The Effect of Midpregnancy Screening for Gestational Diabetes Mellitus on Pregnancy Outcomes: The TESGO Randomized Controlled Trial

Abstract

OBJECTIVE

Newborns delivered by women with gestational diabetes mellitus (GDM) have accelerated intrauterine growth earlier than the current recommended screening period. We aimed to determine whether universal GDM screening using a single oral glucose intolerance test (OGTT) at 18–20 weeks’ gestation improves pregnancy outcomes compared with standard screening at 24–28 weeks’ gestation.

RESEARCH DESIGN AND METHODS

We conducted a dual-center, parallel, randomized controlled trial with a planned interim analysis in singleton pregnant women to compare the effect of midpregnancy screening for GDM at 18–20 weeks’ gestation and standard screening at 24–28 weeks’ gestation. GDM was universally screened and diagnosed using 75-g OGTTs and the International Association of the Diabetes and Pregnancy Study Groups criteria. The primary outcome was a composite measure of primary cesarean delivery, birth weight >90th percentile, neonatal hypoglycemia, cord serum C-peptide >90th percentile, gestational hypertension, preeclampsia, and birth trauma.

RESULTS

The trial was stopped early for futility after the interim analysis. Of the 967 women included in the intention-to-treat analysis, the primary outcome was not significantly different between the two groups. Neonatal hypoglycemia was significantly lower and neonatal adiposity in women with GDM was higher in the midpregnancy screening group compared with the standard screening group. Adverse event rates were similar between the two groups.

CONCLUSIONS

Advancing universal GDM screening to midpregnancy at 18–20 weeks’ gestation may not improve pregnancy outcomes, except for a reduction in neonatal hypoglycemia. Newborns of women diagnosed with GDM through midpregnancy screening had higher neonatal adiposity.

Graphical Abstract

Introduction

Gestational diabetes mellitus (GDM) is associated with increased risks of macrosomia and various adverse pregnancy outcomes (1). Current guidelines generally recommend screening for GDM at 24–28 weeks’ gestation (2,3) as this period coincides with a physiological increase in insulin resistance (4). The standard practice aims to diagnose and treat GDM at a time when glucose tolerance tends to deteriorate, helping to mitigate risks associated with hyperglycemia in pregnancy (5,6). However, previous studies suggested that fetal growth acceleration can begin as early as 20 weeks’ gestation, and significant fetal overgrowth was found at 28 weeks’ gestation in pregnancies affected by GDM (7,8). Moreover, beyond preventing excessive fetal growth, GDM treatment may also reduce risks for several perinatal outcomes, including cesarean delivery, gestational hypertension, preeclampsia, and birth trauma (9,10). These findings raise questions about whether advancing universal GDM screening to midpregnancy could improve these neonatal and perinatal outcomes.

While early screening and intervention of GDM for high-risk women have been explored (11,12), it remains unclear whether changing the timing of universal screening for standard GDM to an earlier gestational age, such as 18–20 weeks, would provide better outcomes compared with the standard screening at 24–28 weeks. Previous randomized controlled trials (RCTs) evaluating early diagnosis and treatment of GDM (Treatment of Booking Gestational Diabetes Mellitus [TOBOGM] study) (11) and early screening for GDM (Early Gestational Diabetes Screening in the Gravid Obese Woman [EGGO] study) (12) were conducted in high-risk populations rather than evaluating whether universal earlier screening for standard GDM would be beneficial.

To address this gap, we conducted The Effect of Midpregnancy Screening for GDM on Pregnancy Outcomes (TESGO) trial, which compared pregnancy outcomes in women universally screened for GDM at different gestational ages. Although most international organizations recommend early screening in high-risk women followed by rescreening for those without early GDM, our study design intentionally avoided repeated testing to reflect routine clinical practice in Taiwan, where screening is conducted only once. Since the single screening could influence outcomes in both women with and without GDM, we analyzed the entire study population to determine whether advancing the timing of a single universal screening could provide significant benefits or risks for pregnancy outcomes.

Research Design and Methods

Study Design

This was an open-label, dual-center, parallel, RCT with an adaptive design. The prespecified adaptation plans allowed the trial to stop early if futility was indicated by the results of a planned interim analysis. We recruited pregnant women from National Taiwan University Hospital (NTUH) and NTUH, Hsin-Chu Branch, between May 2018 and December 2021. Participants were randomized into two groups: the midpregnancy screening group (GDM screened at 18–20 weeks’ gestation) and the standard screening group (GDM screened at 24–28 weeks’ gestation). The details of randomization are described in the trial protocol (Supplementary Material). A planned interim analysis was conducted. The study was reviewed by the Institutional Review Board of NTUH, Taiwan (201710072RINA) and registered with ClinicalTrials.gov (NCT03523143).

