A multistate additive relative survival semi-Markov model

Area of Science:

• Biostatistics
• Epidemiology
• Medical Statistics

Background:

• Medical researchers analyze relationships between variables and event times (e.g., disease progression, death).
• Multistate models are used for multiple times-to-events.
• Semi-Markov multistate models are relevant for chronic diseases, as transition intensities depend on time spent in the current state.

Purpose of the Study:

• To propose a novel Semi-Markov Additive Relative Survival (SMRS) model.
• To combine multistate modeling and relative survival analysis.
• To estimate net survival when causes of death are unknown or not solely disease-related.

Main Methods:

• Development of the Semi-Markov Additive Relative Survival (SMRS) model.
• Application of the SMRS model to a French cohort of kidney transplant recipients.
• Validation using simulated data to assess model effectiveness.

Main Results:

• The SMRS model effectively combines multistate and relative survival approaches.
• The model's usefulness is demonstrated through applications in a kidney transplant cohort.
• Simulated data analysis shows the SMRS model's results approximate those obtained when causes of death are known.

Conclusions:

• The proposed SMRS model offers a valuable tool for analyzing complex survival data in chronic diseases.
• It provides a method to estimate net survival and excess mortality when detailed cause-of-death information is unavailable.
• The SMRS model enhances the analysis of times-to-events in medical research, particularly for conditions with competing risks.