Obesity has been cited as the leading health problem confronting women (ACOG News Release. “Female Ob-Gyns name obesity the greatest threat to women’s health.” December 9, 2003. Available at: www.acog.org/from_home/publications/press_releases/nr12-09-03-1.cfm.). Over the past two decades, the prevalence of obesity has dramatically increased, and it now affects more than one in three women in the United States.1 Obesity has the potential to affect prenatal diagnosis in several ways. Even in the absence of diabetes, maternal obesity may increase the likelihood that an infant will be born with a major structural abnormality, specifically a neural tube defect, heart defect, or omphalocele, or that an infant will be born with multiple anomalies.2 Given the association between obesity and diabetes mellitus, with diabetes further increasing the likelihood of fetal anomalies, these risks may be compounded.3
In addition, obesity may impair visualization of fetal anatomy. Improvements in ultrasound technology, such as tissue harmonics and multi-Hertz transducer technology, have permitted clearer imaging of fetal structures farther from the transducer. However, modern ultrasound equipment has not been able to completely overcome the suboptimal visualization that results in the setting of maternal obesity, with visualization of the fetal heart considered to be a particular challenge.4,5 Whether imaging of fetal anatomy is so limited as to result in poorer detection of anomalous fetuses would be important information to have when counseling patients about the realistic expectations of the ultrasound examination.
Our objective was to estimate the effect of maternal habitus on the detection of fetuses with major structural anomalies during second-trimester standard and targeted ultrasound examinations, using body mass index (BMI) at the first prenatal visit to categorize women as either normal weight, overweight, or obese. It was our intention to evaluate whether fetuses with anomalies would come to attention, realizing that once an abnormality was recognized, additional specialized studies would be performed and neonatal intensive care unit personnel would be present at delivery if needed—neither of which would necessarily occur had an abnormality not been identified. For counseling purposes, a secondary goal was to estimate the likelihood of delivering an infant with a major anomaly if no abnormality had been detected during standard or targeted ultrasound examination and whether this residual anomaly risk would be affected by maternal obesity or diabetes mellitus.
MATERIALS AND METHODS
This was a retrospective cohort study of singleton pregnancies at 18 0/7 weeks through 23 6/7 weeks of gestation that received standard or targeted ultrasonography between January 1, 2003, and December 31, 2007, and delivered liveborn or stillborn infants at our hospital. The study was approved by the Institutional Review Board of the University of Texas Southwestern Medical Center. Body mass index was recorded at the first prenatal visit. Only the findings from the initial ultrasound examination were included, with the rationale that if no abnormalities were detected, women would not be consistently scheduled for further ultrasonographic evaluation before 24 weeks. Standard ultrasonography was performed according to American Institute of Ultrasound in Medicine (AIUM) criteria.6 For study purposes, low-risk pregnancies that did not have a specific indication for targeted ultrasonography were placed in the standard ultrasound group, because targeted ultrasonography would only take place if an abnormality was detected during the standard examination. Also, pregnancies with a specific high-risk indication were placed in the targeted ultrasound group, because targeted ultrasonography would occur regardless of findings during a standard examination. Indications for targeted ultrasonography included pregestational diabetes, family history indicating increased risk for fetal anomaly, exposure to a known or suspected teratogenic agent, maternal serum alpha fetoprotein level exceeding 2.5 multiples of the median, or multiple-marker maternal serum screening that indicated increased risk for trisomy 18. Pregnancies at increased risk for trisomy 21 were offered standard ultrasonography, along with formal genetic counseling and invasive prenatal diagnosis. Targeted ultrasonography was performed whenever an abnormality was suspected during standard ultrasonography in an otherwise low-risk patient; however, because these pregnancies were already in the standard ultrasound group, their results were analyzed separately.
Our county hospital serves a large, medically indigent population. Standard ultrasound examinations are performed by registered diagnostic medical ultrasonographers (American Registry of Diagnostic Medical Sonographers–certified), typically within neighborhood prenatal clinics, with images transmitted electronically to our hospital-based central ultrasonography unit and over read by faculty who specialize in obstetric ultrasonography. This is done in accordance with AIUM Practice Accreditation Standards, eg, a mechanism is in place to address unexpected or emergency findings, a call-back mechanism is in place for patients who leave before the physician reviews the images, and examinations are read within 24 hours. A targeted ultrasonography includes all the components of a standard examination, along with a more detailed evaluation of fetal anatomy that is determined on a case-by-case basis according to the risk factor or other indication for the specialized examination. All targeted examinations are performed within this central unit by faculty from maternal–fetal medicine or radiology with experience and expertise in prenatal diagnosis of fetal anomalies. All of our ultrasound sites are AIUM-accredited.
Anomalies were verified using a prospectively maintained database in which birth defects are recorded upon infant discharge, upon delivery if the infant is stillborn, or upon neonatal death. Stillbirth was defined as fetal death with birth weight of at least 500 grams. An anomalous fetus was considered detected if an abnormality of the relevant organ system was identified. All structural abnormalities potentially life-threatening or requiring surgery were included, regardless of their anticipated detection with either standard or targeted ultrasonography, with the exception of patent foramen ovale and atrial septal defect. For study purposes, major malformations were considered in the same context whether they were isolated or were components of a syndrome. Body mass index was calculated as the weight in kilograms at the first prenatal visit divided by height in meters squared. It was categorized using criteria defined by the World Health Organization and National Institutes of Health: normal less than 25 kg/m2; overweight 25–29.9 kg/m2; class I obesity 30–34.9 kg/m2; class II obesity 35–39.9 kg/m2; and class III obesity 40 kg/m2 or more.7,8 Statistical analyses were performed using SAS 9 (SAS Institute Inc., Cary, NC). Statistical tests included χ2 and Mantel-Haenszel χ2 test for trend.9 Results at P<.05 were considered significant.
Of 11,135 singleton pregnancies that received ultrasonography between 18 and 24 weeks, there were 10,112 standard ultrasound examinations in low-risk pregnancies (91%), and 1,023 targeted examinations performed for a priori indications in high-risk pregnancies (9%). Women in the standard ultrasound group had an average age (mean±standard deviation) of 27.2±6.7 years, 32% were nulliparous, and their ethnicity distribution was as follows: 86% Hispanic, 10% African American, and 2% white, with 2% being of other ethnicity. Women in the targeted/high-risk ultrasound group had an average age of 28.4±5.9 years, 25% were nulliparous, and their ethnicity distribution was 87% Hispanic, 9% African American, and 2% white, with 2% being of other ethnicity.
The number of women in the standard and targeted ultrasound groups is shown in Table 1, according to BMI category. An additional 75 women in the standard ultrasound group received targeted ultrasonography for identified anomalies but are not included in this table, so that each pregnancy is included only once. Only 39% of women had a BMI in the normal range, whereas 34% were overweight and 27% were obese. More women were obese in the targeted ultrasound group than in the standard ultrasound group, 37% compared with 27%, P<.001. Within the targeted ultrasound group, 261 examinations (26%) were performed for the indication of pregestational diabetes.
Also presented in Table 1 is the number of pregnancies with anomalous fetuses, according to BMI category. Anomalies were verified in 181 infants (1.6% of births), 141 in the standard ultrasound group (1.4%) and 40 in the group with a priori indications for targeted ultrasonography (3.9%), P<.001. Within the targeted group, fetuses of mothers with pregestational diabetes were more likely to be anomalous than fetuses whose mothers had another indication for the examination, 16 of 261 (6.1%) compared with 24 of 762 (3.1%), respectively, P=.03.
Low-risk pregnancies in which fetal anomalies were detected with standard ultrasonography did receive targeted ultrasound examinations, and if these fetal anomalies were included with the group receiving targeted examinations for a priori indications, the prevalence of anomalous fetuses in the targeted group increased to 10.5% (115 of 1,098). When pregnancies with fetal anomalies detected during standard ultrasonography were considered in the targeted group, they comprised 6.8% of pregnancies receiving targeted ultrasonography and 73.5% of pregnancies in which fetal anomalies were detected prenatally.
Detection of anomalous fetuses at the initial examination, ie, sensitivity of the ultrasound examination for anomaly detection, was 53% (95% confidence interval [CI] 45–62%) in the standard ultrasound group and 68% (95% CI 51–81%) in the group with a priori indications for targeted ultrasonography. Presented in Figure 1 is the percentage of anomalous fetuses who survived to hospital discharge, according to whether their anomalies were detected during the initial ultrasound examination. There were 11 stillborn infants and 16 neonatal deaths. Anomalies resulting in fetal or neonatal death were more likely to be detected ultrasonographically. Overall neonatal survival rate was 75% if a major anomaly was identified during an ultrasound examination, but was 99% if a major anomaly was present but not noted during the examination, P<.001.
Figure 2 depicts the detection of anomalous fetuses according to BMI category and type of examination. Among low-risk pregnancies, there was a significant decrease in the standard ultrasound detection of anomalous fetuses with increasing BMI, from 66% in women of normal BMI, just under 50% in overweight women, 48% with class I obesity, 42% with class II, and only 25% with class III, P for trend=.03. In high-risk pregnancies with a priori indications for targeted ultrasonography, detection of anomalous fetuses ranged from 83% with normal BMI to 67% with class III obesity, a difference that did not achieve statistical significance. Within the group that received targeted ultrasonography, detection of anomalous fetuses was significantly lower among women with pregestational diabetes than in women with another indication (eg, family history, teratogen exposure, abnormal serum screening), 38% (95% CI 15–65%) compared with 88% (95% CI 68–97%), P<.001, with no significant difference according to BMI. When all targeted ultrasound examinations were evaluated, including those performed for a priori indications and those performed for abnormalities noted during standard examination, there was also a significant decrease in anomaly detection with increasing BMI, ranging from 97% with normal BMI, 91% for overweight women, 75% with class I obesity, 88% for class II, and 75% with class III, P for trend=.02.
To facilitate counseling about the limitations of ultrasonography with increasing habitus, we calculated the residual risk of a major anomaly if none was noted during the initial ultrasound examination. Of the 11,135 pregnancies evaluated, there were 79 fetuses with undiagnosed anomalies, 0.7%. The specific anomalies that were not detected are listed in Table 2 and are shown according to type of examination, maternal BMI category, and presence of diabetes. The table shows that a broad range of anomalies were not detected, some relatively straightforward to identify (endocardial cushion defects), others more challenging (micrognathia), others not consistently identified at 18 to 24 weeks (gastrointestinal atresias), and still others not typically considered detectable (absent ear canal). However, even in the standard ultrasound group, among women of normal BMI, fetuses with the following relatively complex anomalies were detected: ambiguous genitalia, craniosynostosis, vermian agenesis, coarctation of the aorta, double-outlet right ventricle, heterotaxy, and dilated right ventricle with pulmonary insufficiency.
As is shown in Figure 3, the overall residual anomaly risk increased from 0.4% among women of normal BMI to 1.0% among obese women, P for trend=.001. In other words, the specificity of the ultrasound examination was 99.6% for women of normal BMI but only 99% for those with obesity. For both the standard ultrasound group and those who received targeted ultrasonography for a high-risk indication, residual anomaly risk increased significantly with increasing BMI, P for trend <.05. When the targeted ultrasound group was further subdivided according to whether maternal diabetes was present, those without diabetes also had a significantly increased residual anomaly risk with increasing BMI, P<.001. Among women with diabetes, who had a greater prevalence of anomalies as well as lower anomaly detection (shown above), the residual anomaly risk was approximately 4% and did not vary significantly according to maternal BMI category. The residual anomaly rate among women with diabetes was higher than in each of the other groups, all P<.01.
In assessing residual anomaly risk, we did not consider anomalies identified during follow-up ultrasound examinations, because we did not routinely perform additional ultrasound examinations. There were 13 pregnancies with high-risk indications in whom fetal anomalies were undetected during the initial ultrasonography, and in 10 (77%), ultrasonography was performed later in pregnancy, resulting in identification of three additional anomalous infants. The detection of anomalous infants on follow-up examination among women with diabetes was 38% (three of eight cases). In the 66 low-risk pregnancies with undetected fetal anomalies, 21 received ultrasonography later in pregnancy (32%), and a major anomaly was identified in 14% of cases. The detection of these six additional anomalies increased the overall anomaly detection rate by 4%, to 60%.
There are two major findings from this study. The first is that maternal habitus significantly affects the ultrasound detection of anomalous fetuses between 18 and 24 weeks of gestation. With increasing BMI, there was also a significant increase in the residual anomaly risk after a normal standard or targeted ultrasound examination. Because women of normal BMI now comprise a minority of our pregnant population, such findings may have broad implications. Second, detection of anomalous fetuses was lower in pregnancies complicated by pregestational diabetes, all of whom received targeted ultrasonography. This, coupled with the higher prevalence of anomalies in diabetic pregnancies, resulted in a much higher residual anomaly risk after normal ultrasonography in diabetic women.
A strength of this study is that it was performed in a nonreferred population, so we were able to assess the birth prevalence of anomalous infants as well as their detection. Both were comparable to other reports. In one review of 36 ultrasound studies comprising over 900,000 pregnancies, the average anomaly prevalence was 2%, similar to what we identified.10 The review noted that 75% of fetal anomalies occur in otherwise low-risk pregnancies, also similar to our experience (78%). In an overview of the European experience with routine ultrasonography, Levi11 found an average anomaly detection rate of 55% in regular risk and 92% in high-risk patients. These percentages are comparable to our data—53% and 89%, respectively. The prevalence of overweight and obese women in our study (61%) was also the same as has been reported for U.S. women overall in the most recent National Health and Nutrition Examination Survey.12
Several limitations of our study may be noted. We included only liveborn and stillborn infants, not terminated, or aborted (miscarried) pregnancies, because outcome data for such cases would be incomplete in our system. Termination of anomalous pregnancies can vary considerably between and even within centers. In our hospital, they represent a small minority of fetuses with anomalies, due to cultural and religious preferences, which may mean that our detection rates are slight underestimates, because terminated pregnancies are those with the most severe fetal abnormalities in our system. A related limitation is that because fetuses with detected anomalies were less likely to survive to hospital discharge, the residual anomaly risks we calculated may represent risks for anomalies that are less severe. In addition, even with a sample size of more than 11,000 pregnancies, it was not possible for us to draw conclusions about specific anomalies more or less likely to be detected in the setting of obesity or diabetes. The list of anomalies we did not detect (Table 2) includes some that are usually detectable with ultrasonography, such as endocardial cushion defects, and some that most would not consider to be reliably detectable—particularly before 24 weeks—eg, optic nerve atrophy, absent ear canal, hemidiaphragm paralysis, or imperforate anus. We did this for two reasons: 1) there is no recognized list of what is reasonably detectable with ultrasonography, given ongoing improvements in technology and potential medical–legal implications, and perhaps more importantly, 2) because this information is used for counseling, physicians need to be mindful of the many anomalies that are no less devastating despite their lack of reliable ultrasound detection.
Other investigators have also found that visualization of fetal anatomy may be impaired or suboptimal in obese women and in the setting of diabetes. Hendler and colleagues4 found that maternal obesity significantly limits visualization of the fetal heart. Even when advanced ultrasound equipment was used, suboptimal visualization of the fetal heart occurred in approximately 36% of patients, compared with only 16% in nonobese pregnancies. Detection of anomalous infants was not evaluated in that study. In a subsequent study, these investigators found that even on follow-up evaluation, visualization of cardiac anatomy was suboptimal 12%, 17%, and 20% of the time in those with class I, II, and III obesity, respectively.5 Regarding the limitations of ultrasonography in women with diabetes, Wong and colleagues13 reported a significantly lower rate of anomaly detection in diabetic pregnancies than in low-risk pregnancies, 30% compared with 73%. The authors did not specifically take BMI into consideration, but noted that the mean BMI was significantly higher in the diabetic group than among low-risk women. Image quality was considered unsatisfactory in 37% of diabetic women, and even on repeat evaluation, 86% still had unsatisfactory image quality. Wong concluded that anomaly screening in diabetic pregnancies is significantly worse than that for the general population, and the most significant reason for such failure seemed to be related to maternal habitus and unsatisfactory image quality.13 We also found that women with diabetes had a significantly higher likelihood that their fetal anomalies would not be detected at the initial 18–24 week ultrasound examination; however, the maternal BMI did not completely account for this difference, and it may be that the predominantly truncal obesity that complicates diabetes may have played a role.
Whether follow-up ultrasonography could have an effect on anomaly detection in obese women is not something our study was able adequately to address, because we did not routinely perform additional imaging. However, in our study, follow-up evaluation performed in 77% of those with a high-risk indication and 32% with a low-risk indication yielded an overall improvement in anomaly detection of only 4%. As described above, studies in women with diabetes have demonstrated that visualization often remains suboptimal on follow-up imaging.5,13 Similarly, Wolfe and colleagues14 reported that women with BMI above the ninetieth percentile had an overall decrease in visualization of approximately 15%, with no improvement in visualization as gestation advanced or with longer duration of the examination. It is reasonable that anomaly detection would be related to anatomy visualization. Unfortunately, this is a particularly challenging question to address, because there is no standardization of malformations that may or may not be identified, even when a specific anatomic structure is adequately imaged. Another issue our study was neither designed nor able to address is whether targeted imaging might have improved anomaly detection if offered to all women with obesity. Given the magnitude of the obesity epidemic, the cost of performing targeted examinations, the need for personnel with such expertise, and even the risk of injury to providers, we feel that additional study would be needed to justify such an approach.
Based on our findings, we suggest that when performing ultrasonography in overweight or obese pregnancies, counseling may need to be modified to reflect the limitations of both standard and targeted ultrasonography. For example, after a normal ultrasound evaluation, obese women without diabetes may still have up to 1% risk of a major anomaly—including some anomalies not detectable ultrasonographically. Because women with pregestational diabetes also seem to be at particular risk to have anomalies that are not detected and it is not clear whether they would benefit from routine additional imaging, further study may be worthwhile in this population.