About the Revised Atlanta Classification of Acute Pancreatitis (2012)
Two questions decide everything here: is there organ failure, and has it lasted more than 48 hours. Mild acute pancreatitis has no organ failure and no local or systemic complications. Moderately severe has transient organ failure resolving within 48 hours, or local complications, or exacerbation of co-morbid disease. Severe is defined by persistent organ failure beyond 48 hours. Organ failure itself is a modified Marshall score of 2 or more in any one of three systems — respiratory, cardiovascular or renal — which means the classification cannot be finalised until that 48 hours has actually elapsed.
Interpreting the result
Mild disease is the most common form, resolves in the first week, and does not need contrast-enhanced CT for severity assessment — imaging early in mild pancreatitis is usually unhelpful and often misleading, since necrosis is not radiologically evident in the first 72 hours. Moderately severe covers two quite different routes to the same label: transient organ failure that resolves within 48 hours, and local complications without any organ failure. Both carry real morbidity and prolonged admission, and the second frequently needs intervention weeks later, but mortality is far below that of severe disease. Severe means persistent organ failure beyond 48 hours, and it is the category that carries the mortality — higher again when infected necrosis coexists. The practical trap is timing: organ failure present at 24 hours does not yet distinguish moderately severe from severe, and a classification recorded at that point should be flagged as provisional and revisited. Reassessing at 48 hours is not administrative tidiness; it is the moment the classification becomes meaningful.
| Score | Band | What it means | Action |
|---|---|---|---|
| No organ failure, no local complications | Mild acute pancreatitis | The most common form. Usually resolves within the first week; mortality is very low | Supportive care, early enteral feeding as tolerated, and an aetiology work-up. CT is not needed for severity assessment |
| Local complications, no organ failure | Moderately severe acute pancreatitis | Peripancreatic collection, necrosis, pseudocyst or walled-off necrosis without organ failure. Prolonged admission, often delayed intervention | Contrast-enhanced CT after 72–96 hours to characterise the collection; avoid early intervention on sterile collections |
| Transient organ failure (resolves within 48 h) | Moderately severe acute pancreatitis | Organ failure present but resolving within 48 hours, with substantially lower mortality than persistent failure | Monitor closely — transient can become persistent, which reclassifies the patient as severe |
| Persistent organ failure (> 48 h) | Severe acute pancreatitis | The category carrying the mortality, higher again where infected necrosis coexists | Critical care with organ support; CT after 72–96 hours; step-up drainage rather than early open necrosectomy if infection develops |
What the Revised Atlanta Classification needs (5 inputs)
- Modified Marshall — respiratory (PaO₂/FiO₂)
- Scored 0 above 400, 1 for 301–400, 2 for 201–300, 3 for 101–200 and 4 at 101 or below. A subscore of 2 constitutes organ failure.
- Modified Marshall — renal (serum creatinine)
- Scored 0 below 1.4 mg/dL, 1 for 1.4–1.8, 2 for 1.9–3.6, 3 for 3.6–4.9 and 4 above 4.9 mg/dL.
- Modified Marshall — cardiovascular (systolic blood pressure)
- Scored 0 above 90 mmHg, 1 below 90 but fluid responsive, 2 below 90 and not fluid responsive, 3 below 90 with pH under 7.3, and 4 below 90 with pH under 7.2. A patient requiring inotropes is by definition not fluid responsive.
- Duration of organ failure
- Whether organ failure has persisted beyond 48 hours. This single question separates moderately severe from severe, and it cannot be answered before 48 hours have passed — which is why an early classification is provisional.
- Local complications or exacerbation of co-morbid disease
- Acute peripancreatic fluid collection, pancreatic or peripancreatic necrosis (sterile or infected), pseudocyst, or walled-off necrosis. Their presence lifts a patient without organ failure from mild to moderately severe.
