Fetal surveillance in pregnancies complicated by insulin-dependent diabetes mellitus

Mark B. Landon, MD: Oded Langer, MD,b Steven G. Gahbe, MD,. Carol Schick, RN,' and Lois Brustman, MDb

Columbus, Ohio, and Bronx, New York

OBJECTIVE: Our objective was to determine whether maternal vascular disease and/or glycemic control

can be related to tests of fetal condition in diabetic pregnancies.

STUDY DESIGN: A total of 114 women with insulin-dependent diabetes who used a memory-based

glucose reflectance meter were prospectively evaluated. Nonstress testing was begun weekly at 28 to 30

weeks and twice weekly at 32 weeks. A nonreactive nonstress test was followed by a biophysical profile in

all cases.

RESULTS: A total of 1676 nonstress tests was performed (14.7 ± 3.2 tests per patient). Eight percent

(n = 134) were nonreactive, necessitating a biophysical profile. A comparison of ambulatory glucose

profile data, including mean blood glucose level, variation, and excursions from the median, revealed no

significant differences in patients with reactive versus nonreactive nonstress tests. Ten patients, including

eight with vascular disease, were delivered because of abnormal test results of fetal condition.

Nephropathy or hypertension was associated with intervention for fetal well-being in 8 of 20 women (40%)

with these risk factors. Only 2 of 94 patients (2%) without nephropathy or hypertension required delivery

because of abnormal results of fetal testing (p < 0.001). One fetal death occurred. No significant

differences in the various glycemic parameters were found in women delivered for suspected fetal

jeopardy versus the nonintervention group.

CONCLUSION: Pregnancies complicated by vascular disease are at greatest risk for abnormal results of

fetal testing that necessitate early delivery. Women without vascular complications and with maintenance

of good glycemic control rarely have fetal compromise. (AM J OSSTET GVNECOL 1992;167:617-21.)

Key words: Diabetic pregnancy, glycemic control, fetal testing

Although intrauterine fetal death has become an un­common event in pregnancies complicated by insulin­dependent diabetes, intensive antepartum fetal sur­veillance during the third trimester remains the stan­dard practice of obstetricians.' Goals of antepartum fe­tal surveillance include the avoidance of intrauterine death, early detection of fetal compromise, and pre­vention of unnecessary preterm delivery. None of these techniques has been subjected to prospective random­ized clinical trials, but the methods have been incor­porated into complex patient-care programs consisting of a specialized team approach with an emphasis on normalization of maternal blood glucose levels! Inter­vention because of abnormal results of fetal testing is performed in <5% of pregnancies complicated by in­sulin-dependent diabetes mellitus.' Prior studies con­cerning fetal surveillance in pregnancies by insulin-de-

pendent diabetes mellitus have failed to include a de­tailed description of the clinical characteristics of the study population, such as the presence or absence of vascular disease and a detailed description of glycemic control. The current study was therefore performed to determine if a relationship exists between the results of tests of fetal condition and these parameters in women with insulin-dependent diabetes mellitus. We hypothesized that in a population with well-controlled insulin-dependent diabetes mellitus the presence of maternal vasculopathy would be the principal risk fac­tor for fetal compromise.

From the Department of Obstetrics and Gynecology, The Ohio Stale University College of Medicine," and the Department of Obstetric; and Gynecology, Albert Einstein College of Medicine. b

Received for publication March 18, 1992; accepted March 31 ,1992. Reprint requests: Mark B. Landon, MD, The Ohio State University, Department of Obstetrics and Gvneeology, 1654 Upham Dr., 5th Floor, Columbus, OH 43210. 611 /38214

Material and methods

During the period July 1988 to January 1991, all pregnant patients with insulin-dependent diabetes mel­litus using a memory-based glucose reflectance meter and undergoing fetal surveillance during the third trimester were enrolled in this study. Patients were in­structed on the use of self-monitoring blood glucose techniques and were requested to perform seven fin­gerstick determinations daily. The reflectance meter contained an on-board memory that permitted storage of 400 glucose readings with the corresponding times and dates. These data were off-loaded weekly to pro-

