Bayesian cure rate frailty models with application to a root canal therapy study.

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Multivariate survival data commonly arise in biomedical research due to natural or artificial clustering (e.g., multiple teeth per subject).
• A proportion of subjects may be 'cured' or insusceptible to the event of interest, necessitating models that account for this surviving fraction.

Purpose of the Study:

• To propose and evaluate two novel cure rate frailty models for analyzing correlated or clustered failure time data with a surviving fraction.
• To compare the performance of these models against the traditional Cox proportional hazards frailty model in the context of multivariate survival data.

Main Methods:

• Development of two distinct cure rate frailty models: one biologically motivated and another derived from the Cox proportional hazards frailty model.
• Formulation of likelihood functions using piecewise constant hazards.
• Derivation of full conditional distributions for Gibbs sampling within the Bayesian framework.

Main Results:

• The proposed cure rate frailty models effectively handle multivariate survival data with a cure fraction.
• Demonstrated superior performance compared to the standard Cox frailty model for data exhibiting a surviving fraction.
• Successful illustration using a real-world root canal therapy dataset.

Conclusions:

• The developed cure rate frailty models offer a robust approach for analyzing complex multivariate survival data in biomedical studies.
• These models provide a valuable alternative for situations involving a cured or insusceptible subpopulation, enhancing statistical accuracy.