About the Harmless Acute Pancreatitis Score (HAPS)
Three things must all be normal on admission: no rebound tenderness or guarding, a haematocrit below 43% in men or 39.6% in women, and a creatinine below 2 mg/dL. If all three hold, the attack was correctly identified as non-severe in 98% of the validation cohort, which supports ward-level rather than intensive care. The asymmetry is the whole point — this is a rule-out with high specificity and essentially no ability to rule in, so a failed HAPS does not predict a severe course, it only means the score cannot reassure. It is also available within about half an hour of arrival, long before the scores that need 48 hours.
Formula
HAPS is 'harmless' when ALL THREE of the following hold:
• No rebound tenderness and no guarding
• Haematocrit < 43% (men) or < 39.6% (women)
• Serum creatinine < 2 mg/dL
Any one abnormal → not harmless- Peritonism
- Rebound tenderness and/or guarding. Absence required.
- Haematocrit
- Sex-specific: below 43% in men, below 39.6% in women.
- Creatinine
- Below 2 mg/dL (about 177 µmol/L).
- It is a conjunction, not a sum. There is no score to add up and no partial credit — all three must be normal.
- The haematocrit threshold is sex-specific. A haematocrit of 40% is normal in a man and abnormal in a woman, which changes the result.
- All three thresholds are strict: a haematocrit of exactly 43% in a man, or a creatinine of exactly 2.0 mg/dL, counts as abnormal.
- A 'not harmless' result predicts NOTHING about severity. The score has high specificity for a benign course and no useful sensitivity for a severe one, so it cannot be inverted.
- It is an admission snapshot. It is not designed for serial use and does not track a patient's trajectory.
Interpreting the result
A harmless result is a genuinely strong signal: in the validation cohort it correctly identified a non-severe course in 200 of 204 patients, around 98%. It supports ward-level care, standard fluids and analgesia, and early enteral feeding as tolerated, alongside clinical judgement rather than in place of it. A result that is not harmless is much weaker and is routinely over-read. It does not say the patient will develop severe pancreatitis; it says only that this particular reassurance is unavailable, and severity should then be assessed with an instrument built for that purpose — BISAP early, or the revised Atlanta classification once 48 hours have elapsed. The abnormal parameter itself is usually the more useful finding: a raised haematocrit is an indication to resuscitate, a raised creatinine needs its own assessment, and peritonism should prompt a fresh look at whether the diagnosis is right.
| Score | Band | What it means | Action |
|---|---|---|---|
| All three normal | Harmless course predicted | Non-severe course correctly identified in 200 of 204 patients (98%) in the validation cohort | Supports ward-level care; standard fluids, analgesia and early enteral feeding as tolerated |
| Any one abnormal | Harmless course not predicted | Carries no predictive weight for severity — the score has high specificity for a benign course and no useful sensitivity for a severe one | Assess severity with a tool designed for it (BISAP, or revised Atlanta at 48 hours); treat the abnormal parameter on its own merits |
What the HAPS needs (4 inputs)
- Rebound tenderness and/or guarding
- Signs of peritonism on abdominal examination. Their absence is what the score requires. Their presence also raises the question of a perforated viscus mimicking pancreatitis.
- Sex
- Selects the haematocrit threshold, which is 43% in men and 39.6% in women. Using a single cut-off for both misclassifies women with a haematocrit in the low 40s.
- Haematocrit
- Normal is below 43% in men and below 39.6% in women. A raised value reflects haemoconcentration from third-space fluid loss, which is one of the earliest markers of a severe attack.
- Serum creatinine
- Normal is below 2 mg/dL, about 177 µmol/L. Renal impairment at presentation is among the strongest early predictors of a complicated course.
Units. Haematocrit is entered as a percentage; where your laboratory reports it as a fraction (for example 0.41), multiply by 100. Creatinine is entered in mg/dL with a µmol/L toggle — the 2 mg/dL threshold is about 177 µmol/L, and to convert by hand divide µmol/L by 88.4.
What it returns
- Harmless or not harmless
- A binary result. 'Harmless' means all three parameters are normal; 'not harmless' means at least one is abnormal and carries no predictive weight of its own.
- Which parameter failed
- Named explicitly, because the abnormality itself usually needs acting on — haemoconcentration means fluid resuscitation, and peritonism means reconsidering the diagnosis.
How it is calculated
Lankisch and colleagues worked backwards from the question clinicians actually face on admission — not 'how bad will this get?' but 'can I safely avoid intensive care?' — and looked for the smallest set of parameters that would answer it. They derived the algorithm in 394 patients admitted between 1987 and 2003 and tested it prospectively in a further 452 admitted between 2004 and 2006. The three surviving parameters each track a different early mechanism of a severe attack: peritonism reflects the extent of peritoneal irritation and raises the possibility of an alternative intra-abdominal catastrophe; haematocrit reflects haemoconcentration from the massive third-space fluid shift that characterises severe disease; and creatinine reflects the renal hypoperfusion that follows it. Because all three are available within roughly half an hour of arrival, the score answers the admission question at the moment it is being asked rather than two days later.
