Real-world evidence (RWE) is clinical evidence about how a medical product works, its benefits, and its risks, drawn from data collected outside of traditional clinical trials. Instead of studying a drug or device in a tightly controlled lab setting, RWE comes from analyzing health information gathered during routine medical care. The FDA formally defines it as “the clinical evidence about the usage and potential benefits or risks of a medical product derived from analysis of real-world data.”
That definition hinges on an important distinction. Real-world data (RWD) is the raw material: the health records, insurance claims, and patient registries generated every day across the healthcare system. Real-world evidence is what you get when researchers analyze that data to answer a specific clinical question. Think of RWD as the ingredients and RWE as the finished meal.
Why RWE Exists Alongside Clinical Trials
Randomized controlled trials (RCTs) remain the gold standard for proving a treatment works. Researchers randomly assign patients to receive either the treatment or a placebo, then compare outcomes. This design is powerful because randomization eliminates most sources of bias. But RCTs have real limitations. They typically enroll narrow patient populations, sometimes excluding older adults, people with multiple health conditions, or pregnant women. They run in specialized research centers with strict protocols that don’t reflect how medicine is practiced in a typical clinic. And they’re expensive, often costing hundreds of millions of dollars.
RWE fills gaps that trials can’t. It captures how treatments perform in broader, more diverse populations and in the messy reality of everyday healthcare, where patients skip doses, switch medications, or have conditions that would have disqualified them from a trial. Rather than competing with RCTs, RWE provides complementary information. A trial might prove a cancer drug shrinks tumors under ideal conditions. RWE can reveal whether that drug works as well in older patients, in community hospitals, or when combined with medications those trial patients never took.
Where Real-World Data Comes From
The raw data behind RWE comes from sources that already exist in the healthcare system:
- Electronic health records (EHRs): the digital charts doctors and hospitals maintain on every patient, containing diagnoses, lab results, prescriptions, and clinical notes.
- Insurance claims data: billing records submitted to health insurers that capture diagnoses, procedures, and prescriptions across large populations.
- Patient registries: organized databases that track patients with a specific disease or who received a particular treatment over time.
- Patient-generated data: information from wearable devices, mobile health apps, and patient surveys collected outside of clinical settings.
Each source has strengths and trade-offs. EHRs contain rich clinical detail but vary in how different hospitals structure and code their data. Claims data covers millions of patients but only captures what gets billed, missing important clinical nuances. Registries offer focused, high-quality data on specific conditions but cover smaller populations. Researchers often combine multiple sources to build a more complete picture.
How Researchers Generate RWE
Collecting the data is only the first step. Turning RWD into credible evidence requires careful study design. The most common approaches are observational, meaning researchers don’t assign treatments but instead study patterns in data that already exists or is being collected during routine care.
Cohort studies follow two groups of patients, one exposed to a treatment and one not, and track their outcomes over time. These can run forward (identifying patients now and watching what happens) or backward (using historical records to reconstruct what happened). Case-control studies work in reverse: researchers start with patients who developed a particular outcome and look back to identify what exposures or treatments may have contributed. Cross-sectional studies capture a snapshot at a single point in time, measuring both exposure and outcome simultaneously.
Pragmatic clinical trials sit at the intersection of traditional trials and observational research. They randomize patients to different treatments but do so within real clinical settings, with fewer restrictions on who can enroll and how care is delivered. This hybrid approach preserves some of the bias-reducing power of randomization while generating results that better reflect actual practice.
How RWE Is Used in Drug Regulation
The 21st Century Cures Act, passed by Congress in 2016, required the FDA to create a formal framework for evaluating RWE in regulatory decisions about drugs and biologics. Specifically, the law directed the FDA to develop guidance on using RWE to support approvals of new uses for previously approved drugs and to fulfill post-approval safety monitoring requirements.
