Proportional hazards regression with interval censored data using an inverse probability weight

Area of Science:

• Biostatistics
• Medical Research Methodology
• Survival Analysis

Background:

• Interval censored data is increasingly common in medical studies, particularly with biomarker-defined disease progression endpoints.
• This type of data arises from periodic monitoring, where disease progression is identified within an interval between assessments.

Purpose of the Study:

• To propose a novel methodology for estimation and inference of regression coefficients in Cox proportional hazards models using interval censored data.
• To address the challenges posed by periodic monitoring in defining precise event times.

Main Methods:

• The proposed methodology utilizes estimating equations.
• Inverse probability weighting is employed to select unambiguous event time pairs for analysis.
• The approach is designed for Cox proportional hazards models with interval censored data.

Main Results:

• Simulations were conducted to evaluate the finite sample properties of the proposed estimation method.
• The methodology's performance was compared against the conventional partial likelihood estimate, which often ignores interval censoring.

Conclusions:

• The developed methodology offers a robust approach for analyzing interval censored data in medical research.
• This method provides a more accurate estimation of regression coefficients compared to conventional methods that disregard the nature of interval censoring.