Please use this link to access this publication.
Epidemiologic surveys aim to estimate the population prevalence of cannabis use and cannabis use disorder. Prevalence estimates are important for understanding trends, such as the impact of policy change. Existing epidemiologic surveys have produced discrepant and potentially unreliable estimates. The current meta-analysis (PROSPERO CRD42022364818) aims to identify potential sources of unreliability in prevalence estimates of cannabis use and use disorder among the general population (age 12 and older). There was no specific hypothesis about overall prevalence estimate, but we expected significant variability (i.e., heterogeneity) in estimates based on factors such as country, year of data collection, and specific methodological factors (e.g., diagnostic instrument).
Systematic searches identified manuscripts and reports documenting nationally representative lifetime or past-year cannabis use disorder prevalence estimates. Meta-analysis was used to synthesize prevalence estimates, evaluate heterogeneity, and test moderators.
There were 39 manuscripts/reports included in analyses, which resulted in 259 unique prevalence estimates spanning 1980–2013 and an aggregated sample size of 973,281 individuals. Past-year and lifetime prevalence estimates for cannabis use were 12.83% (95% CI [11.15%, 14.71%]) and 38.31% (95% CI [35.92%, 40.76%]), and those for cannabis use disorder were 2.59% (95% CI [2.30%, 2.90%]) and 6.77% (95% CI [4.89%, 9.30%]), respectively. There was significant heterogeneity in estimates, which was partially explained by factors such as country, year of data collection, and methodological characteristics.
The significant heterogeneity in prevalence estimates as a function of methodological characteristics raises concerns about the generalizability of estimates. Recommendations for enhancing the validity and reliability of these estimates are offered.