617

618 Landon et al.

Table I. Results of NSTs according to parameters of glycemic control

(n = 85) (n = 29) Reactive N onreactive*

Glucose (mg/dl, 109.9 ± 20.2 107.5 ± 10.6 mean ± SD)

Reversal magni- 34.4 ± 15.7 35.2 ± 14.3 tude (varia-tion) (mg/dl, mean ± SD)

Excursions 3.1 ± 1.4 3.4 ± 1.6 from median glucose (mg/dl, mean ± SD)

p Value

0.5

0.8

0.3

*Nonreactive test result recorded at >32 weeks' gestation.

duce an ambulatory glucose profile for each study pa­tient. An ambulatory glucose profile reflects glucose patterns by the representation of all self-monitored data as a series of curves representing a single typical day. For each ambulatory glucose profile the mean, median, interquartile range (distance between 25th and 75th percentiles), reversals in the curve, and excursions from the median are determined"

Patients designated as having class F diabetes in­cluded women with a creatinine clearance of < 1 00 mllmin and/or protein excretion >400 mg/24 hours before 20 weeks' gestation in the absence of a urinary tract infection. Chronic hypertension was defined as a persistent diastolic blood pressure >90 mm Hg before 20 weeks' gestation or conditions necessitating antihy­pertensive therapy. The criteria for the diagnosis of preeclampsia were elevation of blood pressure by 30 mm Hg systolic and 15 mm Hg diastolic above baseline and proteinuria in excess of 400 mg/24 hours in in­dividuals without preexisting nephropathy.

All patients performed a daily assessment of fetal activity during the third trimester.s Nonstress tests (NSTs) were performed weekly at 28 to 30 weeks and biweekly at 32 weeks' gestation and thereafter. A re­active NST included at least two accelerations of the fetal heart rate of 15 beats/min lasting 15 seconds within 20 minutes. The test was carried out for a max­imum of 45 minutes. Feeding or drinking was not re­quired in the presence of a nonreactive testing. Acoustic stimulation was performed occasionally. However, a nonreactive NST was followed by a biophysical profile in all cases.6 A biophysical score of 8 to 10 was consid­ered normal and a score of 6 equivocal. Patients with a score of 6 were delivered if pulmonary maturity was documented. If the fetal lung profile was immature, the biophysical profile was repeated in 24 hours. A score of :s4 was considered abnormal and was followed by delivery.

Our management protocol encouraged patients to

September 1992 Am J Obstet Gynecol

be monitored to term (>37 weeks), as long as results of fetal testing and maternal condition remained stable. Patients with vascular disease and normal test results of fetal condition were delivered before term only if hypertension worsened or if lack of fetal growth was documented on serial ultrasonographic examinations. In uncomplicated cases labor was induced at term (>37 weeks) if the cervix was favorable. Amniocentesis was performed only in cases delivered before 39 weeks' gestation. Ripening of the cervix with prostaglandin gel was used in patients at 39 to 41 weeks' gestation with an unfavorable cervix.

Gestational age at delivery was determined from menstrual dates, provided these were in agreement with results first- or early second-trimester ultrasono­graphic examinations. Infants were characterized as having intrauterine growth retardation (IUGR) if birth weight was <10th percentile.7 End points of perinatal outcome included the presence of a low Apgar score (5-minute Apgar score <7), neonatal acidemia (cord arterial pH :s7.20), or admission to the intensive care unit for a condition other than prematurity, hypogly­cemia, or congenital malformation.

Standard X 2 contingency tests and Student t test were used to assess statistical significance between compari­son groups.

