BackgroundUrban birth and upbringing are consistently associated with schizophrenia and other psychoses but the key urban exposures remain unknown. China has previously found prevalence of psychosis higher in rural areas but has shown the largest displacement of population from rural into urban areas of any country in the world over the past 40 years. Studies of psychotic experiences (PE) show inconsistency but this may be due to confounding of PEs by depressive symptoms. This suggests the need to differentiate between PEs on a population continuum with non-affective psychosis and those secondary to common mental disorders when studying urbanicity. Our aims were to investigate effects of exposure to urban birth and upbringing on psychosis in a large Chinese undergraduate sample.MethodsCross-sectional surveys conducted annually during first year of university, 2014–2018, n=39,446. Self-reported categorical measures of psychosis included psychoticism, paranoid ideation, and schizotypal symptoms using SCL-90-R, and lifetime clinical diagnosis of schizophrenia; depressive symptoms using PHQ 9; putative etiological risk factors of family history and childhood maltreatment; urbanicity measured according to birth place in a major city (level 5 of a 5-level rural-urban scale), length of residence in urban location, and length of residence during three critical 5 year periods, birth - 15 years. We studied effects on our 4 psychosis phenotypes of (i) urban birth, (ii) urban living, (iii) critical times of exposure to urban environment, (iv) putative etiological risk factors, before and after adjusting for depression. We investigated associations between etiological risk factors and urban exposures.ResultsWe identified 2,143 (5.4%) participants above a determined cut-off for psychoticism, 2,081 (5.3%) for paranoia, 760 (1.9%)with schizotypal symptoms, and 53 (0.1%) with schizophrenia. Effects of urban exposure on our psychosis phenotypes were only revealed following adjustments for depression in our models: Urban birth was associated with Paranoia (1.46, 1.24–1.70), schizotypal symptoms (1.90,1.48-2.42), and schizophrenia (2.30, 1.14–4.63), but not psychoticism. All four phenotypes were associated with 10–15 years of exposure to urban living, but not shorter periods. Only schizophrenia was associated with critical timings of total exposures of 1–3 years and 4–5 years during the first 5 years of life to an urban environment. There were no associations or negative associations between putative etiological factors and urban exposures.DiscussionWe confirmed that urban birth and living were associated with PEs and schizophrenia in this large sample of Chinese university students, but these findings only emerged after adjusting for depression. Depression is more prevalent in rural Chinese samples and previous studies may have been confounded by effects of PEs secondary to depression. There was a gradient of association between paranoia, schizotypal symptoms and schizophrenia, the latter showing strongest effects in association with urban exposures of birth and length of time in an urban environment. Only schizophrenia showed effects of critical timing of exposure to urban environment during infancy. Finally, we could not identify what exposures in the urban environment contributed to psychosis in our sample - although we could identify the etiological factors that did not. Among Sichuan students, there was no indication that urban effects were due to increased risk from demographic factors of male sex, lower family income, increased genetic risk, or child maltreatment, although these factors showed some effects on psychosis across the entire sample which included previous rural residents.