Sleep macroarchitecture but not obstructive sleep apnea is independently associated with cognitive function in only...

Evidence linking obstructive sleep apnea with cognitive dysfunction predominantly comes from clinical or select community samples. We investigated the independent cross-sectional association of obstructive sleep apnea and sleep macroarchitecture parameters with cognitive function in unselected community-dwelling middle-aged and older men. Four hundred and seventy-seven Florey Adelaide Male Ageing Study participants underwent successful home-based polysomnography. They also completed cognitive testing, including the inspection time task, Fuld object memory evaluation, trail-making test A and B, and mini-mental state examination. Multivariable regression models examined independent cross-sectional associations of obstructive sleep apnea and sleep macroarchitecture parameters with cognitive function. In univariable analyses, a higher apnea-hypopnea index and percentage of total sleep time with oxygen saturation <90% were associated with worse trail-making test A performance (both p < .05). A higher apnea-hypopnea index was also associated with worse trail-making test B performance and slower inspection time (both p < .05). In adjusted analyses, obstructive sleep apnea and sleep macroarchitecture parameters were not associated with cognitive function (all p > .05). In age-stratified analysis in men ≥65 years, greater stage 1 sleep was independently associated with worse trail-making test A performance, whereas greater stage 3 sleep was independently associated with better trail-making test A performance (both p < .05). Our findings suggest that obstructive sleep apnea is not independently associated with cognitive function. In older, but not younger, men, light sleep was associated with worse attention, whereas deep sleep was associated with better attention. Longitudinal population-based cohort studies are needed to determine if obstructive sleep apnea and disrupted sleep macroarchitecture independently predict prospective cognitive dysfunction and decline.