Belfiglio Andrew, Page Shayleigh Dickson, Pettersson Sara, van Rijn Michelle, Vellone Ercole, Westland Heleen, Freedland Kenneth E, Lee Christopher, Strömberg Anna, Wiebe Douglas, Aryal Subhash, Riegel Barbara, Jaarsma Tiny
University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, United States of America.
University of Pennsylvania School of Nursing, Philadelphia, PA, United States of America.
PLoS One. 2024 Sep 16;19(9):e0307440. doi: 10.1371/journal.pone.0307440. eCollection 2024.
Increasingly, studies use social media to recruit, enroll, and collect data from participants. This introduces a threat to data integrity: efforts to produce fraudulent data to receive participant compensation, e.g., gift cards. MOMENT is an online symptom-monitoring and self-care study that implemented safeguards to protect data integrity. Facebook, Twitter, and patient organizations were used to recruit participants with chronic health conditions in four countries (USA, Italy, The Netherlands, Sweden). Links to the REDCap baseline survey were posted to social media accounts. The initial study launch, where participants completed the baseline survey and were automatically re-directed to the LifeData ecological momentary assessment app, was overwhelmed with fraudulent responses. In response, safeguards (e.g., reCAPTCHA, attention checks) were implemented and baseline data was manually inspected prior to LifeData enrollment. The initial launch resulted in 411 responses in 48 hours, 265 of which (64.5%) successfully registered for the LifeData app and were considered enrolled. Ninety-nine percent of these were determined to be fraudulent. Following implementation of safeguards, the re-launch yielded 147 completed baselines in 3.5 months. Eighteen cases (12.2%) were found fraudulent and not invited to enroll. Most fraudulent cases in the re-launch (15 of 18) were identified by a single attention check question. In total, 96.1% of fraudulent responses were to the USA-based survey. Data integrity safeguards are necessary for research studies that recruit online and should be reported in manuscripts. Three safeguard strategies were effective in preventing and removing most of the fraudulent data in the MOMENT study. Additional strategies were also used and may be necessary in other contexts.
越来越多的研究利用社交媒体招募参与者、进行登记并收集数据。这对数据完整性构成了威胁：有人会试图编造虚假数据以获取参与者补偿，例如礼品卡。“时刻”（MOMENT）是一项在线症状监测和自我护理研究，该研究实施了保障措施来保护数据完整性。研究通过脸书、推特以及患者组织，在四个国家（美国、意大利、荷兰、瑞典）招募患有慢性健康问题的参与者。REDCap基线调查问卷的链接被发布到社交媒体账户上。在研究最初启动时，参与者完成基线调查并被自动重定向到LifeData生态瞬时评估应用程序，结果收到了大量欺诈性回复。作为回应，研究实施了保障措施（例如，验证码、注意力检查），并在参与者注册LifeData之前对基线数据进行人工检查。首次启动在48小时内收到了411份回复，其中265份（64.5%）成功注册了LifeData应用程序并被视为已登记。其中99%被判定为欺诈性回复。在实施保障措施后，重新启动在3.5个月内产生了147份完整的基线数据。发现18例（12.2%）为欺诈性回复，这些参与者未被邀请注册。重新启动时的大多数欺诈性案例（18例中的15例）是通过一个注意力检查问题识别出来的。总体而言，96.1%的欺诈性回复来自美国的调查。对于在线招募参与者的研究来说，数据完整性保障措施是必要的，并且应该在论文中报告。三种保障策略有效地防止并清除了“时刻”研究中的大部分欺诈性数据。研究还使用了其他策略，在其他情况下可能也有必要采用。
