Half the mental health apps proven in trials cannot be downloaded

The standard criticism of mental health apps is that the ones people download have no evidence behind them. A systematic review published in npj Digital Medicine examined the mirror image of that problem: what happens to the apps that do have evidence [s1].

The answer is that roughly half of them cannot be obtained.

The audit

The authors set out to identify self-guided mental health apps evaluated in controlled trials, determine whether each was publicly available, and test whether evidence quality or strength predicted availability [s1].

A literature search identified 16,971 records, of which 110 studies investigating 112 apps met inclusion criteria [s1]. Deduplicating to unique products left 81 self-guided apps [s1].

Of those 81, 42 — 52% — were publicly available [s1]. Among the available ones, 79% were listed on both the Apple App Store and Google Play, and 40% were free [s1].

The part that should be uncomfortable

Two tests in this paper matter more than the headline percentage.

The first: study quality, assessed with the Cochrane Risk of Bias 2 tool, had no significant association with whether an app was publicly available (P = 0.851) [s1]. Rigour did not predict release.

The second: a meta-analysis found no significant difference in effect size relative to control between available apps (g = 0.33) and unavailable apps (g = 0.45), with P = 0.228 [s1].

Note the direction. The point estimate for the apps you cannot download is larger than for the ones you can — not significantly so, but certainly not smaller.

The authors' conclusion is that considerations apart from evidence quality or strength determine whether an app reaches the public, and that a gap exists between evidence and availability [s1].

The study did not identify what those considerations are, and the plausible candidates are worth naming as hypotheses rather than findings: research funding that ends when a trial ends, and the continuing cost of app-store compliance, operating-system updates, security patching and support for a product with no revenue model. On that reading, part of this literature documents software that no longer exists.

What is available instead

A second study, published in the same journal days later, examined what fills that space. The authors analysed platform and review data from leading Chinese and US app markets, integrating developer-side and user-side perspectives [s2].

On the developer side, US-based applications demonstrated substantially greater engagement in clinical pipelines and public evidence claims than Chinese applications [s2]. But those indicators were not significantly associated with app store ratings in the US sample, and could not be meaningfully estimated in the Chinese sample because disclosure was very limited [s2].

That is a second disconnection between evidence and the market, at a different point in the chain. In the first study, evidence did not predict whether an app was released [s1]. In the second, among released apps, evidence engagement did not predict how users rated it [s2].

Satisfaction drivers differed by country: Chinese users emphasised core service functions such as online consultation, while US users showed stronger preferences for experiential features such as meditation practice [s2]. Across both contexts, personalised services were consistently rated as insufficient [s2]. The authors argue that development should shift these applications from digital products toward evidence-based, user-centred health tools [s2].

What clinicians say when asked

A systematic review and thematic synthesis in Frontiers in Psychiatry searched five databases for qualitative and mixed-methods studies of healthcare professionals' perspectives on digital mental health tools in clinical practice [s3]. Fifteen studies were reviewed, yielding data from 604 healthcare professionals [s3].

Four themes emerged: that these tools should augment rather than replace face-to-face clinical care; considerations and caveats to their use in practice; using them to enhance clinical care; and perceived barriers and concerns [s3].

Professionals strongly endorsed the view that the tools offer increased access to care, while concerns about therapeutic quality, risk management and workload burden persisted [s3].

Workload burden deserves emphasis, because it is the objection least often addressed by developers. A tool that generates alerts, symptom scores or between-session messages does not reduce clinician time; it redistributes it, usually toward the clinician.

What to watch

The policy discussion around mental health apps is largely about raising the evidence bar. These papers suggest the bar is not the binding constraint. Apps with controlled-trial evidence and acceptable risk-of-bias ratings are being produced at a reasonable rate and then not being released, for reasons that have nothing to do with whether they work.

Sustained funding for maintenance, rather than more trials, is the mechanism that would close that gap — and it is the mechanism no current research funding model provides.

This article is informational and is not medical advice.