Cannabis and Violence: Large Systematic Review Finds Link but Cautions Against Inferring Causation

Overview

This article analyzes a large systematic review from King's College London that links cannabis use to violence, but cautions that the evidence does not establish causation and is limited by study design and biases. Author: The Telegraph.

• Cross-sectional studies show stronger associations than longitudinal studies.
• Confounders such as poverty, alcohol use, other drugs, and upbringing may explain part of the link.
• Odds ratios are not the same as probabilities and can be misleading when violence is not rare.
• Overall certainty in the evidence is very low, so headlines may overstate what the data can support.

Background

The Telegraph headline claims cannabis users are twice as likely to be violent, but the article under review emphasizes that the underlying evidence is nuanced. The piece discusses a systematic review conducted by researchers at King’s College London, which pooled 63 studies involving more than 265 000 people. While the review does identify an association between cannabis use and violence, it stresses that the headline figure does not equate to a causal relationship and may be exaggerated by the ways the studies were designed and how data were gathered.

What the Review Found

The core finding cited in the piece is an association between cannabis use and violence, observed across the 63 included studies. However, the authors of the review note that the strength of this association varied depending on study design. Cross‑sectional analyses, which measure exposure and outcome at a single point in time, tended to show a stronger link, whereas longitudinal studies, which follow individuals over time, reported a weaker association. The Telegraph’s framing of the numbers as “twice as likely” oversimplifies what the data actually show and risks misinterpretation.

Key Cautions: Confounders

The article outlines several factors that commonly correlate with violence and may confound the cannabis‑violence relationship. Poverty, alcohol consumption, use of other drugs, and adverse childhood experiences are all independently linked to higher violence risk. Cannabis users may also be more likely to live in contexts where violence is more prevalent or to encounter situations where crime occurs, simply due to legal and social environments. The review attempted to adjust for some of these factors, but the authors acknowledge that the methods used in many studies left room for residual confounding.

Study Design and Bias

According to the analysis, 46 of the 63 studies were cross‑sectional, limiting causal inference because exposure and outcome are measured simultaneously. Longitudinal studies provide a clearer time order (which comes first) and generally found a weaker link between cannabis use and violence. The article also discusses potential biases in data collection, publication bias, and the heterogeneity of what counts as cannabis use (ever used, past month use, daily use) and what constitutes violence (domestic incidents, fights, or criminal convictions). The pooled analyses across different data sets varied widely, which can reflect differences in definitions and sampling methods. The review’s supplementary materials judged the certainty of the evidence to be very low, indicating substantial uncertainty about the true size of any association.

Interpretation: Odds vs Likelihood

The piece explains a technical nuance: the difference between odds and likelihood. Even when the odds of violence are doubled, this does not necessarily translate to a doubling of the probability of violence, particularly when baseline violence is not rare. This distinction matters for public understanding and policy implications, and highlights why large observational studies can mislead if their statistical metrics are not interpreted correctly.

Conclusions and Implications

None of these limitations negate the possibility that cannabis use and violence are connected in some contexts, but the authors of the review caution against overinterpretation of a single headline figure. The very low certainty status means there is limited confidence in the precise size of the association, though the existence of some real relationship remains plausible. The Telegraph’s presentation of a clear, causal, and large effect is not supported by the evidence as described in the article. The piece emphasizes the need for careful interpretation and for more rigorous, longitudinal research to disentangle causal pathways from correlational patterns, especially in real-world settings with varying legal statuses for cannabis across regions.

Context and Limitations

While cannabis policy discussions often cite risk associations, this analysis reminds readers that complexity underlies the relationship between substance use and violence. The article notes that the review’s authors themselves acknowledge many of the limitations and that headline certainty should not be mistaken for scientific certainty. The PubMed link to the systematic review is provided for readers who wish to examine the primary source in detail.