Initial Assessment and Fluid Resuscitation

Depending on the presenting disease severity, most patients who start to bleed while outside the hospital are likely to present to the Emergency Department (ED), and the initial priority is of prompt and repeated assessment of airway, breathing, and circulation, as patients are at risk of hemodynamic shock and airway compromise. Venous access should be achieved with at least 2 large-bore cannulae, and patients with active bleeding should be monitored in a high-dependency environment with pulse oximetry, cardiac monitoring, automated blood pressure readings, close monitoring of urine output and, ideally, central venous pressure monitoring. As a minimum, all patients should be blood typed and cross-matched with blood sent for hemoglobin, hematocrit, platelets, coagulation time, and electrolytes. Hemodynamic shock is associated with an increased mortality, and prompt restoration of circulating volume takes priority over endoscopy. There are no studies comparing initial resuscitation with crystalloid or colloid in patients with gastrointestinal (GI) bleeding. Extrapolating from studies in other critically ill patients, a large systematic review and meta-analysis comparing the effects or colloids and crystalloids in critically ill patients found no evidence that resuscitation with colloids reduced the risk of death compared with resuscitation with crystalloids, therefore fluid challenge with either product could be used to commence initial volume restoration prior to blood components for nonvariceal upper gastrointestinal bleeding (NVUGIB). Colloids or albumin are preferred in patients with cirrhosis and ascites.

Use of Blood Components

Acute upper gastrointestinal bleeding (AUGIB) is a very common indication for transfusion of blood components. The purpose of blood transfusion is to correct global or regional oxygen delivery and to improve hemostasis. A study from the United Kingdom found that this indication alone accounts for 14% of the national red cell supply, and in a recent United Kingdom national audit of AUGIB, 43% of all presentations with AUGIB received red blood cell (RBC) transfusion. There is widespread variation in transfusion practice, with a multitude of patient-related and clinician-related factors that may influence transfusion thresholds and decisions in AUGIB. The value of RBC transfusion in exsanguinating NVUGIB is self-evident, and the small proportion of patients requiring massive transfusion should be managed in accordance with local major hemorrhage protocols in close liaison with hospital transfusion teams. Patients requiring massive transfusion (>10 units of packed red cells or loss of >1 circulating volume) are likely to develop a dilutional coagulopathy and will require transfusion of platelets and fresh frozen plasma, again guided by local protocols. However, in less severe bleeding the benefit of RBC transfusion is unclear. A retrospective observational study of 4441 patients admitted with AUGIB in the United Kingdom found that for patients presenting with a hemoglobin (Hb) of >8 g/dL, transfusion within 12 hours was associated with a twofold increase in the subsequent risk of rebleeding (odds ratio [OR] 2.26, 95% confidence interval [CI] 1.76–2.90), although confounding by indication could not be excluded. A meta-analysis of randomized controlled trials (RCTs) assessing the effects of RBC transfusion in adults with AUGIB was limited by the poor quality and small number of studies available. A systematic review of 10 RCTs comparing restrictive versus liberal RBC transfusion strategies in 1780 patients from a variety of clinical settings concluded that a restrictive approach led to a 42% reduction in the probability of receiving transfusions with no effect on mortality, rates of cardiac events, morbidity, or length of hospital stay ; 3 of these 10 trials were in the setting of patients with acute hemorrhage, including one small trial in AUGIB. Therefore the existing body of evidence suggests that, for patients presenting with AUGIB who are not heavily bleeding, transfusion can probably be withheld in the presence of Hb levels as low as 7 to 8 g/dL, provided there is no evidence of comorbid cardiovascular disease.

Correction of Coagulopathy

The prognostic value of the international normalized ratio (INR) following presentation with NVUGIB is poorly characterized. In a large United Kingdom national audit a coagulation profile on admission to the ED was conducted in 82% (5535/6750) of patients, and this was prolonged in 18% (996/5535) of presentations (defined as an INR >1.5 or a prothrombin time >3 seconds prolonged); even after excluding all patients taking anticoagulants and those with cirrhosis, 6.2% of patients had a coagulopathy, and this was associated with an increase in both the risk-adjusted OR for mortality and rebleeding. Similarly, In the Canadian Registry on Nonvariceal Upper Gastrointestinal Bleeding and Endoscopy (RUGBE) cohort of 1869 patients with NVUGIB, a presenting INR of greater than 1.5 was associated with almost a twofold increased risk of mortality (OR 1.95, 95% CI 1.13–3.41) after adjustment for confounders, but not an increased risk of rebleeding. Another study in patients with upper gastrointestinal bleeding (UGIB), using a historical cohort comparison, suggested that correcting an INR to less than 1.8 as part of intensive resuscitation led to lower mortality and fewer myocardial infarctions in the intervention group. The prognostic value of a prolonged coagulation time has been extensively documented in traumatic hemorrhage and, similar to NVUGIB, this is likely to be a multifactorial and a proxy measure of disease severity. However, all patients should undergo a routine coagulation screen following presentation with NVUGIB, and the presence of a coagulopathy should prompt correction with appropriate blood components in liaison with the transfusion team (eg, fresh frozen plasma, prothrombin complex). The completeness of correction of a coagulopathy must be considered, as this should not delay the performance of early endoscopy (defined further later in this article). This recent recommendation is based on recognition of the benefits of early endoscopic intervention and the decreased tissue damage associated with newer ligation hemostatic techniques such as endoscopic clips. Moreover, limited observational data also suggest that endoscopic hemostasis can be safely performed in patients with an elevated INR as long as it is not supratherapeutic (ie, up to around 2.5). The INR does not seem to predict bleeding risk in patients with cirrhosis presenting with UGIB.

Nasogastric Lavage

The use of nasogastric tube (NGT) placement before endoscopy in the emergency management of patients with suspected NVUGIB remains controversial. This tube may be placed to confirm an upper GI source of bleeding, as a prognostic index for identifying high-risk lesions or even to facilitate lavage of the upper GI tract to improve mucosal views at subsequent endoscopy. Although upper GI bleeding can be confirmed in the presence of coffee grounds or fresh blood in the nasogastric aspirate, a clear nasogastric aspirate can still miss up to 15% of patients with a true upper GI source, especially duodenal lesions. The most useful role of NGT aspiration appears to be as a prognostic marker of disease severity. The presence of fresh red blood in the NGT aspirate has been found to be an independent predictor of adverse outcome on multivariate analysis, and as a predictor of high-risk lesions in patients who are hemodynamically stable without evidence of hematemesis. A bloody nasogastric aspirate exhibits a specificity for high-risk lesions (75.8%: 95% CI 70.0–80.0) with a negative predictive value of 77.9% (95% CI 73.2–82.0). Nasogastric aspirate may yield the most useful information in hemodynamically stable patients without hematemesis. Therefore, the NGT may be used in the context of sole diagnostic sampling to identify higher-risk patients who may benefit from an earlier endoscopy, and conversely in patients with low-risk lesions who need less urgent endoscopy.

However, use of the NGT in clinical practice is variable; in the Canadian RUGBE cohort only 28% of presentations with AUGIB had a documented NGT on admission, suggesting that clinicians are uncertain as to the utility of this intervention. Further prospective studies will be needed to assess the utility of the NGT in the management of AUGIB and, in particular, if it can be used to determine which patients benefit from earlier endoscopy.