Clinical Insight | CED Clinic: A landmark meta-analysis of 54 randomized trials finds no significant benefit for cannabinoids in treating anxiety, PTSD, depression, psychotic disorders, or opioid use disorder, challenging widespread patient and clinician expectations.

By Dr. Benjamin Caplan, MD | June 27, 2026 | Cannabis Science

That patient and clinician expectations about cannabis for mental health often exceed what the evidence currently supports. The study also teaches a less obvious lesson: negative evidence from high-quality trials is as clinically informative as positive evidence—it tells us where we should exercise caution and where further research is genuinely needed.

Many patients seek cannabis for anxiety, depression, and PTSD, and many clinicians feel pressure to offer cannabis as a ‘natural’ alternative when conventional medications fail or cause side effects. This meta-analysis provides clarity: for most psychiatric indications, we do not yet have sufficient evidence to recommend cannabinoids as first- or second-line therapy. Understanding the evidence gap is clinically important because it shapes realistic patient conversations about risk, benefit, and what the science actually shows.

Current evidence does not support routine prescribing of cannabinoids for anxiety, depression, PTSD, psychotic disorders, OCD, anorexia nervosa, or opioid use disorder. For most psychiatric indications, cannabinoids should not be recommended as established first- or second-line treatment. Further high-quality research is needed to identify any populations or conditions where cannabinoids may have a measurable benefit. Clinicians should manage patient expectations accordingly.

Wilson, Freeman, and colleagues conducted a methodologically rigorous search of published literature through May 2025, screening 5,774 potentially relevant studies. They applied strict inclusion criteria: only randomized controlled trials, no observational studies or case reports. This exclusion of non-RCT data was intentional—RCTs are the gold standard for establishing whether a treatment works, and the researchers wanted to avoid the inflated effect sizes often seen in weaker study designs.

The final pool included 54 RCTs with a total of 2,477 participants. The trials examined cannabinoids (including purified CBD, THC, synthetic cannabinoids, and whole-plant cannabis) across multiple psychiatric and substance-use domains: anxiety disorders, PTSD, depression, psychotic disorders, OCD, anorexia nervosa, opioid use disorder, cocaine use disorder, and cannabis use disorder itself. Study quality varied, but all were randomized designs.

For generalized anxiety disorder, panic disorder, and social anxiety: the meta-analysis found no significant difference between cannabinoid-treated groups and placebo groups. For PTSD, the same null result held. For depression, cannabinoids showed no significant benefit. These findings were consistent across CBD, THC, and combination formulations. The absence of benefit was not due to study quality alone—even higher-quality trials showed no clear signal.

This directly contradicts a widespread perception among patients and some clinicians that cannabis is helpful for anxiety or mood. While individual patients may report subjective improvement, the systematic pooling of RCT data does not support a robust, measurable benefit for these indications.

The review examined cannabinoids for psychotic disorders (including schizophrenia and bipolar disorder with psychosis). No significant benefit was detected. For OCD, the evidence base was sparse, but existing trials showed no clear advantage over placebo. For anorexia nervosa, the data were similarly limited and showed no significant therapeutic effect.

For opioid use disorder, a condition where cannabinoids are sometimes proposed as a harm-reduction or adjunctive tool, the meta-analysis found no robust evidence of benefit. This is a particularly important null finding because it contradicts emerging clinical enthusiasm in some practice settings.

One finding stood out as potentially harmful: in trials examining cannabinoids for cocaine use disorder, cannabinoid-treated participants showed significantly increased craving for cocaine compared to placebo-treated controls. This was not a dramatic effect size, but it was consistent and statistically significant. This finding warrants careful attention and raises questions about cannabinoid use in addiction populations.

The mechanism is unclear—it may relate to endocannabinoid system effects on reward circuitry—but the clinical implication is straightforward: cannabinoids do not appear to reduce cocaine craving and may increase it. This contradicts any therapeutic rationale for offering cannabis to patients struggling with cocaine use.

Cannabis use in the general population is often motivated by perceived mental health benefits. Surveys show that anxiety and sleep problems are among the most commonly cited reasons for cannabis use, especially in jurisdictions where cannabis is legal. This widespread perception stands in contrast to the null RCT findings, creating a significant gap between patient belief and clinical evidence.

The endocannabinoid system plays a role in stress response, fear extinction, and emotional regulation, providing a plausible biological rationale for why cannabinoids might help psychiatric conditions. However, plausible biology is not the same as proven efficacy. Many compounds with sound theoretical rationale fail in clinical trials.

Conventional psychiatric medications (SSRIs, SNRIs, antipsychotics, benzodiazepines) have substantial RCT evidence supporting their use, though they come with their own side effects and limitations. The absence of evidence for cannabinoids does not necessarily mean other treatments are superior—but it does mean cannabinoids should not be presented as established alternatives until the evidence improves.

This meta-analysis reflects a growing maturation in cannabis research: moving away from anecdote and animal models toward systematic synthesis of human RCT data. As more trials are conducted, the evidence picture may shift, but current data do not support routine prescribing.

From a clinical practice standpoint, this meta-analysis clarifies an important gap: our patients often expect cannabis to help anxiety, depression, or PTSD, and the evidence as it stands does not support that expectation. This does not mean we dismiss the patient’s reported benefit—placebo effects are real, and individual responses vary. But it does mean we need to have honest conversations about what the research shows and what it does not.

The practical implication is straightforward: cannabis should not be recommended as first-line or evidence-based treatment for anxiety, depression, PTSD, or most other psychiatric indications on the basis of this review. If a patient is interested in cannabis despite the lack of strong evidence, a shared decision-making conversation is appropriate—acknowledging uncertainty, discussing risks (including potential worsening of psychosis or increased craving in some conditions), and monitoring for benefit or harm. For patients with comorbid pain, sleep, or nausea, the calculus might shift. But for psychiatric symptoms in isolation, the evidence does not yet support cannabinoid prescribing as routine care.

The largest meta-analysis of randomized trials to date shows that cannabinoids do not produce measurable benefits for anxiety, depression, PTSD, psychotic disorders, OCD, or opioid use disorder. This does not mean cannabinoids never help any individual patient—subjective improvement and placebo response are real. But it does mean we cannot recommend cannabinoids as an evidence-based treatment for these conditions in routine practice. The findings should shift how we talk to patients: from ‘cannabis is a natural alternative for anxiety’ to ‘research does not yet support using cannabis for anxiety, though some people report feeling better, and we don’t fully understand why.’ This distinction between individual report and population-level evidence is crucial for informed shared decision-making.

How to Read This Meta-Analysis

This meta-analysis answers a specific, narrow question: Do randomized controlled trials show that cannabinoids improve mental health symptoms compared to placebo? The answer is no for most psychiatric indications. But ‘no’ means different things depending on context.

Understanding the nuance between ‘no evidence of benefit’ and ‘evidence of no benefit’ is important. The review found no significant difference between cannabinoid and placebo groups. This could mean: (1) cannabinoids genuinely do not work for these conditions, or (2) the trials were too small, too short, or used the wrong formulation to detect a real effect. The authors were careful about this distinction.

Five Steps to Interpreting the Findings