Targeting Missing Persons Most Likely to Come to Harm Among 92,681 Cases Reported to Devon and Cornwall Police

Funder: University of CambridgeAbstract Research question Given the information available to Devon and Cornwall Police at the time when they received a missing person report, which knowable variables indicated increased odds of the person coming to serious harm? Data The study examines all 92,681 missing reports received by Devon and Cornwall Police over 11 years from 2008 through 2019, for which 3481 (3.8%) persons came to harm, or about one in 27. Variables known at the time the report was filed included (A) risk levels estimated by police applying the College of Policing risk assessment template (high, medium, low), and (B) the missing person’s age, gender, in-care status, disability, dyslexia, learning disability, hearing or visual impairment, reduced mobility, mental illness, child sexual exploitation risk, reported suicidal, vulnerable adult status and previous reports of being missing. Methods Odds ratios are used to estimate differences in likelihood of missing persons coming to harm in a series of single-variable tests. Matrices are used to assess the accuracy of the current risk assessment process used by police services in England and Wales, as applied in Devon and Cornwall. Limited multivariate analysis was undertaken. Findings Application of the current College of Policing protocol for risk assessment by subjective professional judgements yields substantial error rates, with 89% of the predicted high risk cases having no actual harm and 59% of cases with actual harm after not being predicted as high risk. The odds of harm based on single variables examined are highly conditional on age and gender as third factors. Both men and women over 18 are 4 times more likely to be harmed (6.8%) than those under age 18 (1.7%). Conclusions This study casts substantial doubt on the accuracy of the current subjective risk assessment process for missing persons. As the authorizing body for Approved Professional Practice, the College of Policing could use this evidence to endorse further development and implementation of a multivariate evidence-based risk assessment tool for missing persons that takes into account the age and gender of the missing person as well as all other factors in a single forecasting model.