providing respite, motivation for self-care and stabilisation of health status. Uptake of the intervention was said to vary considerably with some not attending at all or for very few days; however, actual uptake in the group was not specified, and neither was use of other support services during the time period of the study. The control group received customary care but it was unclear what this involved, although it was apparent that this could be nursing home as well as community care.

The only study in this particular analysis reporting a positive effect of respite on carer burden was a quasi-experimental study,41 which was rated to be of high quality. The main difference between this and the other two studies is that the intervention included more flexibility of respite options. At three sites in-home and day care respite were offered with no limitations on access; two sites offered in-home care, both day and evening with a flexible schedule, and two sites offered only day care on weekdays from 8am to 5pm. It was reported that respite workers received special training but beyond this there was no further definition of the intervention. A further notable difference in this study is that all participants used the respite services, with mean use being 220 hours over the period of the study (range 4–1137 hours). A major issue with quasi-experimental studies is the potential for bias in sampling. In this study waiting list control subjects were used but the majority were recruited from just one site. This had an impact on comparability of the intervention and control groups as they differed on race and income (the control group had more ethnic minorities and a lower income). There were no other statistically significant differences between the groups. Consequently, race and household income were controlled in the analysis, and the positive effect of respite remained. Use of other services was not restricted but was also controlled in the analysis. Finally, the sample of care recipients in this study were people with dementia, whereas care recipients in the other two studies had a range of physical and cognitive disabilities.

Four before-and-after studies assessed carer burden, only one of which was rated as being of high quality and which found no effect of day care. 61 One was rated as being of moderate quality67 and demonstrated a positive effect of institutional and day care, and two were rated as being of lower quality,69,70 both showing no effect. The high-quality study61 assessed burden following ADC in a geriatric day hospital. The aim of the programme was to maintain people in their living environments, but being a day hospital as opposed to a day centre participants had access to a more medicalised support team of nurses, physicians, physical therapists, occupational therapists, recreational therapists, a neuropsychologist and a gerontopsychiatrist. All participants in the study had received services from at least two of these categories, which may or may not have involved medical intervention. However, description of the study intervention was incomplete, which was reflected in the quality assessment scoring. Whereas this study assessed a range of physical and cognitive disorders, the study rated as being of moderate quality67 focused on care recipients with dementia. This study showed a positive impact of the intervention on carer burden but again the intervention was poorly described, which in this instance was more problematic, as a range of services were offered, not all of which necessarily involved a respite element. Those specifically aimed at respite provision involved day care and institutional care; 50% of participants received institutional respite and 69% day care, although these were not received in isolation. Other interventions included a social care worker scheme, home help, inpatient access, carers group and B-grade nurse. Participants were assessed by a social worker or community psychiatric nurse on entry to the intervention and an individualised programme of care was devised. As well as poor description of the intervention characteristics and the amount and type of intervention received by participants, other aspects rated on the quality assessment form received only moderate scores (i.e. caregiver characteristics poorly described, incomplete control for population characteristics and methods of analysis inadequately described). As it was not possible to identify which particular aspects of the intervention had a positive effect, a logistic regression analysis was carried out, which suggested that only institutional respite had a positive effect. Those receiving day care appeared to be negatively affected. It is not clear if any other factors were included in this analysis and so there may be some confounding.

Of the studies rated as low quality, one assessed day care70 and the other69 in-home and institutional care; neither of these studies showed any effect of the respective interventions. Warren et al. 70 included both day hospitals (nine sites) and day support programmes (five sites); the majority (69%) of the sample was recruited from the day support programmes and received on average 10.4 hours of respite per week. No differences were detected in any of the outcomes between those attending day hospital and those attending day support and so the sample was analysed as one group. There were no details about the structure or content of the intervention programmes and the characteristics of care recipients were not described. Other aspects of the study receiving low ratings on the quality assessment form were selection methods (not completely described), population characteristics (not controlled for), intention to treat analysis (no attempt made) and conclusions (not all supported by the study).

The other study of low quality,69 which assessed in-home and institutional care, had a large and significant problem of attrition. Details of the carer sample were given for the 130 participants for whom data were available at baseline, but follow-up data at 6 months were only available for 18 carers. The intervention was a nurse-managed co-ordination of the available services. Trained volunteers provided up to 4 hours a week of in-home respite and institutional care was provided in a long-term care facility, but further details of the use and process of the intervention were lacking. However, the maximum potential usage of respite was low. The study design was generally poorly described, as were the selection criteria for the sample, and potential confounders were not addressed in the analysis.

Summary of studies assessing the impact of respite care on carer burden (studies included and excluded from the meta-analysis)

The analysis involving RCTs and quasi-experimental studies failed to show any evidence of an impact of respite care on carer burden. Although the two RCTs included in the analysis were of high and moderate quality, it must be noted that the sample characteristics of one were not generalisable to the population of carers as a whole29 and the other32 acknowledged difficulties in interpretation as uptake of respite was low in the sample. It is possible, therefore, that respite use was insufficient to show any effect in that particular study. The only study included in this analysis to show a positive effect of respite was a quasi-experimental study,41 which was also of high quality. A notable difference between this and the other two studies was the type of respite provision, which involved a range of day care and in-home provision as opposed to day care alone.

Before-and-after studies, on the other hand, demonstrated a positive impact of respite on carer burden. Two of the studies included in this analysis were of low quality;69,70 only one was rated as being of high quality61 and the other was rated as being of moderate quality. 67 The interventions used in these studies, again, consisted mainly of day care provision. However, one study stood out as reporting a large positive impact of respite,67 and this involved a combined intervention of day and institutional care and also a number of other services such as home help, which may also have provided some respite, as well as interventions not providing respite. Although this study offers some consideration of combined approaches to respite provision it also poses some difficulty in interpretation as it is not possible to determine the specific effects of those interventions providing respite.

None of the studies in this analysis was without problems in its applicability to the research question of this review. The range of methods, interventions and sampling used leaves no clear conclusions other than to say that evidence is lacking.

The narrative synthesis of studies not included in the meta-analysis indicates that day care54,68 and home care52 alone may have little or no significant impact on carer burden or self-perceived caring-related stress. At longer-term (6–12 months’) follow-up a number of studies42,43,49 also failed to detect a significant effect of a mixed service on burden. Further evidence for the short-term effects of respite on burden is indicated by the finding57 that burden decreased during a period of inpatient respite, but that levels returned to baseline when care recipients returned home. Some research58 asserts that levels of carer stress may actually increase following a period of respite. A number of studies reported a positive effect of respite on burden35,36 compared with controls, yet follow-up periods were less than 2 months.

In studies evaluating the effects of different types of day care (day care/day hospital34), or mixed respite relative to other interventions (e.g. support groups, education45,46), no differences were observed between groups in terms of burden.

Carer depression and psychological well-being

Carer depression in randomised and quasi-experimental studies

A total of six studies (of which two were RCTs) were included in a meta-analysis of the effects of respite on depression. 29,40,44,47,48,55 Two studies29,55 measured depression at two follow-up times (6 and 12 months and 3 and 12 months respectively). Two separate analyses were therefore carried out examining the effects of both short- and long-term follow-up.

Carer depression: short-term follow-up

Five studies were included in the analysis for short-term follow-up. 29,40,44,47,55 Four of the studies were carried out in the USA29,40,47,55 (three assessing day care29,47,55 and one home care40). Only one was carried out in the UK,44 assessing home care. 44 Three of the studies focused on care recipients with dementia40,47,55 and two focused on frail elderly care recipients. 29,44 Length of follow-up ranged from 1 to 6 months. As the test for heterogeneity was significant in this instance, the effect of fitting a random model was evaluated (Figure 7); the effect just fails to reach significance in favour of respite care. The studies reporting the largest effects in favour of the intervention were quasi-experimental rather than trials.

FIGURE 7.

Carer depression in randomised controlled trials/quasi-experimental studies – short-term follow-up (1–6 months) (random model) (sample sizes in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Carer depression: long-term follow-up

Three studies were included in the meta-analysis of depression at longer-term follow-up. 29,48,55 Two studies were carried out in the USA,29,55 assessing day care in the frail elderly29 and in care recipients with dementia. 55 The third study assessed a combined day and home care intervention for dementia care recipients in the UK. 48 The follow-up period for all studies was 12 months. As the test for heterogeneity was significant in this instance, the effect of fitting a random model was evaluated (Figure 8), which showed a non-significant overall effect.

FIGURE 8.

Carer depression in randomised controlled trials/quasi-experimental studies – long-term follow-up (12 months) (random model) (sample sizes in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Depression and respite setting

Separate meta-analyses were carried out to determine the relative influence of day and home care. Four studies evaluated day care services29,47,48,55 and two home care. 40,44 Using a fixed-effects model significant effects in terms of a reduction in depression were not observed for day care services (Figure 9).

FIGURE 9.

Carer depression in randomised controlled trials/quasi-experimental studies day care (fixed model) (sample sizes in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

As the test for heterogeneity was significant, a random-effects model was used in assessing the effects of home care on carer depression (Figure 10). This resulted in a slighter larger but non-significant effect size.

FIGURE 10.

Carer depression in randomised controlled trials/quasi-experimental studies – home care (random model) (sample sizes in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Carer depression and length of intervention

Three of the studies evaluated the effects of a long-term intervention (as opposed to long-term follow-up of a shorter intervention) provided over a 12-month period29,48,55 and two of the studies40,47 evaluated the effects of short-term respite interventions (the former for 2–8 weeks). Neither long-term (random model; Figure 11) nor short-term (fixed model; Figure 12) interventions were found to significantly impact on depression.

FIGURE 11.

Carer depression in randomised controlled trials and quasi-experimental studies: long-term intervention (random model) (sample sizes in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

FIGURE 12.

Carer depression in randomised controlled trials/quasi-experimental studies: short-term intervention (fixed model) (sample sizes in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Carer depression in longitudinal before-and-after studies

Three single-group longitudinal studies were included in a meta-analysis of depression (all continuously available interventions: 3-month follow-up of home care;64 depression measured during an inpatient respite intervention;66 6-month follow-up of a mixed respite programme69). Two of these studies were carried out in the UK66,69 and one in the USA. 64 The impact of respite on depression is not statistically significant when a random-effects model is fitted (Figure 13).

FIGURE 13.

Carer depression in longitudinal before-and-after studies (0–6 months) (random model) (sample sizes in brackets). CI, confidence interval; ES, effect size.

Quality and design characteristics of studies included in the meta-analysis assessing the impact of respite on carer depression

Three studies used RCT methodology29,40,47 and three a quasi-experimental approach. 44,48,55 The RCTs were rated as being of moderate quality; two assessed day care and one assessed home care. None of these studies showed any significant effect of respite on carer depression. The Rothman et al. study29 has already been discussed in relation to carer burden. Although a fairly well-conducted RCT, the main issues in this study concerned the generalisability of the sample (being nearly all men attending Veteran Administration facilities); the majority being hospital inpatients at the time of recruitment; and the potentially low rate of uptake of day care services (although insufficient details given on this).

