About the APASL ACLF Research Consortium (AARC) Score
APASL and EASL-CLIF do not mean the same thing by acute-on-chronic liver failure, and the AARC score belongs to the APASL definition — which requires an acute hepatic insult causing jaundice and coagulopathy in chronic liver disease and excludes patients who have already decompensated. Within that population it scores five variables from one to three points each — bilirubin, INR, creatinine, lactate and hepatic encephalopathy grade — giving a total from 5 to 15. Grade I is 5–7 with 12.7% 28-day mortality, grade II is 8–10 with 44.5%, and grade III is 11–15 with 85.9%. It achieved an AUROC of 0.80 in derivation and 0.78 in validation across 1,402 patients.
Formula
AARC = bilirubin + INR + creatinine + lactate + HE grade (each 1–3 points, total 5–15)- bilirubin
- 1 point below 15 mg/dL, 2 points at 15–25, 3 points above 25 mg/dL.
- INR
- 1 point below 1.8, 2 points at 1.8–2.5, 3 points above 2.5.
- creatinine
- 1 point below 0.7 mg/dL, 2 points at 0.7–1.5, 3 points above 1.5 mg/dL.
- lactate
- 1 point below 1.5 mmol/L, 2 points at 1.5–2.5, 3 points above 2.5 mmol/L.
- HE grade
- 1 point for West Haven grade 0, 2 points for grades I–II, 3 points for grades III–IV.
- The floor is 5, not 0. Every variable contributes at least one point, so a score of 5 means all five values fall in the most favourable band rather than that nothing is wrong.
- The grade boundaries are 7 and 10. Grade I is 5–7, grade II is 8–10, grade III is 11–15.
- Bilirubin and INR bands assume the patient already meets the APASL entry criteria of bilirubin ≥ 5 mg/dL and INR ≥ 1.5. Applying the score outside that population puts almost everyone in the one-point band and destroys its discrimination.
- The creatinine bands are unusually low, with two points awarded from 0.7 mg/dL. This is not an error — it reflects that creatinine overestimates renal function in patients with reduced muscle mass, which is most of this population.
- No weighting between variables. Lactate and bilirubin contribute equally despite measuring quite different things.
Interpreting the result
Grade I, at 5 to 7 points, carries a 12.7% 28-day mortality and supports careful monitoring with supportive management while the precipitant is treated — but the emphasis belongs on 'monitoring', because this score is explicitly dynamic and a patient can move between grades within days. Grade II, at 8 to 10, carries 44.5% and is the point at which transplant assessment should begin rather than be contemplated; waiting for further deterioration in a syndrome with this mortality rate is how the window is lost. Grade III, at 11 to 15, carries 85.9% 28-day mortality, and at that level transplantation is the only intervention with a realistic prospect of altering the outcome — where it is not available or the patient is not a candidate, palliative input belongs alongside active treatment rather than after it. Read the score in the context of the framework it belongs to: these figures come from a population defined by APASL criteria, in which hepatitis B reactivation and acute hepatitis on chronic disease predominate, and they should not be quoted for a European patient with prior decompensation and renal failure.
| Score | Band | What it means | Action |
|---|---|---|---|
| 5–7 | AARC Grade I | 28-day mortality 12.7% | Supportive management and treatment of the precipitant; reassess daily, since the grade is dynamic and a rising score changes the plan |
| 8–10 | AARC Grade II | 28-day mortality 44.5% | Intensive monitoring with hepatology input; begin transplant assessment now rather than on further deterioration |
| 11–15 | AARC Grade III | 28-day mortality 85.9% | Urgent intensive care and transplant evaluation; where transplantation is not an option, involve palliative care alongside active treatment |
Scroll the table sideways for every column.
What the AARC-ACLF needs (5 inputs)
- Total bilirubin
- One point below 15 mg/dL, two points from 15 to 25, three points above 25. The thresholds sit far higher than in most liver scores, because the APASL definition already requires a bilirubin of at least 5 mg/dL for entry — everyone being scored is jaundiced.
- INR
- One point below 1.8, two points from 1.8 to 2.5, three points above 2.5. Again a high floor, since coagulopathy with an INR of at least 1.5 is part of the entry definition.
- Serum creatinine
- One point below 0.7 mg/dL, two points from 0.7 to 1.5, three points above 1.5. Note how low these bands are — two points is awarded at a creatinine most clinicians would read as normal, which reflects how much renal function matters here and how misleading a normal-looking creatinine is in a patient with low muscle mass.
