Alcoholic hepatitis: Diagnosis and prognosis


  • This study was supported by grant AA21788 from NIAAA (to V. H. S.).

  • Potential conflict of interest: Nothing to report.


Watch a video presentation of this article

Watch the interview with the author

Answer questions and earn CME

Alcoholic hepatitis (AH) is an acute inflammatory syndrome that arises in patients with chronic liver inflammation. AH ranges in severity from mild forms that resolve with conservative management to severe presentations that are associated with mortality as high as 65%.1, 2 Of the people who abuse alcohol for extended duration, 15% to 20% will develop alcoholic hepatitis and/or cirrhosis.3 Age, female sex, and increased body mass index are all independent risk factors for development of AH.4 A level of alcohol consumption for increased risk of AH has not been determined, but observational studies have shown an increased risk of cirrhosis with ingestion of greater than 10 to 20 g of alcohol per day in women and greater than 20 to 40 g per day in men.4 Given the high mortality rate for patients with severe disease, it is imperative to determine which patients require supportive treatment versus more aggressive care. Several scoring systems have been developed to determine prognosis in patients with AH and aid in determining which patients will benefit from interventions beyond supportive care. The most common scores include Maddrey's discriminant function (DF), the model for end-stage liver disease (MELD), the Glasgow alcoholic hepatitis score (GAHS), the Lille model, and the age-bilirubin-INR-creatinine (ABIC) score (where INR is international normalized ratio).


ABIC, age-bilirubin-INR-creatinine; AH, alcoholic hepatitis; ALD, alcoholic liver disease; ALT, alanine aminotransferase; ANI, alcoholic/nonalcoholic steatohepatitis index; AST, aspartate aminotransferase; DF, discriminant function; GAHS, Glasgow alcoholic hepatitis score; GGT, gamma-glutamyltransferase; INR, international normalized ratio; MELD, model for end-stage liver disease; PT, prothrombin time.


Obtaining a history of alcohol abuse is an important clinical assessment in the diagnosis of alcoholic liver disease (ALD). The majority of patients who develop ALD consume >35 units of alcohol a week, where one unit is defined as 10 to 12 g of alcohol.5 This resulted in the recommendation that weekly limits of alcohol intake should be restricted to 21 units for men and 14 units for women.5 Symptoms of AH tend to include constitutional symptoms such as fatigue, anorexia, weight loss, nausea, vomiting, and abdominal pain.3, 4, 6 Clinical exam may demonstrate hepatomegaly, jaundice, and splenomegaly.6 Severe cases may present with signs of portal hypertension, including gastrointestinal bleeding, ascites, and hepatic encephalopathy.3

Several laboratory abnormalities have been associated with alcohol consumption, including elevated aspartate aminotransferase (AST) levels 5 to 10 times the normal limit (AST levels are rarely >300 IU/dL) and an elevated AST/alanine aminotransferase (ALT) ratio of 2:1.7 A bilirubin level >2 mg/dL is generally required to make the diagnosis of AH, but is often >10 mg/dL at diagnosis.4 Other markers of chronic alcohol intake include elevated gamma-glutamyltransferase (GGT) levels and increased mean corpuscular volume. However, these markers are indirect due to the toxicity of alcohol, which significantly reduces their sensitivity to <70% for the aminotransferases and GGT and <50% for mean corpuscular volume.3 Another marker for AH is leukocytosis, which can be confounded by coexisting infection.4

Many experts feel that liver biopsy is not needed to diagnosis AH, but may be obtained when the clinical picture appears unclear. The most common finding is macrovesicular steatosis, which may also be present in patients with nonalcoholic steatohepatitis (Fig. 1). Mallory bodies may be present in up to 65% of patients with AH, and although Mallory bodies are indicative of AH, they are not pathognomonic.4 However, in a recent study by Mookerjee et al.,8 they identified a role of liver biopsy in the diagnosis and prognosis of patients with acute deterioration of alcoholic cirrhosis. In this study, biopsies from 68 patients with acute deterioration of alcoholic cirrhosis were processed with routine stains and K8/18 immunohistochemistry. The authors found that the K8/18 stain, which has a high concordance rate for the diagnosis of balloon degeneration, allowed for grading the severity of AH in this patient cohort.8

Figure 1.

Histology micrograph of liver in patient with AH. Classical findings include macrovesicular steatosis, Mallory bodies, balloon degeneration of hepatocytes, and chicken wire fibrosis. Reprinted with permission from WB Saunders. Copyright 2000, WB Saunders.

Given that many of the clinical, laboratory, and pathological findings of alcoholic steatohepatitis overlap with non-alcoholic steatohepatitis (NASH), Dunn et al.9 developed the alcoholic/nonalcoholic steatohepatitis index (ANI). Logistic regression identified mean corpuscular volume, AST/ALT ratio, body mass index, and sex as the most important variables separating patients with ALD from nonalcoholic fatty liver disease. An ANI of greater than zero favors ALD, and an ANI less than zero favors nonalcoholic fatty liver disease (Fig. 2). ANI had a c statistic of 0.989 in the derivation sample, and 0.974, 0.989, 0.767 in the three validation samples, which included hospitalized, ambulatory, and pretransplantation patients. (The ANI calculator can be found at

Figure 2.

ANI calculator can assist to predict the likelihood of alcoholic liver disease in patients with histological findings of steatohepatitis and uncertainty regarding an alcoholic cause of the steatohepatitis. (See text for web link.)


