Geographic profiling of psychosis prevalence by ethnic group: an application of ecological inference to London neighbourhoods

Abstract In view of ethnic and geographic heterogeneity in psychosis, establishing varying psychosis prevalence both by neighbourhood, and by ethnic group, is important for mental health needs assessment and service planning. This paper uses an ecological inference framework to estimate neighbourhood psychosis rates by ethnic group, when the available information is limited to totals of diagnosed psychosis by neighbourhood. To assess sensitivity in inferences, and possible spatial confounding, alternative models for ethnic-area psychosis rates are used. In an application to 4,835 micro-neighbourhoods across London, the highest psychosis rates are among Black ethnic groups, with a rate ratio averaging 2.35 compared with prevalence among Whites. There is evidence for an ethnic density effect: lower psychosis rates than expected in neighbourhoods with high non-White ethnic concentration.