No association between smoking and obstructive sleep apnea in men: evidence from middle-aged adults followed into later life.

RATIONALE: The relationship between smoking and obstructive sleep apnea (OSA) remains unclear. Observational findings have been inconsistent, and associations may vary depending on study population, design, measurements. OBJECTIVES: To evaluate the cross-sectional and longitudinal associations between smoking and OSA in a population based cohort of Korean Men. METHODS: We analyzed 1,437 Korean men in the Korean Genome and Epidemiology (KoGES) who underwent baseline polysomnography (2011-2014). In the longitudinal analyses, 780 participants without OSA at baseline were included for the primary AHI ≥ 15 outcome. Smoking status was categorized as never, past, and current at baseline. We estimated odd ratios [OR] and relative risks [RR] using logistic regression and modified Poisson regression with adjusted covariates such as age, income, drinking status, exercise, body mass index (BMI), hypertension (HTN), diabetes (DM), and cardiovascular disease (CVD). Longitudinal models evaluated incident OSA (mean follow-up, 7.85 years) among participants with follow-up data. Sensitivity analyses were performed using alternative exposure and outcome definitions. RESULTS: our main outcome at AHI ≥ 15, smoking status was not significantly associated with OSA in either cross-sectional or longitudinal analyses (current smokers, adjusted OR: 0.81, 95% CI: 0.54-1.22); adjusted RR: 0.65, 95% CI: 0.39-1.11) after multivariable analyses. Findings were consistent across alternative smoking definitions and extreme-case bound analyses. In contrast, sensitivity analyses using a lower AHI thresholds (≥ 5) suggested an apparent inverse association for current smokers although findings were not consistent across analyses. CONCLUSIONS: In this cohort of middle-to-older aged men, smoking status was not associated with 8-year incident OSA at AHI ≥ 15. These findings were generally similar across alternative smoking exposure. In contrast, additional sensitivity analyses using a lower threshold (AHI ≥ 5) may be artifactual. Overall, our findings suggest that smoking may have a limited role in the development of OSA in this population.