To assess the factors influencing the prenatal detection rate of structural congenital heart diseases (CHDs).
To assess the factors influencing the prenatal detection rate of structural congenital heart diseases (CHDs).
A retrospective study was conducted at a major obstetric hospital in Australia between 1 January 1996 and 30 June 1999. The medical records of all fetuses and infants born with CHD, except those with isolated patent ductus arteriosus or secundum atrial septal defect, were reviewed. Only pregnancies that had prenatal ultrasound scan assessments for morphological surveys were included. The following factors that may influence the detection rate were assessed: complexity of the lesions; experience of the sonographers (performance in tertiary versus non-tertiary institutions); presence of other structural or chromosomal anomalies; and maternal body mass index (BMI).
The incidence of structural CHD in this series, excluding cases referred from other hospitals, was 7.0 per 1000 (179/25 529). Of the 179 pregnancies with CHD, 151 had prenatal ultrasound scans and were included in the study. The overall detection rate for CHDs in this series was 40.4%. The detection rate for isolated septal defects was poor (13.7%). The detection rates were significantly higher for complex lesions (54%), for lesions with concomitant septal defects (66.7%), and for lesions with abnormal four-chamber views (62.9%). The detection rate was also higher if the scan was performed in the tertiary institution, and if there were other chromosomal or structural anomalies. Maternal BMI did not affect the detection rate in the current series. Stepwise logistic regression analysis showed that three independent variables affecting the detection rate were complexity of the cardiac lesion, experience of the operator, and the detection of chromosomal anomalies.
A high detection rate for major CHDs can be achieved in a screening setting but there is still room for improvement in scanning skills in the four-chamber view and great-artery analysis in both tertiary and local centers. Copyright © 2002 ISUOG. Published by John Wiley & Sons, Ltd.
Congenital heart disease (CHD) is the most common congenital anomaly, with a prevalence of 8.0 per 1000 live births1, 2. In the early 1980s, it was recognized that prenatal diagnosis of structural CHD by transabdominal sonography was possible3–5. Fetal diagnosis had been shown to reduce the birth prevalence of complex structural CHDs, as some women chose not to continue the pregnancies after prenatal diagnoses were established6.
An early study performed in a special fetal cardiology center showed that majority (81%) of prenatally detected CHDs had abnormal four-chamber views7. Recent studies performed in the general obstetric setting, using the four-chamber view as the screening method, among low-risk populations, showed a disappointing detection rate. The reported detection rate of CHDs by routine prenatal screening in fact ranged from 4.5% to 40%8–15. It is important to establish the factors that may influence the detection rates of structural congenital cardiac anomalies. Such information may help in devising methods for improved performance. We postulate that the important factors that may influence the detection rate include the type of lesion, the method of screening, the type of machine used, and the experience of the sonographer8, 15–17. DeVore et al. in 1993 found that other factors such as early gestational age, maternal adipose tissue thickness, and previous lower abdominal surgery also influenced the chances for successful imaging of the fetal heart18.
The aim of the present study was to assess the relative significance of the various factors affecting the prenatal detection rate of structural CHDs in a hospital setting.
The study was conducted at Mater Mothers' Hospital, South Brisbane, Australia. Mater Mothers' Hospital is a tertiary hospital staffed with full-time maternal-fetal medicine subspecialists. Its annual delivery rate is about 7000. It has a 20-bed neonatal intensive care unit and has a catchment area covering about 20 000 births per annum. Routine ultrasound scans for mothers booked for confinement in the hospital were performed either in the hospital (35%) or in private radiology clinics (65%) under a shared-care arrangement. At Mater Mothers' Hospital, qualified sonographers and maternal–fetal medicine subspecialists performed the ultrasound scans. During the study period, the ultrasound machines used were a Toshiba SSA-250 (Toshiba, Tustin, CA, USA) and an ATL-3000 (ATL, Bothell, WA, USA). Satisfactory assessment of the fetal heart included assessment of the four-chamber view and outflow tracts. Color flow mapping was not routinely utilized during the period of study. In the private clinics, qualified sonographers and specialists in obstetric ultrasound performed the ultrasound scans. Private settings were equipped with machines of equivalent or better quality. All sonographers and specialists performing obstetric ultrasound had the Diploma of Medical Ultrasound or Diploma of Diagnostic Ultrasound qualifications issued by the Australasian Society for Ultrasound in Medicine.
