Determinants of environmental tobacco smoke exposure in Chile: a national study using complex survey data and robust count models

Environmental tobacco smoke (ETS) remains a significant public health concern in Chile despite the implementation of comprehensive smoke-free policies. This study aimed to identify sociodemographic, behavioral, and biochemical factors associated with daily ETS exposure across different environments among Chilean adults. A cross-sectional analysis was conducted using data from the 2016–2017 Chilean National Health Survey (ENS), a nationally representative survey with a complex multistage sampling design. The analytical sample included 6,233 individuals aged 15 years or older, of whom 5,520 had biochemical measurements; missing data were addressed using multiple imputation. Six outcome variables representing daily hours (0–24) of ETS exposure were analyzed across home, workplace, and other environments on weekdays and weekends. Given the high proportion of zero values (~88.5%) and evidence of overdispersion (dispersion = 6.2), Poisson, negative binomial, and zero-inflated Poisson models were formally compared using likelihood-based criteria, while quasi-Poisson models were used exclusively to assess overdispersion and the robustness of standard errors. Survey-weighted generalized linear models were fitted, and exponentiated coefficients were interpreted as mean ratios (MR). Negative Binomial models provided the best fit (AIC = 6,300 vs. 11,470 for Poisson), with consistent findings across sensitivity analyses. ETS exposure was higher among men, urban residents, and individuals reporting insomnia. Age was inversely associated with exposure (MR ≈ 0.99 per year), indicating lower expected exposure hours among older individuals. Women showed lower exposure compared to men (MR = 0.69; 95% CI: 0.55–0.87), and rural residence was associated with reduced exposure (MR = 0.61; 95% CI: 0.45–0.84). Insomnia was associated with higher exposure (MR = 1.74; 95% CI: 1.10–2.73). Descriptive analyses indicated that exposure was highest in “other environments,” particularly during weekends. These findings suggest that ETS exposure in Chile is unevenly distributed and shaped by sociodemographic and behavioral factors. Accounting for complex survey design and distributional characteristics of exposure data improves the validity and interpretability of results. Targeted public health strategies focusing on high-risk groups and environments are needed, while integrating behavioral and contextual determinants into tobacco control policies.