The assumption that cannabis risk is concentrated in teenagers may need significant revision. This large meta-analysis reveals a counterintuitive age gradient: the psychiatric cost of cannabis use, measured as earlier onset of psychosis, grows substantially larger the older a person is when psychosis first emerges — a finding with real implications for aging populations in an era of expanding cannabis legalization.
Drawing on 149 peer-reviewed studies encompassing 181 independent samples — 18,272 cannabis users and 52,801 non-users — the analysis calculated a pooled standardized mean difference representing roughly 2.5 years earlier age at psychosis onset among cannabis users overall. The headline figure, however, conceals a striking dose-response-like age gradient. Among individuals whose psychosis onset occurred before age 20, cannabis use was associated with essentially no difference in onset timing compared to non-users. But for those with onset between 25 and 30, cannabis was associated with a 4.43-year earlier onset, and for those whose first episode occurred at 30 or older, the gap widened to 6.30 years. Sample mean onset age emerged as a statistically independent moderator (z = −7.59, p < .0001) in multi-variable meta-regression.
This gradient challenges older frameworks that positioned adolescent exposure as the primary window of vulnerability. A plausible biological interpretation is that cannabis interacts with age-related changes in dopaminergic tone or endocannabinoid system regulation — mechanisms that may amplify psychotic susceptibility differently across the lifespan. It is also possible that late-onset psychosis has distinct genetic architecture that is more sensitive to cannabinoid perturbation, or that older users consume higher-potency products over longer durations. The study cannot establish causality — it remains observational and cannot rule out confounding by predisposing psychopathology or concurrent substance use. Nonetheless, the scale (over 71,000 participants), the consistency across subgroups, and the dose-response age pattern collectively elevate this beyond routine confirmatory evidence. For clinicians and public health planners, the data suggest cannabis risk counseling should not be framed exclusively around youth.