Select the Right Data Collection Method for Real-World Evidence
Choosing the wrong RWE data source undermines your regulatory strategy. This framework helps pharma leaders select the right real-world data collection method based on your programme’s research question.
Selecting the right real-world evidence (RWE) data collection method is one of the most consequential decisions in assessing the usage, benefits, and risks of a medicinal product or medical device.
Each primary RWE data source carries distinct strengths and limitations, yet the choice is routinely made too late in the planning process. Both the U.S. FDA and EMA have significantly raised their standards for what counts as credible RWE.
Understanding Real-World Data and Real-World Evidence
Real-world data (RWD) is health information gathered from routine clinical practice outside controlled trial settings.
RWD includes hospital records, pharmacy dispensing data, insurance claims, disease registries, and data from wearable devices.
In other words, RWD is the raw input from which the evidence (RWE) is built.
Real-world evidence is the clinical knowledge produced when RWD is analysed using a validated, pre-specified study design.
This distinction matters because gathering data in a real-world setting does not automatically produce submission-grade evidence. The quality of the analytical framework determines whether the data output satisfies regulatory standards.
RWE serves multiple roles across the drug development lifecycle. It supports label expansion submissions, post-market safety monitoring, health technology assessment dossiers, and comparative effectiveness analyses.
For medical devices, it also underpins post-market clinical follow-up requirements under EU Medical Device Regulation (MDR). It complements randomised trial evidence rather than replacing it.
Under the U.S Food and Drug Administration (FDA) guidance on real-world evidence, both the quality of RWD and the rigour of the analytical method must meet defined standards before real-world-derived findings can inform a regulatory decision.
The FDA's 21st Century Cures Act framework and the European Medical Agency's parallel guidance formalise these requirements for sponsors.
Why Real-World Data Source Selection Matters
Every major RWD source has limitations in what evidence it can reliably generate. Selecting a a RWE source based on what is already licensed, rather than what the research question requires, is the single most common design error in RWE programmes.
Administrative claims data covers large patient populations with prescription-level detail. It works well for treatment-pattern and adherence studies. It cannot support outcomes research requiring biomarker data, imaging findings, or clinical severity grading.
Electronic health records (EHRs) offer clinical depth but vary considerably in documentation quality across sites and systems. Patient registries give longitudinal integrity but carry selection bias risks if enrolment criteria are not carefully specified.
Wearables and pragmatic trials each apply to narrow question types and carry their own compliance requirements under ICH E6(R3) GCP guidelines.
A mismatch between data source and research question creates errors that statistical methods cannot correct at the analysis stage.
Regulatory reviewers at the FDA, EMA, and UK National Institute of Cost Effectiveness (NICE) identify source-question misalignment during submission review.
Maximise Pharma Innovation With Quality-by-Design
Find out how Quality-by-Design improves reliability and compliance in pharma and how you can maximise innovation through quality-based pharma development systems.
Matching Your Research Question to the Right RWD Source
To match your question to the right RWD source, start by defining the research question precisely before any data access conversation begins.
The question must specify the target population, the exposure, the comparator, the outcome of interest, and the required follow-up duration.
Administrative claims data is the most practical starting point for treatment-pattern and medication adherence questions at the population scale.
Clinical outcome studies requiring biomarker or imaging data need EHRs, with natural language processing (NLP) applied to unstructured clinical notes.
The latest FDA guidance on EHR and claims data requires full documentation of any extraction algorithm, including its assumptions and validation approach.
Rare disease research and studies requiring long follow-up periods are best served by purpose-built registries despite their higher setup cost.
Causal comparative questions need either a pragmatic clinical trial or a target trial emulation with a pre-specified statistical analysis plan.
Post-marketing safety surveillance works best with real-time EHR and claims feeds supported by automated pharmacovigilance platforms.
A significant proportion of clinically meaningful EHR information sits in unstructured text: Physician notes, discharge summaries, and radiology narratives. NLP tools now extract this RWD data at scale and are expected to be validated and documented before use in submissions.
What Regulators Now Expect From RWE
The FDA and EMA's 2024 RWE frameworks, respectively, both require sponsors to apply methodological standards comparable to those used in interventional trials.
This means pre-specifying the analysis plan, reporting all identified sources of bias transparently, and maintaining full data governance documentation.
The NICE real-world evidence framework applies the same standard to health technology assessment submissions in England.
Therefore, building a global evidence strategy requires alignment with all three frameworks from the point of study design, not at the point of submission.
