Cannabis policies are becoming increasingly liberal worldwide, partly driven by interest in the potential therapeutic effects of cannabinoids for mental health conditions and substance use disorders (SUDs). This shift has been accompanied by a growing number of individuals reporting medicinal cannabis use, as well as rising prescription rates of medical cannabinoids, particularly for mental health symptoms. This trend is prevalent in countries such as Australia, the United States, and Canada, despite still limited evidence for their efficacy.
Randomised controlled trials (RCTs) remain the gold standard for evaluating treatment efficacy; however, RCTs investigating cannabinoids for mental health and SUDs are highly heterogeneous and yield mixed findings, complicating evidence synthesis. To address this, the current article by Wilson and colleagues (published in The Lancet Psychiatry on 16th March 2026) systematically reviews and meta-analyses RCTs examining cannabinoids as treatments for mental health conditions and SUDs, which is an important effort to consolidate and extend the existing evidence base.
Prescription cannabinoids for mental health symptoms are increasing despite uncertainties about their efficacy.
Methods
The authors searched multiple databases (1980–2025) for RCTs in all languages on plant-based and pharmaceutical cannabinoids as a treatment for mental disorders and SUDs. Only clinical RCTs were eligible. Reviewers independently screened, selected, and extracted data, resolving disagreements via consensus. Eligible studies assessed the efficacy and safety of the cannabinoids, and mental health outcomes such as remission, symptoms, functioning, and adverse events.
Using validated tools, the risk of bias assessment (the Cochrane risk of bias tool 2.0) and evidence grading were conducted (GRADE framework). The meta-analyses used random-effects models, reporting standardised mean differences and odds ratios. Subgroup, sensitivity, and heterogeneity analyses were also conducted.
Results
- RCTs included: 54 (from 5,774 screened)
- Participants: 2,477 (median n = 32 per study)
- High risk of bias: 44% of included studies; most evidence rated low certainty
- Adverse events: NNTH = 7 (one extra harm for every seven people treated with cannabinoids)
| Condition | RCTs | Cannabinoid(s) | Key findings | Verdict |
|---|---|---|---|---|
| Cannabis use disorder | 12 | THC+CBD (nabiximols), THC, CBD | THC+CBD reduced withdrawal symptoms and weekly cannabis use vs placebo. No significant effect on craving, abstinence, or cannabis-related problems. Withdrawal finding lost significance after removing high-bias studies. GRADE: very low to low certainty. | Mixed / limited |
| Psychotic disorders | 8 | CBD (predominantly), THC | No significant effect on PANSS total, positive, negative, or general symptom scores. No difference in adverse events or withdrawals. | No significant effect |
| Anxiety disorders | 6 | CBD (predominantly), THC | No significant effect on anxiety symptoms at longest follow-up. No difference in adverse events or withdrawals. Includes social anxiety disorder (3 studies) and generalised anxiety (3 studies). | No significant effect |
| Tic or Tourette’s syndrome | 5 | THC+CBD, THC, CBD | THC+CBD significantly reduced tic severity vs placebo. No effect from CBD or THC alone. No effect on premonitory urges. Significantly increased odds of adverse events (OR 4.93). GRADE: very low certainty. | Mixed / limited |
| Insomnia | 4 | CBD, THC+CBD | No significant improvement in overall insomnia symptoms, sleep quality, or sleep latency. Significant increases in sleep duration by device (moderate certainty) and sleep diary (low certainty), though device finding lost significance when high-bias studies removed. High adverse event rate (dry mouth, nausea, dizziness). | Mixed / limited |
| Opioid use disorder | 4 | CBD, THC | No significant effect on withdrawal symptoms or opioid craving. No difference in adverse events. | No significant effect |
| Cocaine use disorder | 3 | CBD | Significantly increased cocaine craving vs placebo (GRADE: very low certainty). Significantly increased adverse events (OR 3.76). | Harm signal |
| PTSD | 3 | THC, CBD, THC+CBD | No significant effect on PTSD symptoms at longest follow-up. No difference in adverse events. Three serious adverse events recorded (all in cannabinoid group). | No significant effect |
| Autism spectrum disorder | 2 | CBD, THC+CBD | Significant reduction in autistic traits overall (GRADE: very low certainty), but neither subgroup (THC+CBD or CBD alone) was significant individually. Both studies at high risk of bias. | Mixed / limited |
| Anorexia nervosa | 2 | THC | No significant difference in weight or physical activity between groups. Insufficient data for adverse events or withdrawals. | No significant effect |
| OCD | 2 | THC, CBD | No significant improvement in body-focused repetitive behaviours or general OCD symptoms. More adverse events in cannabinoid group (16 vs 7). | No significant effect |
| ADHD | 1 | THC+CBD | No significant differences for any outcome. Single small study (n=30). | Insufficient data |
| Bipolar disorder | 1 | CBD | No significant differences for any outcome. Single small study (n=35). | Insufficient data |
| Tobacco use disorder | 1 | CBD (inhaler) | No significant differences for any outcome. Single small study (n=24). | Insufficient data |
| Depression | 0 | — | No RCTs identified, despite being one of the most common reasons cannabinoids are prescribed. | Insufficient data |
Serious adverse events and study withdrawals did not differ significantly between cannabinoids and controls across conditions. Overall OR for all-cause adverse events: 1.75 (95% CI 1.25 to 2.46). 69% of participants were male; median age 33 years. CBD = cannabidiol; THC = delta-9-tetrahydrocannabinol; ASD = autism spectrum disorder; NNTH = number needed to treat to harm. Wilson et al., Lancet Psychiatry 2026.
