Study objectives: Sleep duration has been shown to associate with cognitive function, but little is known about the short-term effect of sleep duration on the previous night. This study examines how usual sleep duration and previous night's sleep duration are associated with cognitive function in older adults.

Methods: The study population consisted of 2949 adults aged 59-92 years (mean 72.6, SD 5.7) derived from three Finnish cohorts. Participants' self-reported usual sleep duration was categorized into short (< 7 h, 19%), mid-range (7- < 9 h, 64%), and long (& GE; 9 h, 17%). Self-reported sleep duration on the night prior to cognitive testing was categorized into shorter (59%), same (35%), and longer (5.9%) than usual sleep duration. Computerized Cambridge Neuropsychological Test Automated Battery (CANTAB & REG;) was used to assess: (1) learning and memory, (2) working memory, (3) information processing, and (4) reaction time.

Results: Participants with self-reported long, but not short, usual sleep duration had poorer learning and memory (p = .004), information processing (p = .003), and reaction time (p = .006) when compared to those with mid-range sleep duration. Those who slept more than usually the night prior to cognitive testing had poorer information processing (p = .019) than those sleeping the same as usually, while sleeping less than usually was not associated with cognitive function.

Conclusions: This study suggests that while long sleep duration was associated with worse cognitive function, sleeping more than usually the night prior to cognitive testing was only associated with information processing, and sleeping less than usually is not associated with cognitive function.