ORIGINAL ARTICLES Influence of hypoalbuminemia or hyperalbuminemia on the serum anion gap MARK FELDMAN, NILAM SONI, and BEVERLY DICKSON DALLAS, TEXAS Background: Conflicting data exist as to what extent hypoalbuminemia reduces the anion gap; estimates range from 1.5 to 2.5 mM per g/dL decrease in serum albumin. Methods: We measured serum albumin, total protein, and electrolyte concentra- tions in 5328 consecutive patients aged 1 month to 102 years. Most patients (3750; 70%) had a normal albumin, but 1158 had hypoalbuminemia (<3.4 g/dL); 420 had hyperalbuminemia (>4.7 g/dL). Relationships between serum albumin or total pro- tein and the anion gap were analyzed by linear regression. Results: 309 (27%) hypoalbuminemic patients had a decreased anion gap, and 257 hyperalbumin- emic patients (61%) had an increased anion gap. Among the entire group of 5328 patients, there were highly significant correlations between either serum albumin or total protein and the anion gap (P < 0.001). The slope of the regression for albumin versus anion gap was 2.3 mM per g/dL. Using this slope, anion gap could be adjusted for abnormal serum albumin levels: anion gap adjusted anion gap  2.3 (4-albumin). The initial assessment of an anion gap as being increased, normal, or decreased changed in 44% of the patients with hypo- or hyperalbuminemia once anion gap had been adjusted with this formula. Conclusions: Before considering whether a disorder associated with an increased or decreased anion gap is present, the anion gap should be first adjusted for abnormal serum albumin con- centrations. Our data suggest that physicians use 2.3 times the change in serum albumin, whereas those of Figge et al suggest 2.5; either approach gives similar results. (J Lab Clin Med 2005;146:317–20) Abbreviations: AG = anion gap; ICU = intensive care unit; KAG = K-included anion gap A t physiological pH, serum proteins are nega- tively charged, 1 and albumin is the most impor- tant of these “unmeasured” anionic proteins. For this reason, disorders associated with hypoalbumine- mia are sometimes associated with a decreased AG. 1– 4 By the same mechanism, patients with hyperalbuminemia would be expected to have an increased AG, although such patients have not been reported. There is conflicting information on how to adjust the AG when abnormal serum albumin concentrations are present. 2,4 Figge et al 2 studied 265 arterial blood samples from 152 patients who had been admitted to an ICU in the Czech Republic, almost all of whom had hypoalbumine- mia. The effect of the serum albumin level on the negative charge in plasma was calculated with a complex, computer-assisted model. 2,5 They calculated that for each gram/deciliter decrease in serum albumin, the AG should decrease by 2.5 mM. However, in a study of 432 From the Departments of Internal Medicine and Pathology, Presby- terian Hospital of Dallas, Dallas, Texas. Submitted for publication June 9, 2005; revision submitted July 29, 2005; accepted for publication July 29, 2005. Reprint requests: Mark Feldman, MD, Department of Internal Med- icine, Presbyterian Hospital of Dallas, 8200 Walnut Hill Lane, Dallas, TX 75231; e-mail: MarkFeldman@TexasHealth.org. 0022-2143/$ – see front matter © 2005 Mosby, Inc. All rights reserved. doi:10.1016/j.lab.2005.07.008 317