Characteristics of Real-World Databases in the United States: A Systematic Review

OBJECTIVES: Regulatory bodies have recently laid guidance on using real-world evidence (RWE) generated through real-world data (RWD) for regulatory decision-making. With several new RWD sources available lately, we aim to summarize the characteristics of RWD within the United States (US) using a systematic literature review (SLR).

METHODS: An electronic search was performed in PubMed in May 2023 to identify studies utilizing ≥ 1 US RWD published between 2019 and 2023.

RESULTS: Of 7,832 retrieved citations, 5,612 studies met inclusion. A total of 107 distinct administrative claims databases were utilized in 4,579 studies. Medicare, Marketscan, Medicaid, Optum, and IQVIA were the most commonly utilized claims. A total of 188 distinct electronic medical records (EMRs) RWDs were utilized in 623 studies. Veteran Affairs, Flatiron, Optum, Cerner, and Epic were the most frequently utilized EMRs. A total of 142 unique registries were used in 643 studies, with SEER, National Trauma Databank, USRDS, NCDR-PINNACLE, and VQI being the most common. Subsequently, 142 unique registries (n=643 studies), 36 unique surveys (n=142 studies), 34 common data model-based consortium data (n=64 studies), and app-based or wearable databases in a few studies were identified. While most studies utilized a single source of RWDs, 529 studies utilized a linkage of 2 RWDs. Claims and registry linkage were the most common. Most RWDs had the providence of the single-payer, single healthcare system and/or fragmented inpatient, emergency, or outpatient data only. Most RWD sources could identify medication use, clinical conditions, healthcare resource use, and costs with quality; however, had limited information on laboratory parameters, genomics/ biomarkers, patient-reported outcomes, imaging data, and social determinants of health.

CONCLUSIONS: This SLR is the first, most up-to-date, and comprehensive review of the RWD landscapes within the US and highlights the relevancy of existing data sources with the regulatory grade RWD.