Estimated Effects of Behavior Change Components in mHealth Interventions for Physical Activity and Related Psychological Outcomes: Bayesian Component Network Meta-Analysis

Background Mobile health (mHealth) interventions for physical activity commonly combine several behavior change and digital design components, making it difficult to determine which components contribute to outcomes. Component estimates also depend on how repeated contrasts within trials and departures from additivity are handled. Objective This study aimed to estimate component-level contributions of 7 operationalized intervention components and examine whether their contributions differed between physical activity and related psychological outcomes. Methods We conducted a systematic review and Bayesian component network meta-analysis of 130 randomized controlled trials involving 25,369 participants after a final study-design eligibility reaudit. Component composition was coded for 310 study arms. The primary target-scope dataset contained 469 component-informative contrast-level observations from 121 trials; a broader 492-observation dataset was analyzed as a sensitivity. A Bayesian additive component metaregression used a multivariate normal likelihood by study, a shared-comparator covariance for multiarm outcome sets, a study-level random effect, and within-study residual heterogeneity. Sensitivity analyses varied working cross-outcome and cross-time correlations, restricted each study to a primary outcome group, excluded cluster-coded trials, compared outcome domains, and explored component interactions. Results In the primary hierarchical model, self-monitoring (Hedges g=0.153, 95% credible interval [CrI] 0.053-0.249) and gamification (g=0.102, 95% CrI 0.020-0.183) had CrIs above 0. Feedback retained a positive point estimate (g=0.127), but its 95% CrI included 0 (−0.004 to 0.259) after the final eligibility reaudit; intervals for the other 4 components also included 0. Across the 7 outcome-domain contrasts, social comparison was lower for psychological than physical activity outcomes (psychological-minus-physical activity difference g=−0.189, 95% CrI −0.325 to −0.052), including in the 66-trial same-study analysis (g=−0.201, 95% CrI −0.386 to −0.021), although social comparison was sparsely represented. For feedback, the corresponding differences were 0.084 (95% CrI −0.088 to 0.254) and 0.046 (95% CrI −0.163 to 0.254), respectively. The low-risk subset was rank deficient and did not support stable estimation of all 7 components. Conclusions Self-monitoring and gamification had the clearest evidence of positive component contributions. Feedback retained a positive point estimate, but its CrI included 0 in the eligibility-locked analysis, and the data did not support a robust divergence in feedback by outcome domain. The isolated social comparison domain contrast arose from sparse evidence and should be considered hypothesis generating. Component rankings should be interpreted alongside small absolute effects, uncertainty, the pragmatic additivity assumption, uncorrected cluster designs in the source summaries, and limited low-risk evidence. Trial Registration PROSPERO CRD420251182885; https://www.crd.york.ac.uk/PROSPERO/view/CRD420251182885

Authors

Publication Details

Journal
Journal of Medical Internet Research
Published
2026-11-19
DOI
https://doi.org/10.2196/88460
Primary Topic
Mobile Health and mHealth Applications
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Estimated Effects of Behavior Change Components in mHealth Interventions for Physical Activity and Related Psychological Outcomes: Bayesian Component Network Meta-Analysis

Long Cheng, Yongheng Zhao, Yimeng Gu, Wang Xi
Journal of Medical Internet Research
Mobile Health and mHealth Applications
article

Estimated Effects of Behavior Change Components in mHealth Interventions for Physical Activity and Related Psychological Outcomes: Bayesian Component Network Meta-Analysis

Long Cheng, Yongheng Zhao, Yimeng Gu, Wang Xi
article en

Abstract

Background Mobile health (mHealth) interventions for physical activity commonly combine several behavior change and digital design components, making it difficult to determine which components contribute to outcomes. Component estimates also depend on how repeated contrasts within trials and departures from additivity are handled. Objective This study aimed to estimate component-level contributions of 7 operationalized intervention components and examine whether their contributions differed between physical activity and related psychological outcomes. Methods We conducted a systematic review and Bayesian component network meta-analysis of 130 randomized controlled trials involving 25,369 participants after a final study-design eligibility reaudit. Component composition was coded for 310 study arms. The primary target-scope dataset contained 469 component-informative contrast-level observations from 121 trials; a broader 492-observation dataset was analyzed as a sensitivity. A Bayesian additive component metaregression used a multivariate normal likelihood by study, a shared-comparator covariance for multiarm outcome sets, a study-level random effect, and within-study residual heterogeneity. Sensitivity analyses varied working cross-outcome and cross-time correlations, restricted each study to a primary outcome group, excluded cluster-coded trials, compared outcome domains, and explored component interactions. Results In the primary hierarchical model, self-monitoring (Hedges g=0.153, 95% credible interval [CrI] 0.053-0.249) and gamification (g=0.102, 95% CrI 0.020-0.183) had CrIs above 0. Feedback retained a positive point estimate (g=0.127), but its 95% CrI included 0 (−0.004 to 0.259) after the final eligibility reaudit; intervals for the other 4 components also included 0. Across the 7 outcome-domain contrasts, social comparison was lower for psychological than physical activity outcomes (psychological-minus-physical activity difference g=−0.189, 95% CrI −0.325 to −0.052), including in the 66-trial same-study analysis (g=−0.201, 95% CrI −0.386 to −0.021), although social comparison was sparsely represented. For feedback, the corresponding differences were 0.084 (95% CrI −0.088 to 0.254) and 0.046 (95% CrI −0.163 to 0.254), respectively. The low-risk subset was rank deficient and did not support stable estimation of all 7 components. Conclusions Self-monitoring and gamification had the clearest evidence of positive component contributions. Feedback retained a positive point estimate, but its CrI included 0 in the eligibility-locked analysis, and the data did not support a robust divergence in feedback by outcome domain. The isolated social comparison domain contrast arose from sparse evidence and should be considered hypothesis generating. Component rankings should be interpreted alongside small absolute effects, uncertainty, the pragmatic additivity assumption, uncorrected cluster designs in the source summaries, and limited low-risk evidence. Trial Registration PROSPERO CRD420251182885; https://www.crd.york.ac.uk/PROSPERO/view/CRD420251182885

Journal of Medical Internet ResearchVol. 28
Mobile Health and mHealth Applications
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon