Real-world data (RWD) is increasingly used to meet regulatory requirements across the medicine lifecycle, including post-marketing surveillance and benefit–risk assessments and more often in earlier regulatory decisions such as label expansions. With this broader role, there is need for more clarity on what acceptable evidence is.

The European Medicines Agency (EMA) Real-World Data Quality Framework (RW-DQF), adopted in March 2026, makes expectations clearer: it is no longer enough to have access to data; regulators expect that companies can show that the data is fit for purpose and suitable for the specific question being asked.

An important shift in RWE: from “usable data” to “defensible data”
In the past, the discussion often focused on whether RWD could be used at all. Now, the question is more specific: is this dataset appropriate for this exact regulatory question?
Under the framework, regulators are likely to place more weight on whether a dataset is:

  • Relevant to the regulatory question
  • Reliable and collected in a consistent way
  • Complete enough for the intended analysis
  • Coherent with internally consistent definitions
  • Timely in relation to the decision

None of these concepts are entirely new, but it is now more explicit, and therefore easier to apply consistently during review. A common assumption has been that large or well-known datasets can be reused across multiple studies with limited adaptation. Under a fit‑for‑purpose lens, the same dataset may be acceptable in one context and fall short in another. The difference is rarely about size; it is about how well limitations are understood, documented, and mitigated.

What this changes in practice

1. Data quality becomes more visible during review
Data quality is no longer a background activity, it directly affects how a submission is experienced by reviewers. The RW‑DQF makes it more likely that assessors will interrogate data provenance and handling in the same way they interrogate study design. In practical terms, common review questions tend to cluster around:

  • Where did the data come from, and how was it captured?
  • What is missing, and is missingness systematic?
  • Are key variables measured consistently across sites, time periods, and care settings?
  • How were records linked, curated, and transformed?
  • What validations were performed, and what did they show?

    Submissions that clearly describe these answers typically reduce follow-up questions and prevent avoidable delays.

2. RWE is moving earlier in decision-making

RWE continues to support post-authorisation commitments, but it is also increasingly used in label expansions and other earlier decisions. As a result, the standard of evidence expected for RWE is becoming more explicit and more consistently applied.

3. Confidence depends on transparency, not just results.

Even strong findings can be undermined if regulators cannot see how the data was processed, linked, validated, and analysed. This does not mean every dataset must be perfect, but it does mean that its strengths, limitations, and handling need to be visible and traceable. That needs to cover the full chain data quality management process from source systems and coding practices through linkage, curation, transformation, and analysis.

What “fit-for-purpose” really requires

Meeting the framework’s expectations is less about a final check but more about end-to-end data quality management, including:

  • Selection data sources that match the regulatory need (and documenting why)
  • Pre-defining “fit-for-purpose” criteria (e.g., key variables, follow-up time, missingness thresholds)
  • Applying quality checks during data ingestion, linkage, and curation and not only at the end
  • Keeping clear records of data transformations and validation steps
  • Being explicit about limitations and mitigation steps.

This kind of approach is sometimes described as “regulatory-grade” RWE. In simple terms, it means data that can be explained, not just analysed.

Looking ahead

The RW-DQF does not introduce entirely new concepts, it makes expectations clearer. Organisations that plan for data quality early, and document it clearly, will be better positioned to navigate regulatory review with fewer avoidable questions and delays. The RW-DQF turns data quality into a central part of regulatory assessment. Not just whether data is used but whether it can be defended. That shift – from having data to justifying it – is likely to shape how real-world evidence is generated and reviewed in the years ahead.