Analytical approach:
The analysis was conducted concurrently with a single trial. The two time horizons were 15 and 24 months. The authors reported that a societal perspective was taken.
Effectiveness data:
The evidence came from a randomised clinical trial (Charlesworth, et al. 2008, see ‘Other Publications of Related Interest’ below for bibliographic details), with 236 carers (18 were lost to follow-up at six months and not included) and a 24-month follow-up. Some estimation of missing values was performed for a large proportion of the sample, who did not have complete data and the analysis was repeated using complete case data only. Of the 157 carers without complete data, 80% were missing four or fewer of the 136 data items. The primary outcome measures in the effectiveness trial were carer wellbeing, which was measured by the Hospital Anxiety and Depression Scale, and carer health-related quality of life, which was measured with the European Quality of life (EQ-5D) questionnaire.
Monetary benefit and utility valuations:
Carers completed the EQ-5D at baseline, 6, 15 and 24 months. These EQ-5D profiles were converted into utilities using UK general population conversion rates.
Measure of benefit:
Quality-adjusted life-years (QALYs) were the summary measure of benefit. A 3.5% discount rate was applied to those accrued from months 15 to 24.
Cost data:
The cost categories included National Health Service (NHS) and social services, voluntary and household sectors, and the costs of the informal care time of carers, family, and friends. Data on the resource use of both the carers and the patients were included and were collected by questionnaire. The cost of the intervention itself was based on facilitators' and carers' reports and was divided into a fixed and variable component. The unit costs came from relevant national or local sources. The price year was 2005, and prices were expressed in UK pounds sterling (£). A 3.5% discount rate was applied to those costs accrued from months 15 to 24 (as there were no data at 12 months).
Analysis of uncertainty:
Confidence intervals (CIs) for the outcomes and costs were generated using a nonparametric bootstrap approach. Additional scenario analyses were performed from the perspectives of the statutory sector (NHS and social services), voluntary and household sectors, and from the societal perspective including both carer and patient QALYs. Cost-effectiveness acceptability curves for each perspective were reported.