Analysis of ordered composite endpoints.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Survival Analysis

Background:

• Composite endpoints are common in clinical trials but often lack endpoint severity ranking.
• Simple methods like time-to-first-event ignore endpoint hierarchy.
• Existing ranking methods are complex and don't easily integrate with regression models.

Purpose of the Study:

• To introduce an ordering score (O) for hierarchical composite endpoints.
• To demonstrate how this score can be analyzed using standard survival analysis software.
• To provide a framework for regression modeling with hierarchical endpoints.

Main Methods:

• Defined an ordering score (O) to operationalize patient ranking based on hierarchical endpoints.
• Showed that differential right censoring corresponds to multiple interval censoring of the ordering score.
• Developed a semiparametric regression model assuming the ordering score is transformable to an exponential random variable.

Main Results:

• The proposed ordering score allows the use of standard survival models for Nonparametric Maximum Likelihood Estimators (NPMLEs).
• The semiparametric regression is equivalent to a proportional hazards model with multiple interval censoring, usable with standard software.
• The semiparametric estimator can be more efficient than simple estimators, especially in staggered entry trials.

Conclusions:

• The ordering score provides a flexible and efficient method for analyzing hierarchical composite endpoints.
• Standard statistical software can be utilized for estimation and analysis, simplifying complex trial data.
• The approach facilitates model fit assessment, flexible generalizations, and covariate interaction analysis in clinical trials.