Real World Evidence FDA: Use in Drug Development and Regulatory Decisions
Real-world evidence (RWE) is clinical evidence about the usage and potential benefits or risks of a medical product derived from analysis of real-world data (RWD). The 21st Century Cures Act (Section 3022) requires FDA to establish a framework for evaluating RWE in regulatory decision-making for drugs. FDA distinguishes between RWD (data collected outside traditional clinical trials) and RWE (evidence generated from RWD analysis). FDA accepts RWE for certain regulatory decisions including label expansions, post-marketing requirements, and safety assessments, but the evidentiary standard depends on the specific regulatory question and the quality, relevance, and reliability of the underlying data.
Key Takeaways
Key Takeaways
- RWD is the raw data (EHRs, claims, registries); RWE is the clinical evidence generated by applying rigorous analytical methods to RWD — not all RWD produces valid RWE.
- The 21st Century Cures Act Section 3022 mandates FDA's RWE framework specifically for drugs, covering label expansions, post-marketing requirements, and safety assessments.
- FDA evaluates RWE fitness based on two dimensions: whether the underlying RWD is fit for use (relevance, reliability, completeness) and whether the study design adequately addresses confounding and bias.
- FDA places weight on prospectively defined protocols, analysis plans, and transparent documentation of any changes made during the study.
Definitions: RWD vs RWE
FDA's "Framework for FDA's Real-World Evidence Program" (December 2018) established the foundational definitions:
Real-World Data (RWD): Data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources. RWD is the raw material. Sources include electronic health records (EHRs), medical claims and billing data, product and disease registries, patient-generated data (including from wearables and mobile devices), and data gathered from other sources that can inform on health status (e.g., social determinants of health).
Real-World Evidence (RWE): Clinical evidence regarding the usage and potential benefits or risks of a medical product derived from analysis of RWD. RWE is the output of applying analytical methods to RWD to answer specific clinical or regulatory questions.
The distinction is critical: not all RWD produces valid RWE. The quality of RWE depends on the quality, completeness, and relevance of the underlying RWD and the rigor of the analytical methods applied.
The Data-to-Evidence Pipeline
| Stage | Description | Example |
|---|---|---|
| Data collection | RWD generated during routine clinical practice | EHR records from 50 hospital systems |
| Data curation | RWD extracted, cleaned, standardized | Mapping EHR data to a common data model (OMOP CDM) |
| Study design | Research question defined, analytical plan created | Retrospective cohort study comparing two treatment strategies |
| Analysis | Statistical methods applied to curated RWD | Propensity score matching, Cox regression, sensitivity analyses |
| RWE generation | Results interpreted in clinical and regulatory context | A comparative effectiveness or safety estimate suitable for regulatory evaluation |
Legislative Mandate: 21st Century Cures Act
Section 3022: Real-World Evidence
Section 3022 of the 21st Century Cures Act (December 2016) added Section 505F to the FD&C Act, requiring FDA to:
- Establish a program to evaluate the potential use of RWE to support approval of new indications for approved drugs and to support or satisfy post-approval study requirements (including accelerated approval confirmatory studies)
- Establish a draft framework within two years (fulfilled by the December 2018 Framework document)
- Issue guidance on the use of RWE
- Convene public meetings and seek stakeholder input
The legislative mandate specifically addresses drugs (not devices). However, CDRH has separately developed policies for using RWD/RWE in device regulatory decisions.
