Gestational diabetes mellitus: Antenatal variables as predictors of postpartum glucose intolerance

Gestational Diabetes Mellitus: Antenatal Variables as Predictors of Postpartum Glucose Intolerance LAUXlE R. GREENBERG, MD, THOMAS HONORE MURPHY, RN, CNM, CDE

R. MOORE, MD, AND

Objective: To determine whether antepartum variables can predict postpartum glucose intolerance. Methods: Glucose tolerance was assessed 6 weeks postpartum in 94 of 238 women with gestational diabetes using a Z-hour, 75-g oral glucose tolerance test (GTT). Selected antepartum variables were analyzed for predictive ability for postpartum glucose intolerance. Results: Of 238 patients, 94 (39%) returned for a GTT. Those returning and those not returning were similar in all variables. Postpartum glucose intolerance occurred in 34%: impaired glucose tolerance in 18%, overt diabetes in 16%. No single maternal, intrapartum, or neonatal variable was predictive of postpartum glucose intolerance in all cases. Predictive variables included: requirement for insulin (insulin versus diet: 25 versus 3% impaired glucose tolerance, 26 versus 0% diabetes; P = .OOl), poor glycemic control (any Z-hour postprandial blood sugar level of 150 mg/dL or higher: 34 versus 5% diabetes; P = .005), and the 50-g GTT value (200 mg/dL or higher: 32 versus 6% diabetes; P = .Ol). For insulin requirement, the relative risk (RR) was 17.28 (95% confidence interval ICI] 2.46-121.451, and for the above three variables combined, the RR was 19.68 (95% CI 2.88134.42). When the insulin dose was at least 100 U/day, all patients had abnormal glucose tolerance postpartum (RR = 34.00, 95% CI 4.93-234.39). Conclusions: Postpartum glucose screening is not warranted for women at low risk who do not require insulin during pregnancy. The incidence of postpartum glucose intolerance in this group is very low. Women with risk factors should receive postpartum screening. Patients receiving at least 100 U/day of insulin have a 100% incidence of postpartum glucose intolerance. (Obstet GynecoZ1995;8697101)

VOL.86,NO.l,JULY

1995

Gestational diabetes mellitus affects 3-8% of pregnant women,’ with up to half of these developing overt diabetes later in life.2 However, most women with gestational diabetes become euglycemic within the first few days postpartum”z4 and do not develop type II diabetes until many years later. Up to 19% of women with gestational diabetes will continue to have abnormal glucose tolerance postpartum, either falling outside the normal range or into the categories of impaired glucose tolerance or overt diabetes mellitus.5,6 Although prior studies reported that patients with abnormal postpartum glucose tolerance had an earlier mean age at diagnosis of gestational diabetes and a higher fasting glucose level than those with normal glucose tolerance,“‘h neither of these factors had adequate predictive power to make postpartum testing unnecessary. The purpose of this study was 1) to identify statistically significant associations between pregnancy characteristics and the occurrence of postpartum glucose intolerance, and 2) to quantitate the relative risk (RR) associated with selected antepartum variables in predicting persistent postpartum glucose intolerance.

Materials and Methods A retrospective chart review was performed, including all patients with gestational diabetes mellitus cared for and delivered at the University of California at San Diego Medical Center from January 1,1987, to April 30, 1992. Screening for gestational diabetes mellitus was performed using a l-hour, 50-g oral glucose challenge at 24-28 weeks‘ gestation. Patients with risk factors for gestational diabetes mellitus (prior gestational diabetes, prior macrosomia, obesity, family history of diabetes, prior stillbirth, and persistent glycosuria) had a screening challenge at the initial prenatal visit and, if negative, again at 24-28 weeks’ gestation. Patients with a l-hour

