Adolescent Cannabis Use and Psychiatric Risk: Commentary on the Largest Study Yet

Study Design and What Makes It Credible

Before discussing the findings, the methodology deserves attention, because this study is better designed than most in this literature.

The cohort — 463,396 adolescents aged 13 to 17 enrolled at Kaiser Permanente Northern California — was followed longitudinally from 2016 through 2023 or until age 25. Cannabis exposure was ascertained through universal, confidential behavioral health screening conducted as part of standard pediatric care visits. This is not retrospective self-report attached to a clinical outcome; it is prospective, time-varying exposure assessment embedded in routine care, which substantially reduces recall bias and allows for temporally accurate exposure classification.

The primary analyses used Cox proportional hazards regression with time-varying cannabis use as the exposure, incident psychiatric diagnoses as outcomes, and adjustment for sex, race/ethnicity, neighborhood deprivation index, insurance type, and time-varying alcohol and other substance use. Incident disorder was defined as a new diagnosis with no prior diagnosis in the relevant category during the lookback period — a critical methodological choice that prevents the results from reflecting symptom-driven cannabis use rather than cannabis-driven psychiatric onset.

The sample is diverse: 29.5% Hispanic, 20.2% non-Hispanic Asian, 7.6% non-Hispanic Black, 33.0% non-Hispanic White. The mean baseline age was 14.5 years. At baseline, 5.7% of adolescents reported past-year cannabis use — a realistic prevalence figure that aligns with national surveillance data for this age group.

This is large-scale, prospective, real-world data from a single integrated health system with longitudinal clinical records. It is not perfect — no observational study is — but it is among the best we have.

The Findings

The adjusted hazard ratios are worth examining carefully, not just as headline numbers:

Several features of these results deserve clinical attention.

The confidence intervals are tight. An adjusted HR of 2.19 (95% CI: 1.97–2.42) for psychosis reflects a precise, well-powered estimate. With over 460,000 adolescents, this study is not producing wide intervals from small-cell instability. The effect is real and is being measured accurately.

The hierarchy of risk reflects known biology. Psychosis and bipolar disorder — conditions with the strongest dopaminergic and endocannabinoid etiological links — show the highest hazard ratios. Depression and anxiety, which are more multifactorial and more responsive to a broader range of environmental inputs, show lower but still significant associations. This is exactly the pattern you would predict from the mechanistic literature.

The temporal precedence is the strongest causal argument available from observational data. Cannabis use preceded psychotic disorder diagnoses by approximately 2.3 years and depression diagnoses by approximately 1.7 years. This does not eliminate the possibility of prodromal self-medication — a genuinely difficult confound in this literature — but it substantially weakens reverse causation as the primary explanation.

Why the Adolescent Brain Is Specifically Vulnerable

The neurodevelopmental case for adolescent vulnerability is well-established and worth stating explicitly, because it explains why these risk estimates are higher in this age group than in adult cohorts.

The endocannabinoid system — comprising CB1 and CB2 receptors, endogenous ligands (anandamide, 2-AG), and degradative enzymes — plays a critical regulatory role in adolescent neurodevelopment. During the period from roughly ages 12 to 25, the brain undergoes active synaptic pruning, particularly in prefrontal and limbic regions, and progressive myelination of long-range cortico-cortical and cortico-subcortical tracts. The endocannabinoid system modulates both processes: it regulates the timing and extent of pruning and guides axonal growth and connectivity.

Exogenous THC binds CB1 receptors with approximately 10 to 100 times the affinity of anandamide. In an adolescent brain that is using endocannabinoid signaling as a developmental compass, sustained CB1 agonism from high-potency THC disrupts this calibration. The downstream effects include:

• Dopaminergic dysregulation in the mesolimbic pathway — THC acutely increases dopamine release in the nucleus accumbens and disrupts the tonic-phasic firing patterns that underlie reward prediction and salience attribution. In vulnerable individuals, this hyperdopaminergic state may precipitate or accelerate a psychotic process.
• Prefrontal-limbic connectivity disruption — Adolescent cannabis use is associated with reduced white matter integrity in fronto-limbic tracts and altered functional connectivity between prefrontal cortex and amygdala. These are exactly the circuits implicated in mood regulation and psychosis.
• Hippocampal effects — CB1 receptors are densely expressed in the hippocampus, a region critical for episodic memory and contextual processing. Heavy adolescent cannabis use is associated with reduced hippocampal volume, and hippocampal dysconnectivity is a consistent finding in both schizophrenia and bipolar disorder.
• HPA axis sensitization — Cannabis exposure during adolescence alters stress reactivity in ways that persist into adulthood, potentially increasing vulnerability to stress-induced psychiatric episodes.

