Cheng, Yvonne W. MD, MPH; Block-Kurbisch, Ingrid MD; Caughey, Aaron B. MD, PhD
Gestational diabetes mellitus (GDM) complicates approximately 2–9% of pregnancies and is associated with maternal and neonatal morbidities.1 These include higher rates of preeclampsia, operative deliveries, macrosomia, shoulder dystocia, and birth injuries.2–5 Long-term consequences of GDM have also been linked with impaired glucose tolerance and obesity in children born to diabetic mothers.6,7 Because the diagnosis and treatment of GDM can improve these undesirable outcomes,8 the American College of Obstetricians and Gynecologists has recommended screening all pregnancies for GDM.1
The diagnostic thresholds of the 3-hour oral glucose tolerance test (OGTT) for GDM were first established in 1964 by O’Sullivan and Mahan, using whole blood.9 In 1979, the National Diabetes Data Group (NDDG) proposed using plasma glucose levels of 105 mg/dL, 190 mg/dL, 165 mg/dL and 145 mg/dL for fasting, 1-hour, 2-hour, and 3-hour postglucose load as the diagnostic thresholds for GDM, defined as having two or more plasma glucose values higher than these cutoffs during the oral glucose tolerance test.10 In 1982, Carpenter and Coustan proposed the more inclusive criteria using 95 mg/dL, 180 mg/dL, 155 mg/dL and 140 mg/dL as diagnostic thresholds (Table 1).11 The Fourth International Workshop-Conference on GDM and the American Diabetes Association both have endorsed the diagnostic criteria proposed by Carpenter and Coustan for use in the interpretation of the 3-hour OGTT.12,13 However, both the NDDG and the Carpenter and Coustan diagnostic criteria have been used by practitioners, and no specific recommendations regarding GDM diagnostic criteria were provided by The Fifth International Workshop Conference on Gestational Diabetes Mellitus.14
Using the more inclusive diagnostic criteria proposed by Carpenter and Coustan, the diagnosis of GDM was increased by 50% in a large health managed care population, with varying magnitude depending on maternal age and race or ethnicity.15 However, the effect on perinatal outcome when using the Carpenter and Coustan instead of the NDDG criteria remains less clear. The Carpenter and Coustan criteria resulted in a 32% increase in the prevalence of GDM without substantial contribution to adverse outcomes in a study conducted in Spain; thus the authors concluded that such a change from NDDG to Carpenter and Coustan diagnostic criteria was not warranted in their population.16 In a recent nested case–control study, it was demonstrated that women with OGTT values above the Carpenter and Coustan cutoffs but below the NDDG thresholds may have a higher risk of macrosomia, hypoglycemia, and hyperbilirubineamia.17 The question thus remains, if perinatal outcomes do not differ between women who would qualify for GDM by only the Carpenter and Coustan criteria (but not according to the NDDG criteria) and those ruled out for GDM diagnosis by either criteria, then endorsing the Carpenter and Coustan thresholds may increase the cost of diagnosis and treatment for GDM without additional perinatal benefit. Alternatively, if this subgroup of women does experience higher risk of perinatal morbidity, then the more inclusive diagnostic criteria proposed by Carpenter and Coustan may be warranted to capture those at risk.
We designed a retrospective cohort study to estimate if women meeting diagnosis for GDM by only the Carpenter and Coustan criteria but not by the NDDG criteria and thus received routine care (represented by the overlapping area in Fig. 1) have increased risk of perinatal morbidity compared with women who did not have GDM. We hypothesize that compared with women without GDM, women who would have GDM by the Carpenter and Coustan diagnostic thresholds but not by the NDDG criteria and thus received routine care still are at risk for perinatal complications often associated with suboptimal glycemic control. If this indeed holds true, then Carpenter and Coustan criteria for the diagnosis of GDM should be used in the clinical setting to decrease undesirable perinatal complications.
MATERIALS AND METHODS
We conducted a retrospective cohort study of all pregnancies screened for GDM and delivered at the University of California, San Francisco between January 1988 and December 2001. Institutional review board approval was obtained from the Committee on Human Research at the University. Screening of GDM was most often performed between 24 and 28 weeks of gestation using the 50-g, 1-hour screening test, with subsequent 3-hour 100-g OGTT for confirmation if screened positive. In women at high risk for GDM, early screening and diagnosis during the first or early second trimester was commenced. In these high-risk patients who screened negative, a repeat 50-g, 1-hour screening test was performed between 24 and 28 weeks of gestation. For the majority of the study population (more than 98%), the threshold for obtaining an OGTT was 140 mg/dL since 1995. Before this, a small minority (1–3%) of the women were considered screened positive when a 50-g, 1-hour screening threshold of 135 mg/dL was used, and less than 1% were considered screened positive when a screening threshold of 130 mg/dL was used.
During the study period, the plasma glucose was measured by the glucose oxidase technique in the clinical laboratory at the University of California, San Francisco during the years 1988 to 1992 and 2000 to 2001. The test was switched to the glucose hexokinase technique between 1992 and 1999. However, internal controls were used to confirm consistency between the two techniques, and the equipment was calibrated three times daily for quality control. Test results of the 50-g, 1-hour screening test and the 100 g, 3-hour OGTT were abstracted from a laboratory database and linked with a perinatal database. The exclusion criteria were multifetal pregnancies, vaginal breech deliveries, delivery before 24 weeks of gestation, known lethal congenital anomalies, and pregestational diabetes mellitus.
During the study period, GDM was diagnosed using the NDDG criteria at the University of California, San Francisco (Table 1). Women who would have been given a diagnosis of GDM based on the Carpenter and Coustan (but not the NDDG) criteria received routine care and did not receive further counseling or nutrition education during the study period, because they were considered “ruled out” for GDM. Women who were diagnosed with GDM were started on glucose self-monitoring as well as nutrition and diet counseling and modification. If glycemic control was not adequately achieved by diet and exercise modifications, subcutaneous insulin therapy would be initiated. All deliveries at this academic institution were performed either by the attending physicians, clinical nurse midwives, or resident physicians with supervision. Maternal, neonatal, and labor information was prospectively documented by the delivering physician and entered into the perinatal database. The database was further maintained with daily chart review by trained abstractors.
Maternal outcomes examined included mode of delivery, third- or fourth- degree perineal lacerations, preeclampsia, and preterm delivery less than 37 weeks of gestation. Neonatal outcomes examined were 5-minute Apgar score less than 7, neonatal acidemia as measured by umbilical cord artery pH less than 7.0 and base excess less than –12, large for gestational age (LGA, defined as birth weight more than 97th centile by gestational age18), macrosomia (birth weight more than 4,500 g), shoulder dystocia (as diagnosed by the delivering attending), birth trauma (examined as a composite variable for brachial plexus injury, facial nerve palsy, clavicular fracture, skull fracture, and head laceration), neonatal hypoglycemia, jaundice, and admissions to the intensive care nursery.
Outcomes were examined using the χ2 test. Multivariable logistic regression analyses were used to control for potential confounders. Covariates included in the multivariable regression models were parity, maternal age, race or ethnicity, gestational weight gain, gestational age at delivery, year of delivery, epidural anesthesia, and induction of labor. Additionally, mode of delivery was included in models for postpartum hemorrhage, perineal lacerations, endomyometritis, shoulder dystocia, and birth trauma; chronic hypertension was controlled for in the model that examined preeclampsia. In the multivariable models, outcomes of women with GDM diagnosis by the Carpenter and Coustan (and not by the NDDG) criteria only were compared with women without GDM. For women diagnosed with GDM by only the NDDG criteria, their outcomes were compared with women with GDM by only the Carpenter and Coustan criteria (Fig. 1). Statistical significance was indicated by P<.05 and 95% confidence intervals (CIs). Statistical analysis was performed using Stata 9.0 (StataCorp LP, College Station, TX).
There were 14,693 women meeting study criteria. Of these, 753 (5.1%) would have been diagnosed with GDM by the Carpenter and Coustan criteria and 480 (3.3%) had GDM diagnosed by the NDDG criteria, yielding the difference between the two, 273 (1.9%) women, as the study group (women who would have GDM according to the Carpenter and Coustan criteria only but not by the NDDG criteria; Fig. 1). Women who were multiparous, aged 35 years or older, African-American, Latina, or Asian race/ethnicity, or with a body mass index of 30 kg/m2 and greater were more likely to be diagnosed with GDM (Table 2, P<.001 for all).
Higher rates of operative vaginal delivery and cesarean delivery were observed in women with GDM by the Carpenter and Coustan criteria only compared with non-GDM patients (Table 3). Of the women who had cesarean delivery, indications for cesarean were different among the GDM diagnostic groups. Labor dystocia was a more common indication for women diagnosed with GDM by NDDG only compared with other groups; likewise, fetal intolerance of labor was a more common indication for cesarean for women who did not have GDM compared with other groups (Table 3). Higher rates of preeclampsia and postpartum hemorrhage were also seen for this group of women (Table 3). Interestingly, women with GDM by the Carpenter and Coustan (but not the NDDG) criteria were more likely to have neonates with birth weight of more than 97th centile or weighing more than 4,500 g, whereas this effect was less pronounced for those women with GDM by the NDDG criteria only (Table 3). Neonatal hypoglycemia was more likely to occur in women with GDM only by the NDDG criteria. The rate of hyperbilirubinemia was also higher in the two subgroups of women with GDM diagnosis by the Carpenter and Coustan criteria only (14.7%) and those with GDM by the NDDG criteria only (12.5%) compared with women without GDM (10.3%, P=.02).
To reduce confounding bias, perinatal outcomes of women who would have GDM by only the Carpenter and Coustan (and not by NDDG) criteria were compared with nondiabetics (as referent) using multivariable regression analysis. Women with GDM by the Carpenter and Coustan criteria only had higher odds of operative vaginal and cesarean delivery (second column, Table 4). Although there were no differences in maternal morbidity, women with GDM by the Carpenter and Coustan criteria only had higher odds of having LGA more than 97th centile (adjusted odds ratio [aOR] 4.28, 95% CI 2.24–8.18), or macrosomic infants (aOR 4.47 95% CI 2.26–8.86) compared with those without GDM. The odds of shoulder dystocia was also higher (aOR 2.34, 95% CI 1.03–4.88), as was hyperbilirubinemia (aOR 1.48, 95% CI 1.01–2.17). However, there were no statistically significant differences in birth trauma or neonatal hypoglycemia (Table 4).
When comparing the group of women who had GDM by Carpenter and Coustan criteria only to those diagnosed with GDM by NDDG criteria, we did not observe a statistically significant difference in the adjusted odds ratio for mode of delivery or maternal morbidity (third column, Table 4). However, among women who had cesarean delivery, indications for cesarean were different (Table 4). Interestingly, compared with women who would have GDM by the Carpenter and Coustan criteria only (but not by the NDDG criteria and were considered as ruled out for GDM during the study period), women with GDM diagnosed by the NDDG criteria (and were treated for GDM) had lowered odds of having macrosomic infants more than 4,500 g (aOR 0.16, 95% CI 0.04–0.64) and shoulder dystocia (aOR 0.21, 95% CI 0.04–0.99). The adjusted odds ratio for birth trauma or metabolic abnormalities did not reach statistical significance (Table 4).
This study examined whether women who would meet diagnostic criteria for GDM according the Carpenter and Coustan thresholds but not according to the National Diabetes Data Group criteria have increased risk of perinatal morbidity. Our findings suggest that these women do have higher odds of having operative deliveries, shoulder dystocia, and LGA or macrosomia compared with women without a diagnosis of GDM. Interestingly, women who were diagnosed with GDM by the NDDG criteria and were treated for GDM had nearly one fifth the odds of macrosomia and shoulder dystocia compared with those women would have GDM diagnosed by the Carpenter and Coustan criteria only (but not according to the NDDG criteria) and were not treated for GDM during the study period. Because macrosomia, shoulder dystocia, and metabolic abnormalities such as hypoglycemia and hyperbilirubinemia are complications associated with gestational diabetes, particularly poorly controlled or untreated diabetes, we speculate that these undesirable outcomes may be caused by persistent in utero exposure to hyperglycemia.
Although the American Diabetes Association has endorsed using the Carpenter and Coustan criteria for the diagnosis of GDM since 2000, this change was not based on evidence from large studies. The most recent American College of Obstetricians and Gynecologists recommendations regarding the management of GDM support the use of either criteria, stating that “there are no data from clinical trials to determine which is superior.”1,13 Our study results are concordant with a recent nested case–control study which suggests an association between macrosomia, neonatal hypoglycemia, hyperbilirubinemia, and pregnancies with plasma glucose levels exceeding the Carpenter and Coustan thresholds but below the NDDG thresholds.17 We also found higher risk of macrosomia in women with plasma glucose levels between the two diagnostic thresholds, similar to a study in a Turkish population.19
One interesting finding of our study was that women who were diagnosed with GDM by the NDDG criteria and were thus treated for GDM during the study period were one fifth less likely to have macrosomic infants or shoulder dystocia when compared with women who would have been diagnosed with GDM using the Carpenter and Coustan criteria, but not according to the NDDG criteria, and thus received routine care and were not treated for GDM during the study period. Interestingly, although women in this group had lower odds of having macrosomia, their risk of operative delivery was not different. Rather, the odds of undergoing cesarean delivery because of labor dystocia as an indication was twofold in this group of women who carried a diagnosis of GDM, although they did not have a higher risk of macrosomia. Perhaps the diagnosis of GDM biased the provider into having a lower threshold for proceeding to cesarean delivery. This topic deserves further investigation.
If we assume the true population at risk for complications related to hyperglycemia were those identified by the Carpenter and Coustan diagnostic criteria, but only women whose plasma glucose values met diagnosis for GDM by the NDDG criteria were treated for GDM, then the women whose plasma glucose values lie between the two diagnostic criteria, our study population, mimic those with undiagnosed and untreated GDM (Fig. 1). This subgroup had higher odds of operative deliveries, macrosomia, and hyperbilirubinemia. From our analysis, we infer that if treated for GDM, the risk of these undesirable outcomes may be reduced. Thus, we speculate that the Carpenter and Coustan diagnostic criteria may be more sensitive than the NDDG criteria for the diagnosis of GDM and recommend that the Carpenter and Coustan criteria be used for the diagnosis of GDM in the clinical setting. These findings are similar to studies that demonstrate that perinatal outcomes are worse in women with one elevated value on the OGTT as compared with women with entirely normal OGTTs.20–22
Although we report that risk of perinatal morbidity is higher in women who would have GDM diagnosed by only the Carpenter and Coustan criteria but not according to the NDDG criteria, our study has limitations. As an observational study, we designated women without GDM as the reference group to examine perinatal outcomes associated with the differences in diagnostic criteria. We further designated those who would have GDM only by the Carpenter and Coustan criteria as the study group, ie, women who would have GDM by the Carpenter and Coustan criteria but were untreated during the study period. Ideally, a prospective randomized controlled trial using either the Carpenter and Coustan or the NDDG criteria for the diagnosis of GDM would best demonstrate the causal association between the different diagnostic criteria and perinatal outcomes. However, this will require large number of patients and can be time-consuming and expensive, particularly given that the prevalence of GDM ranges between 2–9% and most of the undesirable neonatal outcomes associated with GDM are rare events. For example, even with our large cohort, the statistical power was likely limited to examine rare neonatal outcomes such as umbilical artery pH less than 7.0. Additionally, we can speculate that a small number of women who were identified as the study population may be treated for GDM. However, such misclassification would bias our effect estimates toward the null, and the true effect estimates would actually be more pronounced than observed. As an observational study, validity of our study relies on accurate information reporting. Although our database was maintained with daily chart review by trained abstractors to ensure accuracy and minimize missing data, confounding may still exist. We attempted to control for potential confounders using multivariable regression analyses, but there may be residual confounding for which we did not or could not control.
Despite these limitations, our study demonstrates that women who would have GDM using the Carpenter and Coustan criteria but not according to the NDDG criteria (and thus untreated for GDM) were more likely to have operative deliveries, large for gestational age ( more than 97th centile) or macrosomic (more than 4,500-g) neonates as well as shoulder dystocia and hyperbilirubinemia. Of particular interest is that women who were diagnosed with GDM and thus treated according to the NDDG criteria had a reduced risk of macrosomia and shoulder dystocia compared with women with GDM by the Carpenter and Coustan criteria only and were not treated for GDM during the study period, suggesting that treatment of these women may likely lower their risk of complications. Thus, we speculate that the more stringent National Diabetes Data Group criteria for the diagnosis of GDM may miss a subgroup of women still at risk for complications associated with hyperglycemia. In accordance with the recommendations by the American Diabetes Association, we recommend that clinicians should use the Carpenter and Coustan diagnostic thresholds (95 mg/dL, 180 mg/dL, 155 mg/dL, and 140 mg/dL plasma glucose values for fasting, 1-hour, 2-hour, and 3-hour after 100-g OGTT) for the interpretation and diagnosis of gestational diabetes mellitus to capture women at risk and improve perinatal outcomes.
1. American College of Obstetricians and Gynecologists Committee on Practice Bulletins—Obstetrics. ACOG Practice Bulletin. Clinical management guidelines for obstetrician-gynecologists. Number 30, September 2001 (replaces Technical Bulletin Number 200, December 1994). Gestational diabetes. Obstet Gynecol 2001;98:525–38.
2. Naylor CD, Sermer M, Chen E, Sykora K. Cesarean delivery in relation to birth weight and gestational glucose tolerance: pathophysiology or practice style? Toronto Trihospital Gestational Diabetes Investigators. JAMA 1996;275:1165–70.
3. Casey BM, Lucas MJ, McIntire DD, Leveno KJ. Pregnancy outcomes in women with gestational diabetes compared with the general obstetric population. Obstet Gynecol 1997;90:869–73.
4. Acker DB, Sachs BP, Friedman EA. Risk factors for shoulder dystocia. Obstet Gynecol 1985;66:762–8.
5. Adams KM, Li H, Nelson RL, Ogburn PL Jr, Danilenko-Dixon DR. Sequelae of unrecognized gestational diabetes. Am J Obstet Gynecol 1998;178:1321–32.
6. Silverman BL, Metzger BE, Cho NH, Loeb CA. Impaired glucose tolerance in adolescent offspring of diabetic mothers. Relationship to fetal hyperinsulinism. Diabetes Care 1995;18:611–7.
7. Vohr BR, McGarvey ST, Tucker R. Effects of maternal gestational diabetes on offspring adiposity at 4–7 years of age. Diabetes Care 1999;22:1284–91.
8. Crowther CA, Hiller JE, Moss JR, McPhee AJ, Jeffries WS, Robinson JS, et al. Effect of treatment of gestational diabetes mellitus on pregnancy outcomes. N Engl J Med 2005;352:2477–86.
9. O’Sullivan JB, Mahan CM. Criteria for the oral glucose tolerance test in pregnancy. Diabetes 1964;13:278–85.
10. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. National Diabetes Data Group. Diabetes 1979;28:1039–57.
11. Carpenter MW, Coustan DR. Criteria for screening tests for gestational diabetes. Am J Obstet Gynecol 1982;144:768–73.
12. Metzger BE, Coustan DR. Summary and recommendations of the Fourth International Workshop-Conference on Gestational Diabetes Mellitus. The Organizing Committee. Diabetes Care; 21 suppl:B161–7.
13. American Diabetes Association: gestational diabetes mellitus (Position Statement). Diabetes Care 2002;25:S94–6.
14. Metzger BE, Buchanan TA, Coustan DR, De Leiva A, Dunger DB, Hadden DR, et al. Summary and recommendations of the Fifth International Workshop-Conference on Gestational Diabetes Mellitus [published erratum appears in Diabetes Care 2007;30:3154]. Diabetes Care 2007:30:S251–60.
15. Ferrara A, Hedderson MM, Quesenberry CP, Selby JV. Prevalence of gestational diabetes mellitus detected by the National Diabetes Data Group or the Carpenter and Coustan plasma glucose thresholds. Diabetes Care 2002;25:1625–30.
16. Ricart W, Lopez J, Mozas J, Pericot A, Sancho MA, Gonzalez N, et al. Potential impact of American Diabetes Association (2000) criteria for diagnosis of gestational diabetes mellitus in Spain. Diabetologia 2005;48:1135–41.
17. Ferrara A, Weiss NS, Hedderson MM, Quesenberry CP Jr, Selby JV, Ergas IJ, et al. Pregnancy plasma glucose levels exceeding the American Diabetes Association thresholds, but below the National Diabetes Data Group thresholds for gestational diabetes mellitus, are related to the risk of neonatal macrosomia, hypoglycaemia, and hyperbilirubinaemia. Diabetologia 2007;50:298–306.
18. Babson SG, Benda GI. Growth graphs for the clinical assessment of infants of varying gestational age. J Pediatr 1976;89:814–20.
19. Gokcel A, Bagis T, Killicadag EB, Tarim E, Guvener N. Comparison of the criteria for gestational diabetes mellitus by NDDG and Carpenter and Coustan, and the outcomes of pregnancy. J Endocrinol Invest 2002;25:357–61.
20. Stamilio DM, Olsen T, Ratcliffe S, Sehdev HM, Macones GA. False-positive 1-hour glucose challenge test and adverse perinatal outcomes. Obstet Gynecol 2004;103:148–56.
21. Ergin T, Lembet A, Duran H, Kuscu E, Bagis T, Saygili E, et al. Does insulin secretion in patients with one abnormal glucose tolerance test value mimic gestational diabetes mellitus? Am J Obstet Gynecol 2002;186:204–9.
22. McLaughlin GB, Cheng YW, Caughey AB. Women with one elevated 3-hour glucose tolerance test value: are they at risk for adverse perinatal outcomes? Am J Obstet Gynecol 2006; 194:e16–9.