Abstract
Background:
Gestational diabetes mellitus (GDM) is associated with adverse maternal and fetal outcomes, and the oral glucose tolerance test (OGTT) is the recommended test for its diagnosis. We evaluated the role of glycated hemoglobin (HbA1c) in screening and diagnosis of GDM and its correlation with adverse pregnancy outcomes.
Subjects and Methods:
In this prospective observational study, OGTT and HbA1c were performed in 500 antenatal women between 24 and 28 weeks of gestation; the pregnant women were followed up thereafter. Repeat OGTT and HbA1c were done in women with GDM at 6 weeks postpartum.
Results:
Among the 500 women, 45 were diagnosed with GDM, for an incidence of 9%. The mean HbA1c level in women with GDM was 6.2 ± 0.6%, whereas it was 5.4 ± 0.5% in those with normoglycemia. Women with GDM had a higher incidence of pregnancy-related complications compared with normoglycemic women. An HbA1c cutoff of 5.3% had a sensitivity of 95.6% and a specificity of 51.6% for the diagnosis of GDM and would have avoided OGTT in approximately half of antenatal women, while missing 5% of the women. However, those with an abnormal HbA1c will require a confirmatory OGTT, as 50% of normoglycemic women would be misclassified as having GDM by this approach. On repeat testing postpartum, two of 45 women (4.4%) had overt diabetes mellitus, whereas five (11.1%) had impaired glucose tolerance.
Conclusions:
Although HbA1c cannot replace OGTT in the diagnosis of GDM, it can be used as a screening test, avoiding OGTT in approximately 50% of women, if a cutoff of 5.3% is used.
Introduction
G
The oral glucose tolerance test (OGTT) is the established modality of screening for GDM. 8,9 Women with GDM should be subjected to an OGTT at or after 6 weeks postpartum. 10,11 However, the routine application of OGTT is hampered by its high cost, lengthy procedure, need for fasting, and patient's noncompliance. More convenient and simple approaches are therefore sought to minimize the use of OGTT, without compromising the likelihood of diagnosing GDM. Studies have assessed the utility of fasting plasma glucose (FPG) both alone and in combination with glucose challenge test for the diagnosis of GDM; however, both had lower sensitivity than OGTT. 12,13
Glycated hemoglobin (HbA1c) is an alternative method for the screening of GDM. Compared with OGTT, HbA1c has does not require fasting, consumption of a concentrated glucose beverage, or multiple blood draws. The intraindividual coefficient of variation of HbA1c (3.6%) is lesser than both fasting (5.7%) and 2-h glucose 16.6%. 14 In addition, HbA1c has low interindividual variability (<1%) compared with FPG (4%). 15 The inter- and intraindividual variabilities of fasting glucose and 2-h glucose are higher during pregnancy compared with the nonpregnant state 16 ; however, the literature is silent regarding the variability of HbA1c during pregnancy. HbA1c has shown fair correlation with OGTT for the screening and diagnosis of GDM, albeit with lower sensitivity and specificity. 17 –23 Few studies have shown that HbA1c during pregnancy can predict the occurrence of complications like macrosomia and birth injuries. 24 –26 HbA1c has also shown to have reasonable correlation with OGTT results in the postpartum period, but as an independent test, it performed poorly compared with OGTT. 27,28 Here we evaluated the role of HbA1c in screening and diagnosis of GDM and its correlation with adverse pregnancy outcomes.
Subjects and Methods
This prospective observational study was conducted by the Departments of Obstetrics and Gynecology, Endocrinology, and Neonatology of the Postgraduate Institute of Medical Education and Research, Chandigarh, India. Antenatal women with a gestation period of less than 28 weeks were included in the study. Women with history of type 2 diabetes mellitus, GDM in the previous pregnancy, known hemoglobinopathy/hemoglobin variant, or hemoglobin level <10 g/dL at the first visit and those who were diagnosed as having GDM in the index pregnancy by OGTT before 24 weeks were excluded. Informed consent was taken from all participants who were included in the study.
