Title: Estimation of the marginal mean of recurrent events
Authors: Giuliana Cortese - University of Padua (Italy) [presenting]
Thomas Scheike - University of Copenhagen (Denmark)
Abstract: In survival analysis, currently, there is increasing interest in global summary measures over the follow-up period under study. When only a single event is of interest, examples of these measures are the mean lifetime and the residual mean lifetime. However, recurrent events such as cancer relapses or cardiovascular episodes can often be encountered in clinical and epidemiological studies on individuals who may potentially experience a terminal event such as death. With recurrent events data, a global summary measure of great interest is the marginal mean of the cumulative number of recurrent events experienced prior to the terminal event. Statistical efficiency of the IPCW nonparametric estimator for this mean is investigated and a novel efficient augmented estimator, based on dynamic predictions, is presented. In settings with different sources of heterogeneity, the proposed estimator is shown to improve efficiency greatly. In addition, regression models for the mean number of recurrent events can be employed to investigate the role of biomarkers on disease progression. A data example on chronic intestinal failures is discussed to highlight the practical use of these methods.