- Serum lactate
- One point below 1.5 mmol/L, two points from 1.5 to 2.5, three points above 2.5. The variable that distinguishes this score from the EASL-CLIF instruments, included because lactate reflects both tissue hypoperfusion and the failure of hepatic clearance.
- Hepatic encephalopathy grade
- One point for grade 0, two points for West Haven grades I–II, three points for grades III–IV. Banded rather than used directly, which makes it more forgiving of the grading imprecision that affects West Haven at the covert end.
Units. Bilirubin and creatinine may be entered in conventional or SI units and are converted internally; lactate is in mmol/L, which is near-universal. The published thresholds are in mg/dL, and the bilirubin bands in particular are high — 15 mg/dL is 257 µmol/L and 25 mg/dL is 428 µmol/L. The creatinine bands are correspondingly low: 0.7 mg/dL is 62 µmol/L and 1.5 mg/dL is 133 µmol/L. Both sets are easy to misread against a chart in the other unit system, and because the bands are only three wide, a conversion error usually shifts the grade.
What it returns
- AARC score (5–15)
- The sum of five variables each scoring 1 to 3. The minimum possible score is 5, not 0 — a patient with entirely normal values in every band still scores 5, which surprises people expecting a zero floor.
- AARC grade (I, II or III)
- Grade I is 5–7, grade II is 8–10, grade III is 11–15. The grade is what the published mortality figures attach to.
- 28-day mortality
- 12.7% at grade I, 44.5% at grade II, 85.9% at grade III, from the derivation and validation cohorts.
How it is calculated
The score was derived within the APASL ACLF Research Consortium, a multinational Asian registry, in a total cohort of 1,402 patients with acute-on-chronic liver failure as APASL defines it — 480 used for derivation and 922 for validation. Five variables emerged as independently predictive and were banded into three levels each rather than kept continuous, which trades some discrimination for a score that can be computed at the bedside without a calculator. The paper's title states its central finding: liver failure determines the outcome. That framing is the conceptual difference from the European work, where the syndrome is defined by the accumulation of extrahepatic organ failures. APASL treats the liver insult as the driver and the other variables as measures of its consequences, which is why bilirubin and INR carry their own bands here rather than being two among six organ systems.
Facts & figures
| Variable | 1 point | 2 points | 3 points |
|---|---|---|---|
| Total bilirubin (mg/dL) | < 15 | 15–25 | > 25 |
| INR | < 1.8 | 1.8–2.5 | > 2.5 |
| Creatinine (mg/dL) | < 0.7 | 0.7–1.5 | > 1.5 |
| Lactate (mmol/L) | < 1.5 | 1.5–2.5 | > 2.5 |
| Hepatic encephalopathy | Grade 0 | Grades I–II | Grades III–IV |
Scroll the table sideways for every column.
Every variable contributes at least one point, which is why the minimum total is 5. The creatinine bands are far lower than in MELD or CLIF-SOFA, and the bilirubin bands far higher — both follow from the APASL entry criteria, which pre-select a jaundiced, coagulopathic population.
| Element | APASL (AARC) | EASL-CLIF (CANONIC) |
|---|---|---|
| Starting point | Acute hepatic insult in chronic liver disease | Acute decompensation of cirrhosis |
| Entry criteria | Jaundice (bilirubin ≥ 5 mg/dL) and coagulopathy (INR ≥ 1.5), with ascites and/or encephalopathy within 4 weeks | Organ failure by CLIF-SOFA with 28-day mortality above 15% |
| Prior decompensation | Excluded — the patient must not have decompensated before | Included |
| Underlying disease | Chronic liver disease, cirrhotic or not; may be previously undiagnosed | Established cirrhosis |
| What defines severity | Liver failure and its consequences, across five variables | The number of failing organ systems, across six |
Scroll the table sideways for every column.
These are not two scores for one syndrome — they define overlapping but genuinely different populations. A patient with long-standing ascites presenting in renal failure has ACLF by EASL-CLIF and falls outside APASL entirely; a previously undiagnosed hepatitis B patient presenting jaundiced and coagulopathic is squarely APASL.
Evidence
Derivation — APASL ACLF Research Consortium
2017 · n = 480Patients with acute-on-chronic liver failure as defined by APASL criteria, enrolled in the multinational APASL ACLF Research Consortium registry. Of 1,402 patients in total, 480 formed the derivation cohort and 922 the validation cohort.
