Toward a Global View of Alcohol, Tobacco, Cannabis, and Cocaine Use: Findings from the WHO World Mental Health Surveys

Alcohol, tobacco, and illicit drug use contribute substantially to global morbidity and mortality, yet comparable cross-national epidemiological data have been scarce, particularly in less established market economies. Earlier cross-national work relied on indirect measures such as taxation and mortality data and faced issues of comparability, especially for illicit drugs. Standardized survey methods have since been developed, but cross-national implementation has been limited. The WHO World Mental Health (WMH) Survey Initiative provides harmonized population surveys across multiple regions to address these gaps. This study aims to: (a) document lifetime (cumulative) use of alcohol, tobacco, cannabis, and cocaine across 17 countries, with focus on young adults; (b) examine sociodemographic correlates of these drug uses; and (c) characterize age-of-onset (AOO) distributions for initiation. The purpose is to provide a global view of basic epidemiological patterns of substance use to inform burden of disease estimates and policy.

The paper reviews historical challenges in cross-national disease and risk factor research, noting advances via international classifications (e.g., ICD) and standardized measurement of chronic and mental conditions. Traditional cross-national comparisons for alcohol and tobacco used national consumption and mortality correlations, limited by variability in certification and reliability—especially for illicit drugs. Population survey protocols for alcohol, tobacco, and drug use have improved, yet cross-national implementations remain difficult. Prior regional efforts include European surveys and the PACARDO study in Central America and the Dominican Republic, and compilations like WHO’s Global Status Reports on Alcohol and Tobacco, the Global Youth Tobacco Survey (GYTS), and GENACIS. However, few efforts have concurrently assessed alcohol, tobacco, and illicit drug use across diverse regions and age ranges. The WMH program fills this gap by applying harmonized methods across countries, including those with different drug policy regimes (e.g., US prohibition vs Netherlands harm reduction) and both cocaine source and consumer nations.

Design and setting: Eighteen household surveys in 17 countries across the Americas, Europe, Middle East/Africa, Asia, and Oceania followed the WMH protocol. Multistage probability samples were used (nationally representative in most countries; urban-only in Colombia and Mexico). Face-to-face interviews were administered by trained lay interviewers; France used telephone recruitment with supplemental in-person interviewing. Standardized translation, interviewer training, and quality control procedures were applied. Sample: Total n=85,052 adults; Part I included all respondents; Part II (n=43,249) assessed alcohol and illicit drug use and included all Part I cases with any core disorder plus a ~25% random subsample of others. Part II data were weighted for differential selection and poststratified to population sociodemographics; response rates ranged from 46% (France) to 88% (Colombia), average 70%. Measures: CIDI 3.0 assessed lifetime use of alcohol, tobacco (cigarettes, cigars, pipes), cannabis (marijuana, hashish), and cocaine. AOO of first use was collected for each drug except age at first tobacco smoking in New Zealand, Japan, France, Germany, Belgium, Netherlands, Italy, and Spain. Covariates (as of interview) included sex, age, education (none; some/finished primary; some/finished secondary; some college; college+), relationship status (never, previously, married/cohabiting), employment (homemaker, retired, other including unemployed, working/student), and equivalized family income (low: <0.5× median; low-average: ≤median; high-average: >median–≤3× median; high: >3× median). AOO elicitation used an enhanced question series to improve recall plausibility. Training and ethics: Bilingual supervisors were centrally trained; standardized informed consent and confidentiality procedures were used; local IRBs approved protocols. Analytic methods: Cumulative incidence (lifetime use) was estimated as the proportion ever using by interview age. AOO distributions and projected lifetime risk at ages were estimated using the two-part actuarial method (SAS 8.2), which improves timing within-year versus Kaplan–Meier; both assume constant conditional risk at a given age across cohorts. Discrete-time survival analysis (person-year units) estimated age-specific risk and predictors (sex, cohort, sex×cohort). Cohort was operationalized by age at interview (continuous and categorical: 18–29, 30–44, 45–59, ≥60). Cross-sectional associations of lifetime use at interview with sociodemographics were modeled via logistic regression. Standard errors used Taylor series linearization (SUDAAN). ORs and 95% CIs were reported; Wald chi-square tests evaluated significance at two-sided p<0.05.

