Testing for Sufficient Follow-Up in Censored Survival Data by Using Extremes

Area of Science:

• Biostatistics
• Survival Analysis
• Medical Statistics

Background:

• Survival analysis often includes a 'cure fraction' where some individuals never experience the event.
• Accurate analysis requires sufficient follow-up time for all non-cured individuals.
• Existing methods for testing sufficient follow-up are limited.

Purpose of the Study:

• To develop a novel, simple test for sufficient follow-up in survival analysis with a cure fraction.
• To address the limitations of current methods for assessing follow-up adequacy.
• To specifically evaluate this assumption for light-tailed distributions.

Main Methods:

• A new test statistic is proposed, comparing estimators of the noncure proportion with and without the sufficient follow-up assumption.
• A bootstrap procedure is utilized to determine critical values for the test.
• Extensive simulations were conducted to assess the test's finite sample performance.

Main Results:

• The proposed test provides a reliable method for assessing sufficient follow-up in survival data.
• Simulations demonstrated the test's effectiveness in finite sample scenarios.
• The test was successfully applied to real-world leukemia and breast cancer datasets.

Conclusions:

• The novel test offers a valuable tool for survival analysis researchers dealing with cure fractions.
• Ensuring sufficient follow-up is critical for accurate interpretation of results in studies with potential cures.
• The method is practical and applicable to various medical datasets.