Prediction of hospital mortality in patients with left-sided infective endocarditis using a score in the first hours of admission

Abstract

Infective endocarditis is an infection of the endocardium, associated with high morbidity and mortality, particularly when the left heart chambers are involved. Timely diagnosis and appropriate intervention can attenuate these outcomes. The objective of this study was to create and validate the EndoPredict-Px score for early in-hospital mortality prediction in patients with left-sided infective endocarditis. This analysis focused on patients admitted to the emergency department with left-sided infective endocarditis defined by the Duke-ISCVID criteria, from January 2011 to January 2020. The main outcome was in-hospital mortality. Among 530 patients with left-sided infective endocarditis, 160 (30.2%) died during hospitalization. The score included age ≥ 60 years, absence of fever, NYHA functional class III or IV heart failure, suspected embolism, hemoglobin level ≤ 10 g/dL, leukocyte level ≥ 12 × 109/L, platelet level ≥ 150 × 109/L, and creatinine level ≥ 1.3 mg/dL. Patients were categorized into three mortality risk groups. The model displayed an AUROC of 0.76 (95% CI 0.71–0.80) and 0.74 in the bootstrap validation. Both the derivation and validation models showed accurate calibration. In conclusion, the EndoPredict-Px score accurately and early predicted in-hospital mortality in patients with left-sided infective endocarditis.