Maddrey Score in Alcoholic Hepatitis: Why 32 Matters

The Maddrey score, formally called Maddrey’s discriminant function (MDF or DF), is a clinical tool used to gauge the severity of alcoholic hepatitis and guide treatment decisions. Developed in the late 1970s and refined in a landmark 1989 trial, it relies on just two lab values to produce a number that helps doctors decide whether a patient needs corticosteroid therapy. A score above 32 flags severe disease with high short-term mortality risk, making it one of the first and most widely used prognostic benchmarks in liver medicine.

What Goes Into the Calculation

The formula is deceptively simple. It multiplies the difference between a patient’s prothrombin time and a control prothrombin time by 4.6, then adds the serum bilirubin level. In the version used in most clinical practice, bilirubin measured in micromoles per liter is divided by 17.1 to convert it to milligrams per deciliter before being plugged in.¹ Prothrombin time reflects how well the liver is producing clotting factors, while bilirubin measures how effectively the liver is clearing waste. Together they capture two of the most immediate consequences of severe liver inflammation: impaired clotting and deep jaundice.

The score is easy to compute at the bedside with routine bloodwork, which is a big reason it became the standard prognostic tool for alcoholic hepatitis for decades.² No imaging, no biopsy, no complex algorithms needed. A doctor can calculate it within minutes of receiving basic lab results.

Why 32 Is the Threshold That Matters

A Maddrey score of 32 or higher defines what clinicians call “severe” alcoholic hepatitis. Below 32, most patients can be managed with supportive care: fluids, nutrition, and alcohol cessation. Above 32, the picture darkens considerably. Without treatment, patients with untreated severe alcoholic hepatitis and a score at or above 32 face roughly a 50 to 60 percent chance of surviving 28 days.³ That number alone explains why clinicians treat a score above 32 as an urgent signal to consider corticosteroids.⁴

The biology behind that poor prognosis involves a cascade of inflammation. When the liver is overwhelmed by chronic alcohol exposure, damaged liver cells release signals that trigger a surge of inflammatory molecules. Tumor necrosis factor (TNF) is one of the key players, and it rises sharply in the most severe cases of alcoholic hepatitis.⁵ The disease essentially represents the end result of a complex interplay between how the body metabolizes alcohol, the resulting inflammation, and the immune system’s response to it.⁶ At a Maddrey score above 32, that interplay has reached a point where the liver’s own inflammatory response is doing as much damage as the alcohol itself.

Corticosteroids and the Evidence Behind Them

Corticosteroids, specifically prednisolone, became the first-line treatment for severe alcoholic hepatitis largely because of trials that used the Maddrey score as their entry criterion. A meta-analysis pooling individual patient data from five randomized trials of patients with a discriminant function of 32 or higher found that about 80 percent of steroid-treated patients survived to 28 days, compared with roughly 66 percent of those who received no steroids.⁷ That difference is meaningful when you’re talking about a disease that can kill within weeks.

The picture gets more complicated beyond the first month, though. The largest trial to date, known as STOPAH, randomized over a thousand patients with severe alcoholic hepatitis to prednisolone, pentoxifylline, both, or neither. Prednisolone lowered 28-day mortality from about 17 percent in the placebo group to about 14 percent, and when the analysis accounted for disease severity, the benefit became statistically clearer. But by 90 days and one year, the survival advantage had disappeared.⁸ Pentoxifylline, an anti-inflammatory drug that was once considered an alternative, showed no survival benefit at any time point in the same trial.⁹

Steroids also carry real risks. Serious infections occurred in about 13 percent of prednisolone-treated patients versus 7 percent of those not receiving it.¹⁰ Since these patients already have compromised liver function and often poor nutritional status, an infection on top of that can be fatal. This tradeoff is why clinicians don’t simply give steroids to everyone and why monitoring the response closely matters.

