Reliability of HemoCue in patients with gastrointestinal bleeding

HemoCue is routinely used to manage bleeding patients, but few studies have evaluated its accuracy in this population. We compared HemoCue with laboratory determination of blood hemoglobin in patients with gastrointestinal bleeding. A prospective observational study in a 14-bed medicosurgical ICU and an emergency department in an urban general hospital. 94 patients admitted to the emergency department or to the ICU for gastrointestinal bleeding. Blood was drawn at admission to measure laboratory hemoglobin and capillary hemoglobin was measured simultaneously by HemoCue. The unit of hospitalization and the presence or absence of impaired vital signs (tachycardia and/or hypotension and/or shock) were recorded. The mean difference between HemoCue and hemoglobin (bias) was -0.06 g/dl and standard deviation (precision) 0.87 g/dl. (95% CI -1.8 to 1.68). Discrepancies between HemoCue and hemoglobin were greater than 1 g/dl in 21% of cases. Bias was comparable between patients admitted to the ICU and those in the emergency department. The accuracy of HemoCue was not affected by the presence of impaired vital signs or by a hemoglobin level below 9 g/dl or 7 g/dl. Although we demonstrated a low bias between HemoCue and blood hemoglobin determination, large HemoCue vs. hemoglobin differences may still occur, and therefore therapeutic decisions based upon capillary HemoCue alone should be very cautious.