Can brief telephone interventions reduce caregiver burden and depression in caregivers of people with cognitive impairment? - long-term results of the German day-care study (RCT)

Caregivers in the intervention group will show more favorable values in caregiver burden and depressiveness than caregivers in the control group 12 months after the study began.

A further aim was to analyze whether subgroup-specific intervention effects according to the severity of cognitive impairment would be evident in the long-term.

Depending on the cognitive status of the care recipient, the effects will differ. The most favorable effects will hold for the group of caregivers caring for people with mild dementia.

The DeTaMAKS study (dementia – day-care – MAKS therapy; ISRCTN16412551), which began in October 2014 and ended in March 2017, was designed as a cluster-randomized, controlled, two-arm prospective longitudinal study. We used a waitlist control group design, which meant that during the 6-month intervention phase, caregivers in the day-care centers in the intervention group received a brief telephone intervention, whereas the person in their care received the non-pharmacological MAKS therapy at the day-care centers. The dyads comprised of the caregiver and the person with cognitive impairment (PCI) in the control group received no project-specific intervention during this period. After the 6-month intervention phase, the day-care center staff who worked with the control group also received training in MAKS therapy. The brief telephone intervention ended in the intervention groups after 6 months. No caregiver received telephone counseling initiated by the study team between 6 months follow-up and 12 months follow-up. There were also data collected further 12 months later, which are planned to be published by Pendergrass et al.

All study participants had the option to take part in any additional support services that were offered by the German Health Care System. The Ethics Committee of the Medical Faculty of Friedrich-Alexander-University Erlangen-Nuremberg examined and approved all procedures before the beginning of the study (Ref. 170_14 B). Thirty-four participating day-care centers were stratified by study region. By drawing lots, they were randomly assigned to the intervention or control group at baseline. All day-care center users were screened to ensure they met the inclusion criteria for the study. The main inclusion criterion for the PCIs was an existing cognitive impairment that met the threshold for Mild Cognitive Impairment (Mini-Mental State Examination (MMSE) > 23 and a Montreal Cognitive Assessment (MoCA) score ≤ 22) or mild to moderate dementia (MMSE between 10 and 23). PCIs who fulfilled all inclusion criteria were included in the study if their informal caregiver also agreed to participate in the project (please see the study protocol for all inclusion criteria and more details about the study design [14]). The caregiver had to provide home care without payment for the PCI but did not need to be a relative. All included caregivers and PCIs gave their written informed consent and were free to leave the study at any time. According to the Medical Research Council’s (MRC) Framework to design and evaluate complex interventions [19] this trial is in the Phase III (evaluation) since a fully defined intervention is evaluated using a protocol in a controlled study with appropriate statistical power. According to an update of the MRC’s framework “cluster-randomized trials are one solution to the problem of contamination of the control group” in evaluation complex interventions [20].

At the beginning of the study, 453 dyads consisting of a PCI and a caregiver were enrolled and allocated to the control or the intervention group depending on the cluster-randomization procedure. After 6 months (t1), a total of 359 (79.2%) participants (dyads) had completed the intervention period. The participants from 2 day-care centers had to be excluded during the intervention phase (first 6 months)—one because of insufficient administering of the intervention according to the manual, one because of a self-chosen termination of the collaboration agreement with the study center. For detailed reasons for the dropout and the distributions of the dropouts in the control and intervention groups between baseline and the 6-month follow-up, see Behrndt el al. [18]. Three hundred four dyads could be included in the analysis of the follow-up data after 12 months (t2). A total of 15.1% of participants had dropped out of the intervention group, and 14.9% had dropped out of the control group. Figure 1 presents the reasons for the dropout. The main reason was institutionalization (59.3%). One person had to be excluded from the analysis due to a change in caregiver.

The sample characteristics are depicted by group in Table 1. There was no statistically significant difference at the start of the study (t0) in any of the analyzed variables in the sample available at the 1-year follow-up (N = 304). The caregivers had a mean age of 59.5 years (SD = 11.6) ranging from 25 to 87. The persons with cognitive impairment had a mean age of 81.1 (SD = 7.5) ranging from 46 to 99, and 62.2% of them were women. Three-fourths (74.7%) of the caregivers were women, and about half of all caregivers were working (53.0%). The majority of them were the main caregiving person for the cognitively impaired (89.5%), and almost half of them lived together (46.4%). About two-thirds of the caregivers were children or in-laws (66.4%), and more than three-fourths were either married or living with a relationship partner (77.6%).

