Introduction Severe alcoholic hepatitis (AH) is a serious complication of alcohol misuse and has a high short term mortality of up to 40%. Longer term outcome is less well characterised and a single retrospective study has demonstrated that alcohol recidivism is associated with poorer survival. We aimed to prospectively evaluate factors important in long term (>1 year) survival in patients with severe AH.

Method Consecutive patients admitted to our institution between October 2007 and September 2013 with severe AH were recruited. Severe AH was defined as recent heavy alcohol use (>60 g or >80 g ethanol daily for women and men respectively), recent onset jaundice with a bilirubin >80 umol/L and a discriminant function (DF) >32. Patients were followed up for at least 1 year from diagnosis and their alcohol use determined at that time. Patients were treated with prednisolone 40 mg daily for 4 weeks unless they were also recruited to the STOPAH clinical trial in which case treatment allocation remains blinded. Differences in 1 year survival were determined individually for patient characteristics, baseline laboratory parameters and alcohol use. Multivariate analysis with logistic regression was then performed on these terms to determine independent predictors of survival.

Results 87 subjects (mean age 48, 72% male, mean DF 53) were followed up for a median time of 410 days. Overall 1 year survival was 55%. At presentation age, urea, INR and prothrombin time were significantly greater (p < 0.05) in non-survivors than survivors at 1 year but on multivariate analysis only urea and INR remained independent predictors of death. Area under receiver operating characteristic (AUROC) analysis found that DF, MELD and GAHS were poor predictors of 1 year survival (AUROC = 0.41, 0.36 and 0.36 respectively). 71% survived to hospital discharge of which 63% relapsed to heavy drinking within 1 year. In the group of patients who were discharged from hospital, abstinence from alcohol was associated with long term survival (p = 0.04) and multivariate analysis identified abstinence from alcohol as the only independent predictor of >1 year survival (p = 0.03).

Conclusion This is the largest reported prospective cohort of long term outcome in patients with severe AH. Urea and INR at presentation (but not existing scoring systems) are important predictors of 1 year survival. Importantly, only alcohol abstinence predicts long term survival in patients who are successfully discharged from hospital. Therefore ongoing research should focus on maintenance of abstinence from alcohol as the key factor that will improve long term outcome in survivors of severe AH.

Disclosure of interest None Declared.