Aspectos analíticos del analizador de gases Cobas ® b 221 (Roche Diagnostics)

Abstract

Introduction and objectives The Cobas ® b 221 (Roche Diagnostics) is an automatic analyser for blood gas measurement in whole blood, with a wide range of parameters: pH, pO 2, pCO 2, co-oximetry, electrolytes, bilirubin, glucose, lactate and urea. The aim of this study is the assessment of imprecision and to investigate the agreement of the results between the Cobas ® b 221 and other analysers in order to introduce the Cobas ® b 221 as a blood gas analyser in the STAT section of the clinical laboratory. Material and methods For the intra- and inter-day imprecision study, three quality control levels provided by the manufacturer were evaluated. For the transferability study, 130 arterial blood samples and 100 venous blood samples were tested and a comparison between the results provided by the Cobas ® b 221 system and the ABL800 FLEX, Cobas 6000 and ADVIA 120 analysers was made. In the statistical study, parametric and non parametric tests were applied. For the regression analysis and for the evaluation of the concordance of results the Passing-Bablock method and the Bland-Altman method respectively were used. Results In the intra- and inter-day imprecision study, good coefficients of variation were obtained for all the studied parameters. Good correlation coefficients were observed in the correlation study (r>0.90, P<0.001). In the regression study, for most of the cases, a systematic and proportional error was found, which was due to the fact that the original slope and intercept did not include number one nor zero respectively. Moreover, differences between the statistically significant means ( P<0.001) were observed. From the statistical point of view, these data may indicate that the results would not be transferable. Nevertheless, low standard estimation errors (S y/x) and differences between the means lower than the clinically acceptable values were found together with minor systematic errors (mean of the differences between two methods) in the Bland-Altman plots as well as narrow and acceptable concordance limits. From the clinical point of view, these results indicate that the observed differences were not significant, except for glucose, for which, once the values measured with the Cobas ® b 221 analyser using the regression straight line are corrected, transferable results could be obtained. Conclusion Considering the results obtained together with other feasibility features of the Cobas ® b 221 analyser, we conclude that, in spite of the statistically significant differences between the results measured by the Cobas ® b 221 system and the comparison analysers, the differences found for the investigated parameters are not important from a clinical point of view and therefore, we consider that the Cobas ® b 221 blood gas analyser can be adapted to the daily routine of the STAT section in the clinical laboratory.