0029-7844/95/$9.50 SSDI 0029-7844(95)00103-X

97

screen of 140 mg/dL or more were evaluated further with a 3-hour, 100-g oral glucose tolerance test (GTT). Gestational diabetes was diagnosed by the presence of at least two abnormal venous plasma glucose levels: a fasting level of 105 mg/dL or more, a l-hour level of 190 mg/dL or more, a 2-hour level of 165 mg/dL or more, and a 3-hour level of 145 mg/dL or more.‘,x Patients with a l-hour screen of 200 mg/dL or more were given a diagnosis of gestational diabetes without further testing.’ During pregnancy, all patients practiced selfmonitoring of blood glucose levels with reflectance meters; capillary blood glucose levels were determined during fasting and 2 hours postprandially.‘“,” Dietary management followed guidelines determined by the American Diabetes Association.” Insulin therapy was initiated when a fasting glucose level exceeded 100 mg/dL, or when 2-hour postprandial values exceeded 120 mg/dL on more than two occasions.‘” A 2-hour, 75-g oral GTT was scheduled for 6 weeks postpartum for all gestational diabetes patients. The National Diabetes Data Group definitions were used for classification of results.‘” We analyzed 51 antepartum variables for their ability to predict persistent postpartum glucose intolerance. The statistical significance of differences between groups was determined using 2 test, two-tailed Fisher exact test, or Student t test, with P < .05 accepted as significant. We used analysis of variance for multiple comparisons, and P < ,002 was accepted as significant. We used regression analysis for continuous data. The RR for the development of postpartum glucose intolerance was determined for the predictive study variables, with calculation of 95Yc confidence intervals (Cl).‘”

Results We reviewed the charts of 238 mother-infant pairs; of 238 subjects with gestational diabetes, only 94 (39%)) completed postpartum GTTs, despite vigorous attempts at follow-up. A single GTT was performed on each individual at (mean 2 standard deviation) 43.6 +- 9.45 days postpartum. Reasons for failed testing included failure to return for the 6-week postpartum visit, failure to fast, and, infrequently, provider oversight in ordering the test. Frequently, tests could not be rescheduled because a sizable portion of these patients had “pregnancy-only” medical insurance that terminated in the postpartum period. Characteristics of the study population are shown in Table 1. With the exception of the postpartum GTT, antepartum variables were equally available for the returning and nonreturning subjects. No statistically significant differences were found between those who

98

Greenberg

et al

Gestatiorml

Dinhfcs

Table

1. Characteristics

of the Study

Variable

Age (y) Multipams (%) Pregravid weight (lb) BMI l-hour GTT (mg/dL) Fasting glucose level at 3-hour GTT (mg/dL) Initial HbA,< Gestational age at GDM diagnosis (wk) Gestational age when insulin started (wk) Insulin dose at delivery (U/d) Maternal weight at delivery (lb) Cesarean delivery Neonatal birth weight percentile

Population Returning patients (N = Y4, 39% ) 30 z 79 159 i 29.2 i 204 i 112 i

7 36 6.1 64 32

Nonreturning patients (N = 144, 61%) 31 2 77 162 f 29.6 k 195 2 JO9 2

6 41 6.9 50 29

6.0 ? 1.3 25.7 2 5.9

6.3 2 2.8 27.5 i 6.1

28.4 5 5.9

29.5 2 6.3

50 ? 41 188 2 39 34 % 72 +- 26

46 2 37 190 2 44 38% 71 2 26

BMl = body mass index; GTT = glucose tolerance test; HbA,, = hemoglobin A, c; GDM = gestational diabetes mellitus. Data are presented as mean ? standard deviation or %. Comparison\ between returning and nonreturning patients were not statistically significant.

returned for follow-up and those who did not. The study population was 68% Hispanic, 12% white, 9% African-American, 7R Asian, 3% Pacific Islander, and 1% women of other ethnicity, and was comparable for both groups. Of the 94 patients completing postpartum follow-up, 32 (34%) had abnormal glucose tolerance: 17 (18%) had impaired glucose tolerance and 15 (16%) had overt diabetes. Table 2 illustrates the correlations between individual study variables and postpartum diagnosis. No maternal or delivery variables were found to predict postpartum glucose intolerance individually. There was no statistically significant difference in the frequency of postpartum glucose intolerance in subjects with or without a family history of a first-degree relative with non-insulin dependent diabetes mellitus (33 versus 34%), fetal growth profile above the 90th percentile for gestational age (36 versus 33X), or fasting glucose levels of 105 mg/dL or more (26 versus 14%). Patients with postpartum glucose intolerance had gestational diabetes diagnosed earlier in gestation; demonstrated higher l-hour glucose challenge responses, fasting glucose levels, and initial hemoglobin A,, levels; and had earlier initiation of insulin therapy and a greater mean total insulin dose at delivery compared with normal patients. Three factors most useful in the prediction of postpartum glucose intolerance included a screening l-hour oral GTT of 200 mg/dL or greater (32 versus 6% incidence of diabetes; P = .Ol), suboptimal glycemic

