Real-world data are observational data stored in repositories such as electronic health records and disease registries.
The European Medicines Agency (EMA) and the European medicines regulatory network are working to enable better integration of real-world data and real-world evidence into regulatory decisions.
For more information, see:
EMA has established a coordination centre to provide timely and reliable evidence on the use, safety and effectiveness of human medicines, including vaccines, from real-world healthcare databases across the European Union (EU).
This capability is called the Data Analysis and Real World Interrogation Network (Darwin EU).
For more information, see:
A guide is available on how EMA can help generate real-world evidence.
It is meant for EU regulators and decision-makers, including EMA's scientific committees, working parties and groups, as well as national competent authorities, healthcare technology assessment bodies and payers.
It covers:
EMA cannot consider requests from other bodies and institutions, including academia, pharmaceutical companies and contract research organisations.
The guide builds on EMA's experience in using real-world evidence to support regulatory decision-making, described in the report below.
A third report based on studies EMA conducted using real-world data between February 2024 and February 2025 is now published.
It summarises the progress made to enable the use of real-world data and establish its value in regulatory decision-making.
A list of research topics and a portfolio of use cases are also available in the report.
Read third report
The studies included in this report addressed the following research questions:
Highlights include:
For more information, see:
Find below information on the previous EMA real-world evidence reports covering the period from September 2021 to February 2024.
Select the expandable panels to read the complete reports.
Two online catalogues are available from HMA and EMA, one for real-world data sources and one for real-world data studies:
The catalogues serve to:
They enhance and replace two databases previously maintained by EMA:
| Catalogue | Discontinued database |
|---|---|
| Real-word data sources | European Network of Centres for Pharmacoepidemiology and Pharmacovigilance (ENCePP) catalogue |
| Real-world data studies | European Union electronic register of post-authorisation studies (EU PAS Register) |
The catalogues use an agreed list of metadata to describe and connect data sources to studies, using ‘FAIR’ (Findable, Accessible, Interoperable and Reusable) data principles. EMA reviews and updates the list of metadata periodically, based on feedback from catalogue users.
The following documents are also available to support users of the catalogues:
A good practice guide which provides regulators, researchers and other stakeholders with recommendations on how to use the catalogues effectively to identify and assess the suitability of data sources.
A document summarising the outcome of the public consultation on the good practice guide is also available below.
A user guide that supports users submitting data to the catalogues. It provides descriptions of the data fields and definitions, as well as guidance on how to submit and maintain records in the catalogues. EMA updates it periodically to reflect changes in the list of metadata.
EMA encourages all stakeholders to use these catalogues if they are interested in having their data:
Stakeholders include:
EMA has developed a roadmap to produce guidance on real-world evidence to support regulatory decision-making.
A document describing the roadmap and the journey that led to it is available below.
It includes a review of existing real-world evidence guidance that regulators have issued. It also proposes topics for further guidance development.
A reflection paper on non-interventional studies that use real-world data to generate real-world evidence for regulatory purposes is available.
It is relevant for all stakeholders involved in the planning, conduct and analysis of this type of non-interventional studies. This includes marketing authorisation holders and applicants.
The reflection paper was available for public consultation between May and August 2024.
Guidance is available on the methodological, regulatory and operational aspects involved in using registry-based studies to support regulatory decision-making:
EMA works to help integrate real-world evidence into regulatory decision-making across the world, within the International Coalition of Medicines Regulatory Authorities (ICMRA).
In June 2022, ICMRA held a workshop enabling regulators to share experience in obtaining and using real-world evidence for the assessment of medicines. In July 2022, it also issued a pledge to foster global efforts in this area.
For more information, see:
A reflection paper is available aiming to harmonise real-world evidence terminology and optimise the use of real-world data to support regulatory decision-making.
For more information, see:
Questions and answers are available to clarify and facilitate understanding of certain regulatory aspects addressed by existing EMA guidance on the use of real-world data.
These aspects refer to definitions, contexts of use, data quality, transparency and methodological implications of planning and conducting clinical studies using real-world data.
The guidance documents mentioned are available on the following pages on this website:
These questions and answers do not provide additional information to that included in the guidance listed above.
Find the questions and answers in the expandable panels below:
"Real-world data" (RWD) is defined as data that describe patient characteristics (including treatment utilisation and outcome) in routine clinical practice.
"Real-world evidence" (RWE) is defined as evidence derived from the analysis of RWD.
These definitions are included in EMA's reflection paper on the use of real-world data in non-interventional studies to generate real-world evidence for regulatory purposes.
RWD may include a large variety of data from routine clinical practice, such as:
A high-level summary of commonly used data source types is available in the ICH M14 guideline on general principles on plan, design and analysis of pharmacoepidemiological studies that utilize real-world data for safety assessment of medicines.
