Congenital malformations are a leading cause of infant mortality in the United States [6]. Heart defects comprise one quarter to one-third of all birth defects and are a large contributor to infant mortality. Prenatal diagnosis of congenital heart disease (CHD) by way of fetal echocardiography is increasingly common. Tworetsky et al. reported that from 1992 to 1999, 37% of hypoplastic left heart syndrome (HLHS) infants treated at Boston Children’s Hospital were prenatally diagnosed (PREdx) [15], and Sivarajan et al., at the Royal Children’s Hospital of Melbourne, reported that from 2001 and 2005, 77% of HLHS infants were PREdx [14].

Many infants with complex CHD require surgical intervention before hospital discharge. Previous studies investigating whether prenatal diagnosis of CHD favorably impacts survival have demonstrated inconsistent findings [9, 15]. Likewise, the impact on morbidity has been variably reported [3, 4, 16]. In light of these mixed reports, and with the knowledge that prenatal diagnosis is now more common than postnatal diagnosis, we sought to reassess the relations between prenatal diagnosis and various neonatal outcomes. First, we evaluated the effect of prenatal diagnosis on birth characteristics, such as gestational age (GA) and birth weight (BW). Second, the differences in anatomic cardiac diagnoses between PREdx and postnatally diagnosed (POSTdx) infants were investigated. Third, we assessed the impact of prenatal diagnosis on neonatal morbidity, including the need for more aggressive preoperative measures, day of life (DOL) of surgery, and hospital length of stay (LOS). Fourth, the effect of prenatal diagnosis on mortality was determined. Finally, we investigated the trends in prenatal diagnosis and the impact of prenatal diagnosis on DOL of surgery and hospital LOS at our institution during the course of the study period.

Methods

Study Design and Procedures

A retrospective chart review of infants who underwent repair of CHD before hospital discharge between January 1, 2004 and January 1, 2008 at the Morgan Stanley Children’s Hospital of New York (MSCHONY), a tertiary care center with a level IV neonatal intensive care unit, was conducted using the pediatric cardiothoracic surgical database and the Center for Prenatal Pediatrics database. Institutional Review Board approval was obtained. Patient data recorded included timing of diagnosis, type of congenital heart defect, BW, GA, and DOL of surgery. In addition, preoperative information such as use of antibiotics, pressor support, or mechanical ventilator, as well as cardiac catheterization and MRI was collected. Operative and postoperative information including DOL of surgery, bypass and cross-clamp time, and hospital LOS was also recorded. Lesions were categorized based on the surgical severity score described and validated by Jenkins et al., which ranks cardiac diagnoses based on the surgical repair performed before hospital discharge [7].

Univariate descriptive statistics are summarized as means and SDs for normal distributions and as medians and interquartile ranges (IQR) for nonparametric distributions. Differences in proportions between groups were analyzed using 2 × 2 tables and Pearson’s χ2 test statistic. Univariate relations were explored using Kaplan–Meier, and differences in medians were assessed using the log-rank test. In an effort to evaluate more fully the associations between prenatal diagnosis and DOL of surgery, hospital LOS, and neonatal mortality, multivariate models were constructed using both binary logistic regression and the Cox Proportional Hazards technique. Subject characteristics that showed a univariable association with DOL of surgery, hospital LOS, and mortality at the 0.1 level were eligible for inclusion in the multivariable models. All other alpha values were set at 0.05.

Results

During the 4-year study period, 439 infants underwent neonatal cardiothoracic surgery. Characteristics of the cohort are listed in Table 1. Prenatal diagnosis was made in 294 (67%) infants. Median DOL of surgery was 7 days (IQR 5–9), and median length of hospital stay was 17 days (IQR 12–28). Postoperative mortality was 5% in the entire series.

Table 1 Information on total cohort (total n = 439)
Full size table

Birth Characteristics

Differences between the prenatal and postnatal diagnosis group are listed in Tables 2, 3. The prenatal diagnosis group had a lower mean GA (37.9 ± 2.1 vs. 38.6 ± 2.4 wk, p < 0.001) and BW (3.0 ± 0.6 vs. 3.1 ± 0.6 kg, p = 0.002) and were more likely to be born at our tertiary care center where the surgery was performed (OR 156.7; 95% CI 55.6–441.8, p < 0.001) compared with infants diagnosed after birth.

Table 2 Catagoric variables
Full size table
Table 3 Continuous variables
Full size table

Cardiac Diagnosis Breakdown

The prenatal diagnosis group had higher surgical severity scores (p < 0.001) compared with the postnatal group in accordance with the higher prevalence of more severe forms of CHD among the prenatal group (Table 4). Infants with HLHS were more likely to be PREdx (OR 4.1; 95% CI 1.9–8.9), whereas infants with transposition of the great arteries (TGA) were less likely to have a prenatal diagnosis (OR 0.4; 95% CI 0.3–0.7). Infants with total anomalous pulmonary venous return (TAPVR) were also less likely to be diagnosed before birth (OR 0.02; 95% CI 0.0–0.1).

Table 4 Diagnosis breakdown: PREdx versus POSTdx
Full size table

Neonatal Mortality

Prenatal diagnosis did not impact neonatal mortality. Independent factors associated with mortality are listed in Tables 5, 6. SV morphology (p < 0.001), surgical severity score (p = 0.002), postoperative open chest (p < 0.001), 1-min Apgar score (p = 0.05), and bypass time (p < 0.001) were significantly associated with mortality. Multivariate analysis demonstrated that postoperative open chest (OR 1.9; 95% CI, p = 0.12) was the only independent factor associated with mortality.

Table 5 Univariate associations with mortality
Full size table
Table 6 Continuous variables associated with mortality
Full size table

Preoperative Morbidity

PREdx infants were less likely to receive preoperative mechanical ventilation (OR 0.6; 95% CI 0.4–0.9), antibiotics (OR 0.2; 95% CI 0.1–0.4), cardiac catheterization (OR 0.5; 95% CI 0.3–0.9), or emergent surgery (OR 0.2; 95% CI 0.1–0.5) compared with POSTdx infants (Tables 2, 3).

DOL of Surgery

Prenatal diagnosis did not impact DOL of surgery (median DOL of surgery 7 days [IQR 5–8] for prenatal diagnosis vs. 6 days [IQR 5–9] for postnatal diagnosis, p = 0.9). There was no difference in DOL of surgery among groups even when stratified by cardiac lesion (Table 7). Univariate associations between subject characteristics and DOL of surgery are listed in Table 8. Multivariate Cox-proportional hazard modeling demonstrated that the use of preoperative antibiotics (hazard ratio [HR] 0.7; 95% CI 0.6–0.9, p = 0.002) and additional fetal anomalies (HR 0.5; 95% CI 0.4–0.7, p < 0.001) were independently associated with a longer time to surgery when controlling for timing of diagnosis, prematurity, and low BW.

Table 7 Median DOL of surgery by cardiac diagnosis and diagnosis timing
Full size table
Table 8 Univariate associations with DOL at surgery
Full size table

Hospital LOS

Univariate associations between subject characteristics and hospital LOS are listed in Table 9. The prenatal diagnosis group had longer median hospital LOS (20 [range 13–33] vs. 15 days [range 11–25], p = 0.001) than the postnatal diagnosis group. Multivariate Cox-proportional hazard modeling demonstrated that surgical severity score (HR 0.9; 95% CI 0.8–0.9, p < 0.001), other fetal anomalies (HR 0.7; 95% CI 0.5–0.9, p = 0.03), prematurity (HR 0.8; 95% CI 0.6–0.9, p = 0.036), and DOL of surgery (HR 0.9; 95% CI 09–1, p < 0.001) were independently associated with a longer hospital LOS when controlling for prenatal diagnosis and low BW. Prenatal diagnosis was not associated with hospital LOS in this multivariate model.

Table 9 Univariate associations with LOS
Full size table

Prenatal Diagnosis Trends

Trends in prenatal diagnosis, DOL of surgery, and hospital LOS across the 4-year study period are listed in Table 10. Although there was no significant difference in the rate of prenatal diagnosis, median DOL of surgery and hospital LOS decreased significantly from 2004 to 2007.

Table 10 Outcomes over the years
Full size table

Discussion

This study demonstrates that the prevalence of prenatal diagnosis of CHD in the current era is higher than previously reported, reaching as high as 88% for certain defects at our center. There are various explanations for the increase in prenatal detection of CHD. First, standard of care of obstetrical screening for fetal anomalies has expanded to include first-trimester nuchal fold measurements. Abnormalities in these early scans typically result in referral for fetal echocardiography. Second, modern technology has improved image resolution, thus making it easier to detect anomalies during routine obstetric ultrasound screening.

Our study demonstrated an association between prenatal diagnosis and lower GA and BW. This has been reported previously [3]. This finding may be explained in part by institutional practices of scheduling delivery, especially among women who live far from the medical center, to assure availability of maximal neonatal medical support. The differences in BW and GA are small and are unlikely to be of clinical significance, as evidenced by the decrease in neonatal morbidity among the prenatal diagnosis group.

We found that the more severe cardiac anatomic lesions are likely to be PREdx. This finding has been reported previously [3] and can be explained by advances in obstetrical screening methods. The most common indication for referral for fetal echocardiography is a cardiac abnormality seen on the routine obstetrical anatomic scan [5, 12]. Guidelines from the American Institute of Ultrasound in Medicine recommend a complete second-trimester anatomic scan of the fetal chest, including a four-chamber view of the fetal heart and, if technically feasible, views of the two outflow tracts [1]. Therefore, the more grossly abnormal the appearance of the heart in the four-chamber view, the more likely the lesion is to be recognized on routine ultrasound. Conversely, lesions such as TAPVR and TGA that do not significantly change the appearance of the four-chamber view are more likely to be missed. TGA is likely to be overlooked because views of the outflow tracts are not always obtained. The same is true for anomalous pulmonary venous connections because pulmonary veins are not often discerned on fetal studies.

Our study did not demonstrate an association between prenatal diagnosis and surgical mortality. Other studies similarly did not demonstrate a positive impact of prenatal diagnosis on the survival of infants with HLHS and other forms of CHD [8–11]. In 2001, Tworetsky et al. demonstrated a survival advantage to prenatal diagnosis among HLHS infants. In this study, 37.5% of the HLHS infants were PREdx, whereas in our study 88% of HLHS infants were PREdx and 88% survived neonatal surgery. This change in the prevalence of prenatal diagnosis and in the prevalence of survival may partially explain the differences in findings. Although it is true that for some infants prenatal diagnosis is life saving, there are others for whom immediate delivery of targeted cardiac support may not be as crucial. If 88% of HLHS infants are PREdx indiscriminately, which includes a mix of the above phenotypes, then the association between prenatal diagnosis and survival may be difficult to discern.

We have demonstrated that prenatal diagnosis improves certain measures of neonatal morbidity. Others have reported that prenatal diagnosis decreases neonatal metabolic acidosis [2, 4, 13, 16]. Although we found no statistically significant impact of prenatal diagnosis on neonatal acid–base status, we did see decreased use of ventilators, antibiotics, and cardiac catheterization, and fewer emergent surgeries among the prenatal diagnosis group. These findings can be attributed in part to the anticipation of medical needs of patients with a fetal diagnosis of CHD and the institution of medical support before clinical status deterioration, avoiding the poor status that is the usual presentation of the POSTdx infant. The majority of our PREdx cohort (82%) was born at our institution, thus allowing the immediate delivery of medical support, including prostaglandin, which is vital for the maintenance of ductal patency in a ductal-dependent lesion. In addition, the development of cyanosis and respiratory distress in an infant without a prenatal CHD diagnosis often prompts intubation and a sepsis workup, further explaining the higher prevalence of these findings in the postnatal diagnosis group. Furthermore, infants born at outside institutions may have been intubated for the purposes of transportation to our center. The decreased use of cardiac catheterization in the prenatal diagnosis group may be accounted for by the high proportion of POSTdx infants with TGA, who often require balloon atrial septostomy, and those with TAPVR, who may undergo diagnostic cardiac catheterizations to delineate the anatomic pathways of the anomalous pulmonary veins. Similarly, the association with emergent surgery may be explained by the TAPVR anatomic subgroup, many of whom present in distress and are urgently repaired.

We found no association between prenatal diagnosis and DOL of surgery or hospital LOS, even when controlling for lesion severity. Although median hospital LOS was longer in the prenatal diagnosis group in a univariate analysis, the association was no longer significant in multivariate analysis. Copel et al. demonstrated that prenatal diagnosis was associated with an overall increase in neonatal hospital LOS and hospital expenses for live-born infants and postulated that this might be due to increased illness severity in the PREdx cohort [3].

Limitations

Our study consisted of a retrospective review of data. Only infants who survived to cardiac surgery were captured in this analysis; therefore, the effect of prenatal diagnosis on overall neonatal mortality cannot be assessed with the present data. In addition, our sample does not include pregnancies that were terminated. It has been our experience that few fetuses with a prenatal diagnosis of CHD seen at our institution are terminated, which is likely due to selection bias: patients who have decided to continue the pregnancy seek care at a tertiary-level referral center, whereas patients wishing to terminate the pregnancy may do so locally. One could postulate that prenatal diagnosis of CHD may lead to increased terminations of pregnancy, especially among fetuses with more severe lesions and with additional significant anomalies. All of these factors may confound our ability to establish a causal relation between prenatal diagnosis and neonatal outcomes.

Conclusion

Prenatal diagnosis of complex CHD is now more common than postnatal diagnosis; levels were as high as 75% overall in the final year of our study. Prenatal diagnosis is more likely to occur in lesions of higher severity. Despite representing more severe lesions, prenatal diagnosis was associated with decreased neonatal morbidity, including decreased use of mechanical ventilation, antibiotics, and emergent surgery. Prenatal diagnosis did not impact DOL of surgery, hospital LOS, or mortality. Further investigation into the impact of prenatal diagnosis on DOL of surgery and hospital LOS is needed. The positive economic impact of more efficient management of these patients is also an important consideration. Advanced prenatal knowledge of an indication for cardiac surgery may allow for the optimization of factors beyond immediate neonatal resuscitation that affect neonate survival, including labor, delivery, and operative repair.