The association between a low 50-gram, 1-hour glucose challenge test value and neonatal morbidity

Abstract Objective To evaluate the association between a low 50-gram, 1-hour glucose challenge test (GCT) value and adverse maternal and neonatal outcomes among patients receiving care at a single center tertiary care academic hospital. Methods We performed a retrospective cohort study of pregnant patients with a documented result of a 50-gram, 1-hour GCT performed ≥24 weeks 0 days gestation at a single tertiary care academic hospital from 2013–2021. Patients with a low GCT value, defined as cohort specific ≤10th percentile (<82 mg/dL), were compared to patients with a GCT value ≥82 mg/dL who were not diagnosed with gestational diabetes (GDM) to examine adverse maternal and neonatal outcomes. Additionally, these comparisons were repeated across patients with low GCT (<82 mg/dL), those with a GCT ≥82 mg/dL without diagnosis of GDM (heretofore referred to as normal glycemic screening) and patients diagnosed with GDM. Our primary outcome was a composite neonatal morbidity variable, inclusive of stillbirth, neonatal death, neonatal hypoglycemia with neonatal intensive care unit (NICU) admission, neonatal hyperbilirubinemia with NICU admission, respiratory distress with NICU admission, and/or small for gestational age (SGA). Multivariable logistic regression modeling was used to examine the association of low GCT value and the composite neonatal morbidity outcome, compared to those with the normal glycemic screening. Results Of 36,342 eligible patients, 3,789 (10.4%) had a low GCT value of <82 mg/dL, 30,729 (84.6%) had a GCT value ≥82 mg/dL and were not diagnosed with GDM, and 1,824 (5.0%) had a diagnosis of GDM. Patients with a low GCT value were significantly less likely to be diagnosed with hypertensive disorder of pregnancy (HDP) (12.4% vs 16.3%, p < .01), undergo cesarean delivery (22.8% vs 29.9%, p < .01), or experience postpartum hemorrhage (7.8% vs 9.4%, p < .01) as compared to patients with normal glycemic screening. Compared to newborns whose mothers had normal glycemic screening, newborns of mothers with a low GCT value were significantly more likely to experience the composite morbidity outcome (OR 1.17; 95% CI 1.08–1.27); this persisted after adjusting for potential confounders (aOR 1.18; 95% CI 1.09–1.29). Conclusion A low maternal GCT value after 24 weeks gestation is significantly associated with an increased risk of morbidity in the newborn, driven by higher rates of SGA. Patients with a low GCT value may have underlying maternal hypoglycemia or other glycemic dysregulation affecting fetal development and may benefit from enhanced antenatal surveillance.


Introduction
Most investigation of glycemia in pregnancy has focused on the upper end of the spectrum.As a result, definitions of glucose intolerance and the associations between hyperglycemia and maternal, obstetric, and perinatal morbidities are well elucidated [1,2].
In contrast, the clinical relevance of hypoglycemia in pregnancy is poorly understood as it remains significantly understudied [3].There is evidence from both animal and human metabolic studies that while the first two-thirds of gestation promotes maternal anabolism, this shifts in the third trimester to a catabolic state to support the rapidly growing fetus, which can predispose to maternal hypoglycemia [4,5].Studies in pregnant mice demonstrate that during periods of hypoglycemia, nutrients such as glucose and amino acids are shunted away from fetal development, resulting in small pup size as well as cranial, ocular, and skeletal malformations [6,7].However, correlates in human studies, such as the extent and clinical significance of maternal hypoglycemia, are sparse.
