Does Adolescent Cannabis Use Increase the Risk of Psychiatric Disorders?

Young-Wolff et al. (2026) examined whether adolescent cannabis use is associated with an increased risk of developing psychotic, bipolar, depressive, and anxiety disorders by young adulthood. They analyzed data from 463,396 adolescents ages 13–17 who received routine cannabis-use screenings within Kaiser Permanente Northern California between 2016 and 2023 and followed them through age 25. They found that adolescents who reported cannabis use had substantially higher risks of later psychiatric diagnoses. Cannabis use was associated with a 119% higher risk of psychotic disorders (AHR = 2.19), a 101% higher risk of bipolar disorder (AHR = 2.01), a 34% higher risk of depressive disorders (AHR = 1.34), and a 24% higher risk of anxiety disorders (AHR = 1.24).

Why This Article Was Selected for The Policy Scientist

Young-Wolff and her colleagues have produced a substantial body of research on adolescent substance use, making this study a natural extension of an established research agenda. The topic is particularly timely as cannabis legalization expands and adolescent exposure occurs within a rapidly changing policy environment. More broadly, understanding factors associated with the onset of serious mental health conditions is important because psychiatric disorders impose substantial long-term costs on individuals, families, health systems, and labor markets. The study makes a meaningful contribution by examining multiple clinically diagnosed psychiatric outcomes in a very large longitudinal cohort rather than relying on self-reported symptoms. The data set is a major strength, including more than 463,000 adolescents followed through young adulthood. The findings are likely generalizable to many U.S. settings with similar populations and health care access, although caution is warranted for uninsured populations. Methodologically, the longitudinal design and time-varying analyses strengthen the evidence, but the study remains observational rather than a causal inference design. Future research using stronger causal identification strategies would further clarify whether the observed associations reflect causal effects of adolescent cannabis use.

Full Citation and Link to Article

Young-Wolff, K. C., Cortez, C. A., Alexeeff, S. E., Silver, L. D., Pacula, R. L., Slama, N. E., Padon, A., Satre, D. D., Campbell, C. I., Koshy, M. T., Does, M. B., & Sterling, S. A. (2026). Adolescent cannabis use and risk of psychotic, bipolar, depressive, and anxiety disorders. JAMA Health Forum, 7(2), e256839. https://doi.org/10.1001/jamahealthforum.2025.6839

Central Research Question

Young-Wolff and colleagues sought to answer a straightforward but important question: Is adolescent cannabis use associated with an increased risk of developing clinically diagnosed psychotic, bipolar, depressive, and anxiety disorders during adolescence and young adulthood? While previous studies have linked cannabis use to psychiatric symptoms and mental health problems, much of that literature has focused on relatively small samples, cannabis use disorder rather than general cannabis use, or self-reported symptoms rather than clinician-diagnosed conditions. The authors aimed to determine whether adolescents who report cannabis use face a higher risk of receiving formal psychiatric diagnoses before age 26.

The study is particularly relevant because cannabis use often begins during adolescence, a developmental period in which many psychiatric disorders first emerge. At the same time, cannabis has become increasingly accessible and socially accepted throughout the United States. The authors therefore sought to provide stronger evidence regarding the relationship between cannabis use and subsequent mental health outcomes by examining a large population of adolescents over time.

Previous Literature

The authors situate their study within a substantial body of research examining cannabis use and mental health. The strongest prior evidence has focused on psychotic disorders. Several longitudinal studies have found that cannabis use during adolescence, especially frequent use or use of high-potency products, is associated with an elevated risk of schizophrenia and other psychotic disorders. However, not all studies have reached the same conclusions, leaving uncertainty regarding the magnitude and consistency of the relationship.

Research examining bipolar disorder has also suggested a positive association. Previous studies have found that individuals with cannabis use disorder are more likely to receive bipolar disorder diagnoses, and a recent meta-analysis reported substantially higher odds of bipolar disorder among cannabis users. Evidence concerning depression has generally indicated a modest but statistically significant relationship between adolescent cannabis use and later depressive symptoms or diagnoses.

The literature regarding anxiety disorders has been less consistent. Some studies have identified positive associations, while others have found little or no relationship. Moreover, much of the existing research has focused on either a single psychiatric outcome or relatively narrow populations. Few studies have examined multiple psychiatric disorders simultaneously within a large, diverse, longitudinal cohort.

The authors therefore sought to address several limitations in the literature by examining four major psychiatric outcomes, using clinician-diagnosed conditions rather than self-reported symptoms, and following a very large sample of adolescents over time. In doing so, they build upon earlier research while providing evidence from one of the largest studies conducted on this topic.

Data

The study used data from Kaiser Permanente Northern California, an integrated health care system serving more than 4.6 million members. The sample consisted of 463,396 adolescents between the ages of 13 and 17 who completed a confidential health screening questionnaire during routine pediatric visits between 2016 and 2023.

