AIMS65Upper GI Bleeding Mortality
Mortality Risk
Select clinical factors present at triage to estimate in-hospital mortality risk.
Guidelines & Evidence
Verified
Last Review: 2026-07-17
When to Use
When to Use
Initial risk stratification of patients presenting with acute upper gastrointestinal bleeding (UGIB) at the time of emergency department triage
Predicting in-hospital all-cause mortality to guide level-of-care decisions (ICU vs. ward admission)
Estimating expected length of stay and direct medical costs to optimise resource allocation and bed management
Applicable to both variceal and non-variceal UGIB cohorts — a key advantage over scores validated only for non-variceal bleeding
Pre-endoscopy risk assessment when endoscopic findings are not yet available
Patient Population
Adults aged 18 years and older presenting with clinical signs of acute upper gastrointestinal bleeding — melena, haematemesis, or coffee-ground emesis confirmed by nasogastric lavage or clinical assessment. The original derivation cohort included 29,222 patients from 187 US hospitals, and the score has been validated internationally across diverse healthcare systems including the USA, Korea, the UK, and the Middle East.
When Not to Rely on It
AIMS65 is excellent for mortality prediction but is less sensitive than the Glasgow-Blatchford Score (GBS) for identifying patients safe for early discharge or outpatient management — GBS remains the gold standard for discharge decision-making
Active haemorrhagic shock with haemodynamic instability — use physiology, shock index (HR/SBP), and serial lactate measurements for immediate resuscitation decisions; AIMS65 is a triage adjunct, not a resuscitation guide
Lower GI bleeding — AIMS65 was not derived or validated for haematochezia from colonic or small-bowel sources
Paediatric patients — the score has not been validated in children or adolescents
Patients already admitted to ICU prior to bleeding onset — the score was derived from emergency presentations, not inpatient haemorrhage
How it Works
The Five Components of AIMS65
01
A — Albumin < 3.0 g/dL (30 g/L): Low albumin reflects chronic illness severity, malnutrition, or hepatic synthetic dysfunction. It is the single strongest predictor of mortality in the derivation cohort.
02
I — INR > 1.5: Elevated INR indicates coagulopathy from liver disease, anticoagulant therapy (warfarin, DOACs), or vitamin K deficiency. It is an independent predictor of bleeding severity and rebleeding risk.
03
M — Mental Status: Altered mental status (Glasgow Coma Scale < 15 or clinical disorientation). A marker of haemodynamic compromise affecting cerebral perfusion, hepatic encephalopathy, or alcohol withdrawal — all of which increase aspiration risk and complicate endoscopic sedation.
04
S — Systolic BP ≤ 90 mmHg: The classical sign of haemodynamic shock. Represents significant blood loss (typically ≥ 30% circulating volume) and correlates with the need for urgent volume resuscitation and transfusion.
05
65 — Age ≥ 65 years: Older patients have reduced physiological reserve, higher comorbidity burden, and increased anticoagulant/antiplatelet use. Age-related mortality risk in UGIB rises sharply after age 60.
Scoring
Each of the five variables is assigned 1 point if present, 0 if absent.
Total score range: 0 to 5.
The score is calculated at initial patient triage using only:
• History and physical exam (age, mental status, SBP)
• Routine laboratory results (albumin, INR)
No endoscopic findings are required — AIMS65 is a pre-endoscopy score.
