The association between obstructive sleep apnea and sleep spindles in middle-aged and older men: a community-based cohort study

Study Objectives: Sleep spindles show morphological changes in obstructive sleep apnea (OSA). However, previous small studies have limited generalizability, leaving associations between OSA severity measures and spindle metrics uncertain. This study examined cross-sectional associations between OSA severity measures and spindle metrics among a large population-based sample of men. Methods: Community-dwelling men with no previous OSA diagnosis underwent home-based polysomnography. All-night EEG (F4-M1) recordings were processed for artifacts and spindle events identified using previously validated algorithms. Spindle metrics of interest included frequency (Hz), amplitude (μV2), overall density (11-16 Hz), slow density (11-13 Hz), and fast density (13-16 Hz) (number/minute). Multivariable linear regression models controlling for demographic, biomedical, and behavioral confounders were used to examine cross-sectional associations between OSA severity measures and spindle metrics. Results: In adjusted analyses, higher apnea-hypopnea index (AHI/h, as a continuous variable) and percentage total sleep time with oxygen saturation <90% (TST90) were associated with decreased slow spindle density (AHI, B = -0.003, p = 0.032; TST90, B = -0.004, p = 0.047) but increased frequency (AHI, B = 0.002, p = 0.009; TST90, B = 0.002, p = 0.043). Higher TST90 was also associated with greater spindle amplitude (N2 sleep, B = 0.04, p = 0.011; N3 sleep, B = 0.11, p < 0.001). Furthermore, higher arousal index was associated with greater spindle amplitude during N2 sleep (B = 0.31, p < 0.001) but decreased overall density (B = -1.27, p = 0.030) and fast density (B = -4.36, p = 0.028) during N3 sleep. Conclusions: Among this large population-based sample of men, OSA severity measures were independently associated with spindle abnormalities. Further population studies are needed to determine associations between spindle metrics and functional outcomes.