Table 1.

Practical Challenges to Data Pooling and How the Collaboration on Attachment Transmission Synthesis (CATS) Dealt With Them

Challenge and recommendation	Tip
Obtaining the data
Invest time and effort to validate and share (archived) data.	• Plan enough time for this stage. For CATS, it took 18 months. • Establish the meta-analysis as a collaborative group effort by researchers who contribute data. Define roles within the collaboration for which members may volunteer (group coauthor, named coauthor, lead author) and that have academic value. Establish a transparent procedure for assigning roles and for involving members in study plans and ideas (e.g., collaborator meetings, circulating paper proposals). • Make contributing as easy as possible: Inform potential collaborators of exactly which variables are needed, provide a template for the data, be flexible in data format, and offer assistance. • Provide a secure shared folder or server to share the data; data sharing through e-mail is not compliant with privacy laws.
Determine whether data sharing is ethically or legally allowed.	• Include institutional privacy officers from the outset in making a data-protection impact assessment and in determining the infrastructure needs and the minimal set of joint agreements for the collaboration members. • Provide information on where data are located, who controls access, and who has access and under which conditions. • Provide clear directions to collaboration members for checking the local or study-specific ethical and legal conditions under which they are allowed to share their data and for anonymizing their data.
Creating the overall data set
Get insight into the quality of the received data.	• Perform checks for inconsistencies with article, anomalies (e.g., out-of-range scores on questionnaires), and missing data. Try to resolve issues that arise with study authors. • Ask for data quality indicators, such as interrater reliability and internal consistency. • Exclude data that are not up to a priori standards.
Securing access and analysis of the data
Set up a secure and accessible storage facility.	• Determine whether the data need to be accessible to researchers (data analysts) outside the organization. If so, consider building a data commons with secure remote access. If not, store the data with the university secure storage facility. • Place only the final files that need to be accessed by researchers outside the organization on the remotely accessible server. Anonymize the origin of the data sets. Store data files received from primary study authors or files used for data cleaning on a separate secure server. • Partition the remotely accessible server so that researchers have access only to the parts they need to access. • Add the necessary analysis software and other processing software to the remotely accessible server so that the researchers do not have to copy data to their own computers. • Have researchers with access to the data sign a data-sharing agreement.
Defining authorship	• Clearly define contributor roles for the project and provide transparent information about who fulfills these roles and how (e.g., using the Contributor Roles Taxonomy; www.casrai.org/credit.html). • For manuscript preparation, consider the use of tools that allow for simultaneous working on a manuscript so that authors can see and respond to each other’s feedback, comments are given in the same document, and a record of writing contributions is maintained. • Give reasonable deadlines for responding and keep them. • Map contributor roles on authorship roles. In CATS, we have three layers of involvement with articles: Participation in drafting the manuscript leads to named authorship, sharing data and reading and approving the draft before submission leads to group authorship, and sharing data without manuscript involvement leads to a mention as a nonauthor collaborator. These types of involvement were based on the guidelines by the International Committee of Medical Journal Editors (2020).