Abstract
The coronavirus disease 2019 (COVID-19) has disrupted multiple domains of life including sleep. The present study used a longitudinal dataset (N = 671) and a person-centered analytic approach - latent profile analysis (LPA) - to elucidate the relationship between sleep and depression. We used LPA to identify profiles of sleep patterns assessed by Pittsburg Sleep Quality Index (PSQI) at the beginning of the study. The profiles were then used as a predictor of depression magnitude and variability over time. Three latent profiles were identified (medicated insomnia sleepers [MIS], inefficient sleepers [IS], and healthy sleepers [HS]). MIS exhibited the highest level of depression magnitude over time, followed by IS, followed by HS. A slightly different pattern emerged for the variability of depression: While MIS demonstrated significantly greater depression variability than both IS and HS, IS and HS did not differ in their variability of depression over time. Medicated insomnia sleepers exhibited both the greatest depression magnitude and variability than inefficient sleepers and healthy sleepers, while the latter two showed no difference in depression variability despite inefficient sleepers' greater depression magnitude than healthy sleepers. Clinical implications and limitations are discussed.