Use of oral glucose tolerance test in early pregnancy to predict late ...

doi:10.1111/j.1447-0756.2007.00693.x J. Obstet. Gynaecol. Res. Vol. 34, No. 3: 331–336, June 2008Use of oral glucose tolerance test in early pregnancy topredict late-onset gestational diabetes mellitus inhigh-risk womenChadakarn Phaloprakarn and Siriwan TangjitgamolDepartment of Obstetrics and Gynecology, Bangkok Metropolitan Administration Medical College and Vajira Hospital,Bangkok, ThailandAbstractAim: To evaluate if any single plasma glucose level from the four values of the normal 100-g oral glucosetolerance test (OGTT) in early pregnancy (20 weeks of gestation) could predict gestational diabetes mellitus(GDM) diagnosed from a second OGTT in late pregnancy (28–32 weeks).Methods: Glucose levels of pregnant women at high-risk for GDM, who had had a normal early OGTT, andwho underwent the second test in late pregnancy, were studied. Each of the four plasma glucose values of theearly OGTT was determined for sensitivity, specificity, positive predictive value (PPV), and negative predictivevalue (NPV). The receiver operating characteristic curves of these four OGTT values were then constructed tofind the optimal value to predict late-onset GDM.Results: Of 193 pregnant women who had had a normal early OGTT, 154 also had a normal OGTT in latepregnancy while 39 had an abnormal test and were diagnosed with GDM. Among the four glucose values ofthe early OGTT, the 1-h value yielded the best diagnostic performance to predict late-onset GDM. Thesensitivity, specificity, PPV, NPV, and area under the curve achieved from its optimal cutoff level of155 mg/dL (8.6 mmol/L) were 89.7%, 64.3%, 38.9%, 96.1%, and 0.77, respectively.Conclusions: A 1-h glucose value 155 mg/dL at the early OGTT yielded the best diagnostic performance.However, the low specificity and PPV rendered it suboptimal to predict late-onset GDM. Nevertheless, aconsiderable number of high-risk women could avoid the second OGTT in late pregnancy due to its highsensitivity and NPV.Key words: gestational diabetes mellitus, high-risk women, oral glucose tolerance test.IntroductionGestational diabetes mellitus (GDM) is one of the riskfactors for maternal and neonatal morbidities. 1–4 Theoptimal screening and diagnostic approaches for GDMremain controversial since there has been no availabledata from well-designed studies. 1,4–7 The AmericanCollege of Obstetricians and Gynecologists recommendeda two-step procedure to diagnose GDM. 1 Thisincludes a 50-g glucose challenge test (GCT) as aninitial screening step, followed by a diagnostic 100-goral glucose tolerance test (OGTT) if the screening GCTis abnormal. 1From the most recent International Workshop-Conference on GDM in 1997, the prior recommendationof universal screening was changed to a selectiveReceived: May 21 2007.Accepted: September 6 2007.Reprint request to: Dr Chadakarn Phaloprakarn, Department of Obstetrics and Gynecology, Bangkok Metropolitan AdministrationMedical College and Vajira Hospital, 681 Samsen Road, Dusit District, Bangkok 10300, Thailand. Email: chadakarn_pt@yahoo.com© 2007 The Authors 331Journal compilation © 2007 Japan Society of Obstetrics and Gynecology