Study Population

We assessed the eligibility of all pregnant women who visited NTUH and NTUH, Hsin-Chu Branch, for prenatal care. The inclusion criteria included women aged ≥20 years who had a singleton pregnancy and their first prenatal visit before 14 weeks’ gestation. We excluded pregnant women who had preexisting diabetes, twin or multiples pregnancy, or current exposure to any form of glucocorticoids at any dose or could not tolerate an oral glucose tolerance test (OGTT). Written informed consent was obtained from each participant before enrollment. After enrollment, all participants were screened for preexisting diabetes and were randomly assigned to the midpregnancy screening group or the standard screening group on the same day to reduce hospital visits. Participants were excluded if they had a history of diabetes, hemoglobin A_1c (HbA_1c) levels ≥6.5%, or fasting plasma glucose (FPG) levels ≥126 mg/dL (7.0 mmol/L). Participants meeting laboratory criteria for prediabetes at enrollment were included in the study but did not receive any intervention prior to the scheduled GDM screening.

Study Interventions

Participants underwent a single 75-g, 2-h OGTT screening at 18–20 weeks’ gestation or at 24–28 weeks’ gestation according to their allocation. HbA_1c levels were also measured during the OGTT. GDM was diagnosed according to the International Association of the Diabetes and Pregnancy Study Groups criteria (13) if one of the fasting, 1-h, or 2-h plasma glucose (PG) during OGTT was ≥92 mg/dL (5.1 mmol/L), ≥180 mg/dL (10.0 mmol/L), or ≥153 mg/dL (8.5 mmol/L), respectively.

Participants diagnosed with GDM in both groups immediately received nutrition counseling and lifestyle intervention. Self-monitoring of blood glucose was performed at least four times per day, including fasting glucose and 1-h or 2-h postprandial glucose of three main meals. None of the participants wore a continuous glucose monitor throughout the gestational period. Additionally, diet diary recording was recommended, which was used for discussions with dietitians. The treatment goal was according to the suggestion of the Fifth International Workshop-Conference on Gestational Diabetes Mellitus (14). When the glycemic targets were not achieved within 4 weeks of lifestyle intervention, insulin therapy was initiated according to local guidelines (15). Participants with GDM in both groups would receive a 75-g OGTT at 6–12 weeks postpartum.

The protocol for preexisting diabetes screening, GDM screening, glucose monitoring, and ultrasonography is listed in Supplementary Fig. 1. During pregnancy and delivery, the presence of any adverse outcomes was recorded. Neonatal adiposity was defined as the sum of the thickness of the triceps skinfold and subscapular skinfold above the fetal sex- and gestational age–specific 90th percentile. These measurements were conducted using a Harpenden skinfold caliper by trained specialized nurses within 3 days of birth.

Study Outcomes

The primary outcome was a composite measure of primary cesarean delivery, birth weight above fetal sex- and gestational age–specific 90th percentile (based on a nationwide singleton birth weight reference in Taiwan [16]), neonatal hypoglycemia, cord serum C-peptide ≥90th percentile, gestational hypertension, preeclampsia, and birth trauma. Neonatal hypoglycemia was defined as blood glucose values <40 mg/dL (2.2 mmol/L) within 4 h after birth or <45 mg/dL (2.5 mmol/L) from 4 to 24 h after birth. Gestational hypertension was defined as blood pressure ≥140/90 mmHg that developed after 20 weeks’ gestation. Preeclampsia was defined as new-onset hypertension after 20 weeks’ gestation, accompanied by proteinuria, maternal end-organ dysfunction, or uteroplacental dysfunction. Birth trauma was defined as either shoulder dystocia, bone fracture, nerve palsy, or any birth injury during delivery. The primary outcome was assessed during the perinatal period after randomization. Secondary outcomes were rates of preterm delivery, neonatal jaundice needing phototherapy, admission to the neonatal intensive care unit (NICU), stillbirth, fetal death at birth, neonatal adiposity and fetal growth during pregnancy recorded by ultrasonography, and the incidence of maternal diabetes after delivery.

Adverse Events

Adverse events among pregnant women with GDM who received the intervention (safety population) were assessed during the perinatal period, including severe hypoglycemia during pregnancy, maternal death, and low birth weight.

