A Bayesian MCMC approach to survival analysis with doubly-censored data

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Doubly-censored data, common in AIDS incubation and dementia onset studies, involves both interval-censored originating events and right-censored failure events.
• Existing methods often assume a uniform distribution for originating event times, potentially leading to biased parameter estimates.
• Accurate estimation of elapsed times between events and their relationship to risk factors is critical.

Purpose of the Study:

• To demonstrate the importance of using covariates to impute originating event times for more accurate estimation of elapsed times.
• To present a Bayesian Markov Chain Monte Carlo (MCMC) method as a suitable approach for analyzing doubly-censored data.
• To compare the performance of the proposed estimation method against conventional methods via simulations.

Main Methods:

• Development and application of a Bayesian MCMC approach for analyzing doubly-censored failure time data.
• Utilizing additional covariates to impute originating event times.
• Simulation studies to compare the proposed method with existing techniques.

Main Results:

• Imputing originating event times with covariates leads to less biased parameter estimates for elapsed time.
• The Bayesian MCMC method accommodates a rich class of survival models for doubly-censored data.
• Simulations indicate the proposed method's effectiveness compared to conventional approaches.

Conclusions:

• The Bayesian MCMC method offers a robust framework for analyzing doubly-censored data, improving accuracy in survival analysis.
• Accurate imputation of originating event times is essential for reliable estimation of elapsed times and risk factor associations.
• The approach is illustrated with practical applications in AIDS and dementia research.