The Quayhagen et al. study47 was of moderate quality and compared a range of support interventions for a group of carers of care recipients with dementia with a waiting list control group. Day care was provided as one of the interventions with the main aim of providing respite. Two support group sessions for the carer were also included in this arm of the trial. Care recipients received 4 hours a week of day care over a period of 8 weeks, which is a low level of respite provision and which may have been insufficient to have a significant impact on outcomes. The intervention included structured activities designed to stimulate the participant cognitively and socially. The quality assessment identified poor definition of carer characteristics, randomisation methods and whether allocation concealment took place. In addition, it was unclear whether there was any other use of respite services in either the intervention or control groups.

The final RCT40 examined home care and was more specifically focused on respite by providing trained respite carers for up to 6 hours a day for 10 days in a 2-week period. Therefore, although the amount of respite offered was fairly substantial, the length of the intervention was very short. It is debatable whether an intervention of only 2 weeks would impact on depression ratings. In-home care was chosen in this study by assessing preferences for type of care beforehand. The use of other services was controlled during sample selection by including only those who received other respite services for less than 8 hours per week. The sample was considered in relation to level of vulnerability, based on the number of hours a day spent caring and the level of respite received. Vulnerable carers took up more of the respite offered in the intervention (56 hours) whereas non-vulnerable carers took up about half the amount (27.6 hours), although the levels were similar in the two groups when five carers were excluded who did not take up respite at all. A number of physiological measures of stress were also carried out, with the finding that plasma adrenaline declined significantly in the vulnerable carers who received respite but rose in those in the control group. The quality assessment form showed low or moderate ratings for information on care recipient characteristics and incomplete control/description of population characteristics in the analysis.

The only study to be rated as high quality was a quasi-experimental study,55 which demonstrated a positive effect of respite (day care). This study rigorously controlled the amount of respite used during the study to ensure adequate exposure in the intervention group and to ensure that both the intervention and the control group used only minimal amounts of other formal services. The intervention group had to use the service at least twice a week for a period of 3 months. However, being a quasi-experimental study there was the potential for selection bias, as the intervention group was recruited from one area of New Jersey whereas the control group was recruited in Ohio. Although care was taken to choose the sites based on similarities of the populations, the control subjects were recruited via advertisements and were self-selecting. Some differences were identified between the groups at baseline, although these were taken account of in the analysis strategy; although the difference between groups was not significant before controlling for these factors, it was after adjusting the means to take account of the differences. All attributes on the quality assessment were scored highly in this study.

The remaining two quasi-experimental studies44,48 were of low quality. One assessed home care and was the only study to show a positive effect of respite. 44 The intervention provided a care attendant to replace the informal carer duties for an agreed time up to a few hours; however, there was little further detail about the intervention or the number of hours that respite was used, and no indication of the extent of use of other services. Participants in the study were those referred to an Age Concern carer support scheme; the control group elected not to use services for a variety of unreported reasons, giving the potential for significant selection bias. There were inadequate details given in the study concerning participant selection or participant characteristics and so it was difficult to assess this further. No account was taken of any population characteristics in the analysis and it is possible that these may have seriously influenced the results.

The final quasi-experimental study48 assessed day care in the context of specialist dementia services. The comparison groups consisted of a memory clinic, a community mental health team and a day hospital. These interventions were intended to represent the care pathways for dementia patients with different levels of need. In the present analysis the day hospital was compared with the memory clinic, although people attending the memory clinic had mild dementia whereas those attending the day hospital had moderate to severe dementia. As well as having more severe memory problems the day hospital group also had a greater frequency of problem behaviours. The study scored poorly on a number of quality assessment criteria: description of care recipients, carers and the intervention; characteristics of loss to follow-up (likely to affect results); population characteristics (not controlled for); analysis methods (not reported but probably appropriate); intention to treat analysis (no attempt to carry out); and conclusions (some not supported by the data).

Three before-and-after studies assessed carer depression. 64,66,69 Only one64 reported a positive effect of respite. This study assessed a home-based care aide service for care recipients with dementia and was of a moderate quality rating. There was some flexibility in this service as carers decided whether they wanted frequent but short periods of help or less frequent but longer periods of help, so that they could tailor the respite to suit their needs. In addition, continuity was maintained by keeping, as far as possible, the same care aide/care recipient combination. The aide provided help with bathing, dressing, continence care, medication, shopping, preparing meals, household tasks and social activities in and outside the home. There was further flexibility in the programme as the aides themselves could define their role and carry out tasks other than personal care at their own discretion. Once a week the care aides and care recipients all met to provide mutual support, and the carers met monthly for information provision activities. The paper reports an average number of hours of care provided by the service as 15.8 per month, but it is unclear if this figure refers to the sample involved in the study or the average service provision before commencement of the study. Again, details of other service use are lacking. In addition, the quality assessment identified that incomplete carer details were provided and no account was taken of any sample characteristics in the analysis; that variance estimates were provided for some but not all outcomes; and that the selection criteria were not described. Losses to follow-up were also fairly substantial although these were addressed in the analysis. From these observations it was likely that there was some selection bias in the sampling.

The remaining two studies were of low quality. One, that by Theis et al. ,69 was discussed in relation to carer burden. This study had a substantial problem of attrition and the characteristics of both the carer and care recipient samples are unclear. Respite provision (home and institutional) at home was low (up to 4 hours a week) and the level of usage of either in-home or institutional respite is unclear. The final study66 assessed geriatric services providing inpatient respite care; however, the structure of and process involved in the intervention is not described. Limited details of the characteristics of the carer sample are given and no details of the characteristics of the care recipient sample are given. Outcome measures were carried out in the middle of the respite provision, which appeared to be in the middle of a 2-week inpatient stay. It is unlikely that there would be a significant impact on depression scores after this level of respite provision, although there is no information concerning the number of times respite had been utilised, for how long or whether any other services were being accessed. The quality assessment score also identified poor description of aims, study design and analysis.

Summary of studies assessing the impact of respite care on carer depression (studies included and excluded from the meta-analysis)

At short-term follow-up the meta-analysis of the impact of respite care on carer depression was in favour of respite care, although this just failed to reach significance. No effect of respite was seen in the longer term. The RCTs in these analyses were of moderate quality but none found an effect of respite. One was a study carried out in a Veterans’ facility in the US29 and the sample was poorly generalisable, and the other two40,47 had low levels of respite provision, which may be unlikely to have any impact. Only one study was of high quality,55 a quasi-experimental study, which demonstrated a positive effect of day care. The other quasi-experimental studies44,48 showed no significant effects and were of low quality with poor descriptions of methodology and potential sample bias.

Longitudinal before-and-after studies also failed to demonstrate any effects of respite care on carer depression in the meta-analysis. One study included in this analysis was of moderate quality,64 which on its own reported a positive effect of home care, whereas the other two studies were rated as low quality and failed to show any effects. 66,69 The study reporting a positive effect64 assessed a home aide service that incorporated some flexibility for carers to tailor the intervention to suit their needs. Of the two studies reporting no effect, one assessed both home and institutional care69 and the other inpatient respite. 66 These studies were poorly reported and, in one,69 attrition and sample bias were significant problems. The other had limited information from which to assess the reliability of the findings, although it is likely that outcomes were assessed after minimal respite provision. 66

In terms of respite setting there were no significant effects of either day care or home care, although they were in a direction favouring respite. Longer-term interventions seemed to have a greater impact on depression, although this effect was marginal.

Evidence from individual randomised and quasi-experimental studies not included in the meta-analysis largely indicates that respite does not have a significant impact on psychological well-being42,43,54 or depression49,52 when compared with normal care. However, these findings may reflect the type of intervention or the length of follow-up. In a study comparing respite with permanent institutional care, levels of depression were reduced amongst carers who institutionalised the care recipient but not amongst those receiving respite. However, removal of an existing respite service appears to result in increased levels of depression. 50

Two longitudinal before-and-after studies did, however, report a positive effect of respite care on carer depression. 57^,[62,63^] Significant effects were observed at relatively short follow-up periods (≤ 3 months).

Carer anxiety

Carer anxiety in randomised and quasi-experimental studies

Four randomised/quasi-experimental studies measured carer anxiety as an outcome. 29,40,47,48 One of these studies was carried out in the UK,48 the remaining three in the USA. Two focused on day care,29,47 one on frail elders29 and one on care recipients with dementia. 47 One assessed home care in care recipients with dementia40 and one both day and home care, again in care recipients with dementia. 48 No significant effects of respite on anxiety were observed at either short-term (1–6 months) or long-term (12 months) follow-up. Fixed-effects models for both short- and long-term follow-up are shown in Figures 14 and 15, respectively, and a random-effects model for long-term follow-up is shown in Figure 16.

FIGURE 14.

Carer anxiety in randomised controlled trials/quasi-experimental studies – short-term follow-up (1–6 months) (fixed model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

FIGURE 15.

Carer anxiety in randomised controlled trials/quasi-experimental studies – long-term follow-up (12 months) (fixed model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

FIGURE 16.

Carer anxiety in randomised controlled trials/quasi-experimental studies – long-term follow-up (12 months) (random model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Respite setting (day care29,47,48) did not significantly predict anxiety (separate analysis for day care only shown in Figures 17 and 18 for fixed and random effects respectively).

FIGURE 17.

Carer anxiety in randomised controlled trials/quasi-experimental studies – day care (fixed model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

FIGURE 18.

Carer anxiety in randomised controlled trials/quasi-experimental studies – day care (random model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Length of intervention (short, i.e. 2–8 weeks,40,47 or continuous29,48) also did not significantly predict carer anxiety (Figures 19–21).

FIGURE 19.

Carer anxiety in randomised controlled trials/quasi-experimental studies – short-term intervention (2–8 weeks) (fixed model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

FIGURE 20.

Carer anxiety in randomised controlled trials/quasi-experimental studies – long-term intervention (fixed model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

FIGURE 21.

Carer anxiety in randomised controlled trials/quasi-experimental studies – long-term intervention (random model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Quality and study design characteristics of studies included in the meta-analysis assessing the impact of respite care on carers’ anxiety

Three RCTs29,40,47 and one quasi-experimental study48 were included in this analysis. These were all described previously in relation to the analysis of carer depression. The three RCTs were of moderate quality. One47 provided early-stage day care for care recipients with mild to moderate dementia. Day care was provided for 4 hours a week for 8 weeks and involved a range of psychosocial activities. Although the day care group was less anxious at follow-up than a waiting list control group, this was not statistically significant. There were some problems of reporting in this study as carer characteristics and randomisation were poorly reported and it was unclear if allocation concealment had taken place at consent. Rothman et al. 29 also assessed day care but, as discussed previously, this was carried out in a veteran’s facility resulting in poor generalisability in the sampling. The final RCT assessed home care in dementia care recipients and provided in-home help for up to 6 hours a day for a period of 2 weeks. As mentioned previously, this represents a relatively short duration of respite provision, although it was provided fairly intensively in that time. Sample characteristics were poorly reported in this study and no consideration was given to potential sample biases in the analysis.

