Digital mental health interventions: an opportunity to enhance care while preserving human connection

Self-help digital mental health interventions have been suggested as a scalable and low-cost option to improve access to mental healthcare. Nakamura and colleagues¹ conducted a single-blind, two-arm, individually randomized controlled trial of an unguided self-help intervention in 46 primary care canters in Guarulhos, Brazil. Adults aged 60 and above with subthreshold depressive symptoms (5≤ Patient Health Questionnaire-9²≤ 9) were recruited to the 6-week intervention arm (48 automated WhatsApp messages focused on psychoeducation and behavioural activation) and compared to a control group (a single message about depression). The research team adapted the intervention based on their prior work: a community healthcare worker-delivered psychosocial intervention³ and their unguided self-help digital application for older adults meeting clinical thresholds (PHQ≥10).⁴ At 3-months post-randomization, however, Nakamura¹ found no differences between the two groups. This study has many strengths including a rigorous study design, testing a digital mental health intervention in comparison to an active control condition, and studying an at-risk population in a socioeconomically disadvantaged region. In this Comment, I reflect on the wider literature, highlight research opportunities and discuss clinical and health policy implications.

Self-help digital interventions come in varied formats, such as self-help books, mobile and web-based programs or messages. It is important to differentiate guided vs. unguided self-help digital mental health interventions. Guided self-help digital interventions involve some degree of treatment provider involvement, whereas unguided interventions are typically fully-automated therapeutic approaches without any human support. Unguided self-help digital interventions have been argued to provide flexibility and anonymity at a low-cost, offering predesigned content such as videos, exercises or information, that users access at their own pace. Meta-analyses have generally demonstrated that unguided interventions outperform waitlist control groups,^5,6 including in low- and middle-income country settings.⁷ Previous studies suggested that unguided self-help interventions may be effective for individuals with subthreshold depressive symptoms⁵; however, this was not supported by the findings from Nakamura and colleagues.¹ Relative to guided interventions, unguided interventions have poorer engagement rates, higher dropout rates, and worse clinical outcomes, even when tailored to individual profiles.^5,8 Moreover, among patients with depression, the acceptability rates of guided self-help psychotherapy has been ranked considerably lower (1.6%) than any other human-delivered formats via telephone (67.7%), individual (62.5%), and group (51.8%).⁶

Thus, there remain unanswered questions and important research opportunities to examine the potential role of self-help digital mental health interventions to address the global burden of depression (Fig. 1). First, while unguided digital health interventions may be scalable and accessible, are they effective? And if so, for whom? One may argue that the study sample of interest (in this case, older adults) may not be suitable to receive digital health interventions due to poor digital literacy.⁸ However, the lack of evidence for unguided digital interventions has also been found in large trials among young people.⁸ Thus, the inclusion of relevant socioeconomic and clinical baseline variables can facilitate personalized medicine models to inform for whom these interventions may be effective. Second, health behaviourists from diverse interdisciplinary fields have long established that psychoeducation alone does not result in health behaviour change.⁹ Thus, it is unsurprising that simply receiving content about mental health does not influence clinical outcomes. To address this gap, we must develop, test and establish real-time, objective measures to assess therapeutic engagement and health behaviours. As highlighted by studies in cardiovascular medicine,¹⁰ the use of wearables alongside self-report measures is an important area of research, particularly for digital mental health interventions incorporating behavioural activation and sleep hygiene. Finally, the inclusion of repeated, temporally-distinct outcomes permits opportunities for causal mediation analyses to unpack the ‘black box’ of treatment effectiveness (and ineffectiveness).

Fig. 1.

Research opportunities in digital mental health.

The findings from Nakamura and colleagues² and the literature to date have several implications for clinical practice and health policy. First, clinicians, researchers and policymakers should be cautious about widely promoting such interventions without adequate evidence. To address the growing epidemic of loneliness,¹¹ it may be more effective to consider the integration of digital interventions alongside existing and effective solutions. For example, unguided digital mental health interventions could be used to provide psychoeducation alongside a single session or a task-sharing intervention. Second, for decision and policy makers, investments in digital health infrastructure should be complemented by strategies to improve digital literacy to facilitate equitable access to technology. Without such considerations, the most vulnerable populations may be left behind, perpetuating health disparities. Finally, digital interventions could be considered on a continuum of care within a wider collaborative care model where digital intentions may be offered as a first and possibly preventive step for all individuals, including those at risk.

In conclusion, the findings from Nakamura et al.¹ highlight promising avenues for future research of digital interventions. Establishing an equitable and evidence-informed healthcare system that involves a modest but effective level of human support in digital mental health interventions may be a more productive direction for advancing mental healthcare access and outcomes.

Contributors

DRS conceptualized, wrote, submitted and is fully responsible for the contents of this original manuscript.

Declaration of interests

None.