BACKGROUND AND OBJECTIVES

Conventional timing of newborn pulse oximetry screening is not ideal for infants born out-of-hospital. We implemented a newborn pulse oximetry screen to align with typical midwifery care and measure its efficacy at detecting critical congenital heart disease.

METHODS

Cohort study of expectant mothers and infants mainly from the Amish and Mennonite (Plain) communities with limited prenatal ultrasound use. Newborns were screened at 1 to 4 hours of life (“early screen”) and 24 to 48 hours of life (“late screen”). Newborns were followed up to 6 weeks after delivery to report outcomes. Early screen, late screen, and combined results were analyzed on the basis of strict algorithm interpretation (“algorithm”) and the midwife’s interpretation in the field (“field”) because these did not correspond in all cases.

RESULTS

Pulse oximetry screening in 3019 newborns (85% Plain; 50% male; 43% with a prenatal ultrasound) detected critical congenital heart disease in 3 infants. Sensitivity of combined early and late screen was 66.7% (95% confidence interval [CI] 9.4% to 99.2%) for algorithm interpretation and 100% (95% CI 29.2% to 100%) for field interpretation. Positive predictive value was similar for the field interpretation (8.8%; 95% CI 1.9% to 23.7%) and algorithm interpretation (5.4%; 95% CI 0.7% to 18.2%). False-positive rates were ≤1.2% for both algorithm and field interpretations. Other pathologies (noncritical congenital heart disease, pulmonary issues, or infection) were reported in 12 of the false-positive cases.

CONCLUSIONS

Newborn pulse oximetry can be adapted to the out-of-hospital setting without compromising sensitivity or prohibitively increasing false-positive rates.

What’s Known on This Subject:

Newborn pulse oximetry effectively detects critical congenital heart disease among newborns in the hospital setting but is not ideally designed for infants born out-of-hospital.

What This Study Adds:

Newborn pulse oximetry can be modified to align with typical timing of midwifery care in the out-of-hospital setting without compromising the sensitivity of the test or prohibitively increasing false-positive rates.

Approximately 25% of infants with congenital heart defects have a critical congenital heart defect (CCHD), which can lead to sudden and severe illness in the newborn period.1  Pulse oximetry is a portable, inexpensive, and noninvasive test to detect subtle abnormalities in oxygen saturation caused by CCHD before visible cyanosis or signs of illness.24  Notably, studies find that pulse oximetry screening also identifies newborns with significant (but not critical) heart defects, pulmonary issues, or infections.

Infants born in-hospital have routine access to this simple, life-saving screening tool. Although the majority of infants are born in-hospital, the percentage of infants born out-of-hospital in the United States increased by 90% from 2004 to 2019.5  Out-of-hospital birth rates vary greatly by state, with Pennsylvania reporting rates ∼3 times the national average and reaching a new high of 4845 infants (3.6% of all births) in 2018.6  Many out-of-hospital births in Pennsylvania are among the Amish and Mennonite (Plain) communities, Anabaptist groups who typically do not participate in health insurance programs7  and have higher incidence of many genetic disorders, including several that are associated with congenital heart defects.8 

Newborn pulse oximetry screening is currently recommended at 24 to 48 hours of life to reduce false-positive screens because of transitional circulation and correspond with the typical timing of hospital discharge. However, this does not integrate well with usual midwifery practices. Typically, midwives remain with infant and postpartum mother for several hours after delivery. If mother and infant are well, the infant is cared for by the family in the home without direct supervision by the midwife. The midwife returns to assess mother and infant at 2 to 3 days of life or sooner, if concerns arise.

Narayen et al9  used an adapted pulse oximetry screening protocol to coincide with the timing of direct midwifery care (first screen at ≥1 hour of life; second screen at 2 to 3 days of life). The goal of our study was to apply this protocol in the out-of-hospital birth community in Pennsylvania to determine its feasibility to detect critical congenital heart disease.

Midwives completed a training workshop that included rationale for newborn pulse oximetry, probe placement, proper screening methods, and accurate interpretation of results. Midwives were provided free pulse oximeters and color-coded interpretation cards for easy reference in the field.

