A Multi-Stage Drop-the-Loser Design With Superiority Boundaries

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Pharmaceutical Research

Background:

• Multi-arm multi-stage (MAMS) trials offer efficiency but can increase maximum sample size, posing funding challenges for academic research.
• Drop-the-loser designs reduce maximum sample size by dropping treatments at interim stages.
• Existing designs may not optimally balance expected and maximum sample sizes or allow for early trial stopping for superiority.

Purpose of the Study:

• To propose an enhanced multi-stage drop-the-loser design for clinical trials.
• To incorporate early stopping for trial superiority while maintaining reduced maximum sample size.
• To improve overall trial efficiency by lowering both expected and maximum sample sizes.

Main Methods:

• Development of an enhanced multi-stage drop-the-loser design.
• Derivation of analytical expressions for type I error rate, power, and expected sample size.
• Comparison of the proposed design against standard drop-the-loser and traditional MAMS designs using a motivating atrial fibrillation trial.

Main Results:

• The proposed design substantially reduces expected sample size compared to standard drop-the-loser designs.
• The enhanced design lowers the maximum sample size relative to traditional MAMS trials.
• Analytical derivations confirm the design's performance characteristics.

Conclusions:

• The enhanced multi-stage drop-the-loser design offers a superior balance of reduced expected and maximum sample sizes.
• This design provides a valuable, efficient alternative for academic-led clinical trials, particularly in areas like atrial fibrillation research.
• Implementation guidelines and suitable contexts for the proposed design are discussed.