The 50-gram, 1-h oral glucose challenge test (GCT) is the most common screening modality for gestational diabetes (GDM) in the United States [8] and is typically performed at the start of the third trimester as maternal anabolism shifts to catabolism.However, previous studies have reported conflicting results regarding the association between a low GCT value and neonatal outcomes such low birthweight, neonatal hypoglycemia, and neonatal intensive care unit (NICU) admission [9][10][11][12].Therefore, the aim of this study was to examine if a low value on the 50-gram, 1-h GCT test completed at 24 weeks 0 days gestation was associated with increased neonatal morbidity as compared to a GCT value that was above the low value threshold.We hypothesized that amongst our cohort, maternal-infant dyads with maternal GCT values in the lowest 10 th percentile would be at increased risk of morbidity as compared to maternalinfant dyads with maternal GCT values >10 th percentile.

Materials and methods
We performed a retrospective cohort study of all pregnant patients with a 50-gram, 1-h GCT value obtained at or greater than 24 weeks 0 days gestation who delivered at Yale New Haven Hospital, a single urban tertiary-care academic center in New Haven, Connecticut (United States) from October 1, 2013, when our electronic medical record (EMR) system was introduced, to October 2, 2021.Patients with pre-gestational diabetes mellitus and multifetal gestations were excluded.Gestational age was calculated based on the inputted estimated due date in the EMR as ascertained by primary obstetrician according to the American College of Obstetricians and Gynecologists (ACOG) dating criteria [13].During the study period, early risk-based GDM screening based on recommendations from the American College of Obstetricians and Gynecologists [1] with oral glucose loading was performed, and individuals diagnosed with early GDM did not undergo repeat testing after 24 weeks gestation.There were no important changes to our EMR system or GDM screening guidelines at our hospital during the study period.A GCT value of 140-199mg/dL is considered an abnormal screen and prompts the completion of a 100-gram, 3-h glucose tolerance test (GTT).GDM is diagnosed if the GCT result of 200mg/dL or if the GTT has two or more abnormal values per Carpenter-Coustan criteria [14].
The threshold for a low GCT value was determined by calculating the lowest 10 th percentile of GCT values amongst our cohort, which was equivalent to <82 mg/dL.We then categorized patients into three groups: low GCT group (GCT <82 mg/dL), normal glycemic screening group (GCT 82 mg/dL without diagnosis of GDM), and GDM group (diagnosis of GDM).Patients who had a GCT value 82 mg/dL but were not diagnosed with GDM (e.g.patients with a GCT value of 140 to 199 mg/dL who did not meet Carpenter-Coustan criteria on GTT) were included in the normal glycemic screening group.
Our primary outcome was a composite neonatal morbidity measure, which included stillbirth, neonatal death, neonatal hypoglycemia with neonatal intensive care unit (NICU) admission, neonatal hyperbilirubinemia with NICU admission, respiratory distress with NICU admission, and/or small for gestational age (SGA).The composite is a binary field such that meeting criteria for any of the individual components meets criteria of fulfilling the neonatal morbidity composite.Stillbirth was defined as intrauterine demise after 24 weeks of gestation for this study, given our inclusion criteria required a GCT after 24 weeks.Neonatal death was defined as death within the first 28 days of life or hospital discharge, whichever came first.Neonatal hypoglycemia with NICU admission was defined as NICU admission during the birth hospitalization encounter with concomitant presence of billing and diagnostic codes for neonatal hypoglycemia.Neonatal hyperbilirubinemia with NICU admission was defined as NICU admission during the birth hospitalization encounter with concomitant presence of billing and diagnostic codes for neonatal hyperbilirubinemia.Neonatal respiratory distress with NICU admission was defined as NICU admission during the birth hospitalization encounter with concomitant presence of billing and diagnostic codes for neonatal respiratory distress, meconium aspiration syndrome, or bronchopulmonary disorder.SGA was defined as a newborn birthweight <10 th percentile, adjusting for gestational age at birth and infant sex per Aris et al. [15] Maternal and neonatal charts were linked based on delivery admission encounter.
