| Audience | Clinicians, parents, caregivers, educators, policy readers, and lay readers trying to interpret youth cannabis risk carefully |
| Primary Topic | Adolescent cannabis use and later risk of psychotic, bipolar, depressive, and anxiety diagnoses |
| Journal | JAMA Health Forum |
| Study Design | Retrospective cohort study using electronic health record data and time-varying exposure modeling |
| Source | Read the full article |
Adolescent Cannabis Use and Psychiatric Risk, What This Large Study Really Shows, and What It Still Cannot Prove
This large cohort study found that adolescents who reported past-year cannabis use were more likely to later receive diagnoses of psychotic, bipolar, depressive, and anxiety disorders. That makes the paper clinically important. It also makes restraint important, because the study is strongest as evidence of association and warning, not as final proof that cannabis itself directly caused each later diagnosis.
This study teaches that adolescent cannabis use should not be treated as a casual background detail when evaluating young people. In more than 463,000 adolescents screened during routine pediatric care, past-year cannabis use was associated with higher subsequent rates of psychotic, bipolar, depressive, and anxiety diagnoses. The strongest associations were for psychotic and bipolar disorders. For clinicians, that means a teenager reporting cannabis use deserves more careful psychiatric review, not just a brief warning about substances. For families and lay readers, it means youth cannabis exposure belongs in real conversations about vulnerability, development, family history, and emerging symptoms.
This paper matters because discussions about adolescent cannabis often become cartoonish. One side minimizes it as basically harmless. The other treats it as a single-step explanation for severe psychiatric illness. This study supports neither extreme. What it does show is that in a very large real-world pediatric population, adolescent cannabis use was linked with meaningfully higher later psychiatric diagnosis rates, especially for psychotic and bipolar disorders. That is enough to matter in pediatric practice, school health, family counseling, and public health messaging.
| Study Type | Retrospective cohort study |
| Population | 463,396 adolescents aged 13 to 17 years in Kaiser Permanente Northern California |
| Exposure | Self-reported past-year marijuana use during confidential routine pediatric screening, modeled as a time-varying exposure |
| Comparator | Adolescents not reporting past-year cannabis use |
| Primary Outcomes | Incident clinician-diagnosed psychotic, bipolar, depressive, and anxiety disorders |
| Main Results | Adjusted hazard ratios: psychotic disorder 2.19, bipolar disorder 2.01, depressive disorder 1.34, anxiety disorder 1.24 |
| Baseline Use | 5.7% of the cohort reported past-year cannabis use at baseline |
| Year | 2026 |
| DOI | 10.1001/jamahealthforum.2025.6839 |
| Key Limitation | No dose, frequency, potency, route, age of initiation, or product-composition detail |
This is an important association study and a useful counseling paper. It supports taking adolescent cannabis use seriously, especially in youth with psychiatric symptoms or strong family vulnerability. It does not prove that cannabis alone caused later psychiatric diagnoses, and it should not be used as a shortcut around careful clinical thinking.
The investigators used universal confidential adolescent screening embedded in routine pediatric care to ask whether self-reported past-year cannabis use was associated with later clinician-diagnosed psychotic, bipolar, depressive, and anxiety disorders. They followed adolescents through age 25 years or the end of 2023 and modeled cannabis use as a time-varying exposure, which is stronger than relying only on a single baseline snapshot. The models adjusted for sex, race and ethnicity, neighborhood deprivation, insurance type, and time-varying alcohol and other substance use. Sensitivity analyses further adjusted for baseline psychiatric conditions and also examined models that excluded adolescents with psychiatric histories at baseline.
Past-year cannabis use was associated with increased risk across all four psychiatric outcomes studied. The clearest relative associations were for psychotic disorder and bipolar disorder, with adjusted hazard ratios of 2.19 and 2.01. The associations for depressive and anxiety disorders were smaller, and both weakened with age. For depressive disorder, the association was strongest at ages 13 to 15 years and no longer statistically significant at ages 21 to 25 years. A similar age-related weakening was seen for anxiety disorder. Sensitivity analyses attenuated the findings but did not erase the overall signal.