Caregivers from the intervention group received a brief outreach telephone intervention during the 6 months after baseline. The contents of three telephone calls were based on a manual designed specifically for the study. We incorporated tested procedures from stress psychology into the manual to fit the caregivers’ situations [21]. Psychoeducative information regarding dementia and dealing with dementia-specific challenging behaviors were also implemented [22, 23]. Each call lasted up to 1 h. The counselors had received training especially for the intervention and had the main task of supporting the caregivers in developing strategies for self-management [24]. Other goals were helping to reduce the stress due to home care and to help the caregivers deal with challenging behaviors. By improving the caregivers’ skills, the aim of the intervention was to “empower” them. For more detailed information about the telephone-based caregiver intervention and a detailed description of the procedure, see the published study protocol [14] and the report of the immediate effects [18]. After 6 months, telephone support from the study team stopped. Caregivers in both the control and intervention groups were free to use offers of support such as caregiver counseling from the German Health Care System.

During the 6-month intervention period, PCIs in the intervention group also received the multicomponent, non-pharmacological MAKS Therapy [15, 25, 26], which was provided by the day-care centers. MAKS therapy consisted of motor stimulation, the encouragement of activities of daily living, and cognitive stimulation in a social setting. After the 6-month intervention phase, the day-care centers of the intervention and control groups were free to offer MAKS therapy but without project-specific support or reimbursement.

Via computer assisted telephone interviews (CATIs), all caregiver were assessed once at baseline, at the end of the intervention period (after 6 months), and again after six more months (12 months after baseline) regarding the caregiving situation (self-rating), the PCIs’ situation (observer-based rating), and the care situation.

The analyses were computed with IBM SPSS Statistics 21. Descriptive statistics are provided as basic information. T-Tests for independent samples were used to test for significant differences, and Cohen’s d was calculated to obtain an effect size for group differences.

For the main analysis, in order to explore the effects of the RCT on the dependent variables, caregiver burden and depressiveness (Hypothesis 1), two mixed ANOVAs (mANOVA) with repeated measures were carried out. The within-subject variable was time; the between-subject variable was group (intervention versus waitlist control). To avoid over-interpretation of “unadjusted” effects in the main analysis, we determined a priori to carry out another analysis (secondary analysis) in which we controlled for other potential predictors of the dependent variable. This secondary analysis involved multiple linear regressions with either caregiver burden or depressiveness as the dependent variable and different caregiver-related constructs as well as health care system utilization variables as predictors. The secondary analysis was applied only to outcomes that were significant in the main analysis. Potential predictors were checked for multicollinearity before being included in the multiple regressions.

We computed difference scores (t2-t0) for the two main outcomes caregiver burden and depressiveness and compared them across the subgroups that were defined by the different levels of severity of the PCIs’ cognitive impairment (Hypothesis 2): Mild Cognitive Impairment (MCI), mild dementia, and moderate dementia. T-Tests and Cohen’s d were again used to provide interpretations of differences between the subgroups.

Since the matter concerns a cluster-randomized trial we report according to the CONSORT-suggestions the Intraclass correlation coefficient (ICC), which is .046 in this study. There were 34 daycare centers randomized, 17 to each study arm. The mean number of available places in the participating day-care centers was 26.8 for the control and 24.2 for the intervention group (t = .279; p = .782). The mean number of occupied places in the day-care centers was 15.6 for the control and 16.8 for the intervention group (t = −.561; p = .579). We compared the frequency distributions of recruited dyads between control and intervention group, there was also no significant difference (Chi² = 15.333; p = .500).

An analysis of differences between subgroups defined by the severity of cognitive impairment was carried out by analyzing difference scores between the 1-year follow-up and the start of the study in caregiver burden and depressiveness.

For caregiver burden, a significant small-sized effect (d = .443; p = .016) was observed in the group of caregivers who cared for persons with mild dementia. In the group of caregivers with a relatively mildly cognitively impaired person, the effect was descriptively favorable for the intervention group (d = .374) but was not significant (p = .124). However, caregivers who cared for a person with moderate dementia experienced – descriptively – a more intense worsening of burden when in the intervention group 1 year after the beginning of the study (and 6 months after the intervention had ended) compared with the control group, but again, this effect was not statistically significant (p = .238). See Fig. 4 for a graphical representation. The mean difference scores and corresponding t-statistics are included in Table 4 for each group.

