Marginalized models for right-truncated and interval-censored time-to-event data

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Random effects regression models are common for clustered data, but offer subject-specific interpretations.
• Marginalized Multilevel Models (MMM) and Bridge distribution models provide a unified approach for within-cluster correlations and marginal interpretation.
• Generalized Linear Mixed Models (GLMM) are conditional models with subject-specific parameter interpretation.

Purpose of the Study:

• To investigate Marginalized Multilevel Models (MMM), Bridge distribution models, and conditional Generalized Linear Mixed Models (GLMM).
• To extend these models for analyzing right-truncated, interval-censored time-to-event data with clustering and overdispersion.
• To apply these models to the hazard function for survival endpoints.

Main Methods:

• Comparison of conditional GLMM with MMM and Bridge distribution models.
• Application to right-truncated, interval-censored time-to-event data.
• Extension of models to the hazard function for survival analysis.

Main Results:

• MMM and Bridge distribution models are effective for marginal interpretation of covariate effects.
• These models successfully handle clustered, right-truncated, and interval-censored time-to-event data with overdispersion.
• The study demonstrated the utility of MMM and Bridge models in analyzing survival endpoints.

Conclusions:

• Marginalized Multilevel Models (MMM) and Bridge distribution models are valuable for analyzing clustered time-to-event data when marginal interpretation is desired.
• These approaches offer a flexible framework for complex survival data structures.
• The findings support the use of MMM and Bridge models for hazard function modeling in such contexts.