Nguyen Phi-Yen, McKenzie Joanne E, Hamilton Daniel G, Moher David, Tugwell Peter, Fidler Fiona M, Haddaway Neal R, Higgins Julian P T, Kanukula Raju, Karunananthan Sathya, Maxwell Lara J, McDonald Steve, Nakagawa Shinichi, Nunan David, Welch Vivian A, Page Matthew J
Methods in Evidence Synthesis Unit, School of Public Health and Preventive Medicine Monash University Melbourne Victoria Australia.
MetaMelb Research Group, School of BioSciences University of Melbourne Melbourne Victoria Australia.
Cochrane Evid Synth Methods. 2023 Apr 10;1(2):e12008. doi: 10.1002/cesm.12008. eCollection 2023 Apr.
There are many benefits of sharing data, analytic code, and other materials, yet these items are infrequently shared among systematic reviews (SRs). It is unclear which factors influence authors' decisions to share data, code, or materials when publishing their SRs. Therefore, we aimed to explore systematic reviewers' perspectives on the importance of sharing review materials and factors that might influence such practices.
We searched PubMed for SRs published from January to April 2021, from which we randomly allocated 50% to this survey and 50% to another survey on the replication of SRs. We sent an electronic survey to authors of these SRs ( = 4671) using Qualtrics. Quantitative responses were summarized using frequency analysis. Free-text answers were coded using an inductive approach.
The response rate was 9% ( = 417). Most participants supported routine sharing of search strategies (84%) but fewer for analytic code (43%) or files documenting data preparation (38%). Most participants agreed that normative practices within the discipline were an important facilitator (78%). Major perceived barriers were lack of time (62%) and suitable sharing platforms (31%). Few participants were required by funders (19%) or institutions (17%) to share data, and only 12% of participants reported receiving training on data sharing. Commonly perceived consequences of data sharing were lost opportunities for future publications (50%), misuse of data (48%), and issues with intellectual property (40%). In their most recent reviews, participants who did not share data cited the lack of journal requirements (56%) or noted the review did not include any statistical analysis that required sharing (29%).
Certain types of review materials were considered unnecessary for sharing, despite their importance to the review's transparency and reproducibility. Structural barriers and concerns about negative consequences hinder data sharing among systematic reviewers. Normalization and institutional incentives are essential to promote data-sharing practices in evidence-synthesis research.
共享数据、分析代码和其他材料有诸多益处，但在系统评价（SR）中这些内容却很少被共享。目前尚不清楚哪些因素会影响作者在发表其SR时共享数据、代码或材料的决定。因此，我们旨在探讨系统评价者对于共享评价材料的重要性以及可能影响此类做法的因素的看法。
我们在PubMed上搜索了2021年1月至4月发表的SR，从中随机将50%分配到本次调查，50%分配到另一项关于SR复制的调查。我们使用Qualtrics向这些SR的作者（n = 4671）发送了电子调查问卷。定量回答采用频率分析进行总结。自由文本答案采用归纳法进行编码。
回复率为9%（n = 417）。大多数参与者支持常规共享检索策略（84%），但对于分析代码（43%）或记录数据准备的文件（38%）的支持较少。大多数参与者认为该学科内的规范做法是一个重要的促进因素（78%）。主要的感知障碍是缺乏时间（62%）和合适的共享平台（31%）。很少有参与者被资助者（19%）或机构（17%）要求共享数据，只有12%的参与者报告接受过数据共享培训。数据共享常见的感知后果是未来发表机会的丧失（50%）、数据的不当使用（48%）和知识产权问题（40%）。在他们最近的评价中，未共享数据的参与者提到缺乏期刊要求（56%）或指出评价不包括任何需要共享的统计分析（29%）。
尽管某些类型的评价材料对评价的透明度和可重复性很重要，但仍被认为无需共享。结构障碍和对负面后果的担忧阻碍了系统评价者之间的数据共享。规范化和机构激励对于促进循证合成研究中的数据共享实践至关重要。
