Comparing Strategies for Critical Congenital Heart Disease Newborn Screening

In 2013, a pilot community-based project for newborn CCHD screening using pulse oximetry was launched in Taipei,⁵  representing 42% of the total deliveries in Taipei. Between April 1, 2014 and June 30, 2017, the newborn screening program was expanded to 30 birthing facilities, representing 73.7% of birthing facilities and 89.0% (94 239 of 105 869) of the total deliveries in Taipei. Following the same protocol of the newborn screening program and referral criteria,⁵  all newborns delivered in these facilities were eligible for inclusion in this study.

The screening algorithm used in this study was based on that of the previous study.⁵  Pulse oximetry was performed by pediatricians or nurses experienced in routine pulse oximetry measurement for newborns aged 24 to 36 hours. Screening was performed earlier or postponed if a newborn had abnormal symptoms or unstable conditions. Functional SpO₂ was obtained in the right hand and in either foot; a value of ≥95% in either extremity with an absolute difference of ≤3% between SpO₂ in the right hand and either foot was considered a pass result. A screening result was considered “fail” if (1) a value of SpO₂ between 90% and 94% was found in both extremities or a difference of >3% was found in the second and third measurements with 30-minute intervals, or (2) a value of <90% was found in either extremity in the second or third measurement. These criteria were applied to all live births regardless of health status and care setting.

If a newborn has a positive screening result, he will receive a clinical examination first and be referred for an echocardiography as soon as possible. Twelve of the 30 birthing facilities (40%) had pediatric cardiologists and onsite echocardiography. Three of the 8 medical centers capable of further management were referral hospitals. The referral hospitals were responsible for the timely submission of follow-up data on screen-positive infants to the CCHD database. The relevant prenatal diagnosis, findings of the physical examination, and final diagnosis were included in the confirmation report.

All 30 birthing facilities submit data, including demographic data, test results, and referral information, online in a timely manner in the CCHD database maintained by the Taipei Institution of Pathology. Other information, including prenatal examinations, suspicious symptoms associated with CCHD, and the use of oxygen supplementation or mechanical ventilation were also documented in this database. When submitting SpO₂ values to the CCHD database, the result (pass, rescreen, or fail) was determined automatically to avoid judgment mistakes. After cross-verification with the birth certificate registry system and newborn metabolic disease screening database, public health nurses followed up on these unscreened newborns, including missing and refused cases, within 1 month. Those live births with prenatal suspicion of congenital heart disease or CCHD would be excluded to investigate the test accuracy and efficiency of the Taipei protocol. The general characteristics of enrolled infants, such as sex, full-term or preterm, and maternal nationality, are presented in Table 1. This study was approved by the Research Ethics Committee of Taipei City Hospital (reference: TCHIRB-10412126-E).

We used all infants who received oximetry screening (including those with prenatal suspicion) in the database to retrospectively evaluate outcomes on the basis of 9 referral criteria of 10 different screening protocols or recommendations published previously, including AAP recommendation⁴  and AAP updated strategy,¹³  the protocol in a multicenter study in China,^14,15  the New Jersey statewide screening program,²  the Tennessee algorithm,^16,17  the Germany recommendation,¹⁰  the Polish program,¹¹  the Nordic proposed algorithm,⁷  the UK algorithm,^8,18  and the Spanish recommendations.¹⁹  The outcomes included test results, referral rate, CCHD detection rate, false-positive rate, and the presence of false-negative cases.

