The analysis of multivariate recurrent events with partially missing event types.

Area of Science:

• Biostatistics
• Clinical Trial Methodology
• Epidemiology

Background:

• Clinical studies often involve recurrent events, where event types may be incompletely recorded.
• Analyzing such incomplete data is crucial for understanding disease progression and treatment efficacy.

Purpose of the Study:

• To develop a statistical method for analyzing multi-type recurrent events with incomplete data.
• To estimate conditional intensity functions, treatment effects, and event associations.
• To address challenges in summarizing features of recurrent events when types are not always determined.

Main Methods:

• A likelihood approach using joint models for multi-type recurrent events.
• Parameter estimation via a Monte-Carlo Expectation-Maximization (EM) algorithm.
• Application to a real-world asthma study with sub-typed exacerbations.

Main Results:

• The proposed method provides unbiased estimators for regression coefficients and variance-covariance parameters.
• Confidence interval coverage probabilities for regression coefficients are close to nominal levels.
• The method demonstrates robustness even when the frailty variable distribution is misspecified.

Conclusions:

• The developed statistical framework effectively analyzes incomplete multi-type recurrent event data.
• This approach offers reliable estimation of treatment effects and event associations in complex clinical scenarios.
• The method is validated through simulations and applied to a relevant asthma exacerbation dataset.