HMN 2026: What are the key strategies for global drug safety studies

A new study published in Therapeutic Innovation & Regulatory Science provides crucial insights into the design and implementation of multi-national observational studies for post-marketing drug safety assessment, addressing growing needs for global evidence in regulatory decision-making.

The research, which we led along with colleagues from major pharmaceutical companies and the U.S. Food and Drug Administration, examines how combining data from multiple countries can strengthen our understanding of medication safety in real-world populations. Our collaborative team included researchers from Moderna Inc., Takeda Development Center Americas Inc., Pfizer Inc., AbbVie Inc., Merck & Co., Inc., and U.S. Food and Drug Administration.

Why multi-national studies matter

As we explain in our research, relying on data from a single country or database can limit the generalizability of safety findings. Different populations, health care systems, and prescribing patterns across countries can significantly impact how medications perform in practice. Multi-national studies address these limitations by pooling diverse data sources, boosting statistical power to detect rare adverse events, and enabling researchers to verify findings across different populations.

“The utilization of multi-national observational study design has risen in recent years,” we note in the paper. “This study design enables the examination of limited exposures and/or rare outcomes by boosting the statistical power of the investigation. Furthermore, this approach enables researchers to evaluate the generalizability of study findings across diverse populations.”

Postmarketing safety observational studies leveraging large, comprehensive health care databases can provide important insights into adverse drug events, especially for rare outcomes, newly marketed products, or specific patient populations that are not well-represented in clinical trials. By leveraging multi-national data sources, researchers can enhance the robustness and validity of their findings while minimizing potential biases inherent in single-database studies.

Navigating complex challenges

Through our research, we identified critical design considerations that researchers must address when planning multi-national safety studies:

• Data Integration Challenges: Different countries use varying coding systems, health care structures, and data collection methods. We discuss two main approaches for addressing this complexity—Common Data Models (CDMs) like OMOP that standardize data formats, and Common Protocol Models (CPMs) that ensure consistent study methods while allowing flexibility in data sources. CDMs enable seamless integration and enhance comparability, though they require significant initial effort for data mapping and validation. CPMs, conversely, allow use of raw data without extensive preprocessing but may face challenges with data quality variability.
• Defining Outcomes Consistently: Medical coding systems and diagnostic practices vary internationally, which can significantly impact study results. For instance, we note that definitions of congenital heart defects can vary significantly between the U.S. National Birth Defects Prevention Network and the European EUROCAT network, affecting outcomes and comparability of safety assessments. Researchers must develop standardized outcome definitions and rigorous validation procedures to ensure consistency.
• Exposure Assessment: Characterizing drug exposure across countries presents unique challenges. As we describe in our paper, researchers must map unique drug codes from different data sources, ensure temporal consistency in exposure periods, handle missing data appropriately, and account for differences between prescribing data (reflecting provider intentions) and dispensing data (closer to actual patient usage). For example, Denmark’s medical data doesn’t contain days’ supply information, which must be routinely assumed—a complication that can affect duration of medication use calculations.
• Controlling for Confounding: This remains one of the most significant threats to validity in multi-national studies. Health care practices, insurance coverage patterns, and available treatments differ across countries, requiring country-specific statistical models while maintaining consistent analytical methods. We emphasize that neither Common Data Models nor Common Protocol Models can fully remove variability in data structures between different sources or relevant differences in prescribing patterns and health care practices across regions.
• Statistical Considerations: The researchers emphasize the importance of analyzing data separately by country before pooling results, assessing consistency across regions, and conducting sensitivity analyses to evaluate how methodological choices impact findings.

Offering practical recommendations

The study offers actionable strategies for researchers and regulatory agencies:

• Clearly define study objectives that apply across all participating countries while accounting for regional differences
• Develop standardized protocols with transparency through public registration
• Use country-specific causal models to account for differences in confounding variables
• Conduct thorough sensitivity analyses to assess the robustness of findings
• Engage local stakeholders to provide insights into regional nuances and enhance data quality

Looking ahead: Implications for drug safety and regulation

Multi-national studies are particularly valuable for investigating rare populations or outcomes, such as pregnancy exposures or newly approved medications with limited real-world data. There is growing support for forming international consortia to address safety concerns at the therapeutic class level, which would pool data, standardize methods, and accelerate evidence generation—particularly valuable for pregnancy studies where long accrual times often delay critical safety insights.

Recent guidance from the International Council for Harmonization, including the draft M14 guideline on pharmacoepidemiological studies utilizing real-world data, is expected to further increase the use of multi-national studies globally.

As noted in the paper, “The implementation of this guidance across the globe will increase the use of multi-national (with multi-database) pharmacoepidemiological studies.”

“These studies enhance the robustness and validity of findings while minimizing biases inherent in single-database studies,” we write. “By leveraging diverse data sources and innovative analytical approaches, we can advance pharmacoepidemiologic research and improve public health outcomes.” The increasing adoption of multi-national pharmacoepidemiological studies underscores the need for collaborative efforts among researchers, regulatory agencies, and health care stakeholders to generate robust evidence for informing drug safety assessments and regulatory decision-making.

This work also highlights the critical importance of engaging local stakeholders, including experts, regulatory agencies, and health care professionals in different regions. This collaboration provides invaluable insights into regional nuances, enhances data quality, facilitates access to comprehensive safety data, and assists in interpreting safety outcomes within local contexts.

As real-world evidence continues to evolve, leveraging diverse data sources and innovative design and analytical approaches will be crucial in advancing pharmacoepidemiologic research and improving public health outcomes.

This story is part of Science X Dialog, where researchers can report findings from their published research articles. Visit this page for information about Science X Dialog and how to participate.

Li-An Lin, Ph.D., is the Senior Director of Medical Affair Statistics at Scholar Rock. Before joining Scholar Rock, he was a Director and Head of Safety Statistics at Moderna and Biostatistician at Merck following the completion of his Ph.D. at The University of Texas. His research interests encompass Adaptive Trial Design, Causal Inference, Evidence Synthesis, Survival Analysis, and Bayesian Analysis. In recent years, Lian has been actively involved in developing innovative tools for clinical trial data processing, analysis, reporting, and visualization. He is an active member of the ASA Safety Working Group and co-leads the initiative on integrating RCT and RWE for safety decision-making.

The content is provided for information purposes only.