Scope of the Cures Act RWE Mandate
| In Scope | Out of Scope |
|---|---|
| Approved drugs seeking new indications | Medical devices under CDRH's separate device-specific framework |
| Post-marketing study requirements | Questions that cannot be answered with fit-for-purpose RWD and an appropriate study design |
| Supplemental applications for approved drugs | Standalone assumptions that RWE automatically substitutes for controlled evidence |
| Post-marketing safety assessments |
FDA's RWE Framework
The December 2018 Framework Document
FDA's "Framework for FDA's Real-World Evidence Program" identifies three areas where RWE may support regulatory decision-making:
- Supporting or satisfying post-marketing study requirements
- Adding or modifying an indication for use (label expansion)
- Supporting drug safety assessments
RWE Fitness for Use
FDA evaluates RWE fitness for regulatory purposes based on two dimensions:
1. Whether the RWD is fit for use:
| Factor | What FDA Evaluates |
|---|---|
| Relevance | Does the data capture the outcomes, exposures, and covariates needed to answer the regulatory question? |
| Reliability | Is the data collected consistently, accurately, and completely? Are there quality assurance procedures? |
| Completeness | Are there systematic missing data patterns that could bias results? |
| Accrual | Is the data of sufficient volume and duration to support the analysis? |
| Data standards | Does the data use standardized coding systems (ICD, NDC, LOINC, CPT)? |
| Linkage | Can data from multiple sources be linked at the patient level while maintaining privacy? |
2. Whether the study design is fit for use:
| Factor | What FDA Evaluates |
|---|---|
| Causal inference | Does the study design adequately address confounding and bias? |
| Comparator selection | Is the comparator group appropriate and well-defined? |
| Outcome ascertainment | Are outcomes measured validly and consistently across exposure groups? |
| Sensitivity analyses | Are results robust to alternative analytical approaches and assumptions? |
| Transparency | Is there a pre-specified protocol and analysis plan, with changes documented clearly? |
| Reproducibility | Can the analysis be replicated by independent researchers? |
RWD Sources
Electronic Health Records (EHRs)
| Attribute | Details |
|---|---|
| Data elements | Diagnoses, procedures, medications, lab results, vital signs, clinical notes, imaging reports |
| Strengths | Rich clinical detail, longitudinal follow-up, captures clinical context |
| Limitations | Inconsistent coding, missing data, variable data quality across institutions, unstructured data in clinical notes |
| Common data standards | HL7 FHIR, C-CDA, OMOP CDM |
| FDA relevance | Used in safety surveillance, label expansion studies, and natural history studies when fit for purpose |
Medical Claims and Billing Data
| Attribute | Details |
|---|---|
| Data elements | Diagnoses (ICD codes), procedures (CPT/HCPCS), medications (NDC), enrollment, demographics, costs |
| Strengths | Large populations, standardized coding, comprehensive capture of healthcare utilization |
| Limitations | Coding driven by reimbursement rather than clinical accuracy, limited clinical detail, no lab results or vital signs in most claims databases |
| Sources | CMS Medicare/Medicaid, commercial insurers (Optum, MarketScan, IQVIA) |
| FDA relevance | Drug utilization studies, safety signal detection, comparative effectiveness |
Product and Disease Registries
| Attribute | Details |
|---|---|
| Data elements | Disease-specific outcomes, treatment details, biomarkers, patient-reported outcomes, long-term follow-up |
| Strengths | Disease-specific data collection, structured and standardized, often includes outcomes not captured in EHRs/claims |
| Limitations | Selection bias (voluntary enrollment), limited sample sizes for rare diseases, variable data quality |
| Examples | Cystic Fibrosis Foundation Patient Registry, SEER cancer registry, National Cardiovascular Data Registry (NCDR) |
| FDA relevance | Post-market surveillance, natural history studies, rare disease drug development |
Patient-Generated Data
| Attribute | Details |
|---|---|
| Data elements | Activity, sleep, heart rate, blood glucose (from wearables/sensors), patient-reported outcomes, symptom diaries |
| Strengths | Continuous monitoring, captures daily life outside clinical settings, patient-centered outcomes |
| Limitations | Data quality concerns, compliance/adherence to data collection, device variability, privacy issues |
| Sources | Wearable devices, mobile health apps, patient portals |
| FDA relevance | Endpoint development, digital biomarker validation, decentralized trial endpoints |
Sentinel System
FDA uses the Sentinel Initiative as a distributed data approach for active safety surveillance and related evaluations. For this guide, the important point is not a fixed network size but that FDA has established infrastructure for using routinely collected data in post-market surveillance and certain other regulatory contexts.
Use Cases for RWE in Regulatory Decisions
Use Case 1: New Indication (Label Expansion)
FDA has accepted RWE to support supplemental applications for new indications in limited circumstances. The evidentiary standard is high, and RWE is most likely to be accepted when:
- The drug is already approved for a related indication
- The new indication involves a population where randomized trial data exists for similar patients
- RWE is used to supplement, not replace, randomized trial data
- The regulatory question is whether efficacy observed in a clinical trial population extends to a broader population
FDA evaluates label-expansion proposals using RWE on a case-by-case basis. Whether RWE is persuasive depends on the regulatory question, the fit-for-purpose quality of the underlying data, and the study design.
