Time-Dependent Changes in Depressive Symptoms Among Control Participants in Digital-Based Psychological Intervention Studies: Meta-analysis of Randomized Controlled Trials

BACKGROUND: Digital-based psychological interventions (DPIs) have been shown to be efficacious in many randomized controlled trials (RCTs) in dealing with depression in adults. However, the effects of control comparators in these DPI studies have been largely overlooked, and they may vary in their effects on depression management. OBJECTIVE: This meta-analytical study aimed to provide a quantitative estimate of the within-subject effects of control groups across different time intervals and explore the moderating effects of control types and symptom severity at baseline. METHODS: A systematic literature search was conducted in late September 2021 on selected electronic databases: PubMed; ProQuest; Web of Science; and the Ovid system with MEDLINE, PsycINFO, and Embase. The control conditions in 107 RCTs with a total of 11,803 adults with depressive symptoms were included in the meta-analysis, and effect sizes (Hedges g) were calculated using the standardized mean difference approach. Study quality was assessed using the Cochrane risk-of-bias tool for randomized trials version 2. RESULTS: The control conditions collectively yielded small to moderate effects in reducing depressive symptoms within 8 weeks since the baseline assessment (g=−0.358, 95% CI −0.434 to −0.281). The effects grew to moderate within 9 to 24 weeks (g=−0.549, 95% CI −0.638 to −0.460) and peaked at g=−0.810 (95% CI −0.950 to −0.670) between 25 and 48 weeks. The effects were maintained at moderate to large ranges (g=−0.769, 95% CI −1.041 to −0.498) beyond 48 weeks. The magnitude of the reduction differed across the types of control and severity of symptoms. Care as usual was the most powerful condition of all and produced a large effect (g=−0.950, 95% CI −1.161 to −0.739) in the medium term. The findings showed that waitlist controls also produced a significant symptomatic reduction in the short term (g=−0.291, 95% CI −0.478 to −0.104), refuting the previous suspicion of a nocebo effect. In addition, a large effect on depressive symptom reduction in the long term (g=−1.091, 95% CI −1.210 to −0.972) was noted among participants with severe levels of depressive symptoms at baseline. CONCLUSIONS: This study provided evidence that depressive symptoms generally reduced over time among control conditions in research trials of DPIs. Given that different control conditions produce variable and significant levels of symptomatic reduction, future intervention trials must adopt an RCT design and should consider the contents of control treatments when investigating the efficacy of DPIs. The results of waitlist controls confirmed previous findings of spontaneous recovery among people with mild to moderate depressive symptoms in face-to-face studies. Researchers may adopt watchful waiting as participants wait for the availability of digital-based psychological services.