Real-World Evidence to Support Causal Inference: Methodological Considerations for Non-Interventional Studies

Executive Summary

The aim of non-interventional studies, a type of study in which patients receive the marketed drug of interest during routine medical practice and are not assigned to an intervention according to a protocol, is to uncover insights that may be inaccessible through controlled trials. Non-interventional or observational study designs can play a crucial role in assessing treatment effects (i.e., causality) beyond the confines of traditional randomized controlled trials (RCTs). Within these study designs, routine clinical care outcomes are observed among real-world populations, as opposed to research participants of RCTs selected according to narrow inclusion/exclusion criteria. Real-world data (RWD), derived from sources such as electronic health records, claims data, and registries, offers a less constrained environment that better reflects the complexity and diversity of clinical practice. Additionally, real-world studies typically have much larger sample sizes, facilitating subgroup analyses often infeasible in RCTs. Subgroups, in this case, describes an analysis unit of a subset of participants within a given study population. This nuanced understanding can inform health care decision-making by capturing real-world outcomes, patient variability, and long-term effects of interventions observed as part of regular clinical care.

Real-world evidence (RWE) complements RCTs by providing timely insights into effectiveness across diverse populations beyond traditional clinical trials. Regulatory initiatives, such as the U.S. Food and Drug Administration’s (FDA) Advancing Real-World Evidence Program, acknowledge the value of RWE, aiming to modernize evidence generation and incorporate patient perspectives. However, ensuring the credibility of RWE for causal inference requires clear design, fit-for-purpose RWD, communication, and rigorous statistical analysis. Promoting RWE’s capacity for causal inference is essential for advancing evidence-based health care. Regulators recognize that certain limitations accompany the use of RWD to determine or measure causality. Proposed approaches might involve established concepts like target trial emulation and/or other causal frameworks to address confounding and other types of bias and schemas to describe overall study designs. Integrating RWE’s strengths with traditional research methods like RCTs can present a more comprehensive understanding of health care interventions and their real-world impacts.