Risk Stratification for In-Hospital Mortality in Upper GI Bleeding
Five-parameter clinical score predicting in-hospital mortality in upper gastrointestinal bleeding. Simpler than the Glasgow-Blatchford Score with a focus on mortality rather than intervention need. Validated for variceal and non-variceal UGIB.
Check all criteria present at time of assessment:
| Letter | Criterion | Threshold | Points |
|---|---|---|---|
| A | Albumin | <3.0 g/dL | 1 |
| I | INR | >1.5 | 1 |
| M | Mental Status | GCS ≤14 or disorientation | 1 |
| S | Systolic BP | ≤90 mmHg | 1 |
| 65 | Age | ≥65 years | 1 |
Total score ranges from 0 to 5. Each criterion present adds 1 point.
| Score | In-Hospital Mortality | ICU Admission | Recommendation |
|---|---|---|---|
| 0 | ~0.3% | ~3% | Discharge after endoscopy |
| 1 | ~1% | ~11% | Outpatient or short stay |
| 2 | ~3% | ~19% | Inpatient, early endoscopy |
| 3 | ~9% | ~26% | HDU/ICU, urgent endoscopy |
| 4 | ~14% | ~40% | ICU, immediate endoscopy |
| 5 | ~25% | >50% | ICU, immediate intervention |
AIMS65 vs Glasgow-Blatchford Score (GBS): These two scores are complementary tools that serve different purposes. GBS predicts the need for clinical intervention (transfusion, endoscopic therapy, surgery) and is best used to identify patients who can be safely managed as outpatients. AIMS65 predicts in-hospital mortality and ICU admission, making it valuable for triaging sicker patients to appropriate levels of care. Use GBS when the question is "does this patient need intervention?" and AIMS65 when the question is "how likely is this patient to die in hospital?"
Score 0–1: Very low in-hospital mortality. These patients are safe candidates for outpatient management or early discharge following endoscopy, provided endoscopic findings are low-risk (Forrest IIb or III lesions, no high-risk stigmata).
Score ≥3: Requires ICU-level care. Mortality risk is clinically significant (≥9%). These patients should be resuscitated aggressively, monitored in HDU or ICU, and undergo endoscopy urgently after haemodynamic stabilisation.
Variceal vs non-variceal UGIB: AIMS65 has been validated in both populations. In cirrhotic patients with variceal bleeding, AIMS65 score may overlap with Child-Pugh and MELD; combining scores may provide complementary prognostic information.
Simplicity advantage: All five AIMS65 parameters are available at initial triage without endoscopy. This makes it easier to apply than the post-endoscopy Rockall score or the full pre-endoscopy GBS (which requires multiple laboratory and clinical parameters).
Original derivation: Derived from 29,222 hospital admissions for UGIB across 187 US hospitals. The score was designed specifically to predict in-hospital mortality as the primary endpoint, with ICU transfer as a secondary endpoint.
Discrimination: AUC for in-hospital mortality was 0.77 in the original derivation cohort, outperforming the full Rockall score (AUC 0.72) in independent validation. The simplified five-variable design without endoscopic data maintains comparable performance to more complex scores.
Multiple external validations have been performed across Asian, European, and South American cohorts, consistently confirming the score's predictive validity across different healthcare settings and UGIB aetiologies.