Hemoximetry as the "Gold Standard"? Error Assessment Based on Differences Among Identical Blood Gas Analyzer Devices of Five Manufacturers

BACKGROUND: The calibration and testing procedures of a pulse oximeter with arterial blood samples from healthy subjects are based on reference values from the hemoximeter. There are no tests to identify the accuracy of the reference devices. Because of this limitation and since the true values of oxygen saturation (sO(2) in %) in blood samples were not known, we used the differences between two identical devices, A and B, for error assessment. METHODS: Two identical devices, A and B, from five leading manufacturers were investigated. Seventy-two arterial blood samples from 12 healthy volunteers at three different levels of saturation between 100% and 70% sO(2) were randomly evaluated by the test systems. RESULTS: The observed differences (Delta) between Devices A and B, as a measure for the error of the hemoximeters, increased significantly with all manufacturers from level 97 (Delta min, -0.9%; Delta max, 2.6%) to 85 (Delta min, -2.4%; Delta max, 4.3), this effect was even stronger between levels 97 and 75 (Delta min, -4.6%; Delta max, 4.3%). A variance proportion analysis revealed the concentration of the reduced hemoglobin as the main error source for sO(2) measurements. Independent from the sO(2) levels there were also significant differences for the carboxy hemoglobin concentration in the range of 0%-4% and for the methemoglobin concentration in the range of 0%-1%. CONCLUSIONS: The variance of sO(2) measurements between identical devices increased significantly when saturation decreased from the normal level of 97% to the hypoxemic levels of 85% and 75%.