Newborn Pulse Oximetry Screening at a Community Hospital: An 8-Year Experience

A retrospective cohort study was performed to evaluate the POS results for newborns over an 8-year period between September 1, 2012, and September 1, 2020, at Holy Cross Hospital in Silver Spring, Maryland. The study included newborns admitted to the well-infant nursery who received routine POS. Newborns who did not receive routine POS (eg, parent refusal, missed screen, died before the screen, transferred to another hospital before 24 hours) and newborns who were initially admitted to the NICU were excluded from the study. Newborns who received an echocardiogram before POS because of a prenatal diagnosis of CCHD or because they developed symptoms that led to a cardiac workup before POS were also excluded. The study was classified as exempt by the Children’s National Hospital and Holy Cross Hospital Institutional Review Boards.

OZ (OZ Systems, Arlington, TX) is an electronic platform used by the Maryland Department of Health (DOH) for tracking newborn screening results, including those for POS. Holy Cross Hospital began entering all POS data into OZ in September of 2012. Newborns with a failed POS were identified through OZ, and a medical chart review was conducted on this subgroup. A medical chart review was also performed for newborns reported as requiring a second rescreen to pass POS. All initial echocardiograms were performed at Holy Cross Hospital and interpreted by a pediatric cardiologist. If the newborn who failed POS was transferred to Children’s National Hospital for further evaluation and treatment, an additional chart review was performed in that hospital’s medical record.

Oxygen saturation, in-hospital management, clinical diagnosis, and survival to discharge data were collected. The oxygen saturation data were reviewed to ensure patients classified as failing POS in OZ had been categorized correctly per the AAP algorithm. Additional demographic information was also collected, including gestational age, sex, and birth weight. To consider the potential effects of health disparities, race, ethnicity, and insurance type and/or status were collected. The mother was asked to self-select race and ethnicity at registration, which was added to her demographic information, and that of her newborn; this information was used for data collection.

Newborns who had an echocardiogram because of a prenatal diagnosis of CCHD or the development of symptoms before routine POS were coded as “physician override” in OZ and were excluded. Some newborns who passed POS had an echocardiogram performed because of a murmur or development of clinical symptoms. The OZ charts of patients with a passing POS but abnormal echocardiogram were also reviewed to identify any CCHD cases (ie, false-negative results). Children’s National Hospital has worked with the Maryland CCHD Advisory Council and Maryland DOH Newborn Screening Follow-up Program on regional false-negative surveillance, with the goal of evaluating why any infant with CCHD is not identified before hospital discharge. This surveillance includes working with the 3 major cardiac referral centers in the region as well as investigating any deaths that occur because of undetected CCHD. The Newborn Screening Follow-up Program is notified of CCHD identified on birth or death certificates by the Birth Defects Reporting System.⁹  Of the newborns born at Holy Cross Hospital, 98.8% of the families were from Maryland (96.4%) or the District of Columbia (2.4%).

For the purposes of this study, CCHD was defined as any of the POS core conditions, which includes hypoplastic left heart syndrome, pulmonary atresia, tetralogy of Fallot, total anomalous pulmonary venous connection, d-transposition of the great arteries, tricuspid atresia, truncus arteriosus, aortic coarctation, Ebstein anomaly, double-outlet right ventricle, interrupted aortic arch, single ventricle (not otherwise specified), and other critical cyanotic lesions not otherwise specified.⁴  Secondary conditions identified by POS are non-CCHD and other noncardiac conditions, such as infection (including sepsis), hypothermia, hemoglobinopathies, lung disease, persistent pulmonary hypertension of the newborn, and other hypoxemic conditions not otherwise specified.⁴ 

For newborns who failed POS but were not identified as having CCHD, diagnoses were determined on the basis of review of echocardiography reports and medical progress notes. Newborns were classified as having a non-CCHD on the basis of their echocardiographic diagnosis plus required cardiology follow-up. Normal newborn findings, such as a patent foramen ovale, small atrial septal defect, and patent ductus arteriosus, were defined as normal for the purposes of this study. One newborn with a discharge diagnosis of delayed transitional circulation was classified as having mild persistent pulmonary hypertension in this study (discharged at 4 days of life because of a NICU stay for persistent mild hypoxia with predominantly right to left shunting at the patent foramen ovale).

