Using conditional logistic regression to fit proportional odds models to interval censored data

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Interval-censored data presents unique challenges in statistical modeling.
• Proportional odds regression is a valuable tool for analyzing such data.
• Existing methods for fitting proportional odds models to interval-censored data can be complex.

Purpose of the Study:

• To present an easily implemented approach for fitting the proportional odds regression model to interval-censored data.
• To demonstrate the applicability of the method for both interval-censored and current status data.
• To assess the accuracy of the proposed methodology through simulations.

Main Methods:

• The proposed approach utilizes conditional logistic regression routines available in standard statistical packages.
• This method bypasses complexities associated with estimating the baseline odds ratio function.
• The methodology is illustrated using data from an AIDS study comparing combination therapy (ZDV+ddC) versus ZDV alone.

Main Results:

• The conditional logistic regression approach provides a practical solution for fitting proportional odds models to interval-censored data.
• The method is shown to be applicable in various data settings, including ongoing examinations.
• Simulations indicate the procedure is accurate for assessing treatment effects, such as on CD4 cell count changes.

Conclusions:

• The presented approach offers a straightforward and effective way to fit proportional odds regression models for interval-censored data.
• This method enhances the analysis of time-to-event data in epidemiological and clinical studies.
• The use of standard statistical software makes this approach accessible to a wider range of practitioners.