Secondary Logo

Journal Logo

Original Research

False-Positive 1-Hour Glucose Challenge Test and Adverse Perinatal Outcomes

Stamilio, David M. MD, MSCE*†; Olsen, Tandy MD; Ratcliffe, Sarah PhD*; Sehdev, Harish M. MD; Macones, George A. MD, MSCE*†

Author Information
doi: 10.1097/01.AOG.0000109220.24211.BD
  • Free

Gestational diabetes is associated with fetal macrosomia, cesarean delivery, preeclampsia, neonatal hypoglycemia, and perinatal death. Although not proven definitively, diagnosing and treating gestational diabetes might decrease these risks.1–4 In the United States, the most widely accepted screening and diagnosing scheme for gestational diabetes is the National Diabetes Data Group stepwise algorithm, in which a 50-g 1-hour glucose challenge test (GCT) is administered universally at 24–28 gestational weeks, followed by a 100-g 3-hour glucose tolerance test (GTT) in GCT-positive patients. A patient is diagnosed with gestational diabetes if two or more of four values are elevated on the GTT. This scheme has false-negative and false-positive rates. Van Turnhout et al5 reported that the sensitivity and specificity of the GCT are 27% and 89%, respectively. Based on a gestational diabetes prevalence of 5%, they estimated the positive and negative predictive values of the GCT to be 11% and 96%, respectively. Jimenez-Moleon et al6 reported that, depending on the presence or absence of clinical risk factors, the GCT has a positive predictive value of 12–40%. Although an exact estimate of the negative predictive value for the GCT–GTT scheme is difficult to establish, it seems that a significant proportion of GTT-negative patients will develop fetal macrosomia or be identified as diabetic at a later time.7,8 Other gestational diabetes mellitus (GDM) screening methods include risk-factor methods, fasting glucose levels, and single-tier GCT schemes, the most common being the World Health Organization (WHO) protocol. Some studies have shown these methods to be either less accurate or less cost efficient than the two-tier GCT and GTT screening algorithm, but the literature contains conflicting reports.7,9,10

Because the gestational diabetes diagnostic tests are imperfect, it is not surprising that some studies have shown that certain patients with negative or borderline diagnostic test results, with or without a preceding screening test, are at risk for adverse perinatal outcome.11–21 Many clinicians using the National Diabetes Data Group algorithm consider patients with a negative or borderline diagnostic test (GTT) as an intermediate-risk population for adverse obstetric outcome. Therefore, we proposed a cohort study to investigate the association between a false-positive 1-hour glucose challenge test and the development of perinatal complications.


To explore whether or not a false-positive GCT is an independent risk factor for adverse perinatal outcomes, we performed a retrospective cohort study. The University of Pennsylvania institutional review board approved the study. We defined the “exposed” cohort as patients with a false-positive GCT and the “unexposed” cohort as patients with a normal 1-hour GCT. We compared the two cohorts for rates of adverse perinatal outcomes. The primary outcome was a composite variable of perinatal complications known to be associated with diabetes mellitus. The dichotomous composite outcome variable included the following perinatal complications: fetal macrosomia, antenatal death, shoulder dystocia, chorioamnionitis, preeclampsia, intensive care nursery admission, and postpartum endometritis. Cesarean delivery and each component of the composite variable were evaluated as secondary outcomes. To separately evaluate fetal and maternal risks, we also evaluated two secondary dichotomous composite outcome variables that are subvariables of the primary composite outcome variable: a fetal composite variable and a maternal composite variable. The fetal composite variable comprised fetal macrosomia, antenatal death, shoulder dystocia, and intensive care nursery admission. The maternal composite variable comprised chorioamnionitis, preeclampsia, and postpartum endometritis. For purposes of this cohort study, a false-positive GCT was defined as a positive 1-hour 50-g GCT (greater than or equal to 135 mg/dL) followed by a negative 100-g 3-hour GTT. For this study, we used the National Diabetes Data Group criteria, which are based on O’Sullivan’s data for the GTT cutoff glucose values for normality. Glucose values are considered to be elevated when they are greater than or equal to 100, 190, 165, and 145 mg/dL at fasting, 1-hour, 2-hours, and 3-hours postprandial, respectively. Postpartum endomyometritis was defined as fever and uterine tenderness. A fever was defined as temperature greater than 38C. Chorioamnionitis was defined as a patient having at least two of the following: fever, fetal tachycardia (baseline heart rate greater than 169 beats per minute), or uterine tenderness. The macrosomia variable was evaluated at two different levels: fetal birth weight greater than 4000 g and greater than 4500 g. Shoulder dystocia was defined as having occurred if the delivering physician recorded shoulder dystocia in the medical record. Preeclampsia was diagnosed according to the following criteria: 1) systolic blood pressure of 140 mm Hg or more or diastolic blood pressure of 90 mm Hg or more on two separate occasions at least 6 hours apart, and 2) proteinuria of 300 mg or more in 24 hours or at least 1+ to 2+ on serial qualitative examination. In this study, antenatal death was defined as any intrauterine fetal death or intrapartum fetal death after 19 gestational weeks. Intensive care nursery admission serves as a proxy variable for neonatal hypoglycemia, severe hyperbilirubinemia, and infection because these three variables are not recorded in the data set.

