The analysis of multivariate interval-censored survival data

Area of Science:

• Biostatistics
• Survival Analysis
• Clinical Trials

Background:

• Multivariate interval-censored data arise when event times are not precisely known but fall within intervals.
• Accurate analysis is crucial for understanding disease progression and treatment effects in clinical settings.
• Existing methods may struggle with correlated events and interval censoring.

Purpose of the Study:

• To develop a robust statistical method for analyzing multivariate interval-censored survival data.
• To assess the impact of covariates on correlated events in survival analysis.
• To provide a more accurate estimation of parameters and variances compared to existing approaches.

Main Methods:

• A marginal approach based on a discrete proportional hazards model for interval-censored data.
• Development of a robust covariance matrix estimator accounting for event correlations.
• Comparison with a midpoint imputation method via simulation studies.

Main Results:

• The proposed marginal approach yields less biased parameter estimates than midpoint imputation.
• Ignoring event correlation can lead to incorrect variance estimators, even with modest correlation.
• The method demonstrates improved performance in simulation studies.

Conclusions:

• The developed marginal approach is a valuable tool for analyzing multivariate interval-censored survival data.
• Accounting for correlation between events is essential for reliable statistical inference.
• This method enhances the analysis of complex event data in clinical trials.