The Trial Without Walls: How Real-World Evidence Is Rewriting the Rules of Drug Approval

Picture the scene: a regulatory affairs director at a mid-sized oncology sponsor opens her laptop on a Tuesday morning to find a new perspective piece in Nature Medicine arguing that, under the right methodological conditions, real-world data drawn from routine clinical practice can replace the traditional randomized control arm entirely. She reads it twice. Then she pulls up the IND her team filed eighteen months ago for a Phase 3 oncology asset, built around a 600-patient parallel-group RCT, and starts doing uncomfortable arithmetic. If the framework described in that paper holds regulatory water, her trial is already overbuilt, overpriced, and potentially overdue for a redesign she did not budget for.

That discomfort is not hypothetical anymore. It is landing in strategy meetings across the industry right now, and the sponsors who treat it as a theoretical academic exercise will spend the next three years explaining to their boards why they ran a trial they did not need to run.

The core argument in the Nature Medicine piece is deceptively simple: when the real world generates enough high-quality longitudinal data on a disease, that data can be structured to emulate the control arm of a randomized trial. Pre-register the study design, define the eligibility criteria to mirror what an RCT would have enrolled, apply rigorous propensity scoring to minimize confounding, and you have something that begins to approximate the counterfactual you would have built in a clinic. The methodology has a name, trial emulation, and it has been accumulating academic credibility for years. What is shifting in 2026 is the distance between that academic credibility and actual regulatory acceptance.

The Regulatory Gap You Cannot Afford to Misread

The FDA has been signaling openness to real-world evidence as a regulatory input since at least its 21st Century Cures Act mandate, and the agency has backed that signal with action. In July 2021, the FDA approved a new indication for Astellas Pharma’s Prograf (tacrolimus) for prevention of organ rejection in lung transplant patients, relying in part on real-world evidence to support the submission. That precedent matters. It demonstrated that RWE can carry weight in a new indication package, not just as supportive color commentary but as substantive evidentiary input. The global RWE solutions market, now valued at $4.7 billion in 2024 with the drug development and approvals segment holding the largest share, reflects how aggressively the industry has already moved to capitalize on that signal.

But the FDA’s framework for how that evidence gets evaluated remains frustratingly incomplete for sponsors trying to operationalize it. The agency published its draft guidance on “Considerations for the Design and Conduct of Externally Controlled Trials for Drug and Biological Products” in 2023, and it provides useful scaffolding around pre-registration, eligibility alignment, and confounding mitigation. What it does not provide is disease-specific thresholds. It does not tell a sponsor running a rare hematologic malignancy program exactly how many real-world patients constitute sufficient external control arm power, or how the agency will adjudicate endpoint ascertainment from EHR data versus adjudicated trial endpoints. Those gaps are not editorial omissions. They are the regulatory terrain that will determine whether a trial emulation strategy survives a Type B meeting or collapses under reviewer skepticism.

The ICH has begun to close some of that international distance. At the June 2024 ICH Assembly, the body endorsed a Reflection Paper on Real-World Evidence outlining a strategic framework for integrating RWE across a medicine’s development lifecycle. The paper is not binding guidance, and it stops well short of specifying acceptable data sources by indication. But its endorsement signals that the major regulatory jurisdictions are converging on the principle that RWE can inform regulatory decisions, even as they diverge on the mechanics of when and how.

Here is the counterintuitive truth that most sponsors are missing: the academic rigor of trial emulation methodology is not the primary barrier to its regulatory acceptance. The data quality of the underlying real-world sources is.

Where the Methodology Meets the Mess

Consider what happened when researchers compared real-world data against aggregated RCT data for small cell lung cancerlung cancer patients. A 2024 preprint study found significant differences in survival outcomes between the two populations, differences large enough to materially affect how you would interpret a treatment effect if your control arm came from routine practice data rather than a randomized comparator. The patients in real-world datasets carry more comorbidities, receive more variable supportive care, and are subject to documentation patterns that reflect clinical convenience rather than protocol discipline. An overall survival signal that looks robust against an RWD comparator may look considerably less impressive when that comparator is properly characterized.

