Commentary: scoring models for short‐term mortality in alcoholic hepatitis

Abstract

Papastergiou et al. studied patients with biopsy-proven alcoholic hepatitis (AH) with regard to accuracy of scoring systems used to predict outcome and evaluate response to treatment.1 AH is distinguished by a plethora of scoring systems. Some, such as the discriminant function (DF),2 Glasgow alcoholic hepatitis score (GAHS)3 and the ABIC score,4 were developed specifically for use in AH. Others such as MELD have been co-opted from other areas of liver medicine.5 It has been clear for some time that all of these scores are fairly accurate in predicting short-term (28-day) and medium-term (90-day) mortality in AH.6-8 Papastergious' study is a helpful external validation of these scores, and is of particular interest as the cohort all had biopsy-proven AH. Many of the suggested models have been based on groups of patients where only a clinical diagnosis was made, thereby risking the inclusion of patients with other forms of hepatic decompensation. Although this study has validated the negative predictive value of these scoring systems many unanswered questions remain in the area of AH. We await with interest the United Kingdom randomised controlled trial evaluating therapies in AH.9 At present most groups use steroids in severe AH, and stop them if there is no biochemical response after 1 week. However, evidence for efficacy of steroids is disputed by standard meta-analysis and the duration of treatment remains unclear in those who respond. Although this current study does help our understanding of this serious disease, the key answers to its management will only come from randomised controlled trials. Declaration of personal and funding interests: None.