Five hundred forty-seven women satisfied the inclusion criteria, 20 denied consent, and 27 delivered elsewhere, so in total 500 patients were available for final analysis. All women underwent an OGTT between 24 and 28 weeks of gestation, after an unrestricted diet for 3 days and fasting overnight (at least 8 h) before the test. After a fasting blood sample was taken, women were given a solution of 75 g of anhydrous glucose in water over a 5-min period. Blood samples were then taken 1 and 2 h after ingestion of the glucose solution.
Hemoglobin, FPG, and HbA1c were estimated from the fasting sample, and plasma glucose was estimated from the samples taken 1 and 2 h post–glucose load. HbA1c was analyzed by an automated analyzer (model D10; Bio-Rad, Hercules, CA), which uses the principle of ion-exchange high-performance liquid chromatography with a reportable range of 3.8–18.5% (according to the National Glycohemoglobin Standardization Program). Precision of the machine (percentage coefficient of variation) has been reported as 1.16% in normal individuals and 1.22% in diabetes subjects, and it is certified by the National Glycohemoglobin Standardization Program as having documented traceability to the Diabetes Control and Complications Trial reference method. Plasma glucose was estimated by the glucose oxidase method.
The study was approved by the institutional ethical committee.
Diagnosis of GDM was made according to International Association of Diabetes and Pregnancy Study Groups criteria. 8 Management of GDM was done as per standard protocol. These patients underwent repeat HbA1c testing between 32 and 36 weeks of gestation. Decision and timing of termination of pregnancy, as well as mode of delivery, were determined as per obstetric indications and were not modified due to enrollment into the study. All women enrolled in the study were followed up, and pregnancy-related outcomes were noted.
At 6 weeks postpartum, HbA1c measurement and the 75-g OGTT were repeated. Postpartum glucose tolerance status was determined using the World Health Organization criteria for nonpregnant adults. Overt diabetes was defined as FPG of ≥126 mg/dL and/or plasma glucose 2 h post–glucose load of ≥200 mg/dL; impaired glucose tolerance was defined as FPG of 100–125 mg/dL and/or plasma glucose 2 h post–glucose load of 140–199 mg/dL.
The various pregnancy-related outcomes used in our study and criteria used to define them are as follows 29 :
Bad obstetric history was history of one or more pregnancy losses after gestation of fetal viability (>24 weeks) or three or more pregnancy losses at <24 weeks of gestation (both spontaneous and induced).
Conception using assisted reproductive techniques included intrauterine insemination, in vitro fertilization, or ovulation induction.
Oligohydramnios was an amniotic fluid index of ≤5 or a maximum vertical pocket of amniotic fluid of ≤2, whereas polyhydramnios was an amniotic fluid index of ≥25 or a maximum vertical pocket of amniotic fluid of ≥8.
Intrauterine growth retardation (IUGR) was birth weight 2 SDs less than the mean for the gestational age. Macrosomia was birth weight 2 SDs more than the mean for the gestational age. The presence of IUGR or macrosomia was determined from the standard intrauterine growth chart of our institute.
Presence of congenital malformation in the fetus was anything affecting any organ system anatomically, both lethal and nonlethal.
Shoulder dystocia occurred in any delivery requiring additional maneuvers for delivery of shoulders and trunk after the delivery of fetal head.
Birth injury was any injury to the fetus caused during delivery, including injury to nerves/nerve plexuses and injury to bony parts.
Perineal injury was an injury to the maternal perineum exceeding first-degree perineal tear and deliberate episiotomy.
Neonatal resuscitation requirement was any additional measure to resuscitate the neonate other than those routinely used as appropriate for gestational age at delivery.
Neonatal hypoglycemia was any fall in blood glucose levels requiring administration of intravenous dextrose solution or admission/transfer to a special unit for additional care.
Neonatal jaundice was a serum bilirubin level more than the cutoff for phototherapy for the gestational age requiring phototherapy or exchange transfusion.