Results

One hundred fourteen women with insulin-depen­dent diabetes mellitus were studied. Sixty-three women received care at The Ohio State University College of Medicine and 51 at Albert Einstein College of Medicine. According to the White classification, 44 were class B, 32 class C, 24 class D, 12 class F, and 2 class R. Eight women also had chronic hypertension. A total of 1676 NSTs were performed (14.7 ± 3.2 tests per patient). One hundred thiry-four tests (8%) were nonreactive and necessitated a follow-up biophysical profile. The frequency of nonreactive test results was not signifi­cantly different among White classes. Sixty-eight of the nonreactive tests occurred before 32 weeks' gestation. Twenty-nine patients had a total of 66 nonreactive NSTs after 32 weeks. Table I shows the results of NSTs according to various ambulatory glucose profile param­eters of glycemic control. Nonreactive tests are com­monly observed in premature fetuses; therefore the nonreactive group is defined as those patients with non­reactive testing at or beyond 32 weeks' gestation.s No significant differences in glycemic indices were ob­served between reactive and nonreactive groups.

In 104 of 114 patients (91.2%), the last NST before delivery was reactive. The average interval from this test to delivery was 3.2 ± 1.1 days. The mean gesta­tional age at delivery for these women was 38.4 ± 0.9 weeks. Sixty-one patients in this category underwent

Volume 167 Number 3

Glucose control and fetal testing 619

Table II. Cases of abnormal results of fetal testing necessitating delivery

Class

FIR Preeclampsia

D

FIR

B Chronic hypertension B Chronic hypertension

B Preeclampsia

FIR

FIR Preeclampsia F

F Preeclampsia

LI S, Lecithin I sphingomyelin.

Weeks' gestation

28

38

32

33

30

32

33

30

30

34

elective induction of labor or repeat cesarean delivery whereas spontaneous labor began in 43. The overall cesarean section rate for the entire study population was 42%.

Ten patients were delivered because of abnormal re­sults of fetal testing, including eight with nephropathy or hypertension (Table II). Intervention because of ab­normal testing was accomplished in 8 of 20 women (40%) with class F disease or chronic hypertension com­pared with 2 of 94 women (2.1 %) without these risk factors (p < 0.001). The mean gestational age at deliv­ery in the intervention group was 32.0 ± 2.8 weeks. Among these 10 patients, eight underwent cesarean section. A comparison of various parameters of gly­cemic control from the third trimester failed to show significant differences between the 10 patients with ab­normal results of testing leading to delivery and the 104 patients with normal results of testing before de­livery (Table Ill). One fetal death occurred. A 28-year­old, gravida 2, para 1 woman with class C disease was delivered of a stillborn fetus at 36 weeks' gestation. A NST had been reactive 7 days before this event, but the patient failed to keep an appointment for a sched­uled NST 3 days earlier. Of significance, the mean third-trimester glucose level in this patient was elevated at 158 mg/dl.

Neonatal outcome is summarized in Table IV. There were no neonatal deaths. Overall, 5 of 10 neonates in

Test result

Nonreactive NST Biophysical profile score 6 Oligohydramnios Variable decelerations on NST

Nonreactive NST Biophysical profile score 6 Oligohydramnios Nonreactive NST Biophysical profile score 4 Nonreactive NST Biophysical profile score 4 Mature LIS ratio Nonreactive NST Biophysical profile score 6 Mature LIS ratio Nonreactive NST Biophysical profile score 6 Mature LIS ratio Nonreactive NST Biophysical profile score 4 Nonreactive NST Biophysical profile score 4 Vertebral anomaly Nonreactive NST Biophysical profile score 6 Mature LIS ratio

Outcome

820 gm Apgar scores 7 and 8

2600 gm Apgar scores 5 and 6 1608 gm A pgar scores 7 and 8

1730 gm Apgar scores 3 and 6 1050 gm Apgar scores 8 and 9

960 gm Apgar scores 4 and 6

1405 gm Apgar scores 9 and 9

810 gm Apgar scores 3 and 6 860 gm Apgar scores 4 and 5

1990 gm Apgar scores 4 and 7

the intervention group versus 4 of 104 in the nonin­tervention group had 5-minute Apgar scores <7 (p < 0.001). The mean cord arterial pH of 7.21 ± 0.7 in the intervention group was also significantly lower than the mean pH of7.29 ± 0.8 in patients with normal test results before delivery (p < 0.003).