One quasi-experimental study48 was included, also described previously in relation to carer depression. This study was of low quality and reported a negative effect of day care on carer anxiety at 12 months. The study also found that the interpersonal dimension of the patient–carer dyad deteriorated, which may have some impact on carer anxiety. It is unclear why this occurred as the physical and self-care needs and the demanding and disturbing behaviours remained stable in this day care group.

Summary: evidence for the effectiveness of respite in reducing carer anxiety

The meta-analysis of randomised/quasi-experimental studies measuring carer anxiety as an outcome did not demonstrate any significant effects of respite. The RCTs included in the analysis were of moderate quality but either provided little respite or had problems with generalisability of the sample. The quasi-experimental study found a negative effect of respite on carer anxiety but this study was of low quality.

Subgroup analyses were carried out to assess home care, day care and long and short interventions but no significant effects were found. No longitudinal before-and-after studies measuring carer anxiety were identified.

A single quasi-experimental study measuring carer anxiety as an outcome was not included in the meta-analysis as comparison groups were institutional care versus respite. Results indicate that levels of anxiety were reduced for carers who institutionalised the care recipient but remained high for those receiving respite. 53

Carer morale

Carer morale in randomised and quasi-experimental studies

Two studies41,47 measured morale as an outcome in relation to respite care. Quayhagen et al. 47 assessed day care in care recipients with dementia, whereas the study by Kosloski and Montgomery41 assessed combined home and day care. Both studies were carried out in the USA with dementia care recipients. Length of follow-up was 3 and 6 months respectively. There was a positive effect of respite on morale, although this was not statistically significant (Figure 22).

FIGURE 22.

Carer morale in randomised controlled trials/quasi-experimental studies (fixed model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Quality and design characteristics of studies included in the meta-analysis assessing the effects of respite care on carer morale

One RCT47 rated as moderate quality and one quasi-experimental study41 rated as high quality assessed carer morale. Both have been discussed previously in relation to carer burden41 and carer depression47 and both examined care recipients with dementia. The RCT47 provided a low amount of respite (4 hours of day care per week for 8 weeks). There was poor definition of carer characteristics, randomisation methods and allocation concealment, and it was unclear whether there was any other use of respite services in either the intervention or the control groups.

The quasi-experimental study41 showed a trend towards a positive impact of respite care on carer morale although this failed to reach statistical significance. This study also reported a positive effect on carer burden. The intervention included both home and day care and provided some flexibility of service provision with the potential for high usage. Although there were differences between the intervention and control group characteristics at baseline, these were controlled for in the analysis. However, further differences may have been present, but not accounted for, because of the quasi-experimental nature of the study.

Summary of studies assessing the effects of respite on carer morale (included and excluded from the meta-analysis)

Only two studies measuring carer morale were identified, one of which was a randomised trial of moderate quality, the other a high-quality quasi-experimental comparison. Both interventions comprised day care delivered over a 3- to 6-month period, but the quasi-experimental study also included home care with a greater degree of flexibility in the programme. The RCT intervention provided a low amount of respite only. Results of the meta-analysis favoured the respite intervention but were marginally non-significant.

Carer anger/hostility

Carer anger in randomised and quasi-experimental studies

Two studies measured anger or hostility as an outcome following a respite intervention. 47,55 Length of follow-up was 3 months, and 3 and 12 months respectively. Follow-up data at 3 months were used for the meta-analysis. Day care comprised the respite setting in both studies (length of intervention was 8 weeks and 3 months respectively). Day care was found to significantly reduce levels of anger/hostility (Figure 23).

FIGURE 23.

Carer anger/hostility in randomised controlled trials/quasi-experimental studies (fixed model) (sample size in brackets). CI, confidence interval; ES, effect size; RCT, randomised controlled trial.

Quality and design characteristics of studies assessing the effects of respite care on carer anger and hostility

Of the two studies assessing anger and hostility, one was a moderate-quality RCT47 and one a high-quality quasi-experimental study. 55 Both have been discussed previously in relation to carer depression. The RCT was of moderate quality because of poor definition of carer characteristics, randomisation methods and allocation concealment. In addition, respite provision was low.

The quasi-experimental study55 rated well on all criteria in the quality assessment tool. It also controlled the amount of respite provision to ensure adequate exposure to the intervention and limited exposure to other types of respite. The control group was recruited via advertisement and so was self-selecting. As a result there were differences between the intervention and control groups. However, anger was significantly lower at 3 months post intervention in the intervention group (although not at 12 months) when adjusting for differences between groups at baseline.

Summary: evidence for the effectiveness of respite in reducing carer anger

Only two studies measuring carer anger/hostility were identified, one of which was a randomised trial of moderate quality, the other a high-quality quasi-experimental comparison. The randomised trial provided limited respite care whereas the quasi-experimental study controlled the amount of respite provision within the intervention and from other sources. Both interventions comprised day care and findings from the meta-analysis showed a significant positive impact on self-reports of carer anger/hostility towards the care recipient.

Quality of life

Carer quality of life in randomised and quasi-experimental studies

No trials or quasi-experimental studies measuring carer quality of life as an outcome were identified.

Carer quality of life in single-group longitudinal studies

Three single-group (longitudinal before-and-after) studies measuring quality of life post respite were included in a meta-analysis. 65,69,70 Two of the studies were carried out in the USA65,69 and one in Canada. 70 Two assessed day care65,70 and the other a combination of home care and institutional care. 69 Two69,70 focused on frail elders and one on dementia care recipients. 65

Two separate models were studied, to account for different follow-up periods: the first comprised two studies followed up at 6 months65,70 and a third at 12 months,69 and the second comprised two studies followed up at 12 months65,69 and a third at 6 months. 70 The result of these analyses are shown in Figures 24 and 25 respectively (random-effects and fixed-effects models respectively).

FIGURE 24.

Carer quality of life in longitudinal before-and-after studies (two studies at 6 months’ follow-up and one at 12 months’ follow-up) (sample sizes in brackets). CI, confidence interval; ES, effect size.

FIGURE 25.

Carer quality of life in longitudinal before-and-after studies (two studies at 12 months’ follow-up and one at 6 months’ follow-up) (sample sizes in brackets). CI, confidence interval; ES, effect size.

Quality and design characteristics of studies included in the meta-analysis assessing the effects of respite care on carer quality of life

Two of the studies were rated as low quality,69,70 both of which have been discussed previously in relation to carer burden. Both of these studies report a negative effect of respite care on quality of life (in relation to day care at 6 months70 and home and institutional care at 12 months69). In general, the study by Warren et al. 70 was poorly described with limited details of population selection and characteristics and so there was a possibility of selection bias. The other study had a significant problem with attrition, which probably did influence the findings.

The study rated as moderate quality65 measured outcomes at both 6 and 12 months. Unlike the other two studies, which focused on frail elders, this study provided a day care facility for care recipients with mild to moderate dementia. Day care was provided for 1 day per week from 9.30am to 2pm. An educational programme was also provided for carers as part of the package. There was no difference in carer quality of life between baseline and 6 months but quality of life was significantly worse at 12 months. However, the condition of the care recipient deteriorated significantly over the study period with both mini-mental state examination (MMSE) scores and behavioural ratings showing decline. In addition, although the patients themselves rated their own quality of life similarly at the end of the study as at baseline, the carers rated their care recipient’s quality of life as lower, showing that they also perceived there to be a decline in the care recipient’s status. No attempt was made to control for patient decline in the analysis. Use of other support services also increased over the study period, including housekeeping assistance and in-home help with care recipient care. Without an appropriate control group comparison it is difficult to assess whether the decline in quality of life is a result of the use of respite or of other factors.

Summary: evidence for the effectiveness of respite in improving carer quality of life

Only three studies measuring carer quality of life following respite care were identified, and all comprised longitudinal before-and-after evaluations. Two included day care65,70 whereas the third was a mixed respite service. 69 Respite interventions were delivered continuously until follow-up (6–12 months). Carer self-reports of quality of life were actually significantly lower following receipt of respite than before the interventions. The quality of the studies was low to moderate. Two of the studies were poorly described with the potential for selection bias, particularly in one study, which had significant attrition problems. The other study, of moderate quality, showed a significant decline in the condition of care recipients over the period of the study and failed to take account of this in the analysis. As these studies have no comparison groups and are generally of low quality they represent very weak evidence in support of a decline in quality of life after respite care.

Institutionalisation

Rates of institutionalisation were considered when this information was provided alongside carer outcomes.

Recipient institutionalisation in randomised and quasi-experimental studies

Three studies provided information on rates of recipient institutionalisation at follow-up, in addition to carer outcomes. 36,41,55 One study55 provided rates of institutionalisation at 3 and 12 months’ follow-up; separate meta-analyses were therefore carried out to determine the effect of respite on institutionalisation in both the long and short term. Results indicate that carers who had received respite were more likely to institutionalise care recipients at short-term follow-up (10 weeks,36 3 months;55 Figure 26). No care recipients died between baseline and follow-up in either of these studies.

FIGURE 26.

Recipient institutionalisation in quasi-experimental studies: short-term follow-up (10 weeks and 3 months) (fixed model) (sample sizes in brackets). CI, confidence interval; OR, odds ratio.

Rates of institutionalisation at longer-term follow-up (6 months,41 12 months55) were analysed compared with rates in those remaining in the community, including care recipients who had died as well as those institutionalised (Figure 27). It is evident from Figure 27 that death or institutionalisation was more likely in the intervention groups.

FIGURE 27.

Recipient institutionalisation and death in quasi-experimental studies: long-term follow-up (fixed model) (sample sizes in brackets). CI, confidence interval; OR, odds ratio.

Quality and design characteristics of studies included in the meta-analysis assessing institutionalisation after respite care

Two of the studies reporting institutionalisation were quasi-experimental studies of high quality,41,55 both of which have been discussed previously in relation to carer burden and depression. Overall, the analysis found that care recipients were more likely to be institutionalised in the longer term after using respite services, although one study reported that the intervention group were not more likely to be institutionalised after a respite programme involving both home and day care. 41 In this study the programme offered different interventions at different sites but tended to offer a fairly flexible schedule. A positive aspect of this study was that all participants did take up respite services and use of other services was accounted for. The main problem with this study was potential selection bias resulting from the quasi-experimental nature of the study design, particularly as the control group was recruited from one site only. However, known differences between the intervention and control groups were accounted for in the analyses.

