An In-home Respite Care Program to Support Informal Caregivers of People With Dementia

The in-home respite intervention group will receive in-home respite care also called "Baluchonnage". During that period, lasting at least five days (24/24hours), an internally trained or experienced care worker for persons with dementia takes over all caregiving tasks while the caregiver takes a break. This care worker also writes down his/her observations in a diary. The content of this diary enables the caregiver to validate their perceptions, provides strategies to deal with difficult situations, and let them feel that somebody can truly understand their daily challenges. The control group receives usual care (standard dementia care).

Participants of the in-home respite intervention group will be recruited by Baluchon staff when asking for a first or a new period of in-home respite care. First, Baluchon staff will describe the study to the dyads and ask them about their preparedness to participate. When they give their verbal consent contact information will be forwarded to the research team who will then contact them by phone. During this call a home visit will be planned to sign the informed consent and complete the baseline assessment. Participants in the control group will be recruited from several general practitioners spread over the different regions in Belgium. Potentially, in a later phase, control group participants will also be recruited via Memory Clinics in Belgium.

Similar to the recruitment of the intervention group, the physician will give some information about the study. When verbal consent to participate is given to the physician, caregivers will be contacted by phone by the research team to check eligibility. This will be done by listing different support strategies (including "Baluchonnage") and ask the caregiver if they would be interested to receive this type of support. Also previous use of Baluchonnage will be questioned because this is not allowed in the control group. When caregivers appear to be eligible the research team will plan a home visit to sign the informed consent and complete the baseline assessment.

Data collection of caregiver and care-recipient characteristics and research outcomes will be gathered at several assessment moments. The baseline assessment will be conducted during a home visit by a member of the research team. Background characteristics and baseline values of the research outcomes will be gathered during this assessment. For the intervention group this will be in the week preceding the respite period. For the control group this will take place after inclusion. A second assessment will be done in the intervention group two weeks after having the intervention. A third assessment moment will be done 6 months after inclusion and a last follow up assessment will be conducted after 12 months. To fulfill the assessments after the baseline assessment, caregivers will be contacted by telephone. All questionnaires will be available in Dutch and French for use in Belgium. When no valid translation of a certain questionnaire exists in Dutch or French for Belgium a back-forward translation method will be performed.

For caregivers the researchers will obtain the following background characteristics: age, gender, region, marital status, ethnicity, educational level, relationship with the care-recipient, employment status, living situation and time spent in caregiving. In both study groups the reason for (potential) use of respite will be asked. Earlier use of Baluchonnage will also be questioned to intervention group participants. Time spent in caregiving (measured twice) will be calculated using the RUD instrument (Resource Utilization in Dementia). For the care-recipient the following background characteristics will be collected at baseline: age, gender, marital status, ethnicity, region, National Registration Number, educational level, being on a waiting list for long-term nursing home placement, dementia-specific medication use, and severity of dementia. The National Registration Number of the person with dementia will be gathered. This will allow us to trace study participants in the IMA database (Intermutualistisch Agentschap) and gather information on resource use and date of placement in a nursing home over a longer time period than the trial. ADL (Activities of Daily Living) functioning will be measured using the Belgian Katz scale and severity of dementia with the Global Deterioration Scale. Both will be measured twice (baseline and after 6 months).

Sample size calculation is performed using SPSS SamplePower 3®. First, literature was searched to determine the needed effect size and standard deviation (SD) allowing to distinguish a statistical significant and clinical relevant difference in the primary outcome, i.e. burden. Based on the findings of a similar high quality intervention study an effect size of 0.4 was used in the analysis, which implies a difference of six points on the ZBI scale (Zarit Burden Interview), and a SD of 15. Also, an average drop-out rate of 20% was taken into account, a power of 80% and a significance level at 0.05. Based on these values a total of 124 caregiver/care-recipient dyads will be needed for the intervention group. To reduce selection-bias inherent to quasi-experimental studies an allocation ratio of 1:2 will be used to allow matching techniques. As a result 248 caregiver/care-recipient dyads will be needed in the control group.