All data were extracted from the EMR system.Diagnoses of pre-gestational diabetes, multifetal gestation, chronic hypertension, and maternal immunocompromised states (systemic lupus erythematosus, inflammatory bowel disease, and antiphospholipid syndrome) were determined based on diagnoses and billing codes.Maternal characteristics including race and ethnicity and primary language were obtained as documented in the EMR.Race and ethnicity were categorized as Hispanic, non-Hispanic Asian, non-Hispanic Black or African American, non-Hispanic White, or None of the above.Race and ethnicity were included as maternal variables given prior literature suggesting an association between certain race and ethnicities and glycemic dysregulation with increased risks of diabetes [1].Smoking was defined as active tobacco use during pregnancy or those who quit smoking during pregnancy.Metformin and selective serotonin reuptake inhibitor (SSRI) exposure were defined as documentation of active medication use at any point during pregnancy.Noncommercial insurance captured patients with Medicaid or Medicare, self-pay, free or discounted health care programs, or missing or unknown payer.Gestational weight gain was categorized as below, met, or exceeded Institute of Medicine (IOM) recommendations based on pre-pregnancy body mass index (BMI), gestational weight gain, and adjusted for gestational age at delivery per previous work by Woolfolk et al. [16] Hypertensive disorders of pregnancy (HDP) included gestational hypertension, preeclampsia with or without severe features, eclampsia, or hemolysis, elevated liver enzymes, and low platelet count syndrome.Postpartum hemorrhage (PPH) was defined as the presence of billing or diagnostic codes for PPH, a documented estimated blood loss (EBL) of >1000 ml at delivery, or receipt of red blood cell transfusion during the delivery admission.All billing and diagnosis codes used in this study are shown in Supplemental Table S1.
Baseline characteristics and maternal and neonatal outcomes were compared using 2-level and 3-level independent variable comparisons across GCT groups.For the whole study cohort, a 3-level independent variable comparison was used to compare patients with a GCT value <82 mg/mL (low GCT group) vs. patients with a GCT value 82 mg/dL without diagnosis of GDM (normal glycemic screening group) vs. patients with a diagnosis of GDM (GDM group).Then, for the subset of patients who did not have a diagnosis of GDM, we used a 2-level independent variable to compare patients with a low GCT value <82 mg/mL (low GCT group) to patients with a GCT value 82 mg/dL without a diagnosis of GDM (normal glycemic screening group).Bivariate analyses were performed using Chi-square and Fisher exact tests as appropriate for categorical variables.We examined the association between both the 2-level and 3-level independent variable and a composite neonatal morbidity as well as each of the individual components of the composite.Logistic regression modeling was performed for the 2-level independent variable to examine the association of low maternal GCT value <82 mg/dL (using normal glycemic screening group as the referent) and the neonatal morbidity composite outcome.Baseline maternal characteristics with p-values <0.05 in bivariable analyses of the 2-level comparisons were included as covariates in the multivariable logistic regression model.We additionally included statistically significant peri-delivery variables that could clinically impact neonatal morbidity.Unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were calculated for the risk of neonatal morbidity composite among pregnancies with low GCT compared to normal glycemic screening group.Furthermore, adjusted odds ratios were calculated for risk of neonatal morbidity for each covariate that was included in the logistic regression model as compared to a referent.For covariates with a significant proportion of missing observations, a separate "missing" category was created for this variable to be used in multivariable modeling.Because our main goal was to explore the maternal and neonatal outcome differences between patients with low GCT versus normal glycemic screening results, we did not conduct modeling with the 3-level independent variable.All analyses were performed using SAS 9.4 (SAS Institute, Cary, NC).This study received institutional review board approval with a waiver of informed consent.

Results
During the study period, 36,518 patients met eligibility criteria.Among these, 176 (0.5%) patients were excluded due to missing or illogical GCT value, or missing covariate or outcome data.Of the remaining patients, 3789 (10.4%) had a GCT value <82mg/dL, 30,729 (84.6%) had a GCT value 82 mg/dL and were not diagnosed with GDM, and 1,824 (5.0%) were diagnosed with GDM.
