Statistics Seminar - 9/14/21

Abstract: A population-averaged additive subdistribution hazard model is proposed to assess the marginal effects of covariates on the cumulative incidence function to analyze clustered correlated failure time data subject to competing risks. This approach extends the population-averaged additive hazard model by accommodating potentially dependent censoring due to competing events other than the event of interest. Assuming an independent working correlation structure, a generalized estimating equations (GEE) approach is considered to estimate the regression coefficients and a sandwich covariance estimator is proposed to quantify the variability of the regression coefficient estimators. The sandwich covariance estimator accounts for both the correlations between failure times as well as the correlations between censoring times, and is robust to misspecification of the unknown dependency structure within clusters. We further develop goodness-of-fit tests to assess the adequacy of the additive structure of the subdistribution hazard, for each covariate as well as for the overall model. Simulation studies are carried out to investigate the performance of the proposed methods in finite samples, and we illustrate the proposed methods by analyzing the Strategies to Reduce Injuries and Develop Confidence in Elders (STRIDE) study.