Abstract
BACKGROUND AND AIMS: Cannabis use has high prevalence and health burden. Although the effects of cannabis use have been studied in the literature, no systematic review and meta-analysis has measured its association with all-cause mortality. The aim of this systematic review and meta-analysis was to systematically synthesize the evidence on association between cannabis use and all-cause mortality. METHODS: Following the preregistered protocol (PROSPERO: CRD42023396915), we searched in Scopus, PubMed, Web of Science and ProQuest databases until end of October 2023. We included cohort studies comparing individuals using versus not using cannabis and measuring the association with all-cause mortality. A random-effect meta-analysis was conducted calculating the risk ratio (RR) and 95% confidence interval (CI). Heterogeneity and publication bias were measured. Sensitivity and meta-regression analyses were conducted. The Newcastle Ottawa Scale (NOS) was used to assess study quality. RESULTS: Fourteen cohort studies were included (prospective 50%), reporting on 17,545,076 participants (3,000,667 people who use cannabis [PWUC]). The overall RR estimation of all-cause mortality among PWUC versus nonusers was 1.53 (95% CI: 1.09; 2.14, I (2): 98%; τ (2): 0.38). Significantly different RR was observed in prospective versus retrospective designs (2.07 vs. 1.11); cohorts of the general population versus patients (2.53 vs. 1.03). Study sample size was a significant moderator of the association between cannabis use and all-cause mortality, with larger sample size being associated with smaller effect size and less heterogeneity. Based on GRADE assessment, observational evidence, with unadjusted estimates, high heterogeneity with inconsistent results, the overall certainty of evidence seems to be low. CONCLUSION: Cannabis use was associated with an increased risk of all-cause mortality in the general population but not in patients with severe underlying medical co-morbidities. It should be noted that the evidence may currently be biased and new methodologically strong studies need to be conducted.