A sweeping new analysis of nearly 150 randomized controlled trials has found that digital versions of so-called third-wave psychological therapies—including acceptance and commitment therapy, mindfulness-based programs, and compassion-focused interventions—deliver small-to-moderate but statistically reliable improvements in wellbeing and mental health for adults. The systematic review and meta-analysis, published in PLOS Digital Health by Tom C. Gordon, Andrew H. Kemp, Josh Hope-Bell, and Darren J. Edwards, pooled data from 147 trials involving 26,391 participants, making it one of the most comprehensive assessments to date of whether app-based and online process-based therapies actually work outside the therapist’s office.

Third-wave therapies represent a distinct branch of cognitive-behavioral treatment. Rather than challenging the content of negative thoughts directly, as classical cognitive therapy does, they target underlying processes such as psychological flexibility—the capacity to remain open to difficult experiences while continuing to pursue valued actions. Mindfulness training cultivates present-moment awareness, compassion-focused approaches soften harsh self-criticism, and acceptance and commitment therapy teaches people to detach from unhelpful thoughts instead of fighting them. Delivering these techniques digitally, through smartphone apps, web platforms, and guided online modules, promises scalable mental health care at a fraction of the cost of face-to-face treatment, but the evidence base has been fragmented across hundreds of smaller trials with mixed results.

The new synthesis, prospectively registered on the Open Science Framework and conducted across six databases through January 2026, addressed that fragmentation with rigorous random-effects meta-analysis. The headline numbers are consistent: immediately after treatment, digital third-wave interventions produced standardized effect sizes of g = 0.36 for wellbeing across 46 trials, g = 0.44 for depression across 110 trials, g = 0.40 for anxiety across 93 trials, g = 0.46 for stress across 66 trials, g = 0.30 for quality of life across 32 trials, and g = 0.41 for psychological flexibility across 51 trials. In practical terms, these are small-to-moderate effects—roughly the difference between an average person in the treatment group and someone at the 60th to 65th percentile of the untreated population.

Crucially, the benefits were not confined to any single group. The trials spanned healthy samples (61 trials with 15,770 participants), people with diagnosed mental health conditions (41 trials with 6,130 participants), and medical patients managing physical illness (46 trials with 4,491 participants). When the researchers tested whether population type moderated the immediate effects, it did not reach statistical significance, suggesting that the mechanisms these interventions target—flexibility, acceptance, mindful awareness—operate across the spectrum from prevention to clinical treatment. That universality is precisely what process-based therapeutic models predict, and it strengthens the case for digital delivery as a transdiagnostic tool rather than a condition-specific remedy.

But the analysis also delivered a sobering caveat that the authors emphasize: comparator type significantly moderated several of the pooled results. When digital interventions were tested against active controls—other treatments, attention-matched programs, or credible placebo-like conditions—the effect sizes shrank compared with comparisons against inactive controls such as waitlists. This pattern is familiar from psychotherapy research generally: waitlist comparisons inflate apparent benefits because untreated participants often deteriorate or simply fail to improve, while active comparisons reveal the true incremental value of a specific approach. The authors caution that pooled benefits should therefore be interpreted in relation to comparator intensity, a point that matters enormously for clinicians and policymakers deciding whether to fund digital programs as alternatives to existing care rather than merely as stopgaps for long waiting lists.

Heterogeneity was substantial across most outcomes, with I-squared statistics exceeding 75 percent, meaning that the variability among trial results far exceeded what chance alone would explain. This is not unusual in psychotherapy meta-analyses, but it signals that digital third-wave interventions are not a monolithic entity. Programs differ in duration, intensity, content, delivery format, and the populations they serve, and trial quality varies widely. The substantial heterogeneity underscores why the research team went beyond simple pooling to conduct moderator analyses, probing which design features and contextual factors shaped outcomes rather than treating all interventions as interchangeable.

One of the most practically important findings concerned guidance. Human support—coaching, reminders, or feedback from a trained facilitator—did not significantly moderate immediate post-intervention effects overall, but it did predict larger short-term effects on depression (QM = 4.17, p = .048). This nuanced result suggests that guided and unguided digital programs may perform similarly right away for many outcomes, yet human contact may help consolidate gains for depressive symptoms in the weeks following treatment. For developers of mental health apps, the implication is that fully automated products may be viable for some purposes, but adding even modest human support could meaningfully improve outcomes where depression is the target.

Durability of benefits emerged as another key theme. At short-term follow-up, statistically supported effects remained across all outcomes examined—wellbeing, depression, anxiety, stress, quality of life, and psychological flexibility. At medium-term follow-up, however, only depression, anxiety, and stress retained statistical support, with wellbeing, quality of life, and psychological flexibility no longer distinguishable from controls. This fading pattern raises important questions about whether digital interventions produce lasting changes in the underlying processes they claim to target, or whether participants gradually revert to baseline without ongoing practice or booster sessions. The authors argue that future research should directly test whether improvements in psychological flexibility mediate downstream gains in symptoms and wellbeing—a causal chain that process-based theories assume but few trials adequately verify.

Certainty of the evidence, assessed using the GRADE framework, was moderate for most immediate post-intervention outcomes and low for quality of life, reflecting limitations such as risk of bias, heterogeneity, and imprecision across the included trials. Moderate certainty means the true effect is likely to be close to the estimated effect, giving reasonable confidence in the headline findings, while the low rating for quality of life means those estimates should be treated with more caution. The authors’ recommendations for the field are clear: prioritize rigorous active comparators over waitlist designs, extend follow-up periods well beyond the medium term, and design studies capable of testing the mediating mechanisms that would confirm or refute process-based accounts of how these therapies work.

The broader significance of this analysis lies in what it says about the future of scalable mental health care. With demand for psychological support outstripping the supply of trained clinicians worldwide, digital third-wave interventions offer a genuinely evidence-supported option—not a miracle cure, but a reliable, modest improvement accessible to tens of thousands of people simultaneously. The finding that benefits hold across healthy, clinical, and medical populations, combined with evidence that effects persist at least into the short term for every outcome measured, positions these tools as a meaningful complement to traditional care. At the same time, the shrinkage of effects against active comparators and the fading of some benefits at medium-term follow-up are reminders that the field’s next challenge is not proving that digital mindfulness and compassion programs do something, but establishing exactly how much they add, for whom, and for how long.

Subject of Research: Efficacy of digital third-wave psychological interventions for adult wellbeing and mental health

Article Title: Impacts of digital third-wave interventions for adults on wellbeing and mental health outcomes: A systematic review and meta-analysis of randomized controlled trials

Article References: Gordon, T. C., Kemp, A. H., Hope-Bell, J., & Edwards, D. J. (2026). Impacts of digital third-wave interventions for adults on wellbeing and mental health outcomes: A systematic review and meta-analysis of randomized controlled trials. PLOS Digital Health, 5(10), e0001764. https://doi.org/10.1371/journal.pdig.0001764

Image Credits: AI Generated

DOI: 10.1371/journal.pdig.0001764

Keywords: digital mental health, third-wave therapies, meta-analysis, mindfulness, acceptance and commitment therapy, compassion-focused therapy, psychological flexibility, randomized controlled trials, depression, anxiety, wellbeing, GRADE certainty