Descriptive statistics, including demographics, insurance type, and number and percentage of CCHD and secondary conditions were reported. The data were compiled, and basic descriptive statistics were performed through Microsoft Excel (Microsoft Corporation, Redmond, WA).

Over an 8-year period, 65 414 newborns were admitted to the well-infant nursery at Holy Cross Hospital. Of those newborns, 65 403 were eligible for screening after we excluded 5 families who refused screening and 6 infants who died before 24 hours of life. Of the eligible births, 99.0% of infants (n = 64 780) received POS, with <1% missing screening (0.9%, n = 611). Twelve infants were reported as “physician override,” indicating POS was not done because an echocardiogram was performed before 24 hours of life (Fig 1). There was correct provider interpretation of the AAP algorithm for all failed screens; however, 3 newborns had a cardiac evaluation with an echocardiogram after having only one rescreen rather than the recommended 2 rescreens per the current AAP algorithm.

Thirty-one newborns (0.05%) failed POS, a fail rate of 4.6 per 10 000. All the newborns who failed their POS were found to have either congenital heart disease or another medical condition. Of the fails, 38% (n = 12) of newborns were diagnosed with a CCHD, 29% (n = 9) with a non-CCHD, and 32% (n = 10) with a noncardiac condition (Table 1). The diagnoses of CCHD included total anomalous pulmonary venous return (n = 4), coarctation of the aorta (n = 3), pulmonary atresia (n = 1), d-transposition of the great arteries (n = 1), double-outlet right ventricle (n = 1), interrupted aortic arch (n=1), and critical pulmonary stenosis (n = 1).

For newborns who failed POS, 55% (n = 17) were male and the mean gestational age was 39 weeks. The median age at POS was 26.5 hours of life (range 24–44 hours). Of the newborns diagnosed with CCHD, reported race and ethnicity included 2 white non-Hispanic newborns, 4 Black non-Hispanic newborns, one Asian or Pacific Islander newborn, and 4 Hispanic newborns with no race listed. Five of the mothers had Medicaid, and 7 had private insurance (Table 2). There was not a significant difference in race and ethnicity or type of insurance between mothers of newborns with CCHD who failed POS and the general population of mothers who delivered at this hospital during the study time frame (P = .99, P = .76).

The non-CCHDs that were identified included atrioventricular canal defect (n = 1), ventricular septal defect (n = 2), large atrial septal defect (n = 1), mild pulmonary stenosis (n = 2), right-sided heart enlargement (n = 1), and a dilated, hypertrophied right ventricle with small aortic isthmus and decreased left ventricular systolic function (n = 1) (Table 1). The newborn with a small aortic isthmus and decreased left ventricular systolic function had normalization of function and aortic isthmus size on an echocardiogram performed before discharge. Secondary conditions identified included congenital pneumonia (n = 2); meconium aspiration (n = 1); transient tachypnea of the newborn, requiring oxygen (n = 2); and persistent pulmonary hypertension of the newborn (n = 5) (Table 3).

All newborns with a failed POS survived until hospital discharge. The newborns diagnosed with CCHD by POS were all transferred to the nearest pediatric cardiac center, and there were no mortalities during the initial hospitalizations.

On the basis of chart reviews of newborns with abnormal echocardiograms before discharge, tracking from the nearest pediatric cardiac center, and confirmation with the Maryland DOH CCHD Newborn Screening Follow-up Program’s false-negative surveillance initiative, there was 1 infant identified as having a false-negative result. The newborn with a false-negative screen result passed POS with saturations of 99% and 100% but was found at 31 days of life to have a critical aortic coarctation.

