Can police data predict harm? Evaluating risk in missing person cases using a Harm Vulnerability Framework

Assumed risk of harm plays a crucial role in response strategies and resource allocation in missing person investigations, where misclassification or delay can result in serious or fatal consequences. While police adopt a clinical judgement approach to assessing risk, few empirical studies have examined whether routinely captured police data can support structured harm prioritisation tools. Drawing on Lifetime Exposure Theory, logistic regression analyses were conducted on missing incidents (N = 16,454) and individual-level cases (N = 4206) reported to UK police forces to test whether variables from a Harm Vulnerability Framework predicted harmful outcomes. Among child incidents, being high risk, older, female, transgender, likely to commit suicide, mental health issues, and prior harm were associated with harm, although no variables were significant at the individual level. Among adult incidents, being high risk, likely to commit suicide, and prior harm were significant predictors, whilst being high or medium risk, and likely to commit suicide predicted harm at the individual level. Multivariate regression analyses confirmed the predictive value of suicide concern and prior harm across children and adults, although explained variance remained modest. Suicide concern already forms a key aspect of police decision-making. However, harm suffered during a previous missing episode may warrant greater operational attention. Results highlight opportunities for developing child-focused harm prioritisation tools while also revealing limitations in police data. Future research should explore how police and partner-agency interventions influence harm outcomes and how other data sources can strengthen harm assessments.