<![CDATA[Introduction: Adolescent smoking occurs within overlapping individual, interpersonal, school, and communication environments. Examining these domains together can show which associations persist after accounting for co-occurring measured exposures and can identify domains requiring closer surveillance and assessment in local health-promotion planning. This study examined factors associated with screening-defined self-reported current smoking among senior high school students in Semarang City. Methods: This pooled cross-sectional secondary analysis included 8,230 unique senior high school students from routine school-based smoking screening conducted in 2023–2025. Before anonymisation and analysis, the research team performed data cleaning and cross-year record linkage using National Identification Number (NIK) and personal email address as the primary identifiers, with name and school used for additional verification when required, so that each student contributed only one observation. The outcome was screening-defined self-reported current smoking based on the routine screening question “Do you smoke?”, which did not specify a recall period; “yes, every day” and “yes, occasionally” were classified as current smoking. Population-averaged adjusted associations were estimated using generalised estimating equations with a binomial distribution and logit link, Puskesmas (community health centres) as the clustering unit (29 clusters), robust sandwich standard errors, all six prespecified explanatory variables, and screening year. Results: Overall, 2,173 of 8,230 unique students (26.4%) met the screening definition of current smoking; the corresponding screening-defined prevalence was 25.6% in 2023, 26.9% in 2024, and 27.4% in 2025. After adjustment for all prespecified variables, screening year, and clustering within Puskesmas, higher odds were observed for male sex (aOR = 11.71; 95% CI: 8.89–15.45), reported observation of smoking at school (aOR = 1.37; 95% CI: 1.14–1.66), a family member at home who smoked (aOR = 1.97; 95% CI: 1.72–2.25), and reporting that many close friends smoked (aOR = 4.64; 95% CI: 4.01–5.36). The one-source category (aOR = 0.59; 95% CI: 0.40–0.86) and more than one source (aOR = 0.18; 95% CI: 0.13–0.26) were inversely associated with the outcome. Conclusion: The observed adjusted associations identify measured school, household, peer, sex, and communication domains that warrant attention in local surveillance and further assessment, but they do not establish causal pathways or the effectiveness of any intervention. The inverse gradient for the number of smoking-related information sources should be treated as exploratory because message content, timing, frequency, intensity, and direction were not measured. Longitudinal studies using standardised smoking definitions and richer exposure measurement are needed.]]>
