Back to the source: how failure modes in clinical information recording affect research on real-world data

Introduction Electronic Health Records and claims data are increasingly used for research, though not designed for this purpose. These data result from complex interactions between healthcare professionals and Information Technology (IT), but the impact of these interactions on data quality remains insufficiently conceptualized.Objectives We address this gap by analysing how data are recorded and integrated within Clinical Information Systems. (CISs).Methodology Informed by Failure Modes and Effects Analysis, we decomposed the data recording process, identified components involved at each step, and analysed potential failures through focused literature searches. Failures were classified using an established framework to assess their effect on data quality.Results We identified four steps: administrative enrolment, data acquisition during care, data assembly into stay records, and aggregation into a patient record. Risks included IT failures, confusing software settings (e.g. values carried forward by default), human-computer interaction issues and system-level factors such as database management or IT system updates. While the literature provided practical examples of these failures, few studies quantified their impact.Practical Implications Through this study, we elucidate the mechanisms by which CISs influence healthcare data quality and highlight their broader sociotechnical implications of the reuse of real-world data in research.