Clinical trials (in geClinical trials (in general) are designed/powered for efficacy analyses; however, there are important differences between efficacy analyses and safety assessments. Consequently, we are moving away from adherence on p-values (strict or implicit) for safety assessments and moving towards medical judgment within a quantitative framework.
Ronald Fisher proposed the p-value to be interpreted as a measure of statistical evidence, not as part of any formal inferential test.
Fisher suggested that it be used as part of the fluid, non-quantifiable process of drawing conclusions from observations, a process that included combining the p-value in some unspecified way with background information.
– Steven Goodman (1999)
Clinical studies emphasize strict adherence to pre-defined analyses and decision rules. This is necessary for strong control of Type I error rate for concluding superior efficacy and it’s feasible because efficacy endpoints are well-defined a priori. Neither of these conditions apply to safety monitoring. Assessing the large number and diversity of safety events requires some deviation from the traditional significance testing and multiplicity thought-process. We need to apply the rigor of the scientific method, but be willing to weigh evidence based on a collection of qualitative as well as quantitative considerations.
Regulators in the US and EU support a more systematic approach to recognize, assess, and characterize important safety information, which will inform and improve risk management. However, complete elimination of safety risks will never be possible. We need proactive safety assessments to enable effective risk management.
- Patient perspective: Allow distinct patient populations to realize important health benefits of effective drugs
- Regulatory perspective: Improve the way we identify patients at higher risk so that we can better position products in the marketplace
- Sponsor perspective: Avoid premature termination of programs that show promise even in the face of important risks
The goal is to identify, understand and manage risks so that we can deliver effective medicines to the right patients; to protect patients participating in clinical trials and after marketing, but also to save programs with favorable benefit-risk profiles for distinct patient populations.neral) are designed/powered for efficacy analyses; however, there are important differences between efficacy analyses and safety assessments. Consequently, we are moving away from adherence on p-values (strict or implicit) for safety assessments and moving towards medical judgment within a quantitative framework.
Ronald Fisher proposed the p-value to be interpreted as a measure of statistical evidence, not as part of any formal inferential test.
Fisher suggested that it be used as part of the fluid, non-quantifiable process of drawing conclusions from observations, a process that included combining the p-value in some unspecified way with background information.
– Steven Goodman (1999)
Clinical studies emphasize strict adherence to pre-defined analyses and decision rules. This is necessary for strong control of Type I error rate for concluding superior efficacy and it’s feasible because efficacy endpoints are well-defined a priori. Neither of these conditions apply to safety monitoring. Assessing the large number and diversity of safety events requires some deviation from the traditional significance testing and multiplicity thought-process. We need to apply the rigor of the scientific method, but be willing to weigh evidence based on a collection of qualitative as well as quantitative considerations.
Regulators in the US and EU support a more systematic approach to recognize, assess, and characterize important safety information, which will inform and improve risk management. However, complete elimination of safety risks will never be possible. We need proactive safety assessments to enable effective risk management.
- Patient perspective: Allow distinct patient populations to realize important health benefits of effective drugs
- Regulatory perspective: Improve the way we identify patients at higher risk so that we can better position products in the marketplace
- Sponsor perspective: Avoid premature termination of programs that show promise even in the face of important risks
The goal is to identify, understand and manage risks so that we can deliver effective medicines to the right patients; to protect patients participating in clinical trials and after marketing, but also to save programs with favorable benefit-risk profiles for distinct patient populations.