Introduction Following acutely decompensated cirrhosis and associated organ dysfunction, despite recovery, >30% patients are readmitted to hospital and have increased overall mortality. What clinical factors predict likelihood of readmission and whether certain precipitants predispose to re-admission with the same events remain unanswered questions. This study followed patients with acute cirrhosis decompensation admitted to Royal Free London sequentially over 42 months to address these questions.

Method 238 patients admitted to RFH following acutely decompensated cirrhosis (ascites, bacterial infection, haemorrhage, jaundice, alcoholic hepatitis, and/or hepatic encephalopathy (HE)) were followed until death, transplantation or discharge. In discharged patients, observational data was collated to capture readmissions, precipitating events and outcomes. Simple logistic and Cox proportional hazards regression analyses were applied to the data set.

Results 238 patients were admitted with acute decompensation; 36 patients died and 9 were transplanted during their index admission. 195 patients were followed post discharge; 84 (44%) were re-admitted (median time to first re-admission 90 days) and in 38 patients, a second re-admission within 6 months. 11% patients had >3 re-admissions. Index presentation with HE in 22% patients predicted re-admission with HE with odds ratio (OR) 12.8 (2.4–69, P = 0.003). The predictive utility of HE in determining re-admission with HE was 0.83 (p < 0.05) by ROC analysis. By comparison, index presentation with GI haemorrhage (19%) had an OR 7.4; (1.7–32, P = 0.007) for re-admission with further haemorrhage. Of interest, patients discharged with recovery from alcoholic hepatitis (31%) re-presented with new infection [OR 2.78 (1–7.8, P = 0.05)]. 26% of patients had acute kidney injury (AKI) at index presentation which did not predict re-presentation with AKI [OR: 2.21 (0.5–9.66; P = 0.3) and this was also the case for re-admission with infection [OR 1.14]. In this cohort, MELD-Na on index admission was notpredictive of re-admission to hospital.

Conclusion This data re-affirms the high mortality post acute decompensation of cirrhosis whilst also demonstrating considerable early hospital readmission. A novel relationship between index precipitating event and recurrence of this event on re-admission is shown. This suggests certain complications of cirrhosis may be genetically or epigenetically determined. Strategies for primary and secondary prophylaxis should target these predisposing factors and points to important area of future research.

Disclosure of interest None Declared.