Oral glucose tolerance testing is difficult for patients as it requires consumption of a glucose drink, and blood collection over several hours. The glucose drink can also induce nausea and vomiting – especially in pregnant women. This makes diagnostic testing for gestational diabetes mellitus (GDM) especially challenging given there are no alternatives to tolerance testing for GDM at this time. As some guidelines indicate that GDM can be ruled in with a single elevated fasting glucose, we wondered if our lab could take advantage of this requirement to eliminate tolerance testing among those quickly identified to have elevated fasting glucose.
The GDM screening method preferred by the Canadian Diabetes Association and implemented by Alberta Public Laboratories (APL) in Calgary, Alberta, Canada, is a two-step approach. First, a 50-gram screening load of glucose is administered to un-fasted patients. If the 1-hour post-load plasma glucose result falls within a grey zone (140 to 198 mg/dL; 7.8-11.0 mmol/L), the patient is recommended to complete a diagnostic 75-gram oral glucose tolerance test, which includes plasma glucose testing in the fasting state as well as 1 and 2 hours post-load. If any plasma glucose result is elevated, GDM can be ruled in. (http://guidelines.diabetes.ca/cpg/chapter36)
At APL in Calgary, over 90% of the 4850 GDM tolerance tests per year are performed at patient phlebotomy centers where staff collect and centrifuge blood before sending it to a central lab for plasma glucose testing. As reporting occurs several hours after collection, centers are equipped with glucose meters to triage fasting specimens and identify patients who should not undergo tolerance testing. Our team sought to implement a meter fasting whole blood glucose threshold for pregnant women undergoing GDM tolerance testing to reduce un-necessary tolerance testing.
But since diabetes cannot be diagnosed using meters, which are analytically inferior to core-lab analyzers, we set out to determine a threshold that would identify patients likely to have GDM. This required us to evaluate the sensitivity, specificity, positive and negative predictive values of meter whole blood glucose thresholds predicting a diagnostic plasma glucose result (≥95 mg/dL; ≥5.3 mmol/L) reported on the same specimen. To perform this analysis, we merged glucose meter data from our city-wide glucose meter server with plasma glucose from our laboratory information system by accession number.
In close collaboration with endocrinologists involved in diagnosis and treatment of GDM, we implemented a meter threshold with 25% sensitivity and 99.9% specificity for a diagnostic plasma glucose result.We chose this threshold so there would be very few false denials of the drink, which would require re-fasting, re-testing and potentially delay treatment. We also developed an automated lab information system procedure that would identify patients falsely denied the drink i.e. who had a non-GDM fasting plasma glucose result (<95 mg/dL; < 5.3 mmol/L) but had 1 and 2 hour tests manually cancelled. These changes helped reduce the number of patients having to consume the drink annually by 3% (11% of GDM diagnoses; 145 patients per year), eliminated 2 un-necessary phlebotomies in each of these patients, and saved the lab nearly $10,000 CDN.Only 3 false denials (0.06%) occurred, and these patients were quickly called back for re-testing.
So, the answer to original question is a resounding ‘YES’! However, the number of patients saved was relatively small because we had to select a threshold that balanced laboratory savings against the clinical impact of false positives and false negatives. We recommend that other large laboratories with distributed collection sites also explore using glucose meters in similar ways.
Reference(s):
Buse JD, Donovan LE, Naugler CT, Sadrzadeh SMH, de Koning L. 2018. Intervention to Reduce Unnecessary Glucose Tolerance Testing in Pregnant Women. The Journal of Applied Laboratory Medicine. 3(3): 418-428.