Two -Way Comparison: When compared to patients with normal glycemic screening, patients with a low GCT value <82 mg/dL were more likely to be age <19 years, nulliparous, identify as non-Hispanic Black, speak English primarily, have a pre-pregnancy body mass index (BMI) < 30 kg/m 2 , have noncommercial health insurance, and have smoked in pregnancy (Table 1).Compared to patients with normal glycemic screening, those with a low GCT were less likely to have chronic hypertension, maternal autoimmune condition such as systemic lupus erythematosus, inflammatory bowel disease, or antiphospholipid syndrome, or have been exposed to metformin.Patients with a low GCT value had lower frequencies of hypertensive disorder of pregnancy and significant bleeding at time of delivery and had higher frequencies of vaginal birth compared to patients with a normal glycemic screening.The frequency of the neonatal morbidity composite was significantly higher among infants born to mothers with a low GCT value compared to those born to mothers with a normal glycemic screening (21.0%vs 18.5%, p < .01).This was largely driven by increased rates of SGA newborns in the low GCT group compared to the normal glycemic screening group (13.9% vs 9.6%, p < .01).In the logistic regression model, we found that newborns of mothers with a low GCT value during pregnancy were significantly more likely to experience morbidity compared to newborns of mothers with a normal glycemic screening result (OR 1.17; 95% CI 1.08-1.27).After adjusting for potential confounders including maternal age, parity, race and ethnicity, primary language, prepregnancy BMI, chronic hypertension, maternal autoimmune conditions, metformin exposure, insurance status, smoking, hypertensive disorders of pregnancy, PPH, and mode of delivery, the statistical significance persisted between low maternal GCT value and composite neonatal morbidity (aOR 1.18; 95% CI 1.09-1.29)compared to normal glycemic screening (Table 2).In addition, adjusted odds ratios were also calculated for the risk of neonatal morbidity for each of the covariates included in the final logistic regression model (Supplemental Table S2).Of note, there was a significant proportion of participants with missing pre-pregnancy BMI amongst our cohort.When we performed the logistic regression model excluding pre-pregnancy BMI from the above list of variables, our point estimate for our primary independent variable was similar.Therefore, pre-pregnancy BMI was maintained in the final model.
Three-Way Comparison: These bivariate tests of association were generally consistent with analyses of the 3-level independent variable including the GDM group.However, characteristics of the low GCT and GDM groups were more contrasting (Table 1).For example, 4.9% of participants in the low GCT group had chronic hypertension vs 7.1% in the normal glycemic group vs 9.9% in the GDM group.This pattern was seen across most variables.Among the GDM group, the frequency of the neonatal morbidity was 26.1%; p < .01.

Discussion
In this large contemporary U.S. cohort from a tertiary care academic hospital, we demonstrate that a cohortspecific lowest 10 th percentile maternal GCT value of <82mg/dL is associated with significantly increased risk of neonatal morbidity as compared to those with normal glycemic screening in pregnancy.The significant risk was largely driven by the increased frequency of SGA seen among newborns of patients with a low GCT value.
