Magnitude of the Digital Placebo Effect and Its Moderators on Generalized Anxiety Symptoms: Systematic Review and Meta-Analysis.

Background: Digital therapeutics (DTx) have attracted attention as the substitutes or add-ons to conventional pharmacotherapy. The number of clinical trials for DTx has increased recently, and one of the main targets for DTx is psychiatric disorders. Generalized anxiety disorder (GAD) is one of the most common and notable psychiatric disorders, and it's known that the magnitude of placebo effect in the pharmacotherapy is quite large. The randomized controlled trials (RCTs) with digital placebos are the most reliable clinical trials to evaluate the safety and efficacy of DTx. However, the magnitude of the digital placebo effect and its moderators on GAD have not been investigated, although they are critical to assess the true treatment effect of DTx. Objective: The objectives of this study were to identify RCTs with digital placebos as comparators that evaluated GAD assessment scores, to review the characteristics of the RCTs and of the digital placebos in the systematic review, and to investigate the magnitude and its moderators in the meta-analysis. Methods: The RCTs evaluating the GAD assessment scores by setting digital placebos as comparators were identified by searching the database of PubMed, Web of Science, and Scopus in July 2024. The characteristics of the RCTs and of the digital placebos were reviewed systematically. The meta-analysis, including subgroup analyses and meta-regressions, was conducted to investigate the magnitude and its moderators of the digital placebos. Results: A total of 54 RCTs were included in the systematic review and 32 RCTs with 3 GAD assessment scores were included in the meta-analysis with a total of 5311 participants. The magnitude of digital placebos for all the included studies was small to moderate (Hedges g=0.28, 95% CI 0.18-0.38). The subgroup analyses showed the significant difference in the magnitude among target population (P=.03), placebo approach (P=.02), and baseline values (P=.02). The meta-regressions also indicated that the primary psychiatric patients in the target population (P=.01), "Removed" type in placebo approach (P=.04) and high baseline values (P =.02) were moderators for the magnitude of digital placebos. Conclusions: This study showed the small-to-moderate and statistically significant digital placebo effect on GAD assessment scores. Target population, placebo approach, and baseline values were also identified as the moderators of the placebo effect. It would be effective to create the study protocols for the DTx trials with digital placebos by considering the moderators identified in this study.