glucose challenge testing in pregnancy
TRANSCRIPT
Glucose challenge testing in pregnancy
Duncan R. Neilson, Jr., MD, Richard N. Bolton, MD, Robert P. Prins, MD, andCarl Mark III, MD
Portland, Oregon
This study examines results of routine glucose challenge screening in 608 third-trimester patients in aprivate practice setting. Four patient categories were defined on the basis of the results: normal, abnormalscreen with normal glucose tolerance test, borderline glucose tolerance test, and gestational diabetes.When only standard prenatal management practices were used, the fetal outcomes in the four groups
were not measurably different. The costs of these procedures resulted in an expenditure of $722.31 perdiagnosis of gestational diabetes. This seems excessive in view of the negligible impact of these studieson fetal outcome. Our impression was that routine third-trimester dietary counseling, postprandial glucose
monitoring without challenge testing, and evaluation of cases of suspected fetal macrosomia would result
in comparable outcome at significantly reduced cost. (AM J OSSTET GVNECOl 1991 ;164:1673-9.)
Key words: Glucose tolerance, pregnancy, cost-effectiveness, macrosomia
The significance of the diagnosis of gestational diabetes to the pregnant patient and her fetus has been amatter of some concern in our practice. Therefore, inaccord with the recommendations of the American Diabetes Association and the American College of Obstetricians and Gynecologists,1.' our office began routineglucose screening of private prenatal patients in 1985.
Before that time we used the established risk guidelines for glucose screening': patients with a history ofmacrosomia, obesity, intrauterine death, malformations, prior gestational diabetes, glucosuria, or a strongfamily history of diabetes.
We elected to use the standard 50 gm oral glucosechallenge in the nonfasting state with a I-hour plasmalevel detected by the glucose oxidase method. We decided to emphasize specificity, at the possible sacrificeof sensitivity, by using a screen level of 150 mg/dl. 2
Those patients with glucose levels ~ 150 mg/dl weresubjected to a 3-hour oral glucose tolerance test (GTT)after an overnight fast. Actual management of patientswas based on postprandial glucose values that weremeasured in patients with an abnormal GTT. If postprandial glucose values were >125 mg/dl, the patientwas given a printed diet sheet outlining an 1800-calorie,90 gm protein diet. She was retested I to 2 weeks later.Only two patients required insulin to keep the postprandial glucose values in the normal range.
We were interested in determining whether the results of standard glucose challenge testing and 3-houroral GTT correlated with the outcome of the pregnan-
From The Womens CliniC, PC.Presented at the Fifty-seventh Annual Meeting of the Pacific CoastObstetncal and Gynecologzcal Soczety, Sun Valley, Idaho, September9-14, 1990.Repnnt requests: Richard N. Bolton, MD, The Womens Clznic, PC.2220 Lloyd Center, Portland, OR 97232.6/6/28666
cies, specifically in relation to rates of fetal macrosomia,birth weight, birth trauma, and cesarean section rate.We also were interested in the overall cost of this testingin a private practice setting.
Material and methods
Shortly after delivery we obtained the following databy chart review from 608 sequential patients seen fromlate 1985 through early 1987: maternal age, gravidity,parity, estimated date of confinement, glucose screendate, delivery date, delivery mode (vaginal or cesareansection), birth weight, glucose screen value, classification of GTT if performed (normal, borderline, abnormal), GTT values, maternal weight at registration, glucose screen time, pregnancy or delivery complications,and Apgar scores. Data were not available in all categories for all patients, but patients were not excludedbecause of incomplete data, because this reflects therealities of clinical practice. For example, glucosescreening values were available for 578 of the 608 patients, and details of delivery were available for 591.For purposes of calculating birth weight distributionsand cesarean section rates in relation to the variousglucose screening parameters, premature births «37weeks) were excluded.