When appropriate, i.e. if the intervention demonstrates an effect in the study endpoints, within-trial and modeled cost-effectiveness analyses will be conducted in a separate economic evaluation plan. First the analysis will be performed from the perspective of the health care payer taking into account direct health care costs for the government's health care budget (RIZIV/INAMI) as well as patients' co-payment. Assuming that policy makers would consider reimbursement of in-home respite in the future the daily cost of Baluchonnage will thus be included in the analyses. Alongside the costs of Baluchonnage other costs of healthcare resource use including hospital and community care will be included. Additionally, also a full societal viewpoint will be undertaken not only including potential health care costs for the health care budget and patients, but also all other direct and indirect costs for caregivers and patients. Because in this viewpoint everyone affected by the intervention should be considered, caregiver time and costs (i.e. time spent in caregiving and productivity loss) as well as resource use and costs of other sectors (f.e. food delivery) should be included. Based on the KCE (Belgian Health Care Knowledge Center) guidelines for health economic evaluations in Belgium future costs will be discounted at 3% and future QALY's at 1.5%. Health utilities or QALY's of caregivers from both intervention arms will be derived from the Belgian public preference list based on scores on the EQ-5D (general utility instrument). Decision analytic modeling will be carried out to extrapolate effects of the intervention found in the trial to a longer time horizon. The model will be based on results from the trial as well as existing data from literature. Assumptions, hypotheses and sources of information will be represented in a transparent and clear way. Finally, the model will be validated by experts.

Data on resources used by the care-recipient of each study group will be obtained by an adapted version of the RUD instrument at baseline and after six months inclusion. This RUD instrument attempts to include all resource use including: health care resource use containing hospital resources (in-patient and out-patient attendances,…) and community care resources (general practitioner visits, nurse visits,…), caregiver and patient resources (time spent in caregiver,…), and resource use in other sectors (social worker visits, home help visits,…). Duration and frequency of the used services will be multiplied by each unit cost of the corresponding service. These unit costs will be obtained from the Belgian Reimbursement scheme using standard fees for regularly insured patients and other publicly available sources. Time spent in caregiving will be calculated at baseline and after six months using the recall method for which also a part of the RUD instrument will be utilized. Next, the amount of time spent in caregiving will be monetized using the opportunity cost method which estimates the value of lost informal caregiver benefits due to spending time on providing informal care. For informal caregivers not active on the labor market wage rates of similar people will be imputed. Basically, productivity loss should be included but given the average high age of informal caregivers it can be expected that productivity loss will have a negligible impact on costs because most of them are retired. Nevertheless, based on the results of mean age and employment status of the included caregivers the researchers will decide to whether or not include productivity loss as a cost. The costs of the intervention will be obtained from the current unit cost per day of Baluchonnage and multiplied by the amount of days the in-home respite care was delivered. Currently the daily cost of Baluchonnage is set at 350€/day covered by charity and 65€/day direct cost for the patient. Additional costs above the fixed daily tariff for Baluchonnage will also be included such as: travel expenses during the respite period and having a pet (5€ extra per day). Additionally, willingness to pay for one day of in-home respite care by Baluchon will be obtained from both study groups using the contingent valuation method (CVM). This method can be defined as a stated preference method for eliciting a monetary value to a health care program. In this trial a closed response format will be used. Finally, for use in the decision analytic model, costs of nursing home placement will be derived from the average daily cost for staying in a nursing home in Belgium at the time of completing secondary endpoint.

Descriptive statistics will be represented to draw a clear profile of the characteristics of study participants. Therefore the mean, percentages and the standard deviations of all continuous variables will be displayed. To determine possible baseline differences between the groups, mean values of baseline characteristics will be compared using independent sample t-tests for continuous variables if normally distributed or by performing Pearson's chi-square test for categorical variables. To help control for bias and confounding statistical techniques such as propensity score matching will be used. Propensity score matching can be seen as a tool to simulate a RCT setting. In this way the observed effect can be considered an unbiased estimate of the real effect. To investigate possible effects of the intervention on the primary and secondary outcomes, analysis of variance will be conducted if the outcome variables are normally distributed. A P-value of 0.05 will be considered as significant. All analyses will be based on intention to treat also taking drop-outs into account and avoiding overestimation of respite care effects. When the intervention is effective a cost-effectiveness analysis in a separate economic evaluation plan will be performed. Also ICERs (Incremental Cost-Effectiveness Ratio) will be calculated for the mean and upper and lower confidence levels of the costs and consequences. To explore uncertainty one-way-sensitivity analysis will be conducted around the ICER and illustrated in a Tornado diagram. Additionally, on all input variables a probabilistic sensitivity analysis, also called Monte Carlo analysis, will be conducted to test robustness of the model. These results will also be illustrated. Finally, the results and the willingness to pay threshold of the Belgian Health Care System will be presented in a cost-effectiveness acceptability curve. To compare time to nursing home placement of persons who use respite care to those who don't, the researchers will use Kaplan-Meier survival curves to illustrate association between the comparison groups. By additionally conducting a log-rank test statistical difference between the groups in time to placement can be found.