Statistics
A sample size of 42 from the positive group and 420 from the negative group is sufficient to detect a difference in area under the curve (AUC) of 0.065 (considering AUC for OGTT as 0.99) between the AUC of OGTT and HbA1c. This sample size is powered to 80% and allows an α error rate of 0.05%. With a sample size of 500, the power of our study was 85% with an α error rate of 0.05%. The data are expressed as mean ± SD values for continuous variables or as proportions for categorical variables. The baseline characteristics and outcomes were compared using the t test or χ2 test based on the type of data. Multivariate analysis was performed to rule out the effect on pregnancy-related outcomes due to factors other than GDM. Multivariate linear regression was used to analyze the effect on continuous variable pregnancy outcomes with variables as gravida, parity, bad obstetric history, conception with assisted reproductive techniques, and hypothyroidism as covariates. Similarly, multivariate logistic regression was used to analyze categorical pregnancy outcomes. The receiver operator characteristic (ROC) curve was constructed using OGTT as the gold standard and HbA1c as the variable of interest. SPSS version 22.0 software (IBM, Armonk, NY) was used for statistical analysis. A value of P < 0.05 was considered significant.
Results
Among the 500 antenatal women screened, 45 were found to have GDM by OGTT (i.e., an incidence of 9%). Women with GDM were significantly older and multiparous and had higher incidence of bad obstetric history than those with normoglycemia. In addition, they were also more likely to have hypothyroidism and a family history of diabetes mellitus. Although there was no statistically significant difference in rates of infertility in both groups, women with GDM more commonly conceived on assisted reproductive techniques. The baseline characteristics of the subjects are summarized in Table 1.
ART, assisted reproduction techniques; GDM, gestational diabetes mellitus; NS, not significant.
The mean (±SD) FPG, 1-h plasma glucose, and 2-h plasma glucose were 103.3 ± 10.3, 161.4 ± 23.8, and 143.1 ± 26.3 mg/dL, respectively, in women with GDM compared with 74.5 ± 6.8, 123.9 ± 15.9, and 111.7 ± 20.5 mg/dL, respectively, in normoglycemic women. The majority of the women diagnosed as having GDM (43/45) had abnormal FPG, whereas 1-h and 2-h plasma glucose levels were abnormal in 12 women each. The mean (±SD) HbA1c between 24 and 28 weeks was 6.2 ± 0.6% in women with GDM compared with 5.4 ± 0.5% in women with normoglycemia. The mean hemoglobin level in women with GDM was 12 ± 1.07 g/dL, whereas it was 11.6 ± 0.86 g/dL in normoglycemic women.
The ROC curve of HbA1c for the diagnosis of GDM had an AUC of 0.826 (P = 0.000), which suggests good accuracy. An HbA1c level of 5.7% had a sensitivity of 73.3% and specificity of 75.6% for the diagnosis of GDM, whereas at 5.3%, sensitivity increased to 95.6% with a reduction in specificity to 51%. If the HbA1c cutoff value was increased to 6.1%, the specificity increased to 95%, but sensitivity dropped to 46.7%. An HbA1c level of 5.7% had a negative predictive value (NPV) of 96.7% and a positive predictive value (PPV) of 21.5%. If a cutoff of 5.3% was chosen, the NPV was 99%, whereas the PPV was 16%. A higher cutoff of 6.1% would yield an NPV of 94.6% and a PPV of 47.6% (Table 2).
GDM, gestational diabetes mellitus; HbA1c, glycated hemoglobin.
Three-fourths of the women (34/45, 75.55%) diagnosed with GDM were controlled on medical nutrition therapy, whereas the rest needed insulin for achieving normoglycemia. Four were treated with premixed insulin, two with regular insulin, and three with NPH insulin, whereas two received both regular and NPH insulin. The mean HbA1c level at 32–36 weeks of gestation in women with GDM was 5.9% compared with the pretreatment value of 6.2%, indicating a reduction of 0.3% after treatment.
