The hematocrit level in upper gastrointestinal hemorrhage: safety of endoscopy and outcomes

In patients with acute upper gastrointestinal hemorrhage, standard practice is to transfuse packed red blood cells, often to an arbitrary level of hemoglobin or hematocrit (typically 10 g/dL and 30%, respectively) before endoscopy. Therefore, we aimed to determine first whether performing endoscopy in patients with upper gastrointestinal hemorrhage and a low hematocrit is safe and whether it predicts outcomes.

This cohort study included patients with carefully defined upper gastrointestinal hemorrhage captured in our gastrointestinal Healthcare Registry who underwent esophagogastroduodenoscopy. Patients were placed into 2 groups: low hematocrit (<30%) or high hematocrit (>30%). Clinical variables and outcomes, including cardiovascular events, intensive care unit transfer, and death, were measured.

A total of 920 patients meeting entry criteria were identified. Baseline features among those with a low and high hematocrit were identical. Eight cardiovascular events occurred during or after esophagogastroduodenoscopy, including 5 of 587 (1%) in the less than 30% hematocrit group and 3 of 333 (1%) in the greater than 30% hematocrit group (P=.29). Blood transfusions were more common in the low hematocrit group (74% vs 24%, P<.001). However, correlation between the amount of blood transfused and hematocrit level was poor, and the number units of blood transfused was highly variable. There was no significant mortality difference in the 2 hematocrit groups.

Most patients with upper gastrointestinal hemorrhage presented with a hematocrit less than 30%. Performing endoscopy in patients with a low hematocrit was clearly safe; these data strongly imply that waiting for the hematocrit to reach a certain level before endoscopy is not necessary.