Examination of blood haemoglobin concentration measured using the OSM2

Blood haemoglobin concentration is regularly measured automatically by instruments reporting the value in around 1 min. The OSM2 from Radiometer is an example. Results from this instrument have been compared with those of a reference method using the hemiglobincyanide principle. Four healthy, moderately trained, young men (non-smokers) cycled for 2 min to exhaustion. Blood samples were drawn from indwelling catheters in the femoral artery and vein before exercise, during exercise and in the 1 h recovery. Blood haemoglobin concentration was analysed using both methods. The results of the OSM2 were linearly related to those of the control method, with a random variation of 0.14 mmol L(-1) (1.5%). For arterial blood, the OSM2 showed a systematic bias of -0.36 mmol L(-1) (-4%). For femoral venous samples the bias varied depending on the haemoglobin concentration, being negative at low concentrations and positive at high values (-3 to +2%). Consequently, the arteriovenous (a-v) difference differed systematically between the two methods. The varying bias in the results of the OSM2 for femoral-venous samples correlated with pH, pCO(2), O(2) saturation of haemoglobin (sO(2)) and with the haemoglobin concentration itself (cHb). Partial correlation analyses suggest that only the latter two correlations were independent, while correlations of the bias with pH and pCO(2) were removed when correcting for the effect of sO(2) and cHb. In conclusion, the OSM2 measures the blood haemoglobin concentration fairly precisely, but there is a variable bias of up to 4% in absolute value. Finally, the instrument does not report a-v differences reliably.