Cannabis and Schizophrenia: Meta-Analysis Finds Roughly Threefold Risk Elevation, but Causation Remains Contested
Cannabis and Schizophrenia: Meta-Analysis Finds Roughly Threefold Risk Elevation, but Causation Remains Contested
A systematic review of 18 studies pools an odds ratio of 2.88, applies Hill causality criteria, and concludes cannabis “likely contributes” to schizophrenia, though confounding, reverse causation, and publication bias cannot be excluded from the observational evidence base.
Why This Matters
With an estimated 228 million cannabis users worldwide and legalization expanding across jurisdictions, the relationship between cannabis use and serious psychiatric illness has become one of the most consequential questions in public health. Schizophrenia, which typically emerges during adolescence and young adulthood, overlaps precisely with peak cannabis initiation. The endocannabinoid system’s known role in neurodevelopmental signaling makes the biological plausibility of this link scientifically credible. As policy decisions increasingly outpace the evidence base, research that attempts to move beyond correlation toward causal inference carries particular urgency.
Clinical Summary
The relationship between cannabis exposure and psychotic disorders has generated decades of epidemiological data, but the question of causality remains unresolved. This systematic review and meta-analysis, published in Biomolecules in 2025, searched seven databases through July 2024 and identified 18 observational studies meeting inclusion criteria. Ten of those contributed to a quantitative pooled analysis. The mechanistic rationale centers on delta-9-tetrahydrocannabinol (THC) and its interaction with the endocannabinoid system, particularly CB1 receptors in the prefrontal cortex and hippocampus during critical neurodevelopmental windows. THC-driven dopaminergic dysregulation in the mesolimbic pathway is hypothesized to precipitate psychotic symptoms in genetically vulnerable individuals, with adolescent exposure amplifying this risk given ongoing synaptic pruning and myelination.
The pooled odds ratio across 10 studies was 2.88 (95% CI 2.24 to 3.70), with nine of ten individual studies reaching statistical significance. All 18 included studies, regardless of forest plot eligibility, reported an elevated risk. Adolescent use was associated with approximately double the risk relative to the overall pooled estimate, and subgroup data suggested greater risk with higher THC potency, more frequent use, and male sex. However, funnel plot asymmetry indicated likely publication bias or residual confounding. The authors applied Bradford Hill causality criteria qualitatively and judged the relationship “likely contributory,” but acknowledged that reverse causation, in which prodromal psychotic symptoms drive cannabis self-medication, and shared genetic vulnerability between cannabis use disorder and schizophrenia remain plausible alternative explanations. They called for further research, particularly Mendelian randomization and longitudinal studies, to resolve the causal question.
Dr. Caplan’s Take
This review aggregates a signal that has been consistent for years: cannabis use is associated with a meaningfully elevated risk of psychotic disorders, and the association is stronger when use begins in adolescence and involves high-potency products. The roughly threefold odds ratio is clinically significant. What the study gets right is its effort to systematically apply Hill criteria to this literature, which forces a more rigorous conversation about causation than most narrative discussions allow. But the gap between “likely contributory” and “proven causal” is exactly where the hardest clinical conversations happen. When a patient or parent asks me whether cannabis causes schizophrenia, the honest answer is that we have strong, consistent observational evidence of elevated risk, but we cannot yet separate the contribution of cannabis from the contribution of genetic predisposition and early prodromal states.
In practice, I counsel patients with any family history of psychotic disorders, particularly adolescents and young adults, that the available evidence supports treating cannabis use as a meaningful and potentially modifiable risk factor. I do not frame this as proven causation, but I do frame it as a risk that is serious enough to warrant avoidance, especially of high-THC products. For patients already using cannabis who present with subclinical psychotic symptoms, deprescribing or cessation is a priority within the treatment plan.
Clinical Perspective
This meta-analysis sits at a mature but still fundamentally unresolved stage of the research arc. It confirms what multiple prior systematic reviews, including the landmark meta-analysis by Marconi and colleagues in 2016, have reported: a dose-dependent association between cannabis use and psychotic outcomes that is consistent across populations and study designs. What it adds is the explicit application of Hill criteria, which formalizes the causal reasoning rather than leaving it implicit. However, it does not resolve the core inferential limitation of observational data, and the detected funnel plot asymmetry warrants caution about the precision of the 2.88 estimate. The evidence supports informing patients of a robust and consistent association but does not support stating that cannabis definitively causes schizophrenia in otherwise low-risk individuals.
