Diagnostic threshold and performance of anion gap in screening for high anion gap metabolic acidosis.

Abstract

The anion gap (AG) is commonly used to screen for acid-base disorders. It was proposed that the cut-off for high anion gap metabolic acidosis (HAGMA) may be lower with current laboratory techniques, although modern laboratory equipment are still calibrated to familiar reference ranges established with earlier techniques. The appropriate cut-off for HAGMA is unclear. This study aimed to assess the performance of AG as a screening test for HAGMA and to determine the optimal diagnostic threshold of AG for HAGMA. This was a retrospective analysis of a large, anonymised dataset extracted by computerised protocol from 2017 to 2019. All inpatients with blood samples taken for organic acids (lactate, ketone or salicylate) paired with a metabolic panel were included. The target condition was HAGMA secondary to elevated blood lactate, ketone and/or salicylate. Sensitivity for HAGMA was explored at various AG cut-off levels. Of 16,475 patients, 2,621 had organic acidosis. Median age was 65 years, and median estimated glomerular filtration rate was 70 mL/min/1.73 m2. With organic acidosis, the median AG was 23 (interquartile range [IQR] 20-29) mEq/L, while without organic acidosis, the median AG was 16 (IQR 14-19) mEq/L. The area under the curve-receiver operating characteristic of AG for HAGMA was 0.873. Desired sensitivity for HAGMA was set at ≥95%, and this was found with an AG threshold of ≥15 mEq/L (sensitivity 98.1%, specificity 34.0%). The recommended AG threshold value is ≥15 mEq/L with a high sensitivity for HAGMA. The AG should always be interpreted with the clinical context, and it should be repeated as the clinical picture evolves.