A study of 463,396 adolescents , the largest of its kind to examine the link between teenage cannabis use and clinically diagnosed mental illness , has found that any past-year cannabis use at ages 13 to 17 is associated with roughly double the risk of developing psychotic disorders and bipolar disorder by age 26.
The study, published February 20 in JAMA Health Forum, used electronic health records from Kaiser Permanente Northern California, spanning the years 2016 to 2023. Researchers led by Kelly C. Young-Wolff at the Kaiser Permanente Division of Research tracked adolescents who reported any cannabis use during their routine well-child checkups and compared their rates of later psychiatric diagnosis to those who reported no use.
The results, adjusted for demographics, neighborhood deprivation, insurance type, and concurrent alcohol and other substance use, showed adjusted hazard ratios of 2.19 (95% confidence interval 1.97 to 2.42) for psychotic disorders and 2.01 (95% CI 1.82 to 2.22) for bipolar disorder , meaning roughly double the risk for each.
Associations with depressive disorders (AHR 1.34, 95% CI 1.30 to 1.39) and anxiety disorders (AHR 1.24, 95% CI 1.21 to 1.28) were more modest.
A durable signal for severe disorders
The effect for depression and anxiety was not uniform across all ages. When the researchers stratified by age, they found that the elevated risk for depressive disorders declined with age: from an AHR of 1.78 at ages 13 to 15 to 0.97 (not statistically significant) by ages 21 to 25.
In contrast, the elevated risk for psychotic and bipolar disorders persisted across all age groups, suggesting what the authors describe as “a more durable vulnerability” for severe mental illness in cannabis-using adolescents.
The mean time between cannabis use report and psychiatric diagnosis was 2.0 years for psychotic disorders, 2.3 years for bipolar disorder, 1.7 years for depressive disorders, and 1.9 years for anxiety disorders.
Controlling for confounders
The study accounted for sex, race and ethnicity, neighborhood deprivation index, Medicaid status, and time-varying alcohol and other substance use. A sensitivity analysis that additionally adjusted for prior psychiatric conditions attenuated the associations slightly , psychotic disorders dropped to AHR 1.92 and bipolar to AHR 1.73 , but they remained highly statistically significant.
The researchers also calculated E-values, a measure of how strong an unmeasured confounder would need to be to nullify the observed associations. For psychotic disorders, the E-value was 3.79 , meaning an unmeasured confounder would need to be associated with both cannabis use and psychotic disorders by nearly a factor of four to explain away the finding. The authors note this makes the association “relatively robust to unmeasured confounding.”
Caveats and limitations
The study carries important caveats. Cannabis use was self-reported in a clinical screening questionnaire and may underestimate true use, particularly in a healthcare setting where disclosure could have implications. The study population consists of insured adolescents in a single California health system, and results may not generalize to uninsured populations or regions with different cannabis policies.
The study design, while large and well-controlled, cannot fully exclude reverse causation , the possibility that adolescents with prodromal psychiatric symptoms may be more likely to use cannabis as self-medication. The 1.7- to 2.3-year lead time between reported use and diagnosis partly addresses this concern, but does not eliminate it.
Crucially, the study measured only binary any-versus-none cannabis use. It does not capture frequency of use, dose, potency, method of consumption, or whether use escalated to cannabis use disorder , all factors that could significantly moderate risk.
Policy implications
The study arrives amid an ongoing debate about legalization age thresholds and public health messaging regarding adolescent cannabis use. The authors explicitly call for potency restrictions, marketing limits, youth access protections, and clear mental health warnings as part of cannabis regulatory frameworks.
The association between cannabis and psychotic disorders has been documented in multiple prior studies, including meta-analyses by Marconi et al. (2016) and the large-scale EU-GEI study led by Di Forti et al. (2019). The current study extends this evidence to bipolar disorder, a less thoroughly investigated outcome, while providing the largest sample to date , 463,000 adolescents , and a cleaner comparison because it examines any use, not only heavy use or diagnosed cannabis use disorder.
The study is published in JAMA Health Forum, Volume 7, Issue 2, Article e256839 (DOI: 10.1001/jamahealthforum.2025.6839). It was funded by the National Institute on Drug Abuse (grant R01DA0531920). Lead author Kelly C. Young-Wolff is affiliated with the Kaiser Permanente Division of Research; co-authors include researchers from Kaiser Permanente, the Public Health Institute, UCSF, and USC.
Source: JAMA Health Forum and ScienceDaily.