BRIEF REPORT

A Rapid Qualitative Test for SuspectedEthylene Glycol PoisoningHeather Long, MD, Lewis S. Nelson, MD, Robert S. Hoffman, MD

AbstractObjectives: Many hospitals must send out ethylene glycol (EG) samples to a reference laboratory, anddelays in diagnosis and treatment may occur. A qualitative colorimetric test (ethylene glycol test [EGT]kit), already in use by veterinarians, gives results in 30 minutes with little expertise or cost. The EGTreliably detects the presence of EG in spiked human serum samples. The objective of this study was toprospectively assess the sensitivity and specificity of the EGT kit in actual clinical samples submitted forEG testing by the criterion standard gas chromatography (GC).

Methods: Blood samples from patients with suspected toxic alcohol poisoning submitted to a referencelaboratory were tested by GC. An investigator blinded to the GC results tested the same sample withthe EGT kit following the manufacturer’s instructions and using the internal control. Three physiciansalso blinded to the GC results categorized the sample as positive for EG, negative, or inconclusive.Interrater reliability was assessed with a kappa statistic (j). Results of the EGT kit testing were thencompared to those from GC testing.

Results: Data are reported on 24 samples submitted. By GC, 15 samples were confirmed for EG (range27–281 mg ⁄ dL), 5 were confirmed for methanol (ME; range 64–101 mg ⁄ dL), and 4 were negative for bothalcohols. The EGT was unanimously positive in all confirmed EG samples and negative in all ME sam-ples. In one of the negative samples, an ambiguous result occurred and was counted as a false-positive.Interobserver agreement with the EGT was high (j = 0.909; 95% confidence interval [CI] = 0.735 to 1.0).Sensitivity and specificity were 100% (95% CI = 70% to 100%) and 88.8% (95% CI = 52% to 100%),respectively.

Conclusions: The EGT appears to be a reliable qualitative test in cases of suspected human EGpoisoning.

ACADEMIC EMERGENCY MEDICINE 2008; 15:688–690 ª 2008 by the Society for Academic EmergencyMedicine

Key words: ethylene glycol, poisoning, diagnosis, laboratory test

E thylene glycol (EG) is a sweet-tasting, readilyavailable toxic alcohol found most commonly inautomotive radiator antifreeze. Ingestion of as

little as 1.5 mL ⁄ kg 100% EG can be fatal. Definitive ther-apy for EG poisoning is expensive and potentially harm-ful and therefore is not readily initiated without a clearhistory of ingestion, appropriate clinical findings, or aconfirmatory laboratory test. Because of the complexityof the most commonly used tests for EG (i.e., high-per-formance liquid chromatography [HPLC] or gas chroma-

tography [GC]), few hospitals have the capacity for STATin-house testing.1,2 Samples must therefore be sent to areference laboratory, and delays in diagnosis and treat-ment may occur. Delayed treatment of patients with EGtoxicity is associated with a poor outcome, includingrenal failure, and death.3

A qualitative colorimetric test (ethylene glycol test[EGT] kit, PRN Pharmacal Inc., Pensacola, FL), already inuse by veterinarians, gives qualitative results in 30 min-utes with little expertise or cost. Per the manufacturer,the EGT can be used to detect EG in feline or canineblood, plasma, or serum. However, its use for the rapiddiagnosis of human EG poisoning is unstudied. Werecently demonstrated in an in vitro study that the EGTreliably detects the presence of EG in samples of humanplasma to which EG has been added.4 The manufacturerof the EGT cautions that the test will yield false-positiveresults in the presence of propylene glycol, which wealso confirmed in our in vitro study.4 Our objective in thisstudy was to prospectively assess the sensitivity andspecificity of the EGT in comparison to GC using actual

ISSN 1069-6563 ª 2008 by the Society for Academic Emergency Medicine688 PII ISSN 1069-6563583 doi: 10.1111/j.1553-2712.2008.00150.x

From the Department of Emergency Medicine, St. Peter’s Hos-pital (HL), Albany, NY; and the New York City Poison ControlCenter and the New York University School of Medicine (LSN,RSH), New York, NY.Received February 11, 2008; revision received March 28, 2008;accepted April 1, 2008.Presented at the Society for Academic Emergency MedicineAnnual Meeting, Orlando, FL, 2004.Address for correspondence and reprint requests: HeatherLong, MD; e-mail: Heather.long@mindspring.com.

clinical samples submitted to a reference laboratory forEG testing.

METHODS

Study DesignThis was a prospective, observational, blindedlaboratory investigation with convenience samplingusing human blood specimens. The study protocol wasapproved by the institutional review board for the NewYork City Department of Health and Mental Hygiene.

Study Setting and SamplesWe used blood samples submitted to New York CityDepartment of Health and Mental Hygiene’s PublicHealth Laboratory (PHL) for testing. Blood samples frompatients with suspected toxic alcohol poisoning submit-ted to the PHL for confirmatory testing were used. ThePHL is a reference laboratory; as such, specimens aresubmitted at the convenience of the requesting hospital.Time from blood draw to testing and blood handlingtechniques vary with the submitting institution, butbecause of the potential volatility of specimens, onlysealed specimens are accepted for testing. Once receivedby the PHL, specimens are tested immediately if receivedduring normal business hours or at the start of the nextbusiness day if received after hours.

Study ProtocolAt the PHL, specimens from patients with suspectedtoxic alcohol poisoning were tested by GC by a certifiedlaboratory technician using their standard protocol. Asingle investigator (HL) blinded to the GC results testedan aliquot of serum from the same sample with the EGTkit following the manufacturer’s instructions. Each EGTkit contains an internal control. In the final step of test-ing, a color reagent is added to the sample being testedand to the internal control. Color development in the testtube as intense as or greater than that of the control tubeis considered positive for EG. Three physiciansinstructed in the EGT, but blinded to the GC results andto the nature of the study, then compared the colorchange of the test sample with the control tube and cate-gorized the test sample as positive for EG, negative, orinconclusive. Results obtained with use of the EGT werethen compared to those obtained by GC testing.

MeasurementsColor development in the test sample was compared withthe internal control using the EGT. Agreement amongobservers was assessed regarding color development inthe test sample and the intensity of color change whencompared with the control. The results obtainedwith the EGT were compared with those obtained by GC.

Data AnalysisSensitivity and specificity were calculated using stan-dard equations. Interrater reliability among the threephysicians was assessed with a multirater kappa statistic(j). Ninety-five percent confidence intervals (CIs) werecalculated. Data were analyzed with SPSS Version 14(Chicago, IL). The study was terminated after 24 samplesbecause of limited funding.

RESULTS

Results of testing for the 24 samples of blood submittedare shown in Table 1. By GC, 15 samples were con-firmed for EG (range 27–281 mg ⁄ dL), 5 were confirmedfor methanol (ME; range 64–101 mg ⁄ dL), and 4 werenegative for all alcohols. All blood samples in whichGC confirmed the presence of EG were also unani-mously positive by EGT. All samples in which GC test-ing confirmed the presence of ME were unanimouslynegative by EGT. One of the samples that tested nega-tive for both toxic alcohols by GC was interpreted aspositive by one physician and as negative by the othertwo. This ambiguous result was counted as a false-posi-tive for EG (see Sample 14 in Table 1). The lowest EGlevel in a sample that tested positive for EG by GC was27 mg ⁄ dL. That sample was positive for EG by EGT.

Sensitivity and specificity of the EGT were 100%(95% CI = 70% to 100%) and 88.8% (95% CI = 52% to100%), respectively. Interobserver agreement with theEGT results was high (j = 0.909; 95% CI = 0.735 to 1.0).

DISCUSSION

Rapid diagnosis and institution of definitive therapy iscritical to improving patient outcome following EGingestion. Without a clear history of ingestion, the cor-rect diagnosis may be difficult to ascertain, as the differ-ential diagnosis of an elevated anion-gap metabolicacidosis (the hallmark of EG poisoning) is extensive. Sur-rogate markers for EG such as hypocalcemia, an ele-vated osmol gap, crystalluria, and urine fluorescence are

Table 1Comparison of Results Obtained with Criterion Standard High-performance Liquid Chromatography (HPLC) and by theEthylene Glycol Test (EGT) Kit

Sample Number GC Result EGT Result

1 EG 275 Positive2 EG 165 Positive3 ME 64 Negative4 Negative Negative5 EG 177 Positive6 ME 82 Negative7 EG 76 Positive8 EG 132 Positive9 ME 101 Negative10 EG 66 Positive11 EG 222 Positive12 ME 67 Negative13 EG 108 Positive14 Negative Positive*15 EG 116 Positive16 EG 49 Positive17 EG 235 Positive18 EG 89 Positive19 EG 54 Positive20 EG 281 Positive21 ME 75 Negative22 EG 27 Positive23 Negative Negative24 Negative Negative

EG = ethylene glycol; GC = gas chromatography; ME = meth-anol.*Discrepancy.

ACAD EMERG MED • July 2008, Vol. 15, No. 7 • www.aemj.org 689

readily obtainable; however, there are many false-posi-tives and -negatives, and significant toxicity may occur inthe absence of abnormalities of any of these markers.5-8

Furthermore, even when the history is suggestive of EGpoisoning, the diagnosis may be difficult to confirm clini-cally. This occurs in part because the anion gap meta-bolic acidosis develops only after EG is metabolized toglycolic acid, and as such, the anion gap may be normalearly after ingestion.

The most valuable laboratory test for EG is a directmeasurement of the serum EG level. GC and HPLC arethe most common methods for determining these levels.While many hospitals employ GC in their laboratoriesfor certain analytes, EG testing requires a separate, dedi-cated column. Since a dedicated GC column is expensiveto maintain and rarely used, most hospitals elect to sendout EG samples to a reference laboratory. Although ittakes less than 1 hour to run the test, turnaround timemay be many hours or even days. Alternative rapid tests,such as the enzyme-multiplied immunoassay techniqueexist, but suffer from a high false-positive rate, especiallywhen lactate is elevated, as commonly occurs in patientswith an undifferentiated metabolic acidosis.9

Standard therapy for EG poisoning involves inhibitingalcohol dehydrogenase with either fomepizole or etha-nol. Fomepizole is very safe and could be used empiri-cally in a patient with a metabolic acidosis of unknownetiology; however, its cost is prohibitive and it is eithernot stocked by many hospital pharmacies or reserved forconfirmed cases. Ethanol causes central nervous system(CNS) depression and venous irritation and requiresclose monitoring to maintain a therapeutic serum con-centration. Additionally, because ethanol causes cardio-vascular, CNS, and respiratory depression, its empiricuse in patients who are clinically ill and not likely to bepoisoned is concerning. Hemodialysis, alone or incombination with either ethanol or fomepizole, isconsidered definitive therapy. However, many nephrolo-gists require a confirmed diagnosis before institutingtherapy because of the risks and expense of hemo-dialysis.

The high interobserver agreement and sensitivity ofthe EGT suggest it is a reliable test for the presence ofEG in samples of blood submitted for EG testing.Implementation of this rapid qualitative test could aidthe clinician in the early initiation of therapy.

Furthermore, the EGT appears able to detectclinically relevant EG concentrations. An EG level of25 mg ⁄ dL or greater by GC is considered an indicationfor treatment.10 In our previous in vitro study, the EGTwas able to detect levels of EG as low as 1 mg ⁄ dL.4 Thelowest EG level in an actual human sample that testedpositive by GC in this study was 27 mg ⁄ dL. That samplewas also positive for EG by EGT testing. It remainsunknown how the EGT would perform in actual speci-mens with lower EG levels.

LIMITATIONS

None of the submitted samples were mixed (i.e., positivefor both EG and ME) nor were any concomitant ethanollevels documented. This would not be expected to inter-fere with a qualitative test like the EGT, but it remains

unstudied. Furthermore, this was a highly selected groupof laboratory specimens; in actual clinical practice, mostsamples tested for suspected toxic alcohol would beexpected to be negative. As a result, positive and nega-tive predictive values are not reported. All testing byEGT for this study was conducted by a single investiga-tor skilled in performing this test. It is unclear how wellthis test would perform in the clinical setting if it werecarried out by physicians. This may be irrelevant,because implementation of EGT in a clinical settingwould likely require that the test be performed bytrained laboratory technicians, given current restric-tions on bedside testing. Even with such a restriction,the turnaround time would still be less than that associ-ated with a send-out test.

CONCLUSION

Our study suggests that the EGT is a sensitive and spe-cific qualitative test in cases of suspected human EGpoisoning. Continued testing is warranted.

The authors gratefully acknowledge certified public healthlaboratory technicians Raksha Mehta and Tibor Szekely for theirassistance.

References

1. Church AS, Witting MD. Laboratory testing in etha-nol, methanol, ethylene glycol, and isopropanol tox-icities. J Emerg Med. 1997; 15:687–92.

2. Kearney J, Rees S, Chiang WK. Availability of serummethanol and ethylene glycol levels: a national sur-vey [abstract]. J Toxicol Clin Toxicol. 1997; 35:509.

3. Hylander B, Kjellstrand CM. Prognostic factors andtreatment of severe ethylene glycol intoxication.Intensive Care Med. 1996; 22:546–52.

4. Long H, Nelson LS, Hoffman RS. Potential utility ofa rapid ethylene glycol bedside test [abstract]. JToxicol Clin Toxicol. 2002; 40:649.

5. Glaser DS. Utility of the serum osmolal gap in thediagnosis of methanol or ethylene glycol ingestion.Ann Emerg Med. 1996; 27:343–6.

6. Walker JA, Schwartzbard A, Krauss EA, ShermanRA, Eisinger RP. The mission gap: a pitfall in thediagnosis of alcohol intoxication by osmometry.Arch Intern Med. 1986; 146:1843–4.

7. Jacobsen D, Hewlett TP, Webb R, Brown ST,Ordinario AT, McMartin KE. Ethylene glycol intoxi-cation: evaluation of kinetics and crystalluria. Am JMed. 1988; 84:145–52.

8. Winter ML, Ellis MD, Snodgrass WR. Urine fluores-cence using a Wood’s lamp to detect the antifreezeadditive sodium fluorescein: a qualitative adjunctivetest in suspected ethylene glycol ingestion. AnnEmerg Med. 1990; 190:663–7.

9. Eder AF, Dowdy YG, Gardiner JA, Wolf BA,Shaw LM. Serum lactate and lactate dehydrogenasein high concentrations interfere in enzymatic assayof ethylene glycol. Clin Chem. 1996; 42:1489–91.

10. Cheng JT, Beysolow TD, Kaul B, Weisman R, Fein-feld DA. Clearance of ethylene glycol by kidneys andhemodialysis. J Toxicol Clin Toxicol. 1987; 25:95–108.

690 Long et al. • RAPID QUALITATIVE TEST FOR ETHYLENE GLYCOL