Miailhe, Grégoire MD; Le Ray, Camille MD, PhD; Timsit, José MD; Lepercq, Jacques MD
The rate of perinatal mortality in neonates of women who have diabetes has decreased dramatically with improved prenatal care, including aggressive blood sugar control, fetal surveillance techniques, timely labor induction, and advanced neonatal care. Nevertheless, type 1 diabetes mellitus (DM) is still associated with a threefold to fivefold increased risk for stillbirth during pregnancy ranging from 20 to 25 per 1,000 births in recent European studies.1–3 Furthermore, it has been reported recently, in a Norwegian national registry, that the excess risk of stillbirth was confined to term birth.2
Approximately half of all stillbirths can be explained by causes such as fetal growth restriction, preeclampsia, ketoacidosis, umbilical cord problems, acute asphyxia, placental abruption, infection, and congenital malformations, whereas the other half remained unexplained. Unexplained stillbirth has been associated with poor glycemic control, diabetic nephropathy, smoking, and a low socioeconomic status.4 Although the pathophysiology remains unknown, fetal hypoxia5 and cardiac dysfunction,6 both secondary to poor glycemic control, have been associated with fetal death in such pregnancies. Animal studies have shown that chronic fetal hyperglycemia increases fetal oxidative metabolism and decreases arterial oxygen contents.7 Prolonged hypoxia is a major stimulus for fetal erythropoietin production,8 and fetal plasma and amniotic erythropoietin are frequently elevated in diabetic pregnancies.5 A hemoglobin A1C (Hb A1C) level in the third trimester correlates with umbilical erythropoietin at birth, suggesting that maternal hyperglycemia is a significant factor associated with fetal hypoxia.9 Thus, achieving good glycemic control during pregnancy is a critical issue to avoid fetal hypoxia and to reduce the risk of stillbirth.10
Efforts have been initiated to identify fetuses susceptible to stillbirth before it happens. Although there is no consensus regarding antenatal testing, frequently used methods include ultrasonographic examinations of the fetal growth rate, kick counting, and antenatal nonstress tests. Retrospective studies have reported that frequent nonstress tests can identify fetuses at high risk for stillbirth, and anticipated delivery of these neonates may decrease the stillbirth rate,11,12 although this has been questioned by others.13 Currently, most centers apply nonstress testing one to two times weekly from 32 to 34 weeks of gestation until delivery.
Hence, the objective of the present study was to identify risk factors associated with prelabor urgent cesarean delivery for an abnormal nonstress test in women with type 1 DM managed with standardized protocols regarding diabetes care and fetal surveillance.
MATERIALS AND METHODS
All single pregnancies in women with type 1 DM who were consecutively delivered after 22 weeks of gestation in the Department of Obstetrics and Gynecology, Cochin-Saint-Vincent-de-Paul Hospital between 1996 and 2010 were included in this prospectively collected cohort. Pregnancies complicated by major congenital malformations were excluded. After receiving approval from the institutional review board Comité de Protection des Personnes se prêtant à la Recherche Biomédicale Ile de France 3, we performed a nested case–control study within this prospective cohort.
Preconception care and management of diabetes during pregnancy were standardized as previously reported.14 Briefly, preconception care included assessment of diabetes complications, review of dietary habits, folic acid supplementation, and intensification of capillary blood glucose self-monitoring and of insulin therapy. Insulin therapy was given by three or more daily injections or by continuous subcutaneous infusion using an external pump. Capillary blood glucose target values were less than 95 mg/dL (5.3 mmol/L) before and less than 120 mg/dL (6.7 mmol/L) 2 hours after each of the three main meals, respectively. The Hb A1C level was measured by high-performance liquid chromatography (normal 4.3–5.7%) during the first and second trimesters and at delivery. During pregnancy, women were seen every other week at the diabetes clinic and reached a member of the team by phone as needed.
