Estimating the correlation between semi-competing risk survival endpoints

Area of Science:

• Biostatistics
• Survival Analysis
• Medical Data Science

Background:

• Semi-competing risk data, where a terminal event can censor a non-terminal event, present unique analytical challenges.
• Estimating correlations between bivariate time-to-event endpoints is complicated by the inherent censoring mechanisms.
• Traditional correlation estimation methods are inadequate for semi-competing risk scenarios.

Purpose of the Study:

• To develop and evaluate a copula-based methodology for assessing dependence structures in bivariate semi-competing risk data.
• To accurately estimate the correlation between survival endpoints when censoring is present.
• To provide a robust framework for analyzing complex time-to-event data in clinical and epidemiological studies.

Main Methods:

• Utilized a copula-based approach to model the dependence between two time-to-event endpoints.
• Employed various copula functions to capture different association structures.
• Transformed the estimated copula association parameter into Spearman's rank correlation coefficient for interpretability.

Main Results:

• The proposed copula-based methods effectively estimate the correlation between bivariate time-to-event endpoints in semi-competing risk data.
• Simulation studies demonstrated the robustness of the estimation, even with potential misspecification of copula functions and survival distributions.
• The methodology was successfully applied to two real-life datasets, showcasing its practical utility.

Conclusions:

• Copula-based modeling offers a powerful and flexible approach for analyzing dependence in semi-competing risk survival data.
• The developed methods provide reliable estimates of correlation, enhancing the understanding of complex relationships between clinical endpoints.
• This framework is valuable for researchers dealing with censored bivariate time-to-event data in various scientific fields.