Obsfetrics

& Gymdogy

Table

Association Postpartum

2.

of the Study Variables Diagnosis Normal (N = 62)

Maternal Age BMI

With

Impaired GT (N = 17)

Table

Diabetic (N = 15)

Sensitivity, Specificity, and Predictive Value of Antepartum Variables for the Development of Abnormal Glucose Tolerance

diagnosis l-h GTT 3-h GTT: 3-h GTT: 3-h GTE HbA,, at Gestational

(wk) (mg/dL) fasting (mg/dL) 2-h (mg/dL) 3-h (mg/dL) diagnosis age when insulin

started (wk) Insulin at delivery

(U/d)

Delivery variables Cesarean delivery Shoulder dystocia Neonatal variables Birth weight percentile

30 -c 7 27 (43%‘)

32 2 7 6 (38%)

31 k 6 11(73%:)

27 5 5

22 k 7*

23 -t 4’

Specificity

(%)

(70)

42 68

81 71

55 56

64

68

56

188 107 199 153 5.7 31

5 i t ? + -t

45 29 58 63 1.2 4

204 115 230 167 6.0 25

5 k t 2 k +-

42 25 52 56 1.0 8*

264 144 232 187 7.2 26.5

Z 5 2 t 2 f

101’ 47’ 54 69 l.l* 4*

31 i 27

68 2 33*

74 I 49*

35% 8%

28%’ 17%

33% 20%

67 i 26

77 2 26

84 -c 18

GT = glucose tolerance; BMI = body mass index; GTT = glucose tolerance test; GDM = gestational diabetes mellitus; HbA,c = hemoglobin A,,. Data are presented as mean 2 standard deviation, n (%a), or 91. P 5 .002. Comparisons for impaired z’s normal and diabetic ~1s normal; analysis of variance with the Bonferroni correction.

control during gestation (any 2-hour postprandial level of 150 mg/dL or greater, 34 versus 5% diabetes; P = .005), and insulin administration during pregnancy (25 versus 3% impaired glucose tolerance, and 26 versus 0% diabetes in diet-controlled patients; P = .OOl). Patients who required at least 100 U/day of insulin to maintain euglycemia all had postpartum glucose intolerance (20% impaired glucose tolerance, 80%’ diabetes). A stepwise logistic regression was performed to determine the relative contributions of selected risk factors in predicting postpartum glucose intolerance. The factors most predictive individually and chosen for inclusion in the analysis were a previous history of gestational diabetes, a l-hour GTT of 200 mg/dL or more, the requirement for insulin in pregnancy, a total insulin dose at delivery of at least 100 U/day, and any 2-hour postprandial blood sugar level of 150 mg/dL or greater. Of these, the most significant factors were a l-hour GTT of 200 mg/dL or more (Y = 0.244, P = .007), a previous history of gestational diabetes mellitus (Y = 0.180, P = .033), and poor glycemic control with any 2-hour postprandial blood sugar level greater than 150 mg/dL (v = -0.161, P = ,039). Table 3 contains the sensitivities, specificities, and predictive values for postpartum glucose intolerance associated with selected study variables. Individually, each of these factors had a positive predictive value of

86, NO.

1, JULY

lYY5

History of GDM l-h GTT 2200 mg/dL Any

2-h

postprandial

blood

level 1150 mg/dL monitoring) Insulin requiring Insulin ~100 All 3 factors

>200

mg/dL, 2-h postprandial sugar level >150 mg/dL, requiring)

GDM

= gestational

sugar

value

(%I

(home

U/d (1-h GTT

insulin

Positive predictive

Sensitivity variables (y) >30

Metabolic variables Gestational age at GDM

VOL.

3.

97

52

51

19 52

100 96

100 89

blood

diabetes

mellitus;

GTT

= glucose

tolerance

test.

approximately 50%. The negative predictive value of being treated with diet only was high, with 97% of patients with optimal blood sugar control having normal carbohydrate tolerance postpartum. Combining the factors of insulin requirement, poor glycemic control, and an elevated l-hour GTT resulted in improved predictive value (sensitivity 52%, specificity 96%, positive predictive value 89%). Table 4 shows the RRs for the selected study variables The RR for postpartum glucose intolerance resulting from a combination of three antepartum variables (insulin requirement, 2-hour postprandial blood sugar level of 150 mg/dL or more, and l-hour GTT of 200 mg/dL or more) was 19.68 (95% CI 2.88-134.42).