The other study55 assessing institutionalisation in the longer term suffered from similar selection bias as, again, control subjects were recruited from one site, which was in a different geographical area to that of the intervention group; however, again, known differences were accounted for in the analysis. This study demonstrated an increased likelihood of institutionalisation after day care. Both this and the previous study recruited care recipients with dementia, and carers spent similar amounts of time per week caring for their relatives. Because of the long timescale of the study it was felt inappropriate to randomise participants to a control group and so control subjects were recruited from areas where day care provision was limited. One significant difference between the characteristics of the treatment and control groups was the relationship of the carers to the care recipients. A total of 55% of carers were adult children in the intervention group compared with 38% of the control group. A larger percentage of carers in the control group were spouses. In the study by Kosloski and Montgomery,41 discussed above, the majority of carers were spouses. It is possible that the relationship between carer and care recipient impacted on the decision to institutionalise. The Zarit et al. 55 study also provided data on institutionalisation at a shorter term of 3 months but found no greater likelihood of placing the care recipient in long-term care after using respite services, although overall the analysis found an increased likelihood of institutionalisation.

Another study assessing institutionalisation in the short term gave carers the choice of respite care, either in an assisted-living facility or at home. 36 Care was provided for 6–8 hours a day, 2 days a week for 10 weeks. On average, participants received 15.3 hours per week of respite. The quality rating of this study, however, was low. Participants were assigned to either intervention or control and sampling was stratified based on dementia severity. Randomisation was not mentioned, although it would have been feasible to randomly allocate participants to the two groups. However, there were only seven and eight participants in each of the groups and so it is likely that randomisation would not have avoided differences in the characteristics of the samples. Recruitment procedures were not clear and reporting of the characteristics of the carer and care recipient samples was incomplete. It was not clear if there were any differences between the groups and no account was taken of any differences in the analyses. There was incomplete description of the statistics and variance estimates were not reported for all results.

Summary: the impact of respite on care recipient institutionalisation

Three quasi-experimental studies were included in the meta-analysis of recipient institutionalisation. Results suggest that a negative outcome for the care recipient in terms of institutionalisation or death is more likely following a period of respite. Two of the studies were rated as being of high quality and one as being of low quality. No single-group longitudinal studies reporting rates of institutionalisation were identified. Individual studies varied in their findings, with one study having more adult children as carers reporting an increased likelihood of institutionalisation after respite. Carers in the high-quality study who showed no increase in the risk of institutionalisation after respite were mainly spouses.

Narrative evidence also suggests that institutionalisation may be more likely after a period of respite. One study reported that, compared with a carer training programme,33 carers in receipt of respite tended to institutionalise care recipients faster. In contrast, Riordan and Bennett49 found that respite users tended to keep the care recipient in the community for significantly longer than matched control subjects. In a study of preferences for community or institutional care,58 carers were less likely to express a wish to institutionalise the care recipient following respite than before receipt of the service.

Characteristics of longitudinal observational studies

A total of 19 papers reported a longitudinal observational design, representing 13 studies. Seven papers all referred to the Medicare Alzheimer’s Disease Demonstration Evaluation (MADDE) study, but each gave data on different samples taking part in the MADDE. 77–83 The majority of the research (nine studies) was carried out in the USA with one study carried out in the UK,71 one in Canada72 and two in Australia73,87 (Table 7). All except two examined mixed respite provision; the remaining two examined in-home care. Four studies focused on frail elderly with mixed disabilities and the remainder were specific to dementia care recipients. All studies assessed a range of formal service provision including respite and non-respite services, and some also included informal support. It was not possible, therefore, to categorise according to type of respite provision. Length of follow-up ranged from 3 months to 5+ years with the majority being between 1 and 2 years. Although many of the studies used multiple measures, there were generally only one or two related to service use in each analysis. The results of these studies are discussed according to the various outcomes measured.

Institutionalisation

A meta-analysis of institutionalisation in quasi-experimental studies was presented in the previous section. The analysis found institutionalisation and death to be more likely following respite use. This section summarises the evidence from observational longitudinal studies.

Six papers examined the relationship between formal service use and institutionalisation; however, the measure of service use included a range of personal care and support services of varying types and it was difficult to identify the impact of respite. 74,77,84–86,88 The methods and results of these studies were heterogeneous and it is difficult, therefore, to draw any firm conclusions because of the lack of definitions and the variation in study contexts. One further study87 reported on institutionalisation but did not have data linking respite and institutionalisation.

The papers that had a more distinct focus on respite care are summarised below. Seven referred to the same study (the MADDE study) although data were reported for different subsamples. 77–83

Depression

The evidence discussed so far in relation to carer depression was provided by randomised and quasi-randomised studies, representing the highest level of evidence, although the quality of individual studies within this group was generally moderate. Although not all analyses were statistically significant, the direction of the effect was in favour of respite care, with carers having reduced levels of depression. This section summarises the longitudinal observational data in relation to carer depression.

Three studies (all carried out in the USA) examined the relationship between formal help and carer depression. 75,76,83

Burden

Meta-analyses found weak evidence of a positive impact of respite care on carer burden, as significant effects were found only in longitudinal before-and-after studies but not in randomised and quasi-randomised trials. This section assesses the evidence from longitudinal observational studies of the impact of respite on carer burden.

Five studies72,75,76,83,89 looked at the relationship between formal support and measures of burden. In three of the studies there was no clear distinction between respite care and other forms of support service, whereas the study by Cox75 focused specifically on respite care with carers having in-home care, short stay institutional respite care or day care available to them.

Characteristics of observational cross-sectional studies

This section provides a narrative synthesis of the cross-sectional studies that were identified as being relevant to the review but not of an appropriate design to be included in a meta-analysis. Again, they are considered according to each particular outcome, but because of the larger number of studies in this group they are broken down further according to the type of service provision. Depression and anxiety are discussed together in this narrative synthesis as the measures used are frequently combination measures such as the Hospital Anxiety and Depression Scale, which consists of two subscales measuring anxiety and depression, as well as an overall measure of mental health.

A total of 45 papers90–134 with a cross-sectional study design were included in the review (see Appendix 7). These 45 papers represented 40 studies. In five of these studies service use was the outcome variable and the carer and care recipient variables were analysed as predictors of respite use in multivariate analyses. These have been excluded from the following narrative summary but may be found in the summary table in Appendix 7.

The remaining 35 studies are summarised in Table 8 by country of origin, condition of care recipient and outcomes assessed. The majority of papers were published since 1995 (n = 34) with 11 published before 1995. The main outcomes were mental health and burden. The following narrative review will focus on these two outcomes.

Mental health outcomes: observational cross-sectional studies

Depression and anxiety were included in the meta-analyses and depression was also discussed as an outcome in the longitudinal observational studies. There were positive effects of respite care on depression in the meta-analyses of randomised and quasi-randomised studies but there was little support within the longitudinal observational studies. Randomised studies showed no positive effects of respite care on anxiety and there was no further evidence from longitudinal before-and-after or longitudinal observational studies.

In the main this category focused on depression, although some measurement tools also assessed anxiety, for example the Hospital Anxiety and Depression Scale (HADS) (see Appendix 11 for references for measures used in studies). Table 9 shows the direction of association of studies assessing mental health in relation to service use and respite. The service contexts are categorised as being ADC, in-home care, respite (this could include a range of services that are specifically aimed at providing respite) and general service provision (which includes a range of community services, some of which provide respite). This last category may include personal care services, homemaker services or even information provision, legal advice or carer counselling. The studies categorised as providing general services do not usually define the amount and type of services that potentially provide respite.

All studies measuring mental health outcomes (n = 15) used a standard validated depression scale, the most common being the CES-D (used by eight studies).

Burden

Summary of cross-sectional studies

The main outcomes assessed in the cross-sectional studies related to mental health and carer burden. The majority of studies assessing carer mental health focused on general service provision giving a range of personal care and respite services, which are difficult to compare and for which it is difficult to identify the impact of respite. Those examining day care provided heterogeneous methods and results. Studies were small and frequently either did not take account of confounding factors or had limited or no information on levels of service provision. The two studies focusing specifically on respite included one examining informal care and the other examining different types of respite in different European countries, neither of which provide information that is suitable for establishing generalisable conclusions.

Studies measuring the impact of ADC on burden give varying results. A number make comparisons between different types of service provision and do not include a comparison group having no respite care. Studies that did make the appropriate comparisons and which showed no effect of day care were small and provided limited amounts of respite, which could account for the lack of effects. Two of the studies showing a positive effect were larger, although one suffered from possible confounding effects that were not taken into account in the analysis. Again, the extent and quality of the evidence are insufficient to come to any firm conclusions concerning the impact of day care on carer burden.

Studies focusing on the impact on burden of the use of services specifically designed to provide respite do not, however, clearly define the type of respite. Some focus on informal support and some include comparisons of different types of respite service provision or frame the availability of respite hypothetically. The samples used had the potential for considerable bias and, again, there is insufficient evidence to draw firm conclusions.

Studies of the impact of general service provision again suffered from difficulties of definition and the identification of the constituent elements of respite. Studies of the use of in-home support faced similar difficulties in that in-home services tended also to include personal care service provision and home-maker services, with no clear indication of the extent of respite that these services provided. There was also considerable variation in the methods used, making interpretation difficult.

Cross-sectional studies do not, therefore, provide consistent or clear evidence concerning the impact of respite care on carer burden or mental health to support any of the previous findings, although the majority of studies reported either a positive effect or no effect, with little support for negative effects of respite.

The quality ratings of the cross-sectional studies were mixed, with a fair proportion of studies having either high- or lower-quality ratings.

Randomised and quasi-randomised trials: characteristics of studies

Because the search strategy focused on carer outcomes it was not felt that it was appropriate to combine care recipient outcomes quantitatively. Therefore, studies reporting care recipient outcomes are summarised in the following section. Table 11 summarises the outcomes for both carer and care recipient within the randomised and quasi-experimental studies. Further details of the studies can be found in Appendix 4.

Eleven randomised or quasi-randomised studies included care recipient outcomes in the design. The majority were carried out in the UK (n = 4)34,48,49,50 and the USA (n = 3),^[29,51^],47,55 with two from Canada,32,35 one from New Zealand53 and one from Germany54 (Table 12). Six were carried out between 1990 and 2000, two between 1985 and 1989, and three between 2000 and 2002. Six of the 11 studies focused on dementia care recipients,35,47,48,49,53,55 three on frail elderly,^[29,51^],50,54 one on physical disability34 and one on mental health. 32

Outcomes included cognitive function, behaviour problems, functional ability, psychological distress and well-being, self-esteem, health perception, social support and institutionalisation. Five32,34,47,50,53 of the studies showed no effect of respite care, four an effect in favour of respite care35,49,54,55 and two an effect in favour of other interventions or the control group. ^[29,51^],48

Sample sizes were generally small across all of the studies with only three^[29,51^],32,55 having sample sizes greater than 50 per group. Follow-up was short term at around 3 months, although two assessed outcomes at 12 months and one at 9 months.