Tracking Steroid Response With the Lille Score

Starting steroids based on the Maddrey score is only the first step. The next question is whether the steroids are actually working. That’s where the Lille score comes in. After seven days of corticosteroid treatment (some centers check at day four), clinicians calculate a separate score that incorporates six variables: the patient’s age, kidney function, albumin, prothrombin time, bilirubin at admission, and crucially, how much the bilirubin has changed since treatment began.¹¹

A Lille score below 0.45 identifies steroid responders, and these patients typically complete a full 28-day course of prednisolone. A score of 0.45 or higher identifies non-responders, meaning steroids should be stopped because continuing them adds infection risk without providing survival benefit.¹² The change in bilirubin is the key signal: if the liver starts clearing bilirubin more effectively within that first week, it suggests the inflammatory damage is receding. If bilirubin stays flat or climbs, the liver isn’t recovering despite the treatment.

This two-stage approach, using the Maddrey score to decide whether to start steroids and the Lille score to decide whether to continue, has become the standard clinical workflow. Infection risk also plays into the equation; response to steroids as gauged by the Lille model helps clinicians weigh whether the benefit of continued treatment outweighs the rising risk of serious infections.¹³

The Control Prothrombin Time Problem

One persistent criticism of the Maddrey score is that it depends on the “control” prothrombin time, a reference value that varies from lab to lab. Different hospitals use different reagents and instruments, and the control value they report can range anywhere from about 12 seconds to nearly 15 seconds. That might sound like a small difference, but it shifts the final score enough to change treatment decisions for a meaningful number of patients.

A study examining this issue found that the percentage of patients classified as having severe alcoholic hepatitis (a score of 32 or above) dropped from 70 percent to 52 percent simply by using a control PT of 14.8 seconds instead of 12 seconds.¹⁴ That’s a big swing: nearly one in five patients crossing the treatment threshold purely because of which number a particular lab plugs in as the control. In the same study, steroid use tracked with these reclassifications, but the actual 30-day mortality rate didn’t differ significantly between those who received steroids and those who didn’t. The score’s ability to predict survival was statistically similar regardless of which control value was used, but the practical consequence of who gets steroids was not.

This is one of the reasons the score has faced increasing competition from alternatives that don’t have a lab-dependent variable baked in.

How the MELD Score Compares

The Model for End-Stage Liver Disease (MELD) score, originally developed to predict survival in patients undergoing a procedure to relieve portal hypertension, has increasingly been applied to alcoholic hepatitis. It uses bilirubin, creatinine, and the international normalized ratio (INR) for clotting, avoiding the control-PT issue entirely since INR is already standardized across labs.

A large global study found that MELD and its sodium-adjusted variant (MELD-Na) consistently outperformed the Maddrey score in predicting death at both 28 and 90 days. The accuracy of MELD for 28-day mortality was meaningfully higher, and the Maddrey score had the lowest predictive accuracy of any score tested in the study.¹⁵ Other scoring systems, including the Glasgow Alcoholic Hepatitis Score and the ABIC model (which incorporates age, bilirubin, INR, and creatinine), have also been developed and studied as alternatives.¹⁶

Despite this, the Maddrey score persists in clinical guidelines and practice for a few reasons. It was the original benchmark used in the steroid trials, so the treatment thresholds are anchored to it. Most clinicians trained on it know the 32 cutoff instinctively. And while MELD is better at predicting who will die, there’s no universally agreed MELD cutoff that triggers the same clear treatment algorithm the way 32 does for the Maddrey score. A MELD above 21 is commonly cited as indicating severe disease in this context, but the evidence base linking a specific MELD threshold to steroid benefit is thinner than the decades of trials built around the Maddrey cutoff of 32.

When Steroids Fail and Transplant Enters the Picture

For patients with severe alcoholic hepatitis who don’t respond to corticosteroids, the options narrow quickly. The Lille score identifies these non-responders within the first week, but the question of what to do next doesn’t have an easy pharmacological answer. The most dramatic intervention studied is early liver transplantation, performed without requiring the traditional six-month abstinence period that most transplant programs historically demanded.

A pivotal French study compared carefully selected non-responders who received early transplantation to matched non-responders who did not. The six-month survival rate was about 77 percent in the transplant group versus 23 percent in the non-transplant group, a stark gap that persisted through two years of follow-up.¹⁷ This study, published in 2011, fundamentally shifted the conversation about transplantation and alcoholic liver disease. Over a decade of subsequent experience has confirmed that when performed under strict selection protocols, early transplantation offers a clear survival benefit with acceptable rates of return to drinking afterward.¹⁸

Access to early transplantation remains limited, though. Not every center offers it, organ supply is constrained, and many transplant committees still weigh the six-month sobriety rule heavily. The Maddrey and Lille scores together help identify the patients who stand to benefit most, but the real bottleneck is institutional willingness and organ availability.

