Comparison of Measured and Calculated Arterial Bicarbonate Concentrations: Potential Application for Prevention of Random Errors in Blood Gas Results

Abstract

We compared arterial bicarbonate concentrations calculated from pH and PCO2 results with the results of direct measurements of bicarbonate concentrations performed on the same specimen. Blood gas and pH measurements were performed with an Instrumentation Laboratory Model 813 blood gas analyzer. Bicarbonate measurements were performed with (1) an AutoAnalyzer system, (2) a Corning 960 CO2 analyzer, and (3) a Photovolt PVA/4 analyzer. There was good correlation between calculated bicarbonate and bicarbonate measured with the PVA/4 (m = 0.997, b = 0.734, Sy = 1.362, r = 0.978). Except for a large constant error (b), statistical comparisons with the other instruments were generally satisfactory (AutoAnalyzer m = 0.987, b = 3.966, Sy = 1.902, r = 0.961; Corning 960 CO2 analyzer m = 0.916, b = 4.100, Sy = 2.150, r = 0.955). Because there is good correlation between calculated bicarbonate concentration and bicarbonate concentration measured with the PVA/4, measurement of arterial bicarbonate concentration by this instrument should serve as a mechanism to identify random errors in pH and PCO2 results. The Corning 960 and AutoAnalyzer systems could also serve this function provided compensations were made for the large constant errors identified in these comparisons.