For an observational study, the evidence is fairly strong. The sample is very large, the data come from routine care rather than a narrow specialty sample, and the longitudinal design with time-varying exposure modeling improves clinical relevance. Still, it remains observational evidence. That means it is well suited to identifying real-world association and warning signals, but weaker for proving biological direction, isolating causality, or telling us exactly which use patterns or products are driving the risk.
The most important limitation is confounding by vulnerability. Adolescents who use cannabis are not randomly drawn from the population. They may differ in family psychiatric history, trauma exposure, peer environment, temperament, sleep disruption, early subthreshold symptoms, or other factors that also raise later psychiatric risk. The investigators adjusted for several important variables, but no observational model can fully remove those background differences. Reverse causation also remains plausible. Some teens may have begun using cannabis in response to already-emerging anxiety, low mood, sleep trouble, emotional volatility, or subtle psychotic experiences before those symptoms were formally diagnosed.
The exposure measure is also blunt. A yes-or-no question about any past-year marijuana use collapses together very different clinical realities, from experimental use to frequent use of high-THC products. Without detailed information on dose, frequency, potency, route, age of onset, or THC-to-CBD balance, the study cannot tell us whether the observed risk is broadly distributed across all adolescent users or concentrated in heavier-use, earlier-use, or higher-potency subgroups.
Outcome measurement deserves caution too. Diagnoses came from routine electronic health record coding rather than structured research interviews. That makes the paper clinically grounded, but less diagnostically precise than a dedicated psychiatric assessment protocol. The cohort also came from one insured Northern California health system, which may limit how confidently the results generalize to adolescents without regular care or to regions with different market, policy, or social conditions.
This paper does not show that cannabis inevitably causes psychosis, bipolar disorder, depression, or anxiety in adolescents. It does not show that every cannabis product carries the same psychiatric risk, and it does not distinguish occasional lower-intensity use from frequent high-potency use. It also does not answer whether some adolescents were self-medicating already-emerging symptoms, or whether the strongest signal came from a smaller subgroup with unusually high exposure or unusually high vulnerability.
Adolescent cannabis use appears to be associated with higher later risk of several psychiatric diagnoses, with the clearest signals here involving psychotic and bipolar disorders. That is enough to justify concern, screening, and prevention-oriented counseling. What this study does not do is settle causality. A careful reader should come away understanding both halves of the story at once: the signal matters, and the interpretive limits matter too.
? Join the Conversation
How should clinicians and families talk about adolescent cannabis risk without exaggerating the science or minimizing the concern?
Have thoughts on this? Share it:
Frequently Asked Questions
Does this study prove cannabis causes psychosis in teens?
No. It shows a strong association, not definitive causation.
Which psychiatric outcomes had the strongest associations?
Psychotic and bipolar disorders.
Did the study measure how much cannabis adolescents used?
No. The exposure was any self-reported past-year use, not dose or frequency.
Did the paper distinguish product potency or THC versus CBD content?
No. Product composition was not captured in that level of detail.
Could some adolescents have been using cannabis because symptoms were already emerging?
Yes. Reverse causation remains a reasonable concern.
Were diagnoses based on structured psychiatric interviews?
No. They were based on clinician-coded diagnoses in the electronic health record.
Did depression and anxiety findings stay equally strong across age?
No. Those associations weakened with age and were no longer statistically significant at ages 21 to 25 years.
What is the most practical clinical takeaway?
Screen early, ask better psychiatric questions, and treat adolescent cannabis use as clinically meaningful.
Does this paper apply equally to all cannabis products and all adolescents?
No. Individual vulnerability and product characteristics likely matter, but the study could not sort that out in detail.
What kind of future study would improve confidence?
Prospective work with repeated psychiatric assessment and detailed exposure measures, including frequency, potency, route, age of initiation, and product composition.
Read next
