Three Abnormal Glucose Values Mark Highest Risk in Gestational Diabetes

In a retrospective cohort study of 700 Chinese women, researchers found that the number of abnormal time-points on a standard oral glucose tolerance test (OGTT) was associated with the risk of adverse perinatal outcomes. The study, published in Frontiers in Endocrinology, classified women with gestational diabetes mellitus (GDM) based on how many of the three glucose measurements exceeded diagnostic thresholds. The results indicate that women with all three values abnormal faced a significantly higher risk of complications compared to those with one or two abnormal values, although the relationship was not a simple linear progression.

Gestational diabetes is currently diagnosed when any one of three plasma glucose values is abnormal during a 75-g OGTT performed between 24 and 28 weeks of gestation. This “any-one-of-three” diagnostic rule is deliberately inclusive, ensuring that women with mild hyperglycaemia are identified. However, this binary approach assigns a woman with a single mildly elevated value to the same diagnostic category as a woman with severe, multi-time-point hyperglycaemia. The research team sought to determine whether the count of abnormal values provides additional prognostic information beyond the standard binary diagnosis of GDM.

The study was conducted at the Second Affiliated Hospital of Guangxi Medical University in Nanning, China. The analytic cohort comprised 700 women who delivered between January 2024 and December 2025 and had complete OGTT records. Of these, 351 women were normoglycaemic, meaning all three glucose values were below the International Association of Diabetes and Pregnancy Study Groups (IADPSG) thresholds. The remaining 349 women had GDM and were further stratified into three phenotypes: GDM-1 (one abnormal time-point, n=176), GDM-2 (two abnormal time-points, n=134), and GDM-3 (all three abnormal, n=39).

The primary outcome was a pre-specified composite of adverse perinatal events, including preterm birth before 37 weeks, macrosomia (birth weight ≥4000 g), neonatal hypoglycaemia, and neonatal asphyxia. Notably, the researchers excluded neonatal unit transfer from this primary composite because admission to a neonatal unit is heavily influenced by local institutional policies rather than purely by neonatal morbidity. A secondary five-component composite, which included neonatal unit transfer, was also analysed to allow comparison with previous studies.

The incidence of the primary composite outcome varied markedly across the groups. It occurred in 9.4% of normoglycaemic women, 18.2% of women with GDM-1, 18.7% of women with GDM-2, and 41.0% of women with GDM-3. After adjusting for maternal age, pre-pregnancy body mass index, and parity, the odds of experiencing the primary composite were significantly higher in all GDM phenotypes compared to controls. The adjusted odds ratio for GDM-1 was 1.95, for GDM-2 it was 1.87, and for GDM-3 it was 5.42. This pattern suggests that the excess risk is concentrated in the most severe phenotype rather than increasing uniformly with each additional abnormal value.

Metabolic profiles also differed across the phenotypes. Women in the GDM-3 group had the highest fasting plasma glucose, the highest 2-hour glucose levels, and the highest homeostatic model assessment of insulin resistance (HOMA-IR). Conversely, beta-cell function, estimated by HOMA-β, was lowest in the GDM-3 group. These metabolic differences align with the pathophysiological model where greater insulin resistance requires compensatory hyperinsulinaemia, which eventually leads to beta-cell decompensation and simultaneous elevation of multiple glucose values.

The study highlights that the count of abnormal OGTT values should be interpreted as a marker of underlying glycaemic severity rather than an independent causal exposure. When the researchers adjusted for early-pregnancy fasting glucose, gestational weight gain, insulin treatment, and glycaemic control, the association for the secondary composite outcome attenuated. This suggests that part of the observed risk is mediated by the severity of the diabetes and the treatment it provoked. The within-GDM slope per additional abnormal time-point was not statistically significant, further supporting the view that the relationship is not a graded dose-response.

Limitations of the study include its retrospective, single-centre design, which limits the ability to exclude residual and unmeasured confounding. Factors such as prior GDM, chronic hypertension, and socioeconomic status were not recorded. Additionally, the GDM-3 group was small, with only 39 women, leading to wide confidence intervals for phenotype-specific estimates. The near-equal numbers of women with and without GDM in the dataset reflect the source database composition rather than a matched design, meaning the study cannot be used to estimate GDM prevalence in the general population.

Despite these limitations, the findings suggest that the number of abnormal OGTT time-points is a readily available risk-stratification tool. It can be read directly from the diagnostic test at no additional cost. The data do not support de-escalating care for women with a single abnormal value, as they remained at increased risk for preterm birth and fetal distress. Instead, the count may help identify women who could benefit from closer surveillance, such as earlier glycaemic reviews or more frequent fetal growth monitoring.

The authors emphasize that these findings are associative, not causal. Prospective multicentre validation is required before pattern-based stratification can be recommended for clinical practice. The study contributes to the growing body of evidence that OGTT-defined subgroups are biologically and prognostically distinct, moving beyond the simple binary diagnosis of gestational diabetes.

