A Major New Review Found Cannabis Doesn't Curb Cocaine Cravings — It Intensifies Them

If you’ve quietly switched from cocaine to a joint because it felt like the safer choice, the largest review of cannabinoid medicine ever assembled has bad news: it may be making your cravings worse, not better.

That’s the standout finding buried inside a systematic review and meta-analysis — a pooled statistical analysis that combines results from many separate trials into one bigger, more reliable picture — published this year in Lancet Psychiatry. A team led by Wilson and colleagues combed through 54 randomized controlled trials — studies where patients are randomly assigned to get the real drug or a placebo, the design researchers trust most because it strips out guesswork about whether an effect is real — covering 2,477 people across sixteen different mental health and substance use conditions. It is, by a wide margin, the most rigorous test yet of the idea that cannabis is a general-purpose substitute for harder or more dangerous drugs.

The bottom line: cannabis is not the harm-reduction shortcut a lot of people are quietly using it as.

The review found real, if modest, benefit in a short list of places. A combination of CBD and THC eased cannabis withdrawal symptoms and cut weekly cannabis use in people trying to quit. Cannabinoids of any type improved total sleep time in insomnia. There was a small reduction in autistic traits. That’s it — that’s the whole list of conditions with a clear positive signal, out of sixteen tested. For anxiety, anorexia nervosa, psychotic disorders, PTSD, and opioid use disorder, the researchers found no significant effect at all. For ADHD, bipolar disorder, OCD, and tobacco use disorder, there wasn’t even enough data to draw a conclusion. For depression, there was no randomized trial evidence whatsoever — not weak evidence, none.

Then there’s cocaine. Cannabinoids didn’t just fail to help with cocaine use disorder — they paradoxically increased craving. This is worth sitting with, because it cuts directly against a self-medication story that’s become common: people cutting back on cocaine, or trying to get off benzodiazepines or opioids, treating weed as a lower-risk stand-in. If you’re doing that specifically to manage a cocaine problem, this study is telling you the substitution may be backfiring in the exact way you’re hoping it will help.

What “low quality evidence” actually means here

This is the part that gets skipped in most coverage of studies like this, and it’s the part that matters most for a reader trying to decide what to do Monday morning. Forty-four percent of the 54 trials were rated high risk of bias — meaning nearly half the underlying studies had design flaws (small samples, unblinded participants who could likely guess whether they got the real drug, inconsistent outcome measures) that make their individual results less trustworthy. That’s not a technicality. It means the withdrawal and insomnia findings, real as they are, rest on a foundation the authors themselves don’t fully trust yet. Promising signal is not the same thing as established medicine, and treating early-stage evidence as if it were settled science is exactly how people end up over-relying on a treatment that hasn’t earned that confidence.

The safety numbers deserve equal weight. Across the trials, cannabinoids carried an odds ratio of 1.75 for all-cause adverse events compared to placebo — meaning a meaningfully higher likelihood of side effects in the treatment group — with a number-needed-to-harm of 7. That statistic answers a specific question: how many people need to take this before you’d expect one of them to experience a harm from it that they wouldn’t have otherwise. Seven is a low number for a substance being informally prescribed by trial and error, outside any monitored protocol, by people using it for a use case the review found no benefit for.

As OBG Project’s plain-language summary of the paper notes, the study’s own authors concluded that routine use of cannabinoid medicines for mental disorders and substance use disorders is rarely justified on current evidence, and flagged a real concern: that leaning on cannabis could delay or displace treatments that actually work. That’s not reflexive caution from people who dislike cannabis — it’s 54 trials and 2,477 patients telling them the same thing across sixteen conditions, with only a narrow slice showing benefit.

None of this means cannabinoids are worthless. The cannabinoid research building block for the field is getting more precise, not less — CBD+THC for withdrawal and sleep is a real, specific, testable finding, not noise. But precision cuts both ways: it also means the sweeping claim — cannabis as a stand-in for whatever you’re trying to quit — doesn’t survive contact with the data. If you’re a provider fielding a client’s plan to taper off cocaine with weed, this is the study to put in front of them. If you’re the one making that plan yourself, the honest version of this finding is that the drug you picked because it felt lower-stakes may be pulling you toward the thing you’re trying to leave, not away from it. That’s worth knowing before you’re seven weeks into a plan built on a hope this review just measured and found wanting. For the broader state of the evidence, the science and medicine desk will keep tracking what changes as the trials that follow this one get built with the rigor 44% of these didn’t have.

The cannabinoid research building block for the field is getting more precise, not less — CBD+THC for withdrawal and sleep is a real, specific, testable finding, not noise.