Sample Size Calculation

The incidence of the primary outcome among participants with GDM in the standard screening group was assumed to be 56.5% based on our previous prospective cohort study (17,18). A 30% reduction in the primary outcome was considered clinically meaningful to justify changes in the timing of screening, leading to an estimated incidence of 39.55% for the primary outcome among participants with GDM in the midpregnancy screening group. The one-tailed type I error rate and statistical power were set at 0.05 and 80%, respectively. With the O’Brien and Fleming (19) method applied during the interim analysis, 112 participants with GDM were required in each group to detect a statistically significant difference between the two groups. The sample size was calculated using SAS 9.4 (SAS Institute, Cary, NC). Assuming a 10% dropout rate and GDM prevalence of 12% among pregnant women in Taiwan, 1,027 participants were required to be recruited in each group. The interim analysis was planned when ∼514 participants were recruited in each group.

Statistical Analysis

Our primary analysis used an intention-to-treat approach, including all randomized participants. Baseline characteristics were summarized as mean and SD for continuous variables and frequency and percentage for categorical variables. Student t test, χ² test, or Fisher exact test were applied to compare maternal characteristics, glycemic status, rate of GDM, rate of insulin therapy, neonatal characteristics at birth, the biometric parameters of fetal growth during pregnancy, the primary outcome and secondary outcomes, and adverse events in the safety population between the two groups. Risk differences with 95% Wilson CIs were calculated. Modified Poisson regression models with robust SEs were used to estimate risk ratios and 95% CIs for the primary and secondary outcomes (20). Covariates in adjusted models were prespecified baseline characteristics considered potential confounders. Prespecified subgroup analyses were performed for participants with and without GDM, and an additional subgroup analysis compared those with and without risk factors (previous GDM, BMI >30 kg/m², age ≥40 years, first-degree relative with diabetes, previous macrosomia, or polycystic ovary syndrome) within the midpregnancy screening group. Odds ratios for neonatal adiposity, defined as a binary outcome, in relation to baseline characteristics were estimated using logistic regression as part of an exploratory analysis aimed at identifying potential associations. Available case analysis was used for missing data (Supplementary Table 1). Statistical analyses were performed using SAS 9.4.

Termination of the Trial

Statisticians conducted the interim analysis on 10 February 2022, and external experts reviewed the analysis to suggest whether the trial should continue or be stopped according to the prespecified adaptation plans. At interim, the incidence of the primary outcome was 32.81% in the midpregnancy screening group and 29.29% in the standard screening group. The z statistic was −1.13, favoring the standard screening group. Additionally, the conditional power (probability of success if the trial continued to the end) was calculated to be only 0.027%, suggesting that further enrollment would be unlikely to alter the conclusion. Furthermore, the study was conducted during the coronavirus disease 2019 (COVID-19) pandemic, and the recruitment of study participants was difficult at that time. For these reasons, the three external experts decided that the trial should be terminated early for futility. In the interval between the interim analysis and the external experts’ final decision, Taiwan experienced a peak of the COVID-19 pandemic. Although screening for eligible participants continued, no additional pregnant women agreed to participate. Therefore, the final report of the current study comprises the results of the interim analysis.

Data and Resource Availability

The data generated in this study are not publicly available due to consideration of patient privacy and consent. However, they can be made available upon submitting a reasonable request to the corresponding authors.

Results

Study Population

We screened 1,149 pregnant women (Fig. 1), with 109 declining participation and 7 excluded for glucocorticoid use. A total of 1,033 participants were recruited and randomized. Among them, 3 (0.3%) with overt diabetes were excluded, and 63 (6.1%) dropped out during follow-up. The remaining 967 participants, including 483 in the midpregnancy screening group and 484 in the standard screening group, were included in the intention-to-treat analysis.

Figure 1.

Flow diagram of the TESGO trial.

The mean age of all the participants was 33.5 ± 4.2 years. The baseline characteristics of participants were similar in the two groups (Table 1). The mean gestational age at the OGTT was 18.90 ± 0.63 weeks in the midpregnancy screening group and 25.71 ± 1.08 weeks in the standard screening group. In Supplementary Tables 2 and 3, regardless of whether participants had GDM, most baseline characteristics of participants were similar between the two groups, except that FPG and HbA_1c levels at the first prenatal visit in participants without GDM were significantly higher in the midpregnancy screening group.

Table 1.

