Headlines this week said cannabis users are twice as likely to commit violence. The review behind them is real and large. The word "likely" is where the trouble starts, and so is the size of the number once researchers follow people over time.

The paper, by Giulia Trotta and colleagues at King's College London's Institute of Psychiatry, Psychology & Neuroscience, appeared in Psychological Medicine in late September. It pooled 63 studies covering 265,079 people.

What the Review Pooled

The team separated studies of the general population from studies of people in psychiatric care. They used the World Health Organization's broad definition of violence, which covers physical, sexual and psychological violence against another person, inside and outside relationships. When a study reported several kinds of violence, the reviewers took the most serious outcome, according to Earth.com's account of the paper.

That breadth is a strength and a weakness. It gives a wide picture, but a questionnaire answer about a fight and a criminal conviction can both end up counted as "violence."

The Headline Numbers

In the general population, cannabis users had 2.05 times the odds of committing violence compared with non-users. Among psychiatric patients, the figure was 2.49. Users also had about 1.49 times the odds of being a victim of violence, Medical Xpress and Earth.com reported.

A note on wording: an odds ratio is not the same as "twice as likely." The two diverge when an outcome is common, and a reader who sees "twice as likely" may picture a bigger jump than the statistics describe.

The Design Gap

The most important result is buried in the subgroup tables. Studies that measured cannabis use and violence at one point in time found 2.37 times the odds in the general population. Studies that followed people over months or years found 1.17, Earth.com reported. For victimization, the drop was starker: 2.20 in one-time studies against 1.05 in follow-up studies.

Simon Moore, director of the Violence Research Group at Cardiff University, put it sharply in a reaction published by the Science Media Centre on Sept. 29. He noted that the gap between the two designs is nearly twofold, roughly as big as the headline effect itself, and that a confounder no stronger than commonly cited factors such as peer group or neighborhood could explain the follow-up estimate entirely. Results with and without statistical adjustment were similar, he said, but that shows the adjusted factors made little difference, not that context was accounted for.

This is the same lesson that shows up whenever an observational study meets the real world. Blogerroom's piece on REM sleep tied to lower risk for 83 diseases turned on the same question: when a link appears, how much of it survives once time and context are controlled? The authors of the cannabis review say plainly that it shows an association and cannot prove cause.

The Biggest Numbers Rest on Three Studies

The largest figures in the paper involve criminal convictions. Cannabis use was linked with 3.75 times the odds of a violent conviction in the general population and 4.21 times among psychiatric patients. Each of those results came from only three studies, Earth.com noted, so they rest on far thinner evidence than the main findings. Robin Murray, a senior author, still highlighted them in the university's statement.

My view: they make good headlines and weak foundations. The pooled estimate from dozens of studies deserves more trust than a ratio drawn from three.

What Is Missing

The review cannot say whether occasional use carries the same association as daily use. Most studies did not record how much people used, how often, or the THC strength. Many users also drink alcohol or take other drugs, and the original studies did not always account for that fully.

Potency is another gap. The researchers say most included studies were done before high-potency cannabis became common, so the results may understate, or simply not capture, the effect of today's products. Moore added a point about context: because cannabis remains illegal in most places, illicit markets can bring users into contact with people and settings where violence is likelier, and the review did not examine where or with whom cannabis was used.

What It Means

For clinicians the practical message is narrower than the headline. Marta Di Forti, a co-author who also works at the South London and Maudsley NHS Foundation Trust, said heavy cannabis use can be a factor she sees in violence, and that a review this size lets researchers speak with more confidence while avoiding stigma. The authors suggest mental health services should routinely ask about cannabis use when assessing a person's risk of violence.

For everyone else, the lesson is about method. The same dataset can look alarming or modest depending on which subgroup you read, much like the shifting overdiagnosis figures Blogerroom covered in the mammography reanalysis. The fair summary is that cannabis use and violence travel together, the connection shrinks when people are tracked over time, and nobody has yet shown how much is the drug and how much is the life around it. This article is general information and not medical or legal advice.