Comparison of three different methods of evaluation of metabolic acid-base disorders

Crit Care Med. 2007 May;35(5):1264-70. doi: 10.1097/01.CCM.0000259536.11943.90.


Objectives: The Stewart approach states that pH is primarily determined by Pco2, strong ion difference (SID), and nonvolatile weak acids. This method might identify severe metabolic disturbances that go undetected by traditional analysis. Our goal was to compare diagnostic and prognostic performances of the Stewart approach with a) the traditional analysis based on bicarbonate (HCO3) and base excess (BE); and b) an approach relying on HCO3, BE, and albumin-corrected anion gap (AGcorrected).

Design: Prospective observational study.

Setting: A university-affiliated hospital intensive care unit (ICU).

Patients: Nine hundred thirty-five patients admitted to the ICU.

Interventions: None.

Measurements and main results: The Stewart approach detected an arterial metabolic alteration in 131 (14%) of patients with normal HCO3- and BE, including 120 (92%) patients with metabolic acidosis. However, 108 (90%) of these patients had an increased AGcorrected. The Stewart approach permitted the additional diagnosis of metabolic acidosis in only 12 (1%) patients with normal HCO3, BE, and AGcorrected. On the other hand, the Stewart approach failed to identify 27 (3%) patients with alterations otherwise observed with the use of HCO3-, BE, and AGcorrected (16 cases of acidosis and 11 of alkalosis). SID and BE, and strong ion gap (SIG) and AGcorrected, were tightly correlated (R2 = .86 and .97, p < .0001 for both) with narrow 95% limits of agreement (8 and 3 mmol/L, respectively). Areas under receiver operating characteristic curves to predict 30-day mortality were 0.83, 0.62, 0.61, 0.60, 0.57, 0.56, and 0.67 for Sepsis-related Organ Failure Assessment (SOFA) score, SIG, AGcorrected, SID, BE, HCO3-, and lactates, respectively (SOFA vs. the rest, p < .0001).

Conclusions: In this large group of critically ill patients, diagnostic performance of the Stewart approach exceeded that of HCO3- and BE. However, when AGcorrected was included in the analysis, the Stewart approach did not offer any diagnostic or prognostic advantages.

Publication types

  • Comparative Study
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Acid-Base Imbalance / diagnosis*
  • Acid-Base Imbalance / mortality
  • Acid-Base Imbalance / pathology
  • Aged
  • Albumins / analysis
  • Anions / analysis*
  • Area Under Curve
  • Bicarbonates / blood*
  • Female
  • Humans
  • Hydrogen-Ion Concentration
  • Hypoalbuminemia / complications
  • Intensive Care Units
  • Lactic Acid / analysis
  • Linear Models
  • Logistic Models
  • Male
  • Multiple Organ Failure / etiology
  • Multiple Organ Failure / mortality
  • Prognosis
  • Prospective Studies
  • Reference Values
  • Sepsis / complications
  • Sepsis / mortality


  • Albumins
  • Anions
  • Bicarbonates
  • Lactic Acid