Patients who delivered at the University of Pennsylvania Medical Center who were entered into the center’s Triple Marker Screen perinatal database and for whom we had complete follow-up were eligible for the study. The perinatal database contains information on more than 1990 patients who received care at the University of Pennsylvania Medical Center between 1995 and 1997. We used this data set for the present study because it contains complete information on perinatal and maternal outcomes and comprises a representative sample of our complete obstetric population. All eligible patients in the database were screened for GDM with the 1-hour 50-g GCT at 24–28 gestational weeks. An abnormal GCT was followed by a 3-hour 100-g GTT to establish the diagnosis of GDM.

Multiple gestations (twins and higher order) and anomalous fetuses were excluded from the study because these pregnancies are known to be at increased risk for adverse perinatal outcome regardless of diabetes. In addition, the inclusion of multiple gestations would create potential for bias by intracluster correlation and might have increased heterogeneity in the cohorts, thus masking a true difference in the primary outcome of interest. Patients who were not screened for gestational diabetes were excluded from study. Patients with overt diabetes mellitus were also excluded because they are already known to be at increased risk for adverse perinatal outcome and because by definition they were not part of the target population.

Two investigators (DMS, HMS) developed the Triple Marker Screen perinatal database by abstracting data from all triple screen test reports and medical records from the years 1995 to 1997. The two investigators collected the data with a standardized data collection form, entered the data into the database, and then validated the database by evaluating the variables for internal consistency. Demographic, antenatal, intrapartum, and neonatal data were abstracted by the review of maternal and newborn medical records in pairs. We recorded variables of interest that included, but were not limited to, those that might confound the association between a false-positive GCT and the primary composite perinatal outcome variable. Examples of variables are maternal chronic disease (hypertension, sickle cell trait or disease, renal disease, cardiac disease, thyroid disease), parity, maternal age, race, maternal weight, abnormal serum triple screen, maternal drug or tobacco use, gestational age at delivery, preterm premature rupture of membranes (PROM), antenatal hemorrhage, and obstetric history (previous preterm PROM, preterm delivery, preeclampsia, and delivery mode).

The data analysis was completed in three major phases. First, we evaluated the outcome variables, demographic variables and other potential confounding variables in an unadjusted analysis with χ2 or Fisher exact tests for categoric data and unpaired t tests to compare means of continuous variables. Second, we performed a stratified analysis to identify effect modification and to evaluate potential confounding variables of the association between a false-positive GCT and adverse perinatal outcome. Third, we used logistic regression modeling to evaluate the independent effect of having a false-positive GCT on the development of adverse perinatal outcomes, adjusting for effects of all other identified significant dependent variables. Variables with the strongest association (P < .15) with adverse perinatal outcome in the unadjusted analysis and stratified analysis, and those with historic or biologic importance were eligible for inclusion in the initial logistic regression models. Nonsignificant variables were removed from the regression models in a stepwise fashion testing for significance sequentially with the likelihood ratio test. The final explanatory regression model for each outcome variable includes the false-positive GCT variable and all potential confounding variables with significant associations. Logistic regression analysis uses the odds ratio (OR) as an estimator of relative risk, but the unadjusted effect sizes are reported appropriately for a cohort study as relative risk. We tested each logistic regression model for goodness-of-fit with the Hosmer and Lemeshow22 or Pearson goodness-of-fit summary statistics and regression diagnostics. Regression diagnostics consisted of multiple graphs of δ-deviance, δ-χ2 and Pregibon β versus probability of a covariate pattern or leverage. Regression models with poor fit were reanalyzed, and variables were rescaled as indicated by the regression diagnostics. Readers interested in a more detailed description of multivariable analysis might refer to Hosmer and Lemeshow.22

