Regression analysis of multivariate current status data with dependent censoring: application to ankylosing...

Area of Science:

• Biostatistics
• Survival Analysis
• Clinical Data Analysis

Background:

• Multivariate current-status failure time data involve multiple related event times observed at a single examination.
• Dependent censoring occurs when examination times are intrinsically linked to event times, complicating analysis.
• Such data are common in clinical trials and carcinogenicity experiments, necessitating robust statistical methods.

Purpose of the Study:

• To propose a joint frailty model that accounts for dependent censoring in multivariate current-status failure time data.
• To develop a likelihood-based estimation approach for this model.
• To evaluate the finite-sample performance of the proposed method through simulations.

Main Methods:

• Development of a joint frailty model for event times and dependent censoring time.
• Application of a likelihood approach utilizing Gaussian quadrature techniques for parameter estimation.
• Conducting extensive simulation studies to assess the method's properties.

Main Results:

• The proposed joint frailty model effectively handles dependent censoring in multivariate current-status data.
• Maximum likelihood estimates obtained via Gaussian quadrature show reliable performance in simulations.
• The method is illustrated with a real-world case study involving ankylosing spondylitis patient data.

Conclusions:

• The joint frailty model provides a statistically sound framework for analyzing complex survival data with dependent censoring.
• The developed likelihood approach with Gaussian quadrature is a viable method for obtaining accurate estimates.
• This methodology offers valuable insights for clinical studies and other fields dealing with related event and examination times.