Facts & figures
| Score | Available at | Direction it works in |
|---|---|---|
| HAPS | Admission (~30 minutes) | Rules OUT a severe course |
| BISAP | Within 24 hours | Predicts mortality |
| Glasgow-Imrie | 48 hours | Predicts severity |
| Ranson's | 48 hours (11 parameters across two time points) | Predicts mortality |
| Revised Atlanta | 48 hours minimum, by definition | Classifies severity retrospectively |
| CTSI | After contrast CT, usually ≥ 72 hours | Grades morphological severity |
HAPS is the only one of these that returns an answer at the moment the admission decision is made, which is why it complements rather than competes with the others.
| Parameter | Threshold | Mechanism |
|---|---|---|
| Rebound tenderness / guarding | Absent | Extent of peritoneal irritation; also flags a possible alternative diagnosis |
| Haematocrit | < 43% men, < 39.6% women | Haemoconcentration from third-space fluid loss |
| Creatinine | < 2 mg/dL (~177 µmol/L) | Renal hypoperfusion following that fluid shift |
Each is a different early consequence of the same process, which is why all three have to be normal rather than most of them.
Evidence
Derivation and prospective validation — Lankisch et al.
2009 · n = 452Algorithm derived in 394 patients with acute pancreatitis admitted between 1987 and 2003, then tested prospectively in a separate cohort of 452 patients admitted between 2004 and 2006.
In the validation cohort the score correctly identified a harmless course in 200 of 204 patients (98%). The parameters are obtainable within roughly 30 minutes of admission, which was an explicit design goal.
Independent validation against BISAP
2023Prospective comparison of HAPS and BISAP for predicting severity and outcomes in acute pancreatitis.
Reported around 90.9% sensitivity and a negative predictive value of about 96.7% for a negative HAPS predicting a mild course, supporting its use for triage while confirming that a positive score does not reliably identify severe disease.
Revised Atlanta classification — the reference standard it is measured against
2013International consensus revision of the Atlanta classification, defining mild, moderately severe and severe acute pancreatitis by organ failure and its duration.
Provides the severity definitions HAPS is validated against, and cannot itself be finalised until 48 hours have elapsed — the gap HAPS exists to fill.
How it compares
HAPS vs BISAP
Opposite directions and different timings — HAPS rules out a severe course at admission, BISAP predicts mortality within the first 24 hours.
BISAP scores five parameters (BUN, impaired mental status, SIRS, age and pleural effusion) and is built to identify patients at risk of dying. HAPS is built to identify patients who can safely avoid intensive care, and returns an answer roughly 30 minutes after arrival rather than within a day. Using BISAP to reassure, or HAPS to predict severity, misuses both. In practice HAPS informs the initial bed decision and BISAP the subsequent risk assessment.
HAPS vs Revised Atlanta classification
Not comparable — Atlanta classifies severity retrospectively once 48 hours have passed, while HAPS answers a triage question at admission.
The revised Atlanta classification defines mild, moderately severe and severe pancreatitis by organ failure and whether it persists beyond 48 hours, so by construction it cannot be finalised on the day of admission. That 48-hour gap is precisely what HAPS was designed to cover. Atlanta remains the reference standard against which HAPS is validated.
HAPS vs Glasgow-Imrie score
Both predict a non-severe course, but Glasgow-Imrie needs 48 hours and eight parameters where HAPS needs 30 minutes and three.
Glasgow-Imrie scores eight laboratory and demographic parameters and is calculated at 48 hours, making it a severity assessment rather than a triage tool. HAPS trades breadth for immediacy. Many units use HAPS to decide the initial destination and a 48-hour score to confirm the classification.
Pearls & pitfalls
- A 'not harmless' result does not predict severe pancreatitis. The score is a rule-out with high specificity and no useful sensitivity in the other direction, and inverting it is the commonest misuse.
- The haematocrit threshold is sex-specific — 43% in men, 39.6% in women. A haematocrit of 40% flips the result between the two.
- All three thresholds are strict. A haematocrit of exactly 43% in a man, or a creatinine of exactly 2.0 mg/dL, counts as abnormal.
- It is a conjunction, not a sum. Two normal parameters out of three is not a partial pass.
- Peritonism should prompt a diagnostic rethink as well as a scoring decision — a perforated viscus can present with a raised amylase and abdominal pain.
- Use it at admission, not later. It is a snapshot and has no role in tracking a patient whose course has already declared itself.
- A raised haematocrit is an instruction, not just a data point: haemoconcentration indicates significant third-space loss and needs fluid resuscitation.
- It says nothing about aetiology. Gallstones and alcohol account for most cases, and a retained stone with cholangitis needs urgent ERCP regardless of how harmless the pancreatitis looks.
Critical actions
- Apply the score at admission, while the triage decision is still open.
- Where it is harmless, manage on the ward with fluids, analgesia and early enteral feeding as tolerated, and document the reasoning.
- Where it is not harmless, assess severity with an appropriate instrument rather than assuming a severe course.
- Resuscitate a haemoconcentrated patient — a raised haematocrit is the earliest fluid-status signal in this illness.
