UC Merced

Background: The World Health Organization has officially recognized inadequate sleep as a public health issue, yet 30% of Americans do not meet the minimum requirement for sufficient sleep. Digital interventions delivered through websites and smartphone apps are an increasingly prevalent tool to address inadequate sleep, although there is mixed evidence on their efficacy.Objective: This study aimed to answer the following: (1) Are digital interventions aimed at promoting sleep efficacious in improving sleep outcomes? (2) Is the efficacy of digital interventions on sleep outcomes moderated by sleep dimension, hygiene, or measurement, along with other study characteristics? (3a) Is the efficacy of digital interventions on sleep outcomes mediated by self-efficacy? (3b) Do digital interventions using self-efficacy behavior change techniques (BCTs) lead to changes in sleep outcomes? Method: A systematic review and meta-analysis following PRISMA guidelines examined articles on randomized controlled trials for sleep-promoting digital interventions retrieved from three scientific databases. Intervention effect sizes (Cohen’s d), method of sleep measurement (self-report or electronic), and self-efficacy BCTs (the eight BCTs identified by the Human Behavior Change Project as ‘linked’ to changes in self-efficacy) were extracted from all studies. The average bias-corrected effect of digital interventions on sleep outcomes was computed using multi-level meta-analysis (RQ1). Effects of key moderators including method of sleep measurement (RQ2) and the number of self-efficacy BCTs used in an intervention (RQ3) were tested using meta-regression. Results: Forty samples met eligibility criteria. Digital interventions had a moderate-to-large effect size on sleep outcomes, Cohen’s d = 0.670, SE = 0.103, k = 193, t(192)=6.519, p < .001, 95% CI [0.467, 0.872]. Sleep dimension, method of measurement, mode of intervention delivery, and intervention focus significantly moderated the main effect, while sleep hygiene, number of self-efficacy BCTs, funding source, name of digital intervention program, intervention length, health condition at baseline, and comparison group, did not. We were unable to test if the construct self-efficacy mediated the main effect due to insufficient reporting of data necessary to run the analysis. Conclusions: The current study contributes to a growing body of research finding that digital health interventions are an effective tool to improve a range of health behaviors, including sleep. We found evidence that the way sleep is defined and measured can significantly affect the reported efficacy of a digital intervention on sleep, and implications and future directions for all moderators are discussed.