Risk calculator to predict 30-day mortality in left-sided infective endocarditis. The EURO-ENDO score

Abstract

Abstract Background/Introduction Infective endocarditis (IE) is associated with high in-hospital mortality, despite improvements in therapeutic strategies. Nonetheless, there is no prospective risk model to estimate IE mortality. Purpose We sought to develop and validate a calculator to predict 30-day mortality risk regarding to perform surgery or medical treatment alone in left-sided IE. Methods This is a prospective, multicenter registry that included patients between January 2016 and March 2018 with a diagnosis of IE based on ESC 2015 diagnostic criteria. Patients with possible or definite left-sided IE were included in the analyses. Clinical, biological, microbiological and imaging data were collected. The primary end point was 30-day mortality in patients with left-sided IE. The risk calculator was based on multivariable Cox regression models. The accuracy of the logistic regression models was assessed by discrimination and calibration using C-statistic and Hosmer-Lemeshow test. Results Among 3116 patients included, 2171 patients presented left-sided IE and 257 patients (11.8%) died during the first 30 days of IE diagnosis. After multivariable Logistic regression analysis, eleven variables were associated with 30-days mortality and were included in the calculator: previous cardiac surgery, previous stroke/TIA, creatinine >2 mg/dL, S. aureus infection, embolic events on admission, heart failure or cardiogenic shock, vegetation size >14 mm, presence of abscess, severe regurgitation, double left-sided IE and no left valve surgery. There was an excellent correlation between the predicted 30-days mortality in both models with or without performing left valve surgery (area under the receiver operator curve: 0.798 and 0.758, respectively). Moreover, calibration by Hosmer-Lemeshow were 0.085 and 0.09, respectively). Conclusion(s) Our risk score in patients with left-sided IE provides an accurate individualized estimation of 30-day mortality according to perform or not perform left-valve surgery. It allows medical professionals to determine whether submitting patients to surgery or not, and thus improve their prognosis. Funding Acknowledgement Type of funding sources: None.