NONPARAMETRIC TESTING FOR MULTIPLE SURVIVAL FUNCTIONS WITH NON-INFERIORITY MARGINS

Area of Science:

• Biostatistics
• Survival Analysis
• Clinical Trials

Background:

• Non-inferiority trials with multiple treatments present challenges for comparing survival functions.
• Existing nonparametric tests are limited to pairwise comparisons or lack censoring considerations.
• Nonparametric likelihood ratio statistics offer potential for more powerful tests.

Purpose of the Study:

• To develop new nonparametric tests for ordering multiple survival functions with right-censored data.
• To address the limitations of existing methods in complex clinical trial settings.
• To provide a complete solution for multiple treatment comparisons in non-inferiority trials.

Main Methods:

• Development of novel nonparametric likelihood ratio statistics.
• Introduction of a new pool adjacent violator algorithm for ordered alternatives.
• Characterization of limit distributions using Gaussian processes and projections.

Main Results:

• The proposed tests demonstrate superior power compared to a combined-pairwise Cox model approach.
• The new methods provide a comprehensive solution for ordering multiple survival functions.
• Simulation studies validate the effectiveness of the developed procedures.

Conclusions:

• The novel nonparametric tests are effective for multiple survival function ordering in non-inferiority trials with censoring.
• These methods offer a powerful alternative to existing approaches for complex clinical trial data.
• The developed pool adjacent violator algorithm is a key advancement in this statistical area.