Comment

Significant improvement has been made in reducing the frequency of intrauterine fetal death in pregnancies complicated by insulin-dependent diabetes mellitus. In the past, sudden unexplained stillbirths occurred in 10% to 30% of gestations complicated by insulin-de­pendent diabetes mellitus. 3 Such losses are uncommon in patients with well-controlled disease, yet stillbirths may be observed in patients who do not receive optimal care. In a population-based study fetal death was nine times more common in diabetic pregnancies than in a nondiabetic control population." Fetal deaths occur most often in late pregnancy in women with ketoaci­dosis or poor metabolic control or in women with vas­cular disease and preeclampsia. Chronic intrauterine hypoxia represents the likely cause in such cases.lO

Over the past two decades techniques for antepartum fetal surveillance in pregnancies complicated by insulin­dependent diabetes mellitus have been developed to provide a rational basis for management during the third trimester, the period of greatest fetal risk. The

620 Landon et al.

Table III. Comparison of glycemic characteristics

Nonintervention (n = 104) P Value

Mean glucose (mg/dl, mean ± SD)

Reversal magnitude (mg/dl, mean ± SD) (variation)

Excursions from me­dian glucose (mg/dl, mean ± SD)

105.9 ± 8.8 109.9 ±20.2 0.5

33.8 ± 13.3 34.4 ± 15.7 0.9

2.4 ± 0.9 3.1 ± 1.4 0.1

Table IV. Neonatal outcome

Intervention Nonintervention group group

(n = 10) (n = 104)

No. I % No. I % P Value

5 min Apgar 5 50 4 3.8 <0.001 score <7

Cord artery 4 40 6 7.2* <0.01 pH <7.20

Admission to 8 80 8 7.7 <0.001 NICU

*pH obtained in 83 cases. NICU, Neonatal intensive care unit.

primary clinical value of antepartum fetal monitoring tests is their ability to reassure the clinician that the fetus will not die in utero! These tests have few false­negative results and therefore allow safe prolongation of pregnancy with further fetal maturation in a met­abolically stable patient. In our study the low incidence of neonatal acidemia and the requirement for neonatal intensive care in the offspring of patients with reas­suring results of fetal testing indicates that a normal testing sequence is highly predictive of a good outcome.

Our data also support the work of Golde et al.l! that the use of the NST on a semiweekly basis provides an adequate level of fetal surveillance for the diabetic pop­ulation. The biophysical profile is an excellent backup test for the NST. In our experience the biophysical profile was reassuring in 90% of cases when the NST was found to be nonreactive. No depressed infants were delivered after this sequence of testing. Some institu­tions have used the biophysical profile as a primary method of testing, but no study of diabetic women has demonstrated its superiority over the NST.'2

Our review of 879 diabetic pregnancies reported over the last 15 years revealed that intervention for abnormal test results of fetal condition is undertaken in <5% of cases.' Our intervention rate of 8% is com­parable to this figure. It may be slightly increased be­cause of the greater proportion of women with vascular

September 1992 Am J Obstet Gynecol

disease included in this series. We observed only one case of a patient without nephropathy or hypertensive disease requiring delivery because of a suspicious NST. Previous studies have demonstrated that diabetes com­plicated by vascular disease is associated with a high rate of positive contraction stress tests leading to pre­term delivery. IS Recently, Mimouni et al. l4 examined risk factors for perinatal asphyxia in 162 infants of women with insulin-dependent diabetes mellitus. Wors­ening nephropathy was identified as the most signifi­cant risk factor for fetal distress during labor or still­birth. We have previously reported that diabetic women with preexisting vascular disease or preeclampsia also commonly exhibit abnormal umbilical artery wave­forms in association with fetal growth retardation. Se­rial ultrasonography to assess fetal growth is thus an important adjunct in these high-risk women. IS

