RAPID - Phase III SANEST
Dr. Danica Marinac-Dabic, U.S. Food and Drug Administration (Chair)
Dr. Christina Mack, IQVIA (Co-Chair)
Mary Beth Ritchey, MedTechEpi (Co-Chair)
Roseann White, Syntactx (Co-Chair)
Marta Steliac, U.S. Food and Drug Administration (Project Leader)
The Safety Signal Discernment and Biostatistics (SANEST) Working Group (WG) will develop multi-stakeholder consensus on best methodologies to evaluate peripheral arterial disease (PAD) devices through incorporating lessons learned from the paclitaxel (PTX) signal. This work will be informed by application of robust epidemiologic approaches to PTX signal refinement and signal evaluation.
Specifically, the WG deliverables will include consensus-driven Statistical Analysis Plans (SAPs) to help guide future PTX analyses, PTX-relevant decision trees, practical summaries and reference libraries. In addition, a portfolio of the PTX signal driven epidemiologic and statistical approaches will be submitted for publication in a special methodological supplement of a major international peer-review journal. It is anticipated that these foundational principles will be applicable to the specific PTX signal, other signals and other clinical areas.
Portfolio of efforts
The overall work informing the development of SAPs has been focusing on the best approaches to evaluate magnitude and clinical significance of signal and describe similarities and differences in statistical methods and best practices to refine and understand identified signal in real-world and Randomized Control Trials data. In the PTX context, the WG will offer best methods to (1) assess data quality, construct cohorts, and leverage data driven methodologies including Artificial Intelligence/Machine Learning/Block Chain capabilities for signal refinement; (2) assess and understand impact of missing data and how/when to apply instrumental variable analyses, sensitivity analyses and imputation (3) assess and manage misclassifications (4) apply statistical considerations to develop, assess and inform predictive models; and (5) to use this signal refinement and evaluation to inform benefit/risk assessments.