DISCUSSION

From 1991 to 2004, in our obstetric population there was a nearly 15-lb increase in mean maternal weight at the first prenatal visit and greater than a 30% increase in the proportion of women with a BMI greater than 29 kg/m². During this time, we also noted a nearly twofold increase in the rate of major congenital anomalies. There was not, however, an independent association between maternal obesity and major congenital anomalies. Interestingly, we saw a 250% increase in the prevalence of pregestational diabetes during this time period, and its presence, either with or without associated obesity, seems to be a major contributor to the increasing rate of anomalies in our population.

Our results, derived from an inner-city population receiving primary obstetric care in our facilities, differ from many other reports in the literature. Naeye was the first to report an increase in the prevalence of congenital malformations from 4.0% in thin patients to 5.5% in obese patients. Although the rate of congenital anomalies was 4.5% in women of normal weight, this group was not compared with the obese women. Furthermore, those data were derived from a cohort followed up from 1959–1966; the rates and management of diabetes have changed dramatically since that time, and the contribution of diabetes to the occurrence of congenital anomalies was not considered. Cedergren and Ka"lle’n reported a 30% increase in orofacial clefts in the offspring of obese Swedish women; however, diabetic status was not considered in their analysis. Waller et al were the first to report an association between maternal weight and neural tube defects. In this case–control study, they reported an OR of 1.8 for a neural tube defect, and several other groups have reported a similar magnitude of risk. An increased risk of cardiac defects in obese women without diabetes has also been reported in several case–control studies, but varying definitions of obesity have been used. Excluding women with known preexisting diabetes, Watkins et al reported an increased risk of not only neural tube defects and heart defects, but also omphalocele and multiple anomalies in the Atlanta Birth Defects Risk Factor Surveillance Study. Although the point estimates are similar, analyses were based on maternal interview after delivery and were therefore susceptible to selection and recall bias. Two recent meta-analyses demonstrated small obesity-associated increases in the risk of neural tube defects, some cardiac defects, and cleft lip; however, many of the primary sources evaluated in these systematic reviews either excluded people with diabetes or based the classification only on maternal recall. Furthermore, several lumped people with pregestational and gestational diabetes together for the purpose of analysis. No assessment of the magnitude of the contribution of diabetes to the occurrence of fetal anomalies in the setting of obesity was provided.

In one of the few prospective, population-based cohort studies examining this issue, Moore et al examined risk factors and outcomes of nearly 23,000 women undergoing midtrimester maternal serum screening. These investigators interviewed women during the second trimester of pregnancy regarding medical history and risk factor exposure. Infant outcomes were obtained from physician and patient questionnaires. In the absence of diabetes, obese women had no higher risk of having offspring with a major defect, but women who had diabetes and obesity had a threefold risk compared with women who had neither. These findings parallel ours in that the risk in obese women is concentrated in those with diabetes. More recently, a case–control study by Shaw and Carmichael of more than 650 cases of neural tube defects and heart defects failed to demonstrate an association with obesity.

Although one of the goals of our study was to examine temporal changes in the incidence of anomalies and obesity, standard medical practice also changed over the time period of investigation. In the early time periods, BMI was not routinely recorded; therefore, this parameter could not be reliably used or calculated. We relied instead on an alternative definition of obesity for our primary analysis. However, results were similar when only women with available BMI were included; thus, we believe that our definition allows for reasonable representation of this group of women.

Changes also occurred in the reliability with which congenital anomalies are diagnosed prenatally. Ultrasound technology improved dramatically from 1991 to 2004, so it is possible that even major congenital anomalies could have been missed more in the early time periods. However, a recent study by Dashe et al showed a persistently lower detection rate for fetal anomalies in obese women even during the past 6 years. To minimize a potential for ascertainment bias, we limited our investigation to congenital anomalies that would be significant enough to necessitate treatment in the neonatal period or would be recognized in the neonatal period. Furthermore, we included all neonates who had defects diagnosed in the neonatal period and did not limit inclusion to only infants with prenatally diagnosed anomalies. With prenatal ultrasound findings and an examination of each child by a pediatrician, neonatologist, or both before discharge, it is unlikely that major congenital malformations would go undetected. Prenatal ultrasound findings were particularly important in ascertainment of visceral malformations, such as cystic adenomatoid malformations of the lung, which might otherwise be undetected during a neonatal examination. The prevalence of the anomalies we report was similar to the prevalence reported in the EUROCAT survey, a European congenital anomaly surveillance program, for live births and fetal deaths. The relatively low prevalence of anomalies in our study is likely a reflection of limiting our analysis to patients delivered beyond 20 weeks of gestation. Despite the low prevalence of some types of anomalies, given our sample size, a post hoc power analysis demonstrated that we had greater than 90% power to detect a doubling in the prevalence of anomalies in obese women compared with the rates seen in nonobese women.

Among the strengths of our clinical database study are the size of the study group and the availability of individual patient clinical data that are not subject to recall bias. One of the weaknesses of this type of study, however, is the limitation of investigation to data points that were collected during the course of clinical care. In our case, detailed data regarding individual patient glycemic control are not available in the database, and therefore we cannot comment on the association between glycemic control and anomaly rates.

Our findings are noteworthy for a number of reasons. First, although much attention has been given to population-based, epidemiologic data linking maternal obesity with all birth defects, including some organ-specific defects, the pathophysiologic basis and underlying cause of the association has been poorly investigated because of limited data. Lower levels of folic acid, increased serum insulin, chronic hypoxia, and increased inflammatory mediators have all been postulated to contribute to the occurrence of malformations. Our study provides evidence that the defects may not be due solely to the maternal obesity per se but may be due to undiagnosed diabetes. Second, from a public health standpoint, these findings highlight the need to consider refocusing attention on interventions that can best decrease the rate of adverse pregnancy outcomes. Although obesity alone has been associated with myriad poor outcomes, the coexistence of diabetes further increases the risk. Interventions to decrease the prevalence of maternal obesity require high levels of patient motivation, and goal attainment is often delayed. In contrast, diagnosis and treatment of diabetes can readily achieve desired aims as long as a patient is reasonably compliant. In our population, during the 2000–2004 period, more than 9% of all major congenital anomalies and 71% of those in obese women were attributable to diabetes. If euglycemia could be achieved before pregnancy, or at least embryogenesis and organogenesis, the majority of these anomalies could potentially be avoided. This suggests a role not only for weight-reduction strategies, but also consideration of screening for diabetes in obese women contemplating pregnancy and at presentation for antenatal care in those already pregnant. Moreover, given the lack of an independent effect of obesity, it may be that even women who are not overtly obese, but have other risk factors for diabetes, may be at increased risk for fetal anomalies and could benefit from attention to glycemic control.

We conclude that the association between obesity and fetal anomalies may be due to underlying diabetes, which if uncontrolled can lead to hyperglycemia. Because hyperglycemia is a major contributor to developmental malformations, interventions to address obesity and identify women at risk for diabetes and hyperglycemia should be considered in efforts to reduce the occurrence of congenital anomalies.