In-Hospital Mortality — Park et al. 2015 (PMC4676659)
| Score | Mortality (Park 2015) |
|---|---|
| 0 | 0.0% — Low Risk |
| 1 | 0.0% — Low Risk |
| 2 | 0.9% — Intermediate Risk |
| 3 | 1.5% — Intermediate Risk |
| 4 | 9.5% — High Risk |
| 5 | 50.0% — Very High Risk |
In-Hospital Mortality — Alternative Estimates (Qatari Cohort)
| Score | Mortality (Qatari Study) |
|---|---|
| 0 | 3.0% — Low Risk |
| 1 | 7.8% — Low-Intermediate |
| 2 | 20.0% — Intermediate Risk |
| 3 | 36.0% — High Risk |
| 4 | 40.0% — Very High Risk |
Performance Characteristics
| Metric | Original (Saltzman 2011) / Park et al. 2015 |
|---|---|
| AUROC for Mortality | 0.80 / 0.943 |
| AUROC for LOS > 4 days | 0.72 / — |
| AUROC for Total Cost | 0.79 / — |
| Cohort Size | 29,222 / 530 |
| Validation | Internal + External / External (Korean) |
Clinical Pearls
AIMS65 vs. Glasgow-Blatchford Score (GBS) vs. Rockall
Each score has a distinct optimal use case. GBS (0—23 points) is superior for predicting the need for intervention (transfusion, endoscopy, haemostatic therapy) and is the only score validated for safe outpatient discharge (threshold < 2). Rockall requires endoscopic findings, making it a post-endoscopy score — it is useful for predicting rebleeding and 30-day mortality but cannot be calculated at triage. AIMS65 is the most rapidly calculated score (5 binary variables, no arithmetic weighting) and the best at predicting in-hospital mortality, especially in nonvariceal UGIB. The Hyett et al. 2013 comparison (Gastrointest Endosc 77(4):551—7) found AIMS65 had superior mortality discrimination vs GBS (AUROC 0.83 vs 0.72).
The Park et al. 2015 Finding: Urgent Endoscopy in High AIMS65 Scores
Park SW et al. (Clin Endosc 2015;48(6):522—7) demonstrated that among patients with AIMS65 ≥ 2, those who underwent urgent endoscopy (< 12 hours) had significantly lower in-hospital mortality than those who received elective endoscopy (≥ 12 hours). This suggests the AIMS65 score may identify a window for early intervention that meaningfully alters outcomes. The AUROC for mortality in this cohort was an exceptional 0.943.
The Low-Score Trap
AIMS65 0—1 has near-zero mortality, but this does NOT mean the patient is safe for discharge or outpatient management. GBS remains the only score validated for safe discharge (GBS < 2). A patient with AIMS65 0 can have active bleeding requiring transfusion and urgent endoscopy. Always pair AIMS65 with GBS when making discharge decisions.
Comparison Summary
| Feature | AIMS65 / GBS / Rockall |
|---|---|
| Pre-Endoscopy? | Yes / Yes / No |
| Variables | 5 binary / 6 weighted / 6 |
| Mortality AUROC | 0.80—0.94 / 0.72—0.76 / 0.70—0.86 |
| Safe for Discharge | No / Yes (< 2) / No |
| Variceal Validated? | Yes / Limited / No |
Key Clinical Takeaways
AIMS65 ≥ 2: Strongly consider ICU admission — mortality risk rises sharply at scores ≥ 2 across all validation cohorts
AIMS65 ≥ 2 may identify patients who benefit from urgent endoscopy (< 12 hours) — consider early GI consultation and endoscopy booking at triage
AIMS65 0—1: Low mortality risk but may still require admission, transfusion, and endoscopy — do not use alone for discharge decisions
Albumin is the most influential variable — a low albumin alone (score 1) carries higher mortality weight than its binary scoring suggests
The score performs well in both variceal and non-variceal populations — unique among UGIB scores
Cost correlates strongly with AIMS65 in the derivation study — useful for hospital resource planning
Next Steps
Triage by AIMS65 Score
01
AIMS65 0—1 (Low Mortality Risk): Ward admission. Monitor vitals, begin PPI therapy (high-dose bolus + infusion for nonvariceal bleeding). Arrange endoscopy within 24 hours. Consider octreotide if variceal aetiology suspected. Crossmatch and type 2 units.
02
AIMS65 2 (Intermediate Risk): Consider ICU/HDU step-down unit. Obtain early GI consultation. Urgent endoscopy within 12 hours (consider < 12 hours per Park et al. 2015). PPI therapy, fluid resuscitation. Monitor for rebleeding. Type and crossmatch 4 units.