Data were entered into a personal computer spreadsheet program (LOTUS 1-2-3, VIP Technologies), andstatistical analysis was carried out with the LionheartBiometrics (Lionheart Press, Inc., Alburg, Vt.) statistical package. Z, F, and X' distributions were used whereappropriate. A value of p < 0.05 was considered significant.
Interpretation ofGTT results was based on NationalDiabetes Data Group criteria4 as follows: fasting, 105mg/dl; I hour, 190 mg/dl; 2 hour, 165 mg/dl; and 3hours, 145 mg/dl. The test was classified as abnormalif any two values were equal to or in excess of these
1674 Neilson et al.
#Observations
100 -r-------_........-----------...,
80
60
40
20
o -'--:,--r'-.,..-,....""""I--r'--:,-~--:,-~--:,-~--:,-~--:,-I
60 80 100 120 140 160 180 200Plasma Glucose
Fig. 1. Distribution of screening glucose values.
June 1991Am J Obstet Gynecol
160 170 180 190 200Screen Value (MG/DL)
%
60
50
40 -
30
20
10
0150
Glucose Tolerance Testvs. Screen Value
Abnormal GTT
I Borderline GTT .
Fig. 2. Percentages of borderline and abnormal GTTs in relation to glucose screen value.
numbers, and the test was considered borderline if anyone value was excessive.
Results
Population description. The average maternal agewas 27.9, the average gravidity 2.37, and the averageparity 0.8. The overall cesarean section rate was 26.7%.After exclusion of breech presentations and repeat cesarean sections, the rate was 18.6%.
Glucose screen results. The distribution of glucosescreen values one hour after a 50 gm challenge ap-
proximates a normal distribution (Fig. 1) with a meanof 120.7 and standard deviation of29.0. Thus the valueof 150 used as the indication for standard GTT represents 1 SD above the population mean, a value wellwithin the normal population. A value 2 SD above themean would be 179. Fifteen percent of the populationfell above the cutoff value of 150. If 135 had been usedas the cutoff, 28% of the population would have required a GTT. If a cutoff of 130 were used, 35% ofthe population would have required a GTT.
GTT results. Eighty-eight screens were reported as
Volume 16..!\: umber 6, P,nt 1
Pounds
9
8
7
6
5
4
3
2
o
Glucose challenge testing in pregnancy 1675
For gesl ages 37·43 weeks;
Birthweight vs Screen
80 100 120 140 160
Plasma Glucose
180 200
Fig. 3. Birth weight in relation to glucose screen value~.
Table I. Birth weight in relation to GTT result
Birth U'{'lKht (Ib)
Group
Whole populationNormal GTTBorderline C;TTAbnormal GTT
NS, Not significant.
A[elln
7.697.867.507 H9
I SD
1.071.16I.I.~
O.H
COlllpaw,on
ReferenceNSNSNS
2= ISO (15% of the total population). Of these, 83 glucose tolerance tests were performed with the followingresults: 48 normal, 14 borderline, and 21 abnormal.These results, as well as the frequency of abnormal andborderline GTTs, are shown in Fig. 2.
Birth weights in relation to glucose screening. Thebirth weights in relation to glucose screen values areshown in Fig. 3. Note that glucose screening values donot predict birth weight in this population. The averagebirth weight for the entire population i~ 7.69 pounds,with standard deviation of 1.07 pounds. The birthweights for the subgroups with abnormal glucosescreening parameters were not different from this av
erage (Table I).The distributions of birth weights by GTT category
are shown in Figs. 4, 5, and 6. These distributionsindicate that the birth weights from the GTT subgroups are not different from those of the population
with a normal screen. It is important to note that thegestational ages at delivery were similar for all thesubgroups.
Cesarean section rates. Excluding repeat cesareansections and cesarean sections for breech presentation,the rates of cesarean ~ection delivery for the subgroupsare shown in Table II. The cesarean section rates aresimilar in all groups; i.e., glucose screening parametersdid not predict cesarean section delivery rates in thispopulation.
