Ranson Criteria for Acute Pancreatitis: The 11 Signs, Scoring, and the 48-Hour Limitation

The Ranson criteria are a set of 11 prognostic signs for acute pancreatitis, described by John Ranson and colleagues in 1974, that estimate the risk of a severe course and of in-hospital death. They are unusual among bedside clinical scores in one specific way: the score cannot be completed on arrival. Five of the criteria are assessed at admission and the remaining six only at 48 hours, so a Ranson score does not exist as a number until two days into the admission. That single design feature explains most of what is written about the score today — why it remains one of the most recognised names in pancreatitis severity assessment, why it is a poor fit for emergency-department triage, and why newer instruments were built to replace it in exactly the window where a triage decision has to be made.

This page covers the instrument itself: the exact criteria and their numeric cutoffs, the separate gallstone-modified variant, how the total is built, the mortality bands conventionally published alongside it, and what a quality or patient-safety department actually has to solve if it wants the score calculated reliably. For the surrounding hospital deterioration-detection machinery, see CASRAI’s guides on early warning score implementation and rapid response team activation criteria, and the Patient Safety & Infection Prevention hub for the wider cluster.

The five admission criteria

These are assessed on the initial presentation and reflect the early inflammatory burden. Each criterion met scores one point; the criteria are not weighted, so no single sign counts for more than any other. The thresholds below are the original (non-gallstone) set, per the StatPearls review of the Ranson criteria:

• Age older than 55 years.
• White blood cell count greater than 16,000 cells/mm³.
• Blood glucose greater than 200 mg/dL (11 mmol/L).
• Serum AST greater than 250 IU/L. AST, not ALT — substituting ALT is a common abstraction error, and the two are not interchangeable at this cutoff.
• Serum LDH greater than 350 IU/L.

A commonly taught mnemonic for the admission set is GA LAW — Glucose, Age, LDH, AST, WBC. Note what is absent: no vital signs, no mental status, no imaging, and no measure of organ function. The admission half of the Ranson score is essentially age plus four laboratory values, which is why it can be computed retrospectively from discrete data but tells a clinician relatively little at the bedside that the labs themselves do not.

The six 48-hour criteria

These are assessed at 48 hours after admission and reflect ongoing physiologic derangement, third-spacing, and the systemic consequences of the inflammatory response rather than its initial intensity:

• Haematocrit falls by more than 10% from the admission value.
• Blood urea nitrogen increased by 5 mg/dL or more (≥1.8 mmol/L) despite intravenous fluid hydration. The qualifier matters: a BUN that rises while the patient is being volume-resuscitated is the finding, not a BUN that is simply high.
• Serum calcium less than 8.0 mg/dL (<2.0 mmol/L).
• PaO₂ less than 60 mmHg.
• Base deficit greater than 4 mEq/L.
• Fluid sequestration greater than 6 L over the 48-hour period.

The corresponding mnemonic is C HOBBS — Calcium, Haematocrit, Oxygen, BUN, Base deficit, Sequestration. Two of these six require an arterial blood gas (PaO₂ and base deficit), and one — fluid sequestration — is not a laboratory result at all. It is a derived quantity: total fluid intake minus total measured output across the 48 hours. That distinction is the single biggest practical obstacle to calculating the score reliably, and it is dealt with in detail below.

The gallstone-modified variant has 10 criteria, not 11

Ranson’s original series was drawn largely from patients with alcohol-related pancreatitis. A separate set of thresholds was subsequently derived for gallstone (biliary) pancreatitis, and it is not simply the same score with different numbers — it has one fewer criterion, because PaO₂ is omitted entirely. The gallstone-modified version therefore runs 0–10, with five criteria at admission and five at 48 hours:

Criterion	Original (non-gallstone)	Gallstone-modified
Age	> 55 years	> 70 years
WBC	> 16,000 cells/mm³	> 18,000 cells/mm³
Glucose	> 200 mg/dL	> 220 mg/dL
AST	> 250 IU/L	> 250 IU/L (unchanged)
LDH	> 350 IU/L	> 400 IU/L
Haematocrit fall	> 10%	> 10% (unchanged)
BUN rise	≥ 5 mg/dL	≥ 2 mg/dL
Serum calcium	< 8.0 mg/dL	< 8.0 mg/dL (unchanged)
PaO₂	< 60 mmHg	Not included
Base deficit	> 4 mEq/L	> 5 mEq/L
Fluid sequestration	> 6 L	> 4 L

Only three of the eleven thresholds are identical between the two versions. Any calculator, order set, or abstraction specification that treats “Ranson” as a single instrument without branching on aetiology is producing the wrong number for roughly half of the patients it is applied to — and the aetiology has to be known before the score can be computed, which is itself a determination made from imaging and liver chemistries rather than from anything inside the score.