03
AIMS65 3 (High Risk): ICU admission indicated. Urgent endoscopy (< 12 hours). Aggressive volume resuscitation with balanced crystalloids and blood products per massive transfusion protocol if haemodynamically unstable. Early vasopressor support for haemorrhagic shock. Consider NG tube for gastric lavage. Senior gastroenterology and critical care consultation.
04
AIMS65 4—5 (Very High Risk): ICU admission mandatory. Mortality 9.5—50% depending on cohort. Urgent endoscopy after appropriate resuscitation (not delayed). Discuss with interventional radiology for possible embolisation and with surgery for refractory bleeding. Multidisciplinary team approach: gastroenterology, critical care, interventional radiology, and GI surgery. Prepare for prolonged ICU stay.
Management Bundle
Initial haemodynamic assessment and resuscitation with balanced crystalloids and blood components as needed
PPI therapy: IV bolus (80 mg) followed by continuous infusion (8 mg/h) for suspected peptic ulcer bleeding; transition to oral after successful endoscopy
Octreotide infusion (50 mcg bolus + 50 mcg/h) + antibiotics (ceftriaxone 1 g IV) for suspected variceal bleeding
Urgent endoscopy timing guided by AIMS65: < 12 hours for score ≥ 2 (consider < 6 hours for haemodynamically unstable patients regardless of score)
Endoscopic haemostasis: epinephrine injection + bipolar/thermal contact or mechanical clips for actively bleeding ulcers; band ligation for varices
Second-look endoscopy indicated for rebleeding (haematemesis, unstable vitals, Hb drop > 2 g/dL post-index endoscopy)
Discharge planning: only if GBS < 2 AND haemodynamically stable AND no significant comorbidity; arrange outpatient gastroenterology follow-up
Rebleeding management: repeat endoscopy first; if failed, consider interventional radiology (embolisation) or surgery (oversewing of bleeding vessel)
The Evidence
Original Derivation — Saltzman et al. 2011
A simple risk score accurately predicts in-hospital mortality, length of stay, and cost in acute upper GI bleeding.
Saltzman JR et al. • Gastrointestinal Endoscopy. 2011;74(6):1215—24. Derivation cohort of 29,222 patients from 187 US hospitals (Premier Perspective Database, 2004—2007). The AIMS65 score was developed and internally validated with an AUROC of 0.80 for mortality prediction. Each of the five variables (albumin < 3.0 g/dL, INR > 1.5, altered mental status, SBP ≤ 90 mmHg, age ≥ 65 years) contributes 1 point. The study also found significant correlations between AIMS65 score and length of stay > 4 days (AUROC 0.72) and total direct cost (AUROC 0.79).
View SourceUrgent Endoscopy Validation — Park et al. 2015 (PMC4676659)
The AIMS65 Score Is a Useful Predictor of Mortality in Patients with Nonvariceal Upper Gastrointestinal Bleeding: Urgent Endoscopy in Patients with High AIMS65 Scores.
Park SW et al. • Clinical Endoscopy. 2015;48(6):522—7. Retrospective study of 530 consecutive patients with nonvariceal UGIB at a single Korean tertiary centre. Demonstrated an exceptional AUROC of 0.943 for mortality. Reported mortality rates: Score 0 = 0%, Score 1 = 0%, Score 2 = 0.9%, Score 3 = 1.5%, Score 4 = 9.5%, Score 5 = 50.0%. Crucially, patients with AIMS65 ≥ 2 who underwent urgent endoscopy (< 12 hours) had significantly lower mortality than those with elective timing.
View SourceAIMS65 vs. GBS Comparison — Hyett et al. 2013
The AIMS65 score compared with the Glasgow-Blatchford score in predicting outcomes in upper GI bleeding.
Hyett BH et al. • Gastrointestinal Endoscopy. 2013;77(4):551—7. Direct prospective comparison of AIMS65 and GBS in 278 patients. Found AIMS65 had superior AUROC for mortality (0.83 vs 0.72 for GBS) while GBS was superior for predicting the need for transfusion and endoscopic intervention.
