Metabolic acidosis in septic shock: is the Stewart theory the magic bullet?

We read with a great attention the article entitled “Defining metabolic acidosis in patients with septic shock using Stewart approach” [1]. We have some remarks about it. The authors used the Stewart approach to diagnose acid-base disturbances in patients having septic shock. Their reference for metabolic acidosis diagnosis seems to be the calculated standard base excess (SBE). It is interesting to note the lack of correlation between SBE and the strong ion gap (SIG) that represents unmeasured anions. The method used for albumin concentration measurement was different from the 2 methods that we previously compared and that have quite different results [2,3]. They calculated the corrected chloride to diagnose hyperchloremic acidosis. Using our previously published data [3], Fig. 1 shows the reproducibility for chloride and corrected chloride between 2 laboratory analyzers. As expected, the reproducibility between both analyzers was better for chloride (r2 = 0.87; mean difference, 1.7 ± 1.7 mEq/L) rather than for corrected chloride (r2 = 0.68; mean difference, 0.6 ± 2.1 mEq/L). The authors defined an interesting group in which all biologic values were normal except an increased SIG. They stated that SBE missed the presence of unmeasured anions. This increased SIG may also represent the expanded uncertainty of measurement of a calculated parameter that increases with the number of the independent parameters on which it depends (9 for SIG) [3]. We previously discussed the poor value of SIG, particularly its lack of reproducibility [2], a point that was unfortunately not discussed by the authors. What were the clinical characteristics of these patients? Were they different from the other patients besides the lower normal saline administered volume, and what was their prognosis? When using the Stewart approach, the main problem remains the poor reproducibility between analyzers, particularly for calculated parameters. This fact complicates comparisons between studies and may induce a different treatment according to the center. Even if we have to accept a cumulative error in measurement in our clinical practice [4], the used parameters should have a clinical relevant reproducibility to avoid center-dependent patient treatment.