A hidden epidemic? Prevalence and factors associated with shisha smoking in Africa: A multicountry analysis of WHO STEPS survey data (2014-2017)

ABSTRACT Background: Shisha (waterpipe) smoking is increasingly visible in African social settings, yet comparable multicountry evidence remains limited. We estimated current shisha-use prevalence and examined demographic, behavioral, health, and geographic factors associated with use across 10 African World Health Organization (WHO) STEPS surveys. Methods: We conducted a pooled cross-sectional analysis of WHO STEPS data collected between 2014 and 2017 among adults aged 18–69 years. To preserve each survey’s within-country design while avoiding arbitrary domination by country-specific weight scales, sampling weights were rescaled to a mean of one within country before pooling; strata and primary sampling units were nested within country. Design-based prevalence estimates and survey-weighted logistic regression models were used. Covariates were prespecified. Survey year was descriptive only, and FCTC/PEITTP status was treated as contextual rather than as an individual-level causal exposure. Results: The final analytic sample comprised 2113 participants, including 171 current shisha users. The pooled country-normalized survey-weighted prevalence was 8.1% (95% CI 6.5%–10.0%). Descriptively, prevalence was highest among participants aged 18–19 years (25.1%), females (23.4%), urban residents (10.7%), and never-married adults (15.7%). Country-specific weighted prevalence ranged from 0.6% in Malawi to 19.7% in Sudan. In adjusted analyses, males had markedly lower odds than females (AOR = 0.11, 95% CI 0.05–0.23, P < 0.001), and currently married adults had lower odds than never-married adults (AOR = 0.45, 95% CI 0.25–0.82, P = 0.009). Lower odds were also observed in the 25–29, 45–49, 50–54, and 55–59 year groups relative to ages 18–19. Compared with Algeria, Sudan had higher adjusted odds (AOR = 2.38, 95% CI 1.14–4.94, P = 0.020), while Malawi (AOR = 0.13, 95% CI 0.03–0.51, P = 0.004) and Uganda (AOR = 0.20, 95% CI 0.04–0.97, P = 0.046) had lower odds. Residence, education, occupation, cardiometabolic factors, diet, physical activity, and harmful alcohol use were not independently associated after adjustment. None of the four FCTC/PEITTP policy indicators was statistically significant in the crude survey-weighted policy analysis. Conclusion: Current shisha use was substantial but highly heterogeneous across these 10 surveys. The clearest independent patterns were age, sex, marital status, and country. Prevention and surveillance should prioritize younger populations while remaining responsive to country context and emerging patterns among women. The cross-sectional data do not support causal inference about tobacco-control treaty effects.