AUROC 0.80 in the derivation cohort for predicting mortality, and the score was reported as superior to both MELD and CLIF-SOFA in this population. By grade, 28-day mortality was 12.7% at grade I, 44.5% at grade II and 85.9% at grade III.
Internal validation cohort
2017 · n = 922The larger of the two AARC cohorts, used to confirm the grading and its association with 28-day mortality.
AUROC 0.78, close to the derivation figure, supporting the stability of the five-variable model and its grade boundaries.
Framework consensus — APASL 2014 and 2019
2019APASL consensus recommendations on acute-on-chronic liver failure, which define the syndrome the score applies to and were updated in 2019 to incorporate the AARC work.
Established and then refined the APASL definition — acute hepatic insult with jaundice and coagulopathy in chronic liver disease, excluding prior decompensation — and adopted the AARC score and grading for severity assessment.
How it compares
AARC-ACLF vs CLIF-SOFA / EASL-CLIF criteria
They define different populations rather than competing on one — APASL excludes previously decompensated patients and EASL-CLIF includes them, so a patient can meet one definition and not the other.
This distinction is the single most important thing to understand about either score. APASL requires an acute hepatic insult producing jaundice and coagulopathy in chronic liver disease with no prior decompensation; EASL-CLIF starts from acute decompensation of cirrhosis and defines the syndrome by the number of failing organs and the associated 28-day mortality. In the AARC cohort of 1,402 patients the AARC score achieved an AUROC of 0.80 and was reported superior to CLIF-SOFA, which is a valid finding about an APASL-defined population and not a general claim — CLIF-SOFA was never intended for that group. Use the framework that matches your patient, apply that framework's score, and document which one you used.
AARC-ACLF vs MELD-Na
AARC outperformed MELD in the APASL cohort, largely because MELD has no representation of encephalopathy or lactate — but MELD-Na remains the allocation instrument and AARC is not.
MELD and MELD-Na were fitted to three-month waiting-list mortality using bilirubin, INR, creatinine and sodium. They contain nothing for cerebral function and nothing for tissue perfusion, and in a syndrome where encephalopathy and lactate carry independent prognostic weight that omission costs discrimination — which is what the AARC comparison showed. The division of labour is clear enough in practice: calculate MELD 3.0 or MELD-Na because that is what governs transplant priority, and calculate AARC because that is what tells you how urgent the situation is and whether the assessment needs to happen this week.
AARC-ACLF vs Child-Pugh score
Child-Pugh describes chronic hepatic function and can be near-normal in a patient with grade III AARC — it is the baseline, not the acute assessment.
The APASL population includes patients whose chronic liver disease was compensated or entirely undiagnosed before the acute insult, which means Child-Pugh may have been grade A weeks earlier and carries no information about the current emergency. Child-Pugh also has no place for creatinine or lactate, the two variables that most reflect the systemic consequences of acute liver failure. Its value here is retrospective: knowing that the patient was Child-Pugh A before the insult supports the APASL framing that this is a potentially reversible acute event rather than the end of a chronic decline, which is precisely the distinction that makes transplantation worth pursuing urgently.
Pearls & pitfalls
- The minimum score is 5, not 0. Every variable contributes at least a point, so a score of 5 means all five are in the best band — it does not mean no derangement.
- The score only applies within the APASL definition. Using it in a patient with established prior decompensation applies it outside the population it was derived in, and the mortality figures do not transfer.
- The creatinine bands are much lower than you expect: two points from 0.7 mg/dL. That is deliberate, because creatinine overestimates renal function where muscle mass is reduced, which describes most of this population.
- The bilirubin bands are much higher than you expect: one point up to 15 mg/dL. Applying the score to a patient with a bilirubin of 3 places them in the lowest band along with everyone else, and the score stops discriminating.
- Lactate is not optional and is not in any EASL-CLIF instrument. Omitting it or substituting a guess removes the variable that most distinguishes this score.
- Encephalopathy is banded here rather than used directly as a subscore, unlike CLIF-SOFA. Grades I and II score the same two points, which makes the score less sensitive to imprecision at the covert end — a genuine advantage.
- Regrade daily. The consortium presents this as a dynamic score, and a patient can move between grades within days in either direction.
- Do not compare an AARC grade with a CLIF-SOFA ACLF grade as though they were the same scale. Grade I here is not grade 1 there, the populations differ, and the mortality figures are not commensurable.
