False-negative diagnosis of high anion gap in patients with end-stage kidney disease

Abstract The traditional anion gap (AG) equation is widely used, but its misdiagnosis in end-stage kidney disease (ESKD) patients has not been investigated fully. Diagnostic accuracy to detect high AG was cross-sectionally evaluated using 3 AG equations in 1733 ESKD patients with an eGFR less than 15 mL/min/1.73 m2. The prevalence of high AG was 67.9%, 92.1% and 97.4% by the traditional, albumin-adjusted AG (aAG) and full AG equations, respectively. The sensitivity, specificity, accuracy and Kappa coefficient obtained with the traditional AG vs aAG equation were 0.70 vs 0.94, 0.98 vs 0.93, 0.7 vs 0.94, and 0.103 vs 0.44, respectively. Next, we created a subcohort comprising only patients with high full AG and investigated how the traditional AG equation leads to misdiagnoses. Multivariable-adjusted regression analysis in 1688 patients revealed that independent factors associated with a false-negative AG diagnosis were ARB use, eGFR, blood leukocyte count, serum chloride, bicarbonate, ionized calcium, potassium, albumin and phosphate. 93.2% of our subcohort prescribed any of RAAS inhibitors, Loop diuretics or Alkali which could increase either serum chloride or bicarbonate. Frequent use of these possible AG-reducing medications may conceal high AG state in patients with ESKD unless they have incidental inflammation which may increase AG value.