The Real-World Data Quality Framework (RW-DQF) is guidance jointly developed by the European Medicines Agency (EMA) and the Heads of Medicines Agencies (HMA) that sets out how real-world data (RWD) should be assessed for quality before it is used to support regulatory decisions in the EU. Following a public consultation on a draft version, EMA published the finalized framework in late March 2026. This guide covers what the RW-DQF actually says about assessing data quality, not the general concept of real-world evidence, which is covered separately in CASRAI’s Real-World Evidence dictionary entry.
What the RW-DQF Is
The RW-DQF is a methodological framework, not a new legal requirement. It does not create a new regulatory pathway or mandate the use of RWD; instead, it gives regulators, sponsors, and data holders a shared vocabulary and structured approach for evaluating whether a given real-world dataset and the processes behind it are adequate for a specific regulatory question. It sits within the European Medicines Regulatory Network’s (EMRN) broader push to make RWD use in medicines regulation more consistent and defensible, alongside existing tools such as the HMA-EMA Catalogues of Real-World Data Sources and Studies and the coordination work of DARWIN EU (the Data Analysis and Real World Interrogation Network), EMA’s federated network for running RWD studies.
Why It Was Developed
Regulators, sponsors, and health technology assessment (HTA) bodies have increasingly relied on RWD, drawn from electronic health records, insurance claims, prescription/dispensing data, and patient registries, to supplement or, in specific circumstances, substitute for evidence traditionally generated by randomized controlled trials. That reliance exposed a gap: there was no harmonized way across the EU to describe or compare how “good” a given real-world dataset actually was for a particular use. Data quality assessment had been handled inconsistently, sometimes informally, across individual submissions and academic studies. The RW-DQF was built to close that gap by giving the EMRN a common reference point for characterizing, assessing, and documenting data quality in the regulatory context, rather than leaving each sponsor or academic group to define quality on its own terms.
Scope: What Data and What Uses It Covers
The framework is aimed primarily at secondary data, meaning data originally generated through routine clinical or administrative processes and repurposed for research or regulatory evidence generation, rather than data collected prospectively for a specific study protocol. In practice this covers:
- Electronic health records (EHR) and electronic medical records (EMR)
- Administrative and insurance claims data
- Pharmacy and prescription/dispensing data
- Patient (disease) registries
Its intended audience is correspondingly broad: national and EU regulators within the EMRN, marketing authorization applicants and holders, contract research organizations, data source owners and data networks (including DARWIN EU), academic researchers, patient organizations, and HTA bodies that rely on RWD-derived evidence.
The Quality Dimensions the Framework Describes
Rather than a single quality score, the RW-DQF organizes data quality assessment around a set of quality characteristics that reviewers and sponsors are expected to address in combination, generally covering:
- Reliability — how accurately and consistently the data were captured and recorded at the source, including coding practices and validation of key variables.
- Completeness/extensiveness — the extent to which the data needed to answer the research question are actually present, including how missing data are identified and handled.
- Representativeness — how well the population and setting captured in the data correspond to the population and setting relevant to the regulatory question.
- Coherence — internal consistency of definitions, coding, and data structure, including consistency across sites or sources when data are pooled or linked.
- Timeliness — how current the data are relative to the regulatory decision they are meant to inform.
- Relevance — fitness for the specific research question at hand, rather than fitness in the abstract.
Note on sourcing: the specific list and naming of these dimensions above is drawn from secondary reporting on the finalized framework (regulatory-affairs trade press) rather than a direct read of EMA’s own published PDF text, which could not be machine-extracted during drafting. The organizing idea, that RW-DQF assesses RWD against a defined set of quality characteristics rather than a single pass/fail score, is corroborated across independent sources; treat the exact dimension list as indicative and confirm wording against EMA’s published framework document directly (available on the ema.europa.eu regulatory-and-procedural-guideline pages) before citing it in a submission.
Fit-for-Purpose, Not Fixed Thresholds
A central design choice in the RW-DQF is that it deliberately avoids setting fixed numerical thresholds or a universal minimum quality bar. Reporting on the finalized framework describes EMA’s position as encouraging its use as a best-practice framework rather than a set of prescriptive pass/fail requirements. Quality is instead meant to be judged relative to the specific research question being asked of the data: a dataset that is perfectly adequate to answer one question (say, describing treatment patterns in a defined population) may be inadequate for a different one (say, estimating a precise comparative treatment effect) even though nothing about the underlying data has changed. This mirrors the ‘fit-for-purpose’ framing FDA has used in its own RWE guidance and rare-disease evidence work, described in CASRAI’s FDA Rare Disease Evidence Principles (RDEP) guide, though the EU and US frameworks are separate documents from separate agencies and should not be treated as interchangeable.
