Information Bias in Electronic Health Records and Administrative Claims Data: A Targeted Review of the Recent Literature.

Abstract

Randomized clinical trials represent the primary source of evidence for regulatory and health technology assessment (HTA) decision making; however, the integration of real-world evidence (RWE) has increased in recent years. Despite its utility, RWE is often threatened by information bias, and the literature addressing measurement error in RWE remains underdeveloped. To address this gap, we conducted a targeted literature review to identify and synthesize mitigation measures for information bias in RWE generation among studies using electronic health records (EHRs) and administrative claims data. Our review covered articles published between January 2019 and May 2024 and included real-world data (RWD) investigations with a focus on information bias case studies or review articles; to increase the utility of our results, the Food and Drug Administration's guidance on assessing EHRs and medical claims data was also incorporated. Data elements were extracted and categorized to produce a comprehensive information bias mitigation framework. In total, 38 articles and guidance documents were included, primarily focusing on studies conducted in the United States (n = 25) as well as studies using EHR data (n = 31). Findings were synthesized into 15 general recommendations: six targeting study design, four addressing study variables, and five focused on statistical analyses. Prominent themes included validation, data linkage, and quantitative bias analysis. Overall, our findings underscore the diversity and complexity of the information bias in RWD literature. Our resulting framework offers practical recommendations and complements prior work, providing a foundation for future efforts to enhance the validity of RWE in regulatory/HTA decision making.