The FDA has since released multiple guidance documents, including three in 2024 alone covering how to assess EHR and claims data for regulatory purposes, considerations for non-interventional studies, and how to integrate randomized trials into routine clinical practice. These documents lay out what the agency expects in terms of data quality, study design, and transparency when companies submit RWE to support regulatory decisions.
RWE has already played a direct role in drug approvals. In 2021, the FDA approved fosdenopterin for a rare and often fatal genetic condition called molybdenum cofactor deficiency type A. The pivotal study combined data from patients in clinical trials with medical records from an expanded access program across 15 countries, and used a natural history study of untreated patients as the control group. In 2024, atidarsagene autotemcel was approved for metachromatic leukodystrophy, a devastating neurological disease in children, based on an externally controlled trial that used natural history data and outcomes from untreated siblings as the comparator. In both cases, traditional placebo-controlled trials would have been impractical or unethical given the severity and rarity of the diseases.
Beyond approvals, the FDA uses RWE for ongoing safety surveillance. The Sentinel system, for example, monitors insurance claims and EHR data across millions of patients to detect safety signals for marketed drugs. The agency also uses claims data in collaboration with Medicare to evaluate how well annual flu vaccines work in real populations.
The European Approach
The European Medicines Agency (EMA) has built its own infrastructure for generating RWE. In 2022, it launched the Data Analysis and Real World Interrogation Network, known as DARWIN EU. By 2026, this network is expected to include roughly 40 data partners providing information from approximately 280 million patients across Europe. The EMA and national drug regulators in the network can draw on this data at any point during a medicine’s lifecycle, from initial approval through long-term safety monitoring.
Limitations and Quality Concerns
The biggest challenge with RWE is that the data wasn’t collected for research purposes. When a doctor writes a note in a patient’s chart, they’re documenting care, not running a study. This creates several problems.
Missing data is the most commonly reported issue. A patient might switch doctors, stop filling prescriptions, or have a diagnosis that never gets coded properly. Inconsistent data collection across different hospitals, health systems, and countries means datasets can have gaps that are difficult to fill and that can skew results.
Without randomization, selection bias is an inherent concern. Doctors prescribe different treatments to different patients for reasons tied to disease severity, other health conditions, or patient preferences. If sicker patients are more likely to receive a particular drug, that drug may appear less effective in the data, not because it doesn’t work, but because it was given to people who were harder to treat. Researchers use statistical techniques to adjust for these differences, but no method can fully account for factors that weren’t measured or recorded.
Other forms of bias also affect RWE. Information bias arises when data is collected inconsistently or coded differently across sites. Detection bias occurs when one group of patients is monitored more closely than another, making outcomes more likely to be caught in that group. Recall bias can influence studies that rely on patients remembering past events or behaviors. Differences in how outcomes are measured, how treatments are started, and how long patients stay on therapy all contribute to variation between RWE findings and traditional trial results. One large-scale analysis found that three specific design differences explained most of the discrepancies between RCT results and their real-world counterparts: whether treatment began in a hospital (which claims data often misses), whether patients stopped certain baseline therapies at the start of a trial (something that doesn’t happen in routine care), and whether short medication persistence in the real world caused researchers to miss delayed drug effects.
What RWE Means for Patients
For patients, the growing use of RWE translates into several practical benefits. It means treatments can potentially reach the market faster, especially for rare diseases where running a traditional trial with thousands of patients is impossible. It means regulators can detect safety problems with approved drugs sooner by monitoring millions of real patients rather than waiting for voluntary adverse event reports. And it means clinical evidence increasingly reflects the full range of people who actually take medications, including populations that have historically been underrepresented in trials, such as older adults, minority communities, and people managing multiple chronic conditions.
The FDA is actively studying disparities in adverse drug events across minority populations using claims and EHR data, an effort that would be difficult to pursue through traditional trials alone. As data infrastructure improves and analytical methods mature, the role of RWE in shaping which treatments are available, how they’re monitored, and who benefits from them will only grow.