In these RCTs medical cannabinoids were most commonly used for cannabis use disorder symptoms, and findings were most robust herein.
Conclusion
Overall, the quality of the evidence was low. However, the most robust findings were some evidence for symptom improvement in cannabis use disorder (THC+CBD combinations), autism spectrum disorder, insomnia (any cannabinoid), and Tourette’s syndrome (THC+CBD combinations).
The review found no significant effect of medical cannabinoids for psychotic disorders, anxiety disorders, opioid use disorder, PTSD, anorexia nervosa and OCD.
There was insufficient data on ADHD, tobacco use disorder, bipolar disorder and depression, which is perhaps surprising as depression is one of the most common reasons that cannabinoids are prescribed.
Overall, cannabinoids were associated with more adverse events compared to placebo, but serious adverse events and study withdrawals did not differ between groups. However, pooling the 3 trials available on cocaine use disorder suggested that medical cannabinoids may be harmful.
Certain symptoms of mental health disorders improved after medical cannabinoid administration, but overall evidence was low and rarely survived sensitivity analyses.
Strengths and limitations
This new review by Wilson and colleagues (2026) is comprehensive and methodologically rigorous, with a strong statistical approach, use of validated tools, and PROSPERO preregistration, all of which support transparency and reproducibility. A key strength is the use of meta-analyses, which allows us to pool the findings across the 54 RCTs included in this review. The methodology underlying these meta-analyses is clearly described and appears robust. Given the focus on the rise in the prescription of cannabinoids, the decision to include only RCTs makes sense, as this is the gold standard for assessing medication efficacy.
However, this approach does not fully reflect real-world patterns of medicinal cannabis use. In practice, most individuals who use medicinal cannabis do not obtain it through formal healthcare channels, but instead self-medicate, using over-the-counter or illicit products, often high in THC content. Thus, while it is essential that healthcare providers avoid prescribing medical cannabis in the absence of evidence for efficacy, prescription-based access does offer advantages, including clinical supervision and standardised, regulated products.
Moreover, although the study includes RCTs from all countries, the views offered in the article still seem very focused on specific English-speaking countries such as the United States and Australia. For example, the concern that some clinicians receive financial incentives for prescribing cannabis (without potentially being aware of the risks medicinal cannabis may carry) seems mostly a local/regional regulatory issue concerning compensation structures of clinicians, rather than an inherent problem with cannabis as a treatment. For example, in the Netherlands, there is only one medical cannabis company, which does not financially compensate doctors for prescribing their products (this is also prohibited). Potentially, this phenomenon is also a bigger influence on the hindering or delay of alternative treatments, rather than the prescription of cannabis, as the authors now state.
The focus on RCTs of this methodologically robust article is important for clinical practice, but we should not overlook real-world patterns of medicinal cannabis use, due to the high rates of self-medication.
Implications for practice
As prescription rates of cannabinoids for mental health symptoms continue to rise globally, comprehensive reviews and evidence syntheses, such as those by Wilson and colleagues (2026) carry important implications for clinical practice.
A consistent conclusion across the literature is that there is currently no strong evidence supporting the efficacy of medicinal cannabis in improving mental health symptoms of any kind. This may reflect either the generally low quality of existing studies or a genuine absence of therapeutic effect. This is an issue that remains to be clarified. On this basis, there is little justification for clinicians to prescribe medicinal cannabis specifically for mental health symptoms. Even in cases where some benefit has been suggested (e.g., insomnia), more established and evidence-based treatments are already available. However, medicinal cannabis may be considered in cases where patients have exhausted more common treatment options without success. However, such indications are more likely to fall outside the domain of mental health, for example, in the management of chronic pain, where secondary improvements in mental health may occur as a result of symptom relief.
Nonetheless, it is important not to overlook findings from self-report studies, in which patients using medicinal cannabis frequently report perceived improvements in symptoms across a range of mental health conditions. Therefore, the lack of demonstrated efficacy in RCTs should not lead to the dismissal of medicinal cannabis. Rather, it should prompt a shift in focus toward identifying which individual characteristics are associated with differential experiences of benefit and harm. It is also important to consider the risks associated with regular cannabinoid use, including an increased likelihood of developing cannabis use disorder (CUD), particularly symptoms related to withdrawal and tolerance. This risk is especially heightened in vulnerable populations, including individuals with a history of substance use disorders and younger adults.
As a result, clinicians must carefully weigh potential benefits against these risks and consider patient-specific factors when evaluating the appropriateness of this treatment. Interestingly, one of the most consistent findings of efficacy is the effects of cannabinoids in the treatment of CUD, particularly for alleviating withdrawal symptoms. This raises the question whether such approaches constitute genuine treatment effects or merely substitute one cannabis product for another. Finally, if prescriptions of medicinal cannabis are, in some cases, influenced by financial incentives or contribute to the delay or displacement of other treatments, this represents a significant concern for clinical governance. It emphasises the need for careful monitoring of prescribing practices to ensure transparency and adherence to evidence-based care.
Current evidence does not support prescribing medical cannabinoids for mental health conditions. More established, evidence-based treatments should be prioritised.
Statement of Interest
Nora de Bode has no conflicting interests to declare.
Edited by
Dr Dafni Katsampa
Links
Primary paper
Jack Wilson, Olivia Dobson, Andrew Langcake, Palkesh Mishra, Zachary Bryant, Janni Leung, Danielle Dawson, Myfanwy Graham, Maree Teesson, Tom Freeman, Wayne Hall, Gary Chan, Emily Stockings (2026) The efficacy and safety of cannabinoids for the treatment of mental disorders and substance use disorders: a systematic review and meta-analysis. The Lancet Psychiatry, 2026; 13, 304-315