Fetal macrosomia. The numbers of infants weighing2=4000 gm (8.81 pounds or 8 pounds 14 ounces) arepresented in Table III. Glucose tolerance parametersdid not predICt macrosomia in this population.
Maternal age in relation to glucose screening. TableIV shows maternal age parameters in relation to the
glucose screening results. The youngest patient with anabnormal GTT was 22 years of age. Three patients outof the 21 with abnormal GTTs were <28 years old(14o/c). Only one patient was <24, and 8 patients were<30. Therefore, if age were used as a screening indio
cation, the sensitivities shown in Table V would havebeen found.
Maternal weight and glucose screening. Table VI
1676 Neilson et al. June 1991Am J Obstet Gynecol
- NLScreen
- NLGTT
60
40
80
20
Birthweight DistributionsNumber
NL Screen vs. NL Gll100 -r--------------------...,
5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11Pounds
Fig. 4. Birth weight distribution: normal screen versus normal Grr.
- BLGTT
- NLScreenBirthweight Distributions
NL Screen vs. Borderline GTT
60
40
80
20
Number
100 ,...---------------------,
5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11Pounds
Fig. 5. Birth weight distribution: normal screen versus borderline GTT.
Table II. Cesarean section rate in relation toGTT result
Group Cesarean section rate
the others. This may have reflected dietary counseling, which was systematically done only in the thirdtrimester.
shows the maternal weight values in relation to glucosescreening parameters. Maternal weights did not predictabnormal screening results or results of GTTs. If anything, the patients with abnormal glucose tolerancegained less weight in the third trimester than did
Whole populationNormal GTTBorderline GTTAbnormal GTT
18.6%21%15%20%
Comment
It would not be expected that the minimal management used in this study would completely eliminatepathologic conditions associated with abnormal glucosemetabolism in pregnancy, and therefore the results ofthis study suggest that standard glucose challenge testing practices do not predict these conditions (fetalmacrosomia, birth trauma, cesarean section delivery)with enough sensitivity to be cost-effective in a routineprivate practice setting. Instead, dietary advice given
Volume 164J',; umber 6, Part 1
Glucose challenge testing in pregnancy 1677
- NL Screen
- AbnlGTIBirthweight Distributions NL
NumberScreen vs. Abnormal GTI
100 -r-------.;;..~....;;.,;",;;..;;;,.:,..:.:..:;::..:...=...::...:..----.....,
80
60
40
20
o ~-:;::"",,,,,,,,~:::;::==;::~~-,,,,,,,,,,~~,5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11
Pounds
Fig. 6. Birth weight distribution: normal screen versus abnormal GTT.
Table III. Fetal macrosomia (birth weight >4000 gm) in relation to GTT result
Macro;omlC mfants
Group
Screen <150 mg/dlNormal GTTBorderline GTTAbnormal GTT
Total No. ofpatients
501481421
I IExpected
No. % No.
70 14 74.18 17 7.1I 7 2.14 19 3.1
Based on rate for whole population.
Table IV. Maternal age in relation to glucosescreen result
Table V. Screening sensitivity in relation tomaternal age
Age SensitivIty(yr) (%)
",,24 95",,28 86",,30 62
One might cut costs by limiting the screening to olderpatients, but costs would still be high and accuracy questionable. If, as Marquette7 suggests, only patients ",,24years of age were tested, we would have a sensitivity of
27.530.832.930.5
Average age(~r)Group
Screen <150 mg/dlNormal GTTBorderline GTTAbnormal GTT
to patients with elevated postprandial blood glucoselevels in the third trimester might seem to eliminatethe need for challenge testing, This concept is supported by the work of Lind.' Birth trauma did not occurin this study, and there were no significant neonatalcomplications,
Weingold et al.2 state that routine glucose testing iscost-effective but do not substantiate the concept withcost per test or cost per case or cost per case of gestational diabetes diagnosed, Swinker6 reports a cost perpatient screened and a cost per case diagnosed as$10.00 and $173.00, respectively, using 130 mg/dl asthe cutoff value for 3-hour glucose tolerance testing.Marquette et al. 7 report a cost of glucose screening tobe $2.45 per patient, and GTTs to be $11.00 in a 1985study. Lavin" reports a cost of $4.75 per patientscreened and a cost of $328.96 per case of gestationaldiabetes detected. Our cost was $17.75 per screen and$59.15 per GTT, giving our patients a total cost of$10,259.50 for 578 screens, $4904.00 for 83 GTTs, anda cost per diagnosis of gestational diabetes of $722.31.This cost would be significantly higher if we had chosena lower cutoff value for the I-hour screen.
