Cure rate estimation with insufficient follow-up: A median-based bootstrap correction approach

Area of Science:

• Biostatistics
• Clinical Trials
• Medical Research

Background:

• The Kaplan-Meier (KM) estimator is commonly used for estimating cure rates in clinical trials.
• Short follow-up periods in trials can lead to overestimation of cure rates by the KM estimator.
• Existing bootstrap-based correction methods may introduce bias due to skewed distributions.

Purpose of the Study:

• To address the overestimation bias of the Kaplan-Meier estimator in clinical trials with insufficient follow-up.
• To propose a novel, robust method for estimating cure rates that mitigates bias from skewed bootstrap distributions.

Main Methods:

• Development of a median-based approach for bootstrap samples to correct cure rate estimations.
• Simulation studies to compare the proposed method with existing techniques.
• Application of the method to real-world clinical trial data.

Main Results:

• The proposed median-based bootstrap method demonstrated reduced variation due to outliers compared to existing methods.
• The new approach enabled more stable and reliable estimation of cure rates.
• Successful application to clinical trial data from a D-penicillamine study for primary biliary cirrhosis.

Conclusions:

• The median-based bootstrap approach offers a more accurate and stable method for estimating cure rates in clinical trials, especially those with limited follow-up.
• This method effectively addresses biases associated with the Kaplan-Meier estimator and skewed bootstrap distributions.
• The findings have implications for improving the interpretation of cure rates in clinical research.