Nutrition and Caloric Intake

Malnutrition is almost universal in severe alcoholic hepatitis, and there’s been longstanding interest in whether aggressive nutritional support could improve outcomes alongside or instead of steroids. A randomized trial tested intensive enteral nutrition (delivered through a feeding tube) on top of corticosteroids versus corticosteroids alone. The intensive feeding strategy did not significantly reduce six-month mortality compared to standard care.¹⁹

But the same trial revealed something striking when the researchers looked at caloric intake regardless of how it was delivered. Patients who consumed fewer than about 21.5 kilocalories per kilogram of body weight per day had a dramatically higher six-month mortality than those who ate more. The difference was large: roughly 66 percent mortality in the low-intake group versus about 33 percent in those eating above that threshold.²⁰ In practical terms, whether or not a tube is involved, getting enough calories into a severely ill patient with alcoholic hepatitis matters enormously. The challenge is that these patients are often nauseated, encephalopathic, or simply unwilling to eat, and forcing nutrition through a tube doesn’t appear to solve the problem if the underlying liver dysfunction prevents the body from using those calories effectively.

What Determines Long-Term Survival

The Maddrey score and its companion scores are designed to predict short-term risk, primarily within the first month or two. Beyond that window, the single most important factor in long-term survival is whether the patient stops drinking. A study following patients who survived an episode of alcoholic hepatitis found that complete abstinence was reported in about 39 percent of survivors and was independently associated with improved long-term survival, even after accounting for how sick patients were at baseline.²¹ That roughly halved the risk of death compared to those who returned to drinking.

Long-term outcomes also depend on the state of the liver underneath the acute inflammation. Many patients with alcoholic hepatitis already have underlying cirrhosis, which doesn’t reverse even if the acute episode resolves. In those cases, survival hinges on both abstinence and the degree of irreversible scarring that has already accumulated.²² A patient whose Maddrey score was 40 but who stops drinking and has relatively preserved liver architecture may do better over five years than someone whose score was 35 but who has advanced cirrhosis and resumes drinking. The acute score captures the crisis; it doesn’t measure the structural health of the organ or the trajectory of the patient’s relationship with alcohol.

Kidney Complications and Severity Beyond Liver Function

One limitation of the Maddrey score is that it captures liver-specific dysfunction without directly measuring what’s happening to other organs. Kidney injury is one of the most dangerous complications in severe alcoholic hepatitis. Patients with alcoholic hepatitis who develop acute kidney injury reach more severe stages of kidney damage, require dialysis more often, and face higher 90-day mortality than patients with other forms of liver disease who develop the same complication. Among those who progress to hepatorenal syndrome, a condition where failing liver function drags the kidneys down with it, alcoholic hepatitis is an independent predictor of death.²³

The MELD score has an advantage here because it incorporates creatinine, giving it a built-in sensitivity to kidney dysfunction. The Maddrey score is blind to it. A patient with a Maddrey score just below 32 but rapidly rising creatinine could be in far more danger than the score suggests. This gap is one reason clinicians increasingly use multiple scoring systems in parallel rather than relying on any single number, and why newer composite models that fold in kidney function and other organ markers are gaining ground in clinical research.

Patients With Scores Below 32

Most of the attention around the Maddrey score focuses on patients above the 32 threshold, but patients below it aren’t necessarily safe. A study looking at hospitalized patients with alcoholic cirrhosis and Maddrey scores below 32 found that existing prognostic models performed only modestly well in predicting their short-term outcomes, with accuracy trailing off at the three- and six-month marks.²⁴ These are patients who don’t qualify for corticosteroids under the standard algorithm but who still have significant liver disease and real mortality risk.

For this group, treatment is largely supportive: ensuring adequate nutrition, managing complications like fluid retention or infection, and addressing alcohol use through counseling and structured support. The Maddrey score’s binary framing, severe versus not severe, can sometimes create a false sense of reassurance for patients and families when a score lands at, say, 28. The underlying liver disease may still be life-threatening; it simply doesn’t meet the specific threshold where a short course of steroids has been shown to help.