Assessment of Instrument-Specific Variation between Measured and Calculated Bicarbonate

Bicarbonate concentration is widely used to assess acid-base disorders in patients, and the concentration can be directly measured or calculated. When discrepant results between measured bicarbonate (mHCO3-) and calculated bicarbonate (cHCO3-) are observed, it can lead to confusion among clinicians and potential for misdiagnosis. Here we assessed the agreement between cHCO3- from the Radiometer blood gas analyzer and mHCO3- from 3 different chemistry instruments.

Three institutions that measure bicarbonate using chemistry analyzers from 3 different manufacturers and derive bicarbonate on Radiometer blood gas analyzers participated in this study. De-identified patient data include plasma mHCO3- and blood gas cHCO3- results (arterial and venous) collected within +/- 20 min. Correlations and biases were determined.

Deming regression analysis showed good correlations between cHCO3- and mHCO3-. Bland-Altman analysis revealed the greatest bias between Radiometer and Abbott Architect (-2.63 mmol/L), followed by Siemens Advia (0.49 mmol/L) and Beckman AU680 (-0.45 mmol/L).

This is the first study to compare cHCO3- from the Radiometer blood gas analyzer against mHCO3- results from multiple chemistry analyzers. Our results suggest that the Radiometer blood gas analyzer, Siemens Advia, and Beckman AU680 agree well with each other. However, bicarbonate results may be negatively biased when measured on the Abbott Architect compared with the other methods.