ABC | Volume 114, Nº1, January 2019

Original Article Marques et al. In-hospital mortality in infective endocarditis Arq Bras Cardiol. 2020; 114(1):1-8 by hospital discharge codes. The patients were followed until discharge or death (including hospitalization at the surgical center). Demographic and clinical characteristics, type of endocarditis (native valve, prosthetic valve or device‑associated), echocardiographic and microbiological findings, as well as surgical procedure and hospitalization outcomes were retrieved. The sample was characterized using basic descriptive statistic measures. Patients that died during hospital stay were compared with those that survived regarding their demographic and clinical features, microbiological and echocardiographic findings and hospitalization outcomes. The primary outcome was all-cause in-hospital mortality. The other adverse outcomes of interest were heart failure (defined as the presence of typical symptoms and signs caused by a structural and/or functional cardiac abnormality, resulting in a reduced cardiac output and/or elevated intracardiac pressures), septic shock (characterized by the presence of Systemic Inflammatory Response Syndrome to an infectious process, with sepsis-induced organ dysfunction or tissue hypoperfusion and persistently arterial hypotension, despite the administration of intravenous fluids), evidence of locally uncontrolled infection or periannular complication (valve destruction or perforation, increasing vegetation size, abscess formation, pseudoaneurysm, valve aneurysm and intracardiac fistula) and embolic events (ischemic stroke, hemorrhagic stroke, mycotic aneurism, myelitis/meningitis, peripheral ischemia and splenic, pulmonary or hepatic infarction or abscesses, diagnosed through computed tomography and/ or magnetic resonance imaging, performed according to the clinical suspicion of embolism). Healthcare-associated IE was defined as IE manifesting more than 48 hours after hospital admission or IE acquired in association with an invasive procedure performed in the 6 months before diagnosis during hospital stay and/or manipulation in a hospital setting. Valve regurgitation detected at echocardiography included both significant valve regurgitation in native valve IE cases and significant intra and paraprosthetic leaks in prosthetic IE cases. Statistical analysis Categorical variables were presented as frequencies and percentages and were compared using the chi-square test. Continuous variables were expressed as means and standard deviations (SD) and were compared using the independent‑samples t-test, after normal distribution was checked using the Kolmogorov-Smirnov test or skewness and kurtosis. Continuous variables with skewed distributions were presented as medians and interquartile ranges (IQR) and a non-parametric method (MannWhitney U test) was employed. In order to identify predictors of in-hospital mortality, variables with a p value < 0.1 in the univariate analysis were included in a logistic regression using an enter stepwise method. Two models were performed; one model included all Streptococcal Species and the other included the microorganism Streptococcus gallolyticus , since they are variables that are not independent of each other and both had a p value < 0.1 in the univariate analysis. The model predictive performance was tested by assessing its discrimination and its calibration. Discrimination was measured with the area under receiver operating characteristic curve (AUROC) and calibration was measured by using pseudo-R 2 (Nagelkerke R 2 ). The final model defined was that with the highest predictive performance according to the AUROC and pseudo-R 2 . All reported p values were two-tailed, with a p value < 0.05 indicating statistical significance. The statistical analyses were performed using IBM SPSS Statistics software, version 22. Results Population characteristics Between January 2006 and December 2015, 134 cases of infective endocarditis were hospitalized in our center: 101 cases had definite IE and the remaining corresponded to possible IE cases, according to the modified Duke criteria. About 73% of these patients were males, the mean age was 61 ± 16 years. The main clinical characteristics, namely the comorbidities, clinical presentation, microbiology and clinical outcomes of IE cases are summarized in Table 1. About half of the patients had previous arterial hypertension and valvular heart disease and 13.4% were intravenous drug users. Regarding the 13.4% of patients with Human Immunodeficiency Virus (HIV) infection, only 44% of these patients were on antiretroviral therapy at the time of the IE diagnosis; CD4 cells counts were obtained in 13 patients, with a median level of 130 ± 391 CD4 cells. About 12% of the IE cases corresponded to patients with chronic renal disease, and 31% of these were on hemodialysis. The majority of the cases were related to native valves (71.6%), while the remaining were associated with prosthetic heart valves (25.4%) and device-related IE (3%). Healthcare-associated infective endocarditis cases occurred in 22.4% of the patients. About 22% of the cases had negative blood cultures. Antibiotic administration previous to blood culture collection was described in 72% of these cases. In 1 case, the IE diagnosis was made at the autopsy and blood samples were not obtained. The most commonly isolated microorganisms were Staphylococcus aureus (22.4%) and Viridans Group Streptococci (12.7%). A transthoracic echocardiography was performed in all patients, while a transesophageal study was carried out in 118 (88%) patients, with a mean time between admission and test performance of 10 ± 9.5 days (range 0-54 days). The main echocardiographic finding observed was the presence of vegetations (79.1%). Valve regurgitation was observed in 69 cases, with 4 patients having reduced left ventricle ejection fraction (LVEF). Only 11 cases reported the LVEF and the median LVEF was 61% (IQR 18%). Systolic pulmonary artery pressure (SPAP) was reported in 15 cases, with a mean SPAP value of 41 mmHg (SD 27 mmHg). 2