1678 Neilson et al.
Table VI. Average maternal weight versus glucose screening result
Imtlal weight Screen weight Delwery weightGroup (Ib) (Ib) (Ib)
Total gam(Ib)
June 1991Am J Obstet Gynecol
Third-trimestergam (Ib)
Normal screen 140,7 164.7Normal GTT 142.9 162.5Borderline 143.1 162.2
GTTAbnormal 142.2 159.4
GTT
171.8173.2169.4
164.5
31.130.326.2
22
7.110.87.2
5.0
95%. He reported a sensitivity of 83% with this agecriterion. If only patients >30 years were tested, wewould have a sensitivity of 62%.
Elevated maternal weight and excess weight gainwere not predictors of abnormal glucose tolerance, increased birthweight, or cesarean section rates.
Jovanovic and Peterson" report that 27.7% of theirpatients with a glucose screen > 150 mg /dl at 27 to 31weeks' gestation were delivered of infants weighing2::4000 gm. Our number was 15.2% for all patients witha glucose screen> 150 mg/dl, which was comparableto the rate of 14% in patients with a screen of < 150mg/dl.
In conclusion, the results of this study cast doubt onthe current approach to gestational diabetes. We havebeen led to believe that challenge testing will help inthe management of pregnancies and that therapiesbased on these tests are beneficial. This study does notsupport this concept. Fetal weight, morbidity, difficultdelivery, and maternal morbidity were not different inthe various groups defined by glucose challenge testing.Costs of challenge testing are substantial and probablydo not contribute to improvement of fetal or maternaloutcome.
We feel that postprandial glucose measurement atthe time of other routine blood testing in pregnancywould cost less and give us the physiologic values weactually use to manage patients. Routine third-trimesterdietary counseling and aggressive induction of laborfor postdate pregnancies or suspected fetal macrosomiamight help eliminate the pathologic conditions associated with gestational diabetes.
We do not feel, as did Hunter and Milner,'o that"gestational diabetes is a diagnosis looking for a disease"but rather that gestational diabetes is a disease lookingfor a diagnosis.
REFERENCESI. Freinkel N, Josimovich J. Conference Planning Commit
tee, American Diabetes Workshop: Conference on gestational diabetes. Diabetes Care 1980;3:499.
2. Weingold A, Golde S, Spellacy W. Diabetic pregnancy inthe '80s. ACOG Update 1984-5; 10(3): I.
3. American College of Obstetricians and Gynecologists.Management of diabetes mellitus in pregnancy. ACOGTechn Bull no 92 May 1986.
4. National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucoseintolerance. Diabetes 1979;28: 1039-53.
5. Lmd 'I. Antenatal screenmg usmg random blood glucosevalues. Diabetes 1985;34(suppl 2): 17-20.
6. Swinker M. Routine screening for gestational diabetes ina family practice center. J Fam Pract 1983; 17:611-4.
7. Marquette GP, Klein VR, Niebyl JR. Efficacy of screeningfor gestational diabetes. Am J Perinatal 1985;2:7-9.
8. Lavin JP. Screening of high-risk and general populationfor gestational diabetes: clinical application and cost analysis. Diabetes 1985;34(suppl 2):24-7.
9. Jovanovic L, Peterson C. Screening for gestational diabetes: optimum timing and criteria for retesting. Diabetes1985;34(suppI2):21-3.
10. Hunter DJS, Milner R. Gestational diabetes and birthtrauma [Letter). AMJ OBSTET GYNECOL 1985;152:918-9.
Editors' note: This manuscript was revised after thesediscussions were presented.
Discussion
DR. JOHN ENBOM, Corvallis, Oregon. This studygroup involved data on 608 obstetric patients in a 2year period given a 50 gm, I-hour, postprandial glucosechallenge and, further, a 100 gm GTT in those withvalues found to exceed 150 mg/dl on the screen.Eighty-three (15%) had an abnormal screen and werecompliant in follow-up; 48 (7.9%) studies were normal,21 (3.5%) were abnormal, and 14 (2.3%) were borderline. Gestational length appeared to relate to macrosomia, and advancing maternal age with an increasingincidence of operative delivery. No birth trauma, fetalmorbidity, or neonatal hypoglycemia was noted.
The resulting study group, 21 patients with the diagnosis of gestational diabetes out of 608 screened patients, points out the need for a larger patient population or collaborative studies to evaluate the multiplefactors related to morbidity and mortality of gestationaldiabetes. I do find it interesting that maternal weightgain was reduced in the third trimester after diagnosisand the initiation of a diabetic diet; this points out theimportance of motivation when the patient is confronted with a diagnosis that presents potential risk toher fetus.
The authors chose a screening level of 150 mg/dl astheir abnormal cutoff, to increase specificity in thescreened group. This level may have validity, but unlessa lower screening value such as 135 to 140 mg/dl isused to substantiate the specificity and sensitivity of agroup, I don't feel one can rely on the assumption thatthis screening level will not miss a substantial numberof patients with gestational diabetes. The fact that thescreening level was I SD above the mean does not answer the question of the total incidence of gestationaldiabetes in a screened population.
VolulIle 16-1NulIlber (i, Pall I
In 1964, when O'Sullivan and Mahan' first publishedtheir data on pregnancy-induced diabetes, there wasdefinite need for diagnostic criteria for risk management in prenatal care. The contribution of a standardized method of diagnosis and management for gestational diabetes has been invaluable, The continuing importance of this diagnostic area in risk managementhas been emphasized again in an updated ACOG Technical Bulletin from the American College of Obstetricians and Gynecologists," recommendations from theCenters for Disease Control in 1986 \ and the SecondInternational Workshop Conference of Gestational Diabetes Mellitus in 1985, I In all guidelines the emphasisis on diagnostic criteria and, after this diagnosis, a management scheme.
Cost in diagnosis and management are importantareas that I feel were brought out in this paper andneed to be further addressed. The cost of diagnosis inthis article was $722,37 per case of gestational diabetesand would have been higher if a lower screening valuehad been chosen. This is essentially twice the figurestated in Everett's article on screening for gestationaldiabetes.' It would have been beneficial to see the costfigures for screening levels of 135 and 140 mg/dl aswell. Everett suggested further research should be focused on decreasing the cost of screening for gestational diabetes. Wouldn't it be better to screen all patients in a more cost-effective and clinically significantmanner?
