- Design
- Systematic review and meta-analysis of blinded randomised trials
- Population
- 30 trials, 5,680 participants with depressive symptoms
- Primary outcome
- PHQ-9 change within digital placebo arms
- Effect
- Hedges g 0.44 (95% CI 0.29 to 0.59), I² 93.2%
A meta-analysis pooled 30 blinded randomised trials, with 5,680 participants, that used a digital placebo as the comparator in depression trials measured on the PHQ-9.
The pooled improvement within the control arms was Hedges g 0.44 (95% CI 0.29 to 0.59), with very high heterogeneity (I² 93.2%). Two factors explained a good part of the spread: the primary psychiatric disorder and baseline PHQ-9 score. In the multivariable model, heterogeneity fell to 82.1%.
The point is not that digital tools work, nor that they do not. It is that people given an inactive app still improve, so a trial without a well-designed control can overstate benefit. When a patient asks about a depression app, the claimed gain should be read against what a control arm achieves.
- Ask for the control condition when an app cites a trial result.
- Expect some improvement even with an inactive digital comparator, especially at higher baseline PHQ-9.
- Do not read a before-and-after PHQ-9 fall on an app as proof the app caused it.
- Keep monitoring symptoms and risk during digital self-help; it is not a substitute for review.
Why it matters
It lowers the weight a before-and-after score on a digital tool should carry in a consultation.
Don't overread it
The pooled estimate describes change within placebo arms and says nothing on whether any active app works.
The statistics, in plain English
An I² of 93% means the trials differed a lot from each other, so the pooled g of 0.44 is an average over very different settings, not a figure to expect in any one patient. Moderator analyses suggest where the differences come from but do not prove cause.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for psychiatry, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free