Baseline characteristics of participants in the midpregnancy and the standard screening groups in the intention-to-treat population

Characteristic	Midpregnancy screening group (n = 483)	Standard screening group (n = 484)
Age (years)	33.5 ± 4.3	33.5 ± 4.0
Parity
1	249 (51.55)	263 (54.34)
2	198 (40.99)	197 (40.70)
3	33 (6.83)	22 (4.55)
More	3 (0.62)	2 (0.41)
History of GDM	22 (4.55)	15 (3.1)
History of hypertension	2 (0.41)	2 (0.41)
History of preeclampsia	3 (0.62)	2 (0.41)
History of gestational hypertension	4 (0.83)	2 (0.41)
History of preterm delivery	5 (1.04)	2 (0.41)
Family history of diabetes	104 (22.17)	112 (23.73)
Body height (cm)	160.39 ± 5.28	161.03 ± 5.12
Body weight before pregnancy (kg)	56.77 ± 10.02	56.96 ± 9.39
BMI before pregnancy (kg/m2)	22.04 ± 3.61	21.96 ± 3.44
Body weight* (kg)	58.11 ± 10.27	58.20 ± 9.51
BMI* (kg/m2)	22.54 ± 3.68	22.42 ± 3.49
Plasma HbA1c* (%)	5.19 ± 0.25	5.15 ± 0.27
FPG* (mg/dL)	84.41 ± 5.91	83.78 ± 6.59
Gestational age at OGTT† (weeks)	18.90 ± 0.63	25.71 ± 1.08
Center (NTUH, Hsin-Chu Branch)	90 (18.63)	97 (20.04)

Data are n (%) or mean ± SD unless otherwise indicated.

*At the first prenatal visit.

†Diagnostic tests for GDM, which were performed at 18–20 weeks’ gestation for the midpregnancy screening group and 24–28 weeks’ gestation for the standard screening group.

Maternal Glycemic Status During Pregnancy, Rate of GDM, and Neonatal Characteristics at Birth

Maternal glycemic status during pregnancy and neonatal characteristics at birth in both screening groups are summarized in Table 2. The incidence of GDM was 10.99% in the midpregnancy screening group and 14.29% in the standard screening group. HbA_1c levels in the midpregnancy screening group were slightly higher at the first prenatal visit and significantly higher during the OGTT and at 30–32 weeks’ gestation compared with the standard screening group. Neonatal characteristics at birth, including the rate of congenital anomalies, were not significantly different between the two groups.

Table 2.

Glycemic status, rate of GDM, and rate of insulin therapy of participants and clinical characteristics of neonates at birth in the midpregnancy and standard screening groups in the intention-to-treat population

Variable	Midpregnancy screening group	Standard screening group	Risk difference (95% CI)
Pregnant women
n	473	462
PG (mg/dL)
FPG at the first  prenatal visit	84.41 ± 5.91	83.78 ± 6.59	0.63 (−0.16 to 1.42)
During OGTT*
0-h PG	80.41 ± 5.64	80.11 ± 6.38	0.28 (−0.50 to 1.04)
1-h PG	132.27 ± 28.54	136.61 ± 29.93	−4.41 (−8.17 to −0.66)
2-h PG	110.84 ± 24.70	116.03 ± 26.65	−5.24 (−8.53 to −1.94)
FPG at 30–32 weeks’  gestation	81.48 ± 6.86	81.40 ± 7.65	0.08 (−0.87 to 1.03)
HbA1c (%)
At the first prenatal visit	5.19 ± 0.25	5.15 ± 0.27	0.03 (0.00 to 0.07)
During OGTT*	5.09 ± 0.28	4.94 ± 0.29	0.15 (0.06 to 0.23)
At 30–32 weeks’ gestation	5.12 ± 0.28	5.06 ± 0.29	0.06 (0.02 to 0.09)
Prior to delivery	5.41 ± 0.31	5.32 ± 0.24	0.09 (−0.03 to 0.21)
GDM	52 (10.99)	66 (14.29)	−3.29 (−7.55 to 0.97)
Insulin therapy†	2 (3.85)	4 (6.06)	−2.21 (−8.74 to 4.31)
Neonates at birth
n	469	461
Gestational age (weeks)	38.69 ± 1.48	38.68 ± 1.25	0.01 (−0.17 to 0.19)
Male sex	244 (52.03)	235 (50.98)	1.05 (−5.37 to 7.47)
Congenital anomaly	7 (1.49)	10 (2.16)	−0.67 (−2.11 to 0.78)
Birth height (cm)	49.52 ± 2.22	49.42 ± 2.01	0.10 (−0.17 to 0.37)
Birth weight (gm)	3,056.3 ± 412.3	3,043.6 ± 387.6	12.7 (−38.83 to 64.24)
Birth weight z score‡	−0.084 ± 0.914	−0.116 ± 0.918	0.032 (−0.086 to 0.150)
Head circumference (cm)	33.70 ± 1.32	33.71 ± 1.34	−0.02 (−0.19 to 0.16)
Chest circumference (cm)	32.05 ± 1.74	32.03 ± 1.63	0.02 (−0.20 to 0.24)
Subscapular skinfold (mm)	3.66 ± 0.87	3.55 ± 0.78	0.11 (−0.01 to 0.23)
Triceps skinfold (mm)	3.52 ± 0.74	3.42 ± 0.77	0.10 (−0.01 to 0.21)
Apgar score at 1 min	8.69 ± 0.70	8.67 ± 0.76	0.03 (−0.07 to 0.12)
Apgar score at 5 min	8.98 ± 0.49	9.00 ± 0.42	−0.01 (−0.07 to 0.05)
Cord serum C-peptide (ng/mL)	0.79 ± 0.40	0.74 ± 0.38	0.05 (−0.01 to 0.10)
Cord blood glucose (mg/dL)	76.48 ± 15.91	76.10 ± 16.79	0.39 (−1.81 to 2.58)

