Using clustering techniques to identify localities with multiple health and social needs

Development of health promoting policies requires an understanding not just of the interplay between different measures of health but also their relationship with broader education, criminal justice and other social issues. Methods to better utilise multi-sectoral data to inform policy are needed. Applying clustering techniques to 30 health and social metrics we identify 5 distinct local authority types, with poor outcomes for the majority of metrics concentrated in the same cluster. Clusters were distinguished especially by levels of: child poverty; breastfeeding initiation; children's tooth decay; teenage pregnancy; healthy eating; mental illness; tuberculosis and smoking deaths. Membership of the worst cluster (C5) was focused in Northern England which contains 15.7% of authorities analysed (n=324), but 63.0% of those in C5. The concentration of challenges in certain areas creates disproportionate pressures that may exceed the cumulative effects of individual challenges. Such distinct health clusters also raise issues of transferability of effective policies between areas with different cluster membership.