Examining Long-Term Missing (LTM) Persons Cases to Advance the Study of the UK Police Missing Persons Risk Assessment

Long-term missing (LTM) cases remain underexamined within missing persons research despite their operational, safeguarding, and investigative significance. This study analyses a UK police dataset of 4,746 missing person cases, including 237 cases unresolved beyond 28 days, to examine which features of the 19-point police risk assessment are most strongly associated with LTM status. Using a Random Forest machine learning model with synthetic oversampling to address class imbalance, the analysis identified several features that contributed to distinguishing LTM from resolved cases including physical or mental health issues, out-of-character disappearance, an identifiable reason for going missing, suicide risk, and broader case-specific “other” factors. When considered alongside previous studies of high-risk classification and harmful outcomes, only out-of-character behaviour and suicide risk appear consistently across all three categories. In contrast, physical or mental health issues, reasons for disappearance, and “other” factors appear more closely associated with harmful and LTM outcomes than with high-risk classification. The discussion considers whether these findings suggest that some forms of vulnerability may be less visible, more ambiguous, or more difficult to translate into immediate operational risk categories. It also discusses missing person risk assessment as a potential classification–outcome problem and presents a framework to aid interpretation of the various possible permutations. Implications are outlined for police decision-making, investigative reviews, evidence-based practice, and future research.