In this cohort, POS had a specificity of 99.97% with a false-positive rate of 0.03% for CCHD. Assuming accuracy of the false-negative surveillance program, the sensitivity of POS was 92.3%. The positive predictive value of POS for CCHD was 38.7%, and the positive predictive value of POS for all types of congenital heart disease was 67.7%.

The data were analyzed to evaluate the impact of the recommended revised pulse oximetry algorithm. All the newborns who failed POS under the existing AAP algorithm would have also failed under the revised algorithm. Twelve of 31 newborns (39%) who failed their pulse oximetry screen required 2 rescreens in the AAP algorithm (Fig 2). If the revised algorithm’s “no second rescreen” were applied, these 12 patients, including 3 newborns with aortic coarctation, would have failed and been evaluated earlier. In addition, 5 more newborns would have failed POS, requiring additional evaluation. One of these 5 newborns was later identified to have dehydration and fever, requiring transfer to the NICU. The other 4 newborns were not identified as having any other conditions before hospital discharge. In this cohort, eliminating the second rescreen would have resulted in a specificity of 99.97% and a false-positive rate of 0.04% (4 per 10 000).

POS for newborns has been adopted in all 50 states and the District of Columbia. However, long-term data on the outcomes of POS are limited, and little is known about the secondary conditions identified by POS in the United States. This article represents one of the largest longitudinal studies of POS outcomes in the United States. To our knowledge, it also covers the longest time period of published POS data in the United States. In our study, we found POS to be an effective tool to identify both CCHD and secondary conditions. Many of these secondary conditions could lead to serious complications if not identified before newborn discharge.

In this study, 0.05% (4.6 per 10 000 newborns) of newborns who received POS in the newborn nursery failed the screen, a rate similar to data published previously from a tertiary care birth center in Atlanta (0.04% fail rate, 4.4 per 10 000 newborns) and statewide data from New Jersey (0.06% fail rate, 6.7 per 10 000 newborns).^8,12  In contrast, the fail rate in 6 maternity units in the United Kingdom was considerably higher at 0.8%(80 per 10 000 newborns), with a higher false-positive rate, likely because of their common practice of screening before 24 hours of life.¹³ 

POS was found to be highly specific for CCHD in this cohort. A previous Cochrane meta-analysis of 19 studies reported the same specificity of 99.9% (95% confidence interval 99.7%–99.9%).¹⁴  The false-positive rate for CCHD in our study was 0.03%, compared to 0.14% in the Cochrane review and 0.04% in a recently published study from an Atlanta hospital.^8,14  POS sensitivity was found to be 92.3%, which is higher than previously reported sensitivities; the Cochrane review reported a sensitivity of 74.3% (95% confidence interval 69.5%–82%).¹⁴  However, our data are limited by the false-negative results identified at the nearest pediatric cardiac center and through Maryland DOH false-negative surveillance program. There may have been false-negative results that were not detected by these methods, although we expect this system is reasonably accurate in tracking false-negative results.

In previous studies, researchers found that POS had limited ability to detect additional cases of CCHD, likely because of high prenatal detection rates.^8,15,16  It is estimated that between 40% and 70% of diagnoses of CCHD in Maryland are made prenatally.^17,18  In our study, POS detected 12 newborns with CCHD that had not been identified prenatally, a higher find rate and positive predictive value than in previous studies.^8,9,16  This indicates that POS is valuable even in regions with high fetal diagnosis rates, as demonstrated in Saxony, Germany, by Riede et. al.¹⁹  Health care disparity factors, including insurance status, race, and ethnicity, were evaluated in mothers of newborns with CCHD who failed POS. There was no significant difference found in race and ethnicity or type of insurance when compared with the population of mothers who delivered at this hospital. Mothers with both private insurance and Medicaid had newborns who failed POS. However, potential health care disparities would be better assessed in future studies with larger cohorts of newborns diagnosed with CCHD by POS.

