Clinical Question

What is the risk of rebleeding or death in a patient with upper gastrointestinal (GI) bleeding?

Evidence Summary

Upper GI bleeding remains a common problem and reason for hospital admission.¹ A more precise estimate of a patient’s prognosis would be helpful to physicians who are deciding on hospital discharge and the intensiveness of monitoring in inpatient and outpatient settings.

A clinical decision rule has been developed that estimates the likelihood of mortality in patients presenting with upper GI bleeding. A study² by Rockall and colleagues identified all patients presenting to hospitals in four health regions with acute upper GI hemorrhage who subsequently underwent endoscopy. Patients whose bleeding occurred in the hospital and those who did not undergo endoscopy were excluded from the study. Patients were followed prospectively and their risk of rebleeding and death was determined. This clinical decision rule has been validated^3–6 for the prediction of mortality. The largest and best designed of these validations³ was a prospective evaluation in 951 Dutch patients with a median age of 71 years. Although some of these studies^2,3,6 did not find that the prediction of rebleeding was as accurate as the prediction of death, the rule does accurately identify a group with a very low risk of rebleeding (Rockall score of 2 or lower). While other rules^7–9 have been developed, they have not been as well validated as the Rockall² risk score.

The Rockall risk score and its interpretation using combined data from the original study² and the Dutch validation study³ are shown in Figures 1 through3. To use the clinical decision rule, determine the number of points for your patient using Figure 1, then determine the patient’s risk of rebleeding and death using Figure 2. Data in Figure 2 are shown for individual scores, as well as scores grouping patients into low-, moderate-, and high-risk groups. A clinical score using the three nonendoscopic variables is shown in Figure 3; it may be especially helpful to family physicians, but it has not been validated outside of Rockall’s original validation study.²

Obviously, no clinical decision rule should be applied without the usual application of clinical judgment. However, these rules can help support clinical decision making by identifying patients who can be considered for early discharge and patients who are at an increased risk but might otherwise be considered for discharge from the hospital.

Applying the Evidence

Mr. Sailors, a 43-year-old man who is in otherwise good health, presents to the emergency department with a single episode of coffee-ground emesis and a large melanotic stool that morning. He has a six-month history of taking a nonsteroidal anti-inflammatory drug for tendonitis. His blood pressure is 108/60 mm Hg, his heart rate is 108 beats per minute, and he complains of lightheadedness when he sits up on the gurney. After being stabilized, he undergoes endoscopy that reveals a small duodenal ulcer, an adherent clot, and some old blood in the gastrointestinal tract. What is the likelihood that he will rebleed or die?

Answer: This patient gets zero points for age, one point for shock, zero points for comorbidity, one point for diagnosis based on endoscopy, and two points for stigmata of recent hemorrhage. His total risk score is four points, which puts him in the moderate risk category (a 13 percent risk of rebleeding and a 6.8 percent risk of death). Based on this information, the physician decides to observe him closely in an inpatient setting for an additional day or two rather than send him home right after the endoscopy.