Maddrey discriminant function calculator (modified DF)

How this Maddrey discriminant function calculator works

This Maddrey discriminant function calculator multiplies the prolongation of the prothrombin time, the patient's value minus the laboratory's control value in seconds, by 4.6 and adds the total serum bilirubin in mg/dL. That is the modified discriminant function (mDF) written out by Carithers and colleagues in the Annals of Internal Medicine in 1989, in a randomized, double-blind trial of a corticosteroid against placebo in 66 patients at four university hospitals. The function itself comes from Maddrey and colleagues, who reported in Gastroenterology in 1978 that, among 55 patients with alcoholic hepatitis, a stepwise discriminant analysis linked death independently to the prolongation of the prothrombin time and the height of the serum bilirubin at entry. Carithers enrolled patients with a discriminant function above 32 or with spontaneous hepatic encephalopathy. The same cut-off is now the threshold for severe alcohol-associated hepatitis in the guidance of the European (EASL) and American (AASLD) liver associations.

The calculator opens with a worked example: a patient prothrombin time of 22 seconds against a control of 12 seconds is a prolongation of 10 seconds, and 4.6 × 10 = 46.0. Adding a bilirubin of 12 mg/dL (205 µmol/L) gives a modified DF of 58.0, above the cut-off. The result is rounded to one decimal place before it is compared with 32.

The original 1978 function is not the same number. As Forrest and colleagues summarize it, Maddrey's first version combined 4.6 × the prothrombin time with bilirubin in mg/dL, and placebo-treated patients with a value above 93 had a 28-day survival of 25%, against 100% at 93 or below. Use the cut-off of 32 only with the modified formula on this page.

Maddrey score results: what a value of 32 means

A Maddrey score result is read against a single cut-off: a modified discriminant function of 32 or more identifies severe alcohol-associated hepatitis, and a value below 32 does not. The EASL Clinical Practice Guidelines of 2018 state that in its modified version a cut-off value of 32 identifies patients with severe disease and is usually the threshold used for starting specific therapy. EASL adds that, without treatment, the one-month survival of patients with an mDF of 32 or more has improved from 50% in early publications to 85% in recent trials, while patients with an mDF below 32 had a less than 10% risk of death at one month. In the Carithers trial, 11 of the 31 placebo-treated patients (35%) died within 28 days. The 2024 American College of Gastroenterology (ACG) guideline puts the 30-day mortality at an mDF of 32 or more at approximately 20% to 50%.

Guidelines discuss corticosteroids for severe alcohol-associated hepatitis when there is no contraindication such as active infection, and both EASL and AASLD then use the Lille score after seven days to judge response. The 2019 AASLD guidance states that the MDF of 32 or more should be used to assess the need for corticosteroids or other medical therapies, and that a MELD score above 20 should also prompt that consideration. This page gives the score only; treatment decisions follow the full guideline and the patient's clinical picture.

When the Maddrey discriminant function is less accurate

The Maddrey discriminant function is less accurate mainly because it depends on the absolute prothrombin time in seconds. Forrest and colleagues, deriving the Glasgow alcoholic hepatitis score in Gut in 2005, pointed out that prothrombin time values differ significantly between assays, which creates an inherent inaccuracy in the mDF and limits its translation between clinical practices; in their cohort the mDF incorrectly predicted 28-day outcome in 51% of cases. The control value is specific to each laboratory and reagent, so always use the control printed with the patient's result. The function has no kidney term, although creatinine is a major determinant of outcome, and the ACG 2024 guideline lists the use of INR rather than prothrombin time and the inclusion of creatinine among the advantages of the MELD score. In a global study of 2,581 patients from 85 centers in 11 countries, the mDF predicted 28-day and 90-day death less accurately than MELD and the other scores tested.

The prothrombin time also changes for reasons unrelated to the liver, such as anticoagulant treatment or vitamin K deficiency, and bilirubin can rise with hemolysis or biliary obstruction. The score was built and tested in people with a clinical diagnosis of alcohol-associated hepatitis, not in other liver diseases. For the full comparison with MELD, see the guide to Maddrey vs MELD in alcohol-associated hepatitis.