(PS10-C53) Daily Relationships Between Worry and Sleep Disturbance: Disentangling Within- and Between-person Effects

Although numerous studies have examined the relationship between worry and sleep disturbance, whether these associations are due to between- or within-person effects remains unknown. Disaggregation of between- and within- person effects allow for the identification of relationships that occur within persons as well as relationships that occur across persons which is necessary for a complete understanding of the phenomenon of interest. The purpose of this study was to examine the transactional relationships between worry and sleep while disaggregating day- and person-level effects. In this study, we tested the hypothesis that day-level worry would predict day-level sleep and vice versa.

             Participants (N=182) in an ecological momentary assessment study completed self-report measures of worry and sleep for seven consecutive days. Worry was assessed by asking participants to indicate their level of worrisome thinking three times per day (0 = no worry at all, 6 = extreme amount of worry), and we calculated the mean level of worry per day. Sleep was assessed by asking participants to complete a sleep diary each morning which included items used to calculate four sleep indices: sleep quality (SQ), sleep onset latency (SOL), total sleep time (TST), and morning lingering (ML).

We fit a series of multilevel models regressing the four sleep outcomes onto worry. For each model, Level 1 included day-level worry, a control variable for day of the week (weekend vs weekday), and a control variable for previous night day-level sleep disturbance. Level 2 included person-level worry. We fit a series of tobit regression models regressing worry onto four sleep indices. For each model, we controlled for previous day-level worry.

Results of analyses examining worry as a predictor of sleep indicated that greater person-level worry significantly predicted longer SOL (Est = 3.71, SE = 1.44, p < 0.01), worse SQ (Est = -0.16, SE = 0.05, p < 0.01), and longer ML (Est = 7.24, SE = 1.21, p < 0.01). Day-level worry did not significantly predict SOL, SQ, or ML. Neither person-level worry (Est =-2.25, SE = 4.11, p > 0.05) nor day-level worry (Est =2.04, SE = 4.35, p > 0.05) predicted TST.

Results of analyses examining sleep as a predictor of worry indicated that greater person-level SOL (Est = 0.01, SE = 0.002, p < 0.01) and ML (Est = 0.03, SE = 0.002, p < 0.01) predicted higher levels of worry. Day-level SOL and ML did not predict worry. Greater person-level (Est = -0.47, SE = 0.05, p < 0.01) and day-level SQ (Est = -0.10, SE = 0.03, p < 0.01) was associated with less worry. Greater person-level TST (Est = 0.1, SE = 0.04, p < 0.01) was associated with more worry, and greater day-level TST (Est = -0.06, SE = 0.02, p < 0.05) was associated with less worry.

These findings suggest that those with higher levels of worry overall, regardless of their degree of worry on a given day, tend to have longer SOL, worse SQ, and longer ML. Notably, we did not find that day-level worry predicts sleep disturbance, suggesting that momentary variations in worry may not be as important as trait levels of worry in influencing sleep. Additionally, individuals who generally have greater sleep disturbance tend to worry more on a given night. Findings highlight the need to elucidate the complex bidirectional relationships between worry and sleep.