All women were followed monthly at the antenatal clinic. Ultrasonographic scans were performed at 12–14, 22–24, and 32–34 weeks of gestation according to French guidelines. In the absence of complications, delivery was planned at 38–39 weeks of gestation, as previously described.15
Antenatal fetal surveillance was initiated at 32 weeks of gestation and continued until delivery. Twice-weekly home visits were performed by specifically trained midwives. Interventions comprised a targeted questionnaire, arterial pressure measurement, bedside urine analysis, and fetal heart rate monitoring using cardiotocography. With the woman in the lateral tilt position, the fetal heart rate was monitored with an external transducer for 20 minutes and analyzed by the midwife. In the absence of accelerations, tracing was continued for 40 minutes to take into account the variations of the fetal sleep–wake cycle. We classified nonstress tests into two groups as follows: normal (reactive, ie, two or more fetal heart rate accelerations within a 20-minute period according to American College of Obstetricians and Gynecologists Practice Bulletin16) or abnormal (reactive with deceleration, nonreactive with or without deceleration).
Women with an abnormal nonstress test were immediately admitted to Saint Vincent de Paul's hospital where fetal heart rate recordings were analyzed. Fetal heart rate was systematically monitored at admission. According to the severity of nonstress test abnormalities and gestational age, women were either immediately delivered by cesarean delivery or hospitalized for close surveillance when fetal compromise was not immediately life-threatening (absence of deceleration). In those women, fetal surveillance included nonstress tests three times daily. The biophysical profile was not systematically included in the fetal surveillance. A cesarean delivery performed within 48 hours of admission for an abnormal nonstress test was defined as an urgent cesarean delivery, as suggested by Lucas et al.17 In this study, immediate or urgent cesarean deliveries were considered as cases. Control participants comprised pregnancies with a normal nonstress test or recovery of a normal nonstress test during hospitalization.
Maternal characteristics included age, ethnicity, marital status, socioeconomic status, smoking, and prepregnancy body mass index (BMI, calculated as weight (kg)/[height (m)]2). Diabetes characteristics included duration of diabetes, retinopathy, nephropathy, and chronic hypertension. Gestational age was determined from the date of the last menstrual period confirmed by a first-trimester ultrasonography. Chronic hypertension, gestational hypertension, and preeclampsia were defined according to the National High Blood Pressure Education Program Working Group on High Blood Pressure in Pregnancy.18 Polyhydramnios was defined as an amniotic fluid index of 25 cm or more on the third-trimester ultrasonography scan. All the data, including fetal heart rate monitoring, were prospectively collected and analyzed post hoc.
Perinatal mortality was defined as the combined rate of stillbirth and neonatal mortality within the first 28 days of life. Preterm delivery was defined as delivery before 37 weeks of gestation. Birth weight according to gestational age was used to define neonates as small for gestational age, appropriate for gestational age, and large for gestational age according to the French growth standards.19 Neonatal complications included neonatal acidosis, admission to neonatal intensive care unit, hypoglycemia, respiratory distress syndrome, and intraventricular hemorrhage. Neonatal acidosis was defined as an umbilical cord pH less than 7.00. Neonatal hypoglycemia was defined as a plasma glucose level less than 40 mg/dL (2.2 mmol/L). Respiratory distress syndrome was defined as the need for oxygen therapy or invasive ventilation for more than 24 hours.
Factors associated with prelabor urgent cesarean delivery for an abnormal nonstress test were identified by comparing maternal demographic, medical and obstetric characteristics among patients and control participants. Comparisons between the two groups were performed with the t test or χ2 and Fisher’s exact test when appropriate. Hemoglobin A1C level is a commonly used measure associated with pregnancy outcome in women with type 1 DM; therefore, receiver operating characteristics analyses were used to determine the threshold of Hb A1C level during pregnancy most predictive of prelabor immediate or urgent cesarean delivery.