Care recipient outcomes: longitudinal before-and-after studies

Nine before-and-after studies (presented in 10 papers) included care recipient outcomes57–63,65,66,68 (see summary table of before-and-after studies in Appendix 5). Four of the studies were from the USA,57,59,60,65 two from the UK,^[62,63^],66 one from Canada,61 one from Australia68 and one from Hong Kong. 58 The majority of studies were carried out after 1990 (all except two). There was a fairly even split of studies focusing on dementia (n = 5) and the frail elderly (n = 4). The dementia studies were carried out mainly in the USA (n = 4) with one in the UK (see summary table in Appendix 5). Overall, four studies reported a positive effect of respite on the care recipient,59,61,66,68 four no effect57,60^[62,63^],65 and one a negative effect58 (Table 13).

Care recipient outcomes in relation to carer outcomes

The direction of the effects for carers and care recipients were in the same direction for the majority of randomised and quasi-experimental studies. Only two studies47,54 showed different effects for carers and care recipients. In the study by Quayhagen et al. 47 there was no effect of respite on the care recipient whereas the carer had a decrease in hostility. Respite, however, did not affect perceived stress. The study by Zank and Schacke54 measured a number of outcomes for care recipients, showing positive effects in both psychosocial and cognitive tests. Similarly, a number of outcomes were assessed for the carer but no significant effects were found in any of the burden or mental health outcomes.

There was greater disagreement in outcomes for carers and care recipients in the longitudinal before-and-after studies. In two studies there was no effect of respite on care recipient outcomes and a positive effect on carer outcomes;60^,[62,63^] in one study61 there was a positive effect for care recipients but no effect for carers.

It would appear from these comparisons that respite, although often designed to be an intervention aimed at carers, does not necessarily provide benefit to carers to the detriment of care recipients. The consistency of the effects on carers and care recipients suggests that the design of the study and the intervention is likely to be responsible for the types of effects seen, rather than there being differences in outcomes for the two groups of participants.

Longitudinal observational studies: care recipient outcomes

Although a number of the studies measured care recipient characteristics such as severity of mental impairment and functional abilities they generally were not related in any meaningful way to service supports. Only the following studies provided relevant data.

Literature review

Review of the literature to facilitate the economic analysis was incorporated into the search strategy previously identified above. Abstracts yielded from the search that included any economic component were passed to the economists on the research team for further scrutiny and consideration for inclusion or exclusion. A total of 125 papers were identified making reference to economic issues pertaining to respite care. In addition, reference lists of papers were reviewed to identify any further potentially relevant papers to consider for inclusion. This generated a further 30 papers. Relevant web pages (e.g. PSSRU, HTA, Department of Health and health economics research centres) were also reviewed, yielding a further 17 relevant references, one a draft report by a team at the University of York investigating a very similar topic. 148 In total, therefore, 155 peer-reviewed publications, 16 public documents and one draft report were considered for inclusion.

Initial review classified identified papers into two groups: those reporting both costs and consequences and those reporting either costs or consequences but not both. To be included in the economic review studies had to have either a controlled or a matched design, compare at least two options and measure and report both costs and consequences.

Economic evaluation methods

The existence of scarcity and thus opportunity cost considerations underpins the need to undertake economic evaluations. Economic evaluation methods are applied to evaluate the relative efficiency of different methods of providing health-care services, assessing the relationship between resource use, processes and outcomes. 149 The basic task of economic evaluation is to identify, measure and compare the costs and consequences of the alternatives being considered. 149

Three main economic evaluation methodologies exist: (1) cost-effectiveness analysis (CEA) incorporating cost–minimisation analysis and cost–consequences analysis; (2) cost–utility analysis (CUA); and (3) cost–benefit analysis (CBA). All three methodologies adopt a common approach in that they all seek to compare the costs and consequences of health-care interventions. Further, they all deal with costs in very similar ways. The primary difference between them is the differing, although sometimes subtle, ways in which they measure and value consequences. Table 14 presents the three main techniques of economic evaluation and summarises the key differences in their approaches to identifying, measuring and valuing consequences. 149

There has been considerable debate among health economists in recent years regarding encouraging ‘good practice’ in the conduct of economic evaluations within health-care settings. Benchmark standards have been identified in the UK, which recognise the key considerations in the design conduct and critique of economic evaluations. 149 These key considerations were incorporated into the abstraction instrument for synthesis and appraisal of the economic evaluations included in this study.

Country of origin

A total of 70 papers96,103,157–222,235,236 were identified for inclusion in the qualitative review, representing 69 studies. The majority of studies were carried out in the UK and USA (25 and 22 studies respectively). Of the remaining studies, ten were from Canada, three from Australia, two from New Zealand, two from Japan, one from Iceland and four from Sweden.

Note that in the following discussion when the number of studies is referred to these will total 69, whereas when referring to the number of papers or articles the total will be 70.

Condition of the care recipient

Just over half of the studies (n = 38) focused on care recipients who had dementia (see Appendix 9). The other 31 studies focused on frail elders who had a mixed aetiology of both physical and cognitive impairments; those with specific physical disabilities such as stroke; respite care in a palliative care context; or frail older groups for whom it is unclear whether their disabilities were physical or cognitive or both. For this last group of studies it is likely that at least some of the care recipients will experience some combination of both cognitive and physical disability.

Year of publication

The majority of the articles were published relatively recently, 42 being published since 2000 and only six published before 1995. This pattern was consistent across the countries of origin, with only the UK and USA having published studies before 1995.

Sampling and data collection methods

The preferred method of data collection was by individual interview (46 studies). Eleven studies used focus group methodology and five a case study approach. Seven studies used a qualitative approach within a structured survey by reporting on responses to open-ended questions. Two of these were included in the previous quantitative synthesis. 96,103

There was limited information concerning data collection beyond the type of interview and whether it was audio recorded, although some studies did outline the questions posed and the length and location of interviews. Few detailed the approach and recruitment of participants. There was also little justification of data collection methods in relation to an overall qualitative framework. A total of 15 studies did not describe the sampling strategy, four used a random sample, 15 mentioned either purposive or theoretical sampling and 35 used a convenience sample. These last 35 papers included studies in which all users of a particular service were recruited consecutively and studies using volunteers or participants identified by service providers. Those studies using purposive sampling were very variable in the level of detail concerning the criteria used in the sampling and the justification for the criteria.

Theoretical framework and analysis

The use of a theoretical framework to guide data collection and analysis was poorly reported and/or infrequently used. The summary table in Appendix 9 lists the types of framework reported in the articles, although not all represent a theoretical underpinning to qualitative methodology but either a theoretical perspective (such as feminism) or an approach to analysis (such as constant comparative technique). In total, 46 papers did not report any theoretical framework at all; seven reported using a grounded theory approach, five phenomenology and one critical discourse analysis. Two studies claimed to use an interpretivist approach but did not give further details as to which type of interpretivist approach they were adopting. One study claimed to be using content analysis but the analysis was actually thematic. Others (n = 3) reported using a combination of content and thematic analysis although, again, these took mainly a thematic approach. One study proposed a feminist perspective but did not go further to describe the methodological framework. The majority of studies were thematically analysed allowing further integration and thematic synthesis within the present review.

Focus of the study

The papers were categorised according to their main aims, as proposed by the authors. These fell broadly into five categories: views of respite care (n = 28), views of wider community services (n = 14), experiences of caring (n = 6), carer needs (n = 13) and evaluation of specific respite interventions (n = 9). Studies exploring the experience of caring provided information related to carer needs and, to a lesser extent, barriers to uptake of services. Studies examining carer views of community services gave some specific mention of respite services, including views of service use or expressions of need or preferences for types of service provision. The studies explicitly focusing on respite care gave similar information. The studies designed to form an evaluation or part of an evaluation of a specific programme focused on the following.

• Provision of day care (compared with domiciliary stroke team) for stroke survivors; this was an addition to an RCT. 190 Also a pilot study evaluating an ADC programme103 in which open questions were included in the quantitative survey.

• Evaluations of multiservice respite programmes,96,165 in one of which participants were given a choice of a range of services to respond to: unexpected circumstances, routine periodic relief or for special purposes such as short family vacations or outings. 165

  • Two evaluations of case management interventions providing respite relief. 180,192

  • Evaluation of a weekend respite programme geared to individual needs and preferences. 200

  • An action research study to develop and evaluate a palliative care night respite service. 183

  • Evaluation of a continuous intermittent care programme. 163

  • Evaluation of a hospital night respite service. 214

Reporting of the context of the research

Description of the context of service provision within the studies was generally poor. Participants were frequently in receipt of a variety of services but these were often not described. However, as this was not always the focus of a study this is not surprising and does not detract from the quality of the study. But many studies focusing specifically on respite care and those in which participants were recruited by particular respite services failed to describe the characteristics of the services in detail, or the extent of use of the service (duration and/or frequency). Many of the studies discussed ‘formal service use’ or ‘community service provision’ covering a wide variety of home care services designed to support either the carer (e.g. respite) or the care recipient (e.g. personal care). This reflects the difficulty in defining respite care in relation to specific services, as carers often talked about the respite achieved by the input of home health aides, whose main purpose was to provide personal care to the care recipient. Many of the studies were interested in any service likely to provide support to both carer and care recipient.

There was also great variability in the range of participant characteristics reported. The unit of interest in most of these studies in relation to the present review is the carer/care recipient dyad, but authors frequently did not report the characteristics of both parties. As the carers were generally the ones taking part in the interviews their characteristics had greater emphasis. Around half of the studies were designed to explore issues around dementia care, with others aimed at palliative care or physical disabilities such as stroke. For many of the remainder it was often not apparent what disabilities were experienced by the care recipients. Cognitive impairments present particular difficulties in relation to informal care as well as formal services, and defining these characteristics would enable the reader to assess generalisability of the findings.

Overall quality rating

Despite the reservations listed above the majority of the studies were of moderate quality. The scores on the quality ratings were categorised as low, medium or high based on a division of scores into tertiles. This is in line with the method used in the quantitative review, and gives a relative quality rating rather than an absolute quality. The total possible score for any individual study was 30; scores of 8–13 were categorised as low, scores of 14–18 as moderate and scores over 18 as high. Many of the papers scored in the mid-range for the individual items in the scoring system (each item scored from 0–2 – see Appendix 2 for quality assessment instrument). This was because the reports generally gave some information concerning the criteria being assessed, for example context, sampling, data collection and data analysis, whilst not giving full details, thus scoring 1 on individual items. For example, those studies using a purposive or theoretical sampling procedure would state this, while frequently giving little information concerning the basis of the purposive sampling. Some assumptions were then made as to whether the sample included the full range of possible cases based on the characteristics of the sample. Included in the studies using convenience sampling were those that had recruited all participants taking part in a particular service, which, if justified, would receive higher scores than those studies using purposive sampling that was not fully explained.