The potency dimension compounds all of these effects. Average THC concentrations in retail cannabis products have risen from approximately 4% in 1995 to 15-25% in current concentrates and edibles. The adolescents in this 2016-2023 cohort were using products of substantially higher potency than those in earlier cohorts, which may explain why current risk estimates are at the higher end of the historical literature.

The ADHD-Cannabis Intersection

My NIH-funded research has focused specifically on the intersection of cannabis use and adolescent mental health, including the role of ADHD as a risk factor for cannabis initiation and escalation. This finding from Young-Wolff et al. has particular salience for adolescents with ADHD for several reasons.

Adolescents with ADHD initiate cannabis use earlier than neurotypical peers, use more frequently, and transition to cannabis use disorder at higher rates. The mechanisms are multiple: impulsivity lowers the threshold for substance initiation; reward dysregulation and low frustration tolerance make cannabis's acute anxiolytic and euphoric effects especially reinforcing; academic failure and social marginalization create incentives for self-medication. Several studies, including our own work with the CASNY cohort, have documented this elevated initiation risk.

The clinical implication is compounding. ADHD already elevates baseline risk for depression and anxiety. If cannabis initiation in adolescents with ADHD further multiplies psychiatric risk — a 2.19x hazard ratio for psychosis applied to an already elevated baseline risk — the cumulative psychiatric burden becomes substantial. The subset of adolescents with ADHD who use cannabis regularly represents a high-risk population that warrants specific clinical attention and targeted preventive messaging.

I have discussed this with adolescent patients and their parents directly: ADHD is not just a risk factor for academic and functional problems. In the context of cannabis access, it is a risk factor for a significantly more complicated psychiatric trajectory. That is a conversation worth having at every evaluation.

What "Association" Does and Does Not Mean

This study is observational. It cannot establish causation, and the authors are appropriately careful about this. The main threats to causal inference in this literature deserve honest acknowledgment.

Genetic confounding. The genes that increase risk for cannabis initiation and heavy use overlap with those that increase risk for psychotic and mood disorders. A study using siblings as controls would better isolate the cannabis-specific effect from shared genetic liability. This study does not use that design. Some portion of the observed associations reflects shared genetic architecture rather than a causal cannabis effect.

Prodromal self-medication. Individuals in the early, pre-diagnostic phase of a psychotic or bipolar illness may begin using cannabis in response to emerging symptoms before receiving a formal diagnosis. The 2.3-year temporal lead between cannabis use and psychotic disorder diagnosis substantially weakens but does not eliminate this explanation. Subclinical psychotic-like experiences can precede diagnosis by years.

Unmeasured confounding. The study adjusts for sex, race/ethnicity, neighborhood deprivation, insurance type, and concurrent alcohol and substance use. It cannot fully adjust for family psychiatric history, childhood adversity, peer environment, or individual-level psychological vulnerabilities — all of which influence both cannabis use initiation and psychiatric risk.

None of these limitations invalidate the findings. They qualify them. The combination of temporal precedence, biological plausibility, dose-response consistency across the prior literature, and effect sizes of this magnitude in the largest cohort yet studied supports a causal contribution of cannabis to psychiatric risk in a meaningful subset of adolescent users. The responsible clinical position is to treat this as a causal risk factor while acknowledging residual uncertainty — not to use observational design limitations as a reason to dismiss findings that are clinically and publicly important.

Clinical and Public Health Implications

Several things follow from this study for clinicians and for the families they advise.

Risk is not uniform. A family history of psychotic disorder or bipolar illness multiplies the cannabis-associated risk substantially. An adolescent with a first-degree relative with schizophrenia or bipolar I who is using cannabis is in a categorically higher-risk situation than the average adolescent user. This subgroup merits direct clinical counseling — not generic "cannabis is risky" messaging, but a specific and honest conversation about a quantifiable elevation in personal risk.