If patients who maintain a normal fasting and 2-hourpostprandial blood glucose have a low perinatal mortality rate" and postprandial glucose studies after normal dietary intake may be more physiologic, it followsthat evaluation of the parameters of treatment is a moreeffective and cost-efficient method, as suggested bySchneider and Cevert. 7 They have suggested the termpregnancy-induced glucose intolerance and use of apostprandial level of 120 mg/dl. Other authors havesuggested similar screening of postprandial blood glucose levels for diagno~is and follow-up of patients ondiet management,
The question of the later predictive value of the current gold standard in diagnosing diabetes mellitus, the100 gm GTT, has also been brought into question byNaylor,' who suggested only a 36% positive predictivevalue for this diagnosis after delivery in a highly selective group with glucose intolerance during and afterpregnancy and the need to consider higher values inthe diagnosis of gestational diabetes that may correlatewith future diabetes mellitus in a more predictivemanner,
With increasing costs and reduced relative health dollars, we must increase our efficiency in the cost of diagnosis and apply more stringent criteria for the clinicaldiagnosis of gestational diabetes.
I would like to ask Dr, Bolton the following: (I) Doyou still feel that a screening value of 150 mg Idl is themost effective cutoff value, regardless of the cost ofscreening? (2) Were all newborns screened for hypoglycemia or only those that appeared symptomatic? (3)How does your group currently screen for gestational
Glucose challenge testing in pregnancy 1679
diabetes in your office population? Would you suggestdifferent guidelines?
REFERENCES
I. O'Sullivan JB, Mahan CM. Cnteria for the oral glucosetolerance test III pregnancy. DIabetes 1964; 13:27H.
2. American College of Obstetricians and Gynecologi~t~.l\fan
agement of diabetes mellItus in pregnancy. ACOG TechBull May 19H6 no 92.
3. Centers for Disease Control. Public health guidelines forenhancing diabetes control through maternal and childhealth program~ MMWR 1986;35:201-13.
4. Summary and recommendation on the Second International Workshop Conference on Gestational Diabetes Mellitus. Diabetes 1985;34: 123-6.
5. Everett WD. Screening for gestational diabetes: an analY,I,of health benefits and costs. AmJ Prev Med 19H9;5:38-43.
6 Gabbe SG, MestmanJH, Freeman RK, Anderson (~V, Lowensohn Rl. Management and outcome of class A diabetesmellitus. A\IJ OBSTE'I GV:'>.ECOL 1977;127:465-9.
7. Schneider .1M, Cavert LB. Obstetncal management of thepregnant diabetic. In: Depp R, Eschenbach DA, SClarraJ.),eds. Gynecology and obstetrics. New York: Harper andRow, 19H6 vol 3.
8. Naylor CD. Diagnosing gestational diabetes mellitus: is thegold standard valid? Diabetes Care 1989;12:565-72.
DR. BOLTON (Closing). I am very pleased to have notbeen bombarded with technical questions from theperinatologists.
Dr. Enbom, our thesis is that the cutoff value usedfor screening, at least in the ranges we tested, does notseem to make much difference. Although some obstetricians in Portland are now using 130 or 135 mg/dl,it would cost more to do GTTs on women whose screening tests are at that level. As a result of this study, wewill no longer screen women younger than age 24,We will, however, monitor postprandial blood gl.ucosevalues on all mothers throughout pregnancy, In thisstudy the blood glucose values were checked only forbabies who were symptomatic, and we tracked thesethrough the pediatricians. The babies got along quitewell.
The problem in monitoring gestational diabetes is weare on the outside looking in. We indirectly study thefetus by checking maternal glucose, hemoglobin A,,,and fructosamine, while what we really need to knowis the level of fetal insulin, Of course, we do not havea very good handle on that because insulin, being ratheruncooperative, does not cross the placenta. A recentstudy' from Graz, Austria, report~ a large number ofamniotic fluid insulin determinations. It clearly showsthat when fetal insulin is normal there is no problem.The group of gestational diabetic patients in whom thefetus has elevated insulin concentrations is the problem,We need to find a practical, noninvasive method todiscriminate between the two groups, but for the moment, we do not have it.
REFERENCE
I. Hoffmann HM, Weiss PA, Purstner P, et al. Serum fructosamine and amniotic fluid insulin levels in patients withgestational diabetes and healthy control subjects, AMJ OBSTET GV~ECOL 1990; 162: 1174-7.