What it returns
- Severity category
- Mild, moderately severe or severe. Reported as provisional where organ failure is present but 48 hours have not yet elapsed.
- Modified Marshall subscores and which systems are failing
- Reported per system, because 'organ failure' without naming the organ loses the information that drives management.
How it is calculated
The revision was produced by web-based international consultation rather than a single meeting: a working group circulated a draft to eleven national and international pancreatic associations, revised it in response to comments, and repeated the consultation three times, retaining only statements supported by published evidence. That method explains the shape of the result. The classification is definitional rather than statistical — there are no coefficients and nothing is fitted — and its authority comes from consensus across the societies that would have to use it. Two structural ideas underpin it. The first is that acute pancreatitis has two phases, early and late, with different determinants of outcome: organ failure dominates the first week, local complications and their consequences the period after. The second is that severity should be defined by what actually kills patients, which is persistent organ failure, rather than by the presence of a radiological finding.
Facts & figures
| Score | Respiratory (PaO₂/FiO₂) | Renal (creatinine) | Cardiovascular (systolic BP) |
|---|---|---|---|
| 0 | > 400 | < 1.4 mg/dL | > 90 mmHg |
| 1 | 301–400 | 1.4–1.8 mg/dL | < 90, fluid responsive |
| 2 | 201–300 | 1.9–3.6 mg/dL | < 90, not fluid responsive |
| 3 | 101–200 | 3.6–4.9 mg/dL | < 90, pH < 7.3 |
| 4 | ≤ 101 | > 4.9 mg/dL | < 90, pH < 7.2 |
Only three organ systems are assessed. A score of 2 or more in any one of them is organ failure — there is no requirement for multiple systems, and no summing across them.
| Complication | Timing | Contents |
|---|---|---|
| Acute peripancreatic fluid collection | First 4 weeks | Fluid only, no necrosis, no defined wall — arises in interstitial oedematous pancreatitis |
| Pseudocyst | After 4 weeks | Fluid only, with a well-defined inflammatory wall. Genuinely uncommon |
| Acute necrotic collection | First 4 weeks | Variable fluid and necrosis, no defined wall — arises in necrotising pancreatitis |
| Walled-off necrosis | After 4 weeks | Necrosis with a well-defined inflammatory wall |
Each may be sterile or infected. The four-week line and the fluid-versus-necrosis distinction are what the revision tightened, because 'pseudocyst' had been used loosely for collections containing necrosis — which behave completely differently and are not drainable the same way.
Evidence
Derivation — international web-based consensus
2013A working group undertook a web-based consultation begun in 2007, circulating a draft to 11 national and international pancreatic associations and forwarding it to all members. Revisions were made in response to comments and the consultation was repeated three times; only statements based on published evidence were retained in the final consensus.
Definitional rather than predictive — the classification has no sensitivity, specificity or C-statistic, because it defines the outcome that prognostic scores are built to predict rather than predicting anything itself.
Competing framework — determinant-based classification
2012Published in the same period, this alternative classifies severity by two determinants — organ failure and (peri)pancreatic necrosis, each local or systemic — producing four categories including a 'critical' group.
Introduced a fourth, critical category for patients with both persistent organ failure and infected necrosis. Comparative studies have generally found the two systems perform similarly, with the revised Atlanta classification achieving wider adoption.
Guideline adoption — IAP/APA 2013
2013International Association of Pancreatology and American Pancreatic Association evidence-based guidelines for the management of acute pancreatitis.
Adopted the revised definitions and the emphasis on organ failure duration, and set out the management implications including the timing of CT and the step-up approach to infected necrosis.
Guideline adoption — ACG 2024
2024American College of Gastroenterology guidelines on the management of acute pancreatitis.
Uses the revised Atlanta severity categories, and reinforces the recommendations that follow from them — aggressive early fluid resuscitation moderated by more recent trial evidence, early enteral nutrition, and against prophylactic antibiotics.
