Factors Associated With Dropout of Participants in an App-Based Child Injury Prevention Study: Secondary Data Analysis of a Cluster Randomized Controlled Trial

BACKGROUND: Mobile health (mHealth) interventions offer great potential to reach large populations and improve public health. However, high attrition rates threaten evaluation and implementation of mHealth intervention studies. OBJECTIVE: We explored factors associated with attrition of study participants in an mHealth randomized controlled trial (RCT) evaluating an intervention to reduce unintentional child injury risk in China. METHODS: The cluster RCT compared two groups of an app-based intervention for caregivers of 3-6–year-old children (Bao Hu San). The intervention group received unintentional child injury and parenting education, whereas only parenting education was implemented in the control group. The trial included 2920 study participants in Changsha, China, and lasted 6 months. Data on participant engagement (using the app) were collected electronically throughout the 6-month period. Associations between participant attrition and demographic characteristics, and between attrition and intervention engagement were tested and quantified separately for the intervention and control groups using the adjusted odds ratio (aOR) based on generalized linear mixed models. RESULTS: In total, 2920 caregivers from 20 eligible preschools participated, with 1510 in the intervention group and 1410 in the control group. The 6-month attrition rate differed significantly between the two groups (P<.001), at 28.9% (437/1510) in the intervention group and 35.7% (503/1410) in the control group. For the intervention group, the only significant predictor of attrition risk was participants who learned fewer knowledge segments (aOR 2.69, 95% CI 1.19-6.09). For the control group, significant predictors of attrition risk were lower monthly login frequency (aOR 1.48, 95% CI 1.00-2.18), learning fewer knowledge segments (aOR 1.70, 95% CI 1.02-2.81), and shorter learning durations during app engagement (aOR 2.39, 95% CI 1.11-5.15). Demographic characteristics were unrelated to attrition. CONCLUSIONS: Engagement in the app intervention was associated with participant attrition. Researchers and practitioners should consider how to best engage participants in app-based interventions to reduce attrition. TRIAL REGISTRATION: Chinese Clinical Trial Registry ChiCTR-IOR-17010438; http://www.chictr.org.cn/showproj.aspx?proj=17376 INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.1186/s12889-018-5790-1