Data are n (%) or mean ± SD unless otherwise indicated.

*Diagnostic tests for GDM, which were performed at 18–20 weeks’ gestation for the midpregnancy screening group and 24–28 weeks’ gestation for the standard screening group.

†Among participants with GDM.

‡Adjusted by fetal sex and gestational age.

Supplementary Table 3 shows maternal glycemic status during pregnancy and neonatal characteristics at birth for participants with or without GDM in both screening groups. Among participants with GDM, FPG and HbA_1c levels at various time points and the proportions undergoing insulin therapy were similar between the two groups. In particular, women with GDM in the midpregnancy group had mean FPG levels of 84.67 ± 9.52 mg/dL and mean 2-h PG levels of 97.71 ± 16.57 mg/dL after 4 weeks of intervention. Additionally, FPG and 2-h PG levels at 30–32 and 35–37 weeks’ gestation were not significantly different between the two groups (Supplementary Table 4), suggesting that the glycemic control in both groups was comparable. However, participants without GDM in the midpregnancy screening group had statistically significantly higher FPG levels at the first prenatal visit and during the OGTT, as well as higher HbA_1c levels at the first prenatal visit, during the OGTT, and at 30–32 weeks’ gestation, compared with those without GDM in the standard screening group. At birth, neonates of participants with GDM in the midpregnancy screening group had thicker skinfolds compared with those in the standard screening group.

Primary Outcome, Secondary Outcomes, and Adverse Events

As shown in Table 3, the primary outcome was not statistically significantly different between the midpregnancy screening group and the standard screening group (32.81% vs. 29.29%, adjusted risk ratio 1.09, 95% CI 0.89–1.33). The individual components of the primary outcome and secondary outcomes also showed no significant differences between the two groups, except for neonatal hypoglycemia, which was statistically significantly lower in the midpregnancy screening group. After adjusting for covariates, neonatal hypoglycemia remained the only outcome that was significantly different between the two groups. Adverse events in the safety population were not significantly different between the two groups. There was neither severe hypoglycemia during pregnancy nor maternal death in the two groups. There were five (9.62%) neonates with low birth weight in the midpregnancy screening group and three (4.55%) in the standard screening group.

Table 3.

Primary and secondary outcomes of participants in the midpregnancy and standard screening groups and the crude and adjusted risk ratios for the outcomes between groups in the intention-to-treat population