C. Phaloprakarn and S. Tangjitgamolapproach. 5 Based on this recommendation, pregnantwomen who are at high risk for GDM should undergoblood glucose testing as soon as possible to identifypre-existing diabetes or early-onset GDM, 5 with theaim that an appropriate treatment could be immediatelyemployed to optimize the maternal and neonataloutcomes. 8,9 In the circumstance that diabetes is notevident in early pregnancy, blood glucose testingshould be repeated between 24 and 28 weeks of gestationto detect an event of late-onset GDM. 5One previous Thai population-based study reportedthat 4.9% of high-risk pregnancies for GDM who hadhad normal OGTT in early pregnancy would subsequentlybe detected as having GDM by their secondOGTT during late pregnancy. 10 This implied that a highpercentage (approximately 95%) of women would besubjected to the unnecessary second test. Hence, anindicator from the first OGTT which could predict ahigh or low probability of late-onset GDM would beuseful in clinical practice. For example, individualswho are at increased risk for developing GDM determinedby the first OGTT (despite a normal test) wouldbe aware of the disorder, be compliant to a follow-upvisit, or be cautious with their diet and daily lifestyle;likewise, the second test might be omitted in those whoare at lesser risk.One earlier report proposed that the combined use offasting and 2-h plasma glucose values at specific cutofflevels from a normal 75-g OGTT in early pregnancycould exclude women in the high-risk group who wereunlikely to develop GDM in late pregnancy. 11 However,those values could not be utilized in a settingwhere a 100-g OGTT is used. Thus, this study wasdesigned to investigate whether any single plasmaglucose level from the four values of a normal 100-gOGTT in early pregnancy could predict the probabilityof abnormal second OGTT (GDM) during late pregnancyin high-risk women.Materials and MethodsThis study was conducted after approval from theBangkok Metropolitan Administration Ethics Committeefor research involving human subjects. Obstetriccharts of women who attended the antenatal clinic anddelivered at the Department of Obstetrics and Gynecologyin our institution were reviewed. During thestudy period (1 January 2005 to 31 March 2006), allpregnant women without pre-existing overt diabeteswere screened for GDM with a 50-g GCT due to thedepartmental policy. Those without any risk factors forGDM were screened between 24 and 28 weeks of gestation,while women in a high-risk group were tested atan initial visit. High-risk pregnancies for GDM werethose women who had age 35 years, BMI > 27 kg/m 2 , history of GDM in a previous pregnancy, familyhistory of diabetes mellitus in any of first-degree relatives,history of large neonate (4000 g), history ofadverse perinatal outcome (two or more miscarriages,congenital malformation, or stillbirth), or glucosuria.Women who had abnormal GCT results, being definedas plasma glucose levels 140 mg/dL, would bescheduled for a diagnostic 100-g OGTT.A standard OGTT procedure recommended by theAmerican Diabetes Association 12 is routinely practicedin the institution. In brief, after a 3-day intake of highcarbohydratefood and an overnight fast for 10–12 h,blood samples at fasting, and three consecutive hourlysamples after the ingestion of 100 g glucose arecollected for plasma glucose determinations. Glucosemeasurements are performed with the glucose oxidasemethod using an automated analyzer model SynchronLX20 (Beckman Coulter, Fullerton, CA, USA). TheNational Diabetes Data Group (NDDG) criteria areused as the gold standard for diagnosing GDM, whichis made when two or more abnormal glucose valuesare identified: fasting value 105 mg/dL (5.8 mmol/L); 1-h value 190 mg/dL (10.5 mmol/L); 2-h value 165 mg/dL (9.1 mmol/L); and 3-h value 145 mg/dL(8.0 mmol/L). 1Eligibility criteria were all high-risk pregnancies forGDM who had had abnormal 50-g GCT, had hadnormal 100-g OGTT in early pregnancy (20 weeks ofgestation), and underwent a second OGTT during latepregnancy (28–32 weeks) between January 2005 andMarch 2006. Women who were lost to follow up andthose whose obstetric charts had incomplete clinicaldata were excluded from the study. Maternal demographicdata, risk factors for GDM, gestational ages(GA) at which screening and diagnostic tests were performed,50-g GCT, and all four OGTT values were collected.BMI was calculated from prepregnancy weight(kg) divided by square of height (m 2 ).Statistical analysis was performed using the spsssoftware package version 11.5 (SPSS, Chicago, IL,USA). The stata 7.0 (Stata Corp., College Station, TX,USA) was additionally used to generate confidenceintervals (CI). Demographic data and glucose resultswere presented as mean with standard deviation (SD)or median with range for continuous variables, and asnumber with percentage for categorical variables. TheStudent’s t-test or the Mann–Whitney U-test were used332 © 2007 The AuthorsJournal compilation © 2007 Japan Society of Obstetrics and Gynecology

Use of oral glucose tolerance testto compare continuous variables as appropriate, andthe c 2 -square test was used to compare categorical variables.P-value of

C. Phaloprakarn and S. TangjitgamolFigure 1 The receiver operatingcharacteristic curves of the fourplasma glucose values of the oralglucose tolerance test in earlypregnancy for the prediction oflate-onset gestational diabetesmellitus. The number in each lineof the receiver operating characteristiccurves represents theoptimal cutoff level (mg/dL) ofeach glucose value of the earlyoral glucose tolerance test.Table 2 Glucose results in early pregnancy of 193 high-risk women for gestational diabetes mellitus who develop and didnot develop gestational diabetes mellitus in late pregnancyTotal women(n = 193)Women withlate-onset GDM(n = 39)Women withoutlate-onset GDM(n = 154)P-valueGA at GCT (weeks), mean (SD) 12.1 (4.6) 12.1 (4.5) 12.1 (4.6) 0.96†GCT resultPlasma level (mg/dL), median (range) 158 (140–330) 165 (141–265) 156 (140–330)