Several prognostic scores are described below. In general, most of them perform similarly well for predicting short-term mortality in AH. Their performance for long-term survival is less effective, because abstinence is the most important variable for long-term survival.10 The DF has been the mainstay for determining prognosis in patients with AH. A score of ≥32 is associated with significant short-term mortality and has been used as a cutoff for use of glucocorticoid treatment.1 The main drawback to the DF is that it requires a prothrombin time (PT) for the calculation. Prothrombin time varies between different laboratories and may lead to variations in the INR between laboratories.1 This has led to the development of new scoring systems, including the MELD, ABIC, GAHS, and Lille model.

The MELD score was originally developed to assess the short-term prognosis of patients with cirrhosis undergoing elective transjugular intrahepatic portosystemic shunts using objective criteria. In a study by Dunn et al.,1 the MELD score was found to be an independent predictor of mortality in patients with AH based on a retrospective cohort trial1 (Fig. 3). The MELD was also found to be comparable to the DF in predicting mortality within 30 days. A cutoff score of 21 has a sensitivity of 75% and specificity of 75% in predicting 90-day mortality. A benefit of the MELD score over the DF is INR, which is standardized across laboratories, whereas the PT is not.1 The MELD score also includes weighting of the INR and bilirubin level to reduce the influence of values at extremes. Serum creatinine is also used in the MELD score, highlighting the importance of the presence of renal failure in the prognosis of patients with AH. (The MELD calculator to predict AH mortality is available at

Figure 3.

MELD 90-day survival calculator. (See text for web link.) Reprinted with permission from Hepatology.1 Copyright 2005, Wiley.

The ABIC score was derived from a patient cohort from Barcelona, Spain. It has similarities to the MELD score but also includes age as a variable. Additionally, it assigns different weighting for values of INR, bilirubin level, and creatinine level.11 ABIC score predicts mortality in a manner similar to the other scores referred to here.

The GAHS score uses age, white blood cell count, urea level, PT ratio or INR, and bilirubin level to calculate a score between 5 and 12, with scores above 9 associated with poor prognosis.12 The initial study showed an accuracy of 81%, with values on day 1 in predicting day 28 mortality. In a follow-up study, patients with scores of 9 or greater had significant reduction of mortality if they received corticosteroids.12

The Lille model is used to predict the mortality in patients with severe AH who have received glucocorticoids for treatment. This model includes six variables (age, renal insufficiency [INR >1.3 or creatinine clearance <40], albumin, PT, bilirubin level, and bilirubin level on day 7) that are used to determine mortality at 6 months.13 A meta-analysis of the effect of corticosteroids in the short-term survival of patients with severe AH demonstrated that patients who were classified as responders (Lille score <0.56) had significant improvement in survival at 28 days when treated with corticosteroids compared with the nonresponder group (Lille score > 0.56).14 A simpler and very effective alternative is the end of treatment bilirubin response. If the bilirubin does not improve 1 week after corticosteroids, it should be discontinued.15


AH is an acute-on-chronic inflammatory condition that ranges in severity from mild to severe and is associated with mortality rates as high as 65%.1, 2 An accurate history of alcohol abuse is key to the diagnosis but may be difficult to obtain. Laboratory and pathogenic evidence can help in the diagnosis, but there is diagnostic overlap. Use of the ANI calculator can help determine the probability of an alcoholic versus nonalcoholic cause for underlying steatohepatitis. New histologic markers are also being examined to aid in diagnosis. There are multiple different scoring systems that have been developed to determine the prognosis of patients with AH (Table 1). A recent comparison study found no major difference between the MELD, GAHS, Lille, or ABIC score in predicting mortality10 (Table 2). Thus, no one prognostic score can be definitively recommended above another.

Table 1. Overview of the Formulas of the Prognostic Scores Discussed
Prognosis ScoreFormula
  • *

    R = 3.19 − 0.101 × (age in years) + 0.147 × (albumin day 0 in g/L) + 0.0165 × (bilirubin day 0 − bilirubin day 7 (mmol/L)) − 0.206 × (renal insufficiency) − 0.0065 × (bilirubin day 0 in mmol/L) − 0.0096 × (PT in seconds).

DF4.6 × (patient PT − control PT) + bilirubin
MELD3.8 × ln(bilirubin (mg/dL) + 1.2 × ln(INR) + 9.6 × ln(creatinine (mg/dL)) + 6.43
GAHS point2 points3 points
Leukocyte count (109/L)<15≥15
Urea level (mmol/L)<5≥5
PT ratio<1.51.5-2.0>2.0
Bilirubin level (mmol/L)<125125-250250
ABIC(age × 0.1) + (serum bilirubin (mg/dL)) × 0.08) + (serum creatinine (mg/dL) × 0.3) + (INR × 0.8)
Lilleexp(−R)/(1 + exp(−R))*
Table 2. Overview of Prognostic Scores Sensitivity and Specificity of Mortality When Calculated on Day 1 of Diagnosis9, 10, 11, 13
Prognosis Score*8- to 30-Day Mortality84- to 90-Day Mortality180-Day Mortality
  • Cutoff scores: DF, 37; MELD, 21; GAHS, 9;

  • *, 1

    ABIC1, 6.71;

  • 2

    ABIC2, 9; Lille, 0.45.