Clinical background
The anion gap is used to classify metabolic acidosis as high or normal anion gap, a decisive step in the diagnostic algorithm. A high anion gap points to the accumulation of unmeasured acids (lactate, ketones, toxins, uric acid), whereas a normal anion gap suggests hyperchloraemic acidosis. The problem is that the anion gap in its unmodified form presupposes a normal serum albumin concentration. Albumin is quantitatively the most important buffering anion in plasma, and in hypoalbuminaemia, which is common in intensive care patients, the anion gap falls by approximately 2.5 mmol/L per 1 g/dL fall in albumin. This can mask a substantial acidosis and lead to a high anion gap acidosis being misclassified as normal [1].
The albumin-corrected variant was developed precisely to return the anion gap to the scale that applies at a normal albumin, so that the classical reference intervals can be used even in marked hypoalbuminaemia [1].
Calculating the anion gap
The measured anion gap is calculated as:
For albumin correction, a correction term based on the deviation from normal albumin is added:
Here 4.0 g/dL is the reference value for normal serum albumin and 2.5 is the correction factor derived by Figge et al. The factor states how many mmol/L the anion gap changes per g/dL deviation in albumin. If albumin is reported in g/L rather than g/dL, the factor becomes 0.25 and the reference value 40 g/L, which gives the same result [1].
The derivation cohort consisted of 9 healthy subjects and 152 intensive care patients at a university hospital in Albany, New York, with a total of 265 arterial blood samples. Among the intensive care patients included, 49% had a serum albumin below 20 g/L. Figge et al. found a linear relationship in which each g/L fall in albumin caused the measured anion gap to underestimate the total concentration of gap anions by 0.25 mEq/L (r² = 0.94) [1].
Interpretation in practice
The reference interval for the albumin-corrected anion gap is generally 8 to 12 mmol/L, which corresponds to the classical interval for the uncorrected gap at a normal albumin. The purpose of the correction is to restore this scale.
| Corrected anion gap | Interpretation | Clinical action |
|---|---|---|
| ≤ 12 mmol/L | Normal anion gap | Metabolic acidosis, if present, is of the hyperchloraemic type. Consider gastrointestinal losses, renal tubular acidosis, or iatrogenic hyperchloraemia from saline. |
| > 12 mmol/L | High anion gap | Metabolic acidosis with accumulation of unmeasured anions. Investigate with lactate, ketones, creatinine and, where a toxic aetiology is suspected, a toxicology screen (methanol, ethylene glycol, salicylates). |
An albumin-corrected anion gap below 10 mmol/L suggests that no significant hyperlactataemia is present, with a high negative predictive value, provided the samples are drawn at the same time [2]. This can help prioritise a broad work-up, but it does not replace direct lactate measurement when shock is suspected.
Validation and performance
Several studies have validated the albumin-corrected method, particularly with respect to its ability to detect hyperlactataemia as a surrogate for unmeasured anions.
Chawla et al. (2008) studied 143 intensive care patients with 497 simultaneous arterial sample sets (311 with albumin). Mean albumin was 2.5 g/dL, the uncorrected anion gap averaged 9.0 mmol/L and the corrected anion gap 14.1 mmol/L. For the detection of hyperlactataemia (lactate > 2.5 mmol/L) the AUC was 0.70 for the uncorrected anion gap and 0.72 for the corrected, while base excess performed best (AUC 0.79). At a threshold of approximately 12 mmol/L, the corrected anion gap had a sensitivity of 94% but a specificity of only 29%. In severe hyperlactataemia (lactate > 4.0 mmol/L) the sensitivity rose to 100%, but specificity remained low (30%). The negative predictive value was satisfactory (> 88%). The conclusion was that the corrected anion gap can exclude hyperlactataemia but not confirm it, and that albumin and electrolytes must be measured in the same sample for the relationship to hold [2].