Relationship to Existing EU RWD Infrastructure
The RW-DQF does not stand alone. It is meant to be read alongside:
- The HMA-EMA Catalogues of Real-World Data Sources and Studies — a registry-style inventory of RWD sources and the studies run on them across the EU, which the RW-DQF complements by giving reviewers a way to actually evaluate the quality of a catalogued source, not just locate it.
- DARWIN EU — EMA’s federated data analysis and real-world interrogation network, which runs RWD studies on behalf of regulators using a distributed network of data partners; the RW-DQF’s quality criteria are directly relevant to how DARWIN EU studies document and justify the fitness of the data sources they draw on.
- ENCePP Methodological Standards — the European Network of Centres for Pharmacoepidemiology and Pharmacovigilance’s existing methodological guidance, which the RW-DQF is reported to align with rather than replace.
Sponsors preparing an RWD-supported submission should expect reviewers to draw on all of these together: locating and describing a data source via the Catalogues, applying RW-DQF criteria to justify why that source is adequate for the specific question, and, where a DARWIN EU study is involved, documenting how the study’s data quality was assessed within that infrastructure.
What This Means for Sponsors and Data Holders
In practical terms, the RW-DQF pushes data quality management earlier and makes it more explicit and auditable. Applicants relying on RWD in a regulatory submission should be prepared to document, for the specific data source and analysis in question:
- How the data were originally captured (source systems, coding practices, and any validation performed at the point of capture)
- What linkage, curation, or transformation steps were applied between raw capture and the analysis dataset, and what records exist of those steps
- What quality checks were run at each stage, including how missing or inconsistent data were identified and handled
- An explicit justification of why the dataset’s characteristics (population coverage, timeliness, completeness) are adequate for the specific regulatory question being asked, not merely a general assertion that the data source is ‘of good quality’
- The limitations of the data and analysis, documented rather than omitted
This is consistent with the broader shift already visible in comparable RWD-quality expectations from other regulators and payers, including the kind of documentation trail described in CASRAI’s Coverage with Evidence Development (CED) guide and the analytic-design discipline covered in the Target Trial Emulation guide — reviewers increasingly expect the data-quality case to be made explicitly, with the same rigor as the study design itself, rather than assumed from the data source’s reputation.
How the RW-DQF Differs from Generic RWE Guidance
CASRAI’s Real-World Evidence dictionary entry defines RWE and RWD at a conceptual level; it does not cover any specific regulator’s assessment mechanics. The RW-DQF is narrower and more operational: it is a specific EU document, produced jointly by EMA and HMA, that defines how RWD quality is actually assessed within EU medicines regulation, distinct from the FDA’s separate RWE program and guidance documents (some of which are covered in CASRAI’s RDEP guide), and distinct from methodological approaches like target trial emulation, which address study design rather than the underlying data’s quality. A sponsor working across both the EU and US needs to satisfy both agencies’ expectations, which are related in spirit (fit-for-purpose data quality, transparent documentation) but are not the same document and are not guaranteed to require identical documentation.
Frequently Asked Questions
Is the RW-DQF legally binding?
No. It is described in reporting as best-practice guidance rather than a binding legal requirement, consistent with how EMA scientific guidelines generally function: they are not law, but reviewers expect the reasoning behind a submission to be consistent with published guidance, and departing from it without justification increases regulatory risk.
Does the RW-DQF apply to primary data collected directly for a study?
Its stated focus is secondary data, meaning data generated through routine clinical or administrative practice and repurposed for regulatory evidence, such as EHR, claims, prescription, and registry data. It is not primarily aimed at data collected prospectively under a dedicated study protocol.
Does the RW-DQF replace the HMA-EMA Catalogues of RWD Sources and Studies?
No. The Catalogues are an inventory of data sources and studies; the RW-DQF is a set of criteria for assessing the quality of a given source or study once identified. They are complementary, not competing, tools.
Who is expected to apply the RW-DQF in practice?
Reporting describes an intentionally broad audience: EMRN regulators, marketing authorization applicants and holders, contract research organizations, data source owners and networks such as DARWIN EU, academic researchers, patient organizations, and HTA bodies working with RWD-derived evidence.
How does the RW-DQF relate to DARWIN EU specifically?
DARWIN EU is the operational network EMA uses to run federated RWD studies across a set of data partners; the RW-DQF supplies the quality criteria those studies (and any other RWD use in EU regulatory contexts) are expected to be assessed against. DARWIN EU is infrastructure for running studies; the RW-DQF is the standard for judging the data quality behind them.
Because the RW-DQF was finalized in March 2026 and secondary reporting on it is still consolidating, sponsors relying on this guidance for an active submission should confirm current wording directly against EMA’s published framework document rather than this or any secondary summary.
