Instrumental variability of respiratory blood gases among different blood gas analysers in different laboratories.

The aim of this study was to test the hypothesis that differences in oxygen tension (PO2) and carbon dioxide tension (PCO2) values from measurements performed on different blood gas analysers in different laboratories are clinically insignificant. Samples of fresh whole human tonometered blood (PO2 8.1 kPa (60.8 mmHg); PCO2 5.3 kPa (39.9 mmHg)) were placed in airtight glass syringes and transported in ice-water slush. Blood gas analysis was performed within 3.5 h by 17 analysers (10 different models) in 10 hospitals on one day. The mean of the differences between the measured and target values was -0.01+/-0.19 and 0.21+/-0.13 kPa (-0.06+/-1.45 and 1.55+/-1.01 mmHg) for PO2 and PCO2, respectively. The mean of the differences between two samples on one analyser was 0.06+/-0.06 and 0.04+/-0.03 kPa (0.47+/-0.48 and 0.29+/-0.24 mmHg), respectively. For PO2 and PCO2 the interinstrument standard deviations (s(b)) were 0.18 and 0.13 kPa (1.38 and 0.99 mmHg), respectively, whereas the intra-instrument standard deviations (s) were 0.06 and 0.03 kPa (0.47 and 0.26 mmHg), respectively. Both for PO2 and PCO2 the ratios of s(b)2 and s2 were statistically significant (analysis of variance (ANOVA) p<0.001). The standard deviations of a random measurement on a random analyser were 0.19 and 0.14 kPa (1.46 and 1.02 mmHg) for PO2 and PCO2, respectively. We conclude that the variability in measurement of blood gas values among different blood gas analysers, although negligible, depends much more on inter- than intra-instrument variation, both for oxygen tension and carbon dioxide tension. Technical improvements and adequate quality control programmes, including tonometry, may explain why the variability in blood gas values depends mainly on errors in the pre-analytical phase.