Flexible competing risks regression modeling and goodness-of-fit

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Competing risks models are crucial for analyzing time-to-event data with multiple outcomes.
• Traditional methods involve modeling cause-specific hazards, which can be complex.
• The Fine-Gray model offers direct estimation of cumulative incidence but assumes proportional subdistribution hazards.

Purpose of the Study:

• To present a flexible regression modeling approach for estimating covariate effects on cumulative incidence in competing risks.
• To introduce a novel, user-friendly goodness-of-fit test for the proportional subdistribution hazards assumption.
• To demonstrate the application of these methods using real-world bone marrow transplant data.

Main Methods:

• Developed a class of regression models encompassing the Fine-Gray model as a special case.
• Incorporated the ability to model non-proportional hazards directly.
• Proposed a constructive goodness-of-fit test to identify deviations from proportionality.

Main Results:

• The proposed flexible models are easy to fit and accommodate non-proportional hazards.
• The goodness-of-fit test effectively pinpoints the presence and location of non-proportionality.
• Analysis of bone marrow transplant data illustrates the practical utility of the methods.

Conclusions:

• The flexible regression models provide a powerful and adaptable tool for analyzing competing risks data.
• The new goodness-of-fit test enhances the reliability of analyses by assessing key assumptions.
• These advancements facilitate more accurate interpretation of covariate effects in complex survival scenarios.