Risk assessment scales to predict risk of hospital treated repeat self-harm: A cost-effectiveness modelling analysis

Risk scales are used widely for assessing individuals presenting to Emergency Departments (EDs) following self-harm. There is growing evidence that risk scales have limited clinical utility in identifying episodes at highest risk of repeat self-harm. However, their cost-effectiveness in terms of treatment allocation and subsequent repeat self-harm is unknown. We aimed to examine the cost-effectiveness of five risk scales (SAD PERSONS Scale, Modified SAD PERSONS Scale, ReACT Self-Harm Rule, Manchester Self-Harm Rule, Barratt Impulsivity Scale) and single item clinician and patient ratings of risk.

Quality-Adjusted Life Years were estimated for each episode. The five risk scales and the patient rating were compared to the clinician rating. Incremental cost-effectiveness ratios (ICERs) were estimated for each scale, using a range of ICER thresholds. Sensitivity analysis explored different model assumptions.

The formal scales were less cost-effective than the clinician and patient ratings across a range of ICER thresholds (£0-£30,000). The five scales were also less cost-effective than the clinician rating in most alternative scenario analyses. However, the clinician rating would be likely to result in unnecessary treatment costs for over half of patients identified as high risk.

Our primary model depended on the assumption that high-intensity care reduced patients' risk of further self-harm.

The use of formal assessment tools for managing self-harm presentations to EDs did not appear to be cost-effective. While the judgement of a mental health clinician was found to be slightly more cost-effective, it still resulted in incorrect allocation of costs and missed treatment opportunities.