Acute pancreatitis

A 52-year-old man arrives with eighteen hours of severe epigastric pain radiating to his back, repeated vomiting, and a lipase of 4,200 U/L. Everything that follows — whether he needs a CT tonight, how fast the fluids run, whether antibiotics have any role, whether an ERCP is indicated, when his gallbladder comes out, and what to do if his pancreas necroses and infects — is mapped by a small, quotable evidence base. This page gives you that evidence with its numbers attached.^[3][14]

Definition and classification — revised Atlanta 2012

Acute pancreatitis is acute inflammation of the pancreas and one of the most common gastrointestinal diseases leading to hospital admission.^[3] The diagnosis requires at least two of three criteria: characteristic upper abdominal pain, serum amylase or lipase elevated (classically beyond three times the upper limit of normal), and characteristic findings on cross-sectional imaging.^[12][13] Note what this means in practice: imaging is not required to make the diagnosis when pain and enzymes align.^[13]

The 2012 revision of the Atlanta classification — built by international web-based consensus across eleven pancreatic associations — is the vocabulary examiners grade you on.^[1] It identifies two phases: early and late. Severity is classified as mild, moderately severe or severe: mild disease has no organ failure and no local or systemic complications and usually resolves within the first week; moderately severe disease carries transient organ failure (48 hours or less), local complications, or exacerbation of co-morbid disease; severe disease is defined by persistent organ failure — that is, organ failure beyond 48 hours.^[1]

The local complications are named and timed, and the names matter: acute peripancreatic fluid collection, pancreatic pseudocyst, acute necrotic collection, and walled-off necrosis (each sterile or infected).^[1] The organising boundary is roughly four weeks: pseudocyst and walled-off necrosis are the encapsulated, mature descendants of the two acute collections — which is why intervention timing later in this page is expressed in weeks, not days.^[1][4]

The determinant-based classification (Ann Surg 2012) is the complementary scheme worth knowing by name. It grades severity on the actual determinants rather than their correlates: the local determinant is whether (peri)pancreatic necrosis exists and whether it is sterile or infected; the systemic determinant is whether organ failure exists and whether it is transient or persistent — and infected necrosis combined with persistent organ failure outstrips either determinant alone, yielding four severity categories.^[2]

Epidemiology and aetiology

Gallstones are the most common cause of acute pancreatitis, followed closely by alcohol — the two causes that drive the two definitive interventions on this page (cholecystectomy and ERCP-when-cholangitis).^[14] The burden is large and rising: in the United States, acute pancreatitis accounted for approximately 390,000 hospitalisations in 2015, and the global epidemiology review by Petrov and Yadav documents the worldwide burden of acute pancreatitis, chronic pancreatitis and post-pancreatitis diabetes.^[14][28]

Pathophysiology — two phases, two killers

The revised Atlanta divides the disease into an early phase — the first week, dominated by the systemic inflammatory response to acinar injury — and a late phase, in which local complications evolve and persistent organ failure declares severe disease.^[1] WSES frames the clinical consequence: about 20–30% of patients develop the severe form, often with single or multiple organ dysfunction requiring intensive care, and infection of pancreatic and peripancreatic necrosis occurs in 20–40% of those severe cases, worsening organ dysfunction.^[7]

Necrosis itself is a radiological definition before it is a pathological one: Balthazar defined pancreatic necrosis on contrast CT as lack of enhancement of all or a portion of the gland, and the 1990 series quantified what it means — patients with necrosis had a 23% mortality and 82% complication rate, against 0% mortality and 6% morbidity without necrosis.^[11] That 23-to-0 gradient is the reason the whole late-phase machinery of this topic — antibiotics debates, drainage, necrosectomy — exists.

Clinical presentation

The characteristic pain is a severe, constant epigastric pain radiating to the back with nausea and vomiting; the diagnosis then rests on the two-of-three criteria rather than on any single feature.^[12] The examination that matters prognostically is the SIRS screen: the presence of the systemic inflammatory response syndrome on day 1 of admission is highly sensitive in predicting severe disease, and patients with organ failure or SIRS should be admitted to an intensive care or intermediary care setting whenever possible (ACG 2013).^[13][4]

Scoring systems — Ranson, Glasgow-Imrie, BISAP

Ranson (1974)

The historical anchor — eleven prognostic signs measured at admission and at 48 hours — published as "Prognostic signs and the role of operative management in acute pancreatitis".^[8] Its exam value today is heritage: it proved the principle that severity can be predicted from routine early measurements, the principle every successor score refined.

