CoChat

Mobile phone-based interventions for mental health: A systematic meta-review of 14 meta-analyses of randomized controlled trials

Open access

The DOI matches its Crossref record; 70 of the 73 references it cites were found in a registry or an index; 7 of the 81 sources its claims cite were checked for support.

Abstract

Mobile phone-based interventions have been proposed as a means for reducing the burden of disease associated with mental illness. While numerous randomized controlled trials and meta-analyses have investigated this possibility, evidence remains unclear. We conducted a systematic meta-review of meta-analyses examining mobile phone-based interventions tested in randomized controlled trials. We synthesized results from 14 meta-analyses representing 145 randomized controlled trials and 47,940 participants. We identified 34 effect sizes representing unique pairings of participants, intervention, comparisons, and outcome (PICO) and graded the strength of the evidence as using umbrella review methodology. We failed to find convincing evidence of efficacy (i.e., n > 1000, p < 10−6, I2 < 50%, absence of publication bias); publication bias was rarely assessed for the representative effect sizes. Eight effect sizes provided highly suggestive evidence (i.e., n > 1000, p < 10−6), including smartphone interventions outperforming inactive controls on measures of psychological symptoms and quality of life (d s = 0.32 to 0.47) and text message-based interventions outperforming non-specific controls and active controls for smoking cessation (d s = 0.31 and 0.19, respectively). The magnitude of effects and strength of evidence tended to diminish as comparison conditions became more rigorous (i.e., inactive to active, non-specific to specific). Four effect sizes provided suggestive evidence, 14 effect sizes provided weak evidence, and eight effect sizes were non-significant. Despite substantial heterogeneity, no moderators were identified. Adverse effects were not reported. Taken together, results support the potential of mobile phone-based interventions and highlight key directions to guide providers, policy makers, clinical trialists, and meta-analysts working in this area.

Overview

CoChat Summary

TL;DR

This meta-review of 14 meta-analyses representing 145 randomized controlled trials and 47,940 participants found no convincing evidence for mobile phone-based mental health interventions. Eight of 34 representative effect sizes provided highly suggestive evidence, but effects were generally small and weakened as comparison conditions became more rigorous.

Key findings

  • No effect size met criteria for convincing evidence (n > 1000, p < 10−6, I2 < 50%, absence of publication bias); publication bias was tested for only 2 of 34 eligible effect sizes.
  • Eight effect sizes provided highly suggestive evidence: smartphone interventions outperformed inactive controls for anxiety (ds = 0.32 and 0.45), depression (d = 0.32), stress (d = 0.47), and quality of life (d = 0.35); mobile phone and text message-based interventions outperformed non-specific controls for smoking cessation (ds = 0.30 and 0.31); text message-based interventions outperformed active controls for smoking cessation (d = 0.19).
  • Four effect sizes provided suggestive evidence, 14 provided weak evidence, and 8 were non-significant; effects generally diminished from inactive to active and specific active comparisons.
  • No moderators were identified, and adverse events were not reported; only one meta-analysis evaluated adverse events, finding they were not reported in the seven included trials.

Methods

  • Preregistered systematic meta-review following PRISMA; searched six databases through October 31, 2020, for meta-analyses of mobile phone-based interventions on mental health outcomes, requiring at least four RCTs, standardized effect sizes with 95% CIs, and comparison conditions categorized as inactive, active, non-specific, specific, or adjunct.
  • Included 14 meta-analyses representing 145 unique primary RCTs and 47,940 participants; identified 34 representative effect sizes based on the largest sample per PICO pairing and graded certainty using umbrella review criteria.
  • Meta-analysis quality was assessed with the NIH tool (average 7.07/8); risk of bias, publication bias, and moderator tests were extracted when available.

Limitations

  • The paper states that findings are limited by the available meta-analytic and primary study literature, possible underestimation of evidence strength due to lack of publication bias assessment, substantial heterogeneity within and between meta-analyses, rare valid moderator tests, and limited standardized assessment of intervention quality.
  • The paper also notes diminished confidence due to risk of bias from lack of blinding of personnel/participants and incomplete outcome data, and it states that adverse events were almost entirely absent from the meta-analytic literature.
  • Evident from the text: only one meta-analysis tested moderators within an eligible effect size, publication bias was assessed for only 2 of 34 representative effect sizes, and no eligible effect sizes focused exclusively on youth or adolescent samples.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1371/journal.pdig.0000002

