Comparison of three methods to estimate plasma bicarbonate in critically ill patients: Henderson-Hasselbalch, enzymatic, and strong-ion-gap.

We have previously found poor agreement between Henderson-Hasselbalch and enzymatic methods for estimating plasma bicarbonate concentration in critically ill patients. In this study we compared these two established methods with a new method for estimating bicarbonate using the strong-ion-gap equation. The strong-ion-gap is derived from the Stewart approach to acid-base physiology. One hundred data sets were collected from records of routine daily blood samples in critically ill patients. Bland-Altman analyses were used to compare the three methods. We proposed that bias greater than +/- 1 mmol/l and limits of agreement wider than bias +/- 2 mmol/l were clinically important, Comparing the Henderson-Hasselbalch method to the enzymatic method, the bias was 2.1 mmol/l and the limits of agreement were -1.8 mmol/l to 5.9 mmol/l. Comparing the Henderson-Hasselbalch method to the strong-ion-gap method, the bias was -9.1 mmol/l and the limits of agreement were -17.1 mmol/l to -1.1 mmol/l. Comparing the enzymatic to the strong-ion-gap method, the bias was -11.2 mmol/l and the limits of agreement were -18.2 mmol/l to -4.2 mmol/l. This study found poor agreement between the two established bicarbonate assays and worse agreement between the established assays and the strong-ion-gap method. The strong-ion-gap method is currently too inaccurate for clinical application, but may have future use.