The FDA distinguishes between real-world data, which is data about patient health and care collected routinely from sources such as health records, insurance claims, product and disease registries and devices, and real-world evidence, which is the clinical evidence derived from analyzing that data.
Real-world evidence answers questions that randomized trials often cannot: how a treatment performs in the broader population that actually receives it, how it is used in practice, how patients fare over many years, and how rare safety events emerge. It is used in safety monitoring, in health economics and outcomes research, in payer and health technology assessment submissions, and, in some cases, in regulatory decisions.
In the United States, the 21st Century Cures Act of 2016 directed the FDA to develop a program for evaluating real-world evidence in regulatory decisions, and the agency has since published a framework and a series of guidances.
Real-world studies have real limitations. Patients are not randomized, data may be incomplete or recorded for other purposes, and confounding is a constant risk. Good studies are designed with these problems in mind, and their results are read with them in mind too.