About the AIMS65 Score for Upper GI Bleeding Mortality
AIMS65 estimates the risk of in-hospital death in acute upper gastrointestinal bleeding from five equally weighted bedside criteria, each worth one point: Albumin below 3.0 g/dL, INR above 1.5, altered Mental status, Systolic blood pressure 90 mmHg or lower, and age 65 or older. The score runs 0 to 5 and every input is available at presentation, before endoscopy. Mortality climbs steeply with the score — about 0.3% at 0 and 24.5% at 5 in the derivation cohort — and two or more criteria is the usual high-risk threshold. It predicts death, length of stay and cost, not the need for transfusion or endoscopic treatment, which is what the Glasgow-Blatchford score is for.
Formula
AIMS65 = [Albumin < 3.0 g/dL] + [INR > 1.5] + [Altered mental status] + [Systolic BP ≤ 90 mmHg] + [Age ≥ 65]- Albumin < 3.0 g/dL
- 1 point if present, else 0.
- INR > 1.5
- 1 point if present, else 0.
- Altered mental status
- 1 point if present, else 0.
- Systolic BP ≤ 90 mmHg
- 1 point if present, else 0 (threshold inclusive).
- Age ≥ 65
- 1 point if present, else 0 (threshold inclusive).
- Every criterion is worth exactly one point; there is no weighting, so the score is simply the count of criteria met.
- The systolic BP and age thresholds are inclusive: 90 mmHg and 65 years each score.
- Score the values at presentation. Resuscitating the blood pressure or reversing the INR does not lower a score that was already positive on arrival.
Interpreting the result
Read AIMS65 as a mortality gauge. A score of 0 marks the lowest-risk group, with in-hospital mortality around 0.3% in the derivation data; risk rises with each additional point and climbs steeply from two criteria upward, reaching roughly a quarter of patients at the maximum of 5. Two or more is the commonly used high-risk cut-off that should prompt closer monitoring, early resuscitation and senior involvement. What the score does not do is tell you whether a patient can avoid admission or will need endoscopic therapy — for the discharge question, calculate the Glasgow-Blatchford score alongside it.
| Score | Band | What it means | Action |
|---|---|---|---|
| 0 | Low risk | In-hospital mortality ≈ 0.3% in the derivation cohort | Standard management; still calculate Glasgow-Blatchford before considering discharge |
| 1 | Intermediate risk | Mortality low but rising above the zero-criteria group | Admit and monitor; endoscopy within 24 hours |
| 2–5 | High risk | Mortality rises steeply — up to ≈ 24.5% at a score of 5 | Close/high-dependency monitoring, prompt resuscitation, early senior input, timely endoscopy |
What the AIMS65 Score needs (5 inputs)
- Albumin < 3.0 g/dL (the 'A')
- A low serum albumin is the single strongest of the five predictors in most series. It reflects chronic illness, malnutrition and frailty rather than the acute bleed, which is why it carries so much prognostic weight.
- INR > 1.5 (the 'I')
- Raised by anticoagulation and by liver disease alike — the score does not distinguish the cause. Correcting the INR does not retrospectively lower the score, which is scored on the presenting value.
- Altered mental status (the 'M')
- Glasgow Coma Scale below 14, or documented disorientation, lethargy, stupor or coma. It flags either haemodynamic compromise or a serious comorbidity such as encephalopathy.
- Systolic BP ≤ 90 mmHg (the 'S')
- The haemodynamic term, and the threshold is inclusive — exactly 90 mmHg scores the point. It is the criterion most likely to change with resuscitation, so record the presenting value.
- Age ≥ 65 years (the '65')
- Also inclusive — exactly 65 scores the point. Age is a blunt but reliable marker of the reduced physiological reserve that turns a survivable bleed into a fatal one.
What it returns
- AIMS65 score (0–5)
- The number of criteria present. It is a mortality risk estimate, not a rebleeding or intervention predictor.
- Risk band
- 0 low, 1 intermediate, and 2 or more high — the point at which mortality begins to rise steeply.
How it is calculated
AIMS65 was built by mining a large administrative database for the routinely available variables that best predicted in-hospital death after an upper GI bleed, then keeping the handful that were both strongly predictive and easy to capture. The result is an unweighted count rather than a regression equation — a deliberate trade of a little discrimination for a lot of usability. Because four of the five items (albumin, INR, mental status, age) index the patient's background state rather than the bleed, the score is really a measure of who is likely to survive the physiological insult, which is why it tracks mortality far better than it tracks rebleeding.
