Tapia, Amanda

Abstract Introduction Comparing sleep and rest-activity rhythms (RARs) across the spectrum of cognitive aging can reveal novel risk factors and potential mechanisms underlying dementia disorders. However, our current understanding is restricted by differences in sleep measurement, limited longitudinal data, and heterogeneous cognitive aging processes. We aimed to determine (1) which sleep/RAR features exhibit meaningful changes within three distinct cognitive pathways; and (2) which sleep/RAR features differ between cognitive aging pathways prior to and after the onset of mild cognitive impairment (MCI) and/or dementia. Methods Longitudinal self-reported sleep and actigraphy data were obtained from 1,449 participants in the Rush Memory and Aging Project (mean age= 81.2±7.1 years, 75.2% female, 5.2% Black, 94.8% White). We applied flexible cubic spline models to quantify differences in the levels and trajectories of sleep amount (self-reported duration; actigraphy rest interval length and alpha), regularity (actigraphy interdaily stability, intradaily variability, and midpoint standard deviation), and timing (self-reported midpoint; actigraphy midpoint and acrophase) within and between three cognitive aging pathways over 12 years: Normal Cognitive Aging (‘Normal’), Progression to MCI (‘Stable MCI’), and Progression to Dementia (‘Dementia’). Models were adjusted for key sociodemographic and health variables. Results Sleep amount was lowest in the Dementia pathway prior to MCI or dementia, but increased with age, most rapidly following dementia. The greatest effect was observed in self-reported duration (standardized mean change, d[95%CI]=0.912 [0.47, 1.36]. Regularity declined across all pathways, most rapidly after cognitive diagnoses, with the largest increases in intradaily variability occurring among the Stable MCI (d=0.97 [0.70, 1.23]) and Dementia paths (d=1.11 [0.84, 1.38). Shifts toward earlier sleep timing were observed in actigraphy measures across all three pathways (-0.54≤d≤-0.31). Conclusion Shorter sleep amount in cognitively healthy older adults may be a risk factor or prodromal indicator of dementia, while longer sleep amounts and decreasing regularity following dementia may reflect neurodegeneration. Advances in sleep timing may be less useful as a clinical indicator of future cognitive impairment or decline. Differences between self-reported and actigraphy-measured sleep/RAR outcomes may reflect distinct underlying mechanisms, warranting further research. Support (if any) NIH: T32HL082610, RF1AG056331, R01AG052488, R01AG071638, RF1AG070436, R01AG17917