Selection of Quantitative Decision-Making Criteria Using Weighted Decision Error Rates

Area of Science:

• Drug Development
• Decision Analysis
• Biostatistics

Background:

• Drug development involves critical go/no-go decisions.
• Errors include proceeding with a failing drug (incorrect go) or halting a successful one (incorrect no-go).

Purpose of the Study:

• To minimize combined risks of incorrect go and no-go decisions in phase 2/3 drug development.
• To introduce a weighted decision error rate (WDER) for optimizing go thresholds when error types have different costs.
• To explore the impact of prior beliefs and weighting on decision rules and phase 2 sample size.

Main Methods:

• Developing a quantitative decision-making framework.
• Defining and applying a weighted decision error rate (WDER).
• Analyzing the influence of prior beliefs and weighting on optimal decision rules.

Main Results:

• Optimal go thresholds can minimize combined decision error risks.
• WDER guides phase 2 sample size determination, often increasing size and stringency.
• Prior beliefs significantly influence optimal decision rules.

Conclusions:

• A new definition of phase 2 probability of success should focus on correct decision-making, not just advancement.
• The WDER framework enhances robustness and tailoring in drug development decision-making.
• This approach improves risk-benefit assessment for phase transitions.