Earlier onset and higher frequency matter. The data consistently show dose-response and age-of-onset relationships. An adolescent who initiates at 13 with high-potency concentrates used daily is not in the same risk category as an adolescent who uses low-frequency, lower-potency cannabis at 17. Clinical messaging should be calibrated accordingly.

Legalization does not resolve this. Several states have legalized adult recreational cannabis use, and there is evidence that legalization has been accompanied by increased adolescent access and reduced perceived risk. The psychiatric risk data are not affected by the legal status of the product. Regulatory frameworks that reduce adolescent access — minimum age enforcement, limits on product potency, restrictions on advertising to minors — have a sound empirical basis.

Universal screening, as practiced in this study's health system, is the standard we should be working toward. The exposure data in this study came from confidential behavioral health screening at routine pediatric visits. That is not yet universal practice. Adolescents are not volunteering cannabis use information to clinicians they perceive as mandatory reporters. Screening protocols that are explicitly confidential, and that frame the conversation as health-focused rather than punitive, improve ascertainment and open the door for counseling.

In my own clinical practice with adolescents, I ask about cannabis use directly and frame the conversation around their specific risk profile: do they have ADHD, a family history of mood or psychotic disorders, a history of anxiety? The risk discussion is individualized, not generic. That is the model this evidence base supports.

Frequently Asked Questions

Does cannabis use cause psychiatric disorders in teenagers?

The 2026 JAMA Health Forum study establishes strong temporal association — cannabis use preceded psychiatric diagnoses by 1.7 to 2.3 years across disorder types — but cannot prove causation from observational data alone. That said, the combination of temporal precedence, biological plausibility, dose-response consistency, and a cohort of over 460,000 adolescents makes a causal contribution to risk highly probable, particularly for psychotic disorders in genetically vulnerable individuals.

How much does cannabis increase the risk of psychosis in teenagers?

In this study, the adjusted hazard ratio for psychotic disorders was 2.19 (95% CI: 1.97–2.42), meaning adolescents who used cannabis had approximately twice the risk of developing a psychotic disorder compared to non-users. The risk is higher with earlier onset, higher frequency, and higher-potency products, and substantially higher in those with a family history of psychotic or bipolar illness.

Why does cannabis increase psychiatric risk specifically in adolescents?

The adolescent brain is in a sensitive period of synaptic pruning and myelination, with the endocannabinoid system playing a critical regulatory role. THC disrupts this signaling at a period of maximum neuroplasticity, altering dopaminergic maturation, prefrontal-limbic connectivity, and hippocampal development in ways that are least reversible in adolescence.

Does ADHD increase the risk of cannabis use in adolescents?

Yes. Adolescents with ADHD initiate cannabis use earlier, use more frequently, and transition to cannabis use disorder at higher rates due to impulsivity, reward dysregulation, and self-medication of academic and emotional difficulties. This creates compounding risk: ADHD elevates initiation risk, and cannabis use then elevates psychiatric comorbidity risk beyond the already-elevated baseline in ADHD populations.

What should parents tell teenagers about cannabis and mental health?

The evidence supports a direct, risk-calibrated conversation: cannabis is not harmless for developing brains, and the psychiatric risks are real and quantifiable. The risk is highest for those who start young, use frequently, use high-potency products, or have a family history of psychotic or mood disorders. Delay of initiation and reduced frequency are associated with meaningfully lower risk.

Further Reading

• Cannabis and Mental Health — Overview of the research on cannabis, psychosis, mood disorders, and the adolescent brain
• Cannabis and Teen Mental Health: What Our Pediatrics 2026 Study Found — Our own cohort data on cannabis and adolescent academic and emotional outcomes
• Cannabis and Psychosis — Deep dive into the cannabis-psychosis relationship and who is most at risk
• Cannabis and the Teenage Brain — Neurodevelopmental mechanisms and the endocannabinoid system
• Child and Adolescent Psychiatry NYC — Clinical evaluation for adolescents with ADHD, depression, anxiety, and substance use
• ADHD Guide — ADHD and substance use risk in detail