- Re-examine the abdomen where peritonism is present, and consider whether the diagnosis is actually pancreatitis.
- Establish the aetiology in every case, and arrange urgent ERCP if cholangitis coexists.
- Reassess clinically regardless of the score; HAPS informs the initial disposition and nothing after it.
Why this score exists
The name is the argument. Most pancreatitis scores are built to predict severity, which is the harder problem and the one they perform least well at; Lankisch's group inverted the question and asked which patients could be confidently excluded from needing intensive care. That inversion is why the score works with only three parameters and why its accuracy figure is so much higher than the severity scores manage — identifying benign disease in a population where most disease is benign is a genuinely easier task. It also explains the score's one real trap. Because it was optimised in one direction, it carries almost no information in the other, and reading a positive HAPS as a warning of severe pancreatitis inverts a tool that was deliberately built to be asymmetric.
About the creator
First author, 2009 derivation and validation study
Led the development of the score and much of the epidemiological work on acute pancreatitis it draws on.
Co-author; provided the epidemiological and statistical analysis.
Co-author of the derivation and validation study.
Limitations
- Asymmetric by design: high specificity for a benign course, almost no sensitivity for a severe one, so it cannot be used to identify patients who will deteriorate.
- Derived and validated in a single-centre German population over a long period, during which pancreatitis management changed substantially.
- The haematocrit threshold assumes a baseline normal haematocrit; a patient with pre-existing polycythaemia or anaemia will be misclassified in opposite directions.
- An admission snapshot with no capacity to track deterioration, in an illness that characteristically evolves over the first 48 to 72 hours.
- Says nothing about aetiology, and specifically will not identify coexisting cholangitis, which changes management urgently.
- Studied in first attacks; performance in recurrent or chronic-on-acute pancreatitis is less well characterised.
- Some validation studies have used a single haematocrit cut-off of 44% for both sexes rather than the sex-specific thresholds, which makes results across studies not strictly comparable.
If you are the patient
Most attacks of pancreatitis are mild and settle with fluids, pain relief and time, but a minority become serious, and doctors have to decide early which kind they are dealing with. The Harmless Acute Pancreatitis Score is a quick check done soon after arrival at hospital, using an examination of the abdomen and two routine blood tests. If the abdomen is not rigid or exquisitely tender on release of pressure, the blood is not concentrated, and the kidney test is normal, then the attack is very likely to be a mild one — this was correct in about 98% of patients in the study that tested it. That supports being looked after on a normal ward rather than in intensive care. If one of the three is abnormal, it does not mean the attack will be severe. It only means this particular quick check cannot give reassurance, and the medical team will keep assessing in other ways.
Frequently asked questions
What is the Harmless Acute Pancreatitis Score?#
A three-parameter admission check: absence of rebound tenderness and guarding, a haematocrit below 43% in men or 39.6% in women, and a creatinine below 2 mg/dL. If all three hold, a non-severe course is predicted, which was correct in 98% of the validation cohort.
Does a positive HAPS mean the pancreatitis will be severe?#
No, and this is the most important caveat. The score was optimised in one direction only — it has high specificity for a benign course and almost no sensitivity for a severe one. A failed HAPS means the reassurance is unavailable, not that a severe course is coming.
Why is the haematocrit threshold different for men and women?#
Because normal haematocrit differs by sex. The score uses each sex's own normal range — below 43% for men and below 39.6% for women — so a haematocrit of 40% is normal in a man and abnormal in a woman. Using a single cut-off for both misclassifies women in the low 40s.
When should HAPS be calculated?#
At admission, ideally within the first half hour, since that is when the intensive care decision is being made and the parameters are all available. It is a snapshot and is not designed for serial use.
How does HAPS differ from BISAP or Ranson's?#
In direction and timing. BISAP and Ranson's predict mortality and severity, and Ranson's needs 48 hours to complete. HAPS identifies patients who can safely avoid intensive care and returns an answer roughly 30 minutes after arrival. They complement rather than replace one another.
Is HAPS enough on its own to decide against intensive care?#
It supports that decision but should not make it alone. It is one piece of information alongside clinical judgement, comorbidity and the aetiology — and it says nothing about coexisting cholangitis, which needs urgent treatment regardless.
What if two of the three parameters are normal?#
That is not a partial pass. HAPS is a conjunction rather than a sum, so all three must be normal for a harmless prediction. Any single abnormality returns 'not harmless'.
Does HAPS apply to a second or third attack of pancreatitis?#
It was studied in first attacks, and its performance in recurrent or chronic-on-acute pancreatitis is less well characterised. Use it with more caution outside the population it was derived in.
References
Original / primary reference
Reference standard and guidelines
- Banks PA, Bollen TL, Dervenis C, et al. Classification of acute pancreatitis — 2012: revision of the Atlanta classification and definitions by international consensus. Gut. 2013;62(1):102-111.
- Tenner S, Vege SS, Sheth SG, et al. American College of Gastroenterology Guidelines: Management of Acute Pancreatitis. Am J Gastroenterol. 2024;119(3):419-437.
- 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.