Penalized spline smoothing for delay in Pulmonary Tuberculosis diagnosis

Abstract

In this work, Cox's proportional hazards model with penalized spline functions for time-varying covariate effects are presented in order to model the time between the onset of the first symptoms and the diagnosis of Pulmonary Tuberculosis (PTB) in Portugal. Using this approach revealed a better fitting of the model to the data than Cox's classical model. Results showed that an earlier diagnosis tends to be associated with the following: being male, no alcohol consumption, being diagnosed through active screening, having had a previous PTB treatment, being HIV positive, and being a smoker.