This investigation was conducted as a prospective, randomized controlled trial at Ankara Bilkent City Hospital, one of the largest tertiary referral centers in Türkiye, between 2022 and December 2023. The aim was to evaluate two different screening strategies for GDM and to assess their diagnostic yield and impact on maternal and neonatal outcomes. The study protocol was approved by the Institutional Ethics Committee (Approval No: E1-20-800). Written informed consent was obtained from all participants in accordance with the Declaration of Helsinki. NCT number is NCT04585204.

The study compared a two-step OGTT approach (50 g glucose challenge test followed by diagnostic 100 g OGTT using Carpenter–Coustan criteria) with a one-step 75 g OGTT approach based on International Association of Diabetes and Pregnancy Study Groups (IADPSG) recommendations.

Sample size estimation was based on detecting a clinically meaningful difference in GDM prevalence between screening strategies with an alpha level of 0.05 and a statistical power of 80%. The target sample size exceeded 1,400 participants to ensure adequate power for subgroup and ROC analyses.

Our randomized design allowed:

A total of 1,439 pregnant women presenting for routine GDM screening at 24–28 weeks of gestation were screened. Age ≥18 years, singleton pregnancy, no pregestational diabetes, and ability to provide consent were inclusion criteria. Pre-existing type 1 or type 2 diabetes, multiple gestation, chronic hepatic/renal/autoimmune disease, corticosteroid or glucose-altering drug use, refusal of consent, or incomplete OGTT data were exclusion criteria.

Participants were randomized in a 1:1 ratio using a computer-generated sequence prepared by an independent statistician. Allocation was concealed in sequential opaque envelopes. Participants and clinicians were not blinded due to the nature of glucose testing, but outcome assessors and statisticians remained blinded to allocation. Randomization was not stratified; however, maternal age and pre-pregnancy BMI were prespecified variables for adjusted and stratified analyses.

Group A — Two-step OGTT approach (n=719):

Group B — One-step 75 g OGTT approach (n=720):

Following diagnosis, women were classified as diet-controlled GDM (GDM-D), insulin-requiring GDM (GDM-I), or controls (normoglycemic). Women with incomplete OGTT results were excluded from per-protocol analysis but included in intention-to-treat analysis when possible. Missing categorical data (<5%) were excluded case-wise. Sensitivity analyses were conducted to evaluate the potential impact of missing data, and no significant differences were observed compared with the primary analysis.

Primary outcomes:

Secondary outcomes:

Analyses were conducted with SPSS version 25. Continuous variables were tested for normality using the Kolmogorov–Smirnov test. Parametric data are presented as mean ± standard deviation (SD), and non-parametric data as median [minimum–maximum]. Between-group comparisons were performed using the independent samples t-test or one-way ANOVA for parametric variables, and the Mann–Whitney U test or Kruskal–Wallis test for non-parametric variables. Categorical variables are expressed as frequencies and percentages and were compared using the Chi-square test or Fisher’s exact test, as appropriate.

Receiver operating characteristic (ROC) analyses were performed to evaluate the diagnostic performance of OGTT parameters for (i) prediction of polyhydramnios and (ii) differentiation between diet- and insulin-treated patients. AUC values with 95% confidence intervals, optimal cut-off values determined by the Youden Index, sensitivity, and specificity were calculated. ROC curves are presented graphically in Figures 1, 2.

Multivariable logistic regression analyses were conducted to determine independent predictors of GDM diagnosis and insulin requirement, adjusting for maternal age, pre-pregnancy BMI, family history of diabetes, and PCOS. Additional stratified analyses were performed according to maternal age categories and BMI groups to evaluate potential effect modification.

All analyses were two-tailed, and a p value <0.05 was considered statistically significant.

A total of 1,439 women were randomized into two groups: 719 underwent the two-step OGTT approach (50 g screening followed by diagnostic 100 g OGTT) and 720 underwent the one-step 75 g OGTT approach. Of these, 177 (12.3%) were diagnosed with GDM, including 121 (8.4%) diet-controlled (GDM-D) and 56 (3.9%) insulin-requiring (GDM-I) cases, while 1,241 (87.7%) served as controls.

