Abstract: FR-PO631
A Validated Anion Gap Threshold for High Anion Gap Metabolic Acidosis
Session Information
- Fluid and Electrolytes: Clinical - Acid-Base, Magnesium, Calcium, Phosphorus
November 08, 2019 | Location: Exhibit Hall, Walter E. Washington Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Fluid and Electrolytes
- 902 Fluid and Electrolytes: Clinical
Authors
- Chionh, Chang Yin, Changi General Hospital, Singapore, Singapore
- Yeon, Wenxiang, Changi General Hospital, Singapore, Singapore
- Poh, Cheng boon, Changi General Hospital, Singapore, Singapore
- Koduri, Sreekanth, Changi General Hospital, Singapore, Singapore
- Roy, Debajyoti M., Changi General Hospital, Singapore, Singapore
Group or Team Name
- Changi REnal MEdicine REsearch (CREMERE) Group
Background
The anion gap (AG), calculated by AG = [Na+]–[Cl–]–[HCO3–], is often used to screen for acid-base disorders but cut-off levels used by clinical texts were based on empirical data.
We recently sampled 300 healthy volunteers and the mean AG was 13±2mEq/L (manuscript in preparation). The proposed reference range (±2SD; central 95th percentile) was 9-17mEq/L. This study was to define a cut-off value for high AG metabolic acidosis (HAGMA).
Methods
Data from ICU patients from a prior study was used. Blood samples were classified into 2 groups: no HAGMA, or HAGMA present due to lactic acidosis, ketosis, citrate toxicity or severe uremia. The association of AG and HAGMA was tested by the Mann-Whitney U test. ROC analysis of AG was undertaken with optimal cut-off determined by Youden index.
Results
From 1,545 blood samples, 400 had adequate data. The median age was 64.7yrs, weight 60.1kg, 31.6% were females and 170 had HAGMA. With HAGMA, median AG was 19mEq/L (IQR 17–22) vs 15mEq/L (IQR 13–17) without HAGMA (P<0.001).
AG has an AUC of 0.802 and the optimal cut-off was ≥17mEq/L (Fig. 1). The false negative rate (FNR) was 19.5%; false positive rate (FPR) 32.1%.
Other AG thresholds were analysed (Table 1). As HAGMA may be life-threatening, a lower FNR is desired. The best FNR of 13.2% was with AG ≥13mEq/L, corresponding to mean AG of healthy persons, but FPR increased to 53.0%.
Conclusion
The recommended AG cut-off for HAGMA is ≥13mEq/L. This provides the best sensitivity but at the expense of specificity. As false negatives can still occur, acid-base status should be evaluated clinically, aided by repeated measurements of AG.
Table 1: Sensitivity & Specificity of AG Thresholds for HAGMA
| AG cut-off | Specificity | False Positive | Sensitivity | False Negative |
| ≥13 (Population Mean) | 20.0% | 53.0% | 53.0% | 13.2% |
| ≥15 (Population Mean + 1SD) | 44.3% | 45.9% | 45.9% | 15.7% |
| ≥17 (Population Mean + 2SD) | 73.5% | 32.1% | 32.1% | 19.5% |
Fig. 1: ROC curve of AG for HAGMA