Use of Sodium-Chloride Difference and Corrected Anion Gap as Surrogates of Stewart Variables in Critically Ill Patients

Jihad Mallat, Stéphanie Barrailler, Malcolm Lemyze, Florent Pepy, Gaëlle Gasan, Laurent Tronchon, Didier ThéveninPLoS 2013 · DOI

Abstract

Managing acid-base balance in the ICU often requires choosing between simple but flawed traditional metrics and the highly accurate but mathematically complex "Stewart approach." This study demonstrates that two simple bedside calculations—the sodium-chloride difference and the corrected anion gap—serve as excellent, high-accuracy surrogates for complex Stewart variables, allowing clinicians to perform sophisticated physiological assessments in seconds. Specifically, $SID_{app}$ and SIG can be substituted by $[Na^+] - [Cl^-]$ and by $AG_{corr}$ respectively in the diagnosis and management of acid-base disorders in critically ill patients.

Introduction

Critically ill patients frequently suffer from complex acid-base disorders that correlate with poor outcomes. Traditionally, clinicians rely on the "physiological approach" (bicarbonate and pH) or the "base excess" approach. However, these methods often fail in the ICU because they do not account for common conditions like hypoalbuminemia (low blood protein), which can mask underlying metabolic issues.

The "Stewart approach" (physicochemical approach) is considered the gold standard for accuracy because it treats pH as a dependent variable determined by three independent factors: $PCO_2$, the Strong Ion Difference (SID), and weak acids (like albumin). While superior, the Stewart approach requires measuring multiple electrolytes (calcium, magnesium, phosphate) and performing cumbersome calculations that are impractical at a busy bedside. This study sought to determine if simpler "surrogate" formulas could provide the same diagnostic power as the full Stewart equations.

Methods

The researchers conducted a prospective observational study of 341 adult ICU patients in France. The cohort was divided into a Modeling Group (n=161) to build mathematical linear models and a Validation Group (n=180) to test the accuracy of those models.

The study compared:

  • Apparent Strong Ion Difference ($SID_{app}$) vs. the simple Sodium-Chloride difference ($[Na^+] - [Cl^-]$).
  • Strong Ion Gap (SIG) vs. the Anion Gap corrected for albumin and lactate ($AG_{corr}$).

Key Results and Linearity

The most significant finding was the remarkable linear relationship between the complex Stewart variables and their simple surrogates. In the modeling group, $[Na^+] - [Cl^-]$ predicted $SID_{app}$ with an $R^2$ of 0.973, while $AG_{corr}$ predicted SIG with an $R^2$ of 0.96.

Linear regression showing the high correlation ($R^2 = 0.977$) between observed and predicted $SID_{app}$ values.
Linear regression showing the high correlation ($R^2 = 0.96$) between observed and predicted SIG values.

Diagnostic accuracy was equally impressive:

  • A $[Na^+] - [Cl^-]$ value $\leq 34$ mEq/L identified $SID_{app}$ acidosis with 94.5% sensitivity and 98% specificity.
  • An $AG_{corr} > 17$ mEq/L identified SIG acidosis with 95% sensitivity and 93% specificity.

These results held up in the validation group with minimal bias ($0 \pm 1.5$ mEq/L for $SID_{app}$), confirming that these simple surrogates are mathematically robust substitutes across a wide range of patient conditions, including septic shock and kidney failure.

Clinical Implications

The standard Anion Gap is notoriously unreliable when albumin levels are low. By correcting the Anion Gap for both albumin and lactate, it mirrors the "Strong Ion Gap" (SIG)—a metric that identifies "unmeasured anions" (hidden toxins or metabolic byproducts) with high precision. Adopting these simplified formulas could lead to faster, more accurate diagnosis and management of acid-base disorders without the need for complex software or time-consuming manual math.

Limitations

As a single-center study, specific regression coefficients might vary slightly between different laboratory analyzers. Furthermore, the study had limited data for patients with severe hypernatremia ($Na^+ > 145$ mEq/L), and rare unmeasured cations (like lithium) might still cause discrepancies in the SIG surrogate. Despite these edge cases, the physiological principles remain universal for the vast majority of ICU admissions.

Summary updated

Abstract

SID(app) and SIG can be substituted by [Na(+)] - [Cl(-)] and by AG(corr) respectively in the diagnosis and management of acid-base disorders in critically ill patients.
BibTeX
@inproceedings{mallat2013use,
  author = {Jihad Mallat and Stéphanie Barrailler and Malcolm Lemyze and Florent Pepy and Gaëlle Gasan and Laurent Tronchon and Didier Thévenin},
  title = {{Use of Sodium-Chloride Difference and Corrected Anion Gap as Surrogates of Stewart Variables in Critically Ill Patients}},
  booktitle = {PLoS},
  year = {2013},
  doi = {10.1371/journal.pone.0056635},
}
Figure