0289 Sleep and Mortality in Older Adults: A Machine-Learning-Based Comparison with Other Risk Factors. (12th April 2019)

• Record Type: Journal Article
• Authors: Wallace, Meredith; Buysse, Daniel J; Stone, Katie; Redline, Susan; Leng, Yue; Ensrud, Kristine; Ancoli-Israel, Sonia; Hall, Martica H
• Is Part Of: Sleep. Volume 42(2019)Supplement 1
• Page Start: A117
• Page End: A118
• Publication Date: 2019-04-12
• Subjects: Sleep -- Physiological aspects -- Periodicals; Sleep disorders -- Periodicals; Sommeil -- Aspect physiologique -- Périodiques; Sommeil, Troubles du -- Périodiques; Sleep disorders; Sleep -- Physiological aspects; Sleep -- physiological aspects; Sleep Wake Disorders; Psychophysiology; Electronic journals; Periodicals; 616.8498
• DOI: 10.1093/sleep/zsz067.288
• Languages: English
• ISSNs: 0161-8105

Abstract

Abstract: Introduction: Sleep characteristics related to duration, timing, continuity, and sleepiness are associated with mortality in older adults, but are rarely considered in health recommendations. Examining the predictive ability of sleep as a multidimensional construct - rather than a series of separate characteristics—may clarify its importance and influence recommendations for measuring public health. We applied machine learning to: (1) establish the predictive ability of multidimensional self-reported sleep for all-cause and cardiovascular mortality relative to other established risk factors; and (2) identify which sleep characteristics are most predictive. Methods: The analytic sample includes N=8, 668 older adults (54% female) aged 65-99 with self-reported sleep characterization and longitudinal follow-up (≤15.5 years), aggregated from three epidemiological cohorts. We used variable Importance (VIMP) metrics from random survival forests to rank the predictive abilities of five domains and the individual measures they comprise. VIMPs > 0 indicate predictive variables/domains. Results: The predictive ability of the multidimensional sleep domain for all-cause mortality [VIMP (99.9% CI) = 0.94 (0.60, 1.29); 15 predictors] ranked below that of sociodemographic factors [3.94 (3.02, 4.87); 6 predictors], physical health [3.79 (3.01, 4.57); 10 predictors], and medications [1.33 (0.94, 1.73); 10 predictors] but above that of health behaviors [0.22 (0.06, 0.38); 4 predictors]. For cardiovascular mortality, multidimensional sleep was also a significant predictor [1.98 (1.31, 2.64)] and was ranked similarly among the domains. The most predictive individual sleep characteristics across outcomes were time in bed, napping, and wake-up time. Cohort-specific analyses including additional non-sleep measures that could not be harmonized across cohorts indicated our findings are robust. Conclusion: Multidimensional sleep is an important predictor of mortality that should be considered among other more routinely used predictors. Future research should develop tools for measuring multidimensional sleep, especially those incorporating duration, timing, and napping, and test mechanistic pathways through which these characteristics relate to mortality. Support (If Any): MrOS:HL071194;HL070848;HL070847;HL070842;HL070841;HL070837;HL070838;HL070839. SOF: AG05407;AR35582;AG05394;AR35584;AR35583. SHHS: U01HL53916:U01HL53931;U01HL53934;U01HL53937;U01HL53938;U01HL53940;U01HL53941;U01HL64360. Wallace: AG056331. NSRR: HL114473.