Wearable-derived sleep features predict relapse in Major Depressive Disorder.

and circadian function are leading candidate markers for early relapse identification in MDD. Consumer-grade wearable devices may offer opportunity for remote and real-time examination of dynamic changes in sleep. Objective: We used FitBit data from individuals with recurrent MDD to describe longitudinal associations of sleep duration, quality, and regularity with subsequent depressive relapse and depression severity.Design: Data were collected as part of a longitudinal remote measurement technologies (RMT) cohort study in people with recurrent MDD. Participants: A total of 623 people with MDD wore a FitBit and completed regular outcome assessments via email for a median follow-up of 541 days. Multivariable regression models tested for associations between sleep features and depression outcomes. We considered two samples of people with at least one assessment of relapse (n=213) or at least one assessment of depression severity (n=390). Results: Increased intra-individual variability in total sleep time, greater sleep fragmentation, and later sleep mid-points were associated with worse depression outcomes. Adjusted Population Attributable Fractions (PAFs) suggested that an intervention to increase sleep consistency in adults with MDD could reduce the population risk for depression by up to 18-37%. Conclusion: We found consistent associations between wearable-derived sleep features and the probability of depressive relapse and increased depressive symptom severity. Disordered sleep is prevalent and disruptive, and challenging to capture longitudinally via conventional laboratory sleep assessments. Our study demonstrates a role for consumer-grade activity trackers to predict relapse risk and depression severity in people with recurrent MDD.