Summary of main results
The aim of this review was to evaluate the benefits and adverse effects of respite care for people with dementia and their caregivers. See Summary of findings for the main comparison and Summary of findings 2.
Analysis of the available data showed no significant effects on caregiver outcomes when respite care was compared with no respite care in three studies, and there was no evaluable data for people with dementia. When respite care was compared to polarity therapy, a type of touch therapy that uses gentle pressure on energy points and biofields to help the client achieve physiological relaxation, a significant treatment effect was found in favour of polarity therapy for caregiver perceived stress, however, other measures of psychological health and other outcomes showed no significant effects. Again, there were no evaluable data for people with dementia.
Overall completeness and applicability of evidence
There are two possible explanations for these results, firstly, that in reality respite care has no effects on caregivers or, secondly, that any resultant effects are imperceptible due to the small sizes of the trials. In order to establish which of these is the case one must systematically assess the validity of the studies included in the review. There are three main issues to address here, the intervention, the people to whom the intervention was given and the outcomes of the intervention.
The interventions tested in the included studies were very different, although they all met our criteria for respite care by providing relief for the caregiver. The duration of the intervention for three of the trials was extremely short, consisting of two, six and eight weeks for Grant 2003, Wishart 2000 and Korn 2009, respectively. Given the prolonged and degenerative course of the diseases that cause dementia, even the year-long study by Lawton and colleagues could be considered too short. The intensity of the intervention also varied between studies. Participants in the Wishart study were provided with respite care for only two hours per week, in Korn 2009 they received three hours per week, while those in the Grant 2003 trial had 10 days of care (up to six hours per day). The frequency and amount of respite provided in the Lawton 1989 study depended on how much care the caregiver both wanted and felt able to afford. To use an analogy coined by Zarit in his extensive writings on this subject, it is possible that the respite care given in these studies was at a subclinical dosage.
The actual respite care received was also qualitatively different. The intervention in both Grant 2003 and Korn 2009 was in-home respite only, Lawton 1989 allowed the choice between in-home, daycare and institutional respite, and in Wishart 2000 the person with dementia was taken from the home on a walk. Different kinds of respite care are likely to have very different effects on both the caregivers and recipients and may be used in very different ways. Daycare and in-home respite care are likely to be used on a regular basis whereas institutional care is likely to be used on a more infrequent basis, and can be planned or unplanned in the case of, for example, caregiver illness. In-home respite care is said to be the most requested service while out of home daycare may increase the workload for caregivers by requiring them to prepare and transport the person with dementia (Berry 1991). It is recommended that future studies evaluate a single type of respite care and that future reviews consider each type of respite care separately.
A problem specific to the Lawton 1989 study was that caregivers were only given the opportunity to purchase respite care and were not provided with it free-of-charge as with the other three studies, although it should be noted that there were means-tested subsidies available. This introduces a further confounding factor to the study as only those who were able to afford respite care may have used it. This may have partly explained the low utilization of the intervention on offer. A further criticism of the Lawton 1989 study is that on examination of the range of hours of in-home respite used by participants in the control group in the year preceding baseline, some caregiving dyads were receiving full-time in-home care. This shows that the definition of respite care as being a temporary relief for caregivers was confounded in this study.
The Korn 2009 study compared respite care to a very specific alternative treatment, and only one of the four caregiver psychological health and stress outcomes were in favour of the alterative therapy, which could mean that either the trial was underpowered to find a difference on these outcomes or that respite care was equally as good as polarity therapy in improving the psychological and physical health of caregivers. It is unclear whether this trial has wider applicability as it was conducted in a very specific population, American Indians and Native Alaskans.
In terms of participants, the sample sizes in three of the studies were small. If an effect of respite care does exist it is likely to be small and may not be identified in studies of such limited size and quality. The samples of people with dementia were poorly defined, with none of the three included studies using any standard diagnostic tools. There was wide variation in the severity of cognitive impairment, which was likely to translate into a similarly large variation in the need for respite care. It has been shown that many caregivers do not make use of respite care early on in their relative's illness but wait until they have been caring for them for many years (Gottlieb 2000). If a large proportion of the sample were caring for a mildly impaired person then respite care might not have been expected to have a significant impact. Conversely, some researchers have suggested that not using respite care until the care recipients are severely impaired may mean that the caregivers are beyond help. Grant and colleagues split the caregivers in their sample into vulnerable and non-vulnerable subgroups according to the number of hours they spent on caregiving tasks on an average day and the amount of respite they had received in the preceding six months. It is probable that measurable differences in caregiver outcomes are more likely in vulnerable groups.
