|
Abstract:
|
Various semiparametric methods have been proposed for analyzing recurrent event data. Among them, the additive rate model is particularly appealing when the interest is placed on the absolute difference in the occurrence rate of the recurrent events between different subgroups. However, the existing estimation procedure for additive rate model requires that the time-dependent covariates are observed continuously throughout the entire follow-up period, but, in practice, the covariates are usually intermittently observed due to cost and feasibility constraints. In this paper, we propose a novel estimating procedure for additive rate model with intermittently observed time-dependent covariates. The proposed estimating function is constructed by applying the kernel smoothing method to the mean covariate process. Simulation studies show that the proposed estimators are virtually unbiased. Moreover, they perform better than the conventional last-covariate-value-carried-forward approach which usually leads to biased estimation. The proposed method is also applied to a study which assessed the effect of group A streptococcus on pharyngitis.
|