For depressiveness (see Fig. 5), the pattern was basically the same: a significant moderate-sized effect (d = .520; p = .005) was observed in the group of caregivers who cared for persons with mild dementia. A descriptively favorable but nonsignificant (p = .542) effect for the intervention group (d = .147) was seen in caregivers of persons with MCI. Again, an statistical non-significant reversal of the effect was seen in the group of caregivers of people with moderate dementia: There was a seemingly (d = 0.169) more intense worsening of depressiveness in the intervention group 6 months after the end of the intervention compared with the control group, but this change was not statistically significant (p = .382). See Table 4 for mean difference scores and the corresponding t-statistics.

This study has several strengths: First, we applied an RCT with as much blinding as possible when the psychosocial interventions were applied (assessments via computer-assisted telephone interviews (CATIs) with blinded assessors). At the same time, the results demonstrated high ecological validity because the interventions were applied in existing health care structures, that is, day-care centers. Second, the interventions were standardized, manualized, and published so that they could be assessed by other experts and avoid the “black-box phenomenon.” Third, the outcome measurement included well-researched outcomes that were operationalized with established and well-validated instruments. Fourth, the quality of the data was high because there were only a very small number of single missing values that needed to be imputed. Fifth, there was high transparency in the conducting and reporting of the study, that is, the study protocol was published [14], the study was registered, and a second research institution checked the data for plausibility.

However, there were also limitations that should be taken into account when interpreting the results. First, complete blinding was not possible – both the counselors and the participants were aware of who received the intervention – because the intervention concerned phone calls.

Second, outcome data were assessed via self-report. However, possible biases in self-reports (e.g. social desirability or regression toward the mean) would be likely to impact both the control and intervention groups.

Third, there is a noticeable dropout rate due to death or institutionalization, and this might be associated with a certain selection process (the most healthy and less affected care recipients and their caregivers remained in the sample). However, this also affected both the intervention and control groups. The baseline characteristics of the sample analyzed here were compared with the baseline characteristics (t0) of the sample analyzed by Behrndt et al. [18] to evaluate the short-term (6-month = t1) effects. If there was a substantial selection effect, the sample analyzed here (t0-t2) should have more favorable values in important constructs such as burden, cognitive status, ADL, and hours of care than the previously analyzed sample (t0-t1). However, the differences in the sample presented here were rather small to marginal. Subjective burden was 0.2 points lower in both the intervention group and the control group than in the t0-t1 sample. Cognitive status was the same in the intervention groups of the two samples and 0.4 points lower in the control group of the t0-t1 sample. For ADLs, the sample analyzed here was .2 (IG) or .7 (CG) points better than the t0-t1 sample. Finally, there was no difference in hours of care between the two samples in the intervention group, and in the control group, the t0-t2 sample had .2 fewer hours.

Fourth, the dropout rate led to a smaller N, which reduced the power of the statistical tests and may have contributed to the nonsignificant difference between the groups in the mean difference scores for the outcome, which displayed the expected direction of outcome development but failed to achieve statistical significance.

Fifth, the matter concerned a “combined intervention”: in the same period in which the telephone counselling was carried out, the individuals with cognitive impairment received a multi-component, non-drug therapy at the day-care centers. The patient-oriented intervention influenced cognition, activities of daily living and neuropsychiatric symptoms [15]. However, Behrndt et al. [18] in consequence checked whether a change in neuropsychiatric symptoms in the PCI during the therapeutic phase was a predictor for caregiver burden or caregiver depressiveness in the short-term results (right after the intervention stopped: t1). This was not the case. Behrndt et al. conclude: “on the whole, the effect of the caregiver intervention cannot be attributed to the change in PCIs’ neuropsychiatric symptoms” [18]. Thus it is implausible that this construct has significant predictive power for caregiver burden/depressiveness 6 months after the end of both intervention elements. Behrndt et al. suggest though, that “for a final clarification of a potential confounding effect of the MAKS therapy on the caregiver telephone intervention, a new study with separately administered single interventions is necessary” [18].

Sixth, since the intervention effect seemed to wash out as soon as the intervention stopped one needed for a more precise evaluation of the telephone counselling intervention in a future trial an active control group which fulfilled the condition of receiving attention. Only then it would be possible to differentiate whether the contents of the counselling or simply giving someone “support” was responsible for reduction of burden.

Brief telephone counseling for caregivers of PCIs who are attending day-care centers has the advantages of offering an inexpensive, low threshold, time-saving way to offer support for caregivers. The use of trained external counselors secured a quality standard and also time flexibility. However, the data suggest that the effects of this support fade fast as soon as the counseling stops. Caregiver burden and depressiveness increase when the caregiver is caring for a chronically progressive care recipient with cognitive impairment when the caregiver does not receive sufficient support. Thus, to maintain the positive effects of the telephone intervention, it seems urgently important to continue counseling.