That is not a reason to abandon the methodology. It is a reason to take data validation with the same seriousness sponsors currently reserve for statistical analysis plans. The FDA’s guidance on real-world data for regulatory decision-making, analyzed in a 2022 review in npj Digital Medicine, makes this explicit: the agency focuses on the adequacy and relevance of RWD to the specific research question, alongside validation of data quality and provenance of study elements. In operational terms, that means sponsors cannot simply license a claims database, run a propensity score model, and present the output as a validated external control arm. They need to demonstrate, at the source level, that the data was collected in a way that makes the endpoints they are measuring meaningful.

In December 2025, the FDA moved to expand RWE’s operational footprint further, finalizing guidance that allows sponsors to submit de-identified real-world evidence for medical device submissions without patient identification data, drawing from sources including electronic health records. That is a meaningful step toward lowering the operational friction of building RWE packages. But de-identification introduces its own linkage and completeness challenges that sponsors running multi-source RWD datasets will need to resolve before the data is fit for a regulatory submission.

The sponsors who will navigate this most effectively are not the ones with the largest RWE platforms. They are the ones who treat RWD validation as a prospective activity, not a retrospective cleanup job.

What This Means for the People Running the Trials

Operationally, the shift toward trial emulation and external control arms demands a set of competencies that most clinical operations teams have not built. A standard Phase 3 protocol development process does not include a data provenance assessment of the EHR system at the institutions whose records will serve as the external comparator. A standard CRO contract does not include a workstream for validating endpoint ascertainment consistency across the real-world data sources being used alongside the interventional arm. These are not small additions to existing workflows. They require dedicated RWD operations expertise sitting alongside traditional clinical operations, not downstream from it.

For sponsors considering a hybrid design, where a reduced-size randomized arm is supplemented by an RWD-derived external comparator, the pre-IND and Type B meeting strategy needs to be rebuilt from the ground up. The 2023 FDA draft guidance on externally controlled trials provides a starting framework, but it is a floor, not a ceiling. Sponsors should request explicit agency feedback on their proposed eligibility mapping between the real-world population and the intended trial population, their approach to time-zero alignment in the emulation design, and their plan for handling missing data in the RWD sources. Getting those positions documented in meeting minutes before protocol lock is the only way to avoid a complete response letter that questions the comparability of your external control arm two years later.

Evidence density in the RWE package matters as much as evidence density in the clinical package. The Astellas Prograf precedent from 2021 involved a disease area, lung transplant rejection, where the real-world population and the treated population were sufficiently similar that the agency found the RWE credible. Oncology sponsors working in tumor types with highly heterogeneous treatment histories face a materially harder comparability argument, and the burden of demonstrating that their RWD cohort is not systematically different from what a randomized trial would have enrolled falls entirely on the sponsor.

That regulatory affairs director staring at her Phase 3 IND on Tuesday morning faces a genuine decision point. If her disease area has a mature, longitudinal real-world data infrastructure, if her endpoints can be ascertained reliably from EHR and registry data, and if she can pre-register an emulation design that the FDA has at least provisionally blessed in a Type B meeting, then a hybrid or fully externally controlled design could reduce her trial size, compress her timeline, and get a therapy to patients faster. The Nature Medicine framework gives her the methodological vocabulary to make that argument. What she needs now, before she redesigns anything, is a Type B meeting request on her calendar and a data provenance assessment already underway on every RWD source she plans to use. Because the FDA will ask for both, and the sponsors who answer those questions first will have an enormous structural advantage over the ones who are still building their RWE infrastructure when the review clock starts.

References

1. Nature Medicine — “When the real world becomes the trial”
2. RAPS — “FDA Official Offers Examples of RWE Being Used to Secure Drug Approval” (Prograf/tacrolimus, July 2021)
3. FDA Law Blog — “FDA’s Draft Guidance on Externally Controlled Trials Answers Some Questions, Leaves Others Unanswered” (2023)
4. ICH — “Reflection Paper on Real-World Evidence,” Endorsed by ICH Assembly, June 4, 2024
5. medRxiv — “Comparison of Real-World Data with Aggregated RCT Data for Small Cell Lung Cancer” (2024)
6. PMC / npj Digital Medicine — “FDA Guidance on Real-World Data for Regulatory Decision-Making: Data Integrity and Validation Standards” (2022)
7. Drug Discovery News — “FDA Guidance Opens the Door to De-Identified Real-World Evidence in Regulatory Submissions” (December 2025)
8. MarketsandMarkets — “Real World Evidence Solutions Market: $4.7 Billion Valuation, 2024”