Intensive Longitudinal Approaches to Multifaceted Sleep and Depression in Shift Work

Abstract

Prior epidemiological research has established that sleep disturbances and depression are associated. However, notable methodological limitations exist across past studies, including the heavy use of between-person analyses that are often based on sleep and depression measured at a single point in time, use of general “sleep disturbance” over well-defined facets of sleep, overreliance on self-reported sleep, and lack of attention to within-person relationships and their interactions with group-level characteristics (e.g., shiftwork). To address these limitations, the current study examined relationships between multiple sleep facets and depression across various types of measurement (i.e., single time-point, daily; self-report, actigraphy) and moderators (i.e., shiftwork, intraindividual variability of sleep midpoint) in nurses. Participants were 349 nurses (90.83% female; 76.50% White; 89.20% non-Hispanic/Latinx; Mage = 39.13 years) who completed baseline measures of insomnia and depression. Participants then completed daily diaries assessing sleep and depression for 14 days, during which they also wore an Actiwatch (Phillips Respironics, Andover, MA). Simple linear regression was used to examine the relationship between insomnia and depression at a single point in time, while multilevel modeling was used to examine daily, bidirectional relationships between depression and sleep facets (i.e., sleep efficiency, quality, duration, midpoint) across self-report and actigraphy measures. Shiftwork and intraindividual standard deviation (iSD) of sleep midpoint were included as moderators in models. Results revealed higher insomnia severity was significantly associated with higher depression severity (b = -0.35, SE = 0.01, p < .001) at a single point in time, even after covarying for age, gender, race, ethnicity, shiftwork, and midpoint iSD. Daily depression and subjective sleep quality and efficiency were bidirectionally associated (all p’s <.05), even after adding in covariates. Specifically, on days when nurses experienced higher depression than their average, they experienced worse subjective sleep quality (b = -0.10, SE = 0.02, p < .001) and efficiency (b = -0.47, SE = 0.18, p = .008) on the same day. In turn, when nurses experienced worse subjective sleep quality (b = -0.07, SE = 0.02, p < .001) and efficiency (b = -0.01, SE = 0.00, p = .012) than their average, they experienced higher depression on the next day. Daily actigraphy-based sleep facets were not significantly associated with depression. Neither shiftwork nor iSD of sleep midpoint were significant moderators. Results replicate previous cross-sectional findings on sleep and depression and extend research by revealing bidirectional, daily associations between subjective sleep and depression in nurses. Actigraphy-based sleep variables did not reach statistical significance, suggesting that one’s perception of sleep may be more influential on and impacted by depression than how one more objectively slept. Although night shift workers compared to day workers and those with greater variability in sleep timing reported worse sleep facets, shiftwork and sleep variability did not moderate the cross-sectional or daily relationships between sleep and depression. Intensive longitudinal approaches to understanding sleep and depression should continue to be utilized to allow for the establishment of temporal precedence and bidirectionality across granular time scales (e.g., daily).