Risk assessment tools used at the policing stage for health and crime outcomes: A systematic review and meta-analysis.

INTRODUCTION: Risk assessment tools are increasingly employed at the policing stage to consider health-related needs and predict crime outcomes such as repeat intimate partner violence (IPV). Decisions informed by these tools can influence health outcomes, disproportionately affecting certain groups and potentially exacerbating health inequalities. However, their predictive accuracy, quality, and reliability remain uncertain. METHODS: Following PRISMA guidelines, we conducted a systematic review of development/derivation and validation studies on risk assessment tools used at the policing stage to evaluate health and crime outcomes. RESULTS: We identified 29 studies with 256,125 participants, reporting performance measures for 28 different tools for a range of health related and crime outcomes. The most common assessment was risk of IPV (18 studies or 62 %), and three studies (10 %) considered other health-related outcomes. Pooled estimates for outcomes ranged from 0.64 to 0.73, representing poor to moderate performance. The tools for predicting IPV demonstrated the weakest performance with a pooled AUC of 0.64 (95 % CI: 0.62, 0.66). The reporting of other performance measures beyond the AUC, such as true and false positives and negatives, and calibration was inadequate. CONCLUSIONS: Current evidence for the routine use of existing risk assessment tools at the policing stage is limited. Some newer tools, developed with robust methodologies, show high predictive performance. Research should prioritize the development, validation, and implementation of these newer tools, particularly for outcomes with significant morbidity and mortality, such as IPV victimization. Implementing higher quality tools could reduce health inequalities by fostering more consistent decision-making and efficient resource allocation.