The study by Zhu et al. (1) corroborates that the initial fasting plasma glucose (FPG) results of the oral glucose tolerance test (OGTT) can help to circumvent the performance of over half the OGTTs needed to work up pregnant women for gestational diabetes mellitus (GDM) in China. Thus, the algorithm to simplify the diagnosis of GDM suggested by us (2) earlier, which was accepted by the Chinese Ministry of Health (3), has been validated by the authors.
However, a few facts need to be stressed. Strictly speaking, one cannot use a test (FPG, in this case) to screen for any disease (GDM, in this case) when that test is a part of a diagnostic test (OGTT, in this case). Using the same argument, receiver operating characteristic curves become inappropriate to test the value of FPG for screening, as the authors have done. Thus, in our original study, we avoided calling the initial FPG test of the OGTT a “screen” for GDM. Our “rule in-rule out” algorithm using FPG was a practical modification (of the current American Diabetes Association recommendations for GDM) to decide whether a pregnant woman needs to proceed with the OGTT. Therefore, our belief is that it is erroneous to construe the FPG as a screening test, and it is more than just semantics.
It is worth noting that the mean FPG level of Chinese women was much lower than that in our study, involving mainly Arab and South Asian women. Thus, in China, an FPG of 5.1 mmol/L ruled in a diagnosis of GDM in far fewer women compared with our study (12.1 vs. 28.9%), but an FPG of 4.4 mmol/L ruled out a diagnosis of GDM in many more women (38.2 vs. 21.7%), respectively. However, the number of OGTTs potentially avoided has remained about the same: approximately half—and this may argue for expecting similar results in other countries.
We would like to emphasize that for this algorithm to be effective the turnaround time for determining the FPG level is very critical. It must be <1 h, which is often a tall order for many laboratories. Otherwise, the pregnant woman would wait unnecessarily, and the subsequent 1-h sample (for the OGTT) cannot be collected in time. To improve turnaround times, we have successfully experimented with fasting capillary glucose levels (4), using the same logic to avoid the OGTT. Thus, though our FPG algorithm has been reproduced, the more practical fasting capillary glucose algorithm remains to be replicated.
Duality of Interest.
No potential conflicts of interest relevant to this article were reported.