Dinh et al. (2006) retrospectively reviewed 639 sample sets from venous samples at a hospital in Hawaii. At a threshold of ≥ 12 mmol/L the uncorrected anion gap had a sensitivity of 39% and a specificity of 89%, whereas the corrected anion gap had a sensitivity of 75% and a specificity of 59%. The AUC was almost identical: 0.757 for the uncorrected and 0.750 for the corrected. The authors found that the correction gave no diagnostic advantage for the detection of hyperlactataemia. This study used venous samples, which may have influenced the results through regional effects at the sampling site [3].
Mallat et al. (2013) studied 341 intensive care patients in Lens, France, with a modelling group (n = 161) and a validation group (n = 180). They found that the anion gap corrected for albumin and lactate was an excellent surrogate for the strong ion gap (SIG), with R² = 0.96. The AUC for detecting SIG acidosis (> 8 mEq/L) was 0.974 (95% CI: 0.936 to 0.993). At an optimal threshold of > 17 mmol/L the sensitivity was 95% and the specificity 93%. The results were validated in the independent group with good agreement (bias 0.2 ± 1.8 mEq/L). This study shows that the corrected method works well for identifying unmeasured anions when lactate is included in the correction, but that the usual albumin correction without lactate correction performs less well [4].
A review by Morris and Low (2008) recommends base excess and the albumin-corrected anion gap as the most clinically useful methods for acid–base assessment, in preference to both the uncorrected gap and the more complex Stewart approach [5].
Limitations
Albumin must be measured at the same time. If albumin and electrolytes are analysed at different times or from different samples, the correlation between the corrected anion gap and the actual acid–base status is lost. Chawla et al. found that this condition is decisive; in their earlier study with non-simultaneous samples, performance was considerably poorer [2].
Low specificity for hyperlactataemia. Although albumin correction improves sensitivity for detecting unmeasured anions, specificity is low. This is because unmeasured anions in intensive care patients often comprise more than lactate. Chawla et al. found that even after correction for both albumin and lactate, the residual anion gap was raised (mean 12.6 mmol/L), indicating substantial amounts of other unmeasured anions [2]. The anion gap is therefore a screening tool, not a diagnostic test for specific aetiologies.
Does not apply to primary respiratory acidosis. The anion gap assesses the metabolic component of acid–base disturbances and should not be used to classify respiratory disorders.
Potassium and phosphate are not included. The classical formula does not include potassium, which is standard in clinical practice but may cause a small underestimation in marked hyperkalaemia. Phosphate correction may be considered in marked hyperphosphataemia, particularly in renal failure, but is not included in this calculator.
Venous samples may differ. Dinh et al. used venous samples and found no advantage from correction, unlike studies using arterial samples. The difference may be explained in part by regional variation at the venous sampling site [3].
References
- Figge J, Jabor A, Kazda A, Fencl V. Anion gap and hypoalbuminemia. Crit Care Med. 1998;26(11):1807–10. PMID: 9824071
- Chawla LS, Shih S, Davison D, Junker C, Seneff MG. Anion gap, anion gap corrected for albumin, base deficit and unmeasured anions in critically ill patients: implications on the assessment of metabolic acidosis and the diagnosis of hyperlactatemia. BMC Emerg Med. 2008;8:18. PMID: 19087326
- Dinh CH, Ng R, Grandinetti A, Joffe A, Chow DC. Correcting the anion gap for hypoalbuminaemia does not improve detection of hyperlactataemia. Emerg Med J. 2006;23(8):627–9. PMID: 16858097
- Mallat J, Barrailler S, Lemyze M, Pepy F, Gasan G, Tronchon L, Thevenin D. Use of sodium-chloride difference and corrected anion gap as surrogates of Stewart variables in critically ill patients. PLoS One. 2013;8(2):e56635. PMID: 23418590
- Morris CG, Low J. Metabolic acidosis in the critically ill: part 1. Classification and pathophysiology. Anaesthesia. 2008;63(3):294–301. PMID: 18289237