Facts & figures
| Letter | Criterion | Threshold |
|---|---|---|
| A | Albumin | < 3.0 g/dL (30 g/L) |
| I | INR | > 1.5 |
| M | Mental status | Altered (GCS < 14 or disorientation/lethargy/stupor/coma) |
| S | Systolic BP | ≤ 90 mmHg |
| 65 | Age | ≥ 65 years |
The mnemonic AIMS65 encodes both the variables and the age threshold.
| Question | Better tool |
|---|---|
| Will this patient die during admission? | AIMS65 |
| Can this patient avoid admission / intervention? | Glasgow-Blatchford |
| Needs no endoscopic therapy or transfusion (score 0)? | Glasgow-Blatchford |
| Available before endoscopy? | Both |
The two scores are complementary, not competing; many departments calculate both.
Evidence
Derivation — US administrative cohort
2011 · n = 29,222Developed by Saltzman and colleagues from a large multi-hospital US database of adults admitted with acute upper GI bleeding, screening routinely collected variables for those most predictive of in-hospital mortality and retaining albumin, INR, mental status, systolic blood pressure and age.
In-hospital mortality rose from about 0.3% at a score of 0 to 24.5% at a score of 5. The score also predicted length of stay and cost, and discriminated mortality with an area under the ROC curve around 0.77 in the derivation and validation sets.
International multicentre comparison
2017 · n = 3,012Prospective international study across multiple countries comparing the common pre-endoscopy scores head to head for a range of outcomes in upper GI bleeding.
AIMS65 was among the best discriminators of mortality, while the Glasgow-Blatchford score was clearly superior for predicting the composite need for intervention or the safe-discharge endpoint — confirming the division of labour between the two scores.
Comparison with Glasgow-Blatchford and Rockall
2016Independent cohorts comparing AIMS65 with the Glasgow-Blatchford and full Rockall scores for mortality, rebleeding and intervention.
AIMS65 matched or exceeded the other scores for predicting mortality but was consistently weaker than Glasgow-Blatchford for transfusion and intervention need, reproducing the pattern seen in the derivation.
How it compares
AIMS65 Score vs Glasgow-Blatchford Score
Use AIMS65 for the mortality question and Glasgow-Blatchford for the discharge/intervention question — they are complementary, and a low AIMS65 is not a safe-discharge signal.
Glasgow-Blatchford was built to identify patients who need no intervention and might avoid admission, and it discriminates transfusion, endoscopic therapy and rebleeding better than AIMS65. AIMS65 was built to predict in-hospital death and does that at least as well with fewer inputs and no urea. Head-to-head studies consistently show each winning its own outcome, which is why many services calculate both at presentation.
AIMS65 Score vs ABC Score
The newer ABC score generally out-discriminates AIMS65 for 30-day mortality and works in lower as well as upper GI bleeding, at the cost of a few more inputs and explicit comorbidity terms.
ABC (age, blood tests, comorbidities) was derived after AIMS65 specifically to improve mortality prediction and to span both upper and lower GI bleeding. In its derivation and several validations it showed higher discrimination for 30-day mortality than AIMS65, though AIMS65 remains attractive for its brevity and its single, memorable mnemonic.
AIMS65 Score vs Pre-endoscopy Rockall Score
Both are pre-endoscopy mortality tools and perform comparably; AIMS65 avoids the shock and comorbidity categories that Rockall requires, using labs and mental status instead.
Pre-endoscopy (clinical) Rockall uses age, shock and comorbidity for a maximum of 7, while AIMS65 substitutes albumin, INR and mental status for the shock and comorbidity judgements. Discrimination for mortality is similar in most comparisons, so the choice usually comes down to which inputs are fastest to obtain in a given department.
Pearls & pitfalls
- AIMS65 predicts mortality, not rebleeding or the need for intervention — do not use a low score to justify discharge; that is Glasgow-Blatchford's job.
- Score the presenting values. A systolic BP that was 88 mmHg on arrival still scores even after fluids bring it up, and an INR reversed with vitamin K or PCC still scores on its admission value.
- The systolic BP (≤ 90) and age (≥ 65) thresholds are inclusive — exactly 90 and exactly 65 each score a point.
- Albumin is often the heaviest single contributor and reflects chronic illness; a normal albumin in an acutely bleeding patient is reassuring for prognosis even when the bleed looks brisk.
- It was derived largely in non-variceal bleeding populations; use judgement when applying it to known variceal haemorrhage.
- A score of 0 is low risk but not zero risk — clinical deterioration always overrides the number.
Critical actions
- Calculate the Glasgow-Blatchford score in parallel whenever the question is admission or intervention rather than mortality.
- Treat a score of 2 or more as high risk: resuscitate, arrange a monitored bed, and involve senior clinicians early.
- Offer endoscopy within 24 hours of presentation for admitted patients, earlier if haemodynamically unstable, regardless of the AIMS65 band.
- Correct reversible coagulopathy and transfuse to a restrictive target, but do not expect these to change the recorded score.