Few studies have attempted to correlate tests of fetal condition with maternal glycemic control. It has been suggested that excellent control may decrease the in­cidence of abnormal antepartum tests leading to un­scheduled delivery.' In a study antedating the use of blood glucose self-monitoring, Gabbe et al. 16 intervened because of suspected fetal compromise in 21 % of 260 women with insulin-dependent diabetes mellitus and mean fasting glucose levels of 109 to 140 mg/dl ob­tained at clinic visits. Using home monitoring tech­niques Coustan et al. 17 observed a rate of 2.7% in 72 patients with tightly controlled disease. In 52 women without vascular disease and with maintenance mean blood glucose levels ranging from 80 to 87 mg/dl,Jov­anovic et al. 18 reported no interventions for abnormal fetal testing.

In addition to our report, Teramo et al. 19 have pro­vided the only other detailed description of glycemic parameters in patients with insulin-dependent diabetes mellitus undergoing fetal heart rate testing during the third trimester. In their study the mean hemoglobin AI, level during the third trimester was higher in 18 women with pathologic tracings compared with that of 118 with normal patterns. l9 However, mean glucose levels were similar between groups. In the current study parameters of glycemic control determined by verified blood glucose data were not different between the in­tervention and nonintervention groups. Overall, the mean blood glucose level in our study population was 110 ± 19 mg/dl, indicating that most patients under our care had well-controlled disease. Mimouni and et al. 14 could not demonstrate an association between third-trimester glucose control and perinatal asphyxia. However, few glucose measurements were obtained for each patient and the overall mean fasting value was elevated at 127 to 131 mg I dl in their study population. Leveno et al.20 reported that pregnancies of diabetic women whose overall mean plasma glucose levels ex-

Volume 167 Number 3

ceeded 172 mg I dl required earlier intervention for signs of fetal jeopardy, including decreased fetal move­ment. Whereas few patients in our study had poorly controlled disease, we recognize that we did undertake elective delivery in a small number of patients with suboptimal control. This action may have decreased the frequency of abnormal results of fetal testing observed in patients without vascular disease.

Whether rigorous programs of fetal surveillance are necessary for all patients with insulin-dependent dia­betes mellitus remains controversial. It is unlikely that sufficiently large prospective studies to examine the efficacy of such testing will ever be undertaken. Our testing protocol, like others, does carry significant ben­efits. The majority of patients can safely undergo fol­low-up to term without unnecessary intervention. A subgroup of pregnancies with substantial fetal risk is identified. We have confirmed that those pregnancies complicated by vascular disease should be considered at greatest risk for abnormal results of fetal testing necessitating early delivery. In these patients testing should begin as early as 28 weeks' gestation. Patients without vascular complications and those in whom good glycemic control is maintained rarely show fetal com­promise. In these women testing may be initiated at 32 weeks or later. Some have recommended that such seemingly "low-risk" patients not be tested. 21 We do not support this position. However, we certainly agree with the observation of BrudenelJ22 and Doddridge that women with well-controlled disease remain at relatively low risk. Further studies may challenge the need for all diabetic women to be routinely subjected to fetal monitoring in late pregnancy.

REFERENCES

I. Landon MB, Cabbe SC, Sachs L. Management of diabetes mellitus and pregnancy: A survey of obstetricians and maternal-fetal specialists. Obstet Cynecol 1990;75:535-9.

2. Golde S, Platt L. Antepartum testing in diabetes. Clin Obstet Gynecol 1985;28:516-27.

3. Landon MB, Gabbe SC. Fetal surveillance in the preg­nancy complicated by diabetes mellitus. Clin Obstet Cyn­ceol 1991;34:535-43.

