The selection hypothesis posits that the increased rates of psychosis observed among migrants are due to selective migration of people who are predisposed to develop the disorder. To test this hypothesis, we examined whether risk factors for psychosis are more prevalent among future emigrants.

A cohort of 49 321 Swedish military conscripts was assessed at age 18 years on cannabis use, IQ, psychiatric diagnosis, social adjustment, history of trauma and urbanicity of place of upbringing. Through data linkage we examined whether these exposures predicted emigration out of Sweden. We also calculated the emigrants' hypothetical relative risk compared with non-emigrants for developing a non-affective psychotic disorder.

Low IQ [odds ratio (OR) 0.5, 95% confidence interval (95% CI) 0.3–0.9] and ‘poor social adjustment’ (OR 0.4, 95% CI 0.2–0.8) were significantly less prevalent among prospective emigrants, whereas a history of urban upbringing (OR 2.3, 95% CI 1.4–3.7) was significantly more common. Apart from a non-significant increase in cannabis use among emigrants (OR 1.6, 95% CI 0.8–3.1), there were no major group differences in any other risk factors. Compared to non-emigrants, hypothetical relative risks for developing non-affective psychotic disorder were 0.7 (95% CI 0.4–1.2) and 0.8 (95% CI 0.7–1.0), respectively, for emigrants narrowly and broadly defined.

This study adds to an increasing body of evidence opposing the selection hypothesis.