The escalating global mental health burden, affecting an estimated 165 million individuals annually within the European Union, necessitates a fundamental paradigm shift toward digitally-enhanced care models to alleviate systemic strain and limited provider availability. This narrative review provides a holistic socio-technical synthesis of the Digital Mental Health Interventions (DMHI) ecosystem, integrating heterogeneous evidence ranging from randomized controlled trials and meta-analyses to empirical technical software audits. Findings confirm the general non-inferiority of iCBT compared to traditional modalities, although therapeutic efficacy remains highly contingent upon human integration as defined by the Supportive Accountability model. While specialized tools like virtual reality demonstrate superior outcomes for PTSD, severe psychiatric conditions necessitate continued clinical oversight; crucially, a persistent engagement paradox exists wherein high initial adoption is frequently offset by significant long-term attrition. Structurally, the digital inverse care law and the digital divide reinforce systemic health disparities, excluding populations with infrastructure gaps or physical impairments. These systemic inequities are further compounded by an inverse data law, wherein algorithmic datasets remain biased toward healthy populations, leaving marginalized cohorts underrepresented. Furthermore, while digital platforms afford anonymity for managing stigmatized identities, they simultaneously expose vulnerable users to digital microaggressions and prejudice. Ultimately, rapid technological proliferation has outpaced regulatory governance, resulting in privacy vulnerabilities and profound regulatory fragmentation across jurisdictions. DMHI effectiveness is inextricably linked to socio-economic, cultural, and structural contexts, necessitating a transition toward regulated, clinically-led architectures to safeguard social equity and ensure genuine therapeutic value within the digital mental health transition.
Pilarek et al. (Mon,) studied this question.