Introduction Although short-term outcome in severe alcoholic hepatitis (SAH) is well described, its long-term course is largely unknown. Our aim was to assess long-term outcome in SAH.

Methods Cohort study of patients with SAH (Discriminant Function ≥32) admitted to our institute, identified retrospectively 2007–2009 then prospectively until August 2011. Clinical and laboratory parameters were recorded at accession and subsequent follow-up. Kaplan-Meier (KM) and Cox proportional hazards analyses were performed. Data are presented as mean ±SD or median (IQR).

Results 109 consecutive patients with SAH were included (63.3% men, age 49.6 ± 9.4yrs) with median follow-up of 40.7 mths (95% CI 37.2–44.3). At accession median DF was 58 (34), MELD 23 (6) and Glasgow Alcoholic Hepatitis Score 8 (2). 55.0% drank spirits, 86.2% had established cirrhosis and 65.1% received corticosteroid and/or pentoxifylline therapy. Prevalence of hepatic encephalopathy (HE), infection and hepatorenal syndrome (HRS) were 38.5%, 47.7% and 18.3% respectively. Median survival was 22.4 mths (95% CI 11.1–33.7), overall mortality being 57.8% (n = 63). All except two deaths were liver related and 33.3% occurred during the index admission.

In univariate analysis, AST, urea, creatinine, white cell count, and the presence of HE, HRS and infection were statistically significantly associated with mortality (p < 0.10 for all), though only HRS was an independent predictor in multivariate analysis (HR 3.842, 95% CI 2.018–7.312, p < 0.0001). However, all baseline factors were associated with short-term mortality and none predicted death beyond 3 mths.

Of those surviving index hospitalisation (n = 87, 86 with available data), 37 (43.0%) were abstinent at last follow-up. Recidivism occurred in the remaining 49 (57.0%), of whom 33 continued to drink and 16 relapsed after initial abstinence. Abstainers were significantly less likely to require hospital readmission than those with any recidivism (median readmissions/patient 1 vs. 5, p = 0.001). In a further univariate analysis, inpatient paracetamol use and abstinence status were associated with mortality after the index hospitalisation (p = 0.032 and 0.010 respectively). Only abstinence remained an independent predictor in multivariate analysis (HR 0.402 (95% CI 0.183–0.883), p = 0.023). KM analysis showed 3 year survival to be significantly higher in abstainers (75.3%) compared to relapsed and continued drinkers (40.2% and 21.0% respectively, p = 0.005).

Conclusion Recidivism is common after SAH (~65%) and is the main determinant of the high long-term mortality (~60%), which appears unrelated to the severity of the index episode. Novel strategies to improve abstinence following SAH are urgently needed, especially in view of the increasing numbers of deaths due to alcohol-related liver disease in the UK.

Disclosure of Interest None Declared