Toggle Health Problems and D

Low serum Magnesium readings in ICU were wrong 80 percent of the time – July 2017

Total and ionized magnesium testing in the surgical intensive care unit - Opportunities for improved laboratory and pharmacy utilization.

J Crit Care. 2017 Jul 11;42:147-151. doi: 10.1016/j.jcrc.2017.07.026. [Epub ahead of print]
Yeh DD1, Chokengarmwong N2, Chang Y3, Yu L3, Arsenault C4, Rudolf J5, Lee-Lewandrowski E5, Lewandrowski K5.


Serum Magnesium testing for LOW Magneium now seems pointless.
Should be OK, however, to avoid HIGH Magnesium levels

Ionized fraction (iMg) is the physiologically active form of magnesium (Mg); total Mg may not accurately reflect iMg status. Erroneously "low" Mg levels may result in unnecessary repetitive testing.

From 11/2015 to 01/2016, patients ordered for Mg from a pilot ICU also had iMg tested. Weighted kappa statistic was used to assess agreement between Mg categories (low, normal, high). Predictors of unnecessary repeated Mg testing and repletion using data were explored through logistic regression models using GEE techniques to account for repeated measurements in both bivariate and multivariable analyses.

There were 470 Mg/iMg paired measurements from 173 patients. The weighted kappa statistic was 0.35 (95%CI 0.27-0.43) indicating poor agreement in assessment of magnesium status. Of the 34 Mg samples reported as "low", only 6 (18%) were considered "low" using concurrent iMg testing.
In the multivariable models,

  • history of atrial fibrillation (aOR=1.61, 95%CI 1.16-2.21, p=0.004) and
  • concomitant metoclopramide (aOR=1.71, 95%CI 1.03-2.81, p=0.036)

were significant predictors of unnecessary repeat Mg testing.

In the surgical ICU, categorical agreement (low, normal, high) was poor between Mg and iMg.
Over 80% of "low" total Mg values are erroneous and may result in unnecessary additional measurements and repletion.

PMID: 28732315 DOI: 10.1016/j.jcrc.2017.07.026

Created by admin. Last Modification: Saturday September 2, 2017 15:33:23 GMT-0000 by admin. (Version 2)
See any problem with this page? Report it (FINALLY WORKS)