Background: While point-of-care testing (POCT) is being used increasingly as a basis for deciding on perioperative erythrocyte transfusion, no valid standards currently exist concerning the accuracy of Hb concentration measurements. For clinical employment, however, the confidence limits (+/-2 SD) of these measurements should lie close to 5 g/l. The aim of the present study was to evaluate the accuracy and precision of point-of-care testing for blood hemoglobin concentration (cHb in g/l) measurements in critically ill patients.
Methods: Fifty blood samples from 50 postoperative patients requiring intensive care treatment were withdrawn from a cannula in the radial artery into a 2-ml heparinized syringe (containing wet sodium heparinate in the conus), in a 2-ml Monovette with 50 IE lithium heparinate, and into a 2.7-ml cuvette with 1.6 mg potassium EDTA/ml blood. The POCT battery consisted of two blood gas analyzers (ABLTM 625 and 725, Radiometer, Copenhagen), the HemoCue system (Mallinckrodt Medical, Germany), and an automated hematology analyzer (M-2000(R), Sysmex, Germany). The cyanmethemoglobin method served as the reference 'gold standard' procedure. The blood gas analyzer and HemoCue systems were tested using dry and wet heparinized blood samples.
Results: Hemoglobin concentrations of the reference measurements ranged from 73.9 to 159.4 g/l. The automated hematology analyzer method did reveal a small but systematic deviation for higher cHb values. For the blood gas analyzer and HemoCue system procedures there was no systematic deviation of bias for either the first measurement or the averaged data. Bland & Altman analysis revealed a larger scattering for the wet heparinized samples.
Conclusions: The above-stated requirement for POCT systems, i.e. that the confidence limits should lie close to 5 g/l cHb, held true for the dry heparinized samples of the blood gas analyzer (1st measurement and mean of 2), the HemoCue system (mean of 3) and the automated hematology analyzer.