Modeling county level breast cancer survival data using a covariate-adjusted frailty proportional hazards model

摘要

Understanding the factors that explain differences in survival times is animportant issue for establishing policies to improve national health systems.Motivated by breast cancer data arising from the Surveillance Epidemiology andEnd Results program, we propose a covariate-adjusted proportional hazardsfrailty model for the analysis of clustered right-censored data. Rather thanincorporating exchangeable frailties in the linear predictor of commonly-usedsurvival models, we allow the frailty distribution to flexibly change with bothcontinuous and categorical cluster-level covariates and model them using adependent Bayesian nonparametric model. The resulting process is flexible andeasy to fit using an existing R package. The application of the model to ourmotivating example showed that, contrary to intuition, those diagnosed during aperiod of time in the 1990s in more rural and less affluent Iowan countiessurvived breast cancer better. Additional analyses showed the opposite trendfor earlier time windows. We conjecture that this anomaly has to be due toincreased hormone replacement therapy treatments prescribed to more urban andaffluent subpopulations.