Background. Deliberate self-harm can be costly, in terms of treatment and subsequent suicide. Any intervention that reduces episodes of self-harm might therefore have a major impact on the costs incurred by service providers and the productivity losses due to illness or premature death.

Method. Four hundred and eighty patients with a history of recurrent deliberate self-harm were randomized to manual-assisted cognitive behaviour therapy (MACT) or treatment as usual. Economic data were collected from patients at baseline, 6 and 12 months, and these data were complete for 397 patients. Incremental cost-effectiveness was explored using the primary outcome measure, proportion of patients having a repeat episode of deliberate self-harm, and quality of life. The uncertainty surrounding costs and effects was represented using cost-effectiveness acceptability curves.

Results. Differences in total cost per patient were statistically significant at 6 months in favour of MACT (−£897, 95% CI −1747 to −48, P=0·04), but these differences did not remain significant at 12 months (−£838, 95% CI −2142 to 466, P=0·21). Nevertheless, exploration of the uncertainty surrounding these estimates suggests there is >90% probability that MACT is a more cost-effective strategy for reducing the recurrence of deliberate self-harm in this population over 1 year than treatment as usual. The results for quality of life were not conclusive.

Conclusion. Cost-effectiveness acceptability curves demonstrate that, based on the evidence currently available, to reject MACT on traditional grounds of statistical significance and to continue funding current practice has <10% chance of being the correct decision in terms of cost-effectiveness.