Previous studies published between 1999 to 2013 in the United States have examined the association between a low 1-h, 50-gram GCT value and neonatal outcomes, but results were mixed regarding the associations with birthweight, NICU admission, and neonatal outcomes such as hypoglycemia [9][10][11][12][17][18][19].More contemporary international studies have also found conflicting evidence regarding neonatal outcomes such as neonatal hypoglycemia and birth trauma [20][21][22][23][24][25][26][27].A recently published systematic review and meta-analysis capturing 16 studies of both U.S. and international cohorts found that a low GCT value (defined as <90mg/dL) was associated with increased risk of SGA neonate (OR 1.43; 95% CI, 1.28-1.60)as compared to patients with a normal GCT, but did not find differences in other neonatal outcomes such as NICU admission or respiratory distress [28].However, in this meta-analysis, the most recent U.S. study that was included was published in 2013 and is a secondary analysis of a Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network multicenter observational cohort study which recruited from 2002 to 2007 [12].In addition, the largest U.S. study in this meta-analysis was published in 2000 with 7,384 participants and was a case-control study, and the second largest included U.S. study was published in 2001 with 1,076 patients and was a retrospective cohort study.These studies may not adequately reflect a contemporary U.S. birthing cohort.With rising rates of advanced maternal age, obesity, hypertensive disorders, and diabetes, among other comorbidities, all of which may impact placental function and fetal growth [29], our study provides a much-needed reexamination of the association between a low GCT value and neonatal morbidity in a modern large U.S.-based cohort.f PPH is defined as the presence of billing or diagnostic codes for PPH, a documented EBL of 1000 ml at delivery, or receipt of red blood cell transfusion during the delivery admission.g Among vaginal deliveries (n ¼ 25,608).BMI ¼ body mass index, IBD ¼ inflammatory bowel disease, SSRI ¼ selective serotonin reuptake inhibitor, HDP ¼ hypertensive disorders of pregnancy, gHTN ¼ gestational hypertension, PEC ¼ preeclampsia with or without severe features, HELLP ¼ hemolysis, elevated liver enzymes, and low platelet count syndrome, PPH ¼ postpartum hemorrhage, EBL ¼ estimated blood loss, SGA ¼ small for gestational age, NICU ¼ neonatal intensive care unit.
Our finding that patients with a low GCT were more likely to have a vaginal delivery as compared to patients with normal glycemic screening has also been observed in several other studies [20,22,23,25].This is possibly due to higher incidence of SGA among the low GCT group.However, SGA has also been associated with adverse perinatal outcomes including increased risk of operative delivery likely related to placental insufficiency [30][31][32], so our observed association between a low GCT value and increased likelihood of vaginal delivery is likely mediated by multiple pathways.In our cohort, this reduction in cesarean delivery was also seen alongside lower incidence of significant bleeding in the low GCT group, though this variable has been inconsistently evaluated in other studies.
Our results demonstrate a significant association between low GCT value and neonatal morbidity, largely driven by SGA.It is notable that the other elements of our composite are either rare occurrences (stillbirth, neonatal demise) or neonatal sequelae known to be associated with maternal hyperglycemia (neonatal complications of hypoglycemia, hyperbilirubinemia, and respiratory morbidity) [2,33].It is possible that our study was underpowered, that we did not select the proper variables, or that we were limited by our inability to access unmeasured or unrecorded variables in the EMR to detect a greater difference in adverse perinatal outcomes based on GCT values beyond rates of SGA.Regardless, it is interesting that we observed a bimodal increase in our neonatal morbidity composite in the low GCT and GDM groups as compared to the normal glycemic screening group, which had the lowest rates of composite neonatal morbidity.Gestational diabetes and hyperglycemia have been consistently associated with adverse perinatal outcomes, and it is possible that hypoglycemia is similarly disadvantageous in contrast to euglycemia.
It is notable that alongside increased rates of neonatal morbidity and SGA in the low GCT group, several maternal outcomes were noted in the low GCT group, with lower rates of HDP, cesarean birth, and postpartum hemorrhage.As previously acknowledged, the baseline maternal characteristics between the low GCT and normal glycemic screening groups were different, with younger maternal age, lower pre-pregnancy BMI, lower rates of chronic hypertension seen in the low GCT group.It may be that the combination of these maternal factors and pregnancy factors in the low GCT group contribute to a placentally mediated pathway that is protective against HDP and other adverse perinatal outcomes, although confounding or other unmeasured differences may explain these improved maternal outcomes.Regardless, given the proposed links between hyperglycemia and placental inflammation [2,34], it is possible that maternal hypoglycemia in the third trimester may be protective against placental inflammation and therefore result in several outcome measures associated with improved placental health.
