Combination therapy for bloodstream infections with carbapenemase-producing Enterobacteriaceae - Authors' reply.

Abstract

We thank Bernd Salzberger and Gerd Fätkenheuer for their interest in our study.1 They were concerned about the use of Charlson, Pitt, and INCREMENT-CPE scores, as well as the individual variables contained in them in the analyses. Our objective was to provide the best possible estimation of the association of the main exposure ( treatment ) with the outcome considering the effect of confounders, in accordance with the study hypothesis; we did not intend to provide a model to predict outcome ( mortality ). Therefore, we did several multivariate analyses , including models with the individual variables, models with the scores, and models with both, after checking for collinearity. The best models were selected and are shown in tables 2 and 3 of our paper .1 As a result, when the scores were considered, the individual variables were not relevant and the effect of treatment on mortality was adjusted by the scores.