Using frailties in the accelerated failure time model

Abstract

The accelerated failure time (AFT) model is an important alternative to the Cox proportional hazards model (PHM) in survival analysis. For multivariate failure time data we propose to use frailties to explicitly account for possible correlations (and heterogeneity) among failure times. An EM-like algorithm analogous to that in the frailty model for the Cox model is adapted. Through simulation it is shown that its performance compares favorably with that of the marginal independence approach. For illustration we reanalyze a real data set.