Use the Lille model after approximately 7 days of corticosteroid therapy in severe alcohol-associated hepatitis to estimate steroid response and short-term prognosis.

Domain	Evidence summary
Derivation	Derived in 320 adults with severe alcoholic hepatitis treated with prednisolone between 1990 and 2003.
Validation	Externally validated in 118 similar patients treated with corticosteroids.
Outcome	Predicts 6-month mortality and identifies non-response to corticosteroids.
Discrimination	Reported AUROC was approximately 0.89 in derivation and 0.85 in validation cohorts.
Key cutoff	A score of 0.45 or higher is commonly used to identify poor steroid response.
Response bands	Complete response 0.16 or lower, partial response 0.16-0.56, and null response 0.56 or higher. Reported 28-day survival was about 91%, 79%, and 53% respectively.

Apply only after corticosteroids have been started for severe alcohol-associated hepatitis. It is not an admission diagnostic score.
The classic model uses bilirubin at baseline and day 7. earlier day-4 adaptations exist but are separate models.
A high Lille score supports stopping ineffective corticosteroid therapy, particularly when infection, GI bleeding, renal failure or other steroid harms are present.
The model supports prognosis and steroid-response assessment, but does not replace transplant, infection or critical-care assessment.

Day-7 Lille score for severe alcoholic hepatitis