We estimated that approximately 1850 patients in the data set would meet inclusion criteria for the study. In the general obstetric population, the incidence of a positive GCT is approximately 15%, and of those patients with a positive GCT, approximately one third (5%) will actually be diagnosed with gestational diabetes. Therefore, the ratio of unexposed (negative GCT) to exposed (false-positive GCT) patients was projected to be approximately 9:1. The following adverse pregnancy outcomes are listed with their respective incidences in parentheses, as reported for general obstetric populations: macrosomia (10%), shoulder dystocia (1–2%), fetal demise after 16 weeks (1%), preeclampsia (5–10%), chorioamnionitis (1–5% at term; 25% preterm), and endomyometritis (3% with vaginal delivery; 5–20% with cesarean delivery).19 We anticipated that the incidence of a composite outcome variable that includes these variables would not be the sum of the individual incidences because many patients would have multiple outcomes simultaneously. We estimated that with an anticipated 20% baseline rate of any perinatal complication (the composite perinatal outcome variable), and assuming a 10% rate of a false-positive GCT and α = .05, this cohort study has 80% power to detect a relative risk (OR) of 1.5 for perinatal complications. We performed a sensitivity analysis for the sample size calculation, varying baseline outcome variable rate, exposed/unexposed ratio, and the desired effect size (relative risk). By varying the baseline adverse perinatal outcome rate between 5% and 20%, we estimated that the power of the study remains greater than 80% for detecting a relative risk of 2.0 until the baseline risk approaches 6%. When varying the exposed/unexposed ratio between 8:1 and 12:1, the study persistently has greater than 80% power to detect a relative risk of 2.0.


Among the 1998 patients, we identified 1825 patients eligible for the cohort study. Of the 173 patients that were not eligible, 112 did not undergo the GCT, 11 had pregestational diabetes mellitus, 37 had diet-controlled gestational diabetes, and 13 had insulin-dependent gestational diabetes. Among the 1825 eligible patients, 164 had a false-positive GCT (prevalence 8.8%).

Table 1 displays significant results of the unadjusted analysis of demographic and outcome variables. On average, the patients in the false-positive GCT study group were older, of higher parity, more often of nonblack, nonwhite race, and more frequently had chronic hypertension, the sickle cell trait, a high body mass index, and an elevated midtrimester human chorionic gonadotropin (hCG) serum level. The unadjusted analysis of perinatal outcomes revealed that the false-positive GCT cohort had a higher mean birth weight and higher cesarean delivery and shoulder dystocia rates. The study groups were similar with regard to midtrimester serum β-fetoprotein (AFP) and unconjugated estriol levels, prior cesarean delivery, prior preeclampsia, tobacco use, cocaine use, and maternal chronic diseases, such as renal, pulmonary, cardiac, thyroid, and autoimmune (lupus and sarcoidosis) disorders. The cohorts had similar rates of low 5-minute Apgar score (less than 7), meconium, preterm delivery at less than 37, less than 34, and less than 28 weeks, severe preeclampsia, PROM, and umbilical cord pH and base excess.

Table 1
Table 1:
Unadjusted Analysis for False-Positive GCT Cohort

Tables 2 and 3 display the results of the logistic regression analysis. After controlling for all identified confounding variables and effect modification, we determined that a false-positive GCT was a significant independent risk factor for adverse perinatal outcome (the composite perinatal outcome variable). In addition, the false-positive GCT cohort was at increased risk for the maternal adverse outcome composite variable comprising preeclampsia, chorioamnionitis, and endometritis, as well as component adverse outcome variables, including macrosomia, shoulder dystocia, antenatal death, endometritis, and cesarean delivery. The results presented in Tables 2 and 3 represent the final iteration of each of the 12 logistic regression models after the models were tested for goodness-of-fit and revised to achieve well-fit models. Most revisions were accomplished by changing the scale of the suspect variables. All final regression models were proved to be well fit on both summary statistics and regression diagnostics graphs. In addition, an in-depth investigation of covariate patterns for the primary outcome regression model allowed us to identify four covariate patterns with moderately poor fit and high leverage (h > .05). Leverage, denoted by h, is an estimate of a covariate pattern’s deviance from the mean of the data and therefore the magnitude of its influence on the regression model’s estimated parameters. Performing the regression analysis after removing these covariate patterns did not appreciably change the magnitude or direction of the effect sizes (ORs) within the model.

Table 2
Table 2:
Logistic Regression Model for the Primary Composite Perinatal Outcome Variable
Table 3
Table 3:
Logistic Regression Analysis Results for a False-Positive GCT as an Independent Risk Factor for Various Perinatal Outcome Variables

In the stratified and logistic regression analyses, we investigated an extensive number of permutations of variable pairs for effect modification that could significantly alter the association between a false-positive GCT and adverse perinatal outcomes. In the regression models for the primary composite perinatal outcome variable, the maternal composite outcome variable, endometritis, and cesarean delivery, we identified body mass index as having significant effect modification on the false-positive GCT variable with regard to the perinatal outcomes. As body mass index increased, the effect size of a false-positive GCT decreased for the composite adverse perinatal outcome variable (Figure 1). Similarly, elevated midtrimester serum hCG level, elevated midtrimester serum AFP level, and nulliparity were effect modifiers of the false-positive GCT variable for the outcomes of macrosomia, endometritis, and preeclampsia, respectively.

Figure 1
Figure 1:
Figure 1.Stamilio. Glucose Challenge Test and Perinatal Outcomes. Obstet Gynecol 2004.


Patients with pregestational and GDM clearly are at increased risk for adverse obstetric outcome. However, the most commonly used diagnostic testing schemes for gestational diabetes are flawed, with relatively poor negative and positive predictive values. The variability of common clinical practice reflects the inaccuracy of gestational diabetes screening that is reported in the medical literature. With little available outcome research, many obstetric providers treat patients with an abnormal 1-hour GCT and negative 3-hour GTT with more intensive observation or therapy, identifying these patients as “glucose intolerant” or “borderline diabetic.” Still, others maintain that such patients do not warrant additional therapies because their test results do not meet the diagnostic criteria for gestational diabetes. If having a false-positive GCT is identified as an independent risk factor for perinatal complications, patients with a false-positive GCT could benefit from additional therapies, such as more intensive fetal monitoring, nutritional counseling, or a diabetic diet. With this in mind, we developed a retrospective cohort study to determine whether patients with a positive 1-hour GCT and a negative 3-hour GTT, namely a false-positive GCT, are at increased risk for adverse perinatal outcome.

The results of our cohort study suggest that having a false-positive GCT is an independent risk factor for adverse perinatal outcome, including the composite perinatal outcome variable, the composite maternal outcome variable, endometritis, shoulder dystocia, fetal macrosomia, cesarean delivery, and antenatal death. The associations between the exposure variable (false-positive GCT) and antenatal death, fetal macrosomia, and cesarean delivery are of borderline statistical significance (P = .09, .06, and 0.09, respectively). It should be noted that the present study is underpowered to assess some of the component outcome variables, such as antenatal death, fetal macrosomia, shoulder dystocia, chorioamnionitis, and preeclampsia, all of which had a prevalence of 6% or less. Therefore, the reader should use caution in using this study to draw definitive clinical conclusions regarding the association between a false-positive GCT and less prevalent component outcome variables. As one might expect, we identified multiple significant confounding variables and effect modifiers of the association between perinatal outcome and the exposure, false-positive GCT. The ten confounding variables listed in the legend of Table 3 have been associated with perinatal outcome in previous research. Not all of the confounding variables were significant enough to include in all regression models for each of the ten secondary outcome variables. For example, only the body mass index, parity, chronic hypertension, and delivery gestational age variables were included in the logistic regression model for the maternal composite outcome variable.

Although it is not surprising that we found an interaction between maternal body mass index and the false-positive GCT variable in relation to perinatal outcome, the relationship that we unveiled among these three variables is counterintuitive (Figure 1). We had anticipated that the risk of adverse perinatal outcome attributed to a false-positive GCT would increase as body mass index increased. This conjecture was based on the assumption that obese patients with a false-positive GCT would incur additive risk from the two etiologically similar risk factors. Surprisingly, we identified an inverse relationship between body mass index and perinatal risk imparted by a false-positive GCT. In other words, as body mass index increased, the risk of adverse perinatal outcome from a false-positive GCT diminished linearly and actually became statistically insignificant when the body mass index was greater than 25 kg/m2. We can only speculate as to a reason for this type of interaction. Perhaps a nonobese patient with a false-positive GCT has another overriding susceptibility for adverse perinatal outcome and gestational diabetes that is unrelated to weight. Alternatively, the perinatal risk attributed to obesity in our study population might override and mask the risk of a false-positive GCT to the extent that we could not appreciate statistically the risk of a false-positive GCT at higher maternal weights. Regardless of the basis of this seemingly paradoxic interaction, the clinical utility of this finding is that normal-weight patients with a false-positive GCT might incur more perinatal risk from the diabetic or near-diabetic state than do their obese counterparts. In addition, the perinatal risk attributable to obesity might be more important to attempt to modify than the risk attributable to a diabetic or near-diabetic state.

The current literature is replete with research on gestational diabetes screening and obstetric outcome. Unfortunately, the body of literature is difficult to interpret because diabetes testing schemes, study populations, study methods, and results vary among studies. We chose to explore the current hypothesis to provide outcome data that is more generalizable to the North American obstetric population and to hopefully address conflicting results that exist in the current literature. Some research corroborates our findings, but other studies refute an association between an abnormal GCT and adverse obstetric outcome. Rey et al11 reported that patients with an abnormal GCT and a single elevated value on the GTT are at increased risk for fetal macrosomia, neonatal hypoglycemia, and neonatal hyperbilirubinemia. Using a case–control study design, Okun et al12 showed that patients with an abnormal GCT and no elevated values on the GTT are at increased risk for fetal macrosomia. However, Verma et al13 found no association between elevated glucose levels on GCT, GTT, fasting glucose test, or 2-hour postprandial test and fetal macrosomia in patients with a positive GCT and negative GTT. Similar to the above studies that investigated the National Diabetes Data Group screening algorithm, previous authors have shown that nondiabetic “glucose intolerant” patients identified by the WHO diagnostic criteria are at increased risk for shoulder dystocia, cesarean delivery, fetal macrosomia, and preeclampsia.14–20 However, Ramtoola et al,24 using the WHO diagnostic criteria, did not find an increase in perinatal mortality in nondiabetic glucose intolerant patients. Khan et al21 investigated a non–National Diabetes Data Group two-tier scheme to diagnose gestational diabetes that used a 75-g 2-hour glucose screening test and a 75-g 3-hour GTT. The investigators reported that patients with a positive glucose screening test and a negative GTT were at increased risk for fetal macrosomia, cesarean delivery, and preeclampsia.

We would like to make the limitations of our study explicit. First, as with all retrospective studies, we might not have had information on all relevant confounding variables or risk factors because we could only evaluate those recorded in the data set. This data set is comprehensive with respect to clinically important obstetric variables, except for some neonatal outcomes, such as hyperbilirubinemia and hypoglycemia. Fortunately, we were able to evaluate the more important severe neonatal outcomes with a proxy variable, intensive care nursery admission. Second, there is potential for confounding of the association between a false-positive GCT and adverse perinatal outcome. However, we believe that the potential for confounding within this study is very low because we performed a well-thought-out, thorough, and systematic analysis that assessed for confounding and effect modification. Third, systematic error, or bias, could cause us to draw erroneous conclusions regarding the association between a false-positive GCT and adverse perinatal outcome. We minimized the potential for selection bias by using predetermined, strict definitions for the exposed and unexposed cohorts, as well as strict inclusion and exclusion criteria. Misclassification bias might also occur. For example, the diagnosis of endometritis might have been sought more aggressively in obese patients or patients with chronic disease. We minimized the possibility of systematic misclassification bias by applying the same strict definitions of each clinical outcome to the study and control cohorts. In addition, misclassification bias might have been introduced by our definition of an abnormal GTT. If we had used the Carpenter-Coustan25 criteria for an abnormal GTT (95, 180, 155, and 140 mg/dL) instead of the National Diabetes Data Group criteria, the number of false-positive GCT patients would likely have been decreased. Consequently, the adverse outcomes in the false-positive GCT group might have been decreased, potentially abolishing the association between the exposure and adverse outcomes. This possibility is important because it might make this study less generalizable, in that clinicians who use the Carpenter-Coustan criteria might not be able to apply these findings to their own clinical population. However for the large number of clinicians who do use the National Diabetes Data Group criteria in practice, this potential for bias does not discredit or discount the study results. Abstractor bias can occur if the abstractor systematically collects data differently for the two cohorts. This type of bias should be minimized with standardized, detailed data collection forms, abstractor training, and quality assurance chart review in a small sample of charts, all of which were completed when the initial data set was developed. In addition, abstractor bias should be minimized by the fact that the abstractors were not aware of the current hypothesis, outcome variables of interest, or inclusion criteria. Another type of information bias can occur when systematic errors are made in data collection and entry into the data set. We decreased the likelihood of information bias by assessing internal consistency among the variables. After this assessment, identified data errors were corrected. Lastly, variability within the components of the composite outcome variable might mask a true association between a false-positive GCT and adverse perinatal outcome. This is particularly a problem when composite outcome variable components are affected by the exposure in opposite directions. This phenomenon is unlikely in this case because prior studies have shown that the risk of each of the outcome variable components is increased by either diabetes or glucose intolerance. In addition, we investigated variability within the composite variable by analyzing each composite outcome component variable as secondary outcome variables.

In conclusion, the results of our cohort study indicate that, compared with patients with a negative GCT, patients with a false-positive GCT are at increased risk for adverse perinatal outcome, including overall perinatal adversity, endometritis, shoulder dystocia, fetal macrosomia, cesarean delivery, and antenatal death. Until a more reliable diagnostic test for gestational diabetes is developed, patients with a false-positive GCT might benefit from more intensive prenatal care, such as nutritional counseling, a specialized diet, or antenatal fetal assessment. Before changes in clinical practice can be recommended, future research should include cost–benefit analysis of providing additional therapy to this high-risk obstetric population, investigation of whether perinatal outcome is improved with more intensive prenatal therapy, development of a clinical prediction rule for adverse perinatal outcome to help target therapy, and development of an improved diagnostic test for gestational diabetes.


1.Brody SC, Harris R, Lohr K. Screening for gestational diabetes: A summary of the evidence for the U.S. Preventive Services Task Force. Obstet Gynecol 2003;101:380–92.
2.Sameshima H, Kamitomo M, Kajiya S, Kai M, Furukawa S, Ikenoue S. Early glycemic control reduces large-for-gestational-age infants in 250 Japanese gestational diabetes pregnancies. Am J Perinatol 2000;17:371–6.
3.Koukkou E, Taub N, Jackson P, Metcalfe G, Cameron M, Lowy C. Difference in prevalence of gestational diabetes and perinatal outcome in an inner city multiethnic London population. Eur J Obstet Gynecol Reprod Biol 1995;59:153–7.
4.Moses RG, Griffiths RD. Can a diagnosis of gestational diabetes be an advantage to the outcome of pregnancy? J Soc Gynecol Invest 1995;2:523–5.
5.van Turnhout HEW, Lotgering FK, Wallenburg HCS. Poor sensitivity of the fifty-gram one-hour glucose screening test for hyperglycemia. Eur J Obstetr Gynecol Reprod Biol 1994;53:7–10.
6.Jimenez-Moleon JJ, Bueno-Cavanillas A, Luna-del-Castillo JD, Lardelli-Claret P, Garcia-Martin M, Galvez-Vargas R. Predictive value of a screen for gestational diabetes mellitus: Influence of associated risk factors. Acta Obstet Gynecol Scand 2000;79:991–8.
7.Deerochanawong C, Putiyanun C, Wongsuryrat M, Serirat S, Jinayon P. Comparison of the National Diabetes Data Group and World Health Organization criteria for detecting gestational diabetes mellitus. Diabetologia 1996;39:1070–3.
8.Bukulmez O, Durukan T. Postpartum oral glucose tolerance tests in mothers of macrosomic infants: Inadequacy of current antenatal test criteria in detecting prediabetic state. Eur J Obstet Gynecol Reprod Biol 1999;86:29–34.
9.Lavin JP Jr, Lavin B, O’Donnell N. A comparison of costs associated with screening for gestational diabetes with two-tiered and one-tiered testing protocols. Am J Obstet Gynecol 2001;184:363–7.
10.Shamsuddin K, Mahdy ZA, Siti Rafiaah I, Jamil MA, Rahimah MD. Risk factor screening for abnormal glucose tolerance in pregnancy. Int J Gynaecol Obstet 2001;75:27–32.
11.Rey E, Monier D, Lemonnier MC. Carbohydrate intolerance in pregnancy: Incidence and neonatal outcomes. Clin Invest Med 1996;19:406–15.
12.Okun N, Verma A, Mitchell BF, Flowerdew G. Relative importance of maternal constitutional factors and glucose intolerance of pregnancy in the development of newborn macrosomia. J Matern Fetal Med 1997;6:285–90.
13.Verma A, Mitchell BF, Demianczuk N, Flowerdew G, Okun N. Relationship between plasma glucose levels in glucose-intolerant women and newborn macrosomia. J Matern Fetal Med 1997;6:187–93.
14.Moses RG, Griffiths RD. Can a diagnosis of gestational diabetes be an advantage to the outcome of pregnancy? J Soc Gynecol Invest 1995;2:523–5.
15.Tan YY, Yeo GS. Impaired glucose tolerance in pregnancy\Mis it of consequence? Austr N Z J Obstet Gynaecol 1996;36:248–55.
16.Forsbach G, Cantu-Diaz C, Vazquez-Lara J, Villanueva-Cuellar MA, Alvarez y Garcia C, Rodriguez-Ramirez E. Gestational diabetes mellitus and glucose intolerance in a Mexican population. Int J Gynaecol Obstet 1997;59:229–32.
17.Aparicio NJ, Joao MA, Cortelezzi M, Guz M, Sturgeon C, Galimberti DM, et al. Pregnant women with impaired tolerance to an oral glucose load in the afternoon: evidence suggesting that they behave metabolically as patients with gestational diabetes. Am J Obstet Gynecol 1998;178:1059–66.
18.Roach VJ, Hin LY, Tam WH, Ng KB, Rogers MS. The incidence of pregnancy-induced hypertension among patients with carbohydrate intolerance. Hypertens Pregnancy 2000;19:18–9.
19.Jensen DM, Damm P, Sorensen B, Molsted-Pedersen L, Westergaard JG, Klebe J, et al. Clinical impact of mild carbohydrate intolerance in pregnancy: A study of 2904 nondiabetic Danish women with risk factors for gestational diabetes mellitus. Am J Obstet Gynecol 2001;185:413–9.
20.Aberg A, Rydhstroem H, Frid A. Impaired glucose tolerance associated with adverse pregnancy outcome: A population-based study in southern Sweden. Am J Obstet Gynecol 2001;184:77–83.
21.Kahn K, Hashmi FA, Rizvi JH. Are non-diabetic women with abnormal glucose screening test at increased risk of pre-eclampsia, macrosomia and cesarean birth? J Pak Med Assoc 1995;45:176–9.
22.Hosmer DW, Lemeshow S. Assessing the fit of the model. In: Hosmer DW, Lemeshow S, eds. Applied logistic regression. New York: John Wiley & Sons, 1989:135–73.
23.Gabbe SG, Niebyl JR, Simpson JL, eds. Obstetrics\Mnormal & problem pregnancies, 3rd ed. New York: Churchill Livingston, 1996:490–4, 561, 717, 936, 963, 1201, 1203–4.
    24.Ramtoola S, Home P, Damry H, Husnoo A, Ah-Kion S. Gestational impaired glucose intolerance does not increase perinatal mortality in a developing country: Cohort study. BMJ 2001;322:1025–6.
    25.Carpenter MW, Coustan DR. Criteria for screening tests of gestational diabetes. Am J Obstet Gynecol 1982;144:768–73.
    © 2004 The American College of Obstetricians and Gynecologists