Nonparametric inference on median residual life function.

Area of Science:

• Biostatistics
• Survival Analysis
• Medical Statistics

Background:

• Estimating median residual lifetime is crucial for understanding disease prognosis.
• Comparing survival between patient groups requires robust statistical methods.
• Existing methods may rely on complex density function estimations.

Purpose of the Study:

• To propose a simple method for estimating median residual lifetime.
• To develop a test statistic for comparing median residual lifetimes between two groups.
• To validate the proposed methods using simulations and real-world data.

Main Methods:

• Inverting a function of Kaplan-Meier estimators for median residual lifetime estimation.
• Developing a novel test statistic for comparing median residual lifetimes at fixed time points.
• Utilizing extensive simulation studies to assess type I error and power.
• Applying the methods to a long-term National Surgical Adjuvant Breast and Bowel Project (NSABP) dataset.

Main Results:

• The proposed method provides a straightforward estimation of median residual lifetime.
• The test statistic effectively compares median residual lifetimes without density estimation.
• Simulations confirmed the validity of the test statistic.
• Analysis of NSABP data revealed a significant median residual lifetime difference between node-negative and node-positive breast cancer patients for ~10 years without adjuvant therapy.

Conclusions:

• The new approach offers a simple and effective way to estimate and compare median residual lifetimes.
• The findings highlight a persistent survival disparity in breast cancer patients based on nodal status.
• These estimates can serve as a baseline for evaluating the impact of post-operative treatments on breast cancer outcomes.