Mental health acute crisis episodes are associated with high inpatient costs. Self-management interventions may reduce readmission by enabling individuals to manage their condition. Delivery of such interventions by Peer Support Workers (PSWs) may be cost-effective. CORE, a randomized control trial of a PSW self-management intervention compared to usual care, found a significant reduction in admissions to acute mental healthcare for participants receiving the intervention. This paper aims to evaluate the cost-effectiveness of the intervention over 12 months from a mental health service perspective. Analysis methods of increasing complexity were used to account for data missingness and distribution.
Participants were recruited from six crisis resolution teams in England from 12 March 2014 to 3 July 2015 (trial registration ISRCTN: 01027104). Resource use was collected from patient records at baseline and 12 months. The EQ-5D-3L was collected at baseline and 4 and 18 months, and linear interpolation was used to calculate 12-month values for quality-adjusted life-years (QALYs). The primary analysis of adjusted mean incremental costs and QALYs for complete cases are calculated separately using OLS regression. Secondly, a complete-case non-parametric two-stage bootstrap (TSB) was performed. The impacts of missing data and skewed cost data were explored using multiple imputation using chained equations and general linear models, respectively.
Four hundred and forty-one participants were recruited to CORE; 221 randomized to the PSW intervention and 220 to usual care plus workbook. The probability that the PSW intervention was cost-effective compared with the workbook plus usual care control at 12 months varied with the method used, and ranged from 57% to 96% at a cost-effectiveness threshold of £20,000 per QALY gained.
There was a minimum 57% chance that the intervention was cost-effective compared to the control using 12-month costs and QALYs. The probability varied by 40% when methods were employed to account for the relationship between costs and QALYs, but which restricted the sample to those who provided both complete cost and utility data. Caution should therefore be applied when selecting methods for the evaluation of healthcare interventions that aim to increase precision but may introduce bias if missing data are heavily unbalanced between costs and outcomes.