Predictive model of in-hospital mortality in left-sided infective endocarditis

Abstract Introduction and objectives Infective endocarditis (IE) is a complex disease with high in-hospital mortality. Prognostic assessment is essential to select the most appropriate therapeutic approach; however, international IE guidelines do not provide objective assessment of the individual risk in each patient. We aimed to design a predictive model of in-hospital mortality in left-sided IE combining the prognostic variables proposed by the European guidelines. Methods Two prospective cohorts of consecutive patients with left-sided IE were used. Cohort 1 (n = 1002) was randomized in a 2:1 ratio to obtain 2 samples: an adjustment sample to derive the model (n = 688), and a validation sample for internal validation (n = 314). Cohort 2 (n = 133) was used for external validation. Results The model included age, prosthetic valve IE, comorbidities, heart failure, renal failure, septic shock, Staphylococcus aureus, fungi, periannular complications, ventricular dysfunction, and vegetations as independent predictors of in-hospital mortality. The model showed good discrimination (area under the ROC curve = 0.855; 95%CI, 0.825-0.885) and calibration (P value in Hosmer-Lemeshow test = 0.409), which were ratified in the internal (area under the ROC curve = 0.823; 95%CI, 0.774-0.873) and external validations (area under the ROC curve = 0.753; 95%CI, 0.659-0.847). For the internal validation sample (observed mortality: 29.9%) the model predicted an in-hospital mortality of 30.7% (95%CI, 27.7-33.7), and for the external validation cohort (observed mortality: 27.1%) the value was 26.4% (95%CI, 22.2-30.5). Conclusions A predictive model of in-hospital mortality in left-sided IE based on the prognostic variables proposed by the European Society of Cardiology IE guidelines has high discriminatory ability.