Women with GDM had higher incidence of pregnancy-related complications compared with normoglycemic women (Table 3). The presence of GDM was associated with adverse pregnancy outcomes even after multiple logistic regression analysis (Table 4). The HbA1c level at 24–28 weeks had good correlation with the presence of oligohydramnios (odds ratio = 2.343; P = 0.01), polyhydramnios (odds ratio = 2.665; P = 0.004), and occurrence of macrosomia (odds ratio = 2.949; P = 0.007). However, it had no correlation with the occurrence of IUGR or congenital malformations in the fetus. Only 16 women (35.6%) with GDM had spontaneous onset of labor compared with 72.3% in women with normoglycemia. The rates of induced labor (37.7% vs. 22.6%) and cesarean section (26.7% vs. 5.1%) were also higher in women with GDM. The mean period of gestation at delivery was 37 weeks 2 days in women with GDM and 38 weeks 4 days in normoglycemic women. There was significant correlation of HbA1c level with occurrence of shoulder dystocia (odds ratio = 25.165; P = 0.01) and blood loss at delivery (Pearson r value = 0.088; P = 0.05). However, there was no correlation of HbA1c level at 24–28 weeks with birth weight, perineal injury at delivery, or occurrence of instrumental or cesarean delivery.
GDM, gestational diabetes mellitus; NS, not significant.
IUGR, intrauterine growth retardation; NICU, neonatal intensive care unit; OR, odds ratio.
Neonates born to women with GDM were more likely to have neonatal hypoglycemia, neonatal jaundice, and longer duration of hospital stay (Table 3). There was significant correlation of HbA1c value at 24–28 weeks with occurrence of neonatal hypoglycemia (odds ratio = 2.47; P = 0.008), need for neonatal resuscitation (odds ratio = 2.62; P = 0.012), and duration of hospital stay (r = 0.151; P = 0.001). However, it had no correlation with Apgar score at 1 min and 5 min, neonatal jaundice, or duration of neonatal intensive care unit stay.
None of the women with GDM required diet or insulin therapy in the immediate postpartum period. The majority of women with GDM had normal glucose tolerance (84.4%) on during postpartum testing with OGTT, two women (4.44%) were diagnosed as having overt diabetes, two had impaired fasting glucose, two had impaired glucose tolerance, and one had both impaired fasting glucose and impaired glucose tolerance. The ROC curve for HbA1c for diagnosing postpartum glucose intolerance had an AUC of 0.737, which represents fair accuracy (Fig. 1). An HbA1c cutoff of 5.5% had a sensitivity of 85.7% and a specificity of 57.9% for diagnosis of postpartum glucose intolerance (with NPV of 95.9% and PPV of 26.5%), whereas an HbA1c cutoff of 5.35% had a sensitivity of 100% and specificity of 34.2% (with NPV of 100% and PPV of 21%). At an HbA1c cutoff of 6.35%, the sensitivity and specificity were 28.6% and 97.4%, respectively, and NPV and PPV were 88.6% and 63%, respectively.

Receiver operator characteristic (ROC) curve of glycated hemoglobin for the diagnosis of gestational diabetes mellitus. The diagonal line is the threshold line below which the area under the curve is 0.5. The ROC curve created by gestational HbA1c is shown by the line above (AUC = 0.826).
Discussion
The incidence of GDM in our study population was 9%. Women with GDM had a higher incidence of pregnancy-associated comorbidities and adverse maternal and fetal outcomes. They were also more likely to develop complications like IUGR, macrosomia, oligohydramnios, and polyhydramnios, although the rate of congenital malformations did not differ, compared with normoglycemic women. Women with GDM were more likely to undergo induction of labor and cesarean section. Infants born to mothers with GDM were more likely to develop neonatal hypoglycemia and jaundice requiring therapy and required longer hospital stay.