- Reassess clinically over time — AIMS65 is a snapshot at presentation and does not update itself as the patient evolves.
Why this score exists
The score's authors set out to make a mortality tool that any clinician could apply from data already in the chart, and they accepted a simple unweighted count over a more discriminating but fiddlier equation precisely so it would actually be used at the bedside. Its centre of gravity is the patient rather than the lesion — albumin, INR, age and mental status describe who is sick enough to die — which is the reason it should never be read as a statement about whether the bleeding will stop.
About the creator
First author, 2011 derivation study
Derived AIMS65 from a large administrative dataset, deliberately restricting it to five variables available before endoscopy.
Senior author
Contributed the clinical-informatics approach behind the score's derivation from routinely captured hospital data.
Limitations
- It predicts mortality only; it is a poor discriminator of rebleeding, transfusion need and the need for endoscopic intervention.
- As an unweighted count it sacrifices some discrimination relative to a fully weighted model.
- It was derived from administrative data in a predominantly non-variceal population, so its calibration in variceal bleeding is less certain.
- It does not incorporate the bleeding source, ongoing blood loss or the response to resuscitation, all of which matter clinically.
- A low score is reassuring about prognosis but does not authorise discharge — that decision needs a purpose-built score plus clinical judgement.
If you are the patient
AIMS65 is a quick score doctors work out soon after someone arrives with bleeding from the upper gut (the stomach or gullet). It adds up five simple facts — a blood protein called albumin being low, a clotting test (INR) being high, being confused or drowsy, having a low blood pressure, and being 65 or older — giving a number from 0 to 5. The higher the number, the greater the risk of becoming seriously unwell during the hospital stay, so it helps the team decide how closely to watch the patient and how quickly to get senior help. It is a measure of overall risk to life, not a prediction of whether the bleeding will start again, and a low score does not on its own mean it is safe to go home — a different score is used for that.
Frequently asked questions
What is the AIMS65 score?#
AIMS65 is a five-part bedside score for acute upper GI bleeding that predicts in-hospital mortality. Each criterion — Albumin below 3.0 g/dL, INR above 1.5, altered Mental status, Systolic BP 90 mmHg or lower, and age 65 or older — scores one point, for a total of 0 to 5.
What is a high-risk AIMS65 score?#
A score of 2 or more is generally treated as high risk, because mortality begins to rise steeply from two criteria upward — from about 0.3% at a score of 0 to roughly 24.5% at a score of 5 in the original study. A high score warrants close monitoring and early senior involvement.
Is AIMS65 better than the Glasgow-Blatchford score?#
Neither is uniformly better — they answer different questions. AIMS65 predicts death more directly and with fewer inputs, while the Glasgow-Blatchford score is superior for identifying patients who need transfusion or endoscopic treatment, or who could be discharged. Many departments calculate both.
Do I need endoscopy results to calculate AIMS65?#
No. All five inputs — albumin, INR, mental status, systolic blood pressure and age — are available at presentation, which is one of the score's main advantages. It is a pre-endoscopy tool.
Does AIMS65 predict rebleeding?#
Not well. It was designed and validated for mortality, and across studies it discriminates rebleeding and the need for intervention noticeably less well than the Glasgow-Blatchford score. Use it for prognosis, not as a rebleeding forecast.
Should the values be the ones at arrival or after treatment?#
Use the values at presentation. A blood pressure that was 90 mmHg or below on arrival still scores even after resuscitation, and an INR above 1.5 on admission still scores even once it has been corrected.
References
Original / primary reference
Validation and comparison studies
- Hyett BH, Abougergi MS, Charpentier JP, et al. The AIMS65 score compared with the Glasgow-Blatchford score in predicting outcomes in upper GI bleeding. Gastrointest Endosc. 2013;77(4):551-557.
- Stanley AJ, Laine L, Dalton HR, et al. Comparison of risk scoring systems for patients presenting with upper gastrointestinal bleeding: international multicentre prospective study. BMJ. 2017;356:i6432.
- Robertson M, Majumdar A, Boyapati R, et al. Risk stratification in acute upper GI bleeding: comparison of the AIMS65 score with the Glasgow-Blatchford and Rockall scoring systems. Gastrointest Endosc. 2016;83(6):1151-1160.
Clinical practice guidelines
- Laine L, Barkun AN, Saltzman JR, Martel M, Leontiadis GI. ACG Clinical Guideline: Upper Gastrointestinal and Ulcer Bleeding. Am J Gastroenterol. 2021;116(5):899-917.
- Gralnek IM, Stanley AJ, Morris AJ, et al. Endoscopic diagnosis and management of nonvariceal upper gastrointestinal hemorrhage (NVUGIH): ESGE Guideline – Update 2021. Endoscopy. 2021;53(3):300-332.