As few studies reported the use of an underlying methodological framework it was difficult to relate the methods used to a particular typology of qualitative research, and so a general thematic approach was assumed and the requirements for sampling, data collection and analysis, as set out in the quality assessment, were expected. It was also necessary to judge each study on its own terms and assess the reporting in relation to the stated aims. As mentioned previously, the provision of respite was perhaps a subsidiary concern and so detail concerning the types of service received may not be expected to be given in the context of a particular study. This means that little information is available concerning the context of the research for the purposes of the review but does not necessarily mean that a particular study will be assessed as being of poor quality.

When a more quantitative approach was taken and open questions were included in a survey, the quality assessment was not scored on all items. In such cases the scores given were prorated so that they could be compared with the quality assessments of the other studies. Ratings were very dependent on how the authors reported the study; for example, two papers207,208 reporting on the same study scored 19 and 22, respectively, based on differences in reporting. However, with scores of more than 18 both of these papers fall within the category of high quality. It is for this reason that a categorisation approach of low, medium and high was taken rather than giving undue weight to a continuous score on a measure that is not fully validated. Table 16 shows the quality ratings of the qualitative papers by country. Studies carried out in the UK tended to be of lower quality than studies carried out elsewhere. The majority of studies carried out in the UK (60%) were categorised as low quality. Studies carried out in the USA were more evenly distributed across the categories, whereas those carried out in Canada were of generally higher quality. The lower quality studies in the UK tended to be earlier studies, eight out of the 15 having been carried out before 2000, five between 2000 and 2003 and only two after 2003.

Barriers to uptake of respite care

Figure 28 shows a proposed model of the barriers to uptake of respite care emerging from the review of the qualitative studies. It identifies aspects of the carer and the care recipient and the characteristics of service provision that may influence uptake of respite. Personal and cultural attitudes to the caring role and to the function and purpose of respite services may impact on the carers’ perceptions of their needs for respite. The nature of the carer/care recipient relationship may contribute to attitudes to the caring role and the extent and nature of perceived responsibility for care. Guilt is a significant emotion in the uptake and use of respite care and is influenced by the caregiving relationship as well as the actual or anticipated reactions of the care recipient to participation in respite.

FIGURE 28.

Barriers to uptake of respite services.

Once a need for respite is acknowledged, services must be available and the carer must be aware of them for respite to be utilised. A host of other issues more directly related to service provision, such as the appropriateness of service provision and the quality of care, may then be potential barriers to the uptake of services. A change in any one item shown in the model could potentially affect other relationships and the eventual outcome. It must be noted that this is not necessarily a comprehensive model but portrays the findings of the literature to date.

Each of the potential barriers is discussed below.

Reasons for uptake of services

A commonly reported reason for the use of respite services is carer fatigue. However, this was often reported as a last resort, when fatigue had become exhaustion and the carer could no longer cope and was in crisis:

many families delayed using respite until they really needed it, first extending themselves to the point of mental and physical exhaustion.

Brody et al. 165

When formal help had been sought it was usually only when the carer was no longer able to cope, sometimes in response to a crisis or when all other possibilities had been exhausted.

McGarry and Arthur193

Occasionally carers were given the impression by health-care professionals that help would only be available once they had reached the ‘limits of their resilience’:

The doctor told me some of [the diagnosis] and said if you can’t cope with it any longer we will get her into a day clinic and that will give you a day’s rest.

Gillies170

Crisis events could be health events of either the carer or the care recipient, but often the services instituted to deal with sudden health events were continued afterwards if they had proved acceptable:

These caregivers sustained a number of health problems that prevented them from caring for the recipient, including cardiovascular events, cancer, and orthopaedic problems. In some cases, the services that were initiated during their hospitalisation or convalescence were retained.

One male caregiver who sustained a mild stroke experienced a particularly good fit with the sitter acquired by a social worker during his illness. Although he regained his pre-morbid level of functioning, he described the sitter’s on-going role as a companion to his wife and a friend to him, an integral part of the family.

Cotrell167

The experience expressed in the last quote was common not only of respite used during a crisis but also for other situations of one-off pragmatic use that break down the barriers to acceptance of service use. The need for respite for particular social events or occasions such as conferences, weddings and holidays triggered the use of respite, and once the benefit was experienced and proved successful it was used again on a more regular basis. 167 However, it could be a long process of acceptance of the need for relief, and one study165 reported on a counselling intervention that helped carers to identify and accept a need for leisure time and helped them to work out problems preceding the use of respite such as finances, patient management and conflicted relationships. The tendency to delay take-up of respite was sometimes related to cost, which prevented carers using respite until absolutely necessary or prevented the optimum use of services. 165

As well as the need to promote physical well-being, respite was also used to address emotional needs. Some studies reported on carers who recognised their need to distance themselves emotionally to maintain control:

They sought respite to place some emotional distance between them and the care situation because thy perceived that they were becoming irritable and ineffective in their care.

Cotrell167

This was often in response to behaviour problems. Deterioration in the care recipient’s condition was reported as a trigger to the instigation of respite care and was often a precursor to institutionalisation:

Two months later, the situation had deteriorated. A home visit revealed that the mother was incontinent of both bowel and bladder. Mrs J. was overwhelmed, exhausted and depressed. She could not keep up with the house and her mother. In-home respite was immediately arranged and the worker began to discuss with Mrs J. the possibility of institutionalizing.

Brody et al. 165

Apart from the needs of carers there were a number of care recipient issues identified as the focus of the uptake of services. As the care recipient’s condition deteriorates carers express concern for safety and the need for a sitter may be expressed. There are also concerns about maintaining care recipient health in relation to maintaining adequate nutrition. As well as consideration of care recipient physical health, respite was also seen as a means of providing socialisation and stimulation for the care recipient and of facilitating emotional health. Day care was often viewed as an activity centre for elders that could provide socialisation. In-home care, however, also provided an opportunity for stimulation:

she realised that the sitter could provide her husband with the stimulation and supervision he needed and which she herself could not physically provide.

Cotrell167

Positive aspects of respite care

Although some carers had negative experiences with respite care, or had low expectations or fears about the outcome, which acted as barriers to the take-up of services, there were also many reports concerning the positive aspects of respite. These were often the ‘other side of the coin’ to the barriers reported above and illustrate how a positive respite experience could have real benefits for carers and their care recipients.

Shorter-term respite such as day care was felt to give structure to the carer’s week157 along with a sense of normality as the free time matched the ebb and flow of caregiving activities. The respite time gained through this type of care tended to be used for necessary everyday chores:

Analysis highlighted how respite was not a time of relaxation rather an opportunity for cleaning, cooking, shopping, gardening, ironing, etc. and attending to his/her own health/social needs, i.e. the time and space were reinvested into sustaining the overall coping experience.

Upton and Reed211

This focus on catching up with chores was echoed by a number of studies in the review. 161,175,194,199 However, as the above quote suggests, respite was also used to carry out pleasurable activities on a regular basis such as playing golf,205 personal shopping, bowling and sewing,208 participating in religious or diversional activities,161 going to the gym, going for walks and socialising with friends157,171,199 and provided time to devote to the needs of the carer’s own family. 212 Day care also allowed some carers to remain in employment, either paid or voluntary. 175

Longer periods of respite such as institutional care were most associated with ‘recuperating and restoring of batteries’. 207 This type of care, often consisting of around 2 weeks of inpatient care for the care recipient, allowed the carer to take a vacation or visit relatives. Some preferred to stay at home to have uninterrupted time to relax. 161,195 This type of respite was also used for health-care needs such as surgery or emergency hospitalisation. 161

A recurring theme throughout the studies in the review was the essential nature of respite for maintaining and continuing the caring role:

Most of the caregivers perceived that respite enabled them to endure in the caring for their relatives instead of becoming fatigued and burned out.

Piercy and Dunkley202

Several studies reported respite to be instrumental in maintaining carer physical and mental health157,161,186,199 with carers in these studies quoted as saying that without respite they would have had a nervous breakdown. The result of such breakdowns in care would be institutionalisation of the care recipient. 157 These positive effects on mental and physical health were achieved by improved sleep,175 physical relief and the chance to rest and relax,161 the release of the build-up of stress and tension209 and rejuvenation from a sense of freedom. 161 The experience of positive respite was dependent on many of the issues already discussed, such as quality of care and the well-being of the care recipient, which allowed freedom from worry and confidence in the respite event.

There was some disagreement as to whether the positive aspects of respite were framed as being for the benefit of the carer158 or the benefit of the care recipient,170 although the two were frequently viewed as interdependent. One of the main benefits for the care recipient was perceived as the opportunity for socialisation,194 which was felt to play an important role in health improvements, particularly in relation to mental health and well-being:

Overwhelmingly, carers considered socialization the predominant factor for the improvement in their elderly relative’s mental health and overall outlook on life.

Valadez et al. 212

Other physical health improvements were attributed to consistent diet, exercise programmes, daily activities and mental stimulation exercises. 212 The positive effects of stimulation of care recipients from appropriate activities undertaken at respite was a consistent finding in the review:

Some participants stressed the way that services tried to motivate and maintain mental, physical and social awareness through using games and activities.

Innes et al. 175

The process criteria that caregivers identified (for evaluating care) were opportunities for socializing, meaningful interaction, and sensory stimulation.

Perry and Bontinen200

Socialisation and stimulation had benefits for the relationship between carer and care recipient. It enhanced communication by providing topics of conversation as both carer and care recipient felt they had more individual experiences to talk about, and resulted in an enhanced mood in both parties on return from respite. But the relief of strain of the carer also had positive impacts on the relationship as carers felt more tolerant after the respite break. Some carers reported that they did not lose their temper as much after respite. 205 This has obvious implications for the prevention of elder abuse although none of the studies addressed this issue directly.

The relationship between the service provider and both the care recipient and the carer played an important role in providing support:

Several caregivers cited the supportive relationships, even friendships they developed with their relative’s providers, as helpful to their caregiving performance.

Piercy and Dunkley202

This support could result from the passing on of expert knowledge, empathy for the carer’s situation and social support. It was important for carers that the care recipient got on well with the service provider, particularly if care was in-home. This provided the care recipient with the social stimulation gained at day care by other care recipients. It also enhanced the co-operation of the care recipient, which in turn made the carer’s life much easier. It also gave the carer peace of mind when leaving the care recipient with the service provider. This was a consistent theme running through all of the studies, which:

supports the idea of a relationship between the caregiver’s ability to experience relief and the caregiver’s perception that his or her relative is comfortable and safe.

