Is dosage of a meditation app associated with changes in psychological distress? It depends on how you ask.
Simon B Goldberg, Ashley D Kendall, Matthew J Hirshberg, Cortland J Dahl, Inbal Nahum-Shani, Richard J Davidson, Bethany C Bray
Clinical psychological science : a journal of the Association for Psychological Science March 1, 2025 DOI: 10.1177/21677026241266567 via PubMed
Summary
AI-generated from the abstractIn a randomized controlled trial of a meditation app with 662 participants (80.4% had elevated depression or anxiety), the relationship between how much people used the app (dosage) and changes in psychological distress was inconsistent. Across 41 different statistical models, some showed that more use—measured in minutes, days, or activities completed—was linked to greater reductions in distress, but many models found no such link, and a few even suggested the opposite pattern. This variability highlights the challenge of defining and studying dosage in meditation app interventions and points to the need for careful, transparent methods in this area.
Study at a glance
| Characteristics | Randomized controlled trial Preregistered Peer reviewed |
|---|---|
| Sample size | 662 |
| Population | Adults with elevated depression or anxiety (80.4% of sample) |
| Intervention | meditation app |
| Topics | Meditation |
| Keywords | Dosage Mobile health Optimization |
| Citations | 15 |
| Key finding | Across 41 models, the association between meditation app dosage and decreases in psychological distress varied: some models showed higher dosage linked to larger decreases, but many showed no association or even the opposite pattern. |
Abstract
Despite growing popularity, associations between dosage and outcomes in meditation app interventions have not been established. We examined this relationship using a range of operationalizations of dosage (e.g., minutes of use, days of use, number and type of activities completed) and strategies for modeling outcomes (e.g., ordinary least squares regression, multilevel modeling, latent class analysis). We used data from a recently completed randomized controlled trial testing a meditation app (n=662; 80.4% with elevated depression/anxiety) which included psychological distress as its preregistered primary outcome. Across 41 models, whether or not an association was detected as well as the shape and direction of this association varied. Although several models indicated that higher dosage was associated with larger decreases in psychological distress, many models failed to show this relationship and some even showed the opposite. These results may have implications for optimizing and studying dosage in meditation apps and for open science practices.