Infants were first screened at 1 to 4 hours of life (“early screen”) and again at 24 to 48 hours of life (“late screen”) with Masimo Rad 5 or Pronto pulse oximeters (Masimo Corporation, Irvine, CA) by using American Academy of Pediatrics–endorsed screening methods and interpretation criteria described elsewhere and outlined in Supplemental Figs 3 and 4.10  Briefly, pulse oximetry was measured in the newborn’s right hand and either foot. Pulse oximetry <90% in the right hand or either foot constituted a positive screen. Infants with pulse oximetry ≥95% in the right hand or either foot and a ≤3% difference between the 2 readings had a negative screen result. If a measurement of 90% to 94% was obtained in the right hand and either foot or if the difference between the right hand and either foot was >3%, the infant was rescreened 1 hour later. If a repeat screen was required, infants were allowed a total of 2 attempts for the early screen and 3 attempts for the late screen.

Pulse oximetry screening was offered at no cost to all infants delivered by the midwives, irrespective of study participation. Midwives were instructed to have any infant with a positive screen evaluated by a primary care provider or pediatric cardiologist. Low-cost cardiac consultations and echocardiograms were available to midwives in the study to help minimize expenses for families. Midwives were not asked to report the referrals or evaluation but instead were asked to report the final diagnosis.

The Institutional Review Board at Penn Medicine Lancaster General Health (Lancaster, PA) approved the study. Mothers consented for themselves and on behalf of their newborns to participate in the research study from September 2015 to September 2017. Demographic, clinical, and screening data were recorded at the time of screening and submitted after the infant’s final postnatal visit (typically 4 to 6 weeks of age) to determine the outcome.

Infants were excluded from analysis if they were not screened or received an incomplete screen (right arm or foot but not both). For those who were properly screened and included in the analysis, the midwife’s interpretation of pulse oximetry values did not always correspond to the recommended algorithm interpretation, possibly because of human error or the integration of clinical assessment. As a result, reported pulse oximetry readings were analyzed in 2 ways: strict adherence to the screening algorithm (“algorithm”) and on the basis of the midwife’s interpretation of the recorded values (“field”). Early and late screens were analyzed separately and combined as outlined in Supplemental Table 4.

Positive screens with CCHD (ductal dependent lesion or complete mixing) were considered true-positives. Positive screens with no identified CCHD at the time of the final postnatal were considered false-positives. False-positive screens were reviewed to determine if the infant had noncritical cardiac disease, pulmonary disease, sepsis, or other medical condition, but this did not alter their classification as false-positive. Negative screens later found to have CCHD were considered false-negatives. Negative screens with no CCHD were considered true-negatives.

We calculated descriptive statistics for each of the demographic factors with percentages for discrete variables and means and standard deviations for continuous variables. A confusion matrix containing sensitivity, specificity, area under the receiver operating curve (ROC), positive predictive value (PPV), and negative predictive value (NPV), along with respective 95% confidence intervals (CIs), was generated by using a package developed for Stata (Stata Corp, College Station, TX) called diagt (https://ideas.repec.org/c/boc/bocode/s423401.html) for each classification test to be evaluated. False-positive and false-negative rates were calculated by taking 1 minus the specificity or sensitivity, respectively. We evaluated the early, late, and combined “field” and “algorithm” screens. All statistical analyses were conducted in Stata version 16.1.

Demographic information for 3019 mothers enrolled in the study is outlined in Table 1. Midwives were not asked to report the total number of mothers approached to participate in the research study so the exact percentage of mothers who declined is unknown. Most mothers (85%) identified as members of the Plain community and were significantly younger than mothers outside the Plain community (P < .0001). A total of 43% of all mothers had a prenatal ultrasound, and 4% of those had an abnormality detected, typically related to placental issues. Four mothers had fetal cardiac abnormalities (nonstructural) detected as listed in Table 1.

