Reply by Cove and Kellum to Swenson

From the Editorialists:

We thank Prof. Swenson for his well-articulated letter. We conceded in our editorial that the Stewart approach doesn’t provide a strong mechanistic explanation (1), and therefore we also agree it doesn’t eliminate a mechanistic role of bicarbonate. Rather, the Stewart approach provides a unifying explanation of the factors determining plasma pH by incorporating the role of electrolytes, plasma proteins, and carbon dioxide in all forms, including bicarbonate. Whether one considers bicarbonate to be dependent and Pco₂ independent, or vice versa, depends on the system under examination. Thus, we do not think this a useful classification and avoided the terms in our editorial because we’ve shown Pco₂ behaves like a dependent molecule under some circumstances (2). However, the classification does serve the purpose of drawing attention to the fact that Pco₂ and HCO₃ are not independent of each other. Furthermore, the Stewart approach is not at odds with our understanding of membrane ion transporters; instead, it provides a more complete understanding of their function (3).

In the paper by Zanella and colleagues, normal pH was restored after induction of acidosis by removing chloride to increase the strong ion difference; this also led to an increased bicarbonate level (4). Both of these changes can be easily understood with the Stewart approach, whereas a bicarbonate-centric approach focuses only on the bicarbonate change that occurred. Therefore, if we simply considered the information on the blood gas, we could only classify the acid–base changes in the Zanella experiments as metabolic, respiratory, or mixed, but we would be unable to describe what had caused the observed changes. Similarly, in clinical practice, a bicarbonate-centered approach allows us to identify the presence of an acid–base derangement as effectively as the Stewart approach. However, the Stewart approach provides a more precise classification of the acid–base changes, even in the studies referenced by Prof. Swenson (5). Such precision is important in clinical practice where, for example, the impact of choice of resuscitation fluids on acid–base status needs to be considered. We flatly reject the assertion that teaching the Stewart approach is difficult or that it leads to increased costs and blood loss. In fact, the variables needed to apply the approach are routinely measured already (6). The value of the approach comes from the ability to understand the relative contributing effects of simultaneous abnormalities on the final observed pH, such as elevated lactate, hypoalbuminemia, and electrolyte abnormalities, in a way that bicarbonate-centric understanding cannot (7).