Use of oral glucose tolerance testTable 3 Optimal cutoff level of each of the four plasma glucose values of oral glucose tolerance test in early pregnancy withits diagnostic performance to predict late-onset gestational diabetes mellitusFastingglucose value85 mg/dL1-h glucosevalue155 mg/dLOptimal cutoff level2-h glucosevalue131 mg/dL3-h glucosevalue98 mg/dLSensitivity (95% CI) 66.7 (60.0, 73.3) 89.7 (85.5, 94.0) 61.5 (54.7, 68.4) 51.3 (44.2, 58.3)Specificity (95% CI) 55.8 (48.8, 62.9) 64.3 (57.5, 71.1) 70.8 (64.4, 77.2) 61.0 (54.2, 67.9)PPV (95% CI) 27.7 (21.4, 34.0) 38.9 (32.0, 45.8) 34.8 (28.1, 41.5) 25.0 (18.9, 31.1)NPV (95% CI) 86.8 (82.1, 91.6) 96.1 (93.4, 98.8) 87.9 (83.3, 92.5) 83.2 (77.9, 88.5)AUC (95% CI) 0.61 (0.52–0.71) 0.77 (0.70–0.85) 0.66 (0.56–0.76) 0.56 (0.46–0.66)AUC, area under the curve; CI, confidence interval; NPV, negative predictive value; PPV, positive predictive value.diagnostic performance to predict abnormal OGTTin late pregnancy, with its optimal cutoff level of155 mg/dL. From the statistical point of view, anoptimal glucose level in this early OGTT which yieldeda high specificity would be appropriate to identifywomen (with plasma glucose levels above the threshold)who were at greater risk for late-onset GDM. Onthe other hand, a level with a high sensitivity would bebest to exclude women (with plasma glucose levelsbelow the threshold) who were unlikely to have abnormalOGTT in late pregnancy. At the optimal cutoff levelof 155 mg/dL of 1-h glucose value, it yielded highfalse-positive rate (low specificity of only 64.3%) andpoor PPV of 38.9%. These results would render this testinappropriate for an identification of women who werelikely to develop late-onset GDM. In an attempt toreduce the number of false-positive cases (in order toincrease the specificity), we sought to determinethe women’s characteristics, glucose levels, and riskfactors for GDM in this particular group of women.However, we could not identify any specific featuresuggesting an underlying reason for a false-positiveresult. Its high rate might lie on the small number of thestudy population. Based on our observation that thetest yielded low specificity and poor PPV, we could notrecommend an omission of late OGTT for high-riskwomen with the 1-h value of early OGTT 155 mg/dL.On the contrary, the high sensitivity (88.9%) and NPV(96.1%) of this method made it suitable to be a screeningtest (for excluding women with a low probability oflate-onset GDM), as only a few GDM women would bemissed. This would also lessen the number of womenwho would have to undergo the second OGTT in latepregnancy.To date, there has been only one report which investigatedthe diagnostic performance of plasma glucosevalues from normal 75-g OGTT in early pregnancy(16 weeks of gestation) for a prediction of late-onsetGDM (24–28 and 32–34 weeks). 11 Bito et al. studied 155high-risk women for GDM and found that a fastingglucose level of

C. Phaloprakarn and S. Tangjitgamol3. Rosenberg TJ, Garbers S, Lipkind H, Chiasson MA. Maternalobesity and diabetes as risk factors for adverse pregnancyoutcomes: differences among 4 racial/ethnic groups. Am JPublic Health 2005; 95: 1545–1551.4. American Diabetes Association. Gestational diabetes mellitus.Diabetes Care 2004; 27 (Suppl. 1): S88–S90.5. Metzger BE, Coustan DR, The Organizing Committee.Summary and recommendations of the Forth InternationalWorkshop-Conference on Gestational Diabetes Mellitus.Diabetes Care 1998; 21 (Suppl. 2): B161–B167.6. Cosson E, Benchimol M, Carbillon L et al. Universal ratherthan selective screening for gestational diabetes mellitusmay improve fetal outcomes. Diabetes Metab 2006; 32: 140–146.7. Griffin ME, Coffey M, Johnson H et al. Universal vs. riskfactor-based screening for gestational diabetes mellitus: detectionrates, gestation at diagnosis and outcome. Diabet Med2000; 17: 26–32.8. Barahona MJ, Sucunza N, Garcia-Patterson A et al. Period ofgestational diabetes mellitus diagnosis and maternal and fetalmorbidity. Acta Obstet Gynecol Scand 2005; 84: 622–627.9. Bartha JL, Martinez-Del-Frezno P, Comino-Delgado R. Gestationaldiabetes mellitus diagnosed during early pregnancy.Am J Obstet Gynecol 2000; 182: 346–350.10. Boriboonhirunsarn D, Sunsaneevithayakul P, Nuchangrid M.Incidence of gestational diabetes mellitus diagnosed before 20weeks of gestation. J Med Assoc Thai 2004; 87: 1017–1021.11. Bito T, Nyari T, Kovacs L, Pal A. Oral glucose tolerance testingat gestational weeks 16 could predict or exclude subsequentgestational diabetes mellitus during the current pregnancy inhigh risk group. Eur J Obstet Gynecol Reprod Biol 2005; 121:51–55.12. American Diabetes Association. Gestational diabetes mellitus.Diabetes Care 1986; 9: 430–431.13. Fan ZT, Yang HX, Gao XL, Lintu H, Sun WJ. Pregnancyoutcome in gestational diabetes. Int J Gynecol Obstet 2006; 94:12–16.14. Hanna FW, Peters JR. Screening for gestational diabetes: past,present and future. Diabet Med 2002; 19: 351–358.15. Agarwal MM, Punnose J, Dhatt GS. Gestational diabetes:problems associated with the oral glucose tolerance test.Diabetes Res Clin Pract 2004; 63: 73–74.336 © 2007 The AuthorsJournal compilation © 2007 Japan Society of Obstetrics and Gynecology