By distinguishing between the policy-sensitive neonatal unit transfer and other perinatal outcomes, the researchers demonstrated that the association between OGTT phenotypes and adverse outcomes is robust. The transfer-excluded composite produced stronger associations, reinforcing that the findings are not an artefact of local admission practices. This approach provides a more transportable endpoint for future studies in different healthcare settings.

The distribution of phenotypes in this Chinese cohort was consistent with previous observations that post-load hyperglycaemia predominates under IADPSG criteria. Isolated 2-hour elevation was the most common single-time-point pattern, while combined 1-hour and 2-hour elevation without fasting elevation was the most common multi-time-point pattern. This distribution underscores the importance of considering the specific pattern of hyperglycaemia, not just the count, in future research.

In conclusion, while the number of abnormal OGTT time-points does not establish a validated clinical subtype, it serves as a useful indicator of glycaemic severity. It highlights the need for nuanced risk assessment in gestational diabetes management. Further research is needed to determine whether phenotype-tailored surveillance improves outcomes, but the current data suggest that a one-size-fits-all approach to GDM management may overlook significant variations in risk among affected women.

At the study centre, all participants diagnosed with gestational diabetes received standard care under a uniform clinical pathway aligned with national guidelines, ensuring that treatment intensity was not initially determined by the number of abnormal test values. This regimen comprised structured medical nutrition therapy supervised by a dietitian, physical activity advice, regular obstetric clinic reviews every two to four weeks with ultrasound fetal growth assessments, and capillary glucose targets below 5.3 millimoles per litre for fasting and 6.7 millimoles per litre postprandially. Insulin was prescribed only when lifestyle modifications failed to maintain target levels. Despite this uniform protocol, management requirements diverged sharply across phenotypes. Only 1.1 percent of women in the GDM-1 group and 3.0 percent in the GDM-2 group required insulin, compared to 35.9 percent of women with all three abnormal values. Clinicians also judged late-pregnancy glycaemic control to be poor in 35.9 percent of the GDM-3 cohort, compared to approximately 9 percent in the two milder gestational diabetes groups.

Maternal baseline characteristics also revealed clear differences across the study groups. Maternal age rose across categories, with advanced maternal age of 35 years or older increasing from 23.9 percent among controls to 27.8 percent in GDM-1, 34.3 percent in GDM-2, and 46.2 percent in GDM-3. Pre-pregnancy weight followed a similar upward trajectory, while gestational weight gain declined from an average of 13.06 kilograms in normoglycaemic women to 9.42 kilograms in the GDM-3 group. During data preparation, the authors identified and corrected two clerical anomalies in the hospital database. First, the pre-existing pre-pregnancy body mass index field had erroneously incorporated delivery weights rather than pre-pregnancy weights, inflating values by an average of 4.9 kilograms per square metre. Second, an administrative flag for advanced maternal age conflicted with recorded chronological ages in nearly half of flagged entries. Sensitivity analyses refitting all multivariable models with the original uncorrected body mass index field yielded virtually identical odds ratios, confirming that the analytical corrections did not distort the observed risk patterns.

When examining specific secondary complications, the investigators identified distinct risk profiles while addressing sparse event numbers through Firth penalised-likelihood regression and Benjamini-Hochberg false discovery rate control. Caesarean delivery rates rose from 32.8 percent in controls to 39.2 percent in GDM-1, 54.5 percent in GDM-2, and 61.5 percent in GDM-3, yielding adjusted odds ratios of 1.82 for GDM-2 and 2.20 for GDM-3 relative to controls. Severe neonatal complications were heavily concentrated in the small GDM-3 group, including macrosomia with an adjusted odds ratio of 5.89 based on six events and neonatal hypoglycaemia with an adjusted odds ratio of 4.03 based on five events. Beyond overall counts, sub-analyses demonstrated internal heterogeneity among single-value elevations: women with isolated one-hour hyperglycaemia had an adjusted odds ratio of 1.73 for adverse outcomes, whereas isolated two-hour elevations showed an adjusted odds ratio of 1.40 that did not reach statistical significance.

Subject of Research: Number of abnormal oral glucose tolerance test time-points and risk of adverse perinatal outcomes in gestational diabetes: a retrospective cohort study

Article Title: Number of abnormal oral glucose tolerance test time-points and risk of adverse perinatal outcomes in gestational diabetes: a retrospective cohort study

Article References: Su, J., Chen, Y., Tong, J., Chen, H., Huang, L., Huang, Y., & Deng, L. (2026). Number of abnormal oral glucose tolerance test time-points and risk of adverse perinatal outcomes in gestational diabetes: a retrospective cohort study. Frontiers in Endocrinology, 17, Article 1995750. https://doi.org/10.3389/fendo.2026.1995750

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1995750

Keywords: Number, abnormal, oral, glucose, tolerance, test, time-points, risk, adverse, perinatal, outcomes, gestational, diabetes, retrospective, cohort