Variable	Midpregnancy screening group	Standard screening group	Crude risk ratio (95% CI)	Adjusted risk ratio†† (95% CI)
n	448	437
Primary outcome*	147 (32.81)	128 (29.29)	1.12 (0.92–1.36)	1.09 (0.89–1.33)
Individual outcomes in the primary  outcome
Primary cesarean delivery	70 (14.93)	62 (13.42)	1.11 (0.81–1.53)	1.12 (0.81–1.56)
Birth weight >90th percentile	29 (6.18)	22 (4.77)	1.30 (0.76–2.23)	1.40 (0.81–2.41)
Neonatal hypoglycemia	8 (1.71)	19 (4.14)	0.41 (0.18–0.93)	0.41 (0.18–0.92)
Cord serum C-peptide >90th percentile	50 (11.55)	37 (8.71)	1.33 (0.89–1.99)	1.21 (0.81–1.82)
Gestational hypertension	5 (1.06)	3 (0.64)	1.66 (0.40–6.91)	1.74 (0.41–7.50)
Preeclampsia	10 (2.11)	8 (1.70)	1.24 (0.50–3.13)	1.19 (0.49–2.90)
Birth trauma	0 (0)	3 (0.65)	NA	NA
Primary outcome except primary cesarean delivery**	92 (21.00)	86 (20.05)	1.05 (0.81–1.36)	1.01 (0.78–1.32)
Secondary outcomes
Preterm delivery	33 (7.04)	25 (5.41)	1.30 (0.79–2.15)	1.30 (0.78–2.18)
Neonatal jaundice needing phototherapy	72 (15.38)	65 (14.13)	1.09 (0.80–1.48)	1.10 (0.81–1.50)
Admission to the NICU	23 (4.86)	31 (6.64)	0.73 (0.43–1.23)	0.72 (0.43–1.21)
Length of NICU stay (days)	0.51 ± 2.98	1.15 ± 8.91	NA	NA
Stillbirth	4 (0.83)	9 (1.86)	0.45 (0.14–1.44)	0.5 (0.16–1.62)
Fetal death at birth	0 (0)	0 (0)	NA	NA
Small for gestational age†	40 (8.53)	49 (10.63)	0.80 (0.54–1.19)	0.81 (0.54–1.21)
Neonatal adiposity‡	53 (14.52)	39 (11.44)	1.27 (0.86–1.87)	1.29 (0.86–1.93)
Uterine artery resistance§	37 (7.87)	31 (6.71)	1.14 (0.72–1.80)	1.17 (0.74–1.87)
Maternal prediabetes postpartum‖	8 (22.2)	12 (25.0)	0.67 (0.28–1.62)	0.62 (0.26–1.50)
Maternal diabetes postpartum‖	2 (5.6)	5 (10.4)	0.40 (0.08–2.06)	0.44 (0.09–2.25)
HAPO composite outcome¶	137 (30.72)	117 (26.83)	1.14 (0.93–1.41)	1.11 (0.90–1.38)

Data are n (%) or mean ± SD unless otherwise indicated. HAPO, Hyperglycemia and Adverse Pregnancy Outcomes trial; NA, not applicable.

*The primary outcome of the TESGO trial, a composite measure of adverse pregnancy outcomes, including primary cesarean delivery, birth weight >90th percentile, neonatal hypoglycemia, cord serum C-peptide >90th percentile, gestational hypertension, preeclampsia, and birth trauma. Number of participants included in the estimation of crude and adjusted risk ratios were 885 and 862, respectively.

†Defined as birth weight less than the fetal sex- and gestational age–specific 10th percentile based on the nationwide singleton birth weight reference in Taiwan.

‡Defined as sum of thickness of the triceps skinfold and subscapularis skinfold greater than the fetal sex- and gestational age–specific 90th percentile.

§Defined as uterine artery pulsatility index >95th percentile at 11–13⁺⁶ weeks’ gestation or at 22–23⁺⁶ weeks’ gestation.

‖Among women with GDM. The composite outcome of the primary outcomes in the HAPO study, including primary cesarean delivery, birth weight >90th percentile, neonatal hypoglycemia, and cord serum C-peptide >90th percentile.

**A composite measure of birth weight >90th percentile, neonatal hypoglycemia, cord serum C-peptide >90th percentile, gestational hypertension, preeclampsia, and birth trauma.

††Adjusted for age, BMI before pregnancy, family history of diabetes, plasma HbA_1c at the first prenatal visit, and FPG at the first prenatal visit.

Subgroup Analyses

Comparisons of outcomes between the midpregnancy screening group and the standard screening group in participants with or without GDM are summarized in Supplementary Table 5. Neither the primary outcome nor its individual components, including neonatal hypoglycemia, differed significantly between the two groups, regardless of GDM status. Similarly, secondary outcomes showed no significant differences between the two groups, except for neonatal adiposity, which was statistically significantly higher among participants with GDM in the midpregnancy screening group compared with those in the standard screening group. After adjusting for covariates, neonatal adiposity remained significantly different between the two groups among participants with GDM. Subgroup analyses indicated that participants with risk factors had higher risks of adverse pregnancy outcomes than those without risk factors in the midpregnancy screening group (Supplementary Table 6). Additionally, participants with more gestational weight gain or higher FPG or HbA_1c at the first prenatal visit were at risk for neonatal adiposity (Supplementary Table 7).

