How this instrument works
Maddrey's Discriminant Function (DF) combines two markers of liver synthetic and excretory function — how prolonged the prothrombin time is compared with the lab's own control value, and the total bilirubin — into a single score used to gauge the severity of alcoholic hepatitis. It was originally developed by Willis Maddrey and colleagues in the 1970s to identify patients severe enough to be considered for corticosteroid treatment, and the formula was later standardized in the form still used today by Carithers and colleagues in a 1989 trial.
The prothrombin time (PT) comparison is deliberately relative rather than absolute: the calculation uses the difference between the patient's PT and the control PT run by the same laboratory on the same reagents, not a fixed number of seconds. That matters because PT reagents and normal ranges vary somewhat between labs and over time, so a 'normal' PT at one lab isn't necessarily identical to a 'normal' PT at another — using the local control corrects for that.
A DF at or above 32 is the threshold this calculator uses to flag the severe range historically associated with meaningfully higher short-term mortality and used as an entry criterion in corticosteroid trials for alcoholic hepatitis. Some published sources instead use a strict greater-than-32 cutoff rather than an inclusive one; this calculator treats 32 itself as meeting the severe threshold. Either way, DF is one input among several a clinician weighs — alongside markers like the MELD score, signs of infection or gastrointestinal bleeding, and overall clinical status — before deciding on steroid therapy, not a standalone treatment trigger.
- Enter the patient's prothrombin time in seconds.
- Enter the laboratory's control (normal reference) prothrombin time in seconds — this is lab-specific, not a fixed constant, so use the value reported alongside the patient's PT.
- Enter the total bilirubin in mg/dL.
- Read the resulting Discriminant Function value and whether it falls in the severe (DF ≥32) range.
Worked example — patient PT 19s vs. control PT 12s, bilirubin 8 mg/dL
A patient's prothrombin time runs 19 seconds against the lab's control value of 12 seconds — a 7-second prolongation — with a total bilirubin of 8 mg/dL. The Discriminant Function is 4.6 × 7 + 8 = 40.2, which is above the 32-point threshold and falls in the severe range this calculator flags.
For comparison, a patient with a smaller 2-second prolongation (PT 14s vs. control 12s) and a lower bilirubin of 3 mg/dL scores 4.6 × 2 + 3 = 12.2 — well below 32, not in the severe range by this criterion alone. A third case lands exactly on the boundary: a 4-second prolongation (PT 16s vs. control 12s) with bilirubin of 13.6 mg/dL gives 4.6 × 4 + 13.6 = 32.0 exactly, which this calculator's inclusive ≥32 threshold flags as severe.
Questions
What does a Discriminant Function of 32 or higher mean?
It flags the range historically associated with more severe alcoholic hepatitis and higher short-term mortality, and it's the range used as an entry criterion in the corticosteroid trials that shaped current treatment guidance. It is not, by itself, an automatic instruction to start steroids — clinicians weigh it alongside contraindications like active infection or gastrointestinal bleeding, other severity scores such as MELD, and the overall clinical picture.
Why does the formula use a 'control' prothrombin time instead of a fixed number?
Because PT results depend on the specific reagents and instruments a laboratory uses, a given PT in seconds doesn't mean exactly the same thing at every lab. Reporting a control PT alongside the patient's result lets the Discriminant Function measure how prolonged the patient's clotting time is relative to that lab's own normal range, rather than assuming one universal baseline.
Is DF ≥32 or DF >32 the correct threshold for 'severe'?
Both appear in the literature. This calculator uses an inclusive ≥32 cutoff, so a DF of exactly 32.0 is flagged as severe. Some sources instead use a strict greater-than-32 cutoff. The difference only matters for scores landing exactly at 32, and either way the number is meant to support clinical decision-making, not replace it.
Who developed Maddrey's Discriminant Function and when?
Willis Maddrey and colleagues introduced the formula in a 1978 Gastroenterology paper studying corticosteroid therapy for alcoholic hepatitis. The version commonly used today was standardized in a widely cited 1989 trial by Robert Carithers and colleagues in the Annals of Internal Medicine, which helped establish the DF ≥32 threshold used in later steroid trials and guidelines.
Does a high Discriminant Function always mean corticosteroids should be given?
No. DF ≥32 identifies patients severe enough that steroids have historically been considered, but the actual decision also depends on ruling out contraindications — active or uncontrolled infection, gastrointestinal bleeding, renal failure, and other factors — and increasingly on additional scoring tools and response assessments like the Lille score used after a few days of treatment. This calculator computes the number only; the treatment decision is a clinical one.
Is this calculator meant for patients to use on themselves?
No. Maddrey's Discriminant Function is a clinical risk-stratification tool built for use by clinicians managing a patient with confirmed or suspected alcoholic hepatitis, using lab values (prothrombin time, control PT, bilirubin) that come from bloodwork ordered and interpreted in a medical setting. It isn't a symptom checker or a substitute for medical evaluation.
References
- Maddrey WC, Boitnott JK, Bedine MS, et al. — Corticosteroid Therapy of Alcoholic Hepatitis, Gastroenterology. 1978;75(2):193-199 (PMID 352788)
- Carithers RL, Herlong HF, Diehl AM, et al. — Methylprednisolone Therapy in Patients with Severe Alcoholic Hepatitis, Ann Intern Med. 1989;110(9):685-690 (PMID 2648927)
- MDCalc — Maddrey's Discriminant Function for Alcoholic Hepatitis
Read this first: This instrument computes a screening figure from population formulas — it is not a diagnosis, and it cannot see the whole picture a clinician can. Use it to inform a conversation, not to replace one.