We calculated the sample size to achieve a power of 80% with the following hypotheses: 1) a prevalence of at least 0.20 of the studied variable in the control group; and 2) an odds ratio (OR) 3.0 or more associated with the studied variable for the outcome (α, 0.05, case–control ratio, 1/25). Under these assumptions, the calculated total sample size was 416 women. Independent factors associated with urgent cesarean delivery for an abnormal nonstress test and OR were identified by logistic regression. As a result of the sample size, linearity with independent outcome could not be assumed; thus, continuous variables were recoded in nominal variables. For variables with more than 3% missing data, an additional class for missing data was created to avoid a loss of power in multivariable analysis. In case of correlation between two or more variables, the most clinically important was chosen. Variables were kept in the model if P values were lower than .05. Adjusted ORs were reported with 95% confidence interval (CI). All statistical analyses were performed using STATA 11.0.
During the 15-year period, 482 single pregnancies in women with type 1 DM were consecutively delivered after 22 weeks of gestation. Two pregnancies with major congenital malformations (anencephaly, cardiopathy) were excluded. The rate of stillbirth was 2 per 1,000.
Normal nonstress tests were reported in 459 pregnancies, abnormal nonstress tests were reported in 20 pregnancies, and one stillbirth occurred. Among the 20 pregnancies with abnormal nonstress test, two recovered spontaneously and were further analyzed as control participants, and 18 (4%) underwent urgent cesarean delivery and were considered as case participants. The main characteristics of the 479 pregnancies are reported in Table 1. In nine pregnancies, the indication for immediate delivery was a nonreactive nonstress with decelerations in five cases and a reactive nonstress test with decelerations in four cases. In nine women, the initial nonstress test was reassuring, but urgent cesarean delivery was performed within 48 hours because of the appearance of the following abnormalities during surveillance: five had a nonreactive test with decelerations, three had a reactive test with decelerations, and one had a persistent nonreactive test without deceleration.
One stillbirth occurred at 35 weeks of gestation in a 26-year-old woman with a prepregnancy BMI of 28.6. Diabetes duration was 18 years and was complicated with nonproliferative retinopathy. Pregnancy was unplanned and first-trimester Hb A1C level was 8.1%. The woman had no hypertension or ketoacidosis, and fetal death occurred within 3 days of a normal nonstress test. The neonate was a 2,510-g (50th percentile) boy without congenital malformation. The Hb A1C level at delivery was 7.7%, and no cause of stillbirth other than poor glycemic control was identified.
Two neonates delivered by immediate cesarean for an abnormal nonstress test had severe neonatal acidosis. One occurred at 35.5 weeks of gestation in a 32-year-old woman with a prepregnancy BMI of 28.9. Diabetes duration was 1 year and was not complicated. Pregnancy was unplanned and first-trimester Hb A1C level was 8.2%. During pregnancy, glycemic control was poor and Hb A1C level at delivery was 7.0%. A nonreactive nonstress test with decelerations was evident at the time of a maternal diabetic ketoacidosis. According to the severity of the nonstress test and the gestational age, immediate cesarean delivery was performed. The neonate was a 3,490-g (greater than the 95th percentile) girl whose umbilical pH was 6.93. Neonatal course was uneventful. The other occurred at 35 weeks of gestation in a 27-year-old woman with a prepregnancy BMI of 24.1. Diabetes duration was 17 years and was complicated with nonproliferative retinopathy. The mother did not smoke. Pregnancy was planned and first-trimester Hb A1C level was 7.3%. The Hb A1C level at delivery was 6.4%. Placental abruption occurred at home in the absence of hypertension and the woman was immediately transferred to the nearest hospital where immediate cesarean delivery was performed. The neonate was a 3,210-g (90th percentile) boy whose umbilical pH was 6.70. The neonate developed hypoxic–ischemic encephalopathy and died in the neonatal intensive care unit at Day 32.