Use Case 2: Post-Marketing Requirements and Commitments
FDA can require or request post-marketing studies as a condition of drug approval. In some cases, RWE can satisfy these requirements:
| Post-Marketing Study Type | RWE Applicability |
|---|---|
| Post-marketing safety study (21 CFR 314.80) | RWE from Sentinel, claims data, or registries often suitable |
| Post-marketing efficacy study (accelerated approval) | RWE less commonly accepted; generally requires confirmatory clinical trial |
| REMS (Risk Evaluation and Mitigation Strategy) assessment | RWE can supplement REMS evaluation data |
| Pediatric study requirements (PREA) | RWE may supplement clinical trial data in limited circumstances |
Use Case 3: Safety Assessments
RWE is well-established for drug safety assessments. FDA routinely uses RWD/RWE for:
- Signal detection: Identifying potential safety signals from spontaneous reporting (FAERS), claims data, and EHR data
- Signal evaluation: Evaluating identified signals through formal epidemiological studies using RWD
- Labeling changes: Supporting safety-related labeling changes (warnings, contraindications, boxed warnings)
- REMS assessment: Evaluating whether REMS elements are effective at mitigating identified risks
Use Case 4: Drug Utilization Studies
RWE is used to understand how drugs are used in real clinical practice:
- Off-label use patterns and prevalence
- Prescribing patterns across demographics and geographic regions
- Medication adherence and persistence
- Concomitant medication use and polypharmacy
Use Case 5: Natural History Studies
For rare diseases, RWE from registries and EHRs can establish disease natural history, which is critical for:
- Designing clinical trials (selecting endpoints, estimating effect sizes)
- Serving as external control arms in single-arm trials
- Supporting regulatory decisions for rare disease therapies under accelerated approval or breakthrough therapy designation
Study Design Considerations
Observational Study Designs
| Design | Description | Strengths | Limitations |
|---|---|---|---|
| Retrospective cohort | Follow exposed and unexposed groups using historical data | Efficient, large samples, multiple outcomes | Confounding, selection bias, data quality |
| Prospective cohort | Follow exposed and unexposed groups forward in time | Better data quality than retrospective | Expensive, time-consuming, loss to follow-up |
| Case-control | Compare cases (with outcome) to controls (without) | Efficient for rare outcomes | Selection bias, recall bias, confounding |
| Self-controlled | Compare risk periods within the same patient | Controls for time-invariant confounders | Only for transient exposures and acute outcomes |
| Cross-sectional | Measure exposure and outcome at a single time point | Quick, inexpensive | Cannot establish temporality |
Pragmatic Clinical Trials
Pragmatic clinical trials (PCTs) combine elements of RCTs and observational studies, similar in philosophy to decentralized clinical trials. PCTs randomize patients but use RWD for outcome ascertainment and are conducted in routine clinical settings rather than controlled research environments.
FDA's guidance "Use of Real-World Evidence to Support Regulatory Decision-Making for Medical Devices" (August 2017, updated) and draft guidances on RWE for drugs discuss PCTs as a potential source of RWE that may address some confounding concerns of purely observational studies.
| PCT Characteristic | Description |
|---|---|
| Setting | Routine clinical practice (not research centers) |
| Population | Broad eligibility, minimal exclusion criteria |
| Intervention | Standard clinical care, not protocol-driven |
| Outcomes | Collected from EHRs, claims, registries (not case report forms) |
| Randomization | Maintained (key difference from observational studies) |
Addressing Confounding and Bias
The primary challenge in using RWE for regulatory decisions is the risk of confounding and bias inherent in non-randomized data. FDA expects RWE studies to address these challenges through:
| Method | What It Addresses |
|---|---|
| Propensity score matching/weighting | Measured confounders |
| Instrumental variable analysis | Unmeasured confounders (in specific settings) |
| Negative control analyses | Residual confounding detection |
| Sensitivity analyses | Robustness to analytical assumptions |
| Quantitative bias analysis | Systematic assessment of bias impact |
| Active comparator design | Channeling bias, healthy user bias |
| New-user design | Prevalent user bias |
| Prospectively defined protocol and analysis plan | Reduces post-hoc analytical bias and improves transparency |
FDA RWE Guidance Documents
Published Guidances
| Guidance | Date | Scope |
|---|---|---|
| Framework for FDA's Real-World Evidence Program | December 2018 | Overall framework for RWE in drug regulation |
| Use of Electronic Health Record Data in Clinical Investigations | July 2018 | Use of EHR data in clinical investigations |