For more information, see:
In addition, the EMA guideline on registry-based studies defines and describes patient registries.
This guideline highlights that the term "product registry" is sometimes used to indicate a system of data collection set up by marketing authorisation applicants / holders to target patients exposed to a specific medicinal product or class of products.
However, from a regulatory perspective, recruitment of these patients to determine the use, efficacy / effectiveness or safety of medicinal products and follow-up should be considered as a clinical study. In this situation, the term "product registry" should be avoided.
For more information, see:
RWD can be collected for a specific study, in which case it referred to as primary data collection.
Primary data collection has been defined as the collection of data directly from patients, caregivers, healthcare professionals or other persons involved in patient care.
RWD can also be included in subsequent analyses as secondary use of already existing data.
Secondary use of data corresponds to the use of existing data for a different purpose (i.e. a specific study) than the one for which they were originally collected.
The ICH reflection paper on pursuing opportunities for harmonisation in using real-world data to generate real-world evidence acknowledges that different regulatory agencies apply different definitions of RWD / RWE.
Work is ongoing at international level through the development of the ICH E23 guideline on considerations for the use of real-world evidence (RWE) to inform regulatory decision-making with a focus on effectiveness of medicines to address challenges caused by this current lack of convergence.
A high-level summary of commonly used data source types is available in the ICH M14 guideline on general principles on planning, designing, analysing, and reporting of non-interventional studies that utilise real-world data for safety assessment of medicines.
For more information, see:
The relevance of RWD for clinical drug development to support a regulatory application or commitment depends on the research question.
If RWD is proposed to be used, early discussions with regulators through the interaction pathways in place will help understand:
The interaction pathway can include protocol assistance or scientific advice.
If a feasibility assessment is performed, it should ideally be discussed with regulators before developing the full study protocol (see the question and answer on 'What is a feasibility assessment and why should it be performed?').
Examples of where the use RWD to fill knowledge gaps are available in the EMA reflection paper on the use of real-world data in non-interventional studies to generate real-world evidence for regulatory purposes. The examples include:
For more information, see:
In 2023, the European Medicines Regulatory Network published the HMA-EMA Data quality framework for medicines regulation.
The data quality framework provides:
It integrates the definitions and recommendations proposed in several existing data quality frameworks referenced in this HMA-EMA guidance.
The target audience includes anyone involved in the management of a data source, as well as anyone planning to use and assess the data for regulatory purposes.
The framework is intended to be an overarching general resource from which to derive more focused recommendations on metrics and checks for specific domains in the context of regulatory procedures.
A specific chapter (RW-DQF) of the data quality framework outlines how the quality dimensions and assessment criteria laid down in this document can help assess the fitness-for-use of RWD to address specific research questions.
Other existing guidance documents discuss best practices and expectations on data quality in non-interventional studies, including:
For more information, see:
EMA strongly encourages to consider the RW-DQF as a best practice framework for guiding the assessment of RWD quality in regulatory contexts.
While it provides structured methodologies and tools to support robust and transparent evaluation of RWD, it does not prescribe fixed data quality thresholds or acceptance criteria, nor prioritises specific metrics.
Instead, the guidance promotes a contextual approach that enables assessors to determine the appropriateness of a dataset on a case-by-case basis, considering the context of the research purpose and questions.
This flexibility ensures that the framework can be applied across diverse data sources and use cases, while maintaining alignment with evolving regulatory expectations.
The RW-DQF outlines the quality aspects into three types of determinants:
All determinants are essential to assess the relevance of RWD in a given regulatory context.
In the RW-DQF, reliability corresponds to the quality dimension intended to measure the degree to which data correspond to what they intend to represent.
Reliability is defined by several sub-dimensions:
Each sub-dimension is characterised by specific metrics, such as:
Relevance is the quality dimension looking at the extent to which a dataset presents the data elements useful to answer a given research question.
It combines the reliability (are the data correct?) with other quality dimensions into the assessment of the fitness-for-use of given datasets to answer a specific research question to support regulatory decision-making.
These other quality dimensions include:
Existing RWD sources can be identified on their dedicated website, if they have any, or on publicly available catalogues of data sources or studies, if listed there.
The HMA-EMA Catalogues of real-world data sources and studies in the EU features two main components:
The catalogue of RWD sources facilitates the discoverability of real-world data sources. It does so by publishing a description of their characteristics and metadata. This description supports the assessment of the suitability of the data sources used to generate real-world evidence which addresses a specific research question.
As the two catalogues are linked through interoperable interfaces, the catalogue of real-world studies also provides an easy way to view the data sources used in a particular study, and vice-versa.
Data holders are encouraged to register their RWD sources in the HMA-EMA Catalogue of real-world data sources. They should also keep the publicly available information up to date. This increases transparency and facilitates data identification for research.