Baseline demographic characteristics are presented in Table 1. Maternal age was comparable across groups (Diet: 30.4 ± 3.3; Insulin: 30.6 ± 3.6; Control: 29.8 ± 3.5 years; p = 0.640). BMI values were also similar (Diet: 26.7 ± 2.4; Insulin: 26.9 ± 2.6; Control: 26.3 ± 2.3; p = 0.258). Gravida and parity distributions did not differ significantly among subgroups or between screening strategies. These findings confirm that randomization achieved well-balanced baseline demographics.

Additional stratified analyses according to maternal age categories and pre-pregnancy BMI were performed, demonstrating no statistically significant interaction between screening strategy and these variables.

Clinical and obstetric characteristics are presented in Table 2. Women with GDM (both diet and insulin groups) had significantly higher prevalence of family history of diabetes, GDM in a first-degree relative, previous GDM, and previous infant >4,000 g (all p < 0.001). Other comorbidities (hypertension, history of embolism, IVF conception, prior polyhydramnios) were not significantly different between groups (all p > 0.05), confirming comparable baseline obstetric risk profiles between screening strategies.

Lifestyle characteristics are summarized in Table 3. Rates of regular exercise (25–27%), smoking (8–11%), alcohol consumption (2–4%), and vitamin supplementation (87–90%) were similar across groups (all p > 0.05). However, PCOS prevalence was higher among women with GDM (Diet: 17.8%; Insulin: 18.2%; Control: 7.9%; p < 0.001). Gestational weight gain was significantly higher in the insulin group (13.0 ± 3.8 kg) compared to diet-treated (11.7 ± 3.1 kg) and control women (12.0 ± 2.9 kg; p = 0.018). Current pregnancy polyhydramnios was numerically higher in the insulin group but did not reach statistical significance.

A total of 177 women (12.3%) were diagnosed with gestational diabetes mellitus. Although the prevalence of GDM was numerically higher in the one-step 75 g OGTT approach compared with the two-step OGTT approach, this difference did not reach statistical significance (12.9% vs 11.7%, p = .318) (Table 4).

Maternal and neonatal outcomes are summarized in Table 5. No significant differences were observed between screening strategies in polyhydramnios (6.1% vs 6.5%), hypertensive disorders of pregnancy (5.9% vs 6.1%), macrosomia (>4000 g: 9.0% vs 9.4%), preterm birth (<37 weeks: 8.1% vs 8.3%), NICU admission (10.2% vs 10.5%), or cesarean delivery (38.5% vs 39.1%) (all p > 0.05).

Rates of fetal growth–related outcomes remained low overall. The prevalence of small for gestational age (SGA) was 7.4% and intrauterine growth restriction (IUGR) was 4.2%, values consistent with expected ranges reported in comparable obstetric populations. No statistically or clinically meaningful differences were observed between the two-step OGTT and one-step 75 g OGTT screening strategies for these outcomes (all p > 0.05). These findings indicate that the increased diagnostic rate observed with the one-step approach did not translate into improved neonatal growth-related outcomes.

The predictive ability of OGTT parameters for polyhydramnios is presented in Table 6 and illustrated in Figure 1. The strongest predictors were the 100 g OGTT 2 h value (AUC = 0.816) and the 75 g OGTT 2 h value (AUC = 0.801), both demonstrating high discriminatory power. The 75 g 1 h value also showed good predictive performance (AUC = 0.782), whereas the 50 g screening test demonstrated only moderate predictive ability (AUC = 0.701).

For differentiating diet-controlled versus insulin-requiring GDM (Table 7, Figure 2), the 100 g OGTT 2 h value again performed best (AUC = 0.808). The 75 g OGTT 1 h and 2 h values also demonstrated strong discrimination (AUC range: 0.765–0.780), while the 50 g challenge test showed lower performance (AUC = 0.692). These findings highlight that post-load OGTT parameters, particularly 2-hour glucose values, provide the most clinically useful predictive information for both complications and treatment stratification.