The DARWIN EU network connects 20 data partners across 13 EU member states, covering records from 130 million patients. Because the EMA uses this infrastructure for its own regulatory research, its methodology sets a direct benchmark for how European RWE submissions are evaluated.
Selection bias, inadequate sample size, and missing data are the three most frequently cited reasons RWE studies fail regulatory review. All three must be addressed prospectively at the design stage.
Five Actions for Medical Affairs and HEOR Teams
For Medical Affairs and Health Economics and Outcomes Research (HEOR) teams, it is important to establish a formal data source selection protocol requiring written justification of the RWE data source choice before any data access agreement is signed. This single step prevents the most common and costly error in RWE programme design.
Teams should focus on building their NLP capability (in-house or through specialist partners) to unlock clinical information held in unstructured EHR records.
Additionally, teams should embed propensity score matching and target trial emulation as standard tools in your analytical workflow, not optional enhancements.
Bring Regulatory Affairs into the study design process as a co-design function from the outset of any programme intended for regulatory or HTA use.
Lastly, monitor regulatory methodology, especially from the DARWIN EU network, as the analytical standards it applies represent the practical benchmark for European regulatory review.
The Looming Threat of Geopolitical Risk in Clinical Supply Chains
Geopolitical risk is reshaping how clinical supply chains are managed, demanding resilience and foresight. Learn why building supply chain resilience is the key for clinical trial success.
Conclusion: RWE Means the Method Is in the Message
RWD is the raw clinical information collected outside clinical trials, while RWE is the evidence produced by analysing RWD with validated methods. The two terms are not interchangeable, and regulators now expect sponsors to treat them as distinct with separate quality requirements.
Each of the five primary data sources suits a different category of research question.
Claims data suits adherence and treatment-pattern questions; EHRs suit clinical outcome studies; patient and disease registries suit rare disease and longitudinal research. Pragmatic trials and wearables apply to narrower use cases with specific compliance obligations.
Regulatory standards for RWE are converging globally, across the FDA, EMA, and NICE. All three regulatory bodies now expect pre-specified analysis plans, transparent bias reporting, and full data governance documentation.
Source-question misalignment, selection bias, and missing data remain the leading causes of submission failure.
Pharmatica tracks regulatory developments, evidence standards, and analytical methodologies across the global RWE landscape, giving medical affairs and HEOR leaders the intelligence they need to build programmes that get regulatory approval.
Pharmatica: Insight. Connection. Impact.
Frequently Asked Questions
What is the difference between real-world data and real-world evidence?
The difference between real-world data and RWE is that RWD is the input, while RWE is the output. The data quality depends entirely on how rigorously the RWE study is designed and conducted.
Simply, RWD is health information gathered from routine care settings outside clinical trials, including electronic health records, claims, registries, and wearables. Real-world evidence is the clinical knowledge produced by applying validated analytical methods to that data.
What is real-world evidence used for in pharma and medical devices?
RWE in pharma and medical devices is used to support regulatory submissions for new indications and label expansions, post-market safety surveillance, health technology assessment dossiers, and comparative effectiveness research.
For medical devices, RWE also underpins post-market clinical follow-up under EU MDR. It answers questions about real-world patient populations that controlled trials cannot address, complementing rather than replacing RCT evidence.
Which data sources are most commonly used for RWE in pharma?
The five main sources commonly used for RWE in pharma are electronic health records, administrative claims data, disease-specific patient registries, patient-generated data from wearables and apps, and pragmatic clinical trials. Claims data suits population-level treatment-pattern research. EHRs suit clinical outcome studies. Registries suit rare disease and long-term follow-up research.
What are the most common reasons RWE studies fail regulatory review?
The most common reasons RWE studies fail regulatory review are: Selection bias, inadequate sample size, and missing data.
Regulators at the FDA, EMA, and NICE all assess these areas first. Addressing them through propensity score matching, sensitivity analyses, and pre-specified missing data handling at the design stage is the minimum requirement for regulatory use.
What does the FDA's 2024 RWE guidance require from sponsors?
The July 2024 FDA guidance on EHR and claims data requires sponsors to document all data sources fully, validate extraction algorithms including AI-based tools, and keep data access agreements available for inspection.
The FDA’s RWE guidance requires sponsors to address data quality, representativeness, and inherent source bias before submission. AI methods used to extract RWD must be documented with assumptions and validation steps disclosed.
About the author
Did you enjoy the content?
Why not support Nicole Dale by giving this content a like
Comments (0)