Pediatricians screened all newborns before discharge from the hospital. The perinatal database was hand-searched for the presence of CHDs. This database included those admitted to the neonatal ward within 4 weeks of discharge. All babies suspected of having CHDs were scheduled to undergo echocardiography by the pediatric cardiologist and final diagnoses were based on the echocardiographic diagnoses. Babies with clinically small ventricular septal defects (VSDs) might not have undergone echocardiography but were followed up in 3 months. If the cardiac murmur persisted, echocardiography was arranged. The medical records of all fetuses and babies with structural CHDs delivered between 1 January 1996 and 30 June 1999 were reviewed. All structural CHDs detected during the neonatal period, except isolated patent ductus arteriosus and secundum atrial septal defect (ASD), were included. All pregnancies terminated for fetal congenital malformations were reviewed and those with CHDs were included. Postmortem results on all fetuses or babies with CHDs were checked and these were used for the final diagnoses. Pregnancies without prenatal sonographic survey for fetal anatomy were excluded.
Factors that could affect the detection rate were assessed. These included: (1) type and complexity of the cardiac lesions; (2) scans performed in the tertiary institution vs. those performed in non-tertiary institutions; (3) identification of other structural congenital anomalies; (4) identification of chromosomal anomalies; and (5) maternal body habitus.
The detection rates for different types of structural CHDs were analyzed. In order to assess the detection rate according to the complexity of the lesions, the cardiac anomalies were subdivided into two major groups: isolated septal defects and complex heart lesions. Isolated septal defects included isolated VSD, isolated primum ASD, or both lesions. Complex heart lesions included all other structural congenital cardiac anomalies, except patent ductus arteriosus or secundum ASD.
The complex heart lesions were further subdivided into lesions with and without outflow tract defects, and their detection rates compared. Lesions with outflow tract defects included hypoplastic left heart, tetralogy of Fallot, transposition of the great arteries, aortic stenosis, coarctation of the aorta, truncus arteriosus, pulmonary stenosis, pulmonary atresia, aorto-pulmonary window, and other complex lesions with abnormal outflow tracts. Further analysis was performed in babies with outflow tract defects to assess whether there was any difference in the detection rate if there were concomitant septal defects.
Detection rates for routine ultrasound scans performed in the tertiary (Mater Mothers' Hospital) and non-tertiary institutions (local centers) were compared. Babies referred from other hospitals with suspicion of congenital cardiac anomalies, either prenatally or postnatally, were excluded from the analysis. As this was a study of detection based on protocols in the institution, transferred infants would not have been scanned under the same circumstances, and ascertainment bias would be present. Moreover, those detected prenatally and with complex lesions were more likely to be referred. Since the four-chamber view is the most widely used screening method and outflow tract assessment may not be routinely performed in non-tertiary institutions, subanalyses of the detection rates for lesions with normal/abnormal four-chamber views were made. Lesions considered to have abnormal four-chamber views included: (1) lesions with chamber size abnormalities such as hypoplastic left heart or univentricular heart; (2) lesions with abnormal atrioventricular valves such as atrioventricular canal defect, Ebstein's anomaly, mitral or tricuspid valvular atresia; and (3) large septal defects such as isolated VSDs of > 3 mm at birth or VSDs associated with truncus arteriosus or tetralogy of Fallot.
Detection rates for fetuses and infants with isolated structural congenital cardiac anomalies were compared with those that had other structural or chromosomal anomalies identified. This was aimed at assessing the assumption that identification of other structural or chromosomal anomalies could raise the alert for a more detailed assessment, often in a tertiary institution.
To assess the impact of maternal habitus on the detection rate, maternal prepregnancy body weight and body mass index (BMI) were compared between women who had prenatal and postnatal diagnoses of fetal CHDs.
Statistical analyses were performed using Statistical Package for Social Science (SPSS Inc., Chicago, IL, USA). Chi-square test or Fisher's exact test was used, as appropriate, to compare categorical variables. Unpaired Student's t-test was used to compare continuous variables with normal distribution. The Mann–Whitney test was used when the data were distributed in a non-Gaussian fashion. Values of P < 0.05 on two-tailed analyses were considered statistically significant. Stepwise binary logistic regression was performed to identify the independent variable(s) affecting the detection rate of CHDs. Each variable was assigned a value of 0 or 1, depending on whether the variable was absent or present, respectively. The computer software was used to predict the independent variable(s) that predicted prenatal detection of CHDs.
