A comparison of neighbourhood level variation and risk factors for affective versus non-affective psychosis

BACKGROUND: Studies typically highlight area level variation in the incidence of non-affective but not affective psychoses. We compared neighbourhood-level variation for both types of disorder, and the specific effects of neighbourhood urbanicity and ethnic density, using Danish national registry data. METHODS: Population based cohort (2,224,464 people) followed from 1980 to 2013 with neighbourhood exposure measured at age 15 and incidence modelled using multilevel Poisson regression. RESULTS: Neighbourhood variation was similar for both disorders with an adjusted median risk ratio of 1.37 (95% CI 1.34-1.39) for non-affective psychosis and 1.43 (1.38-1.49) for affective psychosis. Associations with neighbourhood urbanicity differed: living in the most compared to the least urban quintile at age 15 was associated with a minimal increase in subsequent affective psychosis, IRR 1.13 (1.01-1.27) but a substantial increase in rates of non-affective psychosis, IRR 1.66 (1.57-1.75). Mixed results were found for neighbourhood ethnic density: for Middle Eastern migrants there was an increased average incidence of both affective, IRR 1.54 (1.19-1.99), and non-affective psychoses, 1.13 (1.04-1.23) associated with each decrease in ethnic density quintile, with a similar pattern for African migrants, but for European migrants ethnic density appeared to be associated with non-affective psychosis only. CONCLUSIONS: While overall variation and the effect of neighbourhood ethnic density were similar for both types of disorder, associations with urbanicity were largely confined to non-affective psychosis. This may reflect differences in aetiological pathways although the mechanism behind these differences remains unknown.