Accuracy and reliability of the i-STAT point-of-care device for the determination of haemoglobin concentration before and after major blood loss

We investigated the accuracy of i-STAT(®) (Abbott Point of Care Inc., Princeton, NJ, USA) haemoglobin (Hb) measurement in surgical patients with an estimated blood loss of ≥25% of total blood volume. Blood tests for i-STAT(®) Hb, laboratory Hb (Sysmex XE-2100(™), Sysmex Corporation, Kobe, Japan) and total plasma proteins were obtained at the start of surgery (T=0) and when an estimated 25% total blood volume loss had occurred (T=1). Thirty-one patients were recruited. The coefficient of variation of the paired i-STAT(®) Hb estimates was 2.8% and 2.9% at T=0 and T=1, respectively. The mean difference between i-STAT(®) and laboratory Hb was -7.6 g/l (standard deviation 6.5) at T=0 and -5.1 g/l (standard deviation 12) at T=1. The mean total plasma protein difference (total plasma protein T=0 minus T=1) was 13.6 g/l (95% confidence interval 10.2 to 17.0). There was poor correlation between total plasma protein and bias in i-STAT(®) measurements. The i-STAT(®) Hb had an acceptable coefficient of variation, but the Hb levels were lower than those estimated by the laboratory. The standard deviation of i-STAT(®) Hb was greater after ≥25% estimated total blood volume loss. Clinicians should not use the i-STAT(®) Hb in isolation for clinical decision-making when considering blood transfusion in a situation of 25% or greater blood loss.