Prediction of in-hospital Mortality and Long-term Survival in Patients Operated for Infective Endocarditis

Background: Decision making in infective endocarditis (IE) patients with respect to surgical intervention is often complex. Martínez-Sellés et al., (International Journal of Cardiology, 2014) developed a risk model to predict in-hospital mortality in these patients based on 7 factors: prosthetic valve IE, age, large intracardiac destruction, S.aureus infection, urgent surgery, sex and euroSCORE. We assessed predictors of in-hospital mortality and 7-year survival in a large single center cohort and tested the validity of the Martínez-Sellés model compared with euroSCORE.

Methods: we analyzed hospital and follow-up data from all patients operated for left sided IE in our center. We performed multivariable logistic regression analysis and Cox-proportional hazard analysis to assess the risk factors of in-hospital mortality and 7-year survival, respectively. To test the validity of the Martínez-Sellés model (PALSUSE) compared with additive euroSCORE in terms of prediction of in-hospital mortality we performed calibration (Hosmer-Lemeshow-test) and discrimination (Receiver-Operating-Characteristic-test; ROC-test) analyses.

Results: A total of 308 patients (age 62.0 ± 13.9 years) who underwent surgery for IE were included in the analysis. In-hospital mortality was 22.6%. Multivariable logistic regression analysis and Cox-proportional hazard analysis showed that advanced age, preoperative hemodialysis, preoperative septic shock, S.aureus infection, liver cirrhosis and longer cardiopulmonary bypass time were independent predictors of in-hospital mortality and 7-year survival. The two models performed moderately and were comparable for the prediction of an individual patient's risk of mortality in the whole study population (the area under the curve for the ROC was 0.723 for PALSUSE and 0.719 for euroSCORE). The overall correct classification was the same for both scores (77.1%). PALSUSE calibrated substantially better than euroSCORE according to Hosmer-Lemeshow-test (Chi square: 2.82, versus 11.25; p: 0.727 versus 0.188, respectively). Both scores slightly underestimated mortality; O/E ratio was 0.96 for both scores.

Conclusions: In-hospital mortality and long-term survival of patients operated for IE were influenced mainly by advanced age, preoperative hemodialysis, preoperative septic shock, S.aureus infection, and liver cirrhosis. The suggested IE score (PALSUSE) had a moderate value in terms of prediction of in-hospital mortality in our cohort.