Analyzing recurrent and nonrecurrent terminal events data in discrete time.

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Joint analysis of recurrent and terminal events is crucial but lacks formal methods for discrete-time data.
• Existing discrete-time survival analysis strategies have limitations in handling correlated event types.

Purpose of the Study:

• To propose a novel discrete-time joint modeling approach for analyzing correlated recurrent and terminal events.
• To develop a flexible methodology that accounts for dependence between event types and allows for time-dependent covariates.

Main Methods:

• Developed a shared frailty model to capture dependence among recurrent events and between recurrent and terminal events.
• Incorporated rich families of transformation models for both event types.
• The method does not assume a specific frailty distribution or a Poisson process for recurrent events.

Main Results:

• The proposed joint modeling approach effectively analyzes correlated recurrent and terminal events on discrete time scales.
• Simulation studies demonstrated the utility and robustness of the method.
• Real-world applications, including biochemist promotion and scleroderma patient data, showcased its practical applicability.

Conclusions:

• The discrete-time joint modeling framework provides a statistically sound and flexible approach for analyzing complex event time data.
• This methodology advances the analysis of recurrent and terminal events, offering broader applicability in various research fields.