ASN's Mission

ASN leads the fight to prevent, treat, and cure kidney diseases throughout the world by educating health professionals and scientists, advancing research and innovation, communicating new knowledge, and advocating for the highest quality care for patients.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on Twitter

Kidney Week

Abstract: FR-PO270

Anion Gap (AG) Improves the Prediction of Ionized Hypocalcemia (HC) in Critical Care

Session Information

Category: Fluid and Electrolytes

  • 902 Fluid and Electrolytes: Clinical

Authors

  • Yap, Ernie, SUNY Downstate, Brooklyn, New York, United States
  • Roche-Recinos, Andrea, Danbury Hospital, Danbury, Connecticut, United States
  • Goldwasser, Philip, VA New York Harbor, Brooklyn, New York, United States
Background

The physiologically active form of total serum calcium (sCa) is ionized (iCa). It is reported in arterial (ABG) and venous (VBG) gas panels. The rest of sCa is bound, mainly to albumin (ALB), or chelated by small anions. Low ALB and high chelating anion levels are common in critical care patients (pts). A popular adjustment of sCa for ALB alone (BMJ 1977) yields a corrected value (cCa) that doesn’t detect abnormal iCa well in such pts, perhaps because it ignores the chelating anions. AG reflects the levels of both such anions and ALB, and has been shown to correlate with iCa (Nordin 1989). We tested whether the diagnosis of HC in critical care could be improved by accounting for both AG and ALB.

Methods

In 769 critical care pts, simultaneous values of sCa (units: mg/dL), ALB (g/dL), and AG (mEq/L) were paired retrospectively with closely-timed (<20 min. apart) values of iCa (mM) derived from 309 ABGs (mean pH=7.390) or 460 VBGs (mean pH=7.363). We defined HC as iCa <1.10 in ABGs and, to adjust for the mean pH difference, as <1.11 in VBGs. The prevalence of HC was 28% (86/309) in ABGs and 21% (95/460) in VBGs. A model to predict HC in the ABG cohort was generated by multiple logistic regression with sCa, ALB, and AG as candidate predictors. Next, this model was validated in the VBG cohort. Areas under the ROC curve (AUC) of the model and of cCa were compared. Data are summarized as means [95%CI].

Results

sCa (p<10-10), ALB (p<.004), and AG (p<10-5) were each significant independent predictors of HC in the ABG cohort, with odds ratios of 0.064 [.029-.142], 2.87 [1.40-5.87], and 1.19 [1.11-1.28], respectively. The overall odds were 6,853,807×0.064sCa× 2.87ALB×1.19AG. This model had a better AUC than cCa in both the ABG cohort (0.89 [.84-.93] vs 0.82 [.77-.88]; p<.005) and the VBG cohort (0.89 [.86-.93] vs 0.82 [.77-.87]; p=.0002). The predicted HC rate and the observed rate showed good agreement in the VBG cohort, divided into 4 prediction subgroups, each 25% wide (Table).

Conclusion

Adjusting sCa for both AG and ALB improves the diagnosis of HC. The absolute probability estimates of the logistic model are intuitive and may help clinicians decide when to obtain an ABG.

Predicted and Observed HC Rates in the VBG Cohort
Prediction group (N)<25% (370)25 to <50% (44)50 to <75% (21)≥75% (25)
Mean predicted rate7%35%64%90%
Observed rate9%48%71%96%

Funding

  • Veterans Affairs Support