Abstract
BACKGROUND/AIMS: In alcohol-associated hepatitis (AH), the Lille score is used to assess futility of steroids. However, the ability of the Lille score to predict 30-day survival in AH is not well-defined. Our aim is to compare the utility of the Lille score in predicting 30-day survival in those with AH treated with steroids. METHODS: Retrospective chart review of 882 patients hospitalized with AH from January 1st, 2012 through December 30th, 2019 was performed. Of these, 201 patients with severe AH met the threshold to receive steroids. Those with data to calculate Lille score < 0.45 on day 4 (n = 29) or 7 (n = 89) who continued steroids were compared to 83 patients with Lille scores ≥ 0.45 on day 4 (n = 18) or 7 (n = 65) who stopped steroids. The primary outcome was 30-day survival. For comparison, a contemporaneous matched control group was also analyzed of 110 patients who were hospitalized with severe AH, but did not receive steroids. RESULTS: In patients with Lille score < 0.45, survival was higher at 30-day when compared to those with Lille score ≥ 0.45 (94.9% vs. 80.72%; p = 0.002). The sensitivity, specificity, positive predictive value and negative predictive value of Lille score (< 0.45) to predict 30-day survival was 95%, 19%, 63%, and 73%, respectively. CONCLUSIONS: In severe AH, those with Lille score < 0.45 at day 4 or 7 have improved 30-day survival compared to those with Lille score ≥ 0.45. In those receiving steroids, Lille score has excellent sensitivity to predict 30-day survival but poor specificity.