The mean HbA1c level at 24–28 weeks was 6.2% in women with GDM, whereas it was 5.4% in women with normoglycemia. In a previous study involving 507 patients, Balaji et al. 21 reported mean HbA1c values of 5.96% and 5.36% in women with GDM and normoglycemia, respectively; however, women with all trimesters were included in the study. In another recent study, Rajput et al. 22 reported mean HbA1c values of 5.73% and 5.34% in women with GDM and normoglycemia, respectively. In the same study, it was shown that an HbA1c level of 5.45% had a sensitivity of 85.7% and a specificity of 61%, whereas a cutoff of 5.95% had a sensitivity of 28% with a specificity of 97%, and the ROC curve of HbA1c for diagnosing GDM had an AUC of 0.805.
In our study, the ROC curve of HbA1c for diagnosis of GDM had an AUC of 0.826, which is very similar to that reported previously by Rajput et al. 22 Choosing an HbA1c cutoff value of 5.3% would have avoided OGTT in approximately half of the antenatal women (232/500, 46.4%) while missing approximately 5% (2/45) of women with GDM. Choosing an HbA1c cutoff value of 6.1% would have avoided OGTT in approximately 95% of subjects (475/500, 95%); however, this would have also resulted in misclassification of more than 50% of women with GDM (24/45, 53.3%) as having normoglycemia. Hence, for the purpose of screening, a lower cutoff value of HbA1c (i.e., 5.3%) would be ideal. As this cutoff had an excellent NPV of 99%, women with an HbA1c level of <5.35% can be followed up without an OGTT. However, because this cutoff also misclassified approximately 50% (223/455) of normoglycemic women as GDM, women with an HbA1c level of >5.3% should be further evaluated with an OGTT.
HbA1c at 24–28 weeks showed correlation with some adverse pregnancy outcomes like shoulder dystocia, macrosomia, neonatal hypoglycemia, need for neonatal resuscitation, and duration of hospital stay, whereas such an association could not be established with others like IUGR or congenital fetal malformations. This could be due to the small sample size and the low incidence of these events in our study cohort.
In our study, a large majority (96%) of the women diagnosed with GDM had an abnormal FPG. This is in contrast to the HAPO study, where FPG identified 51.5% of women with GDM, whereas the 1-h and 2-h plasma glucose post-OGTT identified 35.4% and 13% of women, respectively. It has also been shown previously that FPG has good sensitivity for the diagnosis of GDM at cutoffs ranging from 85 to 93 mg/dL. 30 –32 A 3-h OGTT was previously recommended for the diagnosis of GDM; however, it has previously been demonstrated that the 2-h OGTT is as good, without any adverse perinatal outcome. 33
We evaluated all women with GDM with an OGTT and HbA1c at 6 weeks postpartum, and the majority remained normoglycemic (84.5%). A study by Kim et al. 27 evaluated the usefulness of postpartum HbA1c testing and found fair agreement between HbA1c and glucose levels, similar to our study. They suggested an HbA1c cutoff of ≥5.7% with sensitivity and specificity of 65% and 68%, respectively.
As HbA1c represents mean plasma glucose in the preceding 3–4 months, it is widely believed that it is not suitable for the diagnosis of GDM because of the long time required for changes in HbA1c. Although glycation of hemoglobin occurs over the entire 120-day life span of red blood cells, 34 it has been shown that mean plasma glucose of the last 1 month contributes to 50% of the final result. 35 Therefore, HbA1c may have a role in the screening of GDM, especially if lower cutoffs than those recommended for the diagnosis of diabetes mellitus in nonpregnant adults are used. The strength of our study is that only subjects with normal hemoglobin were included, as anemia causes variation in HbA1c. The limitation of our study is the small sample size.
In conclusion, although HbA1c cannot replace OGTT in the diagnosis of GDM, it can be used as a screening test, avoiding OGTT in approximately 50% of women, if a cutoff of 5.3% is used. Our data need confirmation in a large cohort of subjects.
Footnotes
Author Disclosure Statement
No competing financial interests exist.
All authors have contributed equally.