TABLE 1

Demographic and Clinical Information

Information Regarding Demographics, Pregnancy, Prenatal Care, and Delivery for 3019 Mothers and Infants Enrolled in the Study
Maternal background  
 Plain, n (%) 2555 (85) 
  Amish, n 1658 
  Mennonite, n 889 
  Other Plain, n 
 Not Plain 346 (12) 
 Not specified 118 (4) 
Maternal age, mean ± SD (range)  
 Plain 29.2 ± 5.4 (17.6–51.8)a 
 Not Plain 31.0 ± 5.0 (20.2–45.1) 
Prenatal ultrasound, n (%)  
 Yes 1308 (43) 
 No 1686 (56) 
 Not specified 25 (1) 
Prenatal ultrasound abnormalities, n (%)  
 Yes 48 (4) 
 No 1234 (94) 
 Not specified 26 (2) 
Reported ultrasound abnormalities  
 Cardiac, n (%) 4 (9) 
  Left ventricular focus, n 
  Papillary muscle calcification, n 
  Echogenic foci in heart, n 
  Decreased fetal heart tones, n 
GBS testing, n (%)  
 Yes 2237 (74) 
 No 761 (25) 
 Not specified 21 (1) 
GBS test results, n (%)  
 Positive 420 (19) 
 Negative 1785 (80) 
 Not specified 32 (1) 
GBS treatment methods, n (%)  
 Intravenous antibiotics 106 (25) 
 Oral antibiotics 66 (16) 
 Antibiotics, route not specified 98 (23) 
 Antiseptic washes 98 (23) 
 Probiotics 13 (3) 
 Natural remedies 14 (33) 
 Unknown 35 (8) 
 No treatment 99 (24) 
Location of birth, n (%)  
 Home 1959 (65) 
 Birth center 693 (23) 
 Midwife’s home 305 (10) 
 Hospital or clinic 42 (1) 
 Car or ambulance 5 (0.2) 
 Not specified 15 (0.5) 
Birth information, mean ± SD (range)  
 Gestation at birth, wk 39.5 ± 1.2 (35–43.6) 
 Birth wt, kg 3.6 ± 2.2 (1.9–5.4) 
 Apgar, 1 min 8.5 ± 1.2 (0–10) 
 Apgar, 5 min 9.3 ± 0.6 (4–10) 
 Apgar, 10 min 9.9 ± 0.4 (6–10) 
Infant sex, n (%)  
 Male 1510 (50) 
 Female 1457 (48) 
 Not specified 52 (2) 
Genetic disorder, n  
 Cartilage-hair hypoplasia 
 Ellis-van Creveld syndrome 
 Glutaric aciduria type 1 
 Maple syrup urine disease 
 Spinal muscular atrophy 
 Congenital adrenal hyperplasia 
 Crouzon syndrome 
 Down syndrome 
 Hirschsprung disease 
 Suspected osteogenesis imperfecta (deceased) 
Information Regarding Demographics, Pregnancy, Prenatal Care, and Delivery for 3019 Mothers and Infants Enrolled in the Study
Maternal background  
 Plain, n (%) 2555 (85) 
  Amish, n 1658 
  Mennonite, n 889 
  Other Plain, n 
 Not Plain 346 (12) 
 Not specified 118 (4) 
Maternal age, mean ± SD (range)  
 Plain 29.2 ± 5.4 (17.6–51.8)a 
 Not Plain 31.0 ± 5.0 (20.2–45.1) 
Prenatal ultrasound, n (%)  
 Yes 1308 (43) 
 No 1686 (56) 
 Not specified 25 (1) 
Prenatal ultrasound abnormalities, n (%)  
 Yes 48 (4) 
 No 1234 (94) 
 Not specified 26 (2) 
Reported ultrasound abnormalities  
 Cardiac, n (%) 4 (9) 
  Left ventricular focus, n 
  Papillary muscle calcification, n 
  Echogenic foci in heart, n 
  Decreased fetal heart tones, n 
GBS testing, n (%)  
 Yes 2237 (74) 
 No 761 (25) 
 Not specified 21 (1) 
GBS test results, n (%)  
 Positive 420 (19) 
 Negative 1785 (80) 
 Not specified 32 (1) 
GBS treatment methods, n (%)  
 Intravenous antibiotics 106 (25) 
 Oral antibiotics 66 (16) 
 Antibiotics, route not specified 98 (23) 
 Antiseptic washes 98 (23) 
 Probiotics 13 (3) 
 Natural remedies 14 (33) 
 Unknown 35 (8) 
 No treatment 99 (24) 
Location of birth, n (%)  
 Home 1959 (65) 
 Birth center 693 (23) 
 Midwife’s home 305 (10) 
 Hospital or clinic 42 (1) 
 Car or ambulance 5 (0.2) 
 Not specified 15 (0.5) 
Birth information, mean ± SD (range)  
 Gestation at birth, wk 39.5 ± 1.2 (35–43.6) 
 Birth wt, kg 3.6 ± 2.2 (1.9–5.4) 
 Apgar, 1 min 8.5 ± 1.2 (0–10) 
 Apgar, 5 min 9.3 ± 0.6 (4–10) 
 Apgar, 10 min 9.9 ± 0.4 (6–10) 
Infant sex, n (%)  
 Male 1510 (50) 
 Female 1457 (48) 
 Not specified 52 (2) 
Genetic disorder, n  
 Cartilage-hair hypoplasia 
 Ellis-van Creveld syndrome 
 Glutaric aciduria type 1 
 Maple syrup urine disease 
 Spinal muscular atrophy 
 Congenital adrenal hyperplasia 
 Crouzon syndrome 
 Down syndrome 
 Hirschsprung disease 
 Suspected osteogenesis imperfecta (deceased) 
a

