By Dr. Benjamin Caplan, MD | Board-Certified Family Physician, CMO at CED Clinic | Evidence Watch
Dealing with a condition like this?
Dr. Caplan has worked with 30,000+ patients on conditions like this. A consultation starts with your specific situation — not a generic protocol.
Book a consultation →A 2025 Harvard-affiliated expert review in CNS Drugs finds that most documented mental health harms from cannabis are specifically tied to high-THC recreational products and particular use patterns, not cannabis broadly. Whether cannabis helps or harms depends on cannabinoid composition, dose, frequency, and individual factors like age and genetic vulnerability, meaning blanket statements about cannabis and mental health are not supported by current evidence.
Cannabis and Mental Health: Why the Evidence Looks So Contradictory and What Actually Matters
A Harvard-affiliated expert review argues that product type, dose, use patterns, and who is using cannabis explain much of the confusion in research findings, with high-THC recreational use driving most documented harms while CBD shows early but inconclusive therapeutic promise for certain psychiatric symptoms.
#78
Strong Clinical Relevance
Directly addresses the moderating variables clinicians need to assess when counseling patients about cannabis and psychiatric risk.
THC vs CBD
Risk Stratification
Narrative Review
Medical vs Recreational Use
Cannabis legalization is expanding rapidly across North America and globally, yet clinicians face a contradictory evidence base when patients ask whether cannabis is safe for, or harmful to, their mental health. Headlines declaring cannabis a cause of psychosis coexist with studies suggesting cannabidiol may reduce anxiety and psychotic symptoms. Without a practical framework for reconciling these findings, front-line clinicians risk either overstating dangers uniformly or minimizing real risks in vulnerable populations. This review from McLean Hospital and Harvard Medical School offers a structured lens for the variables that actually drive differential outcomes.
Research on cannabis and mental health has produced seemingly irreconcilable findings: some studies link cannabis to psychosis onset, while others report anxiolytic or antipsychotic benefits. Sagar and Gruber, both affiliated with McLean Hospital and Harvard Medical School, argue that this confusion is not a flaw in individual studies but a systemic consequence of treating “cannabis” as a monolithic exposure. In reality, delta-9-tetrahydrocannabinol (d9-THC) and cannabidiol (CBD) exert quite different, sometimes opposing, effects on the brain. The dose, frequency of use, route of administration, and the ratio of THC to CBD in a given product all modulate outcomes. Layered on top of these product-level variables are individual characteristics including age of first use, genetic predisposition, family psychiatric history, and sex, each of which meaningfully alters risk profiles.
The review covers four psychiatric domains: psychosis, anxiety, mood disorders, and PTSD. Across each, the authors find that documented harms concentrate among heavy, early-onset, high-THC recreational users, particularly those with genetic or familial vulnerability to psychotic disorders. Cannabis use alone does not appear sufficient to cause psychosis but acts as a contributing factor in a multifactorial model. d9-THC shows biphasic effects on anxiety, potentially anxiolytic at low doses but anxiogenic at higher doses, though no clinical consensus exists on precise thresholds. CBD, by contrast, does not appear to worsen any psychiatric domain at any studied dose and shows early clinical promise for anxiety and psychotic symptoms, though human evidence remains preliminary. The authors emphasize that medical cannabis users typically select different products and cannabinoid profiles than recreational users, and their outcomes likely differ accordingly. They call for research that specifies product composition, use parameters, and individual vulnerability rather than treating cannabis exposure as binary.
This review articulates something I have been saying in clinical practice for years: “cannabis” is not one thing, and asking whether it helps or harms mental health without specifying what product, at what dose, in which patient, is the wrong question. Sagar and Gruber bring welcome intellectual clarity to a space often dominated by ideology on both sides. What I appreciate most is their insistence on separating d9-THC effects from CBD effects and distinguishing recreational from medical use contexts. What the review cannot do, and the authors are fairly transparent about this, is provide the precise clinical decision rules clinicians need. We still lack reliable dose-response data in humans for most psychiatric applications.
In my own practice, I already stratify risk along exactly these axes. For patients under 25 or those with a first-degree relative with psychosis or bipolar disorder, I am markedly more cautious with THC-containing products and discuss this risk explicitly. For patients with anxiety who want to explore cannabis, I begin with CBD-predominant formulations at conservative doses and titrate slowly, monitoring carefully. This review reinforces that approach but does not yet give us the controlled trial data to turn it into a formal guideline. We are still working with clinical reasoning more than clinical proof.
This review sits at a critical juncture in the cannabis research arc. The field has moved past the era of crude exposure-outcome analyses and is now grappling with moderator-level complexity. For clinicians, the practical takeaway is that risk counseling must be personalized. A 40-year-old patient using a CBD-dominant tincture for generalized anxiety occupies a fundamentally different risk profile than a 17-year-old using high-potency THC concentrates daily. Research has not yet produced the granular dose-response curves needed for formal prescribing guidance, but the directional evidence is increasingly clear: younger age of onset, higher THC potency, more frequent use, and genetic liability to psychosis are the factors that consistently cluster with adverse psychiatric outcomes across studies.
From a pharmacological standpoint, the biphasic nature of THC’s effects on anxiety deserves particular clinical attention. Low-dose THC may relieve anxiety while higher doses provoke it, but the boundary between “low” and “high” varies by individual tolerance, route of administration, and concurrent CBD content. CBD’s pharmacology is more forgiving in this regard, with no evidence of anxiogenic effects at any dose studied, though its bioavailability varies
