Censored quantile regression model with time-varying covariates under length-biased sampling

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Quantile regression effectively models survival data with time-varying covariates.
• Prevalent cohort data often yields length-biased observations due to informative right censoring.
• Existing methods may not adequately address length-biased data with time-dependent covariates.

Purpose of the Study:

• To propose an estimating equation-based approach for consistent estimation in survival analysis with length-biased, time-dependent covariate data.
• To develop a numerically stable variance estimation procedure.
• To establish large sample properties of the proposed estimator.

Main Methods:

• Developed an estimating equation-based approach for length-biased survival data.
• Incorporated time-dependent covariates into the quantile regression framework.
• Implemented a stable variance estimation procedure inspired by Zeng and Lin (2008).

Main Results:

• The proposed method yields consistent estimators for regression coefficients.
• The variance estimation procedure is shown to be numerically stable.
• Large sample properties, including consistency and asymptotic normality, are theoretically established.

Conclusions:

• The novel method provides reliable analysis for survival data with length-biased observations and time-varying covariates.
• The approach demonstrates robust performance in simulations across various scenarios.
• The method is validated through application to the Oscar dataset.