The impact of gestational diabetes mellitus on perinatal outcomes is well-established, and higher maternal hyperglycemia seems to correspond to worse clinical outcomes. Universal screening for gestational diabetes at 24-28 weeks gestation (or earlier if there are additional risk factors) is recommended using one of two clinically accepted methods: a one-step approach that involves a fasting blood glucose and then blood glucose checks at one and two hours after a 75 g glucose challenge, or the two-step method which involves a blood glucose check one hour after a 50 g challenge and a reflex three-hour glucose tolerance test for those who screen positive. Choice of screening method and diagnostic cutoffs are often institution-specific, without strong evidence at this point for a “best test.”
To that end, a research group in the Pacific Northwest and Hawaii designed a large trial to assess diagnosis of gestational diabetes and four prespecified perinatal outcomes in 23,792 pregnant patients randomized 1:1 to either one- or two-step screening approaches. Providers were made aware of the assigned screening method when they ordered gestational diabetes screening at 24-28 weeks gestation. The pragmatic trial design did not allow investigators to strictly enforce adherence to randomization, and there was disproportionate crossover between patients assigned to the one-step (fasting) screening test who received the two-step test (initial non-fasting) instead (66% vs. 92% adherence respectively). This was likely due to the opportunistic screening for non-fasting patients assigned to the one-step test (which requires initial fasting). In response to this discrepancy, additional patients were enrolled until adequate power was achieved for both groups, and an intention-to-treat (ITT) analysis with inverse probability weighting was applied to compensate for nonadherence. Six percent of patients did not undergo any screening for gestational diabetes and appeared to have overall worse clinical outcomes.
Of the patients who received any kind of screening for gestational diabetes (94%), a diagnosis was made in 16.5 percent of patients randomized to one-step screening (fasting) and 8.5 percent of those randomized to two-step screening (initial non-fasting) (unadjusted relative risk [RR] 1.94, 95% CI 1.79-2.11). There were no differences in any primary clinical outcome including large-for-gestational-age infant, composite perinatal outcome, gestational hypertension or preeclampsia, primary cesarean section, or any secondary clinical outcomes between groups when analyzed with either the preplanned ITT or adjusted ITT with inverse weighting.
If the study question was “what is the best test to screen for gestational diabetes?” then the answer depends on how you look at these data. Twice as many patients received the diagnosis of gestational diabetes with the one-step (fasting) screening method, but without differences in perinatal outcomes. Does that make it a more prognostic test? Or is this just overdiagnosis? The high crossover rates with the one-step (fasting) approach make this a little less certain, but the size of the trial and the fact that the results were similar across several analyses suggest this likely reflects a true finding despite the potential threat to validity. We must also consider the additional burden on the healthcare system and on patients imposed by a “high-risk” label, especially knowing that patients with the diagnosis of gestational diabetes are more likely to have a primary cesarean section without improvements in perinatal outcomes. Given the lack of differences in clinical outcomes and better adherence to the two-step method, maybe we should frame gestational diabetes screening the way we frame colorectal cancer screening: the best test is the one the patient is willing to do. In this case, the two-step strategy seems like the way to go.
For more information, see the topic Gestational Diabetes Mellitus (GDM) in DynaMed.