Comparison of Arterial Bayer Rapidlab 865 and Venous Abbott Cell-Dyn 4000 Hemoglobin Measurements

Abstract

importance, for example in the diagnosis and treatment of anemia and in transfusion protocols [1-3]. Despite the fact that it would be preferable to determine hemoglobin in a hospital using only one method, frequently different methods are used. Necessarily, hemoglobin results are reported into a hospital information system regardless of the sample type and method used. Therefore a good correlation between the methods used is mandatory. Deviations between different methods for the measurement of hemoglobin are described in the literature [4-6]. However, to our knowledge, this is the first published comparison of hemoglobin measured on a blood gas analyzer and a hemocytometry analyzer. For this study we used the Rapidlab 865 blood gas analyzer (Bayer Health Care, Tarrytown, NY, USA) and the Cell-Dyn 4000 hemocytometry analyzer (Abbott Laboratories, Abbott Park, IL, USA). The Cell-Dyn 4000 measures hemoglobin using absorption spectrophotometry. A cyanide-free hemoglobin reagent is used to lyse all cells and convert hemoglobin to a single chromagen with an absorption peak at 544 nm. A 540 nm LED illuminates the sample, and a photodetector measures the amount of light that passes through the sample. The Rapidlab 865 also measures hemoglobin using absorption spectrophotometry. Cells are lysed using ultrasound. Total hemoglobin and several clinically significant derivatives of hemoglobin are determined using an algorithm based on absorption spectrophotometry data using 40 different wavelengths. During 1 year, from April 2002 until March 2003, we tested samples from 1864 patients from our emergency department. Two samples were drawn at the same time from each patient, an arterial blood gas sample (3-mL heparin blood gas syringes [Rapidlyte; Bayer, Mijdrecht, the Netherlands]) and a venous blood sample (3-mL K3EDTA vacuum tubes [BD vacutainer, Plymouth, UK]). The results are shown in Figure 1. There is a linear relation with good correlation between the 2 methods using different sample types (y = 1.014x + 0.164; r = 0.926). In Figure 2 a bias analysis is shown. The Rapidlab 865 has a mean bias of 0.354 g/dL compared with the Cell-Dyn 4000. Despite the fact that the actual difference between the 2 methods is very small, clinical decisions are often based on a LETTER TO THE EDITOR