Scoring and the mortality bands

Each criterion met scores one point, unweighted, for a maximum of 11 (or 10 in the gallstone-modified version). A total of 3 or more is the conventional threshold for predicted severe disease. The mortality bands routinely published alongside the score are:

• 0–2 points: mortality approximately 0–3%
• 3–4 points: approximately 15%
• 5–6 points: approximately 40%
• 7–11 points: approaching 100%

These figures should be read for what they are: bands derived from the original derivation-era cohorts and reproduced ever since, not contemporary outcome rates from a modern intensive-care population. They describe the score’s gradient — that risk rises steeply and non-linearly with each additional criterion met — more dependably than they describe the absolute risk any individual patient faces today. A high Ranson score identifies a patient whose 48-hour trajectory has been bad; it does not establish a cause of death, and the top band in particular rests on very small historical numbers. Do not quote the band figures to families or reproduce them in a discharge summary as a prognosis.

On measured discriminative performance, a 2016 meta-analysis reported that a Ranson score greater than 2 had median sensitivity of 90% and median specificity of 67.4%. That profile — sensitive, not specific — is worth stating plainly: the score is reasonably good at not missing patients who will do badly, and comparatively poor at identifying the ones who will not. Used as a rule-out it behaves acceptably; used as a rule-in for escalating care it over-triages.

Why the 48-hour wait is the defining limitation

Every other criticism of the Ranson criteria is downstream of the timing. The score cannot be finalised until 48 hours after admission, which means:

• It is unusable for emergency-department disposition. The decision the score would be most valuable for — ward versus step-down versus ICU, at hour one — has already been made by the time the score exists. StatPearls states the constraint directly: severity “cannot be determined until 48 hours have passed, which limits its utility in time-sensitive situations, such as the emergency department.”
• By 48 hours, the clinical picture usually speaks for itself. A patient who has developed persistent organ failure by day two is identifiable without a score. This is the substantive complaint: the instrument delivers its answer at the point where it adds least incremental information over direct clinical assessment.
• It is a one-shot measurement, not a trend. Ranson is calculated once and does not repeat. Scores designed for serial recalculation — SOFA being the clearest example — capture trajectory, which is generally more informative than any single cross-sectional value. A rising score over successive days carries information a fixed 48-hour total cannot.
• Missing data silently deflates the total. If no arterial blood gas was drawn, PaO₂ and base deficit cannot be scored, and the conventional handling — scoring an unmeasured criterion as not met — systematically biases the total downward. A score of 2 built from nine available criteria is not the same finding as a score of 2 built from eleven, but nothing in the number itself records the difference.

How it compares to APACHE II and BISAP

Both of the instruments most often placed alongside Ranson were designed around the timing problem rather than around better physiology.

APACHE II scores 12 acute physiologic variables plus points for age and chronic health status. It is not pancreatitis-specific — it is a general critical-illness severity score — and its advantage here is that it can be calculated on admission and recalculated daily, so it yields both an early number and a trend. Its cost is complexity: it requires substantially more data elements than Ranson, including an arterial blood gas, and is impractical to compute by hand at the bedside. In the same 2016 meta-analysis, an APACHE II score greater than 7 had 100% median sensitivity.

BISAP (Bedside Index of Severity in Acute Pancreatitis) went the other way: five elements, scored within the first 24 hours — BUN above 25 mg/dL, impaired mental status, systemic inflammatory response syndrome, age over 60, and pleural effusion on imaging. It is deliberately minimal and deliberately early, which is precisely the niche Ranson cannot occupy. The same meta-analysis reported a BISAP score greater than 2 as having 87.6% median specificity — the mirror image of Ranson’s profile, and the reason the two are sometimes discussed as complementary rather than competing.

The practical summary for a hospital selecting one: BISAP for early triage because it is available at 24 hours and is comparatively specific; APACHE II where an ICU already computes it for other purposes and a daily trend is wanted; Ranson mainly where it is already embedded in an existing pathway, registry, or research dataset.

Where Ranson sits relative to the Revised Atlanta Classification

A point that is regularly conflated: the Ranson criteria predict severity; the Revised Atlanta Classification defines it. They are not alternatives, and a Ranson score does not assign an Atlanta category.

The 2012 revision of the Atlanta classification, produced by international consensus, defines two phases of acute pancreatitis — early (roughly the first one to two weeks) and late — and three severity grades based on organ failure as graded by the modified Marshall scoring system:

• Mild: no organ failure and no local or systemic complications.
• Moderately severe: transient organ failure (resolving within 48 hours), local complications, and/or exacerbation of coexisting disease.
• Severe: persistent organ failure, meaning organ failure lasting 48 hours or more.

Because the Atlanta severity grade depends on whether organ failure persists past 48 hours, it too is a retrospective determination — but it is a determination about what actually happened, not a prediction of what might. For coding, registry submission, and any severity-stratified quality analysis, the Atlanta grade is the classification of record. A Ranson score is a risk estimate that sits upstream of it and should never be entered in its place.