How it compares
Revised Atlanta Classification vs BISAP score
Not alternatives — BISAP predicts severity within the first 24 hours, and the revised Atlanta classification is the severity it is predicting.
BISAP is a five-item bedside score computed on admission and designed to identify patients at risk before the illness has declared itself. The revised Atlanta classification is applied as events unfold and is only definitive at 48 hours. Confusing the two produces both errors: using Atlanta on arrival gives a provisional answer presented as final, and treating a BISAP score as a severity classification reports a prediction as though it were an outcome. In practice both belong in the notes — the prognostic score to guide early disposition, the classification to describe what actually happened.
Revised Atlanta Classification vs Ranson's criteria
Ranson's shares the 48-hour timing but answers the opposite question — it predicts severity from admission and 48-hour values, while Atlanta defines it from organ failure and complications.
Both need 48 hours, which is why they are often confused. Ranson's counts eleven adverse prognostic signs across admission and the first 48 hours to estimate the probability of a severe course; the revised Atlanta classification asks whether organ failure was actually present and whether it lasted. A patient can have four Ranson criteria and turn out to have mild pancreatitis by Atlanta, and that is not a contradiction — it is a prediction that did not come true. Report the classification for what happened and the prognostic score for what was expected.
Revised Atlanta Classification vs Determinant-based classification
A contemporaneous alternative that adds a fourth 'critical' category for persistent organ failure plus infected necrosis; comparative studies find the two perform similarly, and Atlanta has been more widely adopted.
The determinant-based classification builds severity from two determinants — organ failure and (peri)pancreatic necrosis, each considered local or systemic — producing mild, moderate, severe and critical categories. Its conceptual advantage is recognising that persistent organ failure with infected necrosis is a distinct and far worse situation than either alone, which the revised Atlanta classification collapses into 'severe'. Its practical disadvantage is that it requires knowing the infection status of necrosis, which is often not established when severity needs recording. The two systems are not incompatible, and where both are reported the critical category adds information at the top end.
Revised Atlanta Classification vs CT Severity Index (Balthazar)
CTSI grades what the scan shows; the revised Atlanta classification grades what the patient does — and the scan is unreliable in the window where the classification is being decided.
CTSI combines the Balthazar grade of peripancreatic inflammation with the extent of necrosis, and correlates with morbidity and mortality. But necrosis is generally not evident on CT in the first 72 hours, exactly the period in which organ failure declares itself and severity is being determined. So the two operate on different clocks: Atlanta severity is usually settled before CTSI can be scored meaningfully. Where a patient has local complications without organ failure, CTSI adds real detail about the extent of necrosis that the single label 'moderately severe' does not convey.
Pearls & pitfalls
- The classification cannot be finalised before 48 hours. Organ failure at 24 hours could still resolve, and the moderately severe/severe distinction turns entirely on whether it does.
- Organ failure is a modified Marshall score of 2 in ANY ONE system. There is no summing across systems and no requirement for more than one to fail.
- A Marshall subscore of 1 is not organ failure. Three systems each scoring 1 is still mild disease, which is counterintuitive but correct.
- Only three organ systems count — respiratory, cardiovascular and renal. Encephalopathy, coagulopathy and hepatic dysfunction are not part of this definition, unlike SOFA.
- A patient on inotropes is not fluid responsive, and therefore scores at least 2 on the cardiovascular system regardless of the blood pressure achieved.
- Local complications without organ failure make a patient moderately severe, not mild — this is the change from the 1992 system that people most often miss.
- Do not order contrast-enhanced CT in the first 72 hours for severity assessment. Necrosis is generally not evident that early, and a reassuring early scan is a false reassurance.
- 'Pseudocyst' is now a specific and uncommon entity: fluid only, after four weeks, with a defined wall. A collection containing necrosis is walled-off necrosis and behaves entirely differently.
- This is a classification, not a prediction. Using it on admission to decide disposition is applying the outcome rather than a prognostic score — BISAP, Glasgow-Imrie or Ranson's do that job.
