[Advances in the efficacy of digital psychological therapies for depressive disorders]

The overall efficacy of digital psychological therapies for depressive disorders has been supported by multiple systematic reviews and meta-analyses. However, effect sizes vary substantially across studies [Hedges' g (hereinafter referred to as "g")=0.22 to 1.01], and a marked attenuation of treatment effects has been observed when translating findings from efficacy studies (g=0.59) to real-world effectiveness studies (g=0.30). The major sources of this heterogeneity can be considered at 3 levels: intervention characteristics, participant populations, and control conditions. Regarding intervention characteristics, human guidance can improve overall treatment effects; however, it remains unclear whether short-term symptom improvement differs according to the professional qualifications of guides or among different guidance formats, such as on-demand, scheduled, or intensive guidance. Different guidance formats may also influence process-related outcomes, including treatment adherence and therapeutic alliance. Regarding participant populations, existing evidence is derived primarily from young and middle-aged adults and patients with mild-to-moderate depression, whereas evidence remains relatively limited for older adults, patients with severe depression, and those with psychiatric or physical comorbidities. Age, race and cultural background, and baseline symptom severity may influence treatment response. Regarding control conditions, wait-list controls, treatment as usual, and psychological placebo controls may yield different effect-size estimates, with wait-list controls potentially overestimating the treatment effects of digital psychological therapies. Future research should strengthen the assessment of process-related outcomes and long-term treatment effects, optimize the allocation of human guidance resources, expand the evidence base for older adults, patients with severe depression, and individuals with comorbidities, and standardize the selection and reporting of control conditions to improve the interpretability of efficacy evidence and its translational value in real-world settings.