Evidence gaps are not the same as evidence of no effect
The recent 2026 Lancet Psychiatry review is an important contribution and should be taken seriously. It examined 54 randomised trials involving 2,477 participants and concluded that there was little RCT evidence of efficacy for several mental and substance use disorders. It also found no randomised controlled trial evidence for depression,4 and a 202613 review found that controlled evidence for cannabinoids in OCD remains very limited. That matters. But it does not follow that cannabinoids have been shown to be ineffective in depression or OCD, or that all psychiatric use is unjustified. In depression in particular, the review highlights an evidence gap, not evidence of no effect. In other words, absence of RCT evidence is not the same as proof of therapeutic futility.
This distinction matters because clinicians do not treat meta-analyses in the abstract; they treat patients in front of them. In real-world practice, many patients seeking CBMPs for mental health symptoms have chronic anxiety, trauma-related symptoms, poor sleep, autonomic hyperarousal, pain, and marked impairment in quality of life. They have often already tried multiple licensed treatments, psychological interventions, or both. The MCCS framework is clear that CBMPs are not first-line treatments and that conventional evidence-based options should normally have been appropriately explored first. But it is equally clear that prescribing is ultimately based on individual clinical judgment, patient need, risk profile and informed consent.1
Those real-world data are now too substantial to ignore. Project Twenty21 has reported consistent improvements in symptoms, general health and quality of life in large cohorts of UK patients, and its sleep analysis found clinically significant improvements sustained for up to 12 months together with reduced use of sleep medication.5,8 The UK Medical Cannabis Registry has shown similar signals in generalised anxiety disorder and PTSD, with improvements in anxiety, sleep and health-related quality of life and with most reported adverse events mild or moderate.6,9 Most strikingly for current debate, registry data in depression now extend to 24 months and show sustained improvement in depression scores, anxiety, sleep and quality of life in selected patients.10 These studies do not prove causation. They do, however, show that the lived clinical experience of benefit is not fanciful and should not be dismissed as irrelevant simply because it comes from observational data.5,6,8-10
International registry data point in the same direction. A large Australian longitudinal cohort of 3,961 cannabis-naive patients prescribed oral medicinal cannabis reported sustained improvements across multiple validated outcomes over two years and concluded that treatment was safe and well tolerated, while explicitly acknowledging the limits of real-world data for causal inference. This is the correct balance: neither overstatement nor dismissal. Real-world evidence cannot settle every question, but it is highly informative on tolerability, safety, prescribing patterns and patient-reported outcomes in routine care.7