Other public repositories are available, such as:
The HMA-EMA Catalogues of real-world data sources and studies, the EU RDP and Orphanet are currently not technically interoperable. Therefore, data holders are encouraged to register their data sources for rare diseases in all three systems. This enables them to profit from the specific advantages provided by each platform, including:
For more information, see:
The protocol and report of a study conducted for regulatory purposes should be made public.
Amendments to protocols should also be published.
The registration requirements depend on the type of studies, as follows:
In addition, publication of study protocols before the start of data collection or extraction provides information to other researchers about the study methods. In case of secondary use of data, early publication provides confidence that the stated hypotheses have not been influenced by the results.
This can ultimately increase the level of confidence in the evidence that the study generates.
For more information, see:
The feasibility assessment is described and recommended in several pieces of guidance, including:
It is defined in the ICH M14 guideline as a systematic process to identify fit-for-use data to:
It is a preparatory step to support appropriate conduct of a study and can be reflected in the study protocol or included as an annex to the protocol.
Marketing authorisation holders or applicants should performed the feasibility assessment on their own initiative. Regulators can also request it in the course of early discussions on post-authorisation commitments.
A feasibility assessment is important for several reasons:
The information provided in a feasibility assessment depends on the following:
For more information, see:
The feasibility assessment may show that the proposed data source(s) have important limitations, such as a lack of information on confounders or an inadequate size of the study population.
It is expected that some data quality issues inherent to a data source deemed relevant for the study are difficult or impossible to resolve, and therefore that uncertainties on some data quality aspects will remain. This is according to EMA's reflection paper on use of real-world data in non-interventional studies to generate real-world evidence for regulatory purposes.
These uncertainties, their possible impact on study results and the ability to answer the research question should be clearly identified and described in the feasibility assessment. This should also be discussed with regulators.
In multi-database studies, the reliability and relevance of the data sources may differ. Removing some may reduce the amount of information available in the study and the statistical power in pooled analyses or meta-analyses, and this may not be appropriate.
For this reason, the feasibility assessment should address both the reliability and the relevance of each data source identified for the study. If data sources with different levels of quality are used, the main analysis could include the sources considered to have the highest data quality. The other sources could be used in sensitivity analyses.
There is no standard format for the presentation of the feasibility assessment.
The related ICH M14 guideline lists elements to be considered depending on the context, research question, study design and the available data source(s).
The EMA reflection paper on use of real-world data in non-interventional studies and the EMA guideline on registry-based studies describe specific aspects to be addressed in studies using RWD. This includes patient registry data.
The EMA reflection paper on use of real-world data in non-interventional studies to generate real-world evidence for regulatory purposes addresses methodological aspects related to the use of RWD to generate RWE in non-interventional studies (NIS).
The reflect paper makes the distinction between studies with descriptive objectives and studies with causal objectives.
This distinction is based on methodological aspects, depending on whether there is a causal hypothesis to be tested.
NIS can be classified within one of these two categories.
Predictive studies are generally considered as studies with descriptive objectives.
For more information, see:
The methods to be used depend on the research question, the study design and the available data.
As methodological approaches evolve with time, relevant guidance documents (including those mentioned in these Q&As) should be consulted.
Analytical methods to control for confounding should be pre-specified in the protocol.
These can be discussed at an early stage with regulators through the interaction pathways in place, such as protocol assistance or scientific advice.
For more information, see:
The TTE framework is recommended as a strategy to formalise the design and analysis of NIS with causal objectives. This includes studies on effectiveness and safety of medicines.
The framework helps to assess biases, especially time-related bias (immortal time bias) and supports transparency and replicability of studies.
It ensures that all relevant elements of the study design are explained transparently. It also helps regulators to assess the design.
There is no regulatory obligation to use the TTE framework.
Other frameworks that help identify the different components of the research question may be proposed.
Study designs such as test-negative design and self-controlled designs that are not fit for the TTE may also be appropriate for specific research questions.
The so-called estimands framework described in the ICH E9 (R1) Addendum on estimands and sensitivity analysis in clinical trials should be considered in the design of the hypothetical target trial. This refers to aspects such as the attributes of the estimands including intercurrent events, and strategies to handle them.
For more information, see:
Sensitivity analyses should be considered to assess the impact of assumptions made in the primary analysis, biases or data limitations on the study results.
They should address the same research question as the main analysis. primary estimand - see the ICH E9 addendum mentioned above.
Supplementary analyses may be considered to provide additional contextual information. This helps to better understand the study results or the impact of choices in the study design.
Supplementary analyses may address a different estimand than the main analysis.
An update log is available to show the date and summary of changes to this webpage. It does not include updates to linked documents or minor edits like typos or broken link fixes.
The tracking of updates begins in May 2026.
28 May 2026
Section on 'Guidance on use of real-world data: Questions and answers' added to clarify certain regulatory aspects addressed by EMA real-world data guidance.