Rockall score calculator for upper GI bleeding

How this Rockall score calculator works

This Rockall score calculator adds up the five items that Rockall and colleagues built into a risk score for death after acute upper gastrointestinal haemorrhage, published in Gut in 1996. Their prospective, population-based UK study included 4,185 cases over the age of 16 from four months of 1993 and tested the score on 1,625 further cases from three months of 1994. Age, shock and comorbidity are known at the bedside and give the pre-endoscopy score, also called the clinical, initial or admission Rockall score, from 0 to 7. Endoscopic diagnosis and major stigmata of recent haemorrhage are added after endoscopy for the complete, or full, score from 0 to 11. Shock is read from two numbers: a systolic blood pressure below 100 mmHg scores 2 whatever the pulse, and a pulse of 100 or more with a systolic pressure of 100 or more scores 1. Haemoglobin, sex, presentation other than shock and drug therapy were not represented in the final model.

Rockall score points (Rockall 1996). The first three rows make up the pre-endoscopy score.

Comorbidity scores the single highest category that applies, not the sum of each condition. If the patient has not had endoscopy yet, choose “Not yet” and the calculator shows the pre-endoscopy score on its own; picking a diagnosis or stigmata answer switches it to the complete score. For when to use each version, read the guide to the pre-endoscopy Rockall score vs the full score.

Rockall score results and what they mean

Rockall score results map to a mortality figure at each point, and the same number carries less risk after endoscopy than before it. In the 1996 Gut paper, mortality rose from 0.2% at a pre-endoscopy score of 0 to 11.0% at 3 and 50.0% at 7. On the complete score, mortality was 0% at 0 and 1, 0.2% at 2, 10.8% at 5 and 41.1% at 8 or more. The paper concluded that the score could pick out 15% of all cases at presentation and 26% after endoscopy as at low risk of rebleeding and negligible risk of death, who might be considered for early discharge or outpatient treatment. In the companion Lancet paper, 744 of 2,531 patients who had endoscopy (29.4%) scored 2 or less on the complete score; 32 of them (4.3%) rebled and one (0.1%) died. Stanley and colleagues found 4 or more on the pre-endoscopy score and 5 or more on the complete score were the best cut-offs for predicting death.

Mortality and rebleeding by score, as reported by Rockall and colleagues in 1996:

When the Rockall score is less reliable

The Rockall score is less reliable for predicting rebleeding and the need for treatment than for predicting death, the outcome it was built for. Vreeburg and colleagues tested it in Amsterdam hospitals and in Rockall's own validation sample in 1999: the area under the ROC curve for mortality was 0.73 and 0.81, but for rebleeding only 0.61 and 0.70, with poor calibration for rebleeding in both samples. In an international study of 3,012 consecutive patients at six hospitals, published in the BMJ in 2017, the Glasgow-Blatchford score predicted the composite of transfusion, endoscopic treatment, interventional radiology, surgery or 30-day death better (AUROC 0.86) than the full Rockall score (0.70) or the admission Rockall score (0.66), and no score predicted rebleeding well. A pre-endoscopy score of 0 is not a safe discharge rule on its own: in a 2015 New Zealand study, 15 of 42 patients scoring 0 (35.7%) needed transfusion or endoscopic treatment. The mortality figures also come from UK hospitals in 1993 and 1994.