Health Canada collected fewer than 700 adverse reaction reports about legal cannabis products over six years of national surveillance. Most reporters were older adults using cannabis medically, and the passive reporting system cannot tell us how common these harms actually are among the millions of Canadians who consume cannabis. The findings generate hypotheses for future study but should not be mistaken for a measure of cannabis safety or danger.

Cannabis is now widely legal in Canada, yet post-market safety surveillance remains in its early stages and the evidence base for quantifying real-world harms is strikingly thin. Unlike pharmaceuticals that undergo years of premarket clinical trials, cannabis products entered the legal market with minimal controlled safety data. This publication represents the first systematic look at what Health Canada’s national pharmacovigilance system has actually captured, making it a foundational document for understanding the scope and limitations of cannabis safety monitoring in a regulated market. Clinicians, regulators, and patients all need a realistic picture of what these data can and cannot tell us.

Cannabis legalization in Canada created a regulated consumer market without the pre-market safety infrastructure typical of pharmaceutical approvals. To fill that gap, Health Canada relies on the Canada Vigilance Program, a passive spontaneous adverse reaction reporting system that collects voluntary reports from consumers, healthcare professionals, and manufacturers. This study analyzed all cannabis-related adverse reaction reports submitted to the system between October 2018 and December 2024, providing the first published descriptive overview of the program’s cannabis data. The rationale is straightforward: identifying reporting patterns and safety signals is a necessary first step toward understanding potential harms, even when the data cannot establish how frequently those harms occur.

Over the six-year period, 698 unique adverse reaction cases were identified. Reporters were predominantly older adults (mean age 56 years) and medical cannabis users (67.5%). Cannabis extracts accounted for 68.8% of reports. The most frequently cited adverse events included hallucination, headache, nausea, dizziness, and dyspnea. While 62.3% of cases were classified as serious, the dominant seriousness criterion was “other medically important condition,” a broadly defined category that does not equate to life-threatening outcomes. Causality was assessed as “possible” in the majority of cases, meaning a definitive link between the cannabis product and the reported harm was not established. The authors emphasize that the passive design, absence of denominator data on total cannabis exposures in Canada, and inherent reporting biases make it impossible to calculate incidence rates or draw causal conclusions. They call for enhanced surveillance methods, including active monitoring and better integration of exposure data.

This study deserves credit for doing something no one else has done publicly: laying out what Canada’s national cannabis pharmacovigilance system has actually collected and being honest about how limited those data are. Six hundred ninety-eight reports over six years across an entire country tells us much more about the weaknesses of passive reporting systems than about the safety profile of cannabis itself. The gap between what we need to know and what this kind of surveillance can provide is enormous, and the authors rightly acknowledge it.

In my practice, I see the population this study describes: older patients using cannabis medically, often navigating products with variable pharmacokinetics, particularly edible extracts. The adverse events listed here are not surprising to any clinician who works with these patients. What I find more useful than the event counts is the confirmation that our current safety monitoring infrastructure is profoundly inadequate. I use this kind of data not as a guide to specific harms but as reinforcement for the careful, individualized dosing conversations I already have with every patient. We need better tools, not just more reports.

For clinicians working in cannabis medicine, this study is best understood as a map of what one surveillance system has captured rather than a reliable measure of clinical risk. It sits at the earliest stage of the research arc: signal detection. The demographic skew toward older medical users means the data tell us little about harms in younger recreational consumers, heavy users, or populations with psychiatric vulnerability. The dominance of cannabis extracts in the reports may reflect both their market share and the well-known challenge of delayed-onset and unpredictable dosing that oral products present to inexperienced users. These patterns are hypothesis-generating, not confirmatory.

Pharmacologically, the observation that certain adverse events (such as psychotropic effects) were reported more frequently with high-THC products while others appeared more often with high-CBD products is intriguing but must be interpreted cautiously given the lack of dose verification or controlled conditions. Drug interactions are not systematically assessed in this dataset, and clinicians should not use these findings to modify existing prescribing guidance. The most actionable takeaway from this study is the reminder that patients using cannabis extracts, particularly older patients on multiple medications, warrant careful titration counseling and proactive adverse event monitoring in the clinical encounter, because the surveillance system alone is not capturing the full picture.