FDA offers guidance on observational studies as RWE
The US Food and Drug Administration (FDA) this week issued new draft guidance on study design and analysis for non-interventional studies being included as part of a regulatory submission.The draft guidance is one of several guidance documents the agency has issued on the use of real-world data and evidence (RWD/RWE) in regulatory decision making. In August 2023, FDA issued final guidance related to data access and safety monitoring considerations for RWD/RWE, particularly non-interventional clinical study designs. (RELATED: FDA finalizes guidance on real-world evidence in drug approvals, Regulatory Focus 31 August 2023)
“The reliability and relevance of real-world data (RWD) used in a non-interventional study are critical for making appropriate causal inferences and are essential to establishing the data’s fitness for use in generating real-world evidence to support a labeling change or address a safety concern,” FDA wrote in the latest draft guidance.
The current guidance was issued by the FDA’s Center for Drug Evaluation and Research, the Center for Biologics Evaluation and Research, and the Oncology Center of Excellence in response to growing interest in using non-intervention study designs – including observational studies and case-control studies – to contribute to the demonstration of safety and effectiveness of drugs.
The draft guidance covers the agency’s proposed approach to non-interventional studies, study designs, data sources and analytical approaches.
FDA approach
The draft guidance defines non-interventional studies as those in which patients receive a marketed drug during routine medical practice and are not assigned to an intervention per a research protocol. The relevant study designs include observational cohort studies, case-control studies, and self-controlled studies where the patient serve as their own control.
In the guidance, FDA proposes that sponsors finalize the study protocol before beginning the study, including selecting the research question of interest and the rationale for the study design. When proposing a non-interventional study for use in regulatory decision-making, sponsors should also summarize the other study approaches and data sources they considered and why they were not feasible to answer the study questions.
“The discussion should reflect an in-depth understanding of the use of the drug(s) of interest and the outcome(s) of interest, as well as the capture of exposure, outcome(s), and relevant covariates in the proposed study population,” FDA wrote.
FDA suggests that sponsors provide the following information when proposing a non-interventional study:
- Research question and hypothesis
- Rationale for using a non-interventional study approach
- Choice of specific study design
- Proposed data sources and alternatives considered
- Results of preliminary studies to assess data sources as fit for use
- Proposed approach to support causal inference, potential confounding and bias
- Description of the approach for addressing ethical considerations
Study design elements should be developed based on the prespecified research question, according to FDA. Protocols should describe the critical study elements, including:
- Schema of the overall design and a causal diagram
- Source population
- Eligibility criteria and study population
- Conceptual and operational definitions for key variables of interest and validation efforts
- Relevant covariates and strategies to address bias
- Index date for all study arms and strategies to address bias related to “immortal time”
- Start and end of follow-up period, approach to censoring, and anticipated losses to follow-up
“Given that data sources used in a non-interventional study design are often generated for purposes other than research, it is important that sponsors understand the potential limitations of such data sources and determine whether those limitations can be addressed or if another data source should be pursued,” FDA wrote in the draft guidance.
The prespecified statistical analysis plan (SAP) should address the study objectives, as well as provide details on the primary and secondary analyses, FDA wrote. The SAP should provide information on several key elements, including addressing the statistical power of the study, the statistical approach to evaluating the treatment effect, ways to account for potential confounders and unmeasured confounders, and approaches for handling subgroup analyses and missing data, among others.
Comments on the draft guidance can be submitted by 18 June 2024 through regulations.gov using FDA Docket No. FDA-2023-D-5470.
Draft Guidance