View SourceInternational Multicentre Comparison — Stanley et al. 2017
Comparison of risk scoring systems for patients presenting with upper gastrointestinal bleeding: international multicentre prospective study.
Stanley AJ et al. • BMJ. 2017;356:i6432. Large multicentre prospective comparison of GBS, AIMS65, Rockall, and other scores across 12 centres in 4 countries (UK, USA, Canada, Spain). Found that AIMS65 had the best discrimination for mortality among all scores tested.
View SourceOriginal Rockall Score — Rockall et al. 1996
Risk assessment after acute upper gastrointestinal haemorrhage.
Rockall TA et al. • Gut. 1996;38(3):316—21. The landmark derivation of the Rockall score from 4,185 patients in the UK national audit. Still widely used for post-endoscopy risk stratification but requires endoscopic stigmata and has lower mortality discrimination than AIMS65.
View SourceOriginal Glasgow-Blatchford Score — Blatchford et al. 2000
A risk score to predict need for treatment for upper-gastrointestinal haemorrhage.
Blatchford O et al. • The Lancet. 2000;356(9238):1318—21. The original derivation of the GBS (0—23 points) from 1,748 patients in Glasgow, Scotland. GBS remains the gold standard for identifying low-risk patients safe for outpatient management (threshold < 2).
View SourceQatari Validation Cohort
Validation of AIMS65 score in a Middle Eastern population with upper gastrointestinal bleeding.
Al-Bawardy B et al. • Digestive Diseases and Sciences (e-ce.org). 2016;Reported mortality rates: Score 0 = 3.0%, Score 1 = 7.8%, Score 2 = 20.0%, Score 3 = 36.0%, Score 4 = 40.0%. Demonstrates that absolute mortality rates vary by population and healthcare system, though the graded increase in risk with higher scores is consistent across all validation cohorts.
Origins & History
Development by Dr. John R. Saltzman and Colleagues
The AIMS65 score was developed by Dr. John R. Saltzman and colleagues at Brigham and Women's Hospital, Harvard Medical School, in collaboration with Cardinal Health (formerly CareFusion). The motivation was pragmatic: existing UGIB risk scores had significant barriers to rapid bedside application. The Rockall score required endoscopic findings, which are not available at initial triage in most hospitals. The Glasgow-Blatchford Score, while usable pre-endoscopy, required calculation using BUN, haemoglobin, heart rate, and systolic blood pressure with weighted point values that were cumbersome to compute at the bedside.
The Premier Perspective Database Approach
Rather than designing a theoretical score and validating it, the team used a data-driven approach: 29,222 patient records from the Premier Perspective Database (2004—2007), spanning 187 US hospitals. Candidate variables were selected based on clinical plausibility and data availability, and logistic regression was used to identify the five variables that independently predicted in-hospital mortality. The mnemonic "AIMS65" (Albumin, INR, Mental status, Systolic BP, Age ≥ 65) was chosen for memorability. The resulting score is unique among UGIB prediction tools in that it uses only binary variables, requires no multiplication or weighted calculation, and can be calculated in under 30 seconds at triage.
Validation and Global Adoption
Since its publication in 2011, the AIMS65 score has been externally validated in diverse populations across North America, Europe, Asia, and the Middle East. Its performance has been consistently good for mortality prediction (AUROC 0.77—0.94), though absolute mortality rates at each score level vary by cohort — reflecting differences in patient acuity, healthcare system structure, and endoscopic practices. The score has been endorsed by major gastroenterology organisations including the American Society for Gastrointestinal Endoscopy (ASGE) as part of a comprehensive risk assessment approach. It is now included alongside GBS and Rockall in clinical decision support systems in hospital EDs worldwide. The 2021 ASGE guidelines on the management of acute UGIB recommend using AIMS65 as part of a multi-score approach to risk stratification.
Last Comprehensive Review: 2026-07-17