| Submitting Documents Using Real-World Data and Real-World Evidence to FDA for Drug and Biological Products | September 2022 (final) | Submission identification and organization for RWD/RWE-containing drug and biologic submissions |
| Data Standards for Drug and Biological Product Submissions Containing Real-World Data | December 2023 (final) | Data standards expectations for submissions containing RWD |
| Considerations for the Use of Real-World Data and Real-World Evidence to Support Regulatory Decision-Making for Drug and Biological Products | August 2023 (final) | Applicability of IND regulations and core design considerations for studies using RWD |
| Real-World Data: Assessing Electronic Health Records and Medical Claims Data to Support Regulatory Decision-Making for Drug and Biological Products | July 2024 (final) | Considerations for proposing use of EHR and claims data |
| Real-World Data: Assessing Registries to Support Regulatory Decision-Making for Drug and Biological Products | December 2023 (final) | Considerations for using registry data |
Key Concepts from FDA Guidance
Regulatory-grade RWD: FDA has not formally defined "regulatory-grade" RWD, but the concept refers to RWD of sufficient quality, reliability, and relevance to support regulatory decision-making. Key attributes include:
- Provenance and traceability of data
- Data validation and quality assurance procedures
- Standardized data collection and coding
- Adequate capture of relevant clinical variables
- Sufficient sample size and follow-up duration
Transparency and prospective planning: FDA's RWE guidances emphasize having the study protocol and analysis plan specified up front and preserving a transparent record of design decisions and changes. The relevant expectation is methodological transparency, not a single universal public-registration rule for every RWE use case.
Regulatory precedent: FDA evaluates RWE submissions on a case-by-case basis. No single study design or data source is automatically accepted or rejected. The key question is whether the totality of evidence (RWE plus any other available evidence) provides sufficient assurance of safety and effectiveness for the specific regulatory decision.
Practical Considerations
Engaging FDA Early
For sponsors planning to use RWE to support regulatory decisions, early engagement with FDA is essential:
| Interaction Type | When to Use |
|---|---|
| Pre-IND or Pre-NDA meeting | To discuss whether RWE can support a specific regulatory question |
| Type C meeting | For specific questions about RWE study design or data source suitability |
| PDUFA Pre-Submission meeting | Before submitting an sNDA with RWE as a primary evidence source |
Common FDA Review Concerns with RWE
| Reason | Description |
|---|---|
| Inadequate confounding control | Observational study with unmeasured or poorly controlled confounders |
| Data quality concerns | Outcome misclassification, incomplete exposure data, data entry errors |
| Selection bias | Study population not representative of the target population |
| Insufficient sample size | Underpowered to detect clinically meaningful differences |
| Post-hoc analysis | Protocol or analytical decisions were not defined prospectively or changes were not documented transparently |
| Missing data | Systematic missing data without adequate sensitivity analysis |
| Inconsistent with RCT data | RWE contradicts available randomized trial evidence without adequate explanation |
Building Regulatory-Ready RWD Infrastructure
Organizations planning to generate RWE for regulatory purposes should invest in:
- Data governance: Policies for data quality, access, privacy, and security
- Data structuring and standards: Use data structures and submission standards that FDA can review efficiently
- Data quality assessment: Systematic evaluation of data completeness, accuracy, and timeliness
- Analytical expertise: Biostatisticians and epidemiologists experienced in causal inference methods
- Regulatory strategy: Early FDA engagement to align on data fitness and study design
- Transparency: Pre-registration of analytical plans and public availability of study protocols
Key Regulatory References
| Document | Source | Year |
|---|---|---|
| 21st Century Cures Act, Section 3022 | U.S. Congress | 2016 |
| Framework for FDA's Real-World Evidence Program | FDA | 2018 |
| Use of Electronic Health Record Data in Clinical Investigations | FDA | 2018 |
| Submitting Documents Using Real-World Data and Real-World Evidence to FDA for Drug and Biological Products | FDA | 2022 |
| Considerations for the Use of Real-World Data and Real-World Evidence to Support Regulatory Decision-Making for Drug and Biological Products | FDA | 2023 |
| Data Standards for Drug and Biological Product Submissions Containing Real-World Data | FDA | 2023 |
| Real-World Data: Assessing Registries to Support Regulatory Decision-Making for Drug and Biological Products | FDA | 2023 |
| Real-World Data: Assessing Electronic Health Records and Medical Claims Data to Support Regulatory Decision-Making for Drug and Biological Products | FDA | 2024 |