Epidemiologic Features of the Presentation of Critical Congenital Heart Disease: Implications for Screening

METHODS

We conducted a retrospective cohort study of neonates who were admitted with critical CHD to a single institution. Patients were included if they (1) were admitted to the Children's Hospital of Philadelphia (CHOP) between January 1, 2000, and June 30, 2003, at <30 days of age; (2) had CHD; and (3) required surgical or transcatheter intervention for CHD by 30 days of age or died as a result of CHD by 30 days of age. Exclusion criteria were (1) patent ductus arteriosus only in a preterm infant, (2) withdrawal of support before intervention in the absence of evidence of other organ injury, and (3) severe noncardiac anomalies that were the dominant factor in the patient's prognosis. This last criterion was defined as presence of (1) a frequently fatal genetic syndrome such as trisomy 13 or trisomy 18, (2) major malformations of 2 extracardiac organ systems, or (3) a single severe noncardiac malformation that had a high likelihood of being fatal. Potentially eligible patients were identified by cross-referencing the hospital discharge database, other CHD study databases, the cardiothoracic surgery database, and the cardiac ICU database. The hospital discharge database was queried using International Classification of Disease, Ninth Revision codes for CHD in combination with a procedure code for a surgical or transcatheter interventional procedure or discharge status of death.

Demographic information, including race, as collected by the hospital admission clerks, and laboratory data, were obtained from hospital databases. Inpatient charts, including outside records filed with them, were reviewed for the remaining data elements. Information collected included description of cardiac anatomy, genetic syndromes and noncardiac anomalies, referral source, preoperative laboratory data, preoperative clinical events, and timing of presentation and diagnosis of CHD. The institutional review board of CHOP approved the study with waiver of informed consent because the study entailed review of existing data only.

The case definition for preoperative SPC was designed to reflect a sustained period of inadequate tissue oxygenation or perfusion with profound metabolic acidosis or secondary organ dysfunction (Table 1). These features have been generally accepted as identifying a high-risk group in clinical practice at the study institution, although no study has associated this case definition with adverse outcome. Timing of SPC was categorized as occurring at ≤12 hours of age, between 12 and 24 hours of age, or, if after 24 hours of age, the number of days of age. In the absence of documentation that a patient met a given criterion of the definition (eg, no creatinine sent), the patient was considered not to have met that criterion.

Structural cardiac anomalies were classified (class 1–4) according to a scheme that incorporates postoperative physiology (single versus biventricular physiology) and presence or absence of preoperative aortic arch obstruction (Table 2). This diagnosis classification has been shown to be a predictor of in-hospital mortality in neonates who undergo surgery for CHD.¹⁰

The chart of each patient with preoperative SPC was reviewed by 2 cardiologists, blinded to patient identity and outcome, to determine (1) whether SPC was attributable to CHD and (2) if so whether it was potentially preventable by a postnatal screening strategy for CHD. To be potentially preventable, SPC had to occur after 12 hours of life to allow time for screening and be attributable to undiagnosed CHD. Discrepancies in classification were resolved by consensus of the 2 reviewers.

RESULTS

Incidence of SPC in the Cohort and Subsets of the Cohort

Overall, 76 (15.5%) patient met the definition of SPC. Potentially preventable SPC occurred in 33 patients (6.7%; 95% CI: 4.7%–9.3%) and accounted for 43% of all infants with SPC. Figure 1 illustrates the compartmentalization of the cohort into subsets with SPC and potentially preventable SPC. The incidence of SPC in different subgroups of the cohort is shown in Table 4. Among patients who resided locally and were referred by consistent sources, 12.6% met the definition of SPC, compared with 18.2% of all other patients (P = .09). Nine of 237 local, consistent referrals (3.8%; 95% CI: 1.8%–7.1%) experienced potentially preventable SPC, which accounted for 27% of SPC among local, consistent referrals. The frequency of SPC was similar in infants whose CHD was diagnosed prenatally and postnatally as well as in patients with and without complete data available.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Schematic of partitioning of the cohort to identify the subset with potentially preventable SPC. CHD, Congenital heart disease; CoA, Coarctation of the aorta; D-TGA, D-Transposition of the great arteries; HLHS, Hypoplastic left heart syndrome; IAA, Interrupted aortic arch; PA/IVS, Pulmonary atresia with intact ventricular septum; PS, Pulmonary stenosis; SPC, Significant physiologic compromise; VSD, Ventricular septal defect.

View this table:

• View inline
• View popup

TABLE 4

Incidence of SPC in Different Subgroups of the Cohort

Severity of decompensation was examined according to the number of components of the definition of SPC met. Fifty-two patients met only 1 component of the definition; 45 (86.5%) of these qualified on the basis of preoperative acidosis. Twenty-four patients (31.6% of those with SPC; 4.9% of entire cohort; 95% CI: 3.2%–7.2%) met >1 component. Among patients who met at least 2 of the criteria, there was no difference in the proportion from local, consistent referral sources versus all other sources (P = .28); however, severity of SPC was related to timing of diagnosis (prenatal versus postnatal; Table 5). The proportion of patients who met at least 2 criteria for SPC was higher in patients whose CHD was diagnosed postnatally (7.7% vs 1.4%; P = .001). Patients with a prenatal diagnosis had earlier SPC than patients whose CHD was diagnosed after birth and typically qualified on the basis of acidosis alone in the first 12 hours of life (21 [67.8%] of 31 patients). Twenty (60.6%) of 33 patients with potentially preventable SPC met >1 component of the definition of SPC.

View this table:

• View inline
• View popup

TABLE 5

Comparison of the Features of the Patients With SPC in the Subgroups With Prenatal and Postnatal Diagnosis

A similar proportion of patients whose CHD was diagnosed prenatally and postnatally died before any intervention (Table 5). Of the 2 patients in the group with prenatally diagnosed CHD, 1 had birth asphyxia secondary to placental abruption, and the second had tetralogy of Fallot with pulmonary atresia and died of necrotizing enterocolitis, attributed to mesenteric steal from a large number of aortopulmonary collaterals. Of the group with postnatally diagnosed CHD, 3 patients with previously unrecognized hypoplastic left heart syndrome (HLHS) died after presenting in shock as a result of ductal constriction.

Timing of SPC Among Patients With SPC Attributable to Undiagnosed CHD

Timing of SPC is most relevant among patients who develop SPC attributable to undiagnosed CHD, because the timing determines the window available for screening. In 40 patients, SPC was attributed to undiagnosed CHD: 6 met the definition of SPC before 12 hours of age, 5 between 12 and 24 hours of age, and 28 after 24 hours of age (Fig 2). One patient who presented after 24 hours of age had a double aortic arch that was not diagnosed despite several echocardiograms. This patient was not considered potentially preventable by screening. Thus, 33 (83%) of these 40 patients’ SPC were classified as potentially preventable by our a priori definition. Of the 6 patients who had undiagnosed CHD and developed SPC before 12 hours of age, 2 had HLHS with intact atrial septum and 2 had D-transposition of the great arteries with poor intercirculatory mixing. Eleven (33.3%) of the patients with potentially preventable SPC were discharged to home before diagnosis of CHD.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2

Timing of SPC attributable to undiagnosed CHD (n = 40). Separate bars are shown for each CHD class. Age is categorized as <12 hours, between 12 and 24 hours, or, if after 24 hours of age, the number of days of age.

DISCUSSION

Preoperative SPC, as defined by severe acidosis, cardiac arrest, or evidence of renal, neurologic, or hepatocellular injury, occurs in a substantial proportion of patients with critical CHD, 15.5% in this study. The estimated incidence is somewhat lower among neonates who are referred from a local, consistent referral pool, suggesting the potential for some referral bias. Still, the estimated incidence among these patients was substantial (12.6%). Neonates with arch obstruction are particularly vulnerable to SPC, with the highest rate of SPC among patients with single ventricle physiology and arch obstruction (HLHS and variants).

Prenatal diagnosis of complex CHD is increasingly common and accounted for 44.3% of this cohort. Prenatal diagnosis allows proactive treatment of the neonate with CHD, including prostaglandin E₁ to maintain ductal patency if the systemic or pulmonary blood flow is ductal dependent, balloon atrial septostomy in neonates with D-transposition of the great arteries and inadequate intercirculatory mixing, and prompt relief of pulmonary venous obstruction in neonates with obstructed total anomalous pulmonary venous return or HLHS with intact or restrictive atrial septum. It is interesting that, although the overall incidence of preoperative SPC did not differ between groups whose CHD was diagnosed prenatally and postnatally, the severity of SPC seemed to be greater among patients without a prenatal diagnosis. When a more stringent definition of SPC is used (meeting ≥2 criteria), SPC becomes rare in the group with prenatally diagnosed CHD (1.4% vs 7.7% postnatal). Thus, prenatal diagnosis might be effective in preventing more extreme SPC.

Improved postnatal screening for CHD might potentially improve outcomes among neonates with complex CHD. Using our point estimate of 6.7% incidence of potentially preventable SPC among neonates with critical CHD and the estimated incidence of critical CHD of 1 per 1000 live births,^11,12 we estimate that 1 of every 15000 live births has critical CHD that potentially is amenable to postnatal screening. Even after accounting for potential referral bias (ie, using 3.8% with potentially preventable SPC), the estimated incidence of potentially preventable SPC is 1 of every 26000 live births. These estimates are solidly within the range of incidences of other disorders for which universal newborn screening is available in a majority of states.^7–9 For example, phenylketonuria (incidence 1 per 14000 live births) and galactosemia (1 per 53000 live births) both are included in newborn screening programs in all 50 states plus the District of Columbia. Given that our study would not have identified infants who died outside the hospital with critical CHD, the incidence of critical CHD that potentially is amenable to postnatal screening in the population can be expected to be somewhat higher.

The question then arises as to whether postnatal screening for critical CHD is feasible. The timing of SPC among neonates with postnatally diagnosed critical CHD seems amenable to a postnatal screening strategy that could be instituted within the first 12 to 24 hours of life. Screening at 48 hours, just before hospital discharge, is too late, because more than half of patients who develop SPC as a result of unrecognized CHD will do so before that time.

The distribution of lesions indicates that any effective screening strategy must identify neonates with aortic arch obstruction. These patients may be especially difficult to detect on physical examination because of minimal or no cyanosis and frequent lack of pathologic murmurs. The use of 4-extremity blood pressures in neonates has an unacceptable false-positive rate.¹³ Universal echocardiography would likely be prohibitively expensive, and there are insufficient numbers of trained sonographers and cardiologists to implement such a program.¹⁴ Echocardiography would be used more effectively if a simpler screening strategy could be used to select a subset of neonates at high risk for CHD.

Eight published studies examined the use of pulse oximetry as a screening tool for CHD, reporting on a total of 36112 asymptomatic neonates and identifying 28 cases of otherwise unsuspected CHD.^15–22 Among the 22 identified cases with a specified diagnosis, there were 4 cases of noncritical CHD, 15 cases of critical CHD without arch obstruction, and 3 cases of coarctation of the aorta. No cases of HLHS or variants occurred in the screened group in these studies. Two studies reported the sensitivity of their algorithms for coarctation, ranging from 3 (50%) of 6¹⁸ to 16 (94%) of 17.²² Thus, it remains to be determined how effective pulse oximetry screening would be for detection of critical CHD with arch obstruction. Other authors concurred that there are insufficient data to recommend universal pulse oximetry, despite its appeal as an easily accessible, noninvasive technology.^14,23

An alternative to postnatal screening is improved prenatal detection, which might be achievable by increasing the detection of CHD during obstetric ultrasounds that are already being performed. Such an approach would involve education programs for obstetric sonographers and obstetricians.^24,25

There are several potential limitations to this study. First, the cohort is not population based, making it liable to referral bias with potentially higher severity of disease. We performed a sensitivity analysis within our study cohort, comparing local patients from consistent referral sources with those from remote locations or from inconsistent referral sources as discussed already. This analysis indicates some evidence of referral bias in the larger cohort, but the incidence of SPC is still substantial in the restricted cohort. Second, infants who died of unrecognized critical CHD outside the hospital were not captured in our cohort, because admission to CHOP was required for inclusion. Thus, our study would underestimate the incidence of potentially preventable SPC in the population. Third, although the sample size provides adequate power for determining the incidence of SPC in the entire cohort, precision of the estimate is lost in small subsets. Fourth, with the use of existing data, some patients did not have all laboratory tests sent, and some outside charts were missing; however, we found no difference in the incidence of SPC in patients with complete versus incomplete data. In addition, missing data would result only in an underestimation of the incidence of SPC because we assumed the absence of conditions when no data were available. Fifth, there might be information bias among the subset with prenatally diagnosed CHD. These neonates were admitted to the ICU immediately after birth, and most who met the definition of SPC did so because of abnormal blood gases documented in the first 24 hours of life. Most of the patients with postnatally diagnosed CHD were not initially admitted to the ICU and therefore did not have blood gases sent during the same time frame. This bias would tend to reduce the difference in incidence of SPC between the 2 groups. Finally, these estimates might not be generalizable nationally. CHOP has a highly developed fetal diagnosis program, leading to nearly half of neonates’ CHD being diagnosed prenatally. In regions where fetal diagnosis is less common, the incidence of potentially preventable SPC might be higher.

CONCLUSIONS

Preoperative SPC, as defined in this investigation, occurs in a substantial proportion of neonates with critical CHD. An estimated 1 of every 15000 to 26000 live births experience SPC as a result of unrecognized CHD that would be potentially preventable by a postnatal screening strategy instituted within the first 12 hours after birth or by improved prenatal detection. This incidence is comparable to that of other disorders for which newborn screening is widely available. Newborns with aortic arch obstruction are particularly vulnerable to SPC and should be targeted by any new strategies. Future studies should define the effect of SPC on outcome for additional assessment of the potential impact of screening.