A flexible copula model for bivariate survival data with dependent censoring

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Independent censoring is a standard assumption in time-to-event data analysis.
• This assumption is challenging to verify and can be problematic, especially with significant loss to follow-up due to adverse events.

Purpose of the Study:

• To address the challenges of dependent censoring in bivariate survival data analysis.
• To introduce a novel likelihood-based approach for handling dependent censoring.

Main Methods:

• Utilized a flexible Joe-Hu copula to model the interdependence of quadruple times (two events and two censoring times).
• Employed the Cox proportional hazards model to define the marginal distributions of event and censoring times.
• Developed a consistent estimator with desirable asymptotic properties.

Main Results:

• The proposed likelihood-based approach effectively analyzes bivariate survival data under dependent censoring.
• Simulation studies demonstrated the estimator's consistency and asymptotic properties.
• The method was successfully illustrated using prostate cancer data.

Conclusions:

• The developed statistical method provides a robust framework for analyzing time-to-event data when censoring is dependent.
• This approach enhances the reliability of survival analyses in the presence of adverse event-related loss to follow-up.
• The findings have practical implications for medical research, including cancer studies.