Johnson Shoyama Graduate School of Public Policy, University of Regina, Regina, Canada.
Johnson Shoyama Graduate School of Public Policy, University of Saskatchewan, Saskatoon, Canada.
PLoS One. 2021 Nov 1;16(11):e0259486. doi: 10.1371/journal.pone.0259486. eCollection 2021.
This study aims to understand how participants' compliance and response rates to both traditional validated surveys and ecological momentary assessments (EMAs) vary across 4 cohorts who participated in the same mHealth study and received the same surveys and EMAs on their smartphones, however with cohort-specific time-triggers that differed across the 4 cohorts.
As part of the Smart Platform, adult citizen scientists residing in Regina and Saskatoon, Canada, were randomly assigned to 4 cohorts in 2018. Citizen Scientists provided a complex series of subjective and objective data during 8 consecutive days using a custom-built smartphone application. All citizen scientists responded to both validated surveys and EMAs that captured physical activity. However, using Smart Platform, we varied the burden of responding to validated surveys and EMAs across cohorts by using different time-triggered push notifications. Participants in Cohort 1 (n = 10) received the full baseline 209-item validated survey on day 1 of the study; whereas participants in cohorts 2 (n = 26), 3 (n = 10), and 4 (n = 25) received the same survey in varied multiple sections over a period of 4 days. We used weighted One-way Analysis of Variance (ANOVA) tests and weighted, linear regression models to assess for differences in compliance rate across the cohort groups controlling for age, gender, and household income.
Compliance to EMAs that captured prospective physical activity varied across cohorts 1 to 4: 50.0% (95% Confidence Interval [C.I.] = 31.4, 68.6), 63.0% (95% C.I. = 50.7, 75.2), 37.5% (95% C.I. = 18.9, 56.1), and 61.2% (95% C.I. = 47.4, 75.0), respectively. The highest completion rate of physical activity validated surveys was observed in Cohort 4 (mean = 97.9%, 95% C.I. = 95.5, 100.0). This was also true after controlling for age, gender, and household income. The regression analyses showed that citizen scientists in Cohorts 2, 3, and 4 had significantly higher compliance with completing the physical activity validated surveys relative to citizen scientists in cohort group 1 who completed the full survey on the first day.
CONCLUSIONS & SIGNIFICANCES: The findings show that maximizing the compliance rates of research participants for digital epidemiological and mHealth studies requires a balance between rigour of data collection, minimization of survey burden, and adjustment of time- and user-triggered notifications based on citizen or patient input.
本研究旨在了解参与者对传统验证性调查和生态瞬时评估（EMA）的依从率和应答率如何在 4 个队列中有所不同，这 4 个队列均参加了相同的移动健康研究，并在智能手机上接受了相同的调查和 EMA，但每个队列的时间触发因素均不同。
作为智能平台的一部分，居住在加拿大里贾纳和萨斯卡通的成年公民科学家于 2018 年被随机分配到 4 个队列中。公民科学家在连续 8 天内使用定制的智能手机应用程序提供了一系列复杂的主观和客观数据。所有公民科学家均对捕捉身体活动的验证性调查和 EMA 进行了应答。然而，通过使用智能平台，我们通过使用不同的时间触发推送通知，在队列之间改变了对验证性调查和 EMA 的应答负担。队列 1（n = 10）的参与者在研究的第 1 天收到了完整的基线 209 项验证性调查；而队列 2（n = 26）、队列 3（n = 10）和队列 4（n = 25）的参与者则在 4 天内以不同的多部分方式收到了相同的调查。我们使用加权单因素方差分析（ANOVA）检验和加权线性回归模型，在控制年龄、性别和家庭收入的情况下，评估了队列组之间的依从率差异。
对捕捉前瞻性身体活动的 EMA 的依从率在队列 1 至 4 之间有所不同：50.0%（95%置信区间[CI] = 31.4，68.6）、63.0%（95% CI = 50.7，75.2）、37.5%（95% CI = 18.9，56.1）和 61.2%（95% CI = 47.4，75.0）。完成身体活动验证性调查的最高完成率出现在队列 4（均值=97.9%，95% CI = 95.5，100.0）。即使在控制年龄、性别和家庭收入后，这也是如此。回归分析显示，队列 2、3 和 4 的公民科学家完成身体活动验证性调查的依从性明显高于队列 1 的公民科学家，后者在第 1 天完成了完整的调查。
研究结果表明，要使数字流行病学和移动健康研究的参与者的依从率最大化，需要在数据收集的严谨性、调查负担的最小化以及基于公民或患者输入调整时间和用户触发通知之间取得平衡。