• Lifetime prevalence (weighted cumulative incidence): Alcohol use was highest across the Americas, Europe, Japan, and New Zealand, and markedly lower in the Middle East, Africa, and China. Notable estimates: US 91.6%; Germany 95.3%; Ukraine 97.0%; Italy 73.5%; Israel 58.3%; Lebanon 53.3%; South Africa 40.6%; China 65.4. • Tobacco ever-use varied: highest in the US (73.6%), Lebanon (67.4%), Ukraine (60.6%), Mexico (60.2%), Netherlands (58.0); lowest in South Africa (31.9%) and Nigeria (16.8%). • Cannabis ever-use was highest in the US (42.4%) and New Zealand (41.9%), exceeding all other countries; mid-range levels in parts of Europe (e.g., Netherlands 19.8%, France 19.0%, Germany 17.5%, Spain 15.9%); very low in Asia (Japan 1.5%, China 0.3%) and Africa (Nigeria 2.7%). • Cocaine ever-use: US was an outlier at 16.2%; next highest around 4% in Colombia (4.0%), Mexico (4.0%), Spain (4.1%), New Zealand (4.3%); negligible in Middle East, Africa, and Asia. • Age at onset (AOO): Cross-country AOO distributions were remarkably similar. Median AOO for alcohol was 16–19 years (South Africa 20); for tobacco typically ~16–19 years, with exceptions Nigeria 21 and China 20; for cannabis, 18–19 years (Nigeria and Israel 22; Lebanon 21); for cocaine, 21–24 years where estimable. Interquartile ranges: tobacco 15–21, cannabis 16–22, cocaine 19–28. • Young cohort initiation (ages 22–29 assessed by ages 15/21): By age 21, alcohol initiation exceeded 90% in many European countries, Japan (92%), New Zealand (94%), and the Americas (78–93%), but was lower in Middle East/Africa (40–63%). Cannabis by 21: New Zealand 61.8%, US 54.0%, Netherlands 34.6%; minimal in Asia, Middle East, and Africa. Cocaine by 21: US 16.3%; Colombia 3.1%; Netherlands 1.0%. • Cohort effects: Discrete-time survival showed consistently higher initiation risk in more recent cohorts (per 10-year increase in age at interview, ORs <1 across drugs and countries; e.g., Mexico cocaine OR≈0.2, US cannabis OR≈0.6), indicating rising historical risk and extended initiation into adulthood for illicit drugs. • Sex differences and convergence: Women had lower odds than men to initiate all drugs overall, but sex×cohort interactions were <1.0 in most settings for alcohol and tobacco and in many for cannabis and cocaine, indicating narrowing male–female gaps in recent cohorts. Among 18–29-year-olds in several European countries, Japan, China, and New Zealand, sex differences in alcohol (and sometimes cannabis) initiation were minimal. • Sociodemographic correlates (pooled analyses): Younger age groups had much higher odds of cannabis and cocaine use versus ≥60 (e.g., adjusted ORs ~14–20). Females had consistently lower odds for all substances. Higher income was positively associated with both legal and illegal drug use. Being never or previously married was associated with higher odds of cannabis and cocaine (but not alcohol). Education was positively related to alcohol, negatively to tobacco, and largely unrelated to illicit drugs. Employment status showed lower odds among homemakers and retirees; some elevation in “other” for cocaine.

The study demonstrates substantial cross-national heterogeneity in lifetime alcohol, tobacco, cannabis, and cocaine use, with the US generally at the high end for both legal and illegal substances and much lower levels in Asia, the Middle East, and parts of Africa. Despite differing national policies, punitive user-level drug laws were not associated with lower illicit drug use at the country level (e.g., high US use vs lower use in the Netherlands with a harm reduction approach). Cohort analyses indicate that initiation risks have increased over historical time and that the period of risk for initiating illicit drugs now extends further into adulthood, implying that prevention strategies should not focus solely on adolescents. The consistent male excess in drug initiation is attenuating in younger cohorts across many countries, signaling sociocultural shifts in substance use patterns. Sociodemographic patterns—greater use among higher-income individuals and associations of illicit drug use with being never or previously married—mirror prior findings in developed countries and suggest that economic resources and social roles are important correlates. Together, these findings address the study’s objectives by mapping lifetime use patterns, identifying consistent sociodemographic associations, and charting AOO distributions that are remarkably similar across countries yet shifting across cohorts.

Using harmonized WMH surveys across 17 countries, the study provides a global view of lifetime alcohol, tobacco, cannabis, and cocaine use, documenting large cross-national differences, strong cohort increases in initiation (especially for illicit drugs), a lengthening window of risk into adulthood, and narrowing sex differences among younger cohorts. Drug policy stringency alone does not explain national differences in illicit drug use. Future research should expand country coverage, incorporate bioassays where feasible, and use longitudinal designs to refine age-of-onset estimates and examine past-year prevalence, trajectories, and transitions to problematic or dependent use.

• Country coverage was driven by collaborator availability and funding; some regions (e.g., Francophone West Africa) were not represented, limiting global generalizability. • Variable response rates and potential nonresponse bias likely yield underestimates of use, especially for illicit drugs; poststratification may not fully correct bias. • Self-reported use is subject to social desirability, stigma, and legal concerns that vary cross-nationally; despite standardized procedures and self-administered formats for sensitive items (where literacy allowed), differential underreporting may remain. • France used a telephone-based household sampling frame; some countries lacked fully national samples (urban-only in Colombia and Mexico); some surveys had age restrictions for certain analyses. • Retrospective recall of age of onset may be inaccurate, especially among older respondents; differential mortality could affect cohort comparisons (though unlikely to explain large illicit drug differences). • AOO for tobacco was not assessed in several countries (New Zealand, Japan, France, Germany, Belgium, Netherlands, Italy, Spain), limiting comparability. • Lack of comparable urbanicity measures precluded analysis of urban–rural differences. • Cross-sectional design limits causal inference for sociodemographic correlates, which were measured at interview, not at initiation.