Background
Refugees are at an elevated risk of some mental disorders with studies highlighting the contributing role of post-migration factors. Studies of migrant groups show neighborhood social composition, such as ethnic density, to be important. This is the first longitudinal study to examine this question for refugees and uses a novel quasi-experimental design.

Methods
We followed a cohort of 44 033 refugees from being first assigned housing under the Danish dispersal policy, operating from 1986 to 1998, until 2019. This comprised, in effect, a natural experiment whereby the influence of assigned neighborhood could be determined independently of endogenous factors. We examined three aspects of neighborhood social composition: proportion of co-nationals, refugees, and first-generation migrants; and subsequent incidence of different mental disorders.

Results
Refugees assigned to neighborhoods with fewer co-nationals (lowest v. highest quartile) were more likely to receive a subsequent diagnosis of non-affective psychosis, incident rate ratio (IRR) 1.25 (95% confidence interval (CI) 1.06–1.48), and post-traumatic stress disorder (PTSD), IRR 1.21 (95% CI I.05–1.39). A comparable but smaller effect was observed for mood disorders but none observed for stress disorders overall. Neighborhood proportion of refugees was less clearly associated with subsequent mental disorders other than non-affective psychosis, IRR 1.24 (95% CI 1.03–1.50). We found no statistically significant associations with proportion of migrants.

Conclusions
For refugees, living in a neighborhood with a lower proportion of co-nationals is related to subsequent increased risk of diagnosed mental disorders particularly non-affective psychosis and PTSD.