Lille Model Calculator

Lille Model Calculator for Gastroenterology. The validated cutoff is 0.45. A score below 0.45 identifies responders, associated with roughly 85% 6-month survival in the original cohort, and steroids are continued to complete the course. A score of 0.45 or higher identifies non-responders, associated with a marked drop to about 25% 6-month survival; steroids should be discontinued because ongoing treatment offers no benefit and prolongs infection risk. Non-responder status prompts a shift toward alternative strategies, palliative decision-making, and early evaluation for transplant candidacy where programs permit. Some centers also use a lower threshold near 0.16 to distinguish complete responders, and combining Lille with a static MELD score refines individual survival estimates further.

How this calculator works

The Lille Model is a logistic-regression equation that grades how a patient with severe alcohol-associated hepatitis is responding to corticosteroids after one week of treatment. It combines six inputs measured around day 7 of steroids: age, serum albumin, prothrombin time (or INR), serum creatinine/presence of renal insufficiency, day-0 bilirubin, and the change in bilirubin between day 0 and day 7. The formula returns a probability between 0 and 1, where the fall (or rise) in bilirubin over the week is the dominant, dynamic term. Because it is a treatment-response score rather than a baseline severity score, it can only be computed after a steroid trial has begun.

When to use this calculator

Use it exclusively in patients with severe alcohol-associated hepatitis (Maddrey discriminant function >=32) who have already received about 7 days of corticosteroids, to decide whether to continue or stop steroids. It should not be applied at baseline, before steroids are started, or in patients with mild/moderate disease who do not meet the treatment threshold. It also does not apply to patients who were never started on steroids (for example those with active sepsis or GI bleeding at presentation), since the model was derived entirely in steroid-treated cohorts.

Inputs used

  • Age
  • Albumin
  • Creatinine or renal insufficiency
  • Prothrombin time
  • Baseline bilirubin
  • Day 7 bilirubin

Clinical interpretation

The validated cutoff is 0.45. A score below 0.45 identifies responders, associated with roughly 85% 6-month survival in the original cohort, and steroids are continued to complete the course. A score of 0.45 or higher identifies non-responders, associated with a marked drop to about 25% 6-month survival; steroids should be discontinued because ongoing treatment offers no benefit and prolongs infection risk. Non-responder status prompts a shift toward alternative strategies, palliative decision-making, and early evaluation for transplant candidacy where programs permit. Some centers also use a lower threshold near 0.16 to distinguish complete responders, and combining Lille with a static MELD score refines individual survival estimates further.

Worked example

A 52-year-old with day-0 bilirubin 320 umol/L falls to 250 umol/L by day 7, with albumin 26 g/L, prothrombin time modestly prolonged, and no renal insufficiency. Entering these values yields a Lille score of roughly 0.38. Because this is below 0.45, the patient is classified as a responder: steroids are continued to complete 28 days, and the predicted 6-month survival is favorable (on the order of 85%). Had bilirubin instead risen over the week, the score would climb above 0.45, flagging a non-responder for whom steroids should be stopped.

Limitations and safety notes

The score is only meaningful after about 7 days of steroids, so it cannot triage patients at admission or guide whether to start treatment. Accuracy depends on an accurate day-0 bilirubin, and rapidly changing renal function, concurrent sepsis, or transfusion-affected bilirubin can distort the day-7 value and mislabel response. The 0.45 cutoff still misclassifies a minority of patients, and the original model captured roughly 75% of observed deaths rather than all of them, so a low score does not guarantee survival. It was derived in predominantly European steroid-treated cohorts and does not incorporate infection, acute-on-chronic liver failure grade, or alcohol relapse.

Frequently asked questions

When should the Lille score be calculated?

After approximately 7 days of corticosteroid therapy in severe alcohol-associated hepatitis. It requires both a day-0 and a day-7 bilirubin to capture the change in bilirubin, so it cannot be computed at admission or before steroids begin.

What does a Lille score above 0.45 mean?

It defines a corticosteroid non-responder, with roughly 25% 6-month survival in the derivation cohort versus about 85% for responders. Steroids should be stopped, and the patient should be evaluated for alternative approaches including early transplant assessment where available.

How is the Lille Model different from the Maddrey discriminant function or MELD?

Maddrey and MELD are static baseline severity scores used to decide who warrants treatment. The Lille Model is dynamic: it measures actual response after a week of steroids, and its bilirubin-change term drives the prediction. Combining Lille with MELD predicts survival better than either alone.

Should steroids always be stopped if the Lille score is high?

A score of 0.45 or higher indicates continued steroids provide no survival benefit and add infection risk, so most guidelines support discontinuation. The decision should still weigh infection, renal function, and transplant eligibility rather than the number in isolation.

Can the Lille Model be used in patients not on steroids?

No. It was derived and validated entirely in steroid-treated patients, and the change-in-bilirubin term is a response signal to that therapy. Applying it to untreated patients or to those given pentoxifylline instead is off-label and unvalidated.

References

  • Louvet A, Naveau S, Abdelnour M, et al. The Lille model: a new tool for therapeutic strategy in patients with severe alcoholic hepatitis treated with steroids. Hepatology. 2007. PMID: 17518367.
  • Louvet A, Labreuche J, Artru F, et al. Combining Data From Liver Disease Scoring Systems Better Predicts Outcomes of Patients With Alcoholic Hepatitis. Gastroenterology. 2015. PMID: 25935634.

Editorial review and citation methodology

Reviewed by the Quick Medical Calculator Editorial Team. Last reviewed: April 26, 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