Glasgow-Imrie (1984)

The prospective refinement: a system of nine factors available within 48 hours of admission, assessed prospectively on 405 episodes over seven years. The numbers to quote: 31% of 131 episodes with three or more factors were severe, versus 8% of 274 episodes with fewer than three; a revised eight-factor system predicted severity correctly in 79% of episodes.^[9] The threshold of three or more factors survives verbatim in modern trial entry criteria.^[15]

BISAP (2008) — the modern bedside score

The Bedside Index of Severity in Acute Pancreatitis was derived by classification-and-regression-tree analysis on 17,992 cases from 212 hospitals and validated on a further 18,256 cases from 177 hospitals. One point each, within the first 24 hours: BUN over 25 mg/dL, Impaired mental status, SIRS, Age over 60 years, Pleural effusion. Mortality ran from under 1% in the lowest band to over 20% in the highest, and the validation AUC of 0.82 was statistically indistinguishable from APACHE II (0.83) — with five bedside variables instead of a dozen laboratory ones.^[10]

Investigations and the imaging pathway

Laboratory. Serum lipase is the preferred diagnostic test — more sensitive than amylase, particularly in alcohol-induced pancreatitis, and its prolonged elevation gives a wider diagnostic window.^[12] A threshold of two to four times the upper limit of normal is recommended for diagnosis.^[12] In the older comparative series, lipase above three times normal separated pancreatitis from non-pancreatic abdominal pain with 100% sensitivity and 99% specificity, against 72% sensitivity for amylase.^[12]

CT — reserved, and delayed. ACG 2013 is explicit: contrast-enhanced CT or MRI of the pancreas should be reserved for patients in whom the diagnosis is unclear or who fail to improve clinically.^[4] Two reasons make early routine CT poor practice: the diagnosis rarely needs it (two-of-three criteria), and necrosis takes days to declare itself — the gland's failure to enhance is only trustworthy once devitalisation has completed, which is why staging CT is deferred toward the 72–96-hour mark.^[4][11] When CT is performed, its severity statement is Balthazar's: non-enhancing gland is necrosis, and necrosis predicts the 23% mortality group.^[11]

Severity markers with trial pedigree. The trials define "predicted severe" for you: APEC and PYTHON both enrolled on APACHE II of 8 or more, Imrie score of 3 or more, or CRP over 150 mg/L.^[15][18] When a viva asks how you would operationalise predicted severe disease, that triplet is the answer with provenance.

Differential diagnosis

The practical differentials of acute epigastric pain with raised enzymes are the perforated or penetrating peptic ulcer, acute mesenteric ischaemia, aortic catastrophe, biliary colic and cholangitis, and high small-bowel obstruction. The enzyme pitfall cuts both ways: moderate amylase and lipase rises occur in non-pancreatic abdominal pain, which is why the three-times-normal threshold exists — in the comparative series, amylase overlapped between groups in 27% of pancreatitis cases while lipase above three times normal did not overlap at all.^[12]

Early management — fluids, analgesia, nutrition, disposition

Fluids — WATERFALL rewrote this section

The old dogma was aggressive early hydration; ACG 2013 still reflects that era, advising aggressive hydration most beneficial within the first 12–24 hours.^[4] WATERFALL (NEJM 2022) then tested it: 249 patients at 18 centres randomised to goal-directed aggressive resuscitation with lactated Ringer's (a 20 ml/kg bolus then 3 ml/kg/h) or moderate resuscitation (10 ml/kg bolus only if hypovolaemic, then 1.5 ml/kg/h). The trial was halted for harm: no difference in progression to moderately severe or severe disease (22.1% versus 17.3%), but fluid overload in 20.5% versus 6.3% (adjusted RR 2.85). Conclusion: early aggressive fluid resuscitation produced more fluid overload without improving clinical outcomes.^[17] The fellowship answer is goal-directed moderate resuscitation with lactated Ringer's, reassessed at defined intervals.^[17]

Nutrition — enteral first, early when tolerated

In mild disease, oral feeding starts immediately if there is no nausea and vomiting; in severe disease, enteral nutrition is recommended to prevent infectious complications, and parenteral nutrition should be avoided (ACG 2013).^[4] The meta-analysis behind that hierarchy (nine RCTs, 500 patients): total enteral versus total parenteral nutrition in severe acute pancreatitis gave mortality OR 0.31, pancreatic infection OR 0.41, organ failure OR 0.17 and surgical intervention OR 0.17 — all favouring enteral.^[27]