The screening instrument included a question asking whether the adolescent had used marijuana during the previous year. The survey was administered as part of standard pediatric care and was completed confidentially by the adolescents themselves. This design likely improved the accuracy of reporting compared with situations in which parents are present.

The sample was large and demographically diverse. Approximately 29.5% of participants were Hispanic, 20.2% were Asian, 7.6% were Black, and 33.0% were White. The mean age at baseline was 14.5 years. Approximately 5.7% of adolescents reported cannabis use during the previous year.

The study linked survey responses to electronic health records. These records provided information on subsequent diagnoses of psychotic disorders, bipolar disorder, depressive disorders, and anxiety disorders. The authors also collected information on demographic characteristics, neighborhood socioeconomic conditions, insurance status, alcohol use, and use of other substances.

One of the greatest strengths of the data is the combination of universal screening and longitudinal follow-up. Rather than relying on retrospective recall, the study measured cannabis use prospectively and linked those reports to future clinical diagnoses. The resulting dataset provides unusually strong observational evidence regarding adolescent cannabis use and mental health outcomes.

Methods

The authors employed a retrospective cohort design and followed participants from their initial screening date until the occurrence of a psychiatric diagnosis, disenrollment from the health system, death, or the end of the study period. Participants were followed through a maximum age of 25 years.

The primary analytical approach consisted of extended Cox proportional hazards regression models. These models estimated the association between cannabis use and the likelihood of subsequently receiving a psychiatric diagnosis. Cannabis use was treated as a time-varying measure, allowing the authors to update exposure information when adolescents completed additional screenings over time.

The statistical models adjusted for several potential confounding variables, including sex, race and ethnicity, neighborhood deprivation, insurance status, alcohol use, and use of other substances. The authors also conducted multiple sensitivity analyses. These analyses adjusted for preexisting psychiatric conditions, excluded adolescents with prior psychiatric diagnoses, and tested alternative definitions of psychotic disorders.

The methodology strengthens confidence in the findings because it establishes temporal ordering. Cannabis use generally preceded psychiatric diagnoses by approximately 1.7 to 2.3 years. Nevertheless, the study remains observational rather than experimental. Although the authors controlled for many potential confounders, they cannot definitively establish causality. Unmeasured factors such as family history, adverse childhood experiences, or genetic predispositions may influence both cannabis use and mental health outcomes. Future studies employing stronger causal inference strategies would further clarify these relationships.

Findings/Size Effects

The study found statistically significant associations between adolescent cannabis use and all four psychiatric outcomes examined. The strongest relationships were observed for psychotic and bipolar disorders.

Adolescents reporting cannabis use experienced a 119% higher risk of developing a psychotic disorder compared with nonusers, corresponding to an adjusted hazard ratio of 2.19. Cannabis users also experienced a 101% higher risk of bipolar disorder, with an adjusted hazard ratio of 2.01.

The associations were smaller but still substantial for depressive and anxiety disorders. Cannabis use was associated with a 34% higher risk of depressive disorders and a 24% higher risk of anxiety disorders. These effects remained statistically significant after extensive adjustment for demographic characteristics and other substance use.

Additional analyses revealed that the associations with depression and anxiety were strongest during adolescence and weakened with age. By ages 21 to 25, the relationships were no longer statistically significant. In contrast, the elevated risks for psychotic and bipolar disorders remained comparatively strong throughout the observation period.

Sensitivity analyses produced similar results. After controlling for previous psychiatric conditions and excluding adolescents with prior psychiatric diagnoses, cannabis use remained significantly associated with increased risk across all outcomes. These findings suggest that the observed relationships are not solely explained by preexisting mental health problems.

Overall, the evidence indicates that adolescent cannabis use is consistently associated with increased risk of subsequent psychiatric diagnoses, particularly psychotic and bipolar disorders.

Conclusion

This study represents one of the largest longitudinal investigations of adolescent cannabis use and psychiatric disorders conducted to date. Using data from more than 463,000 adolescents, the authors found that cannabis use during adolescence was associated with increased risk of psychotic, bipolar, depressive, and anxiety disorders through young adulthood.

The strongest associations were observed for psychotic and bipolar disorders, where cannabis users experienced approximately double the risk of subsequent diagnosis. More modest but meaningful increases were observed for depression and anxiety. The findings remained robust across multiple sensitivity analyses and after adjustment for numerous potential confounders.

The study contributes to the literature by examining multiple psychiatric outcomes simultaneously, using clinician-diagnosed disorders rather than self-reported symptoms, and employing prospective longitudinal data. Although the observational design prevents definitive causal conclusions, the temporal ordering and extensive robustness checks strengthen the evidence that adolescent cannabis use is associated with elevated psychiatric risk. Future research employing causal inference designs, natural experiments, or other stronger identification strategies would help determine the extent to which these associations reflect causal effects.