RWE to support decision making in HEOR relies on fit-for-purpose data. Much is talked about the quality of research data derived from clinical practice (Real-world Data). To avoid confusion, it helps to keep the inferential use case in mind and work backwards: To support causal conclusions in etiologic studies on drug effects information bias needs to be minimized. What are the measurement error mechanisms that lead to information bias, what are the metrics to quantify them, how do we assess measurement characteristics in validation studies, and how do we assess the impact of imperfect measurement on an effect estimate? Once this is understood the many supporting activities and terms currently used can be better contextualized. These issues will be developed in a logical way to develop a coherent and results-oriented data quality map. Using precise language on data quality and information bias provides clarity in the current mambo-jumbo of buzzwords.

Assessing the measurement characteristics in a given database is time-consuming and expensive. We highlight approaches to expediting data validations through stratified batch validation, through LLM support in establishing electronic health record (EHR)-based reference standards, and through claims diary reviews. To what extent can coding algorithms be shared across data sources and how can their transportability be assessed? We connect our points with recent FDA/EMA guidance documents, discuss implications for federated data analyses, how to prepare decision-makers as reviewers of EHR/claims data analyses, and how to improve your marketing position as a producer of RWD.