Real-World Evidence (RWE), Explained: How Pharma Turns Data Into Decisions

For most of the modern drug era, the randomized controlled trial was the only evidence that counted. That is changing. Regulators, payers, and drug developers now routinely use real-world evidence—clinical insight drawn from data generated during ordinary care—to answer questions trials cannot, faster and at lower cost. This guide explains what RWE actually is, why it has earned a seat at the regulatory table, and where it does and does not work.

What Is Real-World Evidence? (And How It Differs From Real-World Data)

The single most common point of confusion is the relationship between two terms that sound interchangeable but are not.

Real-world data (RWD) is the raw material: information about patient health status and the delivery of care, collected routinely outside the controlled setting of a clinical trial. The FDA's working definition points to sources like electronic health records (EHRs), medical and pharmacy claims, product and disease registries, and data from digital health technologies.

Real-world evidence (RWE) is the output: the clinical evidence about a medical product's use, benefits, or risks that you get by analyzing RWD with rigorous methods. RWD is the ore; RWE is the refined metal.

The distinction matters because the value—and the regulatory weight—lives almost entirely in the refinement step. A spreadsheet of claims records is not evidence of anything. A well-specified study design, applied to fit-for-purpose, well-curated data, with appropriate controls for bias and confounding, is what turns that raw material into something a regulator or a payer will act on.

Why Regulators Made Room for RWE

The turning point in the United States was the 21st Century Cures Act, signed into law in December 2016. Section 505F directed the FDA to build a framework for evaluating how RWE might support two specific things: approval of a new indication for an already-approved drug, and the fulfillment of post-approval study requirements.

The agency published its Framework for FDA's Real-World Evidence Program in December 2018, followed by a series of more granular guidances on topics like assessing the reliability of EHR and claims data, and using registries to support regulatory decisions.

The logic behind the shift is practical. Traditional trials are expensive, slow, and—by design—enroll narrow, idealized populations that often look little like the patients who will actually take the drug. RWE can fill gaps: rare subgroups never enrolled, long-term safety signals that emerge only at scale, and the messy question of how a therapy performs in real clinical practice rather than under protocol.

A frequently cited example: in 2019 the FDA approved an expanded indication for the breast-cancer drug Ibrance to cover men, a population the original trials had not enrolled, supported substantially by real-world data drawn from EHR and post-marketing sources. Around the same period, an immunosuppressant used in lung transplant was supported by a non-interventional study built on a national transplant registry. These are not curiosities—they are templates.

This article is educational and is not regulatory advice. RWE acceptability depends on the specific question, jurisdiction, and current agency guidance; consult qualified regulatory professionals before designing a submission.

The Core Use Cases: Where RWE Earns Its Keep

RWE is not one thing. It powers several distinct workflows across the drug lifecycle.

Epidemiology and disease natural history. Before you can design a trial or build a value story, you need to understand a condition's prevalence, progression, and untreated outcomes. Longitudinal RWD lets teams characterize who gets a disease, how it moves over time, and what "standard of care" actually looks like in practice—sometimes serving as an external control arm when a randomized comparator is impractical or unethical.

Safety and pharmacovigilance. Trials rarely have the size or duration to catch rare adverse events. Post-market surveillance over large real-world populations does, which is why safety monitoring was one of the earliest and least controversial uses of RWD.

Health economics and outcomes research (HEOR). Payers want to know whether a therapy delivers value in routine use: hospitalizations avoided, adherence patterns, total cost of care, comparative effectiveness versus alternatives. HEOR teams lean heavily on claims and EHR data to build these arguments.

Label expansion and regulatory submissions. As the Ibrance example shows, RWE can support extending an approved product to new populations or indications—often the highest-stakes, highest-scrutiny use, demanding the strongest data and methods.

Data Sources and Their Trade-Offs

There is no single "best" source. Each captures a different slice of reality, and the art of RWE is matching the source to the question—what regulators call being "fit for purpose."

Electronic health records are rich in clinical detail: diagnoses, lab values, vital signs, clinician notes. Their weakness is fragmentation. A patient who sees providers across unconnected systems leaves a broken trail, and much of the richest signal is locked in unstructured text that requires careful extraction.

Administrative claims are the mirror image. Because they exist to process payment, they capture nearly every billable encounter across a covered population—broad and longitudinally complete. But they record what was billed, not what was clinically true: no lab results, limited outcomes, and coding shaped by reimbursement rather than research.

Registries (disease- or product-specific) are purpose-built and therefore clean and outcome-focused, but narrow in scope and often costly to maintain.

Digital health and patient-generated data (wearables, apps, remote monitoring) add a continuous, patient-side view that no clinical system captures—at the price of noise, consent complexity, and uneven reliability.

The recurring tension is breadth versus depth versus longitudinality. Claims give you breadth; EHRs give you depth; neither alone gives you a clean, coded, multi-year patient journey. Much of the hard work in this field—and much of where genuine differentiation lives—is in linking sources, de-identifying responsibly, normalizing to common data models, and validating that the result actually answers the intended question.

De-identification, linkage, and secondary use of health data are governed by HIPAA and a patchwork of other regulations. None of the above is legal advice; engage privacy and compliance counsel before working with health data.

What "Good" RWE Actually Requires

A useful mental model: RWE quality is a product of three factors, and a weakness in any one undermines the whole.

1. 01Data relevance — does the dataset contain the right patients, exposures, and outcomes to answer the question?
2. 02Data reliability — was it accurately captured, completely recorded, and consistently coded over time?
3. 03Methodological rigor — is the study design pre-specified, with credible handling of confounding, missingness, and bias?

The trend across the industry is toward "regulatory-grade" RWD: data and processes documented well enough to withstand the scrutiny of a submission. That bar is high, and it is why curation and provenance—not raw volume—increasingly define which datasets are actually usable for the highest-value decisions.

Key Takeaways

• RWD is the raw input; RWE is the analyzed output. The value lives in disciplined refinement, not in the data dump.
• The 21st Century Cures Act (2016) and the FDA's 2018 framework legitimized RWE for new indications and post-approval studies—real approvals now exist.
• The major use cases are epidemiology/natural history, safety surveillance, HEOR, and regulatory submissions including label expansion.
• Every data source trades off breadth, depth, and longitudinality. Claims are broad, EHRs are deep, registries are clean-but-narrow, and linkage is where the hard value is created.
• "Fit for purpose" is the standard: relevance, reliability, and rigor together determine whether RWE holds up.

The Bottom Line

Real-world evidence has moved from a supplement to a strategic asset—but only when the underlying data is coded, longitudinal, and trustworthy enough to carry a decision. The bottleneck is rarely the volume of data; it is the refinement. At Prometheus Bio, that refinement step—turning routinely generated health data into rigorously curated, de-identified ground truth—is the entire point. When the question matters, the quality of the input is what decides the answer.