Between April 2014 and June 2017, 94 239 live births were reported in these birthing facilities in Taipei, and 98.8% (n = 93 093) of infants eligible for screening underwent pulse oximetry screening. Excluding cases with prenatal suspicion (n = 35), the screening results of 93 058 infants are shown in Table 2. The remaining infants were not screened because they died before screening (0.23%), the parents refused screening (0.22%), or the screening was missed (0.77%; Table 2). Through public health tracking and cross-verification with the death registry system, the common causes of death for those who died before screening included extreme preterm births and congenital anomalies, but not CCHD. Among the newborns screened, 355 (0.38%) required a second oximetry measurement (Fig 1); of those, 32.7% (116/355) had SpO₂ <90% in either site in the first measurement. Moreover, 184 of 355 newborns (51.8%) did not pass the second measurement conducted 30 minutes later; of those, 50.5% (93/184) had SpO₂ <90% in either site and were referred for further evaluation immediately. Ninety-one infants received a third oximetry measurement 30 minutes later and 28 of them received a passing assessment. In total, 156 (0.17%) newborns received a failed assessment result. Of those who underwent pulse oximetry screening, 92.6% received complete screening between 24 and 48 hours after birth, and 1572 (1.69%) infants received oximetry measurements within 24 hours of life (Table 3). All 156 newborns with a failed assessment result were referred for further evaluation. The efficiency of referrals is presented in Table 3. The referral rate of these newborns within 48 hours of life in this study was 87.82%. Of referral cases, 117 (75.0%, 117/156) had a final diagnosis before the age of 3 days. Diagnoses other than CCHD in referral cases included respiratory problems (n = 58), other CHD (n = 41), sepsis (n = 2), and others (n = 3; Fig 1 and Supplemental Table 7). Neither abnormal findings nor diseases were found in 10 of the referral cases. The false-positive rate of CCHD screening was 0.12% (114/93 093).

A total of 75 infants were confirmed to have CCHD diagnoses, and 15 of them were preterm. The incidence of CCHD was 0.081% (75/93 093). Of all CCHD cases, 33 had prenatal suspicion of CCHD. Excluding the CCHD cases with prenatal suspicion, 97.6% (41/42) had a diagnosis at an age <72 hours (data not shown). The first 3 common diagnoses were transposition of great arteries, pulmonary atresia, and tetralogy of Fallot (Table 4). Twelve infants died after surgery or because of a refusal of further treatment by the parents, and one-half of them were preterm. The mortality rate of total CCHD cases was 16.0% (Table 4). The mortality rates of CCHD cases detected by oximetry only or by prenatal examinations and oximetry were 16.7% and 15.1%, respectively. Of enrolled live births without prenatal suspicion, 4 infants who passed screening had a final diagnosis of CCHD. The false-negative cases are presented in Figure 1. Three cases passed screening at the second measurement, 1 had SpO₂ <90% in the foot at the first measurement, and 1 case passed the screening at the third measurement. The diagnoses in these 4 false-negative CCHD cases included 2 coarctations of the aorta, 1 transposition of great arteries, and 1 tetralogy of Fallot. The known false-negative rate in this study was 0.004%.

To validate the efficacy and applicability of the Taipei protocol used in this study, all live births (including those with prenatal suspicion) enrolled in this study were used to evaluate outcomes on the basis of 9 different referral criteria of 10 screening protocols or recommendations published previously.^{2,4,7,8,10,11,13–15,17,19}  The outcomes are presented in Table 5. The estimated referral rate ranged from 0.23% to 0.31%, and the estimated lowest false-positive rate was between 0.07% and 0.24% (Table 5 and Supplemental Table 6). There were 2 additional false-negative cases that passed the first test in the Taipei protocol if using all enrolled live births (including those with prenatal suspicion). The CCHD diagnoses of these 2 cases were tricuspid atresia and double outlet of the right ventricle. Fewer CCHD cases were detected by using the foot-first¹⁷  or foot-only screening program^10,11  (Table 5). Applying our database to these different referral criteria, the most CCHD cases would be detected by the Nordic program,⁷  updated AAP strategy,¹³  and Spanish recommendation¹⁹  at the expense of more referral cases and the subsequent cost of transfer and diagnosis (Table 5).

