MELD score is a better prognostic model than Child-Turcotte-Pugh score or Discriminant Function score in patients with alcoholic hepatitis

https://doi.org/10.1016/j.jhep.2004.12.022Get rights and content

Background/Aims

The aim of the present study was to compare MELD score, Child-Turcotte-Pugh (CTP) score, modified Maddrey's Discriminant Function (DF) score, and the related variables in predicting in-hospital mortality of patients with alcoholic hepatitis.

Methods

A retrospective chart review and statistical analyses were done on 202 patients consecutively admitted for alcoholic hepatitis from 1997 to 2002 at the Liver Unit at Rancho Los Amigos Medical Center.

Results

Twenty-nine patients died during the hospitalization. Admission MELD score (OR 1.1, P=0.005), first week MELD score (OR 1.2, P<0.0001), and first week increase in MELD score (OR 1.3, P<0.0001) were independently associated with in-hospital mortality. The area under the receiver operating curve (AUC) for the first week increase in MELD score was higher compared to CTP score (P=0.0004) and DF score (P=0.059). Moreover, the first week MELD score ≥20 had the best sensitivity (91%) and specificity (85%) compared with admission or first week change MELD score.

Conclusions

The present study indicates that in patients with alcoholic hepatitis, admission, first week, and first week change in MELD score are significantly independent predictors for in-hospital mortality. MELD score is a more valuable model than CTP or DF score in patients admitted with alcoholic hepatitis.

Introduction

Alcoholic liver disease encompasses three clinical and histologic phases ranging from fatty liver, alcoholic hepatitis (AH), and alcoholic cirrhosis. Alcoholic hepatitis and/or cirrhosis occur in approximately 15–20% of those who abuse alcohol over a prolonged period of time with different threshold for men and women [1], [2]. Once developed, AH has mortality exceeding fifty percent in some series [3]. Alcohol abstinence and correcting nutritional deficiencies have been the main treatment options. In severe cases, glucocorticosteroids or pentoxifylline have been used to reduce short-term mortality [4], [5], [6], [7]. Identifying individuals with high risk for mortality is paramount in management of AH.

Multiple prognostic factors studied over the last decade include: age, amount of alcohol consumed, aspartate transaminase (AST) to alanine transaminase (ALT) ratio, histological severity, creatinine, and white blood cell count (WBC) [8], [9]. The Discriminant Function (DF) formula proposed by Maddrey et al. is the most used prognostic index clinically. The DF score is derived from a formula that includes total bilirubin and prothrombin time in its calculation [4]. The numerical value derived correlates with prognosis of AH. A DF score of greater than 32 correlates with greater than 50% mortality at one month [4]. This score has also been used to predict possible benefit from corticosteroids therapy in patients with AH [4].

Model for end-stage liver disease (MELD) and Child-Turcotte-Pugh (CTP) scores have been validated as disease severity indices for patients with end-stage liver disease. The numerical value of the MELD score correlates with the three-month mortality rate [10]. Even though individual components of the MELD formula and CTP scoring system have been studied to determine prognostic values in AH, only one study to date has looked at MELD score as a prognostic index in AH [11]. This study involved a population of only thirty-four patients with a confirmed diagnosis of AH. The findings from this study suggest that an admission MELD score of greater than 11 has a sensitivity and specificity of 86 and 81%, respectively, in predicting 30-day mortality [11]. The study is limited by small cohort size and lacks comparison with CTP scores. Herein, we present the findings of a retrospective study comparing MELD, CTP, and DF scores in predicting in-hospital mortality in a large cohort of patients with AH.

Section snippets

Patient population

The study involved 202 patients consecutively admitted from January 1997 to December 2002 to the Liver Unit of the University of Southern California at Rancho Los Amigos Medical Center in Downey, California. The study protocol was approved and informed consent was exempted by the Institutional Review Board. The inclusion criteria were (a) admission diagnosis of AH; (b) history of heavy alcohol abuse; (c) negative serology for hepatitis B surface antigen (HBsAg); (d) negative hepatitis C

Baseline demographics

A total of 224 consecutive patients admitted to the liver unit with the clinical diagnosis of alcoholic hepatitis. Of the 224 patients, 202 patients met the inclusion criteria for the study and analysis. Table 1 summarizes the demographics and clinical characteristics of this cohort of 202 patients. Besides admission diagnosis of alcoholic hepatitis, all 202 patients have history of heavy alcohol use, negative hepatitis B and C serologies, negative HIV antibody, and absence of other causes of

Discussion

Discriminant Function score greater than 32 is shown to be associated with greater than 50% mortality and is currently used as a threshold to start either corticosteroid or pentoxifylline therapy. However, the value of MELD score in predicting outcome in patient with AH remains to be confirmed. Our retrospective study compared MELD, CTP, and DF scores as predictive models to assess in-hospital mortality in a large cohort of patients with AH.

The diagnosis of AH is usually made clinically. Liver

Acknowledgements

Part of this work was presented in 2003 DDW meeting in Orlando, FL, in May 2003. Dr B.A. Runyon's current address is Division of Gastroenterology, Loma Linda University Medical Center, 11234 Anderson Street, Loma Linda, CA 92354; Drs K.-Q. Hu and N. Kyulo's current address is Division of Gastroenterology, University of California, Irvine, 101 The City Drive, Building 53, Rm. 113, Orange, CA 92868. We would like to thank Dr Jim Chen for his invaluable time in reviewing and commenting the

Cited by (0)

View full text