A non-parametric test for interval-censored failure time data with application to AIDS studies

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Interval-censored failure time data analysis is common, with existing methods like score tests under continuous proportional hazards models.
• However, specific challenges arise when failure times are inherently discrete or observed on a discrete scale.

Purpose of the Study:

• To propose a novel non-parametric statistical test for comparing failure time distributions when data is discrete.
• To generalize the widely used logrank test for right-censored data to accommodate discrete failure times.

Main Methods:

• Development of a non-parametric test specifically designed for discrete failure time data.
• The proposed test is a generalization of the standard logrank test, adapted for interval-censored and discrete observations.

Main Results:

• Simulation studies were conducted to evaluate the performance of the proposed test.
• The results indicate that the new non-parametric test performs satisfactorily in comparing discrete failure time distributions.

Conclusions:

• The developed non-parametric test offers a viable method for analyzing discrete failure time data.
• This approach extends the utility of logrank-type tests to situations with discrete time scales, addressing a gap in current methodologies.