Quantile regression for doubly censored data.

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiological Research

Background:

• Registry studies frequently encounter doubly censored data, where observations are censored from both the left and right.
• Existing statistical methods may not adequately address the complexities introduced by double censoring in survival data analysis.

Purpose of the Study:

• To propose a new quantile regression-based analysis strategy for handling doubly censored data.
• To develop computationally efficient estimation and inference procedures for doubly censored data.
• To adapt the proposed method for left-truncated data and address identifiability issues.

Main Methods:

• Utilizing a quantile regression model tailored for doubly censored data.
• Employing the embedded martingale structure for estimation and inference.
• Developing conditional inference procedures to resolve identifiability challenges.
• Demonstrating adaptability to left-truncated data.

Main Results:

• Established asymptotic properties, including uniform consistency and weak convergence, for the proposed estimators.
• Simulation studies confirmed the good finite-sample performance of the new inferential procedures.
• Successfully applied the method to analyze Pseudomonas aeruginosa onset in cystic fibrosis patients.

Conclusions:

• The proposed quantile regression strategy provides a computationally simple and effective approach for analyzing doubly censored data.
• The method is robust and can be extended to handle left-truncated data, offering broader applicability in survival analysis.
• The analysis of cystic fibrosis data highlights the practical utility of this novel statistical approach in real-world epidemiological research.