Discrepancies in osmolal gaps and calculated alcohol concentrations.

Abstract

To estimate unexplained osmolal gaps (UEOG) during ethanol ingestions after accounting for measured ethanol concentrations and to explore the causes of observed discrepancies. DESIGN AND SPECIMENS: A random convenience sample of serum specimens with a request to perform osmometry for the detection of alcohols was studied. Serum concentrations of sodium, glucose, urea nitrogen, ethanol, osmolality, and ethanol were measured by two analytic methods to determine calculated osmolalities and osmolal gaps using two commonly used formulas. In 79 serum specimens submitted for osmometry, mean ethanol concentrations were 199 +/- 154 (SD) mg/dL; 15 specimens had no detectable ethanol. After regression of the calculated ethanol concentrations and the measured ethanol concentrations, the Dorwart and Chalmers formula showed a large constant bias of 45.217 +/- 6.414 (SE) mOsm/kg in predicting measured ethanol concentrations. The Smithline and Gardner formula was similarly as precise (r = .9786), but had a much smaller bias of -3.929 +/- 6.623 (SE) mOsm/kg. Actual mean differences between measured and calculated osmolalities (including ethanol osmols) gave an estimate of UEOG seen during ethanol ingestions. Using the less biased formula, the mean UEOG was 1.4 +/- 8.5 mOsm/kg, giving a 95% upper limit of the reference range of 18 mOsm/kg in this population with ethanol ingestions. Three significant outliers were identified and their UEOG could be explained by the presence of mannitol, isopropanol, or possibly a change in the assumed value for fractional water, normally a constant in these formulas for calculated osmolality. Other specific causes of osmolal gap and UEOG are discussed. The potential range of osmolal gaps seen during ethanol ingestions after accounting for ethanol is demonstrated to be greater than in previous studies, which have included a lower proportion of ethanol-intoxicated subjects in their study sample.