Cannabis use disorder and risk of venous thromboembolism: a retrospective cohort study

Abstract Background Cannabis use has reached historic highs worldwide, yet its long-term vascular safety remains uncertain. Evidence linking cannabis to venous thromboembolism derives largely from hospitalised trauma and surgical cohorts and is inconsistent. We examined whether cannabis use disorder is associated with incident venous thromboembolism over 10 years. Methods Retrospective cohort study in the TriNetX US Collaborative Network, a federated electronic health record database. Adults aged ≥ 18 years with ≥ 2 encounters during 2014 were eligible, and cohort entry was the first emergency department visit in 2015. Cannabis use disorder required ≥ 2 abuse or dependence diagnoses during 2014 and ≥ 1 within the year preceding cohort entry; controls had no cannabis-related diagnosis recorded before entry. Patients were matched 1:1 on 36 prespecified covariates by propensity score and followed from day 90 to day 3,650. Hazard ratios were estimated with Cox models. Prespecified analyses varied the follow-up window, landmark period and exposure definition, and included positive control outcomes. Results Of 5,162 exposed and 1,035,374 unexposed patients, 5,031 pairs were matched, with all standardized mean differences 0.06 or below and mean follow-up of 7.8 and 7.6 years. Cannabis use disorder was not associated with venous thromboembolism over 10 years (hazard ratio 1.01, 95% confidence interval 0.85 to 1.21), deep vein thrombosis (1.06, 0.85 to 1.32), pulmonary embolism (0.97, 0.78 to 1.20) or all-cause mortality (1.00, 0.91 to 1.11). Before matching, cannabis use disorder was associated with higher hazards of venous thromboembolism (1.14, 1.02 to 1.29), pulmonary embolism (1.33, 1.15 to 1.54) and all-cause mortality (1.23, 1.15 to 1.32). Estimates were similar across 1-year and 5-year windows, a 30-day landmark period, alternative exposure definitions and prespecified subgroups, whereas cyclical vomiting syndrome, a positive-control proxy for cannabinoid hyperemesis syndrome, was strongly associated (1.99, 1.44 to 2.74). Conclusions Among adults presenting for emergency care, cannabis use disorder was not associated with an increased 10-year risk of venous thromboembolism. The crude association was attenuated after matching on comorbidity and concurrent substance use. These findings do not exclude a modest effect and do not extend to post-traumatic or perioperative settings.