Maternal BMI, Body Weight, and Weight Change During Pregnancy

Supplementary Figs. 2A and 3A show that maternal BMI, body weight, and weight change did not differ significantly between the two groups for all participants and those with or without GDM.

Fetal Growth During Pregnancy

Supplementary Figs. 2B–F and 3B–F show that fetal biometric parameters during pregnancy in all participants and those with or without GDM were not different between the two groups.

Conclusions

In this study, we investigated the impact of universal GDM screening at 18–20 weeks’ gestation compared with standard screening at 24–28 weeks’ gestation. Since participants in either the midpregnancy screening group or the standard screening group underwent only a single GDM screening without rescreening, aligned with routine clinical practices in Taiwan, those identified as having GDM in the two groups were different. In the midpregnancy screening group, participants with GDM were identified and treated at 18–20 weeks’ gestation, and those with potential late GDM (but without early GDM) were not identified or treated. In the standard screening group, participants with GDM identified were treated at 24–28 weeks’ gestation, and those with potential early GDM were not identified and received delayed treatment or no treatment according to the results of the OGTT at 24–28 weeks’ gestation. We conducted outcome analyses in participants with GDM, those without GDM, and the entire study population given that universal screening would impact not only women with GDM but also those without. The findings showed no significant difference in the primary outcome between the two groups, except for a lower rate of neonatal hypoglycemia in the midpregnancy screening group. Additionally, newborns of participants with GDM in the midpregnancy screening group had a statistically significantly higher risk of neonatal adiposity compared with those of participants in the standard screening group. On the other hand, participants without GDM in the midpregnancy screening group had statistically significantly higher FPG and HbA_1c during pregnancy compared with those without GDM in the standard screening group.

The lack of significant differences in most outcomes between the two groups may be attributed to the relatively short interval between screening times. With only a 6–8-week gap, the impact on outcomes may be limited, especially given that fetal growth becomes more prominent after 24 weeks (8,21,22). Another important consideration is the timing of insulin initiation. In this trial, glycemic control in participants with GDM was evaluated using blood test results and self-monitoring of blood glucose records. Insulin therapy was initiated if glycemic targets were not achieved within 4 weeks of lifestyle intervention. Previous studies reported that 30–50% of women with GDM required insulin therapy, typically initiated 3–4 weeks after diagnosis (23,24). However, in Taiwan, 95.3% of women with GDM achieve glycemic targets with lifestyle intervention alone, and only 4.7% require pharmacotherapy (25), which is similar to our trial, further supporting the effectiveness of lifestyle intervention within the first 4 weeks after diagnosis. Therefore, the delayed pharmacotherapy may have had a relatively minor impact in our study population. To summarize, based on our findings, there is currently insufficient evidence to support advancing universal GDM screening from the current standard of 24–28 weeks’ gestation to an earlier window in midpregnancy.

Recently, the TOBOGM trial highlighted the benefit of early GDM screening and treatment in high-risk women (6). This supports a new model of care for GDM (26), recommending first-trimester screening for early GDM due to higher risks of requiring insulin therapy and encountering pregnancy complications (27), especially in high-risk women. For women without early GDM or those not tested for it (e.g., low-risk women), the model suggests screening for late GDM at 24–28 weeks’ gestation. In our study, midpregnancy screening may have been too early for late GDM, missing cases that develop closer to 24–28 weeks, as indicated by the higher FPG and HbA_1c levels in participants without GDM in the midpregnancy screening group. This suggests that screening for late GDM in low-risk women should be delayed to account for the peak insulin resistance period. Conversely, midpregnancy screening may be too late for timely detection and intervention of early GDM, as indicated by the higher neonatal adiposity in participants with GDM in the midpregnancy screening group. While previous studies suggested fetal growth acceleration in women with GDM from 20 weeks’ gestation (7,8), our data indicate that screening and intervention at 18–20 weeks’ gestation may not be early enough to yield benefit. Given that glycemic levels during normal pregnancy are higher in the first trimester and decline in the second trimester (28) and that higher glycemic levels in women with GDM could occur as early as 13–14 weeks’ gestation (29), an earlier screening period before 18–20 week’ gestation may be necessary. Additionally, we found that participants with risk factors in the midpregnancy screening group had a higher risk of adverse pregnancy outcomes, further supporting early GDM screening in high-risk populations. Overall, our findings support the new paradigm advocating screening promptly for early GDM and type 2 diabetes and screening late for late GDM (26). However, this study did not directly compare the newly proposed model with the current standard screening at 24–28 weeks, highlighting the need for further RCTs to assess the effectiveness of both approaches.

