Maddrey Discriminant Function Calculator

Maddrey Discriminant Function Calculator for Gastroenterology. The single actionable cutoff is 32. A score of 32 or higher defines severe alcohol-associated hepatitis and historically carries a 28-day mortality on the order of 30-50% without treatment; this is the group in which corticosteroids reduced short-term death in the pivotal trials. A score below 32 indicates mild-to-moderate disease with substantially lower short-term mortality where steroids are not indicated and add net harm. Because the PT term dominates, small differences in how the local lab reports PT and its control value can swing borderline scores, so many centers cross-check with MELD (a MELD above roughly 20-21 flags comparable severity) rather than relying on mDF alone at the margin.

How this calculator works

The Maddrey Discriminant Function (mDF), also called the modified discriminant function, estimates short-term mortality in alcohol-associated (alcoholic) hepatitis from two coagulation and cholestasis markers. The modified formula used in practice is 4.6 x (patient prothrombin time in seconds - control PT in seconds) + total serum bilirubin in mg/dL. It derives from stepwise discriminant analysis that isolated PT prolongation and bilirubin as the independent laboratory predictors of death, weighting the PT gap heavily to capture synthetic liver failure.

When to use this calculator

Apply mDF in patients with a clinical diagnosis of alcohol-associated hepatitis (recent heavy drinking, jaundice, elevated AST/ALT usually under 300-400 IU/L with AST:ALT above 1.5-2) to decide who is sick enough to consider corticosteroids. It is a severity trigger, not a diagnostic test. Do not use it in decompensated cirrhosis without superimposed acute hepatitis, in cholestasis from biliary obstruction, or when the bilirubin elevation is not driven by liver injury (e.g., hemolysis), since those inflate the score without reflecting the target physiology.

Inputs used

  • Patient prothrombin time
  • Control prothrombin time
  • Serum bilirubin

Clinical interpretation

The single actionable cutoff is 32. A score of 32 or higher defines severe alcohol-associated hepatitis and historically carries a 28-day mortality on the order of 30-50% without treatment; this is the group in which corticosteroids reduced short-term death in the pivotal trials. A score below 32 indicates mild-to-moderate disease with substantially lower short-term mortality where steroids are not indicated and add net harm. Because the PT term dominates, small differences in how the local lab reports PT and its control value can swing borderline scores, so many centers cross-check with MELD (a MELD above roughly 20-21 flags comparable severity) rather than relying on mDF alone at the margin.

Worked example

A patient with alcohol-associated hepatitis has a PT of 22 seconds, a laboratory control PT of 12 seconds, and total bilirubin of 15 mg/dL. mDF = 4.6 x (22 - 12) + 15 = 4.6 x 10 + 15 = 46 + 15 = 61. A score of 61 is well above the 32 threshold, marking severe disease with roughly 30-50% untreated 30-day mortality, and identifies a patient in whom corticosteroids (after ruling out active infection and GI bleeding) should be considered, typically alongside a MELD and a planned day-7 Lille score reassessment.

Limitations and safety notes

The formula is highly sensitive to PT reagent and instrument variability: because INR standardization does not apply to the raw PT-minus-control term, the same patient can score differently across labs, making values near 32 unreliable. It uses no age, renal function, or sodium, so it can underestimate risk in older patients or those with hepatorenal syndrome, where MELD or GAHS/ABIC discriminate better. It does not predict steroid responsiveness; that requires the day-7 Lille score, and a Lille above 0.45 signals non-response and a case to stop steroids. It also cannot distinguish acute alcohol-associated hepatitis from other causes of a high bilirubin and prolonged PT.

Frequently asked questions

Should I use PT in seconds or INR in the Maddrey formula?

Use PT in seconds with the laboratory's own control PT, not INR. The 4.6 coefficient was derived for the PT-control difference; substituting INR gives an incorrect score. This PT dependence is also why values near the 32 cutoff should be interpreted cautiously and cross-checked against MELD.

Is bilirubin entered in mg/dL or micromol/L?

The commonly used bedside version takes bilirubin in mg/dL and adds it directly. The original Carithers formulation expressed bilirubin in micromol/L divided by 17.1, which is the unit conversion back to mg/dL, so both reduce to the same value. Confirm your calculator's units before trusting a borderline result.

What score means severe disease and triggers steroids?

A Maddrey score of 32 or greater defines severe alcohol-associated hepatitis and is the threshold at which corticosteroid therapy is considered, provided sepsis, uncontrolled GI bleeding, and untreated infection are excluded first.

How does Maddrey relate to MELD and Lille?

Maddrey selects who has severe disease and may benefit from steroids; MELD offers a more reproducible severity estimate that many centers now prefer for prognosis; and the Lille score, calculated after 7 days of steroids, determines whether to continue or stop them based on treatment response.

References

  • Maddrey WC, Boitnott JK, Bedine MS, Weber FL, Mezey E, White RI. Corticosteroid therapy of alcoholic hepatitis. Gastroenterology. 1978. PMID: 352788.
  • Carithers RL, Herlong HF, Diehl AM, Shaw EW, Combes B, Fallon HJ, Maddrey WC. Methylprednisolone therapy in patients with severe alcoholic hepatitis. A randomized multicenter trial. Ann Intern Med. 1989. PMID: 2648927.
  • European Association for the Study of the Liver. EASL Clinical Practice Guidelines: Management of alcohol-related liver disease. J Hepatol. 2018. PMID: 29628280.

Editorial review and citation methodology

Reviewed by the Quick Medical Calculator Editorial Team. Last reviewed: April 20, 2026. The review checks calculator inputs, intended population, interpretation, limitations, and source alignment.

  • Prefer original validation studies for scoring systems and prediction tools.
  • Use current specialty society guidance, transplant allocation policy, public health guidance, or regulator resources when they govern clinical use.
  • Include limitations and safety notes when a calculator is population-specific, context-dependent, or unsuitable as a standalone decision tool.

Related reviewed calculators