Trends in sleep duration in the U.S. from 2004 to 2018: A decomposition analysis

Average sleep duration in the United States declined in recent years, and the decline may be linked with many biopsychosocial factors. We examine how a set of biopsychosocial factors have differentially contributed to the temporal trends in self-reported sleep duration across racial groups between 2004-2005 and 2017-2018. Using repeated nationally representative cross-sections from the National Health Interview Survey, we decompose the influence of biopsychosocial factors on sleep duration trends into two components. One component corresponds to coefficient changes (i.e., changes in the associations between behaviors or exposures and sleep duration) of key biopsychosocial factors, and the other part accounts for the compositional changes (i.e., changes in the distributions of exposures) in these biopsychosocial factors during the study period. We reveal that changes in the coefficients of some biopsychosocial factors are more important than compositional changes in explaining the decline in sleep duration within each racial/ethnic group. Our findings highlight racial differences manifest across multiple biopsychosocial domains that are shifting in terms of association and composition. Methodologically, we note that the standard regression approach for analyzing temporal trends neglects the role of coefficient changes over time and is thus insufficient for fully capturing how biopsychosocial factors may have influenced the temporal patterns in sleep duration and related health outcomes.