Mental Health Apps and Digital Programs: Evidence

What the combined trial results show

A 2024 meta-analysis pooled 176 randomized trials of smartphone apps aimed at depressive or anxiety symptoms, compared with control groups. Apps had small but significant effects on depression (g = 0.28, about 33,500 participants) and generalized anxiety (g = 0.26, about 22,400 participants). [1] The authors translated these into a number needed to treat of about 11 to 12, meaning roughly 11 to 12 people would need to use an app for one extra person to see a benefit over the control. [1]

Results were larger for apps built on cognitive behavioral therapy (CBT) or with chatbot features for depression, and for CBT or mood-monitoring apps for anxiety. [1] Effects on panic symptoms were not significant, and for post-traumatic stress symptoms the effect was small (g = 0.12). The authors caution that results for several conditions came mostly from small trials with high risk of bias. [1]

An earlier review of 66 trials reached a similar picture and also found that apps did not clearly differ from face-to-face or computerized treatment, although few trials made that comparison. [2] It concluded apps are not meant to replace professional services. [2]

Evidence level: promising but limited for depression and generalized anxiety symptoms in studied apps; insufficient evidence for most other conditions.

Why one study of one app does not cover the category

The trials above tested particular programs, often with reminders, guidance or study staff. A product on your phone may share a name or a technique, but it has not been tested unless it has its own trial. The question to ask is whether this exact product has a published trial. See product-specific trials.

A review of app store descriptions looked at 73 top-ranking apps for depression, self-harm, substance use, anxiety and schizophrenia. 64% claimed to be effective at diagnosing or improving symptoms, mood or self-management, yet only two described any primary evidence and one cited published literature. [3]

One example: a cleared prescription app and its trial

In the United States, the FDA cleared Rejoyn through its 510(k) pathway, with an indication of treating major depressive disorder (MDD) symptoms as an add-on to clinician-managed outpatient care in adults aged 22 and older who are on antidepressant medication. [4] The FDA document describes the Mirai trial: a remote, double-blind, randomized trial of 386 people (194 Rejoyn, 192 sham app). The sham was a matched-attention memory exercise without the active tasks or CBT content. [4]

On the primary measure (change in a clinician-rated depression score at week 6), the planned main analysis showed a 2.12 point larger improvement with Rejoyn (95% CI -3.93 to -0.32). A second analysis of those who had at least one treatment session and a follow-up score gave a difference of 1.78 points that did not meet the pre-specified significance threshold (95% CI -3.60 to 0.05). [4] No adverse event was judged related to the app. [4]

This is a helpful example of how modest a well-run app result can be, and it applies to that product, that indication and that population only. It does not show that other depression apps, or this one used outside its indication, work. The FDA document is a summary of the manufacturer's submission, so independent replication, when it exists, is worth reading. For the wider regulatory picture, see regulatory status of digital health products and wellness apps vs digital therapeutics.

How to read an app's claims

• Is there a published randomized trial of this exact product, with a control group that is not just a waiting list?
• Does the outcome measure symptoms or function, or only app use and satisfaction? See meaningful outcomes vs app engagement.
• Does the claim match what the regulator actually allowed? See claim check: "FDA registered" vs approval.