Screening for Critical Congenital Heart Defects in Sweden

Retrospective population-based, nationwide cohort study.

Cases were identified in local surgical registries at the 2 pediatric cardiac surgery centers in Sweden (Lund and Gothenburg) and in cardiac catheter registries in Lund, Gothenburg, and Stockholm (the only centers performing catheter interventions in newborns). Deaths from undetected CCHD were identified from the Cause of Death Register (The National Board of Health and Welfare).¹⁸  Clinical data were collected from medical records. The total number of live-born full-term infants in the background population were retrieved from Statistics Sweden (www.scb.se) and The National Board of Health and Welfare.

In Sweden, 97% of pregnant women undergo prenatal ultrasound screening at 18 to 20 weeks of pregnancy.¹⁹  Most centers used outflow views and the 3-vessel and tracheal view in addition to the 4-chamber view of the fetal heart during the study period, although International Society of Ultrasound in Obstetrics and Gynecology guidelines^16,17  were not recommended until 2016.

All maternity units routinely use a POS protocol measuring in the right hand and in 1 foot before 24 hours of age or before discharge if discharge occurs earlier.^15,20  A single reading with oxygen saturation ≤90% or 3 consecutive readings <95% in the right hand and foot or a >3% hand-foot difference is considered a positive screen, calling for referral for immediate echocardiography.

During the complete study period, at least 1 NPE was performed by a pediatrician after POS and before discharge.

CCHD was defined as CHD requiring surgery or catheter intervention, or resulting in death without such treatment, within 28 days of birth. We included all live-born full-term infants satisfying this definition born in Sweden between January 1, 2014, and December 31, 2019 (N = 649).

We excluded cases with isolated patent ductus arteriosus and 17 cases with miscellaneous diagnoses not usually included in studies on CCHD and in whom early surgery was to some extent elective. These diagnoses were vascular ring (N = 4), ventricular septal defect (VSD) (N = 4), atrioventricular septal defect (AVSD) (N = 1), aortopulmonary window (N = 3), congenitally corrected transposition of the great arteries/VSD (N = 1), cor triatriatum/partial anomalous pulmonary venous return (N = 1), aortico-left ventricular tunnel (N = 1), ductal aneurysm (N = 1), and disconnected left pulmonary artery (N = 1). Two further cases were excluded because of missing POS data (they were, however, included in the live-birth prevalence calculations), leaving 630 cases for analysis. Their diagnoses were: hypoplastic left heart syndrome (HLHS), interruption of the aortic arch, coarctation of the aorta (CoA), aortic stenosis, unbalanced AVSD, double inlet left ventricle, single ventricle, tricuspid atresia, Ebstein anomaly, pulmonary atresia with intact ventricular septum, pulmonary atresia (PA)/VSD, Tetralogy of Fallot (ToF), double outlet right ventricle, pulmonary stenosis (PS), truncus arteriosus, transposition of the great arteries (TGA), and total anomalous pulmonary venous return (TAPVR).

Cases with more than 1 lesion were grouped according to the lesion considered to be the dominant lesion. Infants with TGA were subdivided into simple or complex (ie, those with significant associated defects). Likewise, infants with CoA were subcategorized into isolated and complex. TAPVR was categorized as obstructed or unobstructed. ToF included 2 neonates with absent pulmonary valve syndrome.

The primary outcome variables were the relative contributions of prenatal screening, POS, and NPE to the early diagnosis of CCHD. The contribution of prenatal screening was defined as the number of cases diagnosed before birth, divided by the total number of cases in the cohort. The contribution of POS was defined as the number of cases diagnosed as a result of POS, divided by the total number of cases in the cohort. The contribution of NPE was defined as the number of cases detected at NPE, divided by the total number of cases in the cohort. Live-birth prevalence of CCHD was defined as the number of cases with CCHD per 1000 full-term, live-born infants. Early detection rate was defined as the proportion of all cases with CCHD that were diagnosed before birth or postnatally before discharge or death, if death occurred before discharge. Circulatory failure was defined as need for intensive care with metabolic acidosis (pH <7.26) or elevated lactate (>5 mmol/L) and signs of organ dysfunction. Thirty-day postoperative mortality was defined as death within 30 days of surgery or catheter intervention.

An inconclusive echocardiogram was defined as an echocardiogram that did not result in a diagnosis of CCHD.

Medians with interquartile range (IQR) and range were used for continuous variables and percentages for categorical variables. The primary analysis is descriptive for the complete period and a secondary analysis compares the relative contributions of fetal diagnosis, POS and NPE between 2 time periods using χ² test. Significance level was set to P < .05. Statistics were performed using SPSS Statistics, version 29 (IBM Corp.)

The study was approved by the Regional Ethical Review Board in Gothenburg on February 6, 2017; registration number 1119-16. Additional extensions on the period for patient collection were approved in 2017 and 2020 (T714-17 and 2020-02762).

We identified 630 infants with CCHD satisfying inclusion/exclusion criteria (Fig 1). There were 692 981 live-born, full-term infants during the study period, yielding a prevalence of 0.91 per 1000 births (Table 1).

Among the 630 infants, 66.0% (416/630) were males. Median gestational age was 39 weeks (IQR 38-40) and median birth weight 3.4 kg (IQR 3.1–3.8). Fifty-two (8.3%) had extracardiac anomalies or chromosomal aberrations. Delivery was at 1 of the 2 surgical centers in 39.2% (247/630). Notably, with a prenatal diagnosis, the corresponding proportion was 83.0% (219/264). Prostaglandin E1 was administered to 514/630 (81.6%). Median age at first intervention was 5 days (IQR 3–8; range, 0–28). In 481, the first procedure was cardiac surgery and in 128 it was a catheter intervention, including balloon atrial septostomy (78/128).

The CCHD had been diagnosed prenatally in 41.9% (264/630) (Fig 1, Tables 2 and 3). Neonates with lesions affecting the 4-chamber view were more commonly diagnosed prenatally. For example, 78.0% (46/59) of neonates with HLHS were prenatally diagnosed, in contrast to 41.5% (44/106) of those with simple TGA and 23.4% (30/128) of those with isolated CoA (Table 3).

Seventy-four of 366 (20.2%) prenatally undiagnosed newborns developed early symptoms before POS (Fig 1). The most prevalent diagnoses among them were TGA (33/74) and TAPVR (11/74) (Table 3). Nine of 74 cases with early symptoms died in the hospital before intervention, including 3 who were not diagnosed before death (2 with TGA and 1 with CoA) (Table 4). One neonate with early symptoms was discharged after an inconclusive echocardiogram and was later diagnosed with CoA. Thus 11.1% (70/630) of the complete cohort were diagnosed before death or hospital discharge because of early symptoms.

All prenatally undiagnosed neonates who were not picked up by early symptoms or died early underwent POS (N = 292). The result was positive in 144/292 (49.3%; Fig 1, Table 3). Two neonates with a positive result of POS were discharged after an inconclusive echocardiogram, both later diagnosed with CoA. Therefore, the overall contribution of POS to early diagnosis was 22.5% (142/630). In prenatally undiagnosed and initially asymptomatic cases, POS led to a diagnosis in 48.6% (142/292).

Left-sided obstructive lesions accounted for 85.1% (126/148) of negative POS screenings. Of 325 cases with diagnoses usually considered to be the primary targets for POS²¹  (TGA, TAPVR, truncus arteriosus, ToF, PA with intact ventricular septum, PA/VSD, tricuspid atresia, and HLHS), 99 were not diagnosed prenatally or by early symptoms and therefore underwent POS with 85 positive results (85.9%). Among them were 50 cases of TGA, of whom all except 1 were detected by POS (49/50, 98.0%) (Table 3).

Of 148 cases with negative POS, 96 (64.9%) presented signs or symptoms at the subsequent NPE leading to a timely diagnosis in 86 (Table 3). Thus, the overall contribution of NPE to early diagnosis was 86/630 (13.7%). Fifty-three of 86 (61.6%) diagnosed as a result of NPE had CoA, and of those 33/53 (62.3%) had weak or absent femoral pulses.

The overall proportion with a fetal diagnosis was significantly higher during the past 3 years compared with the first 3 years (151/326; 46.3% vs 113/304; 37.2%; P = .02). No significant time trends were observed in the contribution of POS or NPE, although the contribution of POS was 31% in 2014 and 15% in 2019 (Table 2).

There were 4 deaths from CCHD diagnosed only at autopsy (0.6% of all with CCHD). This translates to 0.6 deaths because of unrecognized CCHD per 100 000 full-term newborns. Two of them had simple TGA with severe desaturation and they both died within 10 hours, before both POS and NPE. Autopsy showed intact atrial septum in both. One had isolated CoA and died within a couple of hours of birth. Finally, 1 with CoA and VSD passed all screening tests (POS and NPE at 24 and 26 hours) but died at 2 days before discharge.

Sixty-four infants (10.2%) were discharged undiagnosed. Twenty-five of them (39.1%) were readmitted in circulatory failure, out of whom 1 (with HLHS) died before surgery. The most prevalent diagnosis was CoA (56/64, 87.5%), 22/56 (39.3%) showing signs of circulatory failure on readmission including 1 infant who received cardiopulmonary resuscitation at home by parents. The other diagnoses were aortic stenosis (1), double outlet right ventricle (1), PS (3), and obstructed TAPVR (2). All 63 survived the operation (N = 60) or catheter intervention (N = 3), but 1 (CoA) died at 11 months of age secondary to neurologic sequalae. Ten infants with signs/symptoms at NPE were discharged undiagnosed. Of these, 6 had a murmur as the only finding (1 AS, 2 CoA, and 3 PS) and 1 had a murmur and weak femoral pulses (CoA). All 7 had revisits scheduled but were not subjected to echocardiography before discharge. The remaining 3 had a murmur and were discharged after an inconclusive echocardiogram and later diagnosed with CoA.

Fifty-two infants passed all screening tests having no early symptoms, negative result of POS, and no findings at NPE suggestive of CHD. All except 1 (CoA/VSD), who died before discharge (described previously), were discharged undiagnosed.

The early detection rate was 89.2% (562/630). Excluding 70 cases detected because of early symptoms, the combined detection resulting from screening (prenatal + POS + NPE) was 78.1% (492/630) (Table 3).

Preoperative mortality is summarized in Table 4 and amounted to 21 cases (21/630, 3.3%), including 9 with comfort care. The 30-day postoperative mortality was 1.6% (10/609). The combined preoperative and 30-day postoperative mortality (excluding comfort care cases) was 22/621 (3.5%).

This study is, to our knowledge, the first to examine the effectiveness of a nationally implemented program for CCHD screening with a near 100% coverage of routine prenatal ultrasound screening, postnatal POS, and NPE. We analyzed the contributions of the different screening tests, highlighting that POS and NPE continue to make important contributions to early diagnosis by picking up cases missed by prenatal screening. With an overall 89% early detection rate using combined screening methods, our findings underscore the need for improved early detection of CoA.

The prevalence of CCHD was 0.91 per 1000 full-term, live-born infants, lower than in previous reports. Differences in reported prevalence rates can be a result of differences in prenatal detection and termination rates. In a previous population-based study from our region between 2014 and 2016, the fetal incidence of CCHD, including 33% terminations, was 1.22 per 1000.¹⁷  Some studies, in contrast to ours, include preterm infants in the analysis. Furthermore, different definitions of CCHD are used across studies, making direct comparisons challenging.

Studies from the United States and China have reported reductions in infant death rates after the introduction of postnatal CCHD screening.^11,22  In our study, only 1 infant (HLHS) died after having been discharged undiagnosed (death after readmission and before surgery). This is an improvement compared with previously reported results from Sweden during a period before POS was implemented.²⁰  Nevertheless, some of the 12 deaths in prenatally undiagnosed newborns might have been possible to prevent had the diagnosis been known prenatally enabling optimal perinatal management. Prenatal detection rates are continuously improving and some programs have reached a near 100% detection of some lesions.^3,17,23,24  In contrast, TGA, TAPVR, and especially CoA are more difficult to detect prenatally,^17,24,25  as found also in our study. Prenatal detection of CoA is improving but achieving a very high prenatal detection rate with acceptable false-positive rates is likely not feasible considering that CoA can occasionally develop postnatally when the ductus arteriosus constricts.^26–28 

A similar national population-based study was conducted in New Zealand in 2016, before the introduction of POS, reporting a higher prenatal detection rate of CCHD than in our study (123/194, 63%).²⁴  Still, 14% of CCHD cases were discharged undiagnosed. In live-born infants with TGA, 40/67 (60%) had a prenatal diagnosis compared with 65/150 (43%) in our study.²⁴  Ten infants with TGA (15%) were discharged undiagnosed compared with no cases in our cohort. The authors concluded that POS should be introduced despite high prenatal detection rates. In TGA, prenatal diagnosis remains particularly important because many newborns with simple TGA will require balloon atrial septostomy soon after birth. Prenatal diagnosis enables the centralization of deliveries to a hospital with the capacity to perform this intervention if needed.

Because a sizeable proportion of cyanotic lesions was detected prenatally and not subjected to POS, the overall contribution of POS to early diagnosis of CCHD was 23%. In comparison, Meberg et al and Riede et al reported 31% (25/81) and 16% (14/90), respectively.^13,29  The relative contribution of POS to the early detection of CCHD will be lower in countries with a well-developed prenatal screening program, as observed by others.^30–33  We found that 49% of newborns with CCHD subjected to POS had a positive result, which is a lower proportion than in some other reports.^9,20,34  The result of POS will depend on case mix, which is influenced by degree of prenatal detection and selection during pregnancy as well as on population differences in prevalence of various defects. In our study, 45% of CCHD patients subjected to POS had CoA, which is not considered a primary target for POS. A better way of comparing results between different studies is the sensitivity of POS for specific cardiac defects. For example, the sensitivity of POS for simple TGA was 97% in our study, for HLHS 50%, and for isolated CoA 17%, which compares well with other reports.^12,34,35 

POS algorithms vary in some aspects including timing, site of measurement, and cutoffs.³⁶  Earlier testing can result in higher false-positive rates, whereas later screening could increase the risk of clinical deterioration before diagnosis.⁹  As demonstrated in this study, 20% of postnatally diagnosed cases developed symptoms (or died) before POS, in almost all within 3 hours of birth. In Sweden, screening occurs within 24 hours, but not before 6 hours. Given the increasing trend of early discharges, it is likely that earlier POS will be implemented. In a UK study on early screening (4-8 hours), false positives decreased between 2 different study periods, which was attributed to increasing experience using POS.³⁰ 

Early detection of aortic arch obstructions continues to be the central remaining issue in CCHD screening. Not only is the sensitivity for CoA low both in pre- and postnatal screening, but early echocardiography before full ductal closure may also fail to detect CoA in some cases as shown in this and other studies.^12,25,37  Our study underscores the importance of careful NPE including predischarge palpation of femoral pulses, with weak or absent femoral pulses being the primary indication for CoA diagnosis in 62% of NPE-detected cases.^20,38,39  If there is uncertainty about femoral pulses, comparison with the right axillary pulse should be made.³⁸ 

The prenatal and postnatal detection rates of CoA have been shown to improve by using prediction models.^40,41  These should be evaluated in larger cohorts. Adding other postnatal screening parameters, such as perfusion index, already displayed on most pulse oximeters, has the potential to detect some newborns with CoA passing current screening methods, but this needs further evaluation.^42,43  Finally, an additional postdischarge examination, perhaps coordinated with Guthrie testing, at 2 to 5 days of age, including palpation of femoral pulses and ideally even a repeat POS and PI, could possibly help improve the timely diagnosis of CoA before clinical deterioration occurs.

A major strength of this study is its national population-based design with complete ascertainment of cases undergoing surgery or catheter intervention, as well as preoperative deaths and out-of-hospital deaths diagnosed only at autopsy. A limitation is the lack of a national registry in which exact results of POS are recorded for all newborns. Therefore, the false-positive rate of POS could not be estimated.

Our findings indicate that POS and NPE are still important methods, in addition to prenatal screening, to detect CCHD. These methods should be combined as they offer complementary strengths. However, in CoA, both pre- and postnatal detection rates are lower compared with other CCHD. Infants with CoA are often discharged undiagnosed, frequently developing circulatory failure before surgery. Therefore, future research on pre- and postnatal screening for CCHD should pay particular attention to this cardiac defect.

The authors thank all colleagues who contributed data on their patients.

AVSD

atrioventricular septal defect

CCHD

critical congenital heart defect

CHD

congenital heart defect

CoA

coarctation of the aorta

HLHS

hypoplastic left heart syndrome

IQR

interquartile range

NPE

neonatal physical examination

PA

pulmonary atresia

POS

pulse oximetry screening

PS

pulmonary stenosis

TAPVR

total anomalous pulmonary venous return

TGA

transposition of the great arteries

ToF

Tetralogy of Fallot

VSD

ventricular septal defect