Plain versus not Plain; P < .0001.

Most mothers (74%) were tested for group B streptococcal (GBS) colonization, and 19% of those tested were positive. A total of 24% of GBS-positive mothers declined medications or home remedies. The remaining GBS-positive mothers used one or more methods (medical and/or holistic) to treat colonization. A total of 25% of GBS-positive mothers received intravenous antibiotics, as recommended by the American College of Obstetricians and Gynecologists.11 

Clinical data from 3019 infants enrolled in the study are outlined in Table 1. Most infants (75%) were born at the family residence or midwife’s home. Infants (50% male; 48% female; 2% not specified) were born at an average of 39.5 weeks’ gestation, with an average 1-minute Apgar of 8.5 (range: 0–10). A total 15 infants had a genetic syndrome evident at birth or detected in the neonatal period, as listed in Table 1. Thirteen of those infants had a genetic condition that occurs at a higher frequency among the Plain communities.8  Two infants had Ellis–van Creveld (EVC) syndrome, and 1 infant had Down syndrome; both are associated with an increased risk of congenital heart disease.

Most infants (2565; 85%) received the early pulse oximetry screen between 1 and 3 hours of life (Fig 1; top panel). After excluding infants who were not screened or who had an incomplete screen (right arm or foot but not both), 35 (1.3%) infants had a positive screen on the basis of the algorithm and 29 (1%) on the basis of field interpretation (Fig 2; top panel). Among infants who had a positive early screen, there were 13 instances in which the algorithm differed from the field interpretation, most commonly because of a positive screen by algorithm guidelines but assigned a negative screen result in the field (n = 7) or an incomplete screen (unable to obtain a reading in one extremity or measurements that warranted a repeat attempt) that was not repeated as recommended (n = 5). In the remaining instance, the newborn had a negative screen but was rescreened later because of documented clinical concerns.

FIGURE 1

Timing of pulse oximetry screening: age of infants at the time of pulse oximetry for early and late screens. A, Early screen. B, Late screen.

FIGURE 1

Timing of pulse oximetry screening: age of infants at the time of pulse oximetry for early and late screens. A, Early screen. B, Late screen.

Close modal
FIGURE 2

Pulse oximetry screening results and outcomes. Pulse oximetry screening results and final outcomes for early, late, and combined screens. Pulse oximetry readings were interpreted strictly according to the provided algorithm (“algorithm”) and also on the basis of the midwife’s interpretation of the screen (“field”). A, Early screen. B, Late screen. C, Combined screen.

FIGURE 2

Pulse oximetry screening results and outcomes. Pulse oximetry screening results and final outcomes for early, late, and combined screens. Pulse oximetry readings were interpreted strictly according to the provided algorithm (“algorithm”) and also on the basis of the midwife’s interpretation of the screen (“field”). A, Early screen. B, Late screen. C, Combined screen.

Close modal

Both algorithm and field interpretation identified 2 true-positive cases (common atrium and coarctation of the aorta with a moderate ventricular septal defect [VSD] and atrial septal defect [ASD]). An additional case (heterotaxy with single ventricle) had a negative early screen by field interpretation (false-negative) and incomplete screen by algorithm (excluded from analysis). Sensitivity was higher for the algorithm interpretation (100%; 95% CI 15.8% to 100%) compared with that of field interpretation (66.7%; 95% CI 9.4% to 99.2%) for the early screen. Specificity, PPV, NPV, ROC area, and false-positive rate were similar between the algorithm and field interpretations (Table 2). The false-negative rate was lower for the algorithm interpretation (0.0%) compared with that of the field (33.3%).

TABLE 2

Summary of Statistical Analysis of Early, Late, and Combined Pulse Oximetry Screens

Early ScreenLate ScreenCombined Screen
