GI bleed scores

Range 0–23. The original score uses urea in mmol/L. BUN (mg/dL) = urea (mmol/L) × 2.8, so 6.5 / 8 / 10 / 25 mmol/L become 18.2 / 22.4 / 28 / 70 mg/dL. The calculator stores the value you type, converts it when you toggle units (no rounding drift), and applies the mg/dL cut-offs to the unrounded value. Ticking suspected cirrhosis also scores hepatic disease (+2).

Very low risk = GBS 0–1. ACG 2021 defines very low risk as a score with a ≤1% false-negative rate for hospital-based intervention or death. It suggests discharge with outpatient follow-up rather than admission (conditional). The International Consensus Group (2019) suggests the same cut-off: these patients may not need hospitalization or inpatient endoscopy. The GBS predicts the need for intervention, not the source of bleeding. It does not replace endoscopy; it decides whether endoscopy can be outpatient.

1 point each: Albumin <3.0 g/dL · INR >1.5 · altered Mental status · SBP ≤90 mmHg · age older than 65. Saltzman 2011 states the age criterion as "older than 65"; some calculators use ≥65.

In-hospital mortality was 0.3% with 0 points and 24.5–31.8% with all 5 (validation and derivation cohorts). AIMS65 is built to flag high mortality risk. The International Consensus Group suggests not using it to pick patients for discharge: even at low cut-offs it misclassifies about 20% of high-risk patients as low risk.

Complete (post-endoscopy) Rockall adds diagnosis — Mallory-Weiss tear, no lesion identified and no stigmata of recent haemorrhage 0, all other diagnoses 1, upper GI malignancy 2 — and stigmata — none or dark spot only 0, blood in the upper GI tract, adherent clot, visible or spurting vessel 2 (maximum 11).

ACG 2021 suggests twice-daily PPI to 2 weeks for "high-risk" ulcer patients without setting a score cut-off. The single trial behind that suggestion enrolled patients with a complete Rockall ≥6, so the tool shows that value for context only. The pre-endoscopy score misclassifies 4–7% of high-risk patients. The International Consensus Group made no recommendation for or against using it to decide on discharge.

• Very low risk (GBS 0–1) without instability or suspected cirrhosis: discharge with outpatient follow-up and outpatient endoscopy rather than admission (conditional). No inpatient drugs are listed for this path.
• Transfusion: restrictive, at Hb 7 g/dL (conditional). Hb 8 g/dL is reasonable with pre-existing cardiovascular disease. Transfuse before Hb reaches 7 if the patient is exsanguinating or hemodynamically unstable. ICG 2019 suggests <8 g/dL without cardiovascular disease and a higher threshold with it.
• Erythromycin: 250 mg IV before endoscopy (conditional). It improves the view and reduces repeat endoscopy and length of stay. Avoid with QT prolongation.
• PPI before endoscopy: ACG 2021 could not recommend for or against. ICG 2019 (statement A9, unchanged from 2010): it may downstage the lesion and reduce the need for endoscopic therapy. ESGE 2026 suggests considering high-dose IV PPI. ICG 2019 and ESGE 2026: it must not delay endoscopy.
• Endoscopy: within 24 h of presentation for patients admitted or under observation (ACG 2021, ICG 2019, ESGE 2021). ESGE 2026 does not recommend endoscopy within 6 or 12 h unless the patient stays unstable despite resuscitation. Do not delay it because of anticoagulation (ICG 2019).
• PPI after endoscopic hemostasis of a high-risk ulcer (strong): high dose for 3 days — 80 mg IV bolus then 8 mg/h, or 80 mg bolus then 40 mg 2–4 times daily (oral if tolerated). Then twice daily until 2 weeks after the index endoscopy for high-risk patients (conditional; the supporting trial enrolled complete Rockall ≥6), then once daily.
• Rebleeding: repeat endoscopic therapy first; over-the-scope clips are an option for recurrent ulcer bleeding (ACG 2021 #12; ESGE 2026 #16). If that fails, transcatheter arterial embolization, then surgery. ESGE 2026 (#14) suggests considering prophylactic embolization in selected high-risk ulcers (e.g., instability at presentation, posterior duodenal wall, ulcer larger than 2 cm).
• Metoclopramide: ACG 2021 found the evidence scant and did not support it; ESGE 2026 suggests it IV in selected severe or ongoing bleeds when erythromycin is unavailable.
• H. pylori: test every bleeding peptic ulcer, treat if positive and confirm eradication. Repeat a negative test taken during the acute bleed (ICG 2019 D5–D6).
• No tranexamic acid: in HALT-IT (12,009 patients with GI bleeding), death from bleeding was 4% on both arms (RR 0.99), with more venous thromboembolism (0.8% vs 0.4%).

• Vasoactive drug as soon as variceal bleeding is suspected, continued 2–5 days or until TIPS. Baveno VIII: a shorter course (even 24 h) can be considered after successful endoscopic hemostasis if a non-selective β-blocker is then started. Octreotide: 50 µg IV bolus, then 25–50 µg/h (AASLD Table 5).
• Antibiotic prophylaxis at presentation (strong). A third-generation cephalosporin is generally preferred, adjusted to local resistance (Baveno VIII). AASLD: ceftriaxone 1 g every 24 h is the most commonly used agent, up to 5 days, stopped once bleeding is controlled if there is no infection. Baveno VIII: duration and need can be individualised — Child-Pugh A patients may need a shorter course or none (weak).
• Transfusion: restrictive, about 7 g/dL unless comorbidities call for more (AASLD 2024); Baveno VIII gives 7–8 g/dL individualized for cardiovascular disease, age, hemodynamics and ongoing bleeding (5.4). With cardiovascular disease the tool uses 8 g/dL — a tool convention within that range. FFP is not routinely recommended; do not transfuse FFP or platelets to numerical INR or platelet targets. Platelets or fibrinogen may be considered if bleeding cannot be controlled (5.24).
• Airway: intubate before endoscopy if consciousness is altered or there is active hematemesis; extubate as soon as possible.
• Erythromycin 250 mg IV, 30–90 min before endoscopy, if the QT is not prolonged (Baveno VIII, weak). AASLD gives 125–250 mg, 30–120 min before.
• Endoscopy within 12 h of presentation after resuscitation; as soon as possible if unstable. Band ligation for esophageal varices.
• Lactulose (oral or enema) is recommended to prevent or treat overt hepatic encephalopathy (Baveno VIII, strong).
• Pre-emptive TIPS (covered stent) within 72 h, ideally <24 h, for esophageal or GOV-1 bleeding with Child-Pugh C 10–13, Child-Pugh B >7 with active bleeding at endoscopy, or HVPG ≥20 mmHg. Missed the window: TIPS within 2 weeks (C 10–13) or 1 week (B >7 with active bleeding) may still help. ACLF, overt hepatic encephalopathy, hyperbilirubinemia, MELD score or severe alcohol-related hepatitis are not absolute contraindications (Baveno VIII 5.33).
• Uncontrolled bleeding: intubate; covered esophageal stent (preferred where available; not FDA approved in the US) or balloon tamponade (24 h maximum) as a bridge; salvage TIPS (Baveno VIII 5.35–5.37; AASLD 34–35). Rebleeding within 5 days: consider rescue TIPS (repeat endoscopy is an alternative within 5 days) (AASLD 35).
• GOV-2 or IGV-1: cyanoacrylate or EUS-guided hemostasis, then TIPS (± embolization) as soon as possible; multidisciplinary plan (Baveno VIII 5.39–5.41). AASLD 2024: cyanoacrylate (not FDA approved in the US for this use), TIPS or balloon-occluded retrograde transvenous obliteration (BRTO) are first-line options (45); TIPS may be preferred with MELD-Na below 20 and no encephalopathy, BRTO with encephalopathy or MELD-Na above 20.
• No tranexamic acid or recombinant factor VIIa (Baveno VIII 5.25, strong).
• Anticoagulants are paused until bleeding is controlled; restart timing is individualized to the strength of the indication (Baveno VIII 5.26).
• Stop a PPI started before endoscopy once a portal-hypertensive source is confirmed, unless there is a strong indication. Manage in intensive or intermediate care. Image (CT or MRI) for splanchnic vein thrombosis and HCC.

Range 0–35. A score ≤8 predicts a 95% probability of safe discharge (no rebleeding, transfusion, hemostatic intervention, death or 28-day readmission). ACG 2023 (recommendation 1, conditional): risk tools such as Oakland ≤8 identify low-risk patients appropriate for early discharge and outpatient diagnostic evaluation; they supplement clinical judgment and do not replace it. BSG 2019 (secondary): calls ≤8 a minor bleed for urgent outpatient investigation and >8 a major bleed. The score was derived with liberal transfusion and has been questioned in populations with high baseline anemia.

• Ongoing, hemodynamically significant hematochezia: CT angiography as the first test (recommendation 7, conditional). It has low yield once bleeding is minor or has stopped. CTA showing extravasation: prompt referral to interventional radiology for angiography and possible embolization (recommendation 8, strong). If the patient is unstable and suspicion of an upper source is high, do upper endoscopy (key concept 3). ACG–SAR 2024: CTA first if unstable; could be considered first in a stable patient with suspected active bleeding; not first-line once bleeding has subsided. BSG 2019 (secondary): if CTA finds no source in an unstable patient, upper endoscopy immediately.
• Shock index (HR ÷ SBP) ≥1 helps predict active bleeding on CTA but has not been shown to be strongly predictive of clinical outcomes (ACG 2023). The tool uses it to flag hemodynamics for review, never as a discharge criterion. BSG 2019 defines instability as a shock index above 1.
• Colonoscopy: for most hospitalized patients (recommendation 6a, strong), done nonurgently; urgent colonoscopy within 24 h has not improved rebleeding or mortality (recommendation 9, strong).
• Transfusion thresholds, not targets: 7 g/dL in hemodynamically stable patients (recommendation 2, conditional); 8 g/dL is used with pre-existing cardiovascular disease (ACG 2023 text). Shock or significant ongoing bleeding are exceptions. BSG 2019 also sets post-transfusion targets; this tool does not use them.
• Warfarin: most patients do not need reversal. Reverse only a life-threatening bleed with an INR substantially above the therapeutic range, with 4-factor PCC rather than FFP (recommendation 3, conditional). No routine vitamin K (ACG-CAG 2022). DOACs: reversal only for the small subset with a life-threatening bleed that does not respond to resuscitation and stopping the drug, last dose within 24 h (recommendation 4, conditional): idarucizumab for dabigatran. ACG 2023 named andexanet for apixaban or rivaroxaban, but it has not been sold in the US since 22 Dec 2025 (FDA); ACG-CAG 2022 allows selective PCC — follow the local protocol.
• No tranexamic acid (recommendation 5, strong).
• Aspirin: hold for primary prevention; ACG 2023 suggests stopping it after diverticular bleeding (11b). Continue secondary-prevention aspirin during admission if possible (key concept 6; 11c). DAPT: continue aspirin and hold the P2Y12 inhibitor initially in severe bleeding; with a stent in the past year, decide jointly with cardiology and restart within 5 days at most.
• Anticoagulants: resume after bleeding stops (recommendation 12, strong), typically within 7 days.

• Warfarin: suggest against FFP and against vitamin K. If reversal is needed, PCC is suggested over FFP. No recommendation was possible for PCC versus no PCC.
• DOACs: suggest against PCC. Dabigatran: suggest against idarucizumab. Rivaroxaban or apixaban: suggest against andexanet alfa — which has not been sold in the US since 22 Dec 2025 (FDA).
• Antiplatelets: suggest against platelet transfusion. Do not hold aspirin given for secondary cardiovascular prevention. If it was stopped, restart on the day endoscopic hemostasis is confirmed.
• Selective use: warfarin — PCC can be considered for a life-threatening bleed, or an INR well above range, or when massive transfusion is undesirable. DOACs — for a life-threatening bleed with a dose within the past 24 h: idarucizumab for dabigatran; selective PCC for factor Xa inhibitors (andexanet withdrawn from the US, Dec 2025).
• Restarting: resume anticoagulation once bleeding is controlled, preferably within or soon after 7 days (ESGE 2021; ACG 2023 for lower GI). ESGE 2026: as soon as clinically indicated by thromboembolic risk. Interrupted secondary-prevention aspirin: restart on the day hemostasis is confirmed (ACG-CAG 2022); ESGE 2021 says as soon as possible, preferably within 3–5 days.
• These are conditional suggestions on low-certainty evidence. Agree reversal for life-threatening bleeding with hematology and local protocols.