Análise de sobrevivência com eventos recorrentes: um estudo de casos de dengue

Abstract

This paper presents an introduction and application of survival analysis, aiming to study dengue cases primarily through modeling the time until the occurrence (and possible recurrence) of infections. It begins with a study of non-parametric descriptions and adopts the semi-parametric Cox model as a basis for estimation, discussing its proportionality and risk assumptions. Subsequently, extensions of the Cox models for recurrent events are presented, including Andersen-Gill, Prentice-Williams-Peterson, and Wei-Lin-Weissfeld models. Following this, two examples are presented: (i) a clinical database of recurrent renal failures and (ii) data on rehospitalization after colorectal cancer surgery, in which the Andersen-Gill model is fitted.