Perry and Bontinen200

As seen in the barriers to the uptake of services the carer’s perception of the quality of care and the acceptability of respite to the care recipient was extremely important and when these aspects were favourable it contributed to a very positive experience for both carer and care recipient.

Ethnicity

Few studies addressed issues of ethnic minority carers. Notably, Bowes and Wilkinson164 carried out case studies of Indian and Pakistani carers; Wykle and Segal222 interviewed 20 white and 20 black carers of people with dementia; Netto198 interviewed 15 Chinese, 15 Pakistani, eight Indian, three Bangladeshi, two Afro-Caribbean and two other Asian ethnicities. Several other studies had small numbers of ethnic minority groups in their samples. On the whole, general issues related to respite care were very similar to those reported in the white populations and were consistent with the model of uptake of services shown in Figure 28. Some of the detail, however, was specific to ethnic minorities.

There was a stronger cultural obligation to care for older relatives amongst ethnic minorities. The study by Bowes and Wilkinson164 portrayed the prevalent view of the extended family in South Asian populations that was supportive of the caring role. In some cases the wider family network took a share in care with the older person staying with different relatives in turn. Netto,198 on the other hand, proposed that the view of ethnic carers as having all the support that they require from within the extended family is a myth. More than one-third of participants in this study received no support from family members:

one of the important findings of this study is that the proportion of sole carers in these communities is, in fact, larger than the proportion of sole carers in the general population, which the GHS recorded as less than a quarter (23 per cent). Dispelling the myth of the supportive family network further and underlining the need for respite services, nearly half (46.7 per cent; 21/45) reported that they had no time off from caring.

Netto198

There was a greater tendency for a lack of relief from the caring role as there was a cultural obligation for older relatives to be co-resident with their adult children compared with white populations, in which many care recipients were not co-resident. As in the white population the majority of carers were female, but the average age of carers tended to be younger, with more carers having the added responsibility of childcare. Burden, therefore, is considered high. However, few used any form of respite, although there were indications in all of the studies of an expressed need by ethnic minority carers for respite. 164,198,222 Not only was there evidence of limited access to respite services but also there was a low awareness of the availability of services. This was coupled with poor knowledge of conditions such as dementia. There is a general issue of information provision to communities who may have language difficulties and whose carers are isolated within the home in a caring role and who do not have access to normal channels of information.

When questioned about the types of service need there were similar requirements as those expressed in the studies of the white population, with particular importance being placed on the appropriateness of care. In this context this referred to ensuring that carers were of the same ethnic group, spoke the same language and were preferably of the same gender as the care recipient. Other important considerations for carers were that food was appropriate for their religion, for example vegetarian, and that the service was sensitive to other cultural and religious differences. Incontinence was a particularly problematic issue as this created a ‘dirty’ environment, preventing the rest of the family from prayer at home; without respite for the care recipient family members would be unable to access other facilities such as the mosque.

Carer needs

Palliative care

Five studies177,183,189,204,209 focused on respite care in a palliative care context. Although many of the issues identified in these studies were similar to those already discussed there were some differences. These studies have been included in the thematic analysis already described; however, the differences are discussed here to include the more general findings in the overall synthesis. One of the main differences is in the approach to caring, which has some impact on the attitudes to respite. Because of the expectation of the impending death of the care recipient the caring experience was more intense. Carers in this situation wanted to spend as much time as possible with the care recipient. This highlighted to a greater extent the distinction between a physical and a mental break. Mental breaks were sought in closer proximity to the care recipient as physical separation could cause anxiety. 209 This is probably very much dependent on how imminent death is seen to be, as the study by Skilbeck et al. 204 did not report these types of views. In this study respite was provided in a hospice for 2 weeks twice a year, which suggests that death was not expected within weeks or months as in the Strang et al. study. 209 Some of the patients in the Skilbeck et al. 204 study had suffered from neurological conditions for many years and so the caring trajectory would be more similar to the long-term caring situations found in most of the studies included in the review.

The mental breaks reported by Strang et al. 209 were achieved by taking part in absorbing activities such as reading a book or watching television. A mental break was also couched in terms of regaining the previous relationship rather than being in the role of carer/care recipient, which helped the carer to forget the gravity of the situation and the impending bereavement. Physical respite was important, however, to allow the carer to catch up on sleep or to carry out chores such as grocery shopping, echoing the needs of carers in general. What was, perhaps, more important to the carers as a form of respite was the sharing of responsibility with both formal and informal respite providers.

Being able to take a break wasn’t as important as being able to have someone share in the responsibility or ease the responsibility.

Strang et al. 209

Related to this and in common with the studies in general was the feeling that respite could only be achieved through peace of mind, achieved when the care recipient was being well looked after. In the palliative care context this required more medical input and well-trained staff as the respite providers were required to have greater specialist knowledge. So, for example, in the study by Jarrett et al. 177 carers received support from Marie Curie nurses but considered their input purely as provision of respite because the specialist training of the nurses gave carers sufficient peace of mind to experience respite. In fact, palliative respite care was viewed as an opportunity to access specialist information and therapies such as complementary therapies.

Views of the service mirrored those already identified. For example, continuity of care was an issue when large numbers of nurses were entering the home. Continuity between home and hospital was also commented on, although in this context it was commented on more positively as the hospice nurses visited the care recipient at home to carry out treatment. However, there was also dissatisfaction with failure to maintain routines, as one carer reported that her father had become incontinent at night since admission to the hospice and another that their relative had not been mobilised sufficiently.

Service users felt considerable benefits of respite. It gave them the opportunity to socialise and talk to people in similar circumstances about issues that they felt their carers could not handle. It also increased self-esteem through being taken at face value rather than as a cancer diagnosis.

The impact of location of study, condition of care recipient and methodological quality rating of study on the findings

The country of origin of the study was one of the criteria used for sampling in the second-stage synthesis of qualitative data; countries were chosen that were culturally similar to the UK and which had service delivery organisation with similarities to that of the UK. As many of the issues raised as barriers to the uptake of respite services were culturally determined issues, such as the relationship between carer and care recipient and the nature of the extended family, relationships with formal carers, the practicalities of dealing with physical and mental disability, attitudes to caring and attitudes to respite, the findings were consistent across locations. The relationship with services is one area in which differences may arise because of the different nature of health and social care provision. However, in sampling these particular countries it was felt that this would not have substantial implications; for example, the existence of Medicare in the USA results in a similar system of services for frail elders as in the UK. However, there is still the potential for differences to exist in relation to the quality of service provision and the interaction between service providers and informal carers and care recipients. The findings, however, showed that the views expressed of service provision were similar across these different contexts. For example, there were similar concerns about continuity in both home care and institutional care, reliability of home care staff, training of staff (particularly in relation to dementia care), staff to patient ratios in nursing homes, the sensitivity of staff and the appropriateness of programming in day care facilities. A major issue for all participants in studies in all countries was the provision of flexible services that are appropriate to carer and care recipient needs.

Underprovision of respite care was a prominent theme across the board with little choice or control available, within generally limited resources. The level of resources and the cost implications were mentioned more frequently in studies outside the UK. There are probably greater limitations in what is covered by Medicare services in the USA and there was more discussion in this country of the difficulties of finding appropriate services even when the carer is willing to pay. This is, however, also an issue in the UK because if carers need to top up limited service provision for particular needs they must seek this support in the private sector. UK studies tended to talk about financial aspects in relation to the needs of carers for information concerning entitlements to allowances, such as attendance allowance, and for financial support generally. Care of the elderly in the UK was viewed as a responsibility of the state and carers viewed their role as deserving acknowledgement of the contribution that they make by receiving better financial support to carry out this role. Many carers lose out financially by giving up employment or reducing employment to undertake their caring responsibilities. However, apart from issues such as costs, the consistency across studies supports the approach of combining the data from studies carried out in different countries to aid understanding of the barriers to uptake of respite services.

Some specific issues have been identified in relation to the characteristics of the care recipients. Palliative care represents a particular aspect of the provision of respite care and this has been considered separately, as have the needs of ethnic minority carers. The remaining major defining characteristic of the care recipient is that of cognitive impairment as opposed to physical impairment. A number of studies have focused specifically on people with dementia or on particular physical conditions such as stroke survivors. Others have included mixed groups of both cognitively and physically disabled care recipients. Other studies have been more vague in their sampling and have included ‘frail elders’ who are likely to comprise care recipients with some degree of both physical and cognitive disability. Many issues are similar in these groups but when there are particular concerns these have been pointed out. For example, continuity of care has been held to be particularly important for cognitively impaired care recipients as constant changes of staff are more difficult for those care recipients who are confused. Similarly, it was pointed out that other changes, such as a change of environment, also add to confusion. Consequently, many carers of cognitively impaired elders prefer in-home respite, although this does not hold for all cases. Training of staff was found to be particularly deficient in relation to dementia, although a perceived lack of training in the care of bed-bound and physically frail patients was also felt to be a barrier to the take-up of services. The difficult behaviour of demented care recipients also caused particular problems for accessing services as it was reported that some services excluded care recipients when they became more severely impaired.

The majority of studies used individual interviews, with fewer studies using focus groups. Although the concepts identified were similar, one-to-one interviews tended to address more sensitive and personal issues whereas focus groups identified more general issues such as needs. There were several studies that used open questions in more structured approaches. Again, the concepts identified were consistent with other approaches but the data tended to be broader rather than focusing in-depth on particular issues. This was advantageous in some respects as these studies tended to access both positive and negative views of respite. However, the majority of studies using open questions addressed views of a particular service. Although addressing a very narrow context they provide useful information about practical needs related to service delivery, highlighting practical issues that may make respite difficult for carers to use.

No systematic differences were found in the issues raised in studies having different quality ratings. The findings of lower-quality studies were consistent with those of higher-quality studies and there was also consistency between those that did and those that did not focus directly on respite care issues.

Implications for care

The evidence suggests that the provision of respite care is, under certain circumstances, of benefit to carers’ well-being. However, the evidence did not identify the relative merits of all of the different types of respite provision, although home care emerged as a preference in a number of qualitative studies and also had some benefits over day care in relation to carer depression. Preferences were dependent on a number of factors and perhaps more important than any particular type of respite provision was a need for flexible respite services that are responsive to changing carer needs, which might include a range of different services. A number of studies in the review provided a range of services such as day care and home care, day care and institutional care, as well as a range of other services likely to provide some respite such as domestic help and personal care services. It would be appropriate for service providers to carry out assessments of needs and preferences before developing services.

The outcomes that arose as indicators of the positive impact of respite on carer well-being were depression, burden, morale, anger and hostility. Anxiety was unaffected by respite provision and quality of life was worse after respite care. Rates of institutionalisation were higher after respite provision but this is likely to be a result of the characteristics of those accessing respite rather than an outcome of respite per se.

There was some evidence to suggest that longer interventions had greater benefits for carers.