The rate of preterm delivery was higher in case participants than in control participants (72% compared with 15%, P<.001). Among 71 pregnancies from the control group, preterm delivery was spontaneous (ie, as a result of preterm labor of premature rupture of the membranes) in 33 (46%) case participants and indicated in 38 (54%) case participants. The main reason for indicated preterm delivery was preeclampsia (n=23, 61% of the case participants). Other reasons were poor glycemic control (n=6), psychiatric disorder (n=1), thrombocytopenia (n=1), acute fatty liver of pregnancy (n=1), placenta previa (n=1), fetal growth restriction (n=1), macrosomia (n=1), acute polyhydramnios (n=1), oligohydramnios (n=1), and previous stillbirth (n=1). The neonatal outcomes are reported in Table 2.
In the univariable analysis, several maternal factors were associated with urgent cesarean delivery for an abnormal nonstress test: lack of preconception care, occurrence of gestational hypertension, or preeclampsia. A low socioeconomic status, smoking during pregnancy, the presence of nephropathy, and development of a polyhydramnios were not associated with urgent cesarean delivery. First-trimester and second-trimester Hb A1C level and Hb A1C level at delivery were higher in women delivered by immediate or urgent cesarean (Table 1). Because a correlation was observed between Hb A1C level during the first and second trimesters and at delivery, Hb A1C level at delivery was chosen as the most clinically relevant variable because it reflects glycemic control during the last months of pregnancy. Receiver operating characteristic curves were constructed for urgent cesarean delivery and Hb A1C level at delivery (Fig. 1). An Hb A1C level of 6.4% at delivery emerged as the optimal cutoff for predicting immediate or urgent cesarean delivery with sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio of 70.6%, 66.7%, 2.1, and 0.4, respectively. Logistic regression models were tested by including risk factors with P values <.05 on univariable analyses. In the final model, Hb A1C level at delivery was dichotomous (less than 6.4% compared with 6.4% or more). In 161 pregnancies (34%, 95% CI 30–39%), Hb A1C level at delivery was 6.4% or more. In the multivariable analysis, an Hb A1C level at delivery of 6.4% or more remained the only factor independently associated with immediate or urgent cesarean delivery for an abnormal nonstress test (2% compared with 8%, P=.003, OR 4.16, 95% CI 1.40–12.32) (Table 3).
In this cohort, the rate of immediate or urgent cesarean delivery for an abnormal nonstress test was 4%. An Hb A1C level at delivery of 6.4% or more was the sole factor independently associated with immediate or urgent cesarean delivery. The rate of stillbirth was 2 per 1,000.
Our study has several strengths: all the data were prospectively collected according to standard definitions; all eligible pregnancies were included and managed according to standardized protocols during the whole study period; there was no loss of follow-up; and finally logistic regression analysis was performed to adjust for potential confounders. The limitations of our study are its single-center setting and its sample size that allowed detection of associations with OR 3.0 or more. A sample size of 1,508 pregnancies would have been required to demonstrate associations with an OR 2.0 or more. However, type 1 DM is uncommon (approximately 3 per 1,000 pregnancies), and stillbirth remains a real concern.
A suboptimal glycemic control (Hb A1C level 4 standard deviations above the normal range) has been identified as a major potentially correctable risk factor for the prevention of stillbirth in women with type 1 DM.4 Similarly, in a nationwide prospective Danish study conducted between 1993 and 1999, the mean third-trimester Hb A1C level was 7.1% in women with serious adverse pregnancy outcomes, including stillbirth, and 6.7% in women with uncomplicated pregnancies.20 Similarly, in a retrospective study conducted in women with type 1 DM, Hb A1C level during the last trimester was higher in women with abnormal nonstress test than in women with normal nonstress test (7.6% compared with 6.9%).21 The mean Hb A1C level at delivery for our entire cohort was 6.2%. It was 7.2% in the case participants and 6.1% in control pregnancies. This may explain why the observed rate of stillbirth was very low, similar to that reported for nondiabetic pregnancies22 and much lower than that reported in women with type 1 DM in recent European studies.1–3 Altogether, we conclude that stillbirth is rarely observed when good glycemic control is achieved and suggest that strict glycemic control might reduce the stillbirth rate in women with type 1 DM. However, as indicated by the area under the curve of the receiver operating characteristics curve (0.75, Fig. 1), a third-trimester Hb A1C value of 6.4% or more was not powerful enough in identifying all fetuses at high risk for abnormal nonstress test. Indeed, based on our results, one-third of pregnancies in women with type 1 DM would have been associated with an increased risk of fetal compromise by this criterion. Conversely, five of 18 urgent cesarean deliveries (28%) were performed in women whose Hb A1C level at delivery was less than 6.4%. In these five women, values of Hb A1C at delivery ranged from 5.6% to 6.3%, ie, above normal values.
