Exploring the heterogeneity in recurrent episode lengths based on quantile regression.

Area of Science:

• Biostatistics
• Epidemiology
• Chronic Disease Research

Background:

• Recurrent episode data are common in chronic disease studies.
• Understanding episode length heterogeneity is crucial for tailored disease management.
• Existing methods lack direct interpretation or make unrealistic assumptions.

Purpose of the Study:

• To propose a novel statistical modeling strategy for recurrent episode data.
• To specifically address the heterogeneity in recurrent episode lengths.
• To overcome limitations of existing approaches in chronic disease research.

Main Methods:

• Utilizing quantile regression to model episode lengths.
• Incorporating time-dependent covariates into the model.
• Treating recurrent episodes as clustered data, handling dependent censoring, truncation, and informative cluster size.

Main Results:

• The proposed method offers direct interpretation of episode lengths.
• The estimation procedure is computationally simple with desirable asymptotic properties.
• Numerical studies show advantages over naive adaptations of existing methods.

Conclusions:

• The developed quantile regression strategy effectively analyzes recurrent episode lengths in chronic diseases.
• This approach provides a more flexible and interpretable tool for disease management.
• The method addresses complex data features like dependent censoring and truncation.