Predictors of clinically significant upper gastrointestinal hemorrhage among children with hematemesis.

OBJECTIVE

The aim of the study was to determine the proportion of children with hematemesis who experience a clinically significant upper gastrointestinal hemorrhage (UGIH) and to identify variables predicting their occurrence.

METHODS

A retrospective cohort study was conducted. All of the emergency department visits by children ages 0 to 18 years who presented with hematemesis between 2000 and 2007 were reviewed. The primary aim of the study was to determine the proportion of children who developed a clinically significant UGIH; the secondary aim was to identify risk factors predictive of a clinically significant UGIH. A significant UGIH was defined by any of the following: hemoglobin drop >20 g/L, blood transfusion, or emergent endoscopy or surgical procedure.

RESULTS

Twenty-seven of 613 eligible children (4%; 95% confidence interval 3%-6%) had a clinically significant UGIH. Clinically significant hemorrhages were associated with older age (9.7 vs 2.9 years; P<0.001), vomiting moderate to large amounts of fresh blood (58% vs 20%; P<0.001), melena (37% vs 5%; P<0.001), significant medical history (63% vs 24%; P<0.001), unwell appearance (44% vs 6%; P<0.001), and tachycardia (41% vs 10%; P<0.001). The frequency of laboratory investigations increased with age (P<0.001). The hemoglobin level was the only laboratory investigation whose results differed between those with and without significant bleeds. The presence of any one of the following characteristics identified all of the children with a clinically significant hemorrhage: melena, hematochezia, unwell appearance, or a moderate to large volume of fresh blood in the vomitus, sensitivity 100% (95% confidence interval 85%-100%).

CONCLUSIONS

The occurrence of a clinically significant UGIH was uncommon among children with hematemesis, especially in well-appearing children without melena, hematochezia, or who had not vomited a moderate to large amount of fresh blood.