What quality and patient-safety teams actually have to solve

If a Ranson score is being captured at all — in an order set, a pancreatitis pathway, a clinical decision support rule, or a registry submission — the implementation problem is not clinical, it is data-structural. The eleven criteria fall into three tiers that behave very differently:

• Directly computable from discrete data (6 of 11): age, WBC, glucose, AST, LDH, and serum calcium are single numeric lab or demographic values with unambiguous cutoffs. These automate cleanly.
• Computable but requiring a paired comparison (2 of 11): the haematocrit fall and the BUN rise are deltas, not values. Automating them requires the rule to anchor correctly on the admission result and select the right 48-hour comparator — and the BUN criterion further requires establishing that IV hydration was in fact running, which is an infusion record, not a lab.
• Requiring data that may not exist (3 of 11): PaO₂ and base deficit require an arterial blood gas that many patients with mild pancreatitis never have drawn. Fluid sequestration requires a complete 48-hour intake and output ledger — the least reliably documented data in most inpatient records, and the criterion most likely to be silently scored as not met by default.

The consequence is a specific and predictable failure mode: an automated Ranson score drifts low, because the three hardest-to-capture criteria all default to zero when the underlying data is absent. Anyone building or validating such a calculation should measure the denominator explicitly — what proportion of scored encounters had all eleven criteria actually evaluable — and report the score alongside it, rather than presenting a total that conceals how much of the instrument was silently unavailable. The same discipline applies as in any quality measure chart abstraction workflow, where a data element that cannot be found and a data element confirmed absent are not the same finding.

Two further governance points. First, a predicted-severity score is not a risk adjuster: if pancreatitis outcomes are being compared across units or over time, the adjustment belongs in a proper observed-to-expected model, not in a stratification by Ranson band. Second, an escalation trigger built on a 48-hour score is not a deterioration-detection mechanism, and should not be counted as one in a governance report — that role belongs to a continuously scored instrument such as a modified early warning score feeding a rapid response pathway.

Frequently asked questions

What are the 11 Ranson criteria?

Five assessed at admission — age over 55, WBC over 16,000 cells/mm³, glucose over 200 mg/dL, AST over 250 IU/L, LDH over 350 IU/L — and six assessed at 48 hours — haematocrit fall over 10%, BUN rise of 5 mg/dL or more despite IV fluids, calcium under 8.0 mg/dL, PaO₂ under 60 mmHg, base deficit over 4 mEq/L, and fluid sequestration over 6 L.

What is a bad Ranson score?

Three or more criteria met is the conventional threshold for predicted severe acute pancreatitis. The published mortality bands rise steeply above that: roughly 15% at 3–4 points, 40% at 5–6, and approaching 100% at 7 or more — though those are legacy derivation-era figures describing a gradient rather than current absolute risk.

Why does the Ranson score take 48 hours?

Six of the eleven criteria are defined as changes measured over the first 48 hours — a haematocrit fall, a BUN rise, cumulative fluid sequestration — so they cannot exist as findings any earlier. The score is measuring trajectory, not a snapshot, and the trajectory needs two days to be observable.

Is Ranson or BISAP better?

They answer different questions. BISAP is available within 24 hours and is comparatively specific, making it the more useful early triage instrument. Ranson is more sensitive but only completes at 48 hours, by which point clinical assessment has usually reached the same conclusion. For a hospital choosing one instrument for early risk stratification, BISAP is the more practical choice.

Does the Ranson score differ for gallstone pancreatitis?

Yes, substantially. The gallstone-modified version has 10 criteria rather than 11 — PaO₂ is omitted — and seven of the remaining thresholds differ, including age (over 70 rather than 55), WBC (18,000), glucose (220 mg/dL), LDH (400 IU/L), BUN rise (2 mg/dL), base deficit (5 mEq/L), and fluid sequestration (4 L).

Is the Ranson score still used?

It remains widely taught and appears in many existing pathways and datasets, but it is no longer the instrument of choice for early severity prediction, and it has never been the instrument that assigns a severity classification — that is the Revised Atlanta Classification’s role. Its practical remaining uses are within pathways where it is already embedded and in research or registry datasets where it provides continuity with historical cohorts.

Related CASRAI guides

• SOFA Score and qSOFA Score: Sepsis-3 Organ Dysfunction Scoring — the serially recalculated organ-failure instrument, and the modified Marshall system’s closest relative in routine use.
• Wells Criteria for DVT and PE — a contrasting criteria-based score designed to be computable at first contact.
• Early Warning Score Implementation — choosing a score, calibrating the threshold, and managing alarm burden.
• Standardized Mortality Ratio (SMR) — why severity scores and mortality comparisons are separate problems.
• Present on Admission (POA) Indicator — how admission-time findings are coded and why the distinction drives reporting.