The Predictive Value of Sepsis Scores for In-Hospital Mortality in Patients with Left-Sided Infective Endocarditis

Background: As infective endocarditis has particular characteristics compared to other infectious diseases, it is not clear if sepsis scores are reported with good accuracy in these patients. The aim of this study is to evaluate the accuracy of the qSOFA and SOFA scores to predict mortality in patients with infective endocarditis. Methods: Between January 2010 and June 2019, 867 patients with suspected left-sided endocarditis were evaluated; 517 were included with left-sided infective endocarditis defined as "possible" or "definite" endocarditis, according to the Modified Duke Criteria. ROC curves were constructed to assess the accuracy of qSOFA and SOFA sepsis scores for the prediction of in-hospital mortality. Results: The median age was 57 years, 65% were male, 435 (84%) had pre-existing heart valve disease, and the overall mortality was 28%. The most frequent etiologies were Streptococcus spp. (36%), Enterococcus spp. (10%), and Staphylococcus aureus (9%). The sepsis scores from the ROC curves used to predict in-hospital mortality were qSOFA 0.601 (CI95% 0.522-0.681) and SOFA score 0.679 (CI95% 0.602-0.756). A sub-group analysis in patients with and without pre-existing valve disease for SOFA >= 2 showed ROC curves of 0.627 (CI95% 0.563-0.690) and 0.775 (CI95% 0.594-0.956), respectively. Conclusions: qSOFA and SOFA scores were associated with increased in-hospital mortality in patients with infective endocarditis. However, as accuracy was relatively lower compared to other sites of bacterial infections, we believe that this score may have lower accuracy when predicting the prognosis of patients with IE, because, in this disease, the patient's death may be more frequently linked to valvular and cardiac dysfunction, as well as embolic events, and less frequently directly associated with sepsis.