Psychiatry · Medical Literature
[Correspondence] Efficacy and safety of cannabinoids
Abstract
We read with interest the systematic review and meta-analysis by Jack Wilson and colleagues1 on the efficacy and safety of cannabinoids for mental and substance use disorders. The authors identify a “crucial need for improved study design that includes larger and more representative participant samples”, but do not elaborate on how this might be achieved. We suggest that addressing this gap requires frameworks that better capture real-world pharmacological and population-level complexity.
Clinical relevance
Do the methodological frameworks used in existing randomized controlled trials (RCTs) of cannabinoids for mental and substance use disorders adequately capture real-world pharmacological and population-level complexity, and how might future study design better address this gap? __SNAPSHOT__ {"studyDesign":"Correspondence (letter) responding to a previously published systematic review and meta-analysis (Wilson et al., Lancet Psychiatry 2026; 13:304–315)","setting":"Not applicable (correspondence); the index systematic review was conducted across institutions in Australia, the United Kingdom, and associated international collaborators","sampleSize":"Not applicable (correspondence); the index review included 54 RCTs with 2,477 total participants","population":"Not applicable as primary data; commentary addresses participants enrolled in RCTs of cannabinoids for mental disorders and substance use disorders as reviewed by Wilson et al.","intervention":"Not applicable (correspondence); the index review examined cannabinoids (CBD, THC, synthetic cannabinoids, and whole-plant cannabis) as primary treatments for mental and substance use disorders","comparator":"Not applicable (correspondence); the index review compared cannabinoids to placebo or active controls within RCTs","primaryOutcome":"Not applicable (correspondence); authors critique the adequacy of existing study designs and propose frameworks incorporating real-world pharmacological and population-level complexity","followUp":"Not reported","fundingSource":"Not reported"}
Key findings
This is a correspondence letter and contains no original primary data, effect sizes, p-values, confidence intervals, or NNT/NNH values. The authors acknowledge the conclusion of Wilson et al. that existing RCTs of cannabinoids for mental and substance use disorders identify a 'crucial need for improved study design that includes larger and more representative participant samples,' but argue the original review does not elaborate on how this might be achieved. The correspondence authors propose that future research frameworks must better capture real-world pharmacological complexity (e.g., variability in cannabinoid formulations, routes of administration, THC:CBD ratios, and dosing patterns used in clinical practice) and population-level complexity (e.g., comorbidities, polypharmacy, and sociodemographic diversity systematically underrepresented in existing RCTs). No quantitative findings are presented. For reference, the index meta-analysis (Wilson et al.) found no significant benefit of cannabinoids for anxiety, depression, PTSD, psychotic disorders, OCD, anorexia nervosa, or opioid use disorder; found some evidence for benefit in cannabis use disorder, insomnia, Tourette's syndrome, and autism spectrum disorder at generally low evidence quality; and reported an overall odds ratio for all-cause adverse events of 1.75 (95% CI 1.25–2.46), with serious adverse events and study withdrawals not differing significantly between cannabinoids and controls.
Practice implications
This letter does not establish new clinical evidence and should not independently change practice. Its clinical value lies in contextualizing the limitations of the RCT evidence base that underpins current cannabinoid prescribing decisions. The correspondents correctly identify a structural tension: the existing RCT evidence base—summarized by Wilson et al. as the largest RCT-only meta-analysis to date—excludes the real-world formulations, dosing patterns, and patient populations that are most commonly encountered in practice. Patients receiving cannabinoids in clinical settings often have multiple comorbidities, concurrent psychotropic medications, and differing sociodemographic profiles compared with the predominantly male (69%), younger (median age 33 years) trial populations in the index review. Consequently, the degree to which negative or null findings from current RCTs translate to the heterogeneous patients seen in psychiatric and addiction medicine practice remains uncertain. Importantly, the absence of evidence for benefit in conditions such as anxiety, PTSD, and depression does not itself constitute evidence of absence in all real-world subpopulations; however, it does not justify routine prescribing in those indications either. This letter is available only as an abstract/brief correspondence; the full text and any additional proposed methodological frameworks are not available for review, which limits comprehensive appraisal. __LIMITATIONS__ As a correspondence, this article presents no original data and undergoes no formal peer review of empirical claims; it is inherently opinion-based. No systematic search, risk-of-bias assessment, or GRADE evaluation is applicable. The proposed 'frameworks' referenced in the abstract are not elaborated in the available text, precluding critical evaluation of their methodological feasibility or scientific rigor. The authors do not quantify the magnitude of the real-world complexity they describe or provide evidence that pragmatic trial designs would yield different efficacy estimates. The correspondence is reactive to a single meta-analysis and does not engage with the broader literature on adaptive trial design, platform trials, or existing real-world evidence (RWE) methodologies in this area. Potential conflicts of interest for the correspondence authors are not reported in the available abstract. __BOTTOM_LINE_CATEGORY__ low_certainty __BOTTOM_LINE__ This correspondence raises a methodologically valid concern: the RCTs informing current cannabinoid prescribing in psychiatry and addiction medicine may inadequately represent real-world pharmacological complexity and patient heterogeneity. However, as a brief opinion piece without original data or specified alternative frameworks, it does not provide actionable guidance to change practice. Clinicians should continue to rely on the Wilson et al. meta-analysis findings, which found insufficient RCT evidence to justify routine cannabinoid prescribing for most mental and substance use disorders, while acknowledging that the evidence base for real-world clinical populations remains an active area of methodological debate.