Reference: Diabetes Care. 2025 Sep 23:dc250084 early online
Practice Point: Early universal screening for gestational diabetes mellitus (GDM) with a single test at 18-20 weeks does not improve maternal or fetal outcomes compared to standard screening at 24-28 weeks.
EBM Pearl: Each item in a composite outcome is similar to another piece of spaghetti thrown against the wall to see if it sticks. More items = higher likelihood to see an effect from an intervention.
Bonus Pearl: Even if adequately powered at the start, trials that are ended early for futility have inherently lower statistical power because of uncaptured events, potentially leaving important undiscovered differences.
Most professional organizations agree that universal screening for gestational diabetes mellitus (GDM) should occur at 24-28 weeks gestation, with earlier screening for those at increased risk. A group out of Taiwan recently wondered if earlier universal screening (treating everyone as though they are high risk) improves maternal and fetal outcomes, citing the idea that there is accelerated growth in fetuses born to gestational parents with GDM that occurs before the 24-28 week diagnosis that could cause potential harm. Unfortunately, the study they published doesn’t really answer this question.
The “TESGO” (The Effect of midpregnancy Screening for Gestational diabetes on pregnancy Outcomes) group randomized 967 patients with a singleton pregnancy and no prior diabetes to either standard screening with a 75 g oral glucose tolerance test at 24-28 weeks or ‘midpregnancy’ screening at 18-20 weeks. Patients in the midpregnancy group who screened negative were not later rescreened. Patients diagnosed with GDM in either group were managed first with lifestyle changes and later with insulin if lifestyle changes were inadequate. The trial was ended early for futility at a planned interim analysis when midpregnancy screening delivered no improvement in the primary outcome, a 7-item composite that combined cesarean delivery, birth weight > 90th percentile, neonatal hypoglycemia, cord serum C-peptide > 90th percentile, gestational hypertension, preeclampsia, and birth trauma.
Even though the results won’t change current practice, they do signal that an earlier, single screen is not overall better when it comes to screening all comers for GDM. To be clear, these data do not speak to the benefit of earlier screening in pregnant patients at high risk for GDM, which is an entirely different question (and answer). But as it pertains to the specific question of whether the 7 outcomes that made up the composite are, as a whole, “better” with earlier single screening, these data can be believed. In general, we are not fans of large, loosely related composites that mix both clinical and surrogate measures. Many-item composites may increase event rates enough to find statistical differences between groups, but even so, the clinical interpretation is often muddled. (In other words, the larger the number of items in a composite outcome, the more likely one of those items will demonstrate a difference. Large composites increase sensitivity at the price of specificity.) So in this case, if a 7-item composite in an adequately powered trial didn’t demonstrate a statistically significant difference, there was no difference to be found. Ah, but was it adequately powered? That has become a trick question.
As with all trials, the authors calculated how many participants were needed to adequately power this trial based on estimated event rates over a period of time. We have to keep in mind that any trial that is ended early for futility automatically limits its power to detect differences, simply because some events that were destined to occur were not captured within the shortened study duration. And this discrepancy is not evenly distributed across outcomes, as many of the more serious events, such as death, naturally occur later in time, regardless of the intervention.
With that in mind, if you are wondering why this study was ended early for futility rather than being allowed to continue despite the similar data at the interim analysis, you’re asking a fair question. The short answer is: babies. We have a much, much lower tolerance for the possibility of harm when it comes to pregnancies and infants. And looking at some potentially important (but only hypothesis-generating) differences between groups at the interim analysis led to a realization that there was a potentially dangerous flaw in the study design: patients assigned to the midpregnancy group who screened negative at 18-20 weeks were not rescreened later, and as such, investigators may have missed the diagnosis in those with late GDM, and thereby left them untreated. Not a great outcome, indeed.
So again, if the question is whether the 7 outcomes that made up the composite are, as a whole, “better” with earlier single screening, the answer seems to be no. A repeat trial that is better powered to find differences and includes a safety second screen is needed to answer the author’s original question of whether earlier universal screening might be a potentially better study.
For more information, see The Effect of Midpregnancy Screening for Gestational Diabetes Mellitus on Pregnancy Outcomes: The TESGO Randomized Controlled Trial in PubMed.
DynaMed EBM Focus Editorial Team
This EBM Focus was written by Dan Randall, MD, MPH, FACP, Senior Deputy Editor at DynaMed. Edited by Alan Ehrlich, MD, FAAFP, Executive Editor at DynaMed and Associate Professor in Family Medicine at the University of Massachusetts Medical School; Katharine DeGeorge, MD, MS, Senior Deputy Editor at DynaMed and Associate Professor of Family Medicine at the University of Virginia; Claire Symanski, PhD, Medical Editor and Team Lead for ENT at DynaMed; McKenzie Ferguson, PharmD, BCPS, Senior Clinical Writer at DynaMed; Rich Lamkin, MPH, MPAS, PA-C, Clinical Writer at DynaMed; Matthew Lavoie, BA, Senior Medical Copyeditor at DynaMed; Hannah Ekeh, MA, Senior Associate Editor II at DynaMed; and Jennifer Wallace, BA, Senior Associate Editor at DynaMed.