James Figge1, Rinaldo Bellomo2, Moritoki Egi3. 1. Department of Medicine, Division of Endocrinology, St. Peter's Health Partners, Albany, NY, United States. Electronic address: James.Figge@SPHP.com. 2. Department of Intensive Care, Austin Hospital, Heidelberg, Victoria, Australia; School of Medicine, University of Melbourne, Parkville, Victoria, Australia. 3. Department of Anesthesiology, Kobe University Hospital, Kobe City, Hyogo 650-0017, Japan.
Abstract
BACKGROUND: Quantitative relationships among plasma [Lactate], [Pi], [Albumin], unmeasured anions ([UA]) and the anion gap (AGK) in lactic acidosis (LA) are not well defined. METHODS: A mathematical model featuring compensatory potassium and chloride shifts and respiratory changes in LA demonstrated: (1) AGK=[Lactate]+Zp×[Pi]+2.4×[Albumin]+constant1+e, where Zp is a function of pH, and e reflects unmeasured anions and cations plus pH-related variations. Eq. (1) can be algebraically rearranged to incorporate the albumin-corrected anion gap, cAGK: (2) cAGK=[Lactate]+Zp×[Pi]+constant2+e. Eq. (1) was tested against 948 data sets from critically ill patients with [Lactate] 4.0mEq/L or greater. AGK and cAGK were evaluated against 12,341 data sets for their ability to detect [Lactate]>4.0mEq/L. RESULTS: Analysis of Eq. (1) revealed r2=0.5950, p<0.001. cAGk>15mEq/L exhibited a sensitivity of 93.0% [95% CI: 91.3-94.5] in detecting [Lactate]>4.0mEq/L, whereas AGK>15mEq/L exhibited a sensitivity of only 70.4% [67.5-73.2]. Additionally, [Lactate]>4.0mEq/L and cAGK>20mEq/L were each strongly associated with intensive care unit mortality (χ2>200, p<0.0001 for each). CONCLUSIONS: In LA, cAGK is more sensitive than AGK in predicting [Lactate]>4.0mEq/L.
BACKGROUND: Quantitative relationships among plasma [Lactate], [Pi], [Albumin], unmeasured anions ([UA]) and the anion gap (AGK) in lactic acidosis (LA) are not well defined. METHODS: A mathematical model featuring compensatory potassium and chloride shifts and respiratory changes in LA demonstrated: (1) AGK=[Lactate]+Zp×[Pi]+2.4×[Albumin]+constant1+e, where Zp is a function of pH, and e reflects unmeasured anions and cations plus pH-related variations. Eq. (1) can be algebraically rearranged to incorporate the albumin-corrected anion gap, cAGK: (2) cAGK=[Lactate]+Zp×[Pi]+constant2+e. Eq. (1) was tested against 948 data sets from critically illpatients with [Lactate] 4.0mEq/L or greater. AGK and cAGK were evaluated against 12,341 data sets for their ability to detect [Lactate]>4.0mEq/L. RESULTS: Analysis of Eq. (1) revealed r2=0.5950, p<0.001. cAGk>15mEq/L exhibited a sensitivity of 93.0% [95% CI: 91.3-94.5] in detecting [Lactate]>4.0mEq/L, whereas AGK>15mEq/L exhibited a sensitivity of only 70.4% [67.5-73.2]. Additionally, [Lactate]>4.0mEq/L and cAGK>20mEq/L were each strongly associated with intensive care unit mortality (χ2>200, p<0.0001 for each). CONCLUSIONS: In LA, cAGK is more sensitive than AGK in predicting [Lactate]>4.0mEq/L.