From a pharmacological standpoint, clinicians should be aware that the risk signal is specifically tied to THC rather than to cannabidiol (CBD), which has shown some antipsychotic properties in early-phase trials. This distinction matters in an era of widely variable product potency. Patients taking antipsychotic medications who continue cannabis use may experience pharmacodynamic antagonism at the dopaminergic level, complicating treatment response. One actionable step clinicians can implement now is routine screening for cannabis use in all patients presenting with first-episode psychosis or attenuated psychosis symptoms, with documentation of product type, potency, frequency, and age of onset of use to support individualized risk counseling.
Study at a Glance
- Study Type
- Systematic review with meta-analysis (PRISMA)
- Population
- Cannabis users and non-users across 18 observational studies, including adolescent and adult subgroups
- Intervention
- Cannabis exposure (varying frequency, potency, and age of onset)
- Comparator
- Non-cannabis users or minimal-use controls
- Primary Outcomes
- Schizophrenia diagnosis or psychosis-like events (ICD/DSM criteria)
- Sample Size
- 18 studies included; 10 contributed to quantitative pooling
- Journal
- Biomolecules
- Year
- 2025
- DOI or PMID
- Not available from provided text
- Funding Source
- Not reported in available text
What Kind of Evidence Is This
This is a systematic review with meta-analysis conducted under PRISMA guidelines, combining quantitative pooling of observational studies with a qualitative application of Bradford Hill causality criteria. It sits near the top of the evidence hierarchy for observational synthesis, but its inferential ceiling is fundamentally constrained by the observational nature of all included studies. No randomized controlled trial can ethically assign cannabis exposure to test psychosis outcomes, which means this design represents the strongest feasible synthesis but cannot establish definitive causation. The most important inference constraint is that confounding by shared genetic liability and reverse causation remain statistically indistinguishable from a true causal effect within this framework.
How This Fits With the Broader Literature
This review aligns with and extends the findings of several major prior syntheses. The Marconi et al. (2016) meta-analysis in Schizophrenia Bulletin similarly reported a dose-response relationship between cannabis use and psychosis risk, and the Di Forti et al. (2019) study across 11 European sites found that daily use of high-potency cannabis was associated with fivefold odds of first-episode psychosis. The current review’s pooled OR of 2.88 is broadly consistent with these estimates. What this paper adds is the systematic application of Hill criteria, an approach more commonly seen in environmental epidemiology than psychiatric research. However, emerging Mendelian randomization studies have produced mixed results on whether the genetic instruments for cannabis use predict psychosis independently, meaning the causal question remains genuinely open. This review does not challenge existing findings so much as it underscores the persistent gap between strong observational association and confirmed causation.
Common Misreadings
The most likely overinterpretation is reading the authors’ conclusion that cannabis “likely contributes” to schizophrenia as equivalent to established causation. An odds ratio of 2.88 from observational data, even when supported by a qualitative application of Hill criteria, does not constitute proof of cause. Hill criteria are an interpretive framework, not a statistical test, and their application is inherently subjective. The funnel plot asymmetry the authors themselves report suggests the true effect size may be smaller than 2.88 once publication bias and confounding are accounted for. Additionally, the finding that adolescent use carries approximately double the risk should not be interpreted as indicating that adult use is safe. Absence of a strong signal in one subgroup is not evidence of absence of risk.
Bottom Line
This meta-analysis reinforces a consistent and clinically important association between cannabis use and schizophrenia risk, with a pooled odds ratio of 2.88 and particular vulnerability during adolescence. It does not, however, establish definitive causation, and detected publication bias suggests the true magnitude of risk may be somewhat lower. For practice now, the evidence supports proactive risk counseling, especially for adolescents, those with family history of psychotic disorders, and users of high-potency THC products, while honestly acknowledging that the causal picture remains incomplete.
References
- Marconi A, Di Forti M, Lewis CM, Murray RM, Vassos E. Meta-analysis of the association between the level of cannabis use and risk of psychosis. Schizophrenia Bulletin. 2016;42(5):1262-1269. doi:10.1093/schbul/sbw003
- Di Forti M, Quattrone D, Freeman TP, et al. The contribution of cannabis use to variation in the incidence of psychotic disorder across Europe (EU-GEI): a multicentre case-control study. The Lancet Psychiatry. 2019;6(5):427-436. doi:10.1016/S2215-0366(19)30048-3
- Hill AB. The environment and disease: association or causation? Proceedings of the Royal Society of Medicine. 1965;58(5):295-300. PMID:14283879
Have thoughts on this? Share it:
