Adjusting for informative cluster size in pseudo-value-based regression approaches with clustered time to event data

Area of Science:

• Biostatistics
• Statistical Modeling
• Epidemiology

Background:

• Informative cluster size (ICS) describes a relationship between cluster size and outcome measures in clustered data.
• Existing statistical methods may yield misleading inferences when ICS is present.
• Adjustments for ICS are crucial for accurate analysis of clustered time-to-event data.

Purpose of the Study:

• To address the challenge of applying pseudo-value regression to clustered time-to-event data with informative cluster size.
• To determine the correct strategy for adjusting for ICS in multistate models.
• To extend methodologies for intracluster group size informativeness.

Main Methods:

• Investigated inverse cluster size reweighting for ICS adjustment in multistate models.
• Employed pseudo-value regression for time-to-event data analysis.
• Conducted theoretical arguments and simulation experiments to validate adjustment strategies.
• Extended methods to account for intracluster group size informativeness.

Main Results:

• Identified and validated specific strategies for adjusting pseudo-value regression for ICS in clustered time-to-event data.
• Demonstrated the practical application of the developed methods.
• The proposed methods provide a robust framework for analyzing complex clustered data.

Conclusions:

• The study provides a clear methodology for handling informative cluster size in pseudo-value regression for clustered time-to-event data.
• Accurate adjustment for ICS is essential for reliable statistical inference in various health-related studies.
• The methods are applicable to real-world scenarios, such as periodontal and rehabilitation studies.