The review did not find any evidence of negative effects of respite on care recipients or any consistent evidence for positive effects, although outcome measurement for care recipients is inadequately addressed and requires further definition.

Further issues of prominence arising mainly in the qualitative review that have implications for care provision are as follows.

1. The need for information. This is a need that came across in many of the qualitative studies and was identified as a barrier to the uptake of respite. Without the knowledge of the availability of services and how to access them, carers cannot make informed choices of how best to carry out their caregiving role. It is an issue for care provision of how to identify and inform carers of the availability of services.

2. Provision of respite early in the caregiving career. The review pointed to the potential for respite only to be offered or accessed at a late stage in the caregiving career, as a result of a crisis situation or immediately before institutionalisation. This relates to point 1 above, in that information provided earlier about respite availability may delay or avoid crisis situations and maintain carer well-being for longer. To achieve this the relevant health professionals need to be knowledgeable about the available options to support and inform carers appropriately.

3. Training of formal carers. Quality of care was an important issue for carers and a number of papers in the qualitative review highlighted poor skills of care staff, particularly in relation to dementia care. Carers expressed a wish for a sensitive approach as opposed to demeaning attitudes sometimes held by care staff towards care recipients. Training and monitoring of staff is an issue for consideration by service providers.

4. Continuity of care. Continuity of care was an important issue in avoiding the disorientation of care recipients but also in avoiding unnecessary stress for carers who may be in a position of repeatedly having to inform new care staff.

5. Benefits for care recipients. Carers placed importance on the benefits of respite care for care recipients, which required good-quality care with appropriate activities and opportunities for socialisation and stimulation.

6. Transport services. Difficulties with transport arose as an issue in many of the qualitative studies, particularly in relation to day care and in more rural communities.

Recommendations for future research

This is an expanding area of research with the majority of studies carried out in the last 10–15 years. However, there were few good-quality RCTs or quasi-experimental studies available for statistical synthesis. Many of the studies reviewed had small sample sizes with inadequate definitions of respite care. However, before larger-scale trials are carried out further developmental work is required to:

• classify available respite services and settings

• quantify carer and care recipient needs and preferences for respite care, taking account of the variations in the characteristics of the carer and care recipient

• define the characteristics of an appropriate intervention (e.g. the type and location, e.g. day care, home care; extent of activities provided within the intervention; duration; range of responsiveness to individual needs)

• define (and validate) outcomes for the care recipient based on type of intervention

• review the robustness of outcome measures for measuring carer burden and well-being

• develop process measures to assess the quality of the interventions.

From the studies reviewed it is not clear that any one type of respite intervention is better than another, but rather that different interventions are appropriate for different types of carer under different circumstances and that flexibility is important. This review has identified some of the factors that affect need, but further qualitative and quantitative work is needed to ensure that these are comprehensive.

Further large-scale trials should ensure that the following factors are considered in the design and execution of the studies:

• use of randomisation and/or appropriate control group comparisons

• outcomes mapped over both the short and long term including and accounting for mortality

• adequate service provision ensured, i.e. in terms of length and duration

• account taken of the uptake of respite and use of other services

• characterisation of the care recipient, i.e. type of disability, severity of condition from a cognitive, physical, affective and behavioural point of view, demographics

• characterisation of the carer and the caring situation, i.e. duration of caring, number of hours per day spent caring, type of caring activities, history of respite use, relationship to care recipient, demographics

• clear and detailed definitions of the intervention used, including skill mix, activities involved, location, duration

• a mixed-methods approach used to assess user views

• process measures utilised to assess the quality of the intervention

• an economic evaluation included

• an assessment of both carer and care recipient outcomes using well-validated measures and measures of utility.

The research should attempt to identify the particular components of the intervention that are likely to provide benefit for carers and care recipients so that they can be replicated in future service development. This will include not only the activities provided as part of the respite package but also the timing and duration of respite provision and the training and skill mix of the care providers. That is, it should address the questions of when respite is best provided in the caregiving career and what amount and duration would be expected to provide optimum benefit, as well as the types of intervention that might be appropriate for different types of carer. Defining these relationships will allow service providers to map services to their populations of carers.

All of the issues mentioned apply equally to ethnic minority groups who will have specific needs distinct from those of the white population. Only a small number of studies were available exploring the needs of ethnic minority groups and examining how culturally sensitive services may be developed. There are a limited number of ethnic-specific services available and research must address how this growing group of carers can be accommodated in the future.

Although the ability of an intervention to provide the carer with a mental break from care is an outcome for development and use in a large study as described above, qualitative work is needed to address and define the concept in more detail. A relatively small number of the studies in the qualitative review addressed this issue but there was some suggestion that a mental break could be achieved in other ways than by physical separation from the care recipient (e.g. by helping the carer to organise the caregiving schedule to allow some regular time to themselves or perhaps by more innovative methods such as cognitive behavioural intervention). The meaning of a mental break and how this can be achieved should be explored more fully.

There is a burgeoning amount of qualitative research in the areas that explore barriers to service use, the views of services and carer needs, and the impact of caring. The findings in relation to identifying barriers to uptake of respite are fairly consistent and future work should focus on identifying how the barriers can be overcome in order to develop interventions designed to break down these barriers. This qualitative work should lead to quantitative investigations and randomised trials of specific interventions. It is likely that there are a number of different types of intervention in this area ranging from individual-level to organisational and community-level interventions.

Related to interventions to overcome barriers to uptake is the need for further research identifying how information can best be provided concerning respite services. What information do carers require and at what point in time and who should deliver it? As part of this it is necessary to assess not only the knowledge of carers but also that of professionals who are likely to be in a position of supporting carers, particularly within the primary care context, which is likely to be the first port of call for many carers. Particular consideration should be given within this research to the needs of ethnic minority groups. From the little information that was available it appeared that ethnic minorities had less access to information than white populations and had less access to respite.

Finally, there was little research available concerning respite care in the palliative care context. The needs and issues that arose had some similarities to those relating to general respite care but they were also quite different in a number of ways. Respite was provided within the remit of palliative care teams, either in a domiciliary setting or in a hospice setting. Research is needed to address the specific respite needs of carers involved in terminal care and how these might be best incorporated into the range of services that are already available.

1. Administration details

1.1 Paper reference number:  ____________

1.2 Extractor initials: __________

1.3 Date information extracted: __________

2. Study identification details

 Title: ____________________________________________________________________________

  _____________________________________________________________________________________

  _____________________________________________________________________________________

2.2 First author (surname, initials):  _______________________

2.3 Date of publication: __________

2.4 Country of origin: UK  □₀

           Other  □₁

  _____________________ (specify)

2.5 Language of publication: English  □₀

             Other  □₁

  _____________________ (specify)

2.6 Primary source: Database □₀

          Hand searching  □₁

          Provided by user group   □₂

3. Study design details

3.1 Type of study (tick more than one option if applicable):  (a) RCT  □₁

                          (b) Quasi-experimental  □₁

                          (c) Cohort  □₁

                          (d) Observational  □₁

                          (e) Qualitative  □₁

                          (f) Description of services  □₁

                          (g) Other  □₁

                          (specify)__________________

3.2 Review category

 (tick more than one option if applicable): (a) Quantitative synthesis  □₁

                   (b) Qualitative synthesis  □₁

                   (c) Economic synthesis  □₁

3.3 Setting in which respite mostly provided

 (tick more than one option if applicable: if intervention compared with another form of respite indicate in text option):

                    (a) Inpatient  □₁

                    (b) Community  □₁

                    (c) Home  □₁

                    (d) Day care  □₁

                    (e) Hospice  □₁

                    (f) Other  □₁

  _______________________________________________________________________________________

3.4 Description of intervention or respite category (including location and duration):

  (a) Intervention 1: _________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  (b) Intervention 2:_________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  (c) Intervention 3:_________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  (d) Intervention 4:_________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

3.5 (a) Was the intervention delivered by health-care professional/s?

  Intervention 1: Yes  □₁  No  □₀ No information  □₂

  Intervention 2: Yes  □₁  No  □₀ No information  □₂

  Intervention 3: Yes  □₁  No  □₀ No information  □₂

  Intervention 4: Yes  □₁  No  □₀ No information  □₂

  (b) If yes, was the health-care team multidisciplinary?

  Intervention 1: Yes  □₁  No  □₀ No information  □₂

  Intervention 2: Yes  □₁  No  □₀ No information  □₂

  Intervention 3:  Yes  □₁  No  □₀ No information  □₂

  Intervention 4:  Yes  □₁  No  □₀ No information  □₂

3.6 Control group receiving no intervention? Yes  □₁  No  □₀

3.7 Planned vs ‘crisis’ respite (a) Planned    □₀

              (b) Crisis    □₁

              (c) No information   □₂

4. Participant details

4.1 Sampling method: (a) Random  □₀

           (b) Convenience  □₁

           (c) Population  □₂

           (d) Other  □₃

           (specify)_____________________

4.2 Inclusion/exclusion criteria:

  (a) Intervention/respite group/s:  __________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  _______________________________________________________________________________________

  (b) Control group:  _________________________________________________________________

  _______________________________________________________________________________________

  OR:

  (c) Total sample:  ________________________________________________________________________

  _______________________________________________________________________________________

4.3 Total number recruited:

  (a) Intervention 1:  _______________  Intervention 2:  _______________

    Intervention 3:  ________________  Intervention 4:  _______________

  (b) Control:   _______________

  OR:

  (c) Total sample:  _______________

4.4 Total number completing study:

  (a) Intervention 1:  _______________  Intervention 2:  _______________

    Intervention 3:  _______________  Intervention 4:  _______________

  (b) Control:   _______________

  OR:

  (c) Total sample:  _______________

4.5 (a) Were participants lost to follow-up? Yes  □₁ No  □₀

  (b) If yes, were sufficient details regarding the reasons for dropout given? Yes  □₁ No  □₀

  (additional information)__________________________________________________________________

5. Outcomes

6. Economic modelling

6.1 Are any cost data provided?

   Yes  □₁ No  □₀ (If yes, please give details)

    _______________________________________________________________________________________

    _______________________________________________________________________________________

    _______________________________________________________________________________________

6.2 Is information regarding health service usage provided (e.g. number of GP visits)?

   Yes  □₁ No  □₀ (If yes, please give details)

    _______________________________________________________________________________________

    _______________________________________________________________________________________

    _______________________________________________________________________________________

6.3 Is information regarding other service usage provided (e.g. caregiver time)?

   Yes  □₁ No  □₀ (If yes, please give details)

    _______________________________________________________________________________________

    _______________________________________________________________________________________

    _______________________________________________________________________________________

6.4 Does the paper provide a good description of respite services?

   Yes  □₁ No  □₀

6.5 Additional information/model used:

    _______________________________________________________________________________________

    _______________________________________________________________________________________

    _______________________________________________________________________________________