Future studies could include maternal and neonatal biospecimens such as placental histopathology and maternal serum and umbilical cord biomarkers to explore potential biological pathways linking low GCT value, maternal metabolism in the third trimester, and fetal growth and development.Additionally, studying the rate of fetal growth and identifying if there is a drop-off in the third trimester may bolster a pathophysiologic mechanism of maternal hypoglycemia in the development of fetal growth restriction, considering physiologic changes over the course of a normal pregnancy.
Our study includes several strengths.First, we had a large and contemporary urban U.S. cohort that is diverse in sociodemographic and medical characteristics.Second, our lowest 10 th percentile was <82mg/dL which is consistent with previously reported values (<70-100 mg/dL) in the literature [9,12,[17][18][19][20][21][22][23][24][25][26][27].Third, our study not only compares patients with a low GCT with normal glycemic screening, but also includes patients with an abnormally elevated GCT and GDM, which are additional populations of interest along the maternal glycemic spectrum that have not been included by any similar studies in the past.This is important as it assesses the associations of GCT value with neonatal morbidity without assuming a directionality to the findings.Finally, our point estimate for the association between a low GCT value and neonatal morbidity was almost unchanged between crude and adjusted odds ratios, indicating a robust finding.
Our study had several important limitations.As with any large dataset, we are reliant on accurate coding of diagnostic and billing codes for the precise categorization of disease states and are subject to unmeasured confounding and misclassification bias.However, within the constraints of a large retrospective database study, this dataset has been repeatedly validated and quality control measures instituted to minimize missing fields and misclassification.For this study, actual GCT and GTT laboratory values as well as timing of when the GCT or GTT were performed were meticulously examined to categorize patients into appropriate glycemic groups, with only a small percentage excluded due to illogical or missing data.Furthermore, we examined data by running variable frequencies, verified procedure and diagnostic codes, cross-checked accurate classification of variables in clinically logical contexts, met with data scientist to review any discrepancy in variables, and abstracted a random selection of charts to verify data fields.However, as with any retrospective cohort study, variables are subject to data-collection errors and not representative of all potentially important variables.For example, factors such as history of SGA newborn, interpregnancy interval, fetal anomaly, food scarcity, access to nutritional services and general healthcare, and conditions that alter glucose metabolism (i.e.history of bariatric surgery) were not validated for research use or otherwise captured for this study.While our study reflects a large and diverse urban cohort, our single institution data affects the generalizability of this study.However, as mentioned above, while our study reflects the population characteristics of one specific institution, our low GCT cutoff is consistent with previously published values, and institutions can perform an analysis of their own internal data to arrive at a cutoff for low GCT value (i.e.lowest 10 th percentile) that may identify patients at risk for fetal growth issues.
In conclusion, a low maternal GCT value was significantly associated with an increased risk for composite neonatal morbidity, largely driven by SGA.Further research is needed to investigate whether a low maternal GCT result is associated with pathologic maternal hypoglycemia or other glycemic dysregulation that can have downstream fetal effects.Pregnant patients with a low GCT value may benefit from enhanced antenatal surveillance given increased risks of neonatal morbidity.

Table 1 .
Baseline maternal demographics, pregnancy outcomes, and neonatal outcomes by glycemic group.
a 2-level independent variable comparison.b 3-level independent variable comparison.c Data available for 35753 patients.d Data available for 31403 patients.e Among cesarean deliveries (n ¼ 10,734).

Table 2 .
Risk of neonatal morbidity composite among pregnancies with low GCT compared to normal glycemic screening group, N ¼ 34,518.Final model adjusted for maternal age, parity, race and ethnicity, primary language, pre-pregnancy BMI, chronic hypertension, maternal autoimmune conditions, metformin exposure, insurance status, smoking, HDP, PPH, and mode of delivery.BMI ¼ body mass index, HDP ¼ hypertensive disorders of pregnancy, PPH ¼ postpartum hemorrhage. h