04/30/2025

**Photo by** **Antoine Schibler** on **Unsplash**

In March, 2025, the European Medicines Agency (EMA) released a “Reflection paper on use of real-world data (RWD) in non-interventional studies (NIS) to generate real-world evidence (RWE) for regulatory purposes” as a guidance for stakeholders on quality planning, conduct and analysis of NISs that intend to use RWD to generate RWE. RWD/RWE has become increasingly popular in clinical research, especially post-COVID, due to its adaptive features and general feasibility and availability while still being reliable in support for marketing applications of investigational products. However, as the paper details, there are considerations stakeholders should take into account in order to maintain the quality and reliability of the data derived from these studies.

In regards to the study design, NISs with RWD should follow the methodological standards found in the European Network of Centres for Pharmacoepidemiology and Pharmacovigilance (ENCePP) Guide on Methodological Standards in Pharmacoepidemiology. Prior to the development of the protocol, a feasibility assessment should be conducted and study objectives selected distinguishing between if they are descriptive or causal. During the development of the protocol, the paper addresses the potential for biases, confounding and effect modifications and provides suggestions to help avoid bias and unknown confounding, and limits to effect modification so it does not negatively impact the generalizability of the results. The biases include:

Selection Bias: the selection of study sites and populations, inclusion and exclusion criteria, and the research question should all be addressed and justified on scientific grounds.
Information Bias: For both primary and secondary data collection, the measurement of outcomes/variables should be accurate and avoid misclassification of variables which can occur during:
- Diagnosis,
- Coding,
- Recording,
- Data Transformations and Aggregations,
- Summarization,
- And Analyses.
Time-related bias: The timepoint of eligibility, treatment initiation and follow ups should be defined in the protocol and be similar for all study participants to avoid time-related biases.

The paper also stresses the importance of good data governance which is defined in the ENCePP Code of Conduct, and transparency to support evaluation, interpretation, and reproducibility. Data quality is broken down in the paper into 5 sections:

Reliability: The RWD should be complete, trustworthy, and credible.
Relevance: This is study-specific and should be discussed during the feasibility assessment to ensure key data elements are available, the study population is an adequate size and representative of the target population, and the study design is capable of fully answering the study question.
Multi-database studies: Heterogeneity of RWD for different sources should be taken into account.
Data Linkage: Sources may need to be linked, the methods of which should be evaluated and discussed.
Data Quality Framework: The paper recommends following the HMA-EMA Data Quality Framework for EU medicines regulation document.

The paper concludes with the statistical analyses section and goes into detail about:

Model Specification,
Estimation and Precision,
Time-Dependant Analyses,
Stratified Analyses,
Sensitivity and Supplementary Analyses,
Missing Data
And Heterogeneity.

This reflection paper closely resembles and address many of the same points as the US Food and Drug Administration’s (FDA) 2024 draft guidance “Real-World Evidence: Considerations Regarding Non-Interventional Studies for Drug and Biological Products”. The reflection paper however provides more in-depth detail on topics such as biases, data quality, and statistical analyses. We may see the final FDA guidance in the upcoming year which could benefit from the information and opinions of the EMA within this reflection paper.

-The Clinical Pathways Team

Enjoy this blog? Please like, comment, and share with your contacts.