What the Evidence Actually Says About Cannabis: Risks, Benefits, and Lingering Uncertainties

A high-profile invited review by international experts synthesizes a decade of research on recreational cannabis risks and medicinal cannabinoid efficacy, offering a measured but incomplete portrait of what can and cannot be concluded from the existing literature.

Why This Matters

Cannabis policy is shifting faster than the science that should inform it. Across North America and Europe, legalization and commercialization are advancing against a backdrop of rising THC potency and expanding medical claims, yet the evidence base for both harms and benefits remains more uncertain than public discourse suggests. This review, published by a multinational team that includes the director of the U.S. National Institute on Alcohol Abuse and Alcoholism, arrives at a moment when clinicians urgently need honest, well-contextualized guidance rather than advocacy from either side of the debate.

Clinical Summary

Published in European Archives of Psychiatry and Clinical Neuroscience, this invited narrative review by Hoch and colleagues synthesizes evidence from systematic reviews and meta-analyses identified through PubMed and Cochrane searches spanning May 2016 to May 2023. The review updates a prior report commissioned by the German Ministry of Health and covers two broad domains: the health risks of recreational cannabis use and the efficacy and safety of cannabis-based medicines. The mechanistic framework is straightforward. Delta-9-tetrahydrocannabinol acts on the endocannabinoid system, which modulates neurodevelopment, reward processing, stress responses, pain signaling, and immune function. This broad biological footprint explains why cannabis touches so many organ systems and why both therapeutic promise and adverse risk profiles are plausible.

On the risk side, the review finds consistent associations between regular, high-THC cannabis use and cannabis use disorder, cognitive impairment across multiple domains (with IQ declining approximately 2 points per year of frequent or dependent use), increased psychosis and schizophrenia risk, mood and anxiety disorders, and suicidal behaviors, particularly among adolescents and young adults. Motor vehicle accident risk was elevated (OR 1.25 to 1.97). On the medical side, cannabis-based medicines showed small to modest benefits for chronic pain, muscle spasticity, and chemotherapy-induced nausea, with CBD demonstrating efficacy for refractory epilepsy. Critically, this is a selective, non-systematic review with no PRISMA diagram, no registered protocol, and no formal risk-of-bias assessment. The authors themselves acknowledge that causality for most associations remains debated, and the review’s conclusions should be read as expert synthesis rather than definitive meta-analytic estimates.

Dr. Caplan’s Take

This review does something valuable: it puts the recreational risk evidence and the medical efficacy evidence side by side, which forces an honest reckoning with the asymmetry between them. The harms literature is largely observational and plagued by confounding, while the benefits literature, though trial-based, consists mostly of small studies with modest effect sizes. Patients come to me convinced that cannabis is either a miracle or harmless. Neither framing survives contact with this evidence. What an honest response requires is acknowledging real signal in the risk data, especially for adolescents and those with psychiatric vulnerability, without overstating causal certainty.

In practice, I treat cannabis use as a clinical variable that demands the same careful assessment as alcohol or any other psychoactive substance. For patients considering medicinal cannabinoids, I focus on the indications where evidence is most robust: chronic pain that has failed other approaches, spasticity, and specific epilepsy syndromes. I counsel strongly against high-THC recreational use in patients under 25 or those with personal or family histories of psychotic disorders. And I am transparent that most mental health applications of cannabis-based medicines remain unproven despite widespread marketing claims.

Clinical Perspective

This review sits at a useful but intermediate point in the research arc. It confirms what multiple prior systematic reviews have found: the association between heavy cannabis use and psychiatric risk is consistent and biologically plausible, but proving causation remains elusive due to shared genetic vulnerabilities, reverse causation, and residual confounding. For medical applications, the review corroborates the conclusion that benefits are real but narrow, concentrated in chronic pain, spasticity, chemotherapy-related nausea, and CBD for refractory epilepsy. Evidence for cannabinoids in depression, anxiety, PTSD, and sleep disorders remains inconclusive, which is important to communicate clearly to patients who may encounter aggressive marketing for these indications.

