Abstract
Background Creators and reviewers of the Clinical Data Interchange Standards Consortium (CDISC®) / Study Data Tabulation Model (SDTM)-compliant datasets often encounter errors due to knowledge gaps or misunderstandings. This study aims to compile real-world cases of unintentional mistakes in SDTM dataset creation and to analyze their causes, solutions, and lessons learned. The objective is to help beginners avoid making similar errors and to propose effective preventive measures. This initiative was undertaken by the “Sub-team Learning from Past Blunders” within the CDISC Japan User Group (CJUG) SDTM team.
Methods Failure cases were collected from the CJUG/SDTM team members through a structured survey. Each case was reviewed to identify its underlying cause, resolution, and key takeaway. Keywords and hashtags were assigned to categorize the cases, and thematic trends were analyzed based on the frequency of specific terms.
Results A total of 42 responses were obtained. The most common issues involved deviations from the SDTM and SDTM implementation guide (SDTMIG) specifications, as well as mapping errors. Frequent keywords included “inadequate” and “confirmation,” highlighting recurring themes across the reported cases.
Conclusions The term “inadequate” encompassed oversights and failures in communication and consideration. Many of these issues stemmed from insufficient confirmation processes, limited expertise, or negligence. Communication gaps between external stakeholders also contributed to these problems. This study underscores the importance of using validation tools such as Pinnacle 21® and enhancing communication to prevent errors. Furthermore, the information required for SDTM creation often extends beyond what is outlined in the official guidelines. Encouraging individual problem-solving skills and learning from past mistakes can significantly improve data quality.
