Severe liver failure rather than cirrhosis is associated with mortality in patients with infectious endocarditis: a...

Data on infectious endocarditis (IE) in patients with liver cirrhosis (LC) are sparse. We aimed to describe the characteristics and predictors of mortality from IE in patients with LC.

Overall, 101 patients with LC and 101 controls with IE matched for sex, age, date of IE, and diabetes were retrospectively selected in 23 liver units between 2000 and 2013.

Mean age was 60.8±10.5 and 60.6±11.5 years in LC and controls, respectively. Causes of cirrhosis (Child-Pugh A/B/C: 10.4%/41.7%/47.9%, MELD score: 17±7.8) were excess alcohol intake (79.6%), viral hepatitis (17.3%), and metabolic syndrome (14.3%). Previous history of cardiopathy was found in 24.8% of LC (prosthetic valve 8.9%) and 37.6% of controls (P=0.07). The most frequent bacteria involved were gram-positive cocci. LC had significantly fewer aminoglycosides (P=0.0007), rifamycin (P=0.03), and valve surgery (P=0.02) than controls. The proportion of patients who died following cardiac surgery was similar between the two groups (9.7% for LC vs. 8.7% for controls, P=1). In-hospital mortality for Child-Pugh C patients was significantly higher than controls (61.4 vs. 23%, P<0.001), but not for Child-Pugh A (33.3%) or B patients (25.0%). A Child-Pugh score of above C10 was the best predictor of in-hospital mortality. In LC, Child-Pugh score (odds ratio=1.5; 95% confidence interval: 1.2-2.0; P=0.002) and history of decompensation (odds ratio=3.1; 95% confidence interval: 1.1-9.0; P=0.003) were independent predictive factors for in-hospital mortality.

Severe liver failure but not cirrhosis is the strongest predictive factor of mortality related to IE in LC. Use of aminosides and rifamycin should be reassessed in LC, and cardiac surgery should be considered for selected patients.