Discussion The 34% incidence of postpartum glucose intolerance in our study is greater than the 6-19% reported by other investigators.5,“16 The study by Kjos et al5 found a 19% incidence of postpartum glucose intolerance in a patient

Table

4.

Relative Risk and 9570 Confidence Interval Persistent Postpartum Glucose Intolerance

History of GDM (N = 20) l-h GTT 2200 mg/dL W = 34) Any 2-h postprandial blood sugar level mg/dL (home monitoring) W = 32) Insulin requiring (N = 31) Insulin 2100 U/d (IN = 7) All 3 factors (l-h GTT >200 postprandial blood insulin requiring) RR = relative diabetes mellitus;

mg/dL,

sugar >150 (N = 181

>150

2-h

RR

95%

1.98 2.86 2.25

1.10-3.55 1.48-5.53 1.21-4.18

CI

17.28 34.00 19.68

2.46-121.45 4.93-234.39 2.88-134.42

GDM

= gestational

mg/dL,

risk; CI = confidence interval; GTT = glucose tolerance test.

Greenberg

for

et al

Gestutionnl

Diabetes

99

population similar to ours. Catalan0 et alh documented a 22% incidence of abnormal postpartum glucose tolerance; however, if patients with nondiagnostic results were excluded, the incidence was only 7%. One reason for the excess incidence of postpartum glucose intolerance in our study compared to that of Catalan0 et al may be the difference in the racial profile of the populations. However, other factors must account for the differing incidences found by Kjos et al” and us. One likely contributor is the limited access to care many of the women in our study population have. Most of these women have medical coverage only for pregnancy and emergencies; thus, medical problems may go undiagnosed until pregnancy brings the patient into the health care system. It is likely that a significant number of our patients had undiagnosed diabetes mellitus before pregnancy, but were considered to have gestational diabetes until a postpartum GTT confirmed the diagnosis of diabetes mellitus. Despite antepartum and early postpartum counseling, most of our patients did not return for postpartum glucose evaluation. Because the patients who did not have postpartum GTTs were similar in all characteristics studied to those who did complete the testing, the conclusions derived from those who returned can be applied reasonably to the entire population. The overall rates of impaired glucose tolerance (18%) and diabetes (16%) would predict an additional 26 patients with impaired glucose tolerance and 23 patients with overt diabetes among the nonreturning subjects, who may be unaware of their condition and at risk for complications associated with untreated hyperglycemia. Diagnosis is also necessary to advise these women of their need for preconception counseling and glycemic control. Thus, postpartum metabolic testing in at-risk patients is of great importance. In this study, we hoped to define factors that would identify a subgroup of patients with the greatest risk of postpartum glucose intolerance. The data indicate that factors strongly predictive of postpartum glucose intolerance are a prior history of gestational diabetes, a l-hour GTT of 200 mg/dL or more, poor glycemic control, the requirement for insulin during pregnancy, and a high total insulin dose. Of these, the need for antepartum insulin is the single most clinically useful variable. Of the 61 patients requiring insulin, 31 (51%) had abnormal glucose tolerance postpartum. None of the patients managed with diet alone during pregnancy developed postpartum diabetes, and only one with impaired glucose tolerance (3%) was identified on postpartum screening. This patient had a previous history of gestational diabetes. Our results suggest three conclusions. First, in the absence of additional risk factors, women with gesta-

