Hospital-Acquired Infection Is the Strongest Predictor of In-hospital Mortality in Patients With Alcoholic Hepatitis

Abstract

Introduction: Alcoholic hepatitis (AH) is a clinical syndrome associated with high mortality and utilization of resources. We performed this study to evaluate predictors of in-hospital mortality (IHM) among AH patients. Methods: A retrospective chart review was done on patients discharged from our hospital with a diagnosis of AH. Patients who met a well characterized definition of AH (serum bilirubin > 3 mg/dL, AST to ALT ratio of > 1.5, recent heavy alcohol use, and absence of other causes of liver disease) were included in the study. Chi-square and t-tests were used to compare categorical and continuous variables. Multivariate logistic regression was used to evaluate the impact of several variables on IHM. Results: 105 patients with AH (median age 46yrs, 56% male, 74% white, 57% with cirrhosis, MELD score 23) were included in the study. Complications present on admission included acute kidney injury (AKI) (44%), encephalopathy (29%), infection (24%), and gastrointestinal bleeding (17%). Complications that developed during the hospitalization included AKI (48%) with 8% requiring dialysis, infection (15%), encephalopathy (12%), and gastrointestinal bleeding (5%). 67 patients (64%) received treatment for AH including steroids (n=33), pentoxifylline (21), or both (n=13).15 patients (14%) died during hospitalization. Compared to those that survived hospitalization, patients with IHM had a higher incidence of AKI on admission (73 vs. 39%, P=0.013), in-hospital AKI (73 vs. 20%, P < 0.0001), in-hospital encephalopathy(40 vs. 8%, P=0.0005), in-hospital infection (47 vs. 10% P=0.0003), pentoxifylline use (60 vs. 28%, P=0.014), and in-hospital renal replacement therapy (33 vs. 3%, P < 0.0001). They were also more likely to meet SIRS criteria (73 vs. 46%, P=0.046) and have a higher admission MELD score (32 vs. 23, P=0.0003). The use of prophylactic antibiotics was similar among patients with IHM and those who survived hospitalization (33 vs. 29%, P=0.73). On multivariate logistic regression, hospital-acquired infection was the single most important predictor for IHM (OR 15, 2.3-99, P=0.005). Prophylactic antibiotic use (OR 1.04, 0.14-8.1, P=0.96) and admission MELD score (OR 1.07, 0.98-1.17, P=0.13) were not predictive of IHM. Conclusion: Patients with AH have high in-hospital mortality and hospital acquired infection appears to be the strongest predictor of IHM. Prophylactic antibiotic use does not affect IHM and so clinicians should be vigilant to reduce the risk of infection and treat appropriately if there is evidence of an infection.