- AUROC 0.80 in derivation and 0.78 in validation is good discrimination, not near-certainty. A grade III patient can survive and a grade I patient can die.
Critical actions
- Confirm the patient meets the APASL definition before applying the score: an acute hepatic insult with bilirubin of at least 5 mg/dL and INR of at least 1.5, ascites and/or encephalopathy within four weeks, and no prior decompensation.
- Measure lactate specifically — it is not part of a routine liver panel and the score cannot be completed without it.
- Identify the precipitant actively: hepatitis B reactivation, superimposed hepatitis A or E, alcohol, drug or herbal injury, and sepsis. In this population the insult is frequently identifiable and often treatable.
- Start antiviral therapy immediately where hepatitis B reactivation is the precipitant, without waiting for the grade to worsen.
- Begin transplant assessment at grade II rather than grade III — the assessment takes time that a 44.5% 28-day mortality does not reliably allow.
- Recalculate daily and record the trend, since the grade is explicitly dynamic and the trajectory informs the transplant decision more than any single value.
- State in the notes which framework you are applying, because a patient described simply as 'ACLF grade II' is ambiguous between two systems with different definitions and different mortality figures.
- Interpret creatinine alongside muscle mass, and consider cystatin C where available, since the low bands make small differences consequential.
Why this score exists
The consortium's argument is in the title of the paper: liver failure determines the outcome. That is a direct challenge to the European framing, in which acute-on-chronic liver failure is characterised by the accumulation of extrahepatic organ failures and the liver is one of six systems. APASL's position is that in their population the hepatic insult is the driver and the other derangements are its consequences, and the population difference is real rather than rhetorical — hepatitis B reactivation and acute hepatitis A or E superimposed on chronic disease are common presentations across Asia and uncommon in the European cohorts, and they produce a patient who is profoundly jaundiced and coagulopathic without necessarily having multiple organ failures. Excluding previously decompensated patients follows from the same logic: if the syndrome is a potentially reversible acute liver injury, a patient already in the chronic decline of decompensated cirrhosis is a different clinical problem. The inclusion of lactate reflects the same emphasis, since it measures the systemic consequence of failing hepatic function rather than the failure of another organ.
About the creator
First author, AARC score derivation
First author of the 2017 study that derived and validated the AARC score and compared it against CLIF-SOFA and MELD.
Senior author; lead author of the APASL ACLF consensus recommendations
Principal architect of the APASL definition of acute-on-chronic liver failure and senior author of both the 2014 and 2019 consensus documents.
Co-author, AARC derivation and validation
Co-authored the derivation study and much of the consortium's subsequent work on organ failure in this population.
Limitations
- Derived and validated entirely within Asian cohorts where hepatitis B reactivation and acute hepatitis A or E predominate; performance in alcohol-related or metabolic liver disease populations is less well established.
- Applies only within the APASL definition, which excludes previously decompensated patients — a substantial proportion of patients in Western practice fall outside it.
- Requires lactate, which is not part of a routine liver panel and may not be available in resource-limited settings where the syndrome is common.
- The bilirubin and INR bands assume a pre-selected jaundiced, coagulopathic population, so the score loses discrimination if applied more broadly.
- Banding all five variables into three levels discards information a continuous model would retain, in exchange for bedside computability.
- No weighting between variables, so lactate and bilirubin contribute equally despite measuring different things and having different reversibility.
- Creatinine is a poor measure of renal function in this population, and the low bands make that imprecision consequential.
- AUROC of 0.80 and 0.78 represents good but not excellent discrimination, and the grade mortality figures are group averages that should not be read as individual prognosis.
- Not an allocation instrument, and no jurisdiction uses it to rank transplant priority.
If you are the patient
The AARC score measures how severe liver failure is in someone who has chronic liver disease and has then had a sudden injury to the liver — for example a flare of hepatitis B, a new infection with another hepatitis virus, a heavy period of drinking, or a reaction to a medication or herbal remedy. Five things are measured: bilirubin and clotting, which show how the liver is functioning; creatinine, which shows how the kidneys are coping; lactate, which shows whether the tissues are getting enough oxygen; and a grading of confusion. Each scores between 1 and 3 points, so the total runs from 5 to 15. Totals of 5 to 7 are grade I, 8 to 10 are grade II, and 11 to 15 are grade III, with the risk over the following month rising substantially across those grades. Two points matter more than the number itself. The score is meant to be repeated each day rather than taken as a one-off, because it can improve as well as worsen — and where the trigger can be treated, such as starting antiviral tablets for a hepatitis B flare, it often does. And this is the point at which the team will start looking into whether a liver transplant is possible, and they will do so early rather than waiting, because arranging it takes time. There is a related score used more in Europe called CLIF-SOFA, which is designed for a slightly different group of patients, so the two are not directly comparable.
