Weed Out the Risk: Pharmacovigilance in Medical Cannabis Users

IntroductionMedical cannabis use has expanded rapidly, yet long-term real-world safety data remain limited. We evaluated adverse-event (AE) frequency, severity, and predictors in a US telehealth registry of medical cannabis patients over one year. MethodsWe analyzed 14,313 adults who completed intake between June-August 2024. Patients reported any of 30 prespecified adverse events (AEs) and rated each on a 0-10 impact scale. Weekly exposure was estimated as (days/week) x (serving size) and categorized into quintiles. We computed AE rates per 100 patients with binomial 95% confidence intervals (CI) and tested linear trends. Univariate logistic regressions assessed 20 candidate predictors within chronic-pain and anxiety subgroups. We then applied LASSO to select multivariate predictors, combining these with age, sex, race/ethnicity, smoking, and unhealthy-weeks in final logistic models. Marginal predicted-probability curves were generated across exposure, stratified by subgroup, sex, race, and age. ResultsOverall, 2.6% of patients reported [&ge;]1 AE. The most common symptoms were increased appetite (23.8%), fatigue (20.3%), and anxiety (19.9%) with mean impact <4/10. In adjusted models, having been to the doctor because of their condition remained the sole AE predictor for patients with anxiety (OR 4.03, 95% CI: 2.44-6.87); age was a significant predictor for patients with chronic pain (OR 0.981, 95% CI: 0.97-0.99). Marginal curves remained flat ([~]2-3% AE probability) across weekly cannabis exposure. Ad hoc analysis of non-missing-at-random data suggests possible AE rates are in line with current literature. DiscussionIn this large cohort, AEs were infrequent and mild, and weekly cannabis frequency did not independently increase odds. Healthcare engagement likely reflects underlying health complexity driving AE reporting. These findings support the safety of medical cannabis.