Analysis of factors predictive of mortality in alcoholic hepatitis and derivation and validation of the Glasgow alcoholic hepatitis score

Introduction: Alcoholic hepatitis is associated with a high short term mortality. We aimed to identify those factors associated with mortality and define a simple score which would predict outcome in our population. Methods: We identified 241 patients with alcoholic hepatitis. Clinical and laboratory data were recorded on the day of admission ( day 1) and on days 6 - 9. Stepwise logistic regression was used to identify variables related to outcome at 28 days and 84 days after admission. These variables were included in the Glasgow alcoholic hepatitis score (GAHS) and its ability to predict outcome assessed. The GAHS was validated in a separate dataset of 195 patients. Results: The GAHS was derived from five variables independently associated with outcome: age ( p = 0.001) and, from day 1 results, serum bilirubin ( p< 0.001), blood urea ( p = 0.019) and, from day 6 - 9 results, serum bilirubin ( p< 0.001), prothrombin time ( p = 0.002), and peripheral blood white blood cell count ( p = 0.001). The GAHS on day 1 had an overall accuracy of 81% when predicting 28 day outcome. In contrast, the modified discriminant function had an overall accuracy of 49%. Similar results were found using information at 6 - 9 days and when predicting 84 day outcome. The accuracy of the GAHS was confirmed by the validation study of 195 patients The GAHS was equally accurate irrespective of the use of the international normalised ratio or prothrombin time ratio, or if the diagnosis of alcoholic hepatitis was biopsy proven or on the basis of clinical assessment. Conclusions: Using variables associated with mortality we have derived and validated an accurate scoring system to assess outcome in alcoholic hepatitis. This score was able to identify patients at greatest risk of death throughout their admission.