Interestingly, the lower rate of neonatal hypoglycemia observed warrants further investigation. Previous studies reported that screening and treatment of women diagnosed with GDM at 24–28 weeks’ gestation cannot mitigate risks of neonatal hypoglycemia (6,9,10). Our finding suggests that earlier glycemic management, even with a small screening interval difference, could benefit neonatal outcomes related to glucose metabolism. However, as neonatal hypoglycemia was a secondary outcome in this trial, this result should be interpreted as hypothesis-generating rather than practice-changing. Conversely, earlier screening at 18–20 weeks’ gestation alone may not be sufficient to reduce the risks of other key outcomes, such as rates of large-for-gestational-age infants, gestational hypertension, and preeclampsia, particularly given the relatively low prevalence of large-for-gestational-age infants and hypertensive disorders of pregnancy in Taiwan (30–32). Given that neonatal hypoglycemia is a major cause of NICU admission and is associated with increased health care costs and maternal-infant separation (33), further investigation of the effect of early screening for GDM using neonatal hypoglycemia as the primary outcome, along with cost-effectiveness and family-centered outcome assessments, would be promising. Another intriguing finding is that in the GDM subgroup, midpregnancy screening was associated with higher risks of neonatal adiposity despite similar baseline characteristics and comparable glycemic control between the two groups. This implies that the increased risk of neonatal adiposity in women with GDM diagnosed earlier may be influenced by factors beyond maternal glycemic levels, such as maternal dyslipidemia, which warrants further investigation.

There are several limitations to our study. First, the trial was terminated early due to futility, potentially reducing the statistical power to detect differences in outcomes. Additionally, the sample size calculation assumed similar GDM detection rates between the two groups, which may have overestimated the rate in early gestation when insulin resistance is less pronounced, thereby contributing to the trial being underpowered. Second, 2.11–13.88% of data for the variables included in the primary outcome were missing primarily due to early pregnancy termination, loss to follow-up, and incomplete data collection (Supplementary Table 8). Among these variables, the proportion of missing data for cord serum C-peptide >90th percentile was 11.55% in the midpregnancy screening group and 13.88% in the standard screening group. The primary reason for missing these data was that some participants chose to store cord blood for personal use and declined laboratory testing. Since the missingness was unrelated to the study groups, and the proportion of missing data was similar between groups, its impact on our findings may be limited. Third, the study was conducted during the COVID-19 pandemic, possibly influencing participant behavior and adherence to lifestyle interventions, which might have affected the study outcomes. Fourth, nutritional data were not collected; therefore, the impact of dietary modifications on pregnancy outcomes could not be assessed. Fifth, we acknowledge that while skinfold thickness measurement using the Harpenden caliper is a widely used and noninvasive method to estimate subcutaneous fat in neonates, it has certain limitations. It does not capture visceral fat, and concerns regarding potential interobserver variability may further complicate interpretation (34). To minimize these limitations, all measurements in our study were performed by trained specialized nurses using standardized protocols. Finally, our study included exclusively Asian pregnant women, necessitating additional research in diverse populations to evaluate the generalizability of our findings.

In summary, advancing universal GDM screening to 18–20 weeks’ gestation may not improve pregnancy outcomes compared with the standard screening at 24–28 weeks, except for a reduction in neonatal hypoglycemia. The findings support current guidelines and the newly proposed model for GDM care, which recommend screening at 24–28 weeks for late GDM. Given that neonatal hypoglycemia is an important cause of NICU admission and is associated with increased health care costs and parental-infant separation, future trials focusing on neonatal hypoglycemia, cost-effectiveness, and family-centered outcomes are warranted to determine the value of universal early screening for GDM.

This article contains supplementary material online at https://doi.org/10.2337/figshare.30047794.

Funding Statement

Support for the TESGO trial was provided by Ministry of Health and Welfare, Taiwan, grants MOHW107-TDU-B-211-123002, MOHW108-TDU-B-211-133002, MOHW110-TDU-B-211-124002, and MOHW111-TDU-B-211-134002; Ministry of Science and Technology, Taiwan, grant MOST 109-2314-B-030-006-MY3; and the Liver Disease Prevention and Treatment Research Foundation, Taiwan.