Hurley Lex, O'Shea Nisha G, Power Julianne, Sciamanna Christopher, Tate Deborah F
The University of North Carolina at Chapel Hill, Department of Health Behavior, Chapel Hill, North Carolina, United States of America.
Research Triangle Institute (RTI) International, Research Triangle Park, North Carolina, United States of America.
PLOS Digit Health. 2025 Mar 25;4(3):e0000766. doi: 10.1371/journal.pdig.0000766. eCollection 2025 Mar.
Digital behavior change interventions (eHealth, mHealth) are known to be capable of promoting clinically significant weight loss among some participants. However, these programs can struggle with declining engagement and adherence over time, which can hamper their effectiveness. This analysis examines the extent that depression symptoms may negatively influence engagement, adherence, and 6 month weight change in an eHealth intervention.
Structural equation modeling is applied to test the effects of baseline depression symptoms on weight change outcomes, mediated through latent constructs of engagement and adherence, respectively. These constructs were highly correlated within this dataset and necessitated two separate models to be tested. Engagement was indicated by 6 month sums of website logins, user-created goals, visiting various webpages, and posts on the online discussion boards. Adherence was indicated by 6 month sums of weeks exercise goals met, days weight logged, and days of complete dietary tracking.
Depression symptoms showed no direct association with weight change (p's ≥ 0.6), but were negatively associated with both constructs of engagement and adherence (p's < 0.001), which in turn were negatively associated with weight change in both models (p's < 0.001). It was determined depression symptoms had a positive indirect association with weight change fully mediated through these variables, meaning less weight loss or possible weight gain (p < 0.001).
This analysis shows that depression symptoms had a significant, undesirable effect on weight loss outcomes within this eHealth intervention, fully mediated through measured participant engagement and adherence. Further research is needed to test these constructs within a longitudinal model to better understand their dynamic interrelationships, and consider means to address depression in future digital interventions.
数字行为改变干预措施（电子健康、移动健康）已知能够促使一些参与者实现具有临床意义的体重减轻。然而，随着时间推移，这些项目可能会面临参与度和依从性下降的问题，这可能会妨碍其有效性。本分析考察了抑郁症状在电子健康干预中对参与度、依从性和6个月体重变化可能产生的负面影响程度。
应用结构方程模型来测试基线抑郁症状对体重变化结果的影响，分别通过参与度和依从性的潜在结构进行中介作用分析。在该数据集中，这些结构高度相关，因此需要测试两个单独的模型。参与度通过6个月内网站登录次数、用户设定的目标、访问的不同网页数量以及在线讨论板上的帖子数量总和来表示。依从性通过6个月内达到的每周锻炼目标次数、记录体重的天数以及完全进行饮食跟踪的天数总和来表示。
抑郁症状与体重变化无直接关联（p值≥0.6），但与参与度和依从性这两个结构均呈负相关（p值<0.001），而在两个模型中，这两个结构又均与体重变化呈负相关（p值<0.001）。经确定，抑郁症状通过这些变量对体重变化具有正向间接关联，意味着体重减轻较少或可能体重增加（p<0.001）。
本分析表明，在该电子健康干预中，抑郁症状对体重减轻结果具有显著的不良影响，通过测量的参与者参与度和依从性完全中介。需要进一步研究在纵向模型中测试这些结构，以更好地理解它们的动态相互关系，并考虑在未来数字干预中解决抑郁问题的方法。