The relationships between the caregivers and those with dementia also differed among the studies. The caregivers in the Grant 2003 study were all the spouses of those with dementia and in Korn 2009 they were spouses and other family members. Caregivers in the Lawton 1989 and Wishart 2000 studies were enrolled irrespective of their relationship with the caregiver. Caregivers who are spouses of patients and those who care for their parents are said to have very different needs (Zarit 1999).
The validity of any randomised controlled trial also depends on the choice of control intervention. Wishart 2000 used a wait-list control group. This type of control has been criticised because the participants in the control group know that they will receive the intervention at some time, and this may have an impact upon their psychological wellbeing (Basham 1986). Grant 2003 and Lawton 1989 both used a no-treatment control. A problem with all of these designs is that the respite care provided as the intervention is not the only respite care available to the caregivers. Some caregivers in the control group of the Lawton 1989 study actually received more hours of respite care than those in the intervention group. This has been criticised in the literature as confounding the study (Gottlieb 2000; Zarit 1998; Zarit 1999) but may rather suggest that the way in which respite care was offered in the study was not as effective as the ways in which caregivers located it independently. In this case one would be evaluating a service designed to deliver respite care rather than respite care per se. Korn 2009 used respite care as the control, with participants engaging in a range of activities, and polarity therapy, a specific form of touch therapy, was the intervention. This is likely to have influenced the results and makes it difficult to compare the effects of respite care in this study with the other included studies.
Regarding outcomes, only one of the studies included any outcomes for the person with dementia (Lawton 1989). One of the most widely quoted statistics in the respite care literature is the reported 22-day increase of days spent in the community by the experimental group in the Lawton 1989 study. As already discussed the analysis in this publication was flawed due to the cluster randomisation process. This is one of the few studies to report a positive effect on rates of institutionalisation. Lawton 1989 also reported measures of functional status and mortality rates for the people with dementia, none of which were significantly different between the intervention group and the control group. In not reporting outcomes for the care recipient in the other three trials, a lack of consideration for the recipient is reflected in the wider literature of respite care; this point should be noted by researchers designing further trials.
In addition to the limitations of the outcomes measured for people with dementia, some of the measures used for assessment of the caregivers may have been inappropriate for that population. For example, the Hamilton scales were designed to monitor the progress of those diagnosed with depression or anxiety. It is unrealistic to expect a change by measuring populations such as those in the Grant 2003 study who largely have subclinical scores. A similar objection can be lodged against the use of the Global Severity Index of the Neuropsychiatric Inventory, which showed low baseline values. The Duke UNC Functional Support Questionnaire was also a questionable choice. It measures perceived social support and may not be sensitive enough to pick up changes in these populations.
Quality of the evidence
Overall the quality of the evidence, based on GRADE, was very low. One study did not report data that could be used in the analysis, the remaining three studies were very small and had short lengths of follow-up. Only Korn 2009 mentioned blinding of the outcome assessor. This means that preconceived ideas about the efficacy of respite care might have been allowed to influence the results.
Potential biases in the review process
We tried to identify all relevant trials through our search, however it is possible that we may have failed to identify some studies.
Agreements and disagreements with other studies or reviews
To counteract the problems regarding insufficient amount of respite and the need for a suitable control group, Zarit and colleagues carried out a quasi-experimental study comparing caregivers living in two different regions of the USA which were similar demographically but which provided different access to daycare facilities (Zarit 1998). The treatment group comprised 121 carers living in New Jersey, which has a well-developed, subsidised daycare programme, who had enrolled their relatives in this programme. The control group comprised 203 caregivers from Ohio and Pennsylvania where there are very limited daycare programmes. The choice of caregivers for the control group was restricted to those who stated that they would use daycare if it was available. Zarit and colleagues claim that the caregivers in the two groups were similar in all respects apart from their access to daycare; the control group also used very small amounts of other types of respite. The treatment group caregivers showed improvements at three months and 12 months on measures of caregiving-related stress and psychological wellbeing. An advantage of this study over some previous research is that large amounts of respite care were utilized by the treatment group, preventing the possibility that respite was received in amounts that were too small to be of value. However, the lack of randomisation to groups means that we cannot be sure that the differences between the groups were due to the daycare or whether they reflected differences between the groups in other ways that might have affected the results. Zarit indicates that the demographic characteristics of the two regions were similar on per capita income, education, proportion of the population over 65 years, unemployment rates, population density and proportion of minorities. The advantage of randomisation is that as well as controlling for factors that are known to affect relevant outcomes it controls for factors that are not known (Higgins 2008).