Discordant Fetal and Post-Natal Diagnosis

Can We Do Better?

• congenital heart disease
• echocardiography
• fetal echocardiography
• pre-natal diagnosis
• quality assurance
• ultrasound

In the United States, fetal echocardiography has come of age as a diagnostic tool. From its beginnings in the 1980s (1) and subsequently, many papers have been written that show, by comparing pre- and post-natal studies, the accuracy of pre-natal echocardiographic diagnosis (2–6). A fetal heart society that focuses on investigating the accuracy of fetal diagnosis and prognosis has now been established (7). High-quality large volume programs have now developed the expertise and sophistication required for extremely accurate diagnosis even in very complex disease cases (2,6,8–10). It is widely accepted that fetal diagnosis of certain cardiac conditions is able to guide treatment during the pre-natal period (11) or provide definitive enough diagnosis to guide decisions regarding termination of pregnancy. However, the general population has been underserved, as many pregnant patients do not have access to expert cardiac ultrasound diagnosis (3,12). Delays and inaccuracies in diagnosis clearly put babies with heart disease at risk for adverse outcome, ranging from mild neurocognitive difficulties to devastation or nonsurvival. Pre-natal echocardiography has the potential to provide a first-line defense in preventing poor outcomes by ensuring birth of affected infants into the appropriate level of care. But, how well are we doing?

In this issue of the Journal, Bensemlali et al. (13) provide important observations concerning the accuracy of fetal diagnosis by ultrasound. This Parisian fetal diagnostic center is one of the oldest, largest, and most advanced in the world, and their studies have been carefully scrutinized and documented over the years. Their echocardiographic diagnostic techniques presumably provide correct information for patients (who may choose to terminate pregnancy or continue), and for the teams who will care for both the mother and the newborn—including allowing for appropriate post-natal level of care and interventional strategies for palliation or repair. It is sobering to note, therefore, that the results of this French study looking only at the accuracy of their own fetal echocardiographic experts nonetheless demonstrated a surprising quota of inaccurate or incomplete diagnoses.

The median age of the fetuses was 26 weeks (range 14 to 40). The overall accuracy of their fetal diagnosis in 1,258 of these neonates with pre-natally diagnosed heart disease was only 70.7%, whereas 29.3% had discordant pre- and post-natal diagnoses. They further stratified their discordant diagnosis into 3 groups characterized as: 1) a completely different category (2.9%); 2) partially different with major impact on post-natal management (7.7%); and 3) different without impact on post-natal management (18.7%) of their cases.

Additionally, they report on 189 cases, where autopsies were done, from either spontaneous abortion or termination of pregnancy. In this series, 45% had been correctly diagnosed by fetal ultrasound, whereas in 8.5% the diagnosis was completely different, and in 14.3% the diagnosis was partially correct but there would have been potential for significant change in post-natal management had the pregnancy continued.

In situations where one cannot afterward reverse an action, however (such as, in termination of pregnancy), the consequences of incorrect diagnosis are extremely serious, a matter that is deeply troubling both to Bensemlali et al. (13) and to us. Extreme care needs to be taken when such therapy is contemplated and how continuing evaluation during pregnancy might change as the fetus grows and develops. Errors that lead to unnecessary monitoring in the post-natal period and to unfounded concern on the part of the expectant parents and caregivers are certainly worrisome and should be minimized. However, perhaps the most troubling statistic in their report (13) is that in one-tenth of the fetuses, fetal echo “failed to predict the need for immediate neonatal intervention.”

Several organizations have recently come out publicly to call for improvement in obstetric screening for congenital heart disease, and indeed the pre-natal “miss” rate seems to be dropping-though we are still far from perfect. This study (13) shows us that the work that needs to be done extends to the referral “specialist.” Perhaps other groups might continue such an evaluation prospectively in the United States, in and in other parts of the world as well to evaluate diagnostic accuracy prospectively to improve the accuracy of fetal echocardiographic diagnosis. In 2008, Benavidez et al. (14) proposed taxonomy for diagnostic errors to categorize errors by type, severity, preventability, and primary contributor specific to pediatric echocardiography. The fetal echocardiography community would be well served to adopt similar standards for improvement and the study by Bensemlali et al. (13) represents an important first step. In analyzing the reasons for their errors, the investigators found that the majority of severe errors were, similar to the post-natal study, in the realm of rare diagnoses and moderate complexity. Many were made in the region of outflow tracts and the great arteries. Incomplete documentation of examination planes (3-vessel view), early gestation at examination, and lack of 3-dimensional imaging modalities were cited as the most likely reasons for discrepancies in most patients.

In the previous (post-natal) experience, additional modalities including cardiac magnetic resonance imaging and cardiac catheterization were used; unfortunately, these are not practical in the fetus. Furthermore, the most basic diagnostic tool, physical examination, is not available in the fetus, making precision of anatomic diagnosis even more difficult—the fetus with double outlet right ventricle will not present in utero with a saturation difference consistent with transposition physiology to lead the examiner toward a diagnosis of subpulmonary over subaortic ventricular septal defect; the fetus with inlet ventricular septal defect rather than perimembranous will not have an electrocardiogram with counterclockwise loop and the fetus with situs inversus totalis will not have right-sided heart tones on auscultation to guide the echocardiographer and provide clues to the diagnosis. Furthermore, the natural history in utero of some of these lesions is still being defined—mild aortic stenosis in mid-gestation may or may not progress to severe disease by term (15). The lack of key physiologic data indeed makes the job of the fetal echocardiographer especially difficult, and the stakes much higher because to be wrong may mean the difference between a critically ill infant needing emergent intervention and an infant stable at birth who will merely require an elective operation. Indeed, in the vast majority of fetuses with discrepancy between pre-natal prediction and post-natal condition it was the physiology that was misjudged—degrees of obstruction to the right or left ventricular outflow tracts and commitment of the great arteries to the ventricular septal defect in double outlet right ventricle as transposition physiology or tetralogy-type physiology. These are indeed difficult distinctions to make in the presence of fetal circulation and without systemic oxygen saturation data that would be readily available in a newborn.

In the early days of fetal echo, discrepancies were common, and we have indeed come a long way. That accurate diagnoses of congenital heart malformations can be made at 12 or 13 weeks of gestation (16,17) would have been unfathomable 25 years ago, as would have been the idea that pre-natal intervention for fetal heart disease could be performed and would result in an alteration of the natural history of the defect (18). But we must not rest on the laurels of the early fetal echocardiographers; as imaging technology improves and studies of human fetal physiology become possible through sharing of experience with rare diseases, we owe it to our patients to strive for perfection. The study by Bensemlali et al. (13) shows that we are as yet far from it, but there is hope, and we can do better.

• American College of Cardiology Foundation