OBJECTIVES: This study systematically reviewed and compared economic evaluation literature on digital health interventions and drug interventions for treating depression.

METHODS: We searched for articles published between October 2013 and October 2023 using the Ovid-MEDLINE, Embase, Cochrane Library, and PsycINFO databases. We conducted a qualitative comparison of the costs, effects, and modeling components of each intervention. Cross-tabulation analysis was used to examine the frequency of different components in each type of intervention. The CHEERS checklist evaluated the reporting quality of the selected literature and generalized linear model analysis investigated factors affecting this quality.

RESULTS: We selected 42 articles: 23 on digital health interventions and 19 on drug interventions. Qualitative analysis revealed that direct medical expenses for digital health interventions included license fees or program maintenance costs, while drug interventions included costs related to suicide or suicide attempts. Quantitative cross-tabulation analysis showed significant differences in economic evaluation components based on intervention type, including the type of economic evaluation (CUA, CEA or other types) (p=0.005), the type of comparative alternative (active control, placebo control or other) (p=0.000) and funding sources (government, company or other) (p=0.002). Generalized linear model analysis indicated that the modeling approach (model-based or trial-based) (OR=1.11, p=0.012) and the type of economic evaluation (OR=1.21, p=0.002) significantly influenced the CHEERS scores.

CONCLUSIONS: This study confirms that calculation of program costs of digital health interventions varies widely between studies, unlike the relatively fixed costs for drug interventions. Additionally, there is a notable lack of model-based studies for digital health interventions, and the existing ones often use simplistic models. Our findings highlight the need for more research on cost and effectiveness measurements that reflect the unique characteristics of digital health interventions. Moreover, developing and refining modeling methods to accurately capture these characteristics is essential, necessitating more diverse and sophisticated model-based studies.