Maternal obesity and diabetes mellitus have both been implicated in the development of congenital defects. Several studies suggest that maternal obesity is associated with an increased risk of neural tube defect (NTD) outcomes, particularly spina bifida and anencephaly, 1–5 although not all studies are consistent. 6–8 Waller and colleagues 5 also found that obese women had higher reproductive risk of other central nervous system defects and defects of the great vessels, the ventral wall, and the intestines.
Obese women are at higher risk for type 2 diabetes and, during pregnancy, at higher risk of developing gestational diabetes. Beginning in the 1940s, there were reports that women with pre-existing diabetes had a higher risk of having a baby with an anomaly, but the particular pattern of defects found has varied. 9–17 Some animal and human studies reported an increase in rates of caudal dysgenesis, 12,18–20 although many of these studies were small or lacked appropriate concurrently collected control data. Defects of the cardiovascular, skeletal, and central nervous systems, including NTDs, were also frequently cited, 12,14–18,20 as well as renal and urogenital defects in some studies. 15,20
Some of these studies found similar cardiovascular, musculoskeletal, or central nervous system defects associated with gestational diabetes. 14,15 A recent case-control study also reported an increased risk of upper and lower spine and rib defects, limb defects, holoprosencephaly, and renal and urinary anomalies. 21 In general, the adverse effect of gestational diabetes has been much weaker than that associated with pre-existing diabetes.
We used data from a large prospective cohort of pregnant women to examine the individual effects of obesity and diabetes as well as the extent to which these factors might interact in the pathogenesis of nonchromosomal congenital defects.
Subjects and Methods
Between October 1984 and June 1987, 24,559 women from more than 100 obstetrics practices who underwent early second-trimester amniocentesis or alpha-fetoprotein screening studies were invited to participate in a prospective study of early pregnancy exposures and pregnancy outcome. Details of the sampling and data collection methods have been previously published. 22 For these analyses, we included 22,951 women with complete data on height, prepregnancy weight, all confounders of interest, and pregnancy outcome.
During the initial telephone interview, conducted between the 15th and 20th gestational week, each woman was asked about a number of potential exposures before and during the first trimester of the pregnancy, including her weight 3 months before conception and at the time of conception. As a part of the detailed medical and reproductive history, women were asked about their diabetes history, including the specific diagnosis and the date of onset. Each woman was also asked for a history of medication usage during the first 8 weeks of pregnancy, including the use of insulin and oral hypoglycemic medications. The interviewer asked for the exact drug name, how long the woman had been taking it, when she had first started the drug, and whether its use had been discontinued or changed during pregnancy.
At about the time of expected delivery, an outcome questionnaire was mailed to the delivering physician to ascertain details of any congenital anomalies, other defects, deaths or fetal losses, newborn complications, maternal illnesses, and other pregnancy complications such as gestational diabetes or pregnancy-induced hypertension. About 77% of physicians returned the completed questionnaire, whereas the rest were completed by the mother. When necessary, a nurse interviewer made a follow-up telephone call to the mother or physician for clarification of outcome information.
Each woman was assigned to one of the following categories of diabetes using national standards for diabetes classification: (1) no past or current diabetes, (2) type 1 diabetes, (3) type 2 diabetes, or (4) gestational diabetes only. 23 Women who developed diabetes at any time during pregnancy, regardless of their need for exogenous insulin, were classified as having gestational diabetes. Because the prevalence of type 1 diabetes is low and type 2 diabetes is usually seen in middle-aged and older adults, we had few women with either type 1 (N = 36) or type 2 (N = 32) diabetes. Therefore, we combined women in these two categories.
At least two independent coders examined all outcome questionnaires in which there was any report of a defect, fetal abnormality, or death. Coding of congenital defects was carried out using the 6-Digit Code List for Reportable Congenital Anomalies of the Centers for Disease Control and Prevention, 24 with coders blinded to the subject’s exposure information. In the event of disagreement, a third and, if necessary, a fourth coder was consulted. In cases of multiple defects, each defect was coded separately unless the defects constituted a known syndrome. Defects associated with birth trauma, infection, an existing chromosomal defect, and single gene defects were excluded. Finally, for these analyses, we considered only major, nonchromosomal anomalies. Major defects were those that resulted in death, disability, or substantial social stigma and those that required follow-up medical or surgical intervention.
Initially, we assigned each defect to one of the following categories: (1) NTD, (2) other neurologic, (3) cardiovascular, (4) craniofacial, (5) musculoskeletal, (6) gastrointestinal, (7) urogenital, (8) respiratory, (9) skin, and (10) other defects. On the basis of previous reports of defects associated with obesity and/or diabetes, we examined the following six defect categories for the current analyses: (1) NTD, (2) other neurologic, (3) craniofacial, (4) musculoskeletal, (5) urogenital, and (6) cardiovascular. Multiple defects within a category were counted only once, whereas two distinct defects in different categories were counted separately. Twelve babies had defects in more than one of these six categories.
We examined the effects of two primary exposure variables: body mass index (BMI) 3 months before conception [calculated as weight (kg) per height squared (m 2)] and diabetes before or during pregnancy. We first examined the prevalence of birth defects by category of BMI and diabetes. We then calculated prevalence ratios (PRs) and 95% confidence intervals (95% CIs) for the separate effects of obesity and diabetes and used multiple logistic regression to adjust for the potential confounding effects of maternal age, education, first-trimester cigarette use, alcohol intake, and mean supplemental folate and retinol intakes during weeks 3–8 of the pregnancy (calculated from last menstrual period). Finally, we examined the individual and joint effects of obesity and diabetes on birth defect outcome by calculating separate adjusted PR estimates within strata of obesity and diabetes.
Table 1 shows the characteristics of the 22,951 women by the presence of diabetes and level of BMI. Women with any form of diabetes tended to be somewhat older than those without, but otherwise, the two groups were similar. The more obese subjects were slightly older and somewhat less well educated than the nonobese. Multivitamin use during the first trimester was similar in most groups although slightly less frequent among the obese.
The prevalence of major defects by category of BMI and diabetes is shown in Table 2. Overall, the prevalence of any major defect was not very different in the three BMI categories (1.37%, 1.10%, and 1.47% for those with BMI <25, 25–<28, and ≥28 kg/m 2, respectively). There was no evidence of an excess risk of NTDs, other neurologic defects, urogenital defects, or cardiovascular defects, although both craniofacial and musculoskeletal defects were more frequent among the most obese women. Women with pre-existing diabetes had a much higher prevalence overall of offspring with major defects than did women without diabetes (5.88%vs 1.34%), and those with gestational diabetes had no appreciable excess of defects (1.38%). Although the one NTD case that occurred among the women with pre-existing diabetes was more than expected, no conclusion is possible from these small numbers. The prevalence of craniofacial and other musculoskeletal defects was increased among women with any diabetes history relative to that seen in women without diabetes.
Because the pregnancies of moderately overweight women (BMI 25–<28) in this study population did not result in an excess of offspring with major defects of any type, we combined women in the lower two BMI categories for all subsequent analyses. In Table 3, we see that the pregnancies of obese (BMI ≥28) women were about as likely to result in an offspring with a major defect than were those of nonobese women (PR = 1.1). The adjusted relative risk estimates for both craniofacial and musculoskeletal defects were higher among obese than nonobese women [PR = 2.2 (95% CI = 0.91–5.4) for craniofacial defects and PR = 1.5 (95% CI = 0.69–3.4) for other musculoskeletal defects]. The pregnancies of women with pre-existing diabetes were more than four times as likely to result in offspring with some form of anomaly (PR = 4.4; 95% CI = 1.6–12.1) compared with those of nondiabetic women; again, the category-specific relative risks were strongest for craniofacial and musculoskeletal defects. Women with gestational diabetes also had a 2.6-fold increased risk of musculoskeletal defects in their offspring (95% CI = 0.82–8.5).
In Table 4, we provide the prevalence of specific types of major defects according to the combined classification of obesity and diabetes. Because the number of defects was small in each category of diabetes and because the offspring of women with gestational diabetes had an increased risk of some types of major defects, we combined all types of diabetes for this analysis. The pregnancies of 6 of the 453 (1.32%) nonobese women with diabetes resulted in offspring with a major defect. These 6 babies had a total of seven defects, one each of the following types: NTD, lower limb reduction, club foot, preaxial polydactyly, amniotic bands (resulting in a deformation), genital defect, and congenital neoplasm. Among the 1,853 obese women without diabetes, 24 had an offspring with a major defect (1.30%). While the prevalence of defects overall was not elevated for women with diabetes only or with obesity only [relative to the prevalence in nonobese, nondiabetic women (1.34%)], certain types of defects were more common. For example, obese, nondiabetic women more frequently had offspring with hydrocephaly, orofacial clefts, club foot, abdominal wall defects, and cardiac defects of septal closure. Finally, the offspring of women who were both obese and diabetic had a higher risk overall (4.13%) of having a major defect.
To ascertain whether the apparent excess risk of defects among obese and/or diabetic women might be attributable to confounding (for example, by older ages at the time of conception), we used multiple logistic regression analysis to adjust for maternal age, education, first-trimester smoking, alcohol use, and the intake of supplemental folate per day in weeks 3–8 of the pregnancy. Overall, as shown in Table 5, there was no increased risk of major defects in the offspring of obese women without diabetes (PR = 0.95; 95% CI = 0.62–1.5) and no increased risk among diabetic women without obesity (PR = 0.98; 95% CI = 0.43–2.2). The offspring of women who were both obese and diabetic were about three times as likely (PR = 3.1; 95% CI = 1.2–7.6) to have a major defect.
Because women with diabetes and/or obesity may be more likely to have offspring with certain types of defects, particularly craniofacial and musculoskeletal defects, we then combined these two categories of defects. In this analysis, the pregnancies of obese women without diabetes were 60% more likely (95% CI = 0.80–3.0) to result in offspring with a craniofacial or musculoskeletal defect, whereas those of nonobese women with diabetes were 1.8 times as likely (95% CI = 0.57–5.8) to result in such a defect. Women who were both obese and diabetic had a sevenfold increased risk (PR = 7.0; 95% CI = 2.1–22.7) of a craniofacial or musculoskeletal defect in their offspring than were nonobese, nondiabetic women.
Although obese women without diabetes in this study had no excess risk of total anomalies in their offspring, they did have an increased frequency of some specific types of defects, including orofacial clefts; club foot; cardiac septal defects; and, to a lesser extent, hydrocephaly and abdominal wall defects. The offspring of diabetic women (and gestationally diabetic women) who were not obese had an excess of musculoskeletal defects. These results are supportive of some earlier studies, 12,13–15,21 but our results differ from these studies in finding an even higher risk of major nonchromosomal defects among the offspring of women who were both obese and diabetic. We did not find any excess of multiple anomalies in the offspring of obese or diabetic women in this study.
To quantify the potential synergistic effect of obesity and diabetes, we calculated the proportion of disease among those with both exposures that is attributable to the interaction 25 and found that approximately 65% of total major defects among women who were obese and had some form of diabetes was attributable to the interaction of the two factors.
One earlier case-control study concluded that the effect of obesity was independent of diabetes 3; these authors found a 90% increased risk of NTDs associated with obesity after excluding 11 women with known diabetes. Because many NTD cases may die or be terminated early in a pregnancy at a time when gestational diabetes is unlikely to have been diagnosed, however, it is difficult to control for the possible contribution of an underlying (and undetected) metabolic defect associated with gestational diabetes in these analyses. We were also likely to have incomplete ascertainment of gestational diabetes in the mothers of our NTD cases, given that most NTD cases in our dataset resulted in an elective abortion. Nevertheless, because we included women with a history of gestational diabetes in an earlier pregnancy (even if the current pregnancy did not progress to the point of recurrent gestational diabetes) in our exposure group, we may have had more ability to detect an effect of an underlying metabolic abnormality.
The women in the current study differ in some ways from women in other studies. 26 Overall, about 70% of these women had some college education (compared with about 20–50% of women in other studies). 3–5 Although obese women are less likely in general to be college educated than nonobese women (58%vs 71%, respectively, in this sample), the higher levels of education in our study population overall may point to other exposure differences for women in this study (for example, different dietary habits, different patterns of vitamin use, or differences in other unmeasured factors affecting the development of congenital defects). Although these differences could account for such things as different defect rates across studies, they do not affect the validity of the internal comparisons in this study.
There is always the possibility that some bias in reporting or observation might explain the results observed. Given the prospective design of this study, it is unlikely that biased reporting of body weight or diabetes history occurred. The reporting of gestational diabetes at the end of the pregnancy could have been biased by the occurrence of an adverse outcome, however. Although we cannot rule this out entirely, the fact that there was not a general perception, especially in the 1980s, that gestational diabetes was associated with the occurrence of birth defects argues against this possibility.
Observation bias might also be a concern in this study if, for example, the physicians had looked more closely for defects among the offspring of women who were obese or who had diabetes. Many of the defects that were seen in excess among obese and/or diabetic women, however, were defects that are unlikely to be missed by the delivering physician (for example, cleft lip and palate or lower limb defects).
The ascertainment of defects in this study was confined to those observed at birth or shortly thereafter. The follow-up questionnaire was sent to the physician within a few weeks of the expected delivery time. There was some variability in the time of the mailing, and not all physicians returned the questionnaire immediately. Routine follow-up was done to increase compliance, and this follow-up was done without any knowledge of the woman’s exposure history. The absence of ongoing defect surveillance in this cohort has probably also led to an overall underascertainment of anomalies, particularly for those defects that are more difficult to detect at birth. Thus, we are likely to have more complete ascertainment of musculoskeletal defects than cardiovascular defects. In terms of bias, however, it is unlikely that the completeness of defect ascertainment differed for obese and nonobese women or for those with and without a history of diabetes.
The mechanism by which diabetes may be associated with the development of fetal defects has been the subject of a number of investigations. In some 27–29 but not all 30 studies, the degree of glucose control has been inversely correlated with birth defect risk. If glucose control is causally related to the development of many of the defects seen in this study, the degree of control early in the pregnancy would be of primary concern. 31 Women with pre-existing diabetes may have significant fluctuations in circulating glucose and other factors; however, women who develop gestational diabetes later in pregnancy may also have had undetected metabolic problems earlier in the pregnancy. In addition, obese women, even in the absence of overt diabetes, have been found to have impaired glucose metabolism. 32
Other mechanisms, such as an underlying vascular pathology, 16 could also explain the increased risk of defects in offspring of diabetic women. Those who are also obese may have different dietary habits from nondiabetic, nonobese women; thus, nutritional differences might also explain some of the adverse effects of these factors. In this study, obese women were more likely to lose weight in the 3 months before conception than were nonobese women (6.0% of women with a BMI ≥28, 3.6% with a BMI ≥25 but <28, and 1.0% of those with a BMI <25 lost more than 5 pounds in the 3 months before pregnancy). Voluntary weight loss itself could lead to inadequate nutrient intake during a critical period of development for the fetus, even before a woman knew that she was pregnant. We controlled for the first-trimester intakes of folate and vitamin A in these analyses, but we cannot rule out the possibility that obese and/or diabetic women had other differences in dietary intake that might affect fetal development.
In our multivariate analyses, we combined defects into broad categories that may be etiologically dissimilar. There is no uniform classification scheme for birth defects in the existing literature, and different studies have implicated different types of defects within a category in association with obesity and diabetes. Thus, in our multivariate models, we combined defects into broad categories (for example, musculoskeletal or cardiovascular). This broad classification would most likely lead to an underestimation of the true effects of these factors on more specific types of defects. We found that the joint exposure to obesity and diabetes led to a threefold increased risk of all major defects. Thus, with even the broadest classification of defects, there was evidence that these two factors acted together to promote the development of congenital anomalies.
We thank the knowledgeable and helpful staff of the Boston Collaborative Drug Surveillance Program, who oversaw the initial data collection, and Agostino Visioni and Di Gao for their careful and skillful data management.
1. Källén K. Maternal smoking, body mass index, and neural tube defects. Am J Epidemiol 1998; 147: 1103–1111.
2. Werler MM, Louik C, Shapiro S, Mitchell AA. Prepregnant weight in relation to risk of neural tube defects. JAMA 1996; 275: 1089–1092.
3. Shaw GM, Velie EM, Schaffer D. Risk of neural tube defect-affected pregnancies among obese women. JAMA 1996; 275: 1093–1096.
4. Watkins ML, Scanlon KS, Mulinare J, Khoury MJ. Is maternal obesity a risk factor for anencephaly and spina bifida? Epidemiology 1996; 7: 507–512.
5. Waller DK, Mills JL, Simpson JL, Cunningham GC, Conley MR, Lassman MR, Rhoads GG. Are obese women at higher risk for producing malformed offspring? Am J Obstet Gynecol 1994; 170: 541–548.
6. Haddow JE, Palomaki GE. Is maternal obesity a risk factor for open neural tube defects? Am J Obstet Gynecol 1995; 172: 245–246.
7. Wald N, Cuckle H, Boreham J, Terzian E, Redman C. The effect of maternal weight on maternal serum alpha-fetoprotein levels. Br J Obstet Gynaecol 1981; 88: 1094–1096.
8. Johnson AM, Palomaki GE, Haddow JE. The effect of adjusting maternal serum α-fetoprotein levels for maternal weight in pregnancies with fetal open spina bifida. Am J Obstet Gynecol 1990; 163: 9–11.
9. White P. Pregnancy complicating diabetes. Am J Med 1949; 7: 609–616.
10. Mølsted-Pedersen L, Tygstrup I, Pedersen J. Congenital malformations in newborn infants of diabetic women. Lancet 1964; 1: 1124–1126.
11. Greene MF. Prevention and diagnosis of congenital anomalies in diabetic pregnancies. Clin Perinatol 1993; 20: 533–547.
12. McCarter RJ, Kessler II, Comstock GW. Is diabetes mellitus a teratogen or coteratogen? Am J Epidemiol 1987; 125: 195–205.
13. Becerra JE, Khoury MJ, Cordero JF, Erickson JD. Diabetes mellitus during pregnancy and the risk for specific birth defects: a population-based case-control study. Pediatrics 1990; 85: 1–9.
14. Ramos-Arroyo MA, Rodriguez-Pinilla E, Cordero JF. Maternal diabetes: the risk for specific birth defects. Eur J Epidemiol 1992; 8: 503–508.
15. Janssen PA, Rothman I, Schwartz SM. Congenital malformations in newborns of women with established and gestational diabetes in Washington State, 1984–1991. Paediatr Perinat Epidemiol 1996; 10: 52–63.
16. Miodovnik M, Mimouni F, Dignan PSJ, Berk MA, Ballard JL, Siddiqi TA, Khoury J, Tsang RC. Major malformations in infants of IDDM women. Vasculopathy and early first-trimester poor glycemic control. Diabetes Care 1988; 11: 713–718.
17. Mølsted-Pedersen L, Pedersen JF. Congenital malformations in diabetic pregnancies. Acta Pediatr Scand 1985; S320: 79–84.
18. Eriksson U, Dahlstrom E, Larsson KS, Hellerstrom C. Increased incidence of congenital malformations in the offspring of diabetic rats and their prevention by maternal insulin therapy. Diabetes 1982; 31: 1–6.
19. Martínez-Frías ML. Epidemiological analysis of outcomes of pregnancy in diabetic mothers: identification of the most characteristic and most frequent congenital anomalies. Am J Med Genet 1994; 51: 108–113.
20. Kučera J. Rate and type of congenital anomalies among offspring of diabetic women. J Reprod Med 1971; 7: 61–70.
21. Martínez-Frías ML, Bermejo E, Rodriguez-Pinilla, Prieto L, Frías JL. Epidemiological analysis of outcomes of pregnancy in gestational diabetic mothers. Am J Med Genet 1998; 78: 140–145.
22. Milunsky A, Jick H, Jick SS, Bruell CL, MacLaughlin DS, Rothman KJ, Willett W. Multivitamin/folic acid supplementation in early pregnancy reduces the prevalence of neural tube defects. JAMA 1989; 262: 2847–2852.
23. The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 1997; 20: 1183–1196.
24. Centers for Disease Control and Prevention. CDC 6-digit code list for reportable congenital anomalies. In: Metropolitan Atlanta Congenital Defects Program: Surveillance Procedure Manual and Guide to Computerized Anomaly Record. Atlanta: Centers for Disease Control and Prevention, 1998.
25. Rothman KJ. Interactions between causes. In: Modern Epidemiology. Boston: Little, Brown and Company 1986;321.
26. Rothman KJ, Moore LL, Singer MR, Nguyen UDT, Mannino S, Milunsky A. Teratogenicity of high vitamin A intake. N Engl J Med 1995; 333: 1369–1373.
27. Ballard JL, Holroyde J, Tsang RC, Chan G, Sutherland JM, Knowles HC. High malformation rates and decreased mortality in infants of diabetic mothers managed after the first trimester of pregnancy. Am J Obstet Gynecol 1984; 148: 1111–1118.
28. Mironiuk M, Kietlińska Z, Jezierska-Kasprzyk K, Pi¸ekosz-Orzechowska B. A class of diabetes in mother, glycemic control in early pregnancy and occurrence of congenital malformations in newborn infants. Clin Exp Obstet Gynecol 1997; 24: 193–197.
29. Kitzmiller JL, Gavin LA, Gain GD, Jovanovic-Peterson L, Main EK, Zigrang WD. Preconception care of diabetes. Glycemic control prevents congenital anomalies. Diabetes Care 1991; 265: 731–736.
30. Mills JL, Knopp RH, Simpson JL, Jovanovic-Peterson L, Metzger BE, Holmes LB, Aarons JH, Brown Z, Reed GF, Bieber FR, Van Allen M, Holzman I, Ober C, Peterson CM, Withiam MJ, Duckles A, Mueller-Heubach E, Polk BF, and the National Institutes of Child Health and Human Development Diabetes in Early Pregnancy Study. Lack of relation of increased malformation rates in infants of diabetic mothers to glycemic control during organogenesis. N Engl J Med 1988; 318: 671–676.
31. Mills JL, Baker L, Goldman AS. Malformations in infants of diabetic mothers occur before the seventh gestational week. Implications for treatment. Diabetes 1979; 28: 292–293.
32. Bonadonna RC, Defonzo RA. Glucose metabolism in obesity and type II diabetes. In: Bjorntorp P, Brodoff BN, eds. Obesity. Philadelphia: JB Lippincott Company, 1992; 474–501.