Association of nocturnal oxygen saturation with coronary artery calcification: cross-sectional evidence from the population-based SESSA study among Japanese men

Objective We investigated the associations of nocturnal oxygen saturation (SpO 2 ) with prevalence of coronary artery calcification (CAC) and its level of burden, and whether the associations differ by obesity and mediate through diabetes mellitus (DM), dyslipidaemia and hypertension. Design Observational study. Setting Kusatsu, Shiga. Participants We analysed data from 510 Japanese men (aged 46–76 years) with no history of myocardial infarction. Primary and secondary outcomes Mean SpO 2 , minimum SpO 2 and oxygen desaturation index (ODI) were measured using overnight pulse oximetry (primary exposures), while CAC was assessed using CT. Prevalence of CAC (>0) and its burden (CAC 0, >0–100, >100–400 and>400) were evaluated as outcomes. Body mass index≥25.0 kg/m 2 defined as obesity. Results Mean±SD age was 66.7±7.2 years. The prevalence of CAC was 64.7% (CAC scores>0–100, 31.4%; >100–400, 20.0% and>400, 13.3%). In multivariable binary logistic regression, the OR and 95% CI for the prevalence of CAC were 1.25 (1.02 to 1.53) per 1-SD decrement in mean SpO 2 and 1.25 (1.01 to 1.55) per 1-SD increment in ODI. The associations lost their significance while further adjusted for DM and dyslipidaemia. Similar trends were observed for the level of CAC burden in multivariable ordinal logistic regression. Obesity did not show significant interaction with SpO 2 on CAC. In mediation analysis, the OR (95% CI) for natural indirect effect; percentage mediated through dyslipidaemia in association of 1-SD decrement in mean SpO 2 with prevalence of CAC were 1.06 (1.01 to 1.10); 25.4%. These estimates for mediation through DM and dyslipidaemia for 1-SD increment in ODI in the associations were 1.07 (1.01 to 1.12); 29.6% and 1.04 (1.00 to 1.08); 17.5%, respectively. Conclusions Lower mean SpO 2 and higher ODI are associated with CAC among Japanese men, independent of age, lifestyle factors and obesity. The associations became non-significant after adjusting for diabetes and dyslipidaemia, but were mediated through these factors.