All newborns with CCHD detected by POS were able to be safely transferred to a pediatric cardiac center and eventually discharged from the hospital. There were also no mortalities in those newborns with a secondary condition who failed POS. This is consistent with a previously published study that revealed that mandatory policies for POS was associated with decreased infant cardiac deaths.³ 

Previous studies have given rise to concerns about false-positive results from POS increasing clinical burden and echocardiograms.^14,20  However, in our study, false-positive results identified important non-CCHD and 10 noncardiac conditions. These secondary conditions could be harmful to the newborn if not diagnosed and treated in an appropriate time frame, including sepsis, pneumonia, and persistent pulmonary hypertension of the newborn. Secondary conditions identified through POS have also been discussed in other studies, including in Atlanta, where there is a high rate of prenatal diagnoses of CCHD, and in the United Kingdom, where there are more false-positive screen results, likely because of the practice of screening before 24 hours of life.^8,21,22  Our study supports the recommendation that non-CCHD conditions also should be part of public health tracking to get a full understanding of the benefits of POS.⁴ 

Lastly, we found that the elimination of the second rescreen did not add a significant burden to the medical system, leading to 0 to 1 additional evaluation a year. An additional 5 newborns would have failed (total n = 36), including 1 infant who eventually required a transfer to the NICU and who potentially could have been identified earlier. This change did not affect the high specificity (99.9%) and only minimally increased the false-positive rate from 0.03% to 0.04%. These findings are consistent with a those of a previous study from a hospital in Atlanta that also showed an identical 0.01% increase in the false-positive rate.⁸  All of the 31 newborns who failed in our cohort would have also failed under the recommended revised algorithm. Twelve of the newborns, including 3 newborns with aortic coarctation, would not have had to wait an additional hour for a rescreen to be evaluated. In newborns with ductal-dependent systemic blood flow, prompt diagnosis and initiation of prostaglandin is essential in preventing end-organ injury. This additional analysis provides timely data given that health departments and hospitals across the country are making decisions on whether to implement switching to the revised POS algorithm.

The other change in the revised algorithm of requiring oxygen saturation of ≥95% in both the right hand and foot, rather than either the right hand or foot (AAP algorithm), was not able to be assessed with the saturation data available. However on the basis of New Jersey’s previous experience using this part of the algorithm, their false-positive rate remained low at 0.06%, similar to our cohort’s false-positive rate of 0.03%, as well as that of other states.¹² 

There are several limitations to our study. Although >65 000 newborns were included, this is a single-center study, which limits generalizability. The study data were based on records from OZ and the medical chart, which may be subject to errors in documentation. POS saturations for some newborns who passed were not available for review in OZ (only pass and fail reported), so we could not confirm that the algorithm was applied correctly in every instance. When applying the elimination of the second rescreen, we were only able to identify conditions documented in the medical record and were unable to assess for underlying cardiac disease because these patients did not receive echocardiograms before discharge. Lastly, although 1 false-negative result was identified through medical chart review and the Maryland DOH surveillance program, there is a possibility that there were false-negative results after discharge that were not identified.

POS is an effective tool for identifying CCHD and secondary conditions. POS was successfully implemented at a large community hospital, with few missed screens, and was found to be highly specific in this cohort. Eliminating the second rescreen in the POS algorithm would have resulted in a minimal increase in false-positive results in this population and faster evaluation of newborns with CCHD. Future population-level studies are needed to evaluate outcomes from the recommended revised POS algorithm.

We acknowledge Nancy Nagel, Vice President for the Division of Women and Children at Holy Cross Hospital, for her efforts in developing POS at Holy Cross Hospital and her guidance during the initiation of this study. We acknowledge Lindsay Attaway for her work with the Children’s National Heart Institute’s POS program. We also acknowledge Terese Finitzo and her team at OZ Systems for their assistance in gathering data for this article and ongoing work on a platform to track newborn screening outcomes. We also acknowledge Johnna Watson from Maryland DOH for her dedication to improving POS and tracking in the state of Maryland. Lastly, we wish to acknowledge the C.R. Beyda Professorship for their funding of this project.

AAP

American Academy of Pediatrics

CCHD

critical congenital heart disease

DOH

Department of Health

POS

pulse oximetry screening