Published summary: 2026-03-07
Where the evidence converges
Across 22 longitudinal cohorts, cannabis use was linked to about 29 percent higher odds of later depression, with exposure measures often weak and overall bias rated medium
Strong evidence · 25 studies
Observational evidence only. Even with baseline adjustment, unmeasured confounding can explain part of the signal. Cannabis exposure was inconsistently measured, often as any use versus none, with lit
A meta-analysis found cannabis use associated with 2
Strong evidence · 25 studies
The authors acknowledged lack of homogeneity in how cannabis exposure was measured across studies, and in some cases insufficient control for known suicide risk factors (depression, other substance us
Largest cannabis GWAS (184,765 people) found 8 significant variants, CADM2 as the top gene, 11% heritability, and Mendelian randomization evidence that schizophrenia risk causally increases cannabis use
Strong evidence · 25 studies
European-ancestry cohorts only. Lifetime cannabis use is a binary measure that does not capture dose, frequency, or recency. Mendelian randomization assumptions may be violated. 11% variance explained
Where questions remain
Cannabis use before age 15 was associated with 2
Moderate evidence · 158 studies
Observational cohort study cannot fully establish causation even with extensive confounder adjustment. Some confounders may be unmeasured. The study covered outcomes only through age 15-16, missing lo
Among drug treatment patients, 64% had alcohol dependence, 44% antisocial personality, and 24% depression
Moderate evidence · 158 studies
Treatment-seeking populations may not represent all drug-dependent individuals. Race was used as a category, which may oversimplify complex social and cultural factors. DSM-III-R criteria are now outd
In 232 first-episode schizophrenia patients, cannabis preceded psychosis in one-third, coincided with onset in another third, and began after symptoms in the final third, suggesting multiple pathways
Moderate evidence · 158 studies
Retrospective timing of symptom and drug use onset is subject to recall bias. The three groups are approximate and the boundaries between them may be artificial. The sample included only those who dev
In 50 bipolar patients, alcohol abuse correlated with time spent depressed while cannabis abuse correlated with time in mania, suggesting each substance interacts differently with bipolar mood states
Moderate evidence · 158 studies
Small sample (50 patients). Correlation does not establish causation. Cannabis could trigger mania, or manic episodes could increase cannabis use, or both. The subgroup analysis was exploratory and sa
Research gaps
- Only 37 randomized controlled trials exist out of 840 studies — most evidence is observational or from reviews.
- Long-term prospective studies tracking outcomes over 5+ years are largely absent from the literature.
- Research on diverse populations (different ages, ethnicities, and medical backgrounds) remains limited.
Key studies
About One-Third of People With Cannabis-Induced Psychosis Later Develop Schizophrenia
Distinguishing cannabis-induced psychosis from the onset of a primary psychotic disorder is one of the most consequential diagnostic challenges in psychiatry. This meta-analysis quantifies the stakes: about a third of cannabis-induced psychosis cases may actually be the first presentation of schizop
Psychological interventions showed minimal benefit for people with both schizophrenia and substance use disorders
About 42% of people with schizophrenia have co-occurring substance use disorders, yet they are frequently excluded from clinical trials. This comprehensive meta-analysis reveals that current evidence-based psychological treatments are essentially ineffective for this doubly burdened population, high
Meta-analysis: youth cannabis use linked to 28-87% higher odds of depression, anxiety, and suicidal behavior
Young people are the largest cannabis consumer demographic and also the population most vulnerable to mental health disruptions. This meta-analysis quantifies the magnitude of the mental health risks associated with cannabis use during this critical developmental period.
Meta-Analysis Confirms Cannabis Can Cause Paranoid Symptoms
Paranoia is common and has significant social consequences. This meta-analysis provides converging evidence from both experimental (causal) and observational designs that cannabis, particularly THC-dominant products, can cause paranoid symptoms.
Cannabis Use Was Linked to Suicidal Behavior Even After Accounting for Depression, Though Depression Partly Explains the Connection
The cannabis-suicide link has been debated, with critics arguing depression explains it entirely. This meta-analysis shows the association persists even after accounting for depression in many studies, while also acknowledging that depression does partly confound the relationship.
Recreational Cannabis Use Was Linked to Worse Sleep Across 120 Studies
Many people use cannabis specifically to improve sleep, yet this large-scale synthesis of the general population evidence suggests recreational use is actually associated with worse sleep outcomes. The disconnect between observational and experimental findings highlights the complexity of the cannab
How the research developed
Pre-2000
2 studies published. Predominantly observational and review studies.
2000–2009
34 studies published. Includes 2 RCTs, 4 strong-evidence studies.
2010–2014
63 studies published. Includes 1 meta-analyses, 4 RCTs, 6 strong-evidence studies.
2015–2019
139 studies published. Includes 4 meta-analyses, 2 RCTs, 21 strong-evidence studies.
2020–present
602 studies published. Includes 16 meta-analyses, 29 RCTs, 113 strong-evidence studies.
How this summary was assembled
This consensus synthesizes 840 peer-reviewed studies: 21 meta-analyses (3%), 37 randomized controlled trials (4%), 367 observational studies (44%), 122 reviews (15%), 28 case studies (3%), 265 other study types (32%). Studies span from the earliest available research through 2025. Evidence strength ratings reflect study design, sample size, and replication across multiple research groups.
These counts describe the source summary at publication. They may differ from the current library.