After excluding those live births with prenatal suspicion, the referral and false-positive rates of pulse oximetry CCHD screening in this study were 0.17% and 0.12%, respectively. In this study, 92.6% of enrolled newborns were screened within 24 and 48 hours after birth. Among the referral cases, 87.8% were referred for diagnosis within 48 hours of life. The coverage rate in this study was 98.8%. These results are compatible with those in our previous pilot study.⁵  Other studies have revealed a varied referral rate from as low as 0.06% to 12.6% and a varied false-positive rate from 0.05% to 0.80% because of different screening criteria and protocols.^2,8,9,12  The sensitivity of pulse oximetry screening of total cases and after excluding cases with prenatal suspicion in our study were 92.6% and 87.5%, respectively, which is higher than previous reports (∼65%–85%).^1,15,19  Our results supported that using pulse oximetry for CCHD screening is a feasible method with acceptable sensitivity.²⁰ 

In our study, the incidence of CCHD in total live births in Taipei was 0.08%, which is higher than that (0.06%) in a meta-analysis of studies involving healthy late preterm and term infants.³  This difference may be due to the following reasons. Firstly, our cohort of live births included preterm births, which have a higher incidence of severe CHD than term births.²¹  Secondly, approximately 24% (18/75) of total CCHD cases were referred prenatally from other cities to the medical centers in Taipei because of abnormal results of prenatal examinations (data not shown). With a positive screening result, 117 referral cases (75.0%, 117/156), including 41 CCHD cases (97.6%, 41/42) that had no prenatal suspicion, received confirmatory examinations within 72 hours of age (data not shown). The mortality rate of CCHD cases detected by oximetry seemed to be higher than those detected by prenatal examinations (Table 4). However, the mortality rate was similar if excluding those CCHD cases who were born preterm (10.2% vs 9.6%). Compared with previous studies under a hospital-based circumstance,^9,22,23  our results proved that the Taipei protocol, which integrated both screening and diagnostic aspects, is efficient and effective in CCHD diagnosis of referral cases under multicenter or community-based setting and would be also practicable for nationwide CCHD screening in Taiwan.

The screening algorithm also detected noncardiac illness in 0.068% (63/93 058) of all infants enrolled. In addition to other congenital heart diseases, respiratory disorders and sepsis were the major diagnoses of false-positive cases (60/114, 52.6%). CCHD screening using pulse oximetry provided an additional advantage of significant noncardiac illness with a true hypoxic condition as secondary targets^7,8,24,25  and led to early intervention and better outcomes for infants.^26,27  Four CCHD cases passed pulse oximetry screening in this study (false-negative rate, 0.004% Fig 1). According to our screening algorithm, newborns with SpO₂ <90% initially in either extremity would undergo a second test after 30 minutes. One false-negative case that had a positive result in the first measurement due to SpO₂ <90% in the foot passed the second test (Fig 1).

Up to the present, the optimal screening algorithm for newborn CCHD screening is unclear.⁸  The most common discrepancies among the different algorithms and recommendations are measurement times and sites. The Spanish recommendation,¹⁹  the U.K.,¹⁸  and China study¹⁴  suggested 1 retest was needed. The site of testing also differed between studies; most studies have used both preductal and postductal SpO_2,^2,7,19  but some have used postductal SpO₂ only.^10,11  There was also a small difference of acceptable SpO₂ value between protocols (Table 5). We examined the efficacy of our current protocol and estimated outcomes by applying the referral criteria of other screening protocols/recommendations to this database (Table 5). By using the referral criteria of AAP original recommendation⁴  and New Jersey²  program, more CCHD cases were detected although there were higher referral rates and false-positive rates than our protocol. Based on Nordic referral criteria,⁷  which suggested first SpO₂ ≤90% as fail, there were more CCHD cases detected than the AAP recommendation,⁴  New Jersey program,²  and our protocol. The AAP updated strategy,¹³  released in 2020, as well as the Spanish recommendation,¹⁹  suggested an SpO₂ of at least 95% in both the upper and lower extremities as a pass result and only 1 repeat screen for cases that neither pass nor fail initially. According to our estimated results, the updated AAP strategy¹³  and Spanish recommendation¹⁹  seemed to be the most suitable newborn CCHD screening protocols.

