Statistical Methodology in Safety Monitoring
*Melvin Munsaka, Takeda 

Keywords:

In the efforts to better promote public health and protect patient safety, there is growing interest in developing systematic approaches for safety evaluation of pharmaceutical products. This happens not only in post-marketing safety surveillance, but also in pre-marketing safety monitoring. Recent regulatory guidance, such as CIOMS VI, ICH E2C (periodic benefit risk evaluation report) and the FDA IND safety reporting guidance, highlight the importance and provide recommendations on aggregate safety monitoring. Safety monitoring in clinical development in a broad sense, spans from individual trial level to program level, from unblinded to blinded, from expected to unexpected adverse events and from patient profiles to aggregate safety tables and figures, supporting DMC and benefit-risk assessments. It also serves to lay out the foundation for Integrated Analysis of Safety preparation and benefit-risk assessment in the Clinical Overview and possibly in Advisory Committee Meetings. Quantitative safety scientists can closely engage with clinical and regulatory scientists and play an important role in these efforts. To better enable this, the ASA Biopharmaceutical Section in 2015 established a Safety Working Group. One of the focus areas of the working group is safety monitoring. This presentation will discuss part of the work of the working group. The discussion will focus on the various statistical methodological approaches for safety monitoring in pharmaceutical development life cycle. It will include methods for blinded versus unblinded, frequentist versus Bayesian, premarketing versus post marketing, static versus dynamic, visual analytics in safety monitoring, and trial-level versus program-level safety data aggregation.