100

Greenberg

et al

Ccstatiord

Diabrk5

tional diabetes optimally controlled with diet alone do not require postpartum GTTs. However, some caution is necessary because all women with gestational diabetes in our study practiced home glucose monitoring both during fasting and postprandially with each meal. Thus, our patients with “optimal control” represent a group that can be confidently characterized as normoIn patients unable to maintain all blood glycemic. glucose levels in the normal range, or in those whose blood glucose records are incomplete, postpartum glucose evaluation would be advisable because up to 56% may have postpartum glucose intolerance. Second, all patients with established risk factors for postpartum glucose intolerance should receive postpartum metabolic evaluation. These include a history of recurrent gestational diabetes, episodes of documented hyperglycemia, antepartum insulin treatment, and patients with a screening l-hour GTT exceeding 200 mg/dL. Third, when the total antepartum insulin dose exceeds 100 U/day, all patients can be expected to have postpartum glucose intolerance. Postpartum patients with impaired glucose tolerance should be followed closely for at least 1 year, and thereafter should have a GTT every year, because the rate of conversion from impaired glucose tolerance to diabetes is estimated to be l-5% per year.14 In women with gestational diabetes whose glucose tolerance returns to normal in the postpartum period, further investigation is necessary to determine which pregnancy-related variables may be helpful in predicting the risk of developing carbohydrate intolerance later in life. In these women, the appropriate interval for metabolic evaluation remains to be determined.

References 1. Barden TP, Knowles Clin Obstet Gynecol 2. O’Sullivan JB. Body

HC Jr. Diagnosis of diabetes in pregnancy. 1981;24:3-19. weight and subsequent diabetes mellitus.

JAMA 1982;248:949-52. 3. MacDonald HN, Good W, Schwarz of glucose tolerance in pregnancy

K, Stone J. Serial observations and the early puerperium. Br J

Obstet Gynaecol 1971;78:489-97. 4. Lind T, Harris VG. Changes in the oral glucose tolerance during the puerperium. Br J Obstet Gynaecol 1976;83:460-3.

test

5. Kjos SL, Buchanan TA, Greenspoon JS, Montoro M, Bernstein GS, Mestman MD. Gestational diabetes mellitus: The prevalence of glucose intolerance and diabetes mellitus in the first two months postpartum. Am J Obstet Gynecol 1990;163:93-8. 6. Catalano PM, Vargo KM, Bernstein IM, Amini SB. Incidence and risk factors associated with abnormal postpartum glucose tolerance in women with gestational diabetes. Am J Obstet Gynecol 1991;165:914~6. 7. Summary and Workshop-Conference 1985;34.123-6. 8. Gabbc

recommendations on Gestational

SG, Mestman

JH,

Freeman

of

the Second international Diabetes Mellitus. Diabetes

RK,

et al. Management

and

outcome 1977;127:465 9. Carpenter tional 10. Landon

of

class A diabetes 9. MW, Coustan DR.

mellitus. Criteria

Am

J Obstet

for screening

tests

diabetes. Am J Obstet Cynecol 1982;144:768-73. MB, Gabbe SG. Glucose monitoring and insulin

tration in the pregnant 1985;28:496-506. 11. Coustan DR. Management Coustan DR, eds. Diabetes

diabetic

patient.

Clin

Cynecol

Obstet

of gesta-

ance

In: Reece Principles

EA, and

1979;28:1027-30. 13. Coustan DR, Carpenter MW. Detection and treatment of gestational diabetes. Clin Obstet Gynecol 1985;28:507-15. 14. National Diabetes Data Group: Classification and diagnosis of

tive risks, Gardner

VOL.

86, NO.

odds ratios, MJ, Altman

1, IULY

19~5

and DC,

standardized eds. Statistics

glucose intervals

delivery.

Obstet

Gynecol

London:

of abnormal

glucose

British toler-

1990;75:397-401.

Gynecol

Livingstone, 1988:441-52. Principles of nutrition and dietary with diabetes mellitus. Diabetes

mellitus and other categories of 1979;28:1049. JA, Gardner MJ. Calculating confidence

after

guidelines.

adminisAddress

of gestational diabetes. mellitus in pregnancy:

practice. New York: Churchill 12. American Diabetes Association: recommendations for individuals

diabetes Diabetes 15. Morris

confidence intervals and statistical Medical Journal, 1989:50-63. 16. Oats JN, Beischer NA. The persistence

intolerance.

reprint

requests

to:

Laurie Greenberg, MD OB/GYN ConsuItnnts suite 200 4094 Fourth Avenue Sm Diego, CA 92103

Received October 31, 1994. Receievd in rwised form March Accepted March 13, 1995.

7, 1995.

for rela-

ratios and rates. with confidence-

In:

Copyright

0

1995

by

The

American

College

of Obstetricians

and

Gynecologists.

Greenberg

et al

Gestational

Diabetes

101