- Exacerbation of co-morbid disease also qualifies for moderately severe. It is in the definition and routinely forgotten.
Critical actions
- Record the modified Marshall subscores per system rather than a bare 'organ failure present', so the classification can be checked and the failing organ is named.
- Reassess at 48 hours and record the definitive classification then; flag any earlier assessment as provisional.
- Establish the aetiology in every patient — biliary ultrasound at minimum, plus triglycerides and calcium where the cause is not obvious.
- Perform cholecystectomy during the index admission for mild gallstone pancreatitis; delaying it exposes the patient to recurrence.
- Start enteral nutrition early as tolerated rather than keeping the patient nil by mouth, and use the enteral rather than parenteral route where feeding is needed.
- Defer contrast-enhanced CT to 72–96 hours unless the diagnosis itself is in doubt or the patient deteriorates unexpectedly.
- Do not give prophylactic antibiotics. Treat infected necrosis when clinically suspected or proven, and use a step-up drainage approach in preference to early open necrosectomy.
- Escalate the care setting on persistent organ failure rather than on the pancreatitis label alone.
Why this score exists
The working group was explicit that the 1992 classification had become an obstacle rather than a help. Its two-category structure meant that any patient with a local complication was labelled severe, which grouped a patient with an asymptomatic peripancreatic collection alongside one in multi-organ failure and made mortality figures across studies uninterpretable. The revision's central judgement was that severity should track what determines death — persistent organ failure — rather than what shows up on a scan, and that a middle category was needed for the large group with genuine morbidity and low mortality. The second judgement was procedural: organ failure had to be defined by a stated score with a stated threshold, because leaving it to clinical impression is what allowed the same patient to be classified differently in different units. The choice to retain only statements supported by published evidence, across three rounds of international consultation, is why the document reads as definitions rather than recommendations.
About the creator
First author, Acute Pancreatitis Classification Working Group
Led the working group that produced the 2012 revision of the Atlanta classification.
Co-author; radiological definitions
Contributed the imaging definitions that distinguish fluid collections from necrotic collections and the standardised CT reporting template.
Senior author
Senior author of the revision and of much of the subsequent guideline work built on it.
Limitations
- Definitional rather than predictive, so it cannot guide early management decisions — that is what the prognostic scores are for.
- Cannot be finalised before 48 hours, which is precisely the window in which triage and escalation decisions have to be made.
- Only three organ systems are assessed. A patient with severe coagulopathy or encephalopathy and none of the three Marshall systems failing is not classified as having organ failure.
- The modified Marshall cardiovascular criterion depends on judging fluid responsiveness, which is the one subjective element in an otherwise objective definition.
- Groups together, as 'severe', patients with persistent organ failure alone and those with persistent organ failure plus infected necrosis, whose outcomes differ substantially — the gap the determinant-based classification's critical category was created to fill.
- Local complications and transient organ failure both produce 'moderately severe' despite representing different clinical problems with different trajectories.
- The four-week boundary between acute collections and their walled-off successors is a convention rather than a biological threshold, and maturation varies between patients.
- Says nothing about aetiology, recurrence risk or the transition to chronic pancreatitis.
If you are the patient
Acute pancreatitis is graded as mild, moderately severe or severe, and which one applies depends on two things: whether any major organ stops working properly, and if so, for how long. Mild means no organ problems and no collections of fluid or damaged tissue around the pancreas; this is the most common form and usually settles within a week. Moderately severe means either a temporary organ problem that recovers within two days, or a fluid collection or area of damaged tissue around the pancreas, or a flare-up of another condition you already had. Severe means an organ problem that lasts longer than two days, and this is the group where the illness is genuinely dangerous. The important practical point is that your team often cannot say which category applies on the day you arrive — they have to see whether things settle over the first 48 hours, so an early answer is provisional and may change in either direction. Two other things commonly surprise people. A scan is usually not done in the first three days, because the changes doctors are looking for are not visible that early and a scan then can be falsely reassuring. And eating is usually encouraged as soon as you can manage it, rather than being kept nil by mouth — that approach changed some years ago and is now what the evidence supports. If gallstones caused the attack, you will usually be offered gallbladder surgery during the same hospital stay to stop it happening again.