Frequently asked questions
What is the AARC score?#
A five-variable score for grading acute-on-chronic liver failure as defined by APASL. Bilirubin, INR, creatinine, lactate and hepatic encephalopathy grade each score 1 to 3 points, giving a total from 5 to 15. Grade I is 5–7, grade II is 8–10 and grade III is 11–15, with 28-day mortality of 12.7%, 44.5% and 85.9% respectively.
What are the AARC grade cut-offs and mortality?#
Grade I is 5 to 7 points, with a 28-day mortality of 12.7%. Grade II is 8 to 10 points, with 44.5%. Grade III is 11 to 15 points, with 85.9%. These figures come from the APASL ACLF Research Consortium cohort of 1,402 patients, in which the score achieved an AUROC of 0.80 in derivation and 0.78 in validation.
How does the APASL definition of ACLF differ from EASL-CLIF?#
APASL requires an acute hepatic insult in a patient with chronic liver disease, producing jaundice with a bilirubin of at least 5 mg/dL and coagulopathy with an INR of at least 1.5, plus ascites and/or encephalopathy within four weeks — and it excludes patients who have decompensated previously. EASL-CLIF starts from acute decompensation of cirrhosis and defines the syndrome by the number of failing organ systems. A patient with long-standing ascites presenting in renal failure has ACLF by EASL-CLIF and falls outside APASL entirely.
Why is the minimum AARC score 5 rather than 0?#
Because each of the five variables contributes at least one point, even when the value falls in the most favourable band. A score of 5 therefore means all five are in their best band — it does not mean there is no derangement, and it does not mean the patient does not have ACLF, since meeting the APASL entry criteria is a precondition for scoring at all.
Is AARC better than CLIF-SOFA?#
In the APASL population, the AARC score outperformed both CLIF-SOFA and MELD with an AUROC of 0.80. But that comparison was made in patients defined by APASL criteria, which is not the population CLIF-SOFA was derived for, so it is not evidence that one score is generally superior. The two frameworks define different syndromes, and the right choice depends on which definition your patient meets.
Why does the AARC score include lactate?#
Because lactate reflects both tissue hypoperfusion and the failure of hepatic clearance, and it carried independent prognostic weight in the derivation cohort. It is the variable that most distinguishes AARC from the EASL-CLIF instruments, none of which include it. Practically, it means the score cannot be completed from a routine liver panel — lactate has to be requested specifically.
Why are the creatinine bands so low in the AARC score?#
Two points are awarded from 0.7 mg/dL, which most clinicians would read as normal. That is deliberate: creatinine systematically overestimates renal function in patients with reduced muscle mass, which describes most patients with chronic liver disease, so a value in the apparently normal range already represents meaningful impairment. It also means small measurement differences can shift the grade, and cystatin C is worth considering where available.
Should the AARC score be repeated?#
Yes — the consortium presents it as a dynamic score. Patients move between grades within days in both directions, particularly where the precipitant is treatable, as with antiviral therapy for hepatitis B reactivation. A single admission score is a starting point; the trajectory is what should inform the transplant decision.
Can the AARC score be used for transplant allocation?#
No. Allocation runs on MELD 3.0 or an equivalent national model, and AARC was neither derived nor validated for ranking waiting-list priority. What the grade should drive is timing: assessment belongs at grade II rather than grade III, because a 44.5% 28-day mortality does not reliably leave time to arrange it later.
References
Consensus recommendations
- Sarin SK, Choudhury A, Sharma MK, et al. Acute-on-chronic liver failure: consensus recommendations of the Asian Pacific Association for the Study of the Liver (APASL): an update. Hepatol Int. 2019;13(4):353-390.
- Sarin SK, Kedarisetty CK, Abbas Z, et al. Acute-on-chronic liver failure: consensus recommendations of the Asian Pacific Association for the Study of the Liver (APASL) 2014. Hepatol Int. 2014;8(4):453-471.