549 Associations of Bedtime, Wake-time and Employment Status by Gender and Race

Introduction Poor sleep quality has been reported in the unemployed compared with employed. How sleep varies by employment status has been rarely examined at a population level. Therefore, we investigated sleep-wake patterns among employed, unemployed but actively seeking a job, and not-in-the-labor-force participants by gender and race/ethnicity. Methods Methods We used data from the American Time Use Survey (ATUS), a nationally representative sample of US residents aged ≥15years, which records weekday/weekend activities in a 24-hour period (4:00am-4:00am). This sample was restricted to participants aged 25–60 years (n=130,062). This analysis utilized functional nonparametric regression based on dimension reduction and neighborhood matching. We modeled the relationship between participant-specific sleep-wake trajectories, coded by minute, and employment status. Implementing the counterfactual approach, we estimated the effects of each employment scenario on participant-level expected sleep trajectory. This approach allowed the examination of hypothetical sleep-wake trajectories for each participant if their employment status differed from the observed. We then marginalized these findings to gender and race/ethnic subpopulations, controlling for confounders and secular trends. Results Mean age was 42□0.01 years, nearly half (51%) of participants were women and 68% were Whites. The proportions of employed, unemployed, and not-in-the-labor-force were 79%, 16.5% and 4.5%, respectively. On average, unemployed and not-in-the-labor-force participants had a later bedtime and wake-time compared with employed. With the exception of Whites, each individual race/ethnicity group showed pronounced differences in sleep-wake patterns by employment status. Of note, the likelihood of still being asleep up to 9:00am was greater when unemployed in comparison to had they been employed. Compared with employed, differences in sleep-wake patterns were pronounced among Blacks and Hispanics had they been unemployed, but attenuated if they were out-of-the-labor-force. Gender alone was not a strong moderator of the relationship between sleep-wake patterns and employment status. Unemployed participants had bedtime after 11pm, regardless of gender or race/ethnicity. Conclusion Using the counterfactual approach, we predicted sleep-wake patterns among individuals had they been employed, unemployed, or out-of-the-labor-force by gender and race/ethnicity. Though cross-sectional, our data