Frequently asked questions
What is the revised Atlanta classification of acute pancreatitis?#
A 2012 international consensus classification that grades acute pancreatitis as mild, moderately severe or severe. Mild has no organ failure and no local or systemic complications. Moderately severe has transient organ failure resolving within 48 hours, local complications, or exacerbation of co-morbid disease. Severe is defined by persistent organ failure lasting more than 48 hours.
How is organ failure defined in the revised Atlanta classification?#
By the modified Marshall scoring system, applied to three organ systems — respiratory (PaO₂/FiO₂), cardiovascular (systolic blood pressure and fluid responsiveness) and renal (serum creatinine). A score of 2 or more in any one of those three systems constitutes organ failure. Scores are not summed across systems, and a subscore of 1 does not count.
What is the difference between transient and persistent organ failure?#
The 48-hour mark. Organ failure that resolves within 48 hours is transient and gives moderately severe acute pancreatitis; organ failure lasting longer than 48 hours is persistent and defines severe acute pancreatitis. Because of this, the classification cannot be finalised until 48 hours have elapsed, and any earlier assessment should be recorded as provisional.
What changed from the 1992 Atlanta classification?#
Three things. A moderately severe category was added, so that patients with local complications or transient organ failure — real morbidity, low mortality — are no longer grouped with those in persistent multi-organ failure. Organ failure was given an objective definition with a stated threshold rather than being left to clinical impression. And the local complication definitions were tightened, separating collections containing only fluid from those containing necrosis, which behave very differently.
What are the local complications in the revised Atlanta classification?#
Four, distinguished by timing and contents. In the first four weeks: acute peripancreatic fluid collection (fluid only, no wall) and acute necrotic collection (fluid and necrosis, no wall). After four weeks: pseudocyst (fluid only, with a defined wall) and walled-off necrosis (necrosis with a defined wall). Each may be sterile or infected.
When should a CT scan be done in acute pancreatitis?#
Generally after 72 to 96 hours, not on admission. Necrosis is usually not radiologically evident in the first 72 hours, so an early scan can be falsely reassuring and rarely changes management. CT is not required at all for severity assessment in mild disease. Earlier imaging is appropriate where the diagnosis itself is in doubt or the patient deteriorates unexpectedly.
Is the revised Atlanta classification a prognostic score?#
No — it is a classification of what happened, not a prediction of what will. It defines the severity that prognostic scores such as BISAP, Glasgow-Imrie and Ranson's are built to predict. Using it on admission to guide disposition is applying an outcome measure to a decision it was not designed for.
How does the determinant-based classification differ?#
It builds severity from two determinants — organ failure and (peri)pancreatic necrosis, each local or systemic — and produces four categories, adding a 'critical' group for patients with both persistent organ failure and infected necrosis. That distinction is real, since those patients do considerably worse than either factor alone would suggest, but it requires knowing the infection status of the necrosis, which is often not established when severity needs to be recorded. Comparative studies generally find the two systems perform similarly.
References
Clinical practice guidelines
- Working Group IAP/APA Acute Pancreatitis Guidelines. IAP/APA evidence-based guidelines for the management of acute pancreatitis. Pancreatology. 2013;13(4 Suppl 2):e1-e15.
- Tenner S, Vege SS, Sheth SG, Sauer B, Yang A, Conwell DL, Yadlapati RH, Gardner TB. American College of Gastroenterology Guidelines: Management of Acute Pancreatitis. Am J Gastroenterol. 2024;119(3):419-437.