0177 Measuring Sleep Health Across Survey, Actigraphy, and Diary Methods in Young Adults

Abstract

Abstract Introduction Sleep health is a relatively new construct including multiple dimensions of nighttime and daytime sleep across the 24-hour day. To date, few studies have investigated sleep health across various types of sleep measures. We collected sleep health data using survey, actigraphy, and daily diary, affording the unique ability to compare the latent factor structure of the sleep health construct across methods in a single sample of healthy young adults. Methods Undergraduates (N=540; 50% female; 71% white; mean age=18) self-reported sleep health via the 6-item RUSATED survey (Buysse, 2014), which queries “typical” sleep patterns including: daytime alertness and nocturnal sleep regularity, satisfaction, timing, efficiency, and duration (0=rarely/never to 2=usually/always); total score=0–12 (higher=better sleep health). A subsample (N=114) provided 7-day actigraphy and daily diaries, which were used to derive weeklong averages of the aforementioned sleep dimensions. Confirmatory factor analysis was conducted to assess model fit of three factors (survey, actigraphy, diary), each including within-method indicators of the six sleep dimensions; note the actigraphy model included diary satisfaction. Acceptable model fit was assessed using established criteria (e.g., Bentler, 1990): RMSEA < .08, CFI > .90, and TLI > .90. Maximum likelihood estimation was used. Results Average RUSATED total score was 7.1 (SD=2.5). Average actigraphy [diary] sleep dimensions were: Alertness (i.e., proportion of days napped)=21% [25%]; Duration=6.2hr [7.1hr]; Timing (i.e., midpoint)=5:00am [3:30am]; Regularity (i.e., SD of midpoint)=56 min [40 min]; Efficiency= 82% [95%]; Satisfaction=“average” to “good”. The survey model demonstrated acceptable fit (RMSEA [90%CI] = .06 [.03, .09], CFI=.96, TLI=.92) after correlating regularity and timing, as suggested by modification indices; factor loadings [λ(SE)] ranged from .27(.05) for efficiency to .71(.04) for satisfaction. Fit was poor for the diary model, whereas the actigraphy model failed to converge. Conclusion Sleep health models demonstrated adequate fit using survey but not actigraphy or diary data. The lack of acceptable fit for the latter may reflect differences in the measurement timeframe (e.g., “typical” sleep vs. 7-day averages) or the smaller sample size for the actigraphy/diary measures. Determining how to best measure sleep health and ultimately apply it to health-relevant outcomes is a valuable research agenda. Support T32HL07560; T32MH018269