PYTHON (NEJM 2014) refined the timing question: 208 high-risk patients randomised to nasoenteric feeding within 24 hours versus an oral diet at 72 hours with tube feeding only on demand. The composite of major infection or death was identical — 30% versus 27% — and 69% of the on-demand group tolerated oral feeding and never needed a tube. Early routine nasoenteric feeding was not superior.^[18] The synthesis: feed early by mouth when tolerated; if tube feeding is needed, go enteral, never parenteral-first.^[4][18][27]

Disposition

Organ failure or SIRS earns intensive or intermediary care admission whenever possible (ACG 2013) — the corollary of day-1 SIRS being highly sensitive for severe disease.^[4][13]

Antibiotics — the two negative trials every examiner knows

Prophylaxis is dead. Two double-blind, placebo-controlled trials settled it.^[19][20]

• Isenmann 2004 (Gastroenterology): ciprofloxacin 400 mg twice daily plus metronidazole 500 mg twice daily versus placebo in predicted severe disease (CRP over 150 mg/L and/or necrosis on CT). Infected pancreatic necrosis developed in 12% on antibiotics versus 9% on placebo; mortality 5% versus 7%. No benefit.^[19]
• Dellinger 2007 (Ann Surg): meropenem 1 g intravenously every 8 hours within 5 days of symptom onset versus placebo in 100 patients with confirmed necrotizing pancreatitis. Pancreatic or peripancreatic infection in 18% versus 12%, mortality 20% versus 18%, surgery 26% versus 20% — no significant difference on any endpoint, and the authors did not support early prophylactic antimicrobial use.^[20]

The guideline consequence is categorical: routine prophylactic antibiotics in severe acute pancreatitis and/or sterile necrosis are not recommended (ACG 2013).^[4] Antibiotics are for established infection — infected necrosis, cholangitis, or another documented focus — and in infected necrosis, agents that penetrate pancreatic necrosis may usefully delay intervention.^[4]

PROCAP (Lancet Gastroenterol Hepatol 2022) supplies the stewardship layer: a procalcitonin-guided algorithm (test days 0, 4, 7 and weekly; stop or withhold below 1.0 ng/mL) cut antibiotic prescribing from 63% to 45% without increasing infection or harm.^[21]

ERCP timing and the gallbladder

ERCP is for cholangitis, not for pancreatitis severity. ACG 2013: patients with acute pancreatitis and concurrent acute cholangitis should undergo ERCP within 24 hours of admission.^[4] ACG 2024 keeps the same logic: biliary pancreatitis may need early ERCP when complicated by cholangitis.^[3] The pivotal trial is APEC (Lancet 2020): 232 patients with predicted severe gallstone pancreatitis without cholangitis (entry on APACHE II ≥8, Imrie ≥3, or CRP >150 mg/L), randomised to urgent ERCP with sphincterotomy within 24 hours versus conservative treatment. The composite of mortality or major complications was 38% versus 44% (RR 0.87) — no benefit. The only component that differed was cholangitis itself (2% versus 10%), confirming that conservative management still catches and treats cholangitis when it declares. Interpretation: urgent ERCP without cholangitis does not help; reserve it for cholangitis or persistent cholestasis.^[15]

Cholecystectomy — same admission for mild disease. PONCHO (Lancet 2015) randomised 266 patients recovering from mild gallstone pancreatitis to same-admission versus interval cholecystectomy: recurrent gallstone-related complications fell from 17% to 5% (RR 0.28), with a very low rate of cholecystectomy complications.^[16] Both ACG-era guidance and current reviews agree: cholecystectomy during the initial admission for mild acute biliary pancreatitis.^[13][16] ACG 2024 states the rationale plainly: biliary pancreatitis typically requires surgery to prevent recurrent disease.^[3] The deliberate exception is severe or necrotizing biliary pancreatitis, where the cholecystectomy waits for the necrosis strategy to play out — operating into unevolved necrosis is the error this caveat exists to prevent.

Necrotizing pancreatitis — delay, drain, debride

The framework before the trials: sterile necrosis is managed non-operatively in most patients; infected necrosis usually requires an intervention that can be percutaneous, endoscopic, or open surgical (WSES 2019).^[7] Two rules precede every intervention decision: asymptomatic necrosis and pseudocysts do not warrant intervention regardless of size, and in stable infected necrosis, drainage should be delayed preferably for 4 weeks to allow a wall to form (ACG 2013).^[4]

PANTER (NEJM 2010) — step-up beats open

88 patients with suspected or confirmed infected necrosis randomised to open necrosectomy versus a step-up approach — percutaneous drainage first, then minimally invasive retroperitoneal necrosectomy only if needed. The composite of major complications or death: 69% open versus 40% step-up (RR 0.57); new-onset multiple organ failure 40% versus 12%; and 35% of the step-up group were cured with drainage alone. Mortality was statistically identical (16% versus 19%) — the step-up wins on morbidity, not mortality.^[22]