Cites

70 of 70 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Figures

Fig 1 Risk of bias summary aggregated across meta-analyses. Random Seq = random sequence generation; Allocat Conceal = allocation concealment; Blind Person/Partic = blinding of personnel and participants; Blind Outcome = blinding of outcome assessor; Incomp Data = incomplete outcome data; Select Report = selective reporting; Other = other bias.
Fig 1 Risk of bias summary aggregated across meta-analyses. Random Seq = random sequence generation; Allocat Conceal = allocation concealment; Blind Person/Partic = blinding of personnel and participants; Blind Outcome = blinding of outcome assessor; Incomp Data = incomplete outcome data; Select Report = selective reporting; Other = other bias.CC BY · PubMed Central
Fig 2 Density plots displaying distribution of representative effect sizes separated by comparison type. Inactive = no active comparison (e.g., assessment only, waitlist control); Non-specific = non-specific controls (i.e., not intended to be therapeutic); Active = active comparison that may or may not have included therapeutic ingredients; Specific = specific active controls (i.e., intended to be therapeutic); Adjunct = mobile phone-based intervention tested as adjunct to another intervention.
Fig 2 Density plots displaying distribution of representative effect sizes separated by comparison type. Inactive = no active comparison (e.g., assessment only, waitlist control); Non-specific = non-specific controls (i.e., not intended to be therapeutic); Active = active comparison that may or may not have included therapeutic ingredients; Specific = specific active controls (i.e., intended to be therapeutic); Adjunct = mobile phone-based intervention tested as adjunct to another intervention.CC BY · PubMed Central
Fig 3 Forest plot of representative effect sizes separated by comparison type. Color based on strength of evidence and size based on inverse variance. Non-specific = non-specific controls (i.e., not intended to be therapeutic); Specific = specific active controls (i.e., intended to be therapeutic); Adjunct = mobile phone-based intervention tested as adjunct to an active treatment.
Fig 3 Forest plot of representative effect sizes separated by comparison type. Color based on strength of evidence and size based on inverse variance. Non-specific = non-specific controls (i.e., not intended to be therapeutic); Specific = specific active controls (i.e., intended to be therapeutic); Adjunct = mobile phone-based intervention tested as adjunct to an active treatment.CC BY · PubMed Central
Fig 4 Forest plot of representative effect sizes separated by outcome domain. Color based on strength of evidence and size based on inverse variance. Non-specific = non-specific controls (i.e., not intended to be therapeutic); Specific = specific active controls (i.e., intended to be therapeutic); Adjunct = mobile phone-based intervention tested as adjunct to another intervention.
Fig 4 Forest plot of representative effect sizes separated by outcome domain. Color based on strength of evidence and size based on inverse variance. Non-specific = non-specific controls (i.e., not intended to be therapeutic); Specific = specific active controls (i.e., intended to be therapeutic); Adjunct = mobile phone-based intervention tested as adjunct to another intervention.CC BY · PubMed Central

Verification

Claims the paper makes

6 Valid · 1 Not valid · 66 Not checked · 64 Own results and interpretations

74 more claims · 1 with their cited sources checked

Create a free account to see all 137 claims and to verify the 74 references still open.

See all claims — free account
  • It may state two separate claims as one. 40
  • Probably not a substantial claim. 28

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

5 of 5

Citations

Cites

Cited by

See all 253

Metadata

Authors
Goldberg, S. B., Lam, S. U., Simonsson, O., Torous, J., & Sun, S.
Affiliations
  • University of Wisconsin–Madison (US)
  • Beth Israel Deaconess Medical Center (US)
  • Brown University (US)
Corresponding author
Goldberg, Simon B.
Venue
PLOS Digital Health
Published
2022-01-18
Volume
1
Issue
1
Pages
e0000002
ISSN
2767-3170 (ISSN-L)
OpenAlex
W4206302417
Cited by
237
References
73
Citations by year
2026: 31 · 2025: 61 · 2024: 62 · 2023: 55 · 2022: 28
FWCI
45.48
Type
journal article
Language
en
Open access
Locations
  • PLOS Digital Health (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • OSF Preprints (OSF Preprints) (submittedVersion, cc-by)
  • OSF Preprints (OSF Preprints) (submittedVersion)
Funders
National Center for Complementary and Alternative Medicine (K23AT010879, K23AT011173)
Topics
  • Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
  • Health Sciences › Health Professions › General Health Professions › Mobile Health and mHealth Applications
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Mental Health Research Topics
Keywords
mobile phone intervention, smartphone-based intervention, text messaging intervention, mental health, psychological symptoms, quality of life, smoking cessation, intervention efficacy, publication bias, umbrella review, meta-review

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-06, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-06 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.