Antenatal fetal testing is a part of the management of pregnancies in women with type 1 DM. However, available evidence is insufficient to determine the need for these tests and the optimal timing and frequency to perform them. Retrospective studies showed that frequent nonstress tests can identify fetuses at high risk of stillbirth and that anticipated delivery of these neonates may decrease the stillbirth rate.11,12 However, these studies included mostly women with gestational DM treated with insulin and their conclusions may not apply to women with type 1 DM. Other methods of fetal surveillance such as computerized analysis of fetal heart rate and ultrasonographic Doppler examination of the umbilical and uterine arteries have not been extensively studied in women with type 1 DM. A poorer diastolic function of the fetal right ventricle has been reported in women with poorly controlled pregestational diabetes,23 but this result needs further evaluation.
An important issue to consider is the timing of delivery because an increased risk of stillbirth at greater than 37 weeks of gestation among women with type 1 DM has been recently reported.2 The American Diabetes Association stated that: “an emerging consensus suggests that well-monitored diabetic women achieving excellent glycemic control without obstetric complications can await spontaneous labor up to 39–40 weeks gestation.”24 The National Institute for Health and Clinical Excellence guidance recommends that all women with diabetes are offered elective birth after 38 completed weeks of gestation.25 In our study, planned delivery at 38–39 weeks of gestation in women with no complications and carefully followed resulted in a stillbirth rate similar to that of the general population.
The practice of performing fetal surveillance in all women with type 1 DM remains controversial. It has been stated that when good glycemic control is achieved, in the absence of nephropathy, preeclampsia and abnormal fetal growth, fetal compromise is unlikely to occur.13 However, glycemic control was considered as good in 28% of the 18 pregnancies delivered by immediate or urgent cesarean. We thus suggest that twice-weekly nonstress tests from 32 weeks of gestation until planned delivery at 38–39 weeks of gestation remain a safe strategy in women with type 1 DM.
1. Persson M, Norman M, Hanson U. Obstetric and perinatal outcomes in type 1 diabetic pregnancies: a large, population-based study. Diabetes Care 2009;32:2005–9.
2. Eidem I, Vangen S, Hanssen KF, Vollset SE, Henriksen T, Joner G, et al.. Perinatal and infant mortality in term and preterm births among women with type 1 diabetes. Diabetologia 2011;54:2771–8.
3. Dunne FP, Avalos G, Durkan M, Mitchell Y, Gallacher T, Keenan M, et al.. ATLANTIC DIP: pregnancy outcomes for women with type 1 and type 2 diabetes. Ir Med J 2012;105(suppl):6–9.
4. Lauenborg J, Mathiesen E, Ovesen P, Westergaard JG, Ekbom P, Mølsted-Pedersen L, et al.. Audit on stillbirths in women with pregestational type 1 diabetes. Diabetes Care 2003;26:1385–9.
5. Teramo K, Kari MA, Eronen M, Markkanen H, Hiilesmaa V. High amniotic fluid erythropoietin levels are associated with an increased frequency of fetal and neonatal morbidity in type 1 diabetic pregnancies. Diabetologia 2004;47:1695–703.
6. Russell NE, Holloway P, Quinn S, Foley M, Kelehan P, McAuliffe FM. Cardiomyopathy and cardiomegaly in stillborn infants of diabetic mothers. Pediatr Dev Pathol 2008;11:10–4.
7. Philipps AF, Porte PJ, Stabinsky S, Rosenkrantz TS, Raye JR. Effects of chronic fetal hyperglycemia upon oxygen consumption in the ovine uterus and conceptus. J Clin Invest 1984;74:279–86.
8. Jelkmann W. Erythropoietin: structure, control of production, and function. Physiol Rev 1992;72:449–89.
9. Widness JA, Teramo KA, Clemons GK, Voutilainen P, Stenman UH, McKinlay SM, et al.. Direct relationship of antepartum glucose control and fetal erythropoietin in human type 1 (insulin-dependent) diabetic pregnancy. Diabetologia 1990;33:378–83.
10. Mathiesen ER, Ringholm L, Damm P. Stillbirth in diabetic pregnancies. Best Pract Res Clin Obstet Gynaecol 2011;25:105–11.
11. Kjos SL, Leung A, Henry OA, Victor MR, Paul RH, Medearis AL. Antepartum surveillance in diabetic pregnancies: predictors of fetal distress in labor. Am J Obstet Gynecol 1995;173:1532–9.
12. Brecher A, Tharakan T, Williams A, Baxi L. Perinatal mortality in diabetic patients undergoing antepartum fetal evaluation: a case-control study. J Matern Fetal Neonatal Med 2002;12:423–7.
13. Landon MB, Vickers S. Fetal surveillance in pregnancy complicated by diabetes mellitus: is it necessary? J Matern Fetal Neonatal Med 2002;12:413–6.
14. Lepercq J, Coste J, Theau A, Dubois-Laforgue D, Timsit J. Factors associated with preterm delivery in women with type 1 diabetes: a cohort study. Diabetes Care 2004;27:2824–8.
15. Lepercq J, Le Meaux JP, Agman A, Timsit J. Factors associated with cesarean delivery in nulliparous women with type 1 diabetes. Obstet Gynecol 2010;115:1014–20.
16. American College of Obstetricians and Gynecologists. Antepartum fetal surveillance. ACOG Practice Bulletin 9. Washington, DC: ACOG; 1999.
17. Lucas DN, Yentis SM, Kinsella SM, Holdcroft A, May AE, Wee M, et al.. Urgency of caesarean section: a new classification. J R Soc Med 2000;93:346–50.
18. Diagnosis and management of preeclampsia and eclampsia. ACOG Practice Bulletin No. 33. American College of Obstetricians and Gynecologists. Obstet Gynecol 2002;99:159–67.
19. Mamelle N, Munoz F, Martin JL, Laumon B, Grandjean H. Fetal growth from the AUDIPOG study: II. Application for the diagnosis of intrauterine growth retardation. J Gynecol Obstet Biol Reprod (Paris) 1996;25:71–7.
20. Jensen DM, Damm P, Moelsted-Pedersen L, Ovesen P, Westergaard JG, Moeller M, et al.. Outcomes in type 1 diabetic pregnancies: a nationwide, population-based study. Diabetes Care 2004;27:2819–23.
21. Teramo K, Ammälä P, Ylinen K, Raivio KO. Pathologic fetal heart rate associated with poor metabolic control in diabetic pregnancies. Obstet Gynecol 1983;61:559–65.
22. EURO-PERISTAT project, with SCPE, EUROCAT, EURONEOSTAT. European Perinatal Health Report, 2008. Available at: http://www.europeristat.com
. Retrieved September 12, 2012.
23. Wong SF, Chan FY, Cincotta RB, McIntyre HD, Oats JJ. Cardiac function in fetuses of poorly-controlled pre-gestational diabetic pregnancies—a pilot study. Gynecol Obstet Invest 2003;56:113–6.
24. Kitzmiller JL, Jovanovic L, Brown F, Coustan D, Reader DM. Managing preexisting diabetes and pregnancy: technical reviews and consensus recommendations for care. Alexandria (VA): American Diabetes Association; 2008. p. 585.
25. Guideline Development Group. Management of diabetes from preconception to the postnatal period: summary of NICE guidance. BMJ 2008;336:714–7.
Figure. No available...Image Tools