POINTER (NEJM 2021) — postponement is safe

104 patients with infected necrotizing pancreatitis randomised to immediate drainage within 24 hours of diagnosis versus drainage postponed until walled-off necrosis. Comprehensive Complication Index: 57 versus 58 — no difference; mortality 13% versus 10%; but the postponed group needed fewer interventions (mean 2.6 versus 4.4). Postponement to encapsulation is not only safe, it is leaner.^[24]

TENSION (Lancet 2018) — endoscopic versus surgical step-up

98 patients with infected necrosis and an indication for intervention randomised to an endoscopic step-up (EUS-guided transluminal drainage, then endoscopic necrosectomy if needed) versus a surgical step-up (percutaneous catheter drainage, then video-assisted retroperitoneal debridement). Major complications or death: 43% versus 45% — no superiority; but the endoscopic route had fewer pancreatic fistulas and shorter hospital stay, foreshadowing the shift toward endoscopic-first practice where expertise exists.^[23]

The stent detail and the evidence quality

The LAMS comparison (Gut 2023) — lumen-apposing metal stents against double-pigtail plastic stents in the endoscopic step-up — found no reduction in the need for endoscopic necrosectomy (64% versus 53%) and comparable bleeding (9% versus 22%, RR 0.44): the expensive stent is not the active ingredient; the step-up logic is.^[26] And the Cochrane review (8 RCTs, 311 participants) keeps everyone honest: low to very low quality evidence favours the minimally invasive step-up over open necrosectomy for adverse events, serious adverse events, organ failure and costs.^[25]

Complications and prognosis

Local complications are the four Atlanta collections; systemic complications are organ failure, transient or persistent — and the grade follows the persistence.^[1] The prognostic anchors, oldest to newest: necrosis versus no necrosis, 23% versus 0% mortality (Balthazar 1990)^[11]; BISAP band mortality from under 1% to over 20% (Wu 2008)^[10]; and the contemporary trial mortalities for infected necrosis — 16–19% in PANTER, 10–13% in POINTER, 13–18% in TENSION — which tell you both that the step-up era works and that infected necrosis remains a lethal disease.^[22][24][23]

Guideline landscape

The evidence base is unusually convergent. IAP/APA 2013 set the GRADE-based frame — 38 clinical questions, 21 of 38 recommendations rated strong, 34 of 38 with strong plenary agreement.^[6] ACG 2013 supplied the quoted rules on CT reservation, ICU disposition, ERCP within 24 hours for cholangitis, no prophylactic antibiotics, enteral nutrition, and delayed intervention.^[4] The AGA Institute Guideline on Initial Management (2018) covers the early phase^[5], ACG 2024 updates the whole pathway^[3], and WSES 2019 owns severe disease, infected necrosis and the operative questions.^[7] Beneath all of them sit the revised Atlanta and determinant-based classifications — the definitions every recommendation is written in.^[1][2]

Revision summary

• Two of three criteria diagnose; lipase first; enzymes never stage severity.^[12]
• Atlanta 2012: mild / moderately severe / severe on organ-failure persistence beyond 48 hours; four local collections divided by the 4-week encapsulation boundary.^[1]
• BISAP for the bedside (five variables, first 24 hours, mortality under 1% to over 20%); Glasgow-Imrie three-or-more-factors threshold lives on in trial entry criteria.^[10][9][15]
• CT reserved for unclear diagnosis or failure to improve; non-enhancement is necrosis; necrosis means the 23% mortality group.^[4][11]
• Fluids: moderate, goal-directed, lactated Ringer's — WATERFALL halted for overload harm (20.5% vs 6.3%).^[17]
• Feeding: immediate oral in mild disease, enteral over parenteral in severe (mortality OR 0.31); PYTHON — on-demand oral at 72 hours matched early tube feeding, and 69% never needed a tube.^[4][27][18]
• No prophylactic antibiotics — Isenmann (12% vs 9%) and Dellinger (18% vs 12%); PROCAP's procalcitonin algorithm safely cut prescribing 63% to 45%.^[19][20][21]
• ERCP within 24 hours only for cholangitis — APEC: 38% vs 44% without it; same-admission cholecystectomy for mild biliary disease — PONCHO: 5% vs 17% recurrence.^[15][16]
• Infected necrosis: delay to encapsulation (preferably 4 weeks; POINTER 57 vs 58 CCI), step-up drainage first (PANTER 40% vs 69%), endoscopic or surgical minimally invasive necrosectomy (TENSION 43% vs 45%), open necrosectomy last.^[24][22][23]