Between 1 January 1996 and 30 June 1999 a total of 211 fetuses and babies with CHDs were identified. Thirty-two pregnancies/babies were referred from other hospitals for tertiary assessments because of the suspicion of congenital cardiac anomalies in the prenatal or postnatal period. The incidence of CHD in this series, excluding cases referred from other hospitals, was 7.0 per 1000 (179/25529). One hundred and fifty-one fetuses and babies with CHDs other than patent ductus arteriosus or secundum ASD had prenatal ultrasound scans for morphological survey. A total of 89% of the scans were performed at 17–24 weeks, and the rest were performed after 24 weeks.
The overall detection rate for CHDs in the study population was 40.4% (61/151). After excluding those pregnancies and babies referred from other hospitals, the detection rate was 36.1% (43/119). Twenty-six cases were terminated and the major diagnoses on postmortem reports corresponded to the prenatal diagnoses. The detection rates for various congenital cardiac anomalies are listed in Table 1. The majority of the complex cardiac lesions had detection rates of 50% or more. A total of 34% of babies with CHDs had isolated septal defects and the majority of these lesions were small muscular VSDs. There were two cases of isolated primum ASDs and they were diagnosed after birth.
|Congenital heart disease||Detection rate|
|VSD and/or ASD||7/51||13.7|
|Hypoplastic left heart||9/15||60|
|Tetralogy of Fallot||5/9||55.6|
|Transposition of the great arteries||10/16||62.5|
|Severe aortic stenosis||3/7||42.9|
|Coarctation of aorta||2/7||28.6|
|Other complex lesions||11/14||76.9|
Figure 1 represents a flow chart analyzing the detection rates according to the complexity of the structural CHDs. Only 13.7% of the VSDs were detected, and all of these defects were > 3 mm at birth. About half (54%) of the complex cardiac lesions were detected during the prenatal period. The detection rate was significantly higher than that for isolated septal defects. For those complex cardiac lesions with abnormal outflow tracts, the detection rate was higher if there were concomitant VSDs (67% vs. 40%).
Comparisons of the detection rates for various CHDs between the tertiary and non-tertiary institutions were made (Table 2). All cases referred with suspicion of structural CHDs were excluded from this analysis. The detection rates for all types of structural congenital cardiac anomalies were higher for women who had the routine ultrasound scans performed in the tertiary institution. The overall detection rate of CHDs for routine ultrasound scans performed in the tertiary institution was significantly higher than that for scans performed in non-tertiary institutions (61% vs. 21%). The odds ratio between the two types of institution for detection of isolated septal defects was 8.3, and for complex CHD was 4.7.
|Congenital heart disease||Prenatal detection rate||P||OR (95% CI)|
|Tertiary institution||Non-tertiary institutions|
|VSD and/or ASD||4/12||33.3||2/35||5.7||< 0.05*||8.3 (1.3–53)|
|Complex CHD (total number)||23/32||71.9||14/40||35.0||< 0.01||4.7 (1.7–13)|
|Complex heart with normal outflow tracts||7/7||100||0/2||0||< 0.05*|
|Complex heart with abnormal outflow tract(s)||16/25||64.0||14/38||36.8||< 0.05||3.0 (1.1–8.7)|
|Hypoplastic left heart||6/7||85.7||1/4||25||0.09*|
|Tetralogy of Fallot||1/1||100||2/6||33.3||NS|
|Transposition of the great arteries||4/6||66.7||4/7||57.1||NS|
|Coarctation of aorta||0/2||0||0/2||0||NS|
|Other complex lesions||3/3||100||4/7||57.1||NS|
|Overall||27/44||61.4||16/75||21.3||< 0.001||5.9 (2.6–13)|
Figure 2 represents a flow chart of the performance in tertiary vs. non-tertiary institutions according to the ultrasound features. The detection rate for lesions with abnormal four-chamber views was significantly higher when compared to those with normal four-chamber views both in the tertiary as well as non-tertiary institutions. When comparing the performance of the tertiary institution with non-tertiary institutions, the detection rate for lesions with abnormal four-chamber views was significantly higher at the tertiary setting (78% vs. 47%). For lesions with normal four-chamber views, the detection rate at the tertiary institution was still higher (16.7% vs. 6.7%), but the difference did not reach statistical significance.
Table 3 shows the influence of the identification of other structural or chromosomal anomalies on the detection rates of CHDs. The overall detection rate was significantly improved with the identification of other congenital anomalies (81% vs. 34.4%). These were true for both isolated septal defects (50% vs. 3.0%), and complex cardiac lesions (100% vs. 46.1%). On the contrary, there were two fetuses with multiple structural congenital anomalies that were not detected prenatally (right diaphragmatic hernia, cleft lip). Both fetal cardiac lesions in these pregnancies were also not detected.
|Isolated lesions: detection rate (n (%))||†Chromosomal anomalies: detection rate ( n (%))||*P||OR (95% CI)||‡Other congenital structural anomalies: detection rate ( n (%))||*P||OR (95% CI)|
|All CHDs||42/122 (34.4)||15/19 (78.9)||< 0.001||7.1 (2.2–22.9)||17/21 (81.0)||< 0.001||8.1 (2.6–25.6)|
|Isolated septal defect||1/33 (3.0)||5/9 (55.6)||0.001||40 (3.7–435)||4/8 (50)||0.003||32 (2.8–362)|
|Complex lesions||41/89 (46.1)||10/10 (100)||0.001||2.2 (1.7–2.7)||13/13 (100)||< 0.001||2.2 (1.7–2.7)|
A total of 22/211 infants with CHDs had chromosomal abnormalities. Nineteen cases (86%) were detected during the prenatal period. Most of these were trisomy 21 (12); the others being trisomy 18 (five), trisomy 13 (three), triploidy (one) and 22q deletion (one). The overall detection rate of structural CHDs was significantly improved if the chromosomal anomalies were detected prenatally (79% vs. 34.4%).
Comparisons of maternal characteristics, such as prepregnancy weight (66.0 kg vs. 66.1 kg) and BMI (24.8 kg/m2 vs. 24.0 kg/m2), between women whose babies' CHDs were diagnosed prior to delivery vs. those that were detected only after delivery showed no differences. Maternal obesity did not appear to be an important confounding variable in this study.
Using stepwise logistic regression analysis, complexity of the cardiac lesions, experience of the operator (tertiary scan), and detection of chromosomal anomalies were the three independent variables that affected the detection rate (Table 4).
|Independent variables||P||Exp (B)||95% CI for Exp (B)||Correlation matrix|
|Complex lesion||0.0002||15.7||3.7–66||− 0.90|
|Other structural anomaly||0.058||5.7||0.94–34||− 0.37|
|Maternal obesity||NS||1.8||0.66–4.8||− 0.36|
As the detection rate of CHD by routine prenatal screening in the general obstetric setting is unsatisfactory, we attempted to determine factors affecting the detection rates of structural CHDs. Such knowledge could help to identify the problems associated with routine screening for CHDs, so that methods could be suggested to overcome these problems. We found that there were three independent factors: the complexity of the cardiac lesions; the experience of the operators; and the identification of chromosomal anomalies affecting the detection rate of CHDs.
In this study, the more complex the lesion, the better the detection rate. Current screening methods for structural cardiac anomalies include assessment of the four-chamber view and visualization of outflow tracts in certain planes as complex cardiac lesions are more likely to have one or more abnormalities seen in these views. Conversely, the low detection rate for isolated septal defects is consistent with data published by others19. All small septal defects (< 3 mm) were diagnosed after delivery. This may not be an important medical issue, as the majority of patients are asymptomatic and do not require surgery. Nevertheless, women need to be informed about this so that they do not have unrealistic expectations, since septal defects are the commonest congenital cardiac lesion.
The detection rate of CHD in the tertiary institution was better than that in the non-tertiary institutions (61% vs. 21%). Similar findings were observed in the RADIUS Study, where the detection rate reported before 24 weeks was 24% at tertiary institutions, and 0% at non-tertiary institutions20. The detection rates for isolated septal defects and complex cardiac lesions were both significantly higher in the tertiary institution. The possible reasons that might account for this difference include difference in ultrasound equipment used, type of screening methods employed, or experience of the sonographers. As most of the private radiological institutions were equipped with similar or better quality machines, we believed this was not an important confounding variable. After controlling for screening method, the detection rate for lesions with abnormal four-chamber views was still significantly higher in the tertiary institution (78% vs. 47%). Thus, sonographers' experience might account for the differences observed.
The detection rates for cardiac lesions with normal four-chamber views were unsatisfactory (Figure 2) even after exclusion of all small muscular VSDs and ASDs (tertiary 33.3%; non-tertiary 17.6%, P = 0.6). This suggested that reliance on the four-chamber view as the only screening method may be insufficient. Assessment of the outflow tracts and aortic arch is possible with appropriate training20. The addition of outflow tract assessment has been shown to improve the detection rate for some of the CHDs involving the great vessels21. Thus, more effort should be concentrated on the detection of these lesions by incorporation of outflow tract assessment in routine screening. Up to two-thirds of isolated outflow tract lesions were not detected prenatally in the tertiary institution, despite routine assessment of the outflow tracts. Mild outflow tract obstructions such as mild valvular stenosis of the great arteries, and coarctation of the aorta may be difficult to diagnose with two-dimensional visualization of the great arteries. Transposition of the great arteries can be detected by the assessment of crossing-over of the great vessels. Thus, the detection rate may further improve by inclusion of the assessment of crossing-over of the great vessels, comparison of great vessel sizes at the valvular level and color flow mapping. The addition of color Doppler mapping may help, especially in outflow tract obstructions. This has been shown to be useful in assisting prenatal diagnosis of CHDs, but its use as a screening tool has not been fully evaluated22–25.
The detection rate for congenital cardiac lesions improved if other structural or chromosomal anomalies were detected. First, these pregnancies probably received more detailed ultrasound surveys by experienced sonographers, including full fetal echocardiographic examination. It has been shown that the use of fetal echocardiography can improve the detection rate of structural CHDs18. Second, sonographers would make more effort to detect cardiac abnormalities, knowing that these pregnancies were likely to be associated with congenital cardiac anomalies. Third, pregnancies with major chromosomal abnormalities were also associated with more complex cardiac abnormalities.
DeVore et al. found that early gestational age, maternal adipose tissue thickness, and previous lower abdominal surgery influenced the successful imaging of the fetal heart18. We failed to show that maternal obesity affected the detection rate, probably because most women would have another scan at a more advanced gestation. The role of gestation was not assessed but the majority of women were screened between 18 and 22 weeks. The presence of an abdominal scar was also not assessed because the information was incomplete; some private obstetricians failed to input this into hospital records. Those women with an incomplete examination automatically had another scan at a more advanced gestation.
Despite such findings, there were 23 CHDs with abnormal four-chamber views that were overlooked. Therefore, there is a need to improve scanning skills in the four-chamber view in tertiary and particularly in local centers.
It is obviously not feasible or cost-effective to screen every pregnancy in a tertiary institution. To improve screening performance, however, one should aim to upgrade the expertise of sonographers in non-tertiary institutions through adequate supervision and training. Although this is difficult to implement, a study from the UK has shown that this is possible6. Targeted fetal echocardiography at a tertiary institution for high-risk populations may increase the yield, but the majority of women who have babies with structural congenital cardiac anomalies do not have such clinical risk factors26.
We acknowledge that there were a number of limitations associated with this retrospective study. The overall detection rate might be overestimated because some structural congenital cardiac anomalies such as ASD and hypoplastic left heart syndrome might not present in the first few days of life. Nevertheless, most of the comparisons made in this study were still valid, as babies in both groups were as likely to have been missed.
In conclusion, the complexity of the cardiac lesions, experience of the operators, and identification of chromosomal anomalies are significant factors affecting the detection rate of CHDs. To improve future detection rates, further emphasis on the training and supervision of sonographers needs to be made and there is room for improvement in scanning skills in both tertiary and local centers. Since routine screening by the four-chamber view appears to be inadequate for outflow tract lesions, specific assessment of outflow tracts is necessary to improve the detection rate by routine ultrasound.