Clinicians should be aware that even in medical contexts, the risk of cannabis use disorder is non-trivial, and drug interactions with cannabinoids, particularly CBD’s inhibition of CYP3A4 and CYP2C19, can alter levels of commonly prescribed medications including anticoagulants, antiepileptics, and certain antidepressants. Adverse effects in clinical trials were generally mild to moderate but included somnolence, dizziness, and gastrointestinal symptoms. One concrete, implementable recommendation: screen all patients reporting cannabis use, whether recreational or medical, for frequency, THC concentration, age of onset, and psychiatric history, and document this assessment as systematically as you would for alcohol or tobacco use.

Study at a Glance

Study Type

Invited narrative review

Population

General population (recreational risks); patients with chronic pain, spasticity, epilepsy, nausea, and psychiatric conditions (medical efficacy)

Intervention

Recreational cannabis use (risk domain); cannabis-based medicines including THC, CBD, and combination products (efficacy domain)

Comparator

Non-users (risk domain); placebo or active comparators in cited trials (efficacy domain)

Primary Outcomes

Psychiatric, cognitive, physical, and social harms (risk domain); symptom reduction and adverse events (efficacy domain)

Sample Size

Not reported; number of included reviews not specified

Journal

European Archives of Psychiatry and Clinical Neuroscience

Year

2025 (online September 2024)

DOI or PMID

Not available in extracted text

Funding Source

Not reported in extracted text

What Kind of Evidence Is This

This is an invited narrative review, not a systematic review or meta-analysis. It occupies a position in the evidence hierarchy below systematic reviews and well above expert opinion alone, but its value depends entirely on the representativeness and quality of the literature the authors chose to cite. The absence of a registered protocol, PRISMA-compliant search, and formal risk-of-bias assessment means the review is not reproducible, and its conclusions could be influenced by selection or framing bias in ways that are difficult for the reader to detect.

How This Fits With the Broader Literature

The findings here are broadly consistent with the 2017 National Academies of Sciences report on cannabis health effects and with the conclusions of Whiting et al. (2015), which found moderate evidence for cannabinoid efficacy in chronic pain and spasticity but limited evidence for most other indications. The psychiatric risk findings align with the extensive work of Di Forti and colleagues on high-potency cannabis and psychosis. What this review adds is temporal currency, covering literature through mid-2023, and the explicit juxtaposition of harm and benefit evidence in a single document. It does not meaningfully challenge any established consensus, nor does it resolve the persistent causal ambiguity that characterizes the observational harm literature.

Common Misreadings

The most likely overinterpretation is treating the reported associations between cannabis use and psychiatric outcomes as established causal relationships. The authors are careful to note that causality is debated for most outcomes, yet the review’s structure and tone may lead readers to infer stronger causal conclusions than the underlying observational data support. Conversely, on the medical side, the characterization of “small to modest” benefits may be misread as “clinically insignificant.” In some patients, a small average effect size can translate into meaningful individual benefit, particularly for treatment-refractory conditions. Both directions of misreading reflect the same error: collapsing population-level summary statistics into individual-level certainties.

Bottom Line

This narrative review provides a useful, current synthesis of cannabis-related evidence but does not generate new data or resolve longstanding causal questions. Heavy, early-onset, high-THC cannabis use is consistently associated with meaningful health risks, and medicinal cannabinoids offer modest benefits for a narrow set of indications. For clinical practice today, the review reinforces the need for careful screening, honest patient counseling, and restraint in recommending cannabis-based medicines outside their best-supported indications.

References

1. Hoch E, Lorenzetti V, Hall W, Volkow N, et al. Cannabis, cannabinoids, and health: an updated overview of the evidence. European Archives of Psychiatry and Clinical Neuroscience. 2025 (online September 2024).
2. National Academies of Sciences, Engineering, and Medicine. The Health Effects of Cannabis and Cannabinoids: The Current State of Evidence and Recommendations for Research. Washington, DC: National Academies Press; 2017. DOI: 10.17226/24625.
3. Whiting PF, Wolff RF, Deshpande S, et al. Cannabinoids for medical use: a systematic review and meta-analysis. JAMA. 2015;313(24):2456-2473. DOI: 10.1001/jama.2015.6358.
4. Di Forti M, Quattrone D, Freeman TP, et al. The contribution of cannabis use to variation in the incidence of psychotic disorder across Europe (EU-GEI): a multicentre case-control study. Lancet Psychiatry. 2019;6(5):427-436. DOI: 10.1016/S2215-0366(19)30048-3.