OBJECTIVES : Real-world data (RWD) and real-world evidence (RWE) play an increasing role in healthcare decision-making, including regulatory decisions, clinical care, guideline development, and reimbursement/payer decisions. As such, the capture, completeness, and quality of these data underlying these analyses are imperative. Demographic data are critical to the identification of disparities in disease risk and care quality. However, the extent to which these characteristics are collected and recorded in RWD sources is not well known. The objective of our research was to assess and compare the degree of missingness of race, ethnicity, language, and sex data through a review of commercially available datasets across multiple countries.

METHODS : Commercial EHR datasets from the United States, France, the United Kingdom, Spain, Belgium, and Romania were used for this analysis. The data dictionaries for each were reviewed to determine whether they included categories for race, ethnicity, language, and sex. The Aetion Evidence Platform (AEP) was used to assess the percentage of patients for whom demographic data was missing.

RESULTS : Of the seven EHR datasets, only the US included information about patient race (29.4% missing) and none included information on language. Datasets from Spain, Belgium, Romania, and the UK listed “Ethnicity” as a variable in the data dictionary; however, this variable was only populated in the US and UK datasets, with 32.01% and 69.43% of respective patients missing ethnicity data. Information on sex was captured in all datasets, with varying degrees of missingness ranging from 0.67% (US) to 41.34% (Spain).

CONCLUSIONS : Major gaps in capture and completeness of race, ethnicity, language, and sex data in RWD make it difficult for researchers to study variations in care delivery, quality, and outcomes across countries and populations. More research is needed to examine the root causes of these gaps and their impact on decisions guided by RWE.