AGREEMENT BETWEEN SURVEY AND ACTIGRAPHY-ASSESSED SLEEP TIMING AMONG ADOLESCENTS IN THE FAMILY LIFE, ACTIVITY, SUN, HEALTH, AND EATING (FLASHE) STUDY

Abstract Introduction Epidemiologists often deploy questionnaires or wearable monitors to quantify sleep. The extent to which self-report and device-derived sleep measures agree among adolescents are not well known. This study describes the agreement between survey and actigraphy-assessed sleep timing among adolescents in the 2014 Family Life, Activity, Sun, Health, and Eating (FLASHE) study. Methods FLASHE survey and motion substudy participants provided data for analyses. FLASHE participants (12–17y) completed a survey that captured self-reported usual sleep and wake times on weekdays and weekends. A subset of participants also wore an ActiGraph GT3X+ accelerometer on the wrist for 24-hours for seven days and completed a daily log for time in/out of bed. Actigraphy-assessed sleep periods were estimated using the Sadeh algorithm. Total sleep time (TST) and sleep midpoint for weekday and weekend, social jetlag, and chronotype were computed for survey and actigraphy and means and standard deviations were examined. Paired t-tests were used to test mean differences between measures overall and stratified by sex and school level (middle vs. high school). Results The analytic sample included 372 U.S. adolescents: 49% were female, 53% were high schoolers, 28% had overweight/obesity, and 62% identified as non-Hispanic White. Compared to actigraphy, surveys overestimated TST by an average of 2.6h on weekdays (8.7h [SD:1.4h] vs. 6.1h [SD:1.5h], p<0.001) and 3.1h on weekends (9.5h [SD:1.5h] vs. 6.4h [SD:1.8h], p<0.001), and resulted in earlier weekday sleep midpoints (02:49 [SD:1h:24 min] vs. 03:40 [SD:1h, 42 min], p<0.001); differences were not significant for weekend midpoints. Survey estimates were larger than actigraphy for social jetlag (1.7h [SD:1.2] vs. 0.9h [SD:1.3], p<0.001) and resulted in an earlier chronotype (04:16 [SD:1h, 38min] vs. 04:30 [SD:1hr, 43 min], p=0.004). Findings remained consistent when stratified by sex and school level except for chronotype: estimates were not significantly different among females or high schoolers. Conclusion The large discrepancies between survey and actigraphy-based sleep timing highlight the importance of understanding what type of data each assessment method is capturing in adolescents (self-report vs. objective measures; average vs. daily sleep). Differences between self-report and device-derived sleep data should be considered during study development and when comparing results across studies. Support (If Any)