How to use it
Day 0. Enter bilirubin, the patient’s PT and your lab’s control PT, INR and creatinine. The tool gives the Maddrey discriminant function (DF) and MELD. If the hepatitis is severe, five safety questions appear. A green recommendation only shows when all are answered and none is a contraindication.
Day 4 or 7. Switch to the Lille tab. Day-0 values carry over (check them) and you add the age and the day-4 or day-7 bilirubin.
Reading the result
| Finding | Meaning |
|---|---|
| MELD ≤20 and DF <32 | Not severe; steroids not indicated |
| MELD >20 | Severe AH: prednisolone 40 mg daily for 28 days if no contraindication (ACG 2024) |
| MELD 25–39 | Greatest expected benefit |
| MELD >50 | Benefit uncertain: still seen at 51, none above 51 (Arab 2021). Weigh risks and benefits carefully (ACG 2024); discuss transplant |
| DF ≥32 with MELD ≤20 | Severe by DF only; steroid benefit has only been shown above MELD 20 |
| Lille ≥0.45 (day 4 or 7) | Non-response: stop steroids |
Which MELD? The steroid evidence uses the original MELD (bilirubin, INR, creatinine), shown here without the cap of 40 used for transplant allocation, because the studies report MELD up to and above 50. MELD-Na and MELD 3.0 are shown capped at 40, as in allocation, with the uncapped value when higher. They do not change the steroid decision, but MELD 3.0 matters for transplant: Ontario uses it for liver transplant referral, listing and allocation (Ontario Health 2025–2026). For the full MELD 3.0, MELD-Na and Child-Pugh calculation, 90-day survival and Ontario referral thresholds, use the MELD 3.0 and Child-Pugh tool.
Dialysis means two or more dialysis sessions, or 24 hours of continuous renal replacement, in the past week. Creatinine then counts as 4.0 mg/dL for MELD and MELD-Na. MELD 3.0 caps creatinine at 3.0 mg/dL, with or without dialysis.
Guidelines and thresholds. ACG 2024 defines severe AH by MELD above 20. The DF (≥32) is older; it defined severe disease in the original steroid trials and in the Lille cohort. We found no current Canadian guideline on alcohol-associated hepatitis.
Evidence since ACG 2024
None of these has changed a guideline yet; the tool still follows ACG 2024 (prednisolone 40 mg daily for 28 days).
- Tapered prednisolone (STASH, 2025). 40 mg tapered by 10 mg a week over 4 weeks cut infections by day 90 (19.7% vs 33.1%) with no difference in mortality. One trial, 98% men, mostly from India.
- Early spontaneous improvement (CORTISAVE, 2026). If bilirubin falls more than 10% by day 5–10 without treatment, methylprednisolone did not improve 90-day survival (79% vs 83% on placebo). The trial stopped at half its planned size, so a benefit is not excluded.
- Larsucosterol (AHFIRM, 2025) missed its primary end point (90-day death or transplant). G-CSF added to prednisolone improved 90-day survival in a small trial from India (GPreAH 2025: 88% vs 64%, 42 per group); it is not standard care and is not in this tool.
Pitfalls
- The control PT matters. Each second changes the DF by 4.6. Never assume 12 s.
- Rule out infection first, and keep looking. In STOPAH, serious infections were almost twice as common on prednisolone (13% vs 7%).
- Contraindications are often temporary. ACG 2024 says steroids can be started once infection, bleeding or kidney injury is controlled. Recalculate then.
- Prednisolone, not pentoxifylline. ACG 2024 recommends against pentoxifylline and recommends IV N-acetylcysteine with steroids.
- Lille uses PT in seconds (not INR) and day-0 albumin in g/L. Renal insufficiency means creatinine above 1.3 mg/dL (115 µmol/L).
- The cut-off is 0.45. Louvet described non-responders as scoring above 0.45; this tool treats exactly 0.45 as non-response, the cautious choice.