Estimating Time to Event From Longitudinal Categorical Data: An Analysis of Multiple Sclerosis Progression

Area of Science:

• Neurology
• Biostatistics
• Medical Statistics

Background:

• The Expanded Disability Status Scale (EDSS) is crucial for measuring multiple sclerosis (MS) progression.
• Traditional survival methods are insufficient for EDSS data, failing to utilize end-of-follow-up information.
• There is a need for advanced statistical models to accurately analyze repeated ordinal EDSS measurements.

Purpose of the Study:

• To introduce and validate a Markov transitional model for analyzing time to progression in multiple sclerosis (MS) based on EDSS scores.
• To demonstrate the model's ability to derive covariate-specific survival curves from repeated ordinal data.
• To provide a flexible and implementable statistical framework for MS progression analysis.

Main Methods:

• Development and application of a Markov transitional model for ordinal data.
• Estimation of regression coefficients and manipulation of transition matrices.
• Utilizing large sample theory and resampling methods for confidence intervals.
• Explicitly defining methods for various progression endpoints (e.g., EDSS level, increase, consecutive visits).

Main Results:

• The proposed Markov model effectively analyzes repeated EDSS data, outperforming traditional survival methods.
• Covariate-specific survival curves can be accurately derived.
• Pointwise confidence intervals demonstrate good performance in simulations.
• The model is applicable to different definitions of MS progression, including time to two consecutive EDSS > 3 visits.

Conclusions:

• The Markov transitional model provides a robust and adaptable approach for analyzing multiple sclerosis (MS) progression using EDSS data.
• This method enhances the understanding of disease trajectories and the impact of covariates.
• The model is readily implementable in standard statistical software, facilitating its widespread use in clinical research.