4. Mazze RS, Lucido D, Langer 0, et al. The ambulatory glucose profile: a representation of verified self monitored blood glucose data. Diabet Care 1987; 10: 111-7.

5. Rayburn WF, McKean HE. Maternal perception of fetal movement and perinatal outcome. Obstet Gynecol 1980;56:161-6.

Glucose control and fetal testing 621

6. Manning FA, Morrison J, Lange JR, Harman CR, Cham­berlain PF. Fetal assessment based upon fetal biophysical profile scoring: experience in 12,620 referred high risk pregnancies. I. Perinatal mortality by frequency and etiol­ogy. AMJ OBSTET CYNECOL 1985;151:343-50.

7. Brenner WE, Edelman DA, Hendricks CH. A standard of fetal growth for the United States of America. AM J OBSTET GYNECOL 1976; 126:555-64.

8. Lavin J, Miodovnik M, Barden T. Relationship of non­stress test reactivity and gestational age. Obstet Cynecol 1984; 148:591-6.

9. Connell FA. Vadheim C, Emanual J, et al. Diabetes in pregnancy: a population-based study of incidence, refer­ral for care, and perinatal mortality. AM J OBSTET Cy­NECOL 1985;151:598-602.

10. Madsen H. Fetal oxygenation in diabetic pregnancy. Dan Med Bull 1986;33:64-74.

11. Colde SH, Montaro M, Good-Anderson B, et al. The role of non-stress tests, fetal biophysical profile, and contrac­tion stress tests in the outpatient management of insulin­requiring diabetic pregnancies. AM J OBSTET GYNECOL 1984; 148:269-73.

12. Johnson JM, Lange JR, Harmon CR, et al. Biophysical profile scoring in the management of the diabetic preg­nancy. Obstet Gynecol 1988;72:841-6.

13. Kitzmiller JL, Brown ER, Phillippe M, et al. Diabetic ne­phropathy and perinatal outcome. AM J OBSTET GY:-.iECOL 1981; 141 :741-8.

14. Mimouni F, Miodovnik M, Siddiqi TA, et al. Perinatal asphyxia in infants of insulin-dependent diabetic mothers. J Pediatr 1988;113:345-53.

15. Landon MH, Gabbe SG, Bruner JP, et al. Doppler um­bilical artery velocimetry in pregnancy complicated by in­sulin-dependent diabetes mellitus. Obstet Cynecol 1989;73:961-5.

16. Gabbe SG, Mestman JH, Freeman RK, et al. Management and outcome of pregnancy in diabetes mellitus, classes B to R. AM J OBSTET GYNECOL 1977;129:723-8.

17. Coustan DR, Berkowitz RL, Hobbins JC. Tight metabolic control of overt diabetes in pregnancy. Am J Med 1980;68:845-52.

18. Jovanovic L, Druzin M, Peterson C. Effect of euglycemia on the outcome of pregnancy in insulin-dependent dia­betic women as compared with normal control subjects. Am J Med 1981;71 :921-31.

19. Teramo K, Ammala P, Ylinen K, et al. Pathologic fetal heart rate: association with poor metabolic control in di­abetic pregnancies. Obstet Cynecol 1983;61 :559-65.

20. Leveno KJ, Hauth JC, Gilstrap LC, et al. Appraisal of "rigid" blood glucose control during pregnancy in the overtly diabetic woman. AM J OBSTET GYNECOL 1979;135:853-62.

21. Drury MI, Stronge JM, Foley M, et al. Pregnancy in the diabetic patient: timing and mode of delivery. Obstet Gy­necol 1983;62:279-83.

22. Brudenell M, Doddridge MC. Obstetric management in the diabetic pregnancy. London: Churchill Livingstone, 1989:53-69.