Cumulative risk regression in case-cohort studies using pseudo-observations.

Area of Science:

• Epidemiology
• Biostatistics
• Survival Analysis

Background:

• Case-cohort studies are efficient for risk factor ascertainment, particularly with competing risks.
• Traditional Cox regression for case-cohort data estimates rate ratios, not cumulative risks, which is a limitation in competing risk scenarios.
• Existing methods do not directly assess the association between covariates and cumulative risks in case-cohort studies with competing risks.

Purpose of the Study:

• To develop and evaluate a novel regression analysis for cause-specific cumulative risks in case-cohort studies.
• To extend the methodology to analyze absolute mortality risks directly from case-cohort survival data.
• To address limitations of existing methods in handling competing risks within case-cohort designs.

Main Methods:

• Utilizing pseudo-observations for regression analysis of cumulative risks.
• Adjusting for case-cohort sampling using inverse probability weighting within a generalized estimating equation framework.
• Developing theoretical large-sample properties and conducting simulation studies for small-sample evaluation.

Main Results:

• The proposed method accurately estimates cause-specific cumulative risks in case-cohort studies, even with competing risks.
• The methodology also enables direct analysis of absolute mortality risks from case-cohort data.
• Simulation studies confirm the validity and efficiency of the developed estimator.

Conclusions:

• The novel pseudo-observation approach provides a robust method for analyzing cumulative risks in case-cohort studies, particularly beneficial for competing risks.
• This method overcomes limitations of traditional rate ratio estimations in such designs.
• The approach was successfully applied to investigate risk factors for atrial fibrillation, demonstrating its practical utility.