Rockall Score Calculator

Peptic ulcers (gastric and duodenal) account for approximately 30–50% of all upper GI bleeds. They are scored as “all other diagnoses” (1 point) in the Rockall system, though outcomes vary significantly depending on endoscopic stigmata.

• Forrest Ia (spurting haemorrhage): highest rebleeding risk (~55% without intervention)
• Forrest Ib (oozing haemorrhage): rebleeding risk ~40% without therapy
• Forrest IIa (visible vessel, non-bleeding): rebleeding risk ~43% without therapy
• Forrest IIb (adherent clot): rebleeding risk ~22%
• Forrest IIc (flat pigmented spot): low rebleeding risk (~10%)
• Forrest III (clean-based ulcer): rebleeding risk <5%

Risk factors include H. pylori infection, NSAID use, anticoagulant therapy, and physiological stress. Endoscopic therapy (adrenaline injection, thermal coagulation, clips) substantially reduces rebleeding for high-risk stigmata.

Variceal bleeding accounts for approximately 10–20% of upper GI bleeds but carries significantly higher mortality (15–25% per episode). In the Rockall system, varices are scored as “all other diagnoses” (1 point), which some clinicians feel underestimates their true risk. Patients with varices almost always have liver failure or portal hypertension, which adds 2 points for comorbidity.

Management is distinct from non-variceal bleeding: it includes vasoactive drugs (terlipressin or octreotide), endoscopic band ligation, antibiotic prophylaxis, and consideration of transjugular intrahepatic portosystemic shunt (TIPS) for refractory cases. The Child-Pugh class is a better predictor of outcome specifically for variceal bleeding.

Gastric varices (particularly GOV2 and IGV1) are technically more challenging to manage endoscopically and may require cyanoacrylate injection or interventional radiology.

Mallory-Weiss tears are longitudinal mucosal lacerations at the gastro-oesophageal junction, typically following forceful vomiting or retching. They account for approximately 5–10% of upper GI bleeds. The Rockall Score assigns 0 points for this diagnosis, reflecting the typically self-limiting nature of these lesions.

The vast majority of Mallory-Weiss tears stop bleeding spontaneously (80–90% of cases). Endoscopic therapy is indicated only for active bleeding at the time of endoscopy. Rebleeding is uncommon (<5%), and mortality is very low in otherwise healthy patients. However, in patients with portal hypertension or coagulopathy, Mallory-Weiss tears can bleed more significantly.

Malignant lesions (gastric carcinoma, oesophageal carcinoma, duodenal malignancy, gastrointestinal stromal tumours) receive the highest diagnostic score (2 points) in the Rockall system. This reflects the poor overall prognosis associated with GI malignancy and the higher rates of rebleeding due to tumour neovascularisation and friable tissue.

Bleeding from malignancy is often difficult to manage endoscopically and may require radiological embolisation, radiotherapy, or surgical intervention. Endoscopic haemostasis is often a temporising measure. These patients typically require multidisciplinary team input and consideration of overall prognosis when making treatment decisions.

Malignancy-related bleeding accounts for approximately 2–5% of upper GI bleeds but carries disproportionately high mortality.

Erosive oesophagitis is a common cause of minor upper GI bleeding and is scored as “all other diagnoses” (1 point). Severe oesophagitis (Los Angeles grade C or D) can cause clinically significant haemorrhage, particularly in patients on anticoagulants or with coagulopathy.

Other less common causes include: Dieulafoy lesions (submucosal arteriolar bleeding, often intermittent and difficult to identify), gastric antral vascular ectasia (GAVE / “watermelon stomach”), angiodysplasia, aortoenteric fistulae (particularly in patients with prior aortic graft surgery), haemobilia, and haemosuccus pancreaticus. Aortoenteric fistulae are rare but carry extremely high mortality and may present with a “herald bleed” before catastrophic haemorrhage.

The pre-endoscopy Rockall Score (range 0–7) is useful for initial triage but has limited discriminatory power for rebleeding and mortality compared to the full score. Clinicians sometimes document the pre-endoscopy score and then do not recalculate after endoscopy. The endoscopic components — diagnosis and stigmata of recent haemorrhage — substantially improve the accuracy of risk prediction and should always be incorporated once available.

A pre-endoscopy score of 0 is the one exception where the clinical score alone has strong prognostic value, as these patients have consistently demonstrated very low adverse outcomes across validation studies.

The Rockall Score was developed in 1996 when anticoagulant use was far less prevalent than today. It does not include any variable for anticoagulant or antiplatelet therapy, yet these medications significantly increase both the risk of initial bleeding severity and rebleeding. A patient on dual antiplatelet therapy or a DOAC may have a misleadingly low Rockall Score if they are young, haemodynamically stable, and without significant comorbidity.

When managing anticoagulated patients, always consider the bleeding risk independently and factor in the need for reversal agents, withholding of therapy, and timing of resumption — none of which the Rockall Score addresses.

The comorbidity variable has two tiers: 1 point for cardiac failure, ischaemic heart disease, or “any major comorbidity,” and 2 points for renal failure, liver failure, or disseminated malignancy. A common error is assigning only 1 point to patients with advanced chronic kidney disease or compensated cirrhosis, when they should receive 2 points. Conversely, well-controlled hypertension or stable diabetes alone do not qualify as “major comorbidity” for scoring purposes.

The original study did not provide a precise definition of “major comorbidity,” which introduces subjectivity. When in doubt, err on the side of the higher score, as this will lead to more conservative management.

Variceal bleeding is scored as “all other diagnoses” (1 point) and “no SRH” or “blood in upper GI tract” (0 or 2 points) depending on findings. However, variceal bleeding carries a 15–25% mortality per episode and a high rebleeding rate that may not be adequately captured by the Rockall Score alone. Patients with known varices or cirrhosis benefit from additional risk stratification using the Child-Pugh or MELD scores alongside the Rockall.

The Rockall Score was validated in a general upper GI bleeding population, and variceal bleeding has a distinct pathophysiology and management pathway that the score was not specifically designed to address.

No scoring system replaces bedside clinical assessment. A patient with a low Rockall Score can still deteriorate rapidly if they are actively bleeding at presentation, if the bleeding source was not identified at endoscopy, or if there are factors not captured by the score (e.g., anticoagulant use, social circumstances, ability to return if symptoms worsen). Equally, a high score should prompt increased vigilance but does not mandate ICU admission in all cases.

The Rockall Score is a risk-stratification aid — it provides population-level probabilities that should inform, not replace, individualised clinical decision-making.