Instead of measuring SpO₂ in both the right hand and 1 foot, some studies have suggested measuring SpO₂ in 1 foot first,¹⁷  or only,^10,11  on the basis of the rationale of differential cyanosis in ductal-dependent CCHD. A single-extremity staged screening for CCHD was initiated in Tennessee in 2012, with the first checkpoint of SpO2 <97% in 1 foot.¹⁷  The authors of the Tennessee study suggested screening SpO₂ at either foot first to eliminate “unnecessary” pulse oximetry determinations. The authors of the Polish program¹¹  and Germany recommendation¹⁰  suggested SpO₂ in 1 foot only, and the checkpoint was SpO₂ <95% and 96%, respectively. In this database, 2 more CCHD cases were missed when using the Tennessee algorithm¹⁷  and German recommendation,¹⁰  and 7 more using the Polish program.¹¹  A review of the SpO₂ records in this database revealed that 5.3% (4895/93 093) of screened cases had a lower SpO₂ of the right hand (<97%), whereas SpO₂ in the foot was ≥97%. Through reviewing our database, there was a similar proportion of screened cases with a first test result of ≥97% in 1 foot as well as <97% in the right hand in the medical centers, regional hospitals, and district hospitals (4.99%–5.79%). Three of 75 CCHD cases (4.0%) had a first test result of ≥97% in 1 foot. This finding suggested that a single-extremity screening is insufficient.

Our estimated results supported the Spanish recommendation¹⁹  and the AAP updated algorithm¹³  (Table 5). In addition, 1 more CCHD case would be detected if cases with SpO₂ ≤90% in the first measurement were immediately referred (Nordic proposed criteria).⁷  Therefore, the Taipei protocol is suggested to be modified by revising the definition of the passing assessment as SpO₂ ≥95% in both extremities with a difference of ≤3% between SpO₂ in extremities (Fig 2). The modified screening strategy had 3 points that were different from the previous Taipei protocol: (1) immediately refer if the SpO₂ in either extremity is ≤90% in the first measurement, (2) retest if the SpO₂ in either extremity is <95% or if the difference between SpO₂ in the extremities is >3%, and (3) abolish the third measurement. With these modifications, we aim to increase sensitivity and decrease the false-negative rate at the expense of an acceptable increased referral rate (compared with the original Taipei protocol, the modification referred an additional 76 cases and detected 4 additional CCHD cases; Table 5).

This study has several limitations. First, it needed further study to investigate the pathophysiology or predisposing factor which could explain the finding that ∼5% of screening cases had lower SpO₂ in the right hand than that in either foot. Second, the number of false-negative cases may have been underestimated because of the difficulty in follow-up. Third, it is a limitation of this study that we did not have the full comparable data to all protocols, such as enrolled criteria (all live births, newborns with gestational age ≥34 weeks, or asymptomatic newborns) and the time window of screening.

In conclusion, our findings indicate that pulse oximetry is an effective and feasible tool for universal neonatal CCHD screening in developing countries. The Taipei protocol is an efficient and effective protocol for CCHD screening in a community setting. Other screening protocols that measured SpO₂ in either foot only or first seemed less useful for CCHD detection. Our results support the Spanish recommendations¹⁹  and the AAP updated strategy.¹³  In addition, we suggested that the pass criteria in the Taipei protocol should be modified as SpO₂ ≥95% in both extremities (the right hand and 1 foot) with a difference of ≤3%. Immediate referral on measuring SpO₂ ≤90% in either extremity and abolishing the third measurement could reduce the false-negative rate and burden on clinical practice. The proposed revisions to the protocol would enable effective nationwide CCHD screening in Taiwan and could be used by other regions with similar medical care systems.

We thank all the staffs of the 30 birthing facilities that participated to conduct and collect screening data through the project. This project was partially funded by United Way of Taiwan. We thank Yu-Ling Liu, Yuan-Fang Chung, Yi-Ya Wang, Li-Ju Lin, and Lily Wang for data collection and validation.

AAP

American Academy of Pediatrics

CCHD

critical congenital heart disease

SpO₂

oxygen saturation