Stability Of Actigraphy-measured Sleep Characteristics Among Adults

PURPOSE: Epidemiological studies of sleep using actigraphy often rely on a single week measure of sleep to assess associations with prospective outcomes. However, it remains unclear whether sleep characteristics obtained from 7-day aggregates for estimating long-term sleep patterns (e.g., 12 months) are stable over time. METHODS: Participants were 3147 UK Biobank participants aged 40-69 yrs (52% women, 57% Overweight/Obese) that completed a 7-day accelerometry protocol and 3-5 yrs later, repeated the 7-day accelerometry protocol in 4 additional occasions, 3 months apart. Actigraphy-based sleep characteristics were scored using the Van Hees 2015 sleep algorithm including, sleep duration (hrs/night), wake after sleep onset (WASO; hrs/night), midpoint of sleep, and social jetlag (i.e., absolute difference in weekend vs. weekdays midpoints of sleep; hrs/week). We computed Intraclass Correlation Coefficients (ICCs) and 95% confidence intervals of 7-day aggregates to characterize, 1) short-term stability (i.e., 3 months apart) and 2) long-term stability (i.e., 3-5 yrs apart), of sleep characteristics. RESULTS: Participants slept for 6.7 ± 0.9 hrs/night and spent 0.8 ± 0.4 hrs/night awake after sleep onset. The midpoint of sleep was 3:30 am and participants had on average 0.9 hrs/week of social jetlag. The ICCs were similar for short- and long-term stability and ranged from 0.6 to 0.7 except for social jetlag that had ICCs of approximately 0.20. ICCs for short-term stability were 0.68 (0.67, 0.70) for sleep duration, 0.69 (0.67, 0.70) for WASO, 0.65 (0.63, 0.67) for midpoint of sleep, and 0.21 (0.18, 0.24) for social jetlag. Long-term ICCs were 0.63 (0.60, 0.65) for sleep duration, 0.64 (0.62, 0.67) for WASO, 0.59 (0.56, 0.61) for midpoint of sleep, and 0.24 (0.19, 0.29) for social jetlag. Both short- and long-term stability were similar among men, women, and adults of different age groups. CONCLUSION: The results suggest that sleep duration and indicators of sleep timing and quality are moderately stable over long periods of time (e.g., 3 yrs). The poor stability of social jetlag suggests that a single 7-day aggregate actigraphy measure of this metric may not adequately characterize long-term social jetlag in etiologic studies of adults.