Background to the diagnosis of gestational diabetes mellitus

The observation that diabetes could develop during pregnancy and resolve thereafter dates back at least to the 19th century. The first published case description of diabetes diagnosed in pregnancy with subsequent resolution post partum was in 1828 by Heinrich Gottleib Bennewitz . He described a patient with symptoms of severe hyperglycemia, exhibiting excess production of sugar, who delivered a baby of “such robust and healthy character…you would have thought Hercules had begotten.” The baby was macrosomic and stillborn, and the mother’s symptoms resolved post partum. Subsequently, the term “gestational diabetes” was coined in 1957 by ER Carrington .

The first systematic evaluation of the oral glucose tolerance test (OGTT) for the diagnosis of diabetes in pregnancy was in the 1950s in Boston. O’Sullivan et al. used a four-sample, 100-g, 3-h OGTT and initially applied the US Public Health Service criteria, which were in common use at the time for OGTTs performed outside pregnancy . These required three of four values to exceed the threshold for the diagnosis of diabetes. Women in this study were enrolled due to additional risk factors for diabetes and were tested on several occasions during pregnancy if they had an initial normal result. The frequency of gestational diabetes mellitus (GDM) was low at 0.9%.

The need to define OGTT criteria for the pregnant population was recognized and the Boston group conducted further studies. The most widely reported cohort study involving 752 women was published in 1964 by O’Sullivan and Mahan . Given its iconic status in the world of GDM diagnosis, further description seems warranted. O’Sullivan and Mahan determined the 97th centile levels for whole blood glucose on 752 women who underwent a full (100-g, four-sample, 3-h) OGTT. Initially, they enrolled 986 women who underwent a 50-g, non-fasting “glucose challenge test” (GCT) at their first antenatal visit. Of these, 752 returned for further testing and formed the basis of their cohort. Importantly, the GCT results were not used to triage women for the full OGTT. O’Sullivan and Mahan arbitrarily decided that two abnormal values would be needed for GDM diagnosis and rounded the 2- and 3-h values for equally arbitrary reasons. The subsequent 97th centile whole blood glucose values (fasting 90, 1-h 165, 2-h 145, and 3-h 125 mg/dL; 5.0, 9.2, 8.1, and 6.9 mmol/L, respectively) were then applied to a separate cohort of 1013 women who had been followed up post partum, and the retrospective diagnosis of GDM was found to be predictive of subsequent diabetes in the mother. Importantly, this concept of “prediabetes” in the mother was of paramount interest, not the ability of OGTT results to predict adverse pregnancy outcomes.

These “O’Sullivan criteria” have since been adjusted to convert the original measurements of “whole blood sugar” (glucose and other non-glucose carbohydrate molecules using the Somogyi–Nelson method) to plasma glucose concentrations. Varying results have been obtained , but the most methodologically correct and widely accepted criteria for GDM diagnosis using venous plasma glucose (VPG) can be traced back to the original O’Sullivan cohort and are commonly referred to as the “Carpenter and Coustan criteria” .

A growing body of evidence has since accumulated associating abnormal glucose tolerance in pregnancy with adverse perinatal outcomes.