Gestational diabetes in a high-risk population: Using the fasting plasma glucose to simplify the diagnostic algorithm

Mukesh M. Agarwal, Gurdeep S. Dhatt, John Punnose, Gertrude Koster

Research output: Contribution to journalArticlepeer-review

31 Citations (Scopus)


Objective: To evaluate the value of fasting plasma glucose (FPG) in screening a high-risk population for gestational diabetes mellitus (GDM). Study design: During an 8-month period, 1685 pregnant women underwent the one-step 75 g oral glucose tolerance test (OGTT) as a part of a universal screening program. The receiver operating characteristic (ROC) curve was used to analyze the performance of the FPG. Results: 333 (19.8%) women had GDM (WHO criteria). The area under the ROC curve of FPG to detect GDM was 0.639 (95% CI 0.603-0.674), which reflected the degree of the FPG histogram overlap in women with and without GDM. A FPG threshold of 4.7 mmol/l reached the minimally acceptable sensitivity of 78.1% with a corresponding unacceptable specificity of 32.2%. 508 (31%) women were below this threshold, at a negative predictive value of 85.6%. The FPG at higher thresholds with acceptable specificity had poor sensitivity and positive predictive value to be useful. Conclusion: Though the high false positive rate at any FPG threshold with adequate sensitivity makes the FPG an inappropriate test for GDM screening, the FPG has the potential to avoid nearly one-third of the cumbersome OGTTs at the expense of missing one-fifth of pregnant women with milder GDM.

Original languageEnglish
Pages (from-to)39-44
Number of pages6
JournalEuropean Journal of Obstetrics and Gynecology and Reproductive Biology
Issue number1
Publication statusPublished - May 1 2005
Externally publishedYes


  • Fasting glucose
  • Gestational diabetes
  • Screening

ASJC Scopus subject areas

  • Reproductive Medicine
  • Obstetrics and Gynaecology


Dive into the research topics of 'Gestational diabetes in a high-risk population: Using the fasting plasma glucose to simplify the diagnostic algorithm'. Together they form a unique fingerprint.

Cite this