Predictors of 90-day mortality in patients with severe alcoholic hepatitis: Experience with 183 patients at a tertiary care center from India.

BACKGROUND: Severe alcoholic hepatitis (AH) is not an uncommon indication for hospital admission in India. However, there is limited data from India on predictors of mortality in patients of severe AH. We analyzed the data on patients with severe AH admitted to our institute and compared various parameters and severity scores in predicting 90-day mortality.

METHODS: In this prospective study, we analyzed patients with severe AH (defined as discriminant function ≥ 32) admitted from January 2015 to February 2017 to our institute. All patients were administered standard treatment according to various guidelines, and their 90-day mortality was determined. Various hematologic, biochemical factors, and severity scores were compared between survivors and patients who died.

RESULTS: A total of 183 patients (98% males, median age 41 years [range 20-70 years]) were included in our study. The median model for end-stage liver disease (MELD) was 26 (15-40). Ascites were present in 83% and hepatic encephalopathy in 38%. Only 21 (12%) could be offered steroid therapy, due to contraindications in the remaining. By 90 days, only 103 (56%) patients survived while 80 (44%) died. All patients died due to progressive liver failure and its complications. On multivariate analysis, presence of ascites, hepatic encephalopathy, high bilirubin, low albumin, high creatinine, high INR, and low potassium independently predicted 90-day mortality. All the scores performed significantly in predicting 90-day mortality with no statistically significant difference between them. MELD score had a maximum area under the curve 0.76 for 90-day mortality. A combination of Child class and presence of acute kidney injury (creatinine ≥ 1.35) was good in predicting 90-day mortality.

CONCLUSION: Our patients had severe AH characterized by a median MELD score of 26 and had a 90-day mortality of 44%. Most patients were not eligible to receive corticosteroids. Presence of Child C status and high serum creatinine value (≥ 1.35 mg/dL) accurately predicted mortality. Newer treatment options need to be explored for these patients.