Modelo predictivo de mediastinitis postoperatoria en cirugía cardiovascular

Abstract

IntroductionThe most common incision for cardiovascular surgery is median longitudinal sternotomy. Postoperative mediastinitis is deep infection of this. On being a serious problem, predicting it can help to provide guidelines for prevention and to reduce its incidence. ObjectivesTo design a predictive model for postoperative mediastinitis with risk factors. MethodsA retrospective cases-controls analytical study conducted, that enrolled 100% of cases with postoperative mediastinitis and 4 controls for each case. The study included 20 years of work in Cardiocentro Ernesto Guevara. An adjusted logistic regression was performed to obtain the model. ResultsThe incidence of postoperative mediastinitis was 0.98%, with numerous risk factor predictors, with the most important being: chronic obstructive pulmonary disease, postoperative hyperglycaemia, mechanical artificial ventilation for more than 24hours, transfusion of more than 2 units of blood products, pneumothorax, and endovascular sepsis. These were the significant factors in the logistic regression model. Validation was made by data splitting method. ConclusionsA predictive model with discriminative power was obtained with 6 predictors.