Regression analysis in an illness-death model with interval-censored data: A pseudo-value approach

Area of Science:

• Epidemiology
• Biostatistics
• Survival Analysis

Background:

• Pseudo-values are established for right-censored data analysis.
• Interval-censored data, common in dementia studies, presents analytical challenges.
• Existing methods may not fully capture complex event trajectories.

Purpose of the Study:

• To extend the pseudo-value approach for interval-censored data.
• To incorporate competing risks within an illness-death model.
• To analyze key dementia-related parameters: survival, restricted mean survival time, and absolute risk.

Main Methods:

• Developed a semi-parametric estimator using penalized likelihood and splines.
• Extended pseudo-value methodology to accommodate interval-censored outcomes.
• Applied the method to an illness-death model with competing risks.

Main Results:

• The proposed pseudo-value method effectively handles interval-censored data in dementia research.
• Simulation studies validated the properties of the new semi-parametric estimator.
• The approach was successfully applied to the PAQUID cohort data.

Conclusions:

• The extended pseudo-value method provides a robust framework for analyzing interval-censored survival data in epidemiological cohorts.
• This statistical innovation enhances the analysis of dementia progression and related outcomes.
• The method offers valuable insights for public health research on aging and neurodegenerative diseases.