About the Glasgow Alcoholic Hepatitis Score
GAHS runs from 5 to 12 and is assembled from age, white cell count, blood urea, prothrombin time ratio and bilirubin. Its value lies in one cutoff: at 9 or above, corticosteroids improved 28-day survival to 78% from 52% untreated, and 84-day survival to 59% from 38%, while patients scoring below 9 showed no survival benefit from steroids at all — even though every patient in that validation cohort already met Maddrey's severity threshold. On the same derivation patients, GAHS classified 28-day outcome correctly 81% of the time against 49% for Maddrey's discriminant function.
Formula
GAHS = age points + white cell count points + urea points + PT ratio points + bilirubin points- age points
- 1 or 2.
- white cell count points
- 1 or 2.
- urea points
- 1 or 2, scored in mmol/L.
- PT ratio points
- 1, 2 or 3.
- bilirubin points
- 1, 2 or 3, scored in µmol/L.
- Minimum possible score is 5 (all variables at their lowest band), maximum is 12.
- The original derivation paper found GAHS equally accurate whether PT ratio or the international normalised ratio was used for the coagulation term.
- The score can be recalculated using labs from day 6–9 of admission with similar accuracy to day 1, which is useful for tracking a patient whose baseline values were borderline.
Interpreting the result
A GAHS below 9 carries a better prognosis, and in the corticosteroid validation study this group showed no survival benefit from steroids at all — worth knowing before exposing a patient to their infection risk for no expected gain. A GAHS of 9 or above marks a genuinely poor prognosis if untreated, and this is precisely the group in whom the validation study found a clear, statistically significant survival benefit from corticosteroids at both 28 and 84 days. Because the validation cohort was itself restricted to patients already meeting Maddrey's DF ≥ 32, the practical read is: use DF to flag severe disease, then use GAHS to decide whether steroids are likely to be worth the risk within that severe group.
| Score | Band | What it means | Action |
|---|---|---|---|
| 5–8 | Better prognosis | No demonstrated survival benefit from corticosteroids in this band, even among DF ≥ 32 patients | Corticosteroids not indicated on the strength of this score; treat supportively |
| 9–12 | Poor prognosis | Untreated 28-day survival 52%, rising to 78% with corticosteroids (p=0.002); 84-day survival 38% untreated vs 59% treated (p=0.02) | Consider corticosteroids after excluding sepsis and GI bleeding; reassess with Lille score at day 7 |
Scroll the table sideways for every column.
What the GAHS needs (5 inputs)
- Age (years)
- 1 point under 50, 2 points at 50 or over.
- White cell count (×10⁹/L)
- 1 point below 15, 2 points at 15 or above.
- Blood urea
- Scored as urea in mmol/L: 1 point below 5, 2 points at 5 or above. Enter BUN in mg/dL and use the unit toggle if that is what your laboratory reports — BUN and urea are not numerically interchangeable, and mislabelling this field has been a real source of scoring errors on calculator pages.
- Prothrombin time ratio
- Patient PT divided by control PT. 1 point below 1.5, 2 points at 1.5–2.0, 3 points above 2.0.
- Total bilirubin
- Scored in µmol/L: 1 point below 125, 2 points at 125–250, 3 points above 250. Use the unit toggle if your result is in mg/dL.
What it returns
- GAHS (5–12)
- Sum of the five component scores. Can be calculated from day-1 admission values or recalculated on days 6–9, both of which were validated in the derivation study.
- Prognosis band
- Poor prognosis (≥ 9) or better prognosis (< 9), the same cutoff used in the corticosteroid-response validation study.
How it is calculated
GAHS was derived by stepwise logistic regression on a cohort of 241 patients with alcoholic hepatitis, testing which admission-day and day-6-9 variables independently predicted death at 28 and 84 days. Five variables survived — age, blood urea and bilirubin from day 1, and bilirubin, prothrombin time and white cell count from days 6–9 — and were converted into the integer point bands used here. Unlike Maddrey's discriminant function, which is built from only two variables (PT and bilirubin), GAHS folds in a marker of renal function (urea) and a marker of the systemic inflammatory response (white cell count), which is very likely why it discriminated outcome more accurately in the same cohort.
Facts & figures
| Score | Overall accuracy, 28-day outcome |
|---|---|
| GAHS (day 1) | 81% |
| Modified Maddrey's discriminant function | 49% |
Scroll the table sideways for every column.
Both scored on the same 241-patient cohort. Similar results held for day 6–9 values and for 84-day outcome.
Evidence
Derivation and internal validation — Forrest et al.
2005 · n = 241241 patients with alcoholic hepatitis, with clinical and laboratory data recorded on admission (day 1) and on days 6–9. Stepwise logistic regression identified five variables independently associated with 28- and 84-day outcome. The resulting score was then tested in a separate validation dataset of 195 patients.
GAHS on day 1 had 81% overall accuracy for 28-day outcome, versus 49% for the modified Maddrey's discriminant function scored on the same patients. Similar results were found using day 6–9 data and when predicting 84-day outcome, and accuracy was confirmed in the separate 195-patient validation dataset. The score performed equally well whether INR or PT ratio was used, and regardless of whether the alcoholic hepatitis diagnosis was biopsy-proven or clinical.
Corticosteroid-response validation — Forrest et al.
2007 · n = 225225 patients with Maddrey's discriminant function ≥ 32 (i.e., already meeting the conventional severe threshold) from five UK hospital centres, reviewed for survival relative to GAHS band and corticosteroid use. 144 of 225 (64%) also had a GAHS of 9 or above.
Among patients with GAHS ≥ 9, 28-day survival was 78% with corticosteroids versus 52% without (p=0.002), and 84-day survival was 59% versus 38% (p=0.02). Among patients with GAHS below 9, there was no significant survival difference between treated and untreated patients at either timepoint.
How it compares
GAHS vs Maddrey's Discriminant Function
Use Maddrey's DF as the entry threshold for severe disease, and GAHS to decide whether steroids are likely to help within that group — GAHS was built specifically to correct the older score's blind spots.
Maddrey's DF uses only PT and bilirubin and, in the GAHS derivation cohort, achieved only 49% overall accuracy for 28-day outcome versus GAHS's 81% on the same patients. More importantly for treatment decisions, a separate study of 225 patients who all met Maddrey's DF ≥ 32 found the corticosteroid survival benefit was confined to those who also had a GAHS ≥ 9 — patients below that GAHS cutoff showed no benefit despite meeting the older score's severity threshold. The two scores are complementary rather than competing: DF for the conventional severity gate, GAHS for the treatment-benefit question.
GAHS vs Lille model
They answer sequential questions rather than competing — GAHS helps decide whether to start corticosteroids, and the Lille model decides at day 7 whether to continue them.
GAHS is calculated at presentation and identifies patients with severe alcohol-related hepatitis likely to benefit from corticosteroids, particularly in combination with a Maddrey discriminant function of 32 or more. The Lille model is calculated after seven days of treatment and quantifies response, with a score of 0.45 or above defining non-response and prompting withdrawal of steroids to avoid continued immunosuppression without benefit. Used together they frame the decision as start-then-review rather than a single irreversible call, which matches how the treatment is actually given.
Pearls & pitfalls
- Urea, not BUN, is what the point bands are written against (mmol/L). If your lab reports BUN in mg/dL, convert or use the unit toggle — treating the two as numerically interchangeable understates the score.
- GAHS < 9 showed no corticosteroid survival benefit in the validation cohort, even though every patient in that cohort already had a Maddrey's DF ≥ 32. A high DF alone should not be read as sufficient reason to treat.
- The corticosteroid-response evidence comes from patients who were already DF ≥ 32; GAHS was not independently validated for treatment decisions in patients below that DF threshold.
- The score can be recalculated on days 6–9 using the same accuracy as day 1 — useful if the admission values were borderline or incomplete.
- GAHS performed equally well with either PT ratio or INR for the coagulation term, so either is acceptable, but be consistent about which one is entered.
Critical actions
- Calculate GAHS in any patient already flagged severe by Maddrey's DF (≥ 32), specifically to inform the steroid decision.
- Exclude sepsis and gastrointestinal bleeding before starting corticosteroids, regardless of GAHS band.
- At GAHS ≥ 9, discuss corticosteroids using the concrete survival figures from the validation study (78% vs 52% at 28 days) rather than a vague statement of benefit.
- At GAHS < 9, do not treat a high Maddrey's DF alone as sufficient justification for steroids — the validation data found no benefit in this group.
- Plan a day-7 Lille assessment for any patient started on corticosteroids, to decide on continuation.
Why this score exists
The Glasgow group built GAHS explicitly as a response to a shortcoming they identified in Maddrey's discriminant function: that it uses only two laboratory variables and, in their own re-analysis, correctly predicted 28-day outcome only about half the time. Rather than proposing another severity score in the abstract, their 2007 follow-up study went further and asked the more useful clinical question directly — not just who is sicker, but who actually benefits from steroids — and found that the GAHS ≥ 9 versus < 9 split answered it with results that reached statistical significance in a multicentre population already selected as severe by the older score.
About the creator
First author, 2005 derivation study
Derived the Glasgow Alcoholic Hepatitis Score and showed that corticosteroid survival benefit was confined to patients scoring 9 or above. A different Forrest from the endoscopic classification.
Limitations
- Derived and validated entirely in UK cohorts; performance in other populations and healthcare settings has not been established to the same standard.
- The corticosteroid-benefit evidence (GAHS ≥ 9 vs < 9) comes from a cohort restricted to patients already meeting Maddrey's DF ≥ 32, so its treatment-decision use outside that severity band is less well supported.
- Requires white cell count and urea in addition to the variables Maddrey's DF needs, which is a small extra data burden in settings where those results are not already to hand.
- Like other alcoholic hepatitis scores, it assumes the diagnosis has already been made and does not itself distinguish alcoholic hepatitis from other causes of jaundice and coagulopathy.
- Has not been tested against the newer, larger STOPAH trial population in the same head-to-head way Maddrey's DF has.
If you are the patient
The Glasgow Alcoholic Hepatitis Score is a number from 5 to 12 that your medical team works out from your age, a white blood cell count, a kidney-function blood test (urea), how well your blood clots, and your bilirubin level. A score of 9 or higher means a more serious illness, but it also carries genuinely good news if steroid treatment is being considered: in a study of patients who were already being considered for steroids, those with a score of 9 or above lived longer with steroid treatment than without it — 78 in 100 were alive at 28 days with steroids, compared with 52 in 100 without. For people whose score was below 9, steroids made no measurable difference to survival in that same study, which is why your team may recommend against steroids even if another test suggested severe disease — the Glasgow score is specifically trying to identify who steroids will actually help.
Frequently asked questions
What is the Glasgow Alcoholic Hepatitis Score?#
A score from 5 to 12 for alcoholic hepatitis, calculated from age, white cell count, blood urea, prothrombin time ratio and bilirubin. A score of 9 or above indicates a poor prognosis and identifies patients likely to benefit from corticosteroids.
What GAHS score means you need steroids?#
A GAHS of 9 or above, specifically in patients who also meet Maddrey's discriminant function threshold of 32 or higher. In that combined group, a UK multicentre study found a clear survival benefit from corticosteroids; below GAHS 9, no benefit was found even when the discriminant function was also severe.
Is GAHS more accurate than Maddrey's discriminant function?#
In its derivation cohort, yes — GAHS had 81% overall accuracy for predicting 28-day outcome, compared with 49% for the modified Maddrey's discriminant function scored on the same 241 patients.
Do I need urea or BUN for the GAHS?#
Urea, scored in mmol/L. If your laboratory reports blood urea nitrogen (BUN) in mg/dL, convert it or use the unit toggle — urea and BUN are related but not numerically interchangeable, and using the wrong one under-scores the point band.
Does GAHS below 9 mean steroids will not help?#
In the validation study of 225 patients who all had a Maddrey's DF of 32 or above, patients with a GAHS below 9 showed no significant survival difference between corticosteroid-treated and untreated groups, at either 28 or 84 days.
What variables make up the Glasgow Alcoholic Hepatitis Score?#
Five, each scored 1 to 3: age, white cell count, blood urea, prothrombin time ratio and bilirubin. The total therefore runs from 5 to 12, not from 0 — a patient with every variable in the most favourable band still scores 5, which surprises people expecting a zero floor.
When should the GAHS be calculated?#
On day 1, and again on day 7. The score was designed to be dynamic, and the day-7 value carries information the admission value does not, in the same way the Lille model formalises steroid response at day 7. A single admission score treats a syndrome that evolves over the first week as though it were fixed.
Do I enter urea or BUN, and does it matter?#
It matters a great deal, because the two differ by a factor of roughly 2.14. The published GAHS thresholds are in mmol/L of urea, while US laboratories report blood urea nitrogen in mg/dL. This calculator accepts BUN in mg/dL and converts internally, but a hand-calculation that puts a BUN value against a urea threshold will misclassify the patient. If a score computed elsewhere looks implausibly low, unit confusion is the first thing to check.
References
Original / primary reference
Validation and treatment-response study
Treatment-response evidence and guidelines
- Thursz MR, Richardson P, Allison M, et al. Prednisolone or pentoxifylline for alcoholic hepatitis (STOPAH). N Engl J Med. 2015;372(17):1619-1628.
- European Association for the Study of the Liver. EASL Clinical Practice Guidelines: Management of alcohol-related liver disease. J Hepatol. 2018;69(1):154-181.
- Crabb DW, Im GY, Szabo G, Mellinger JL, Lucey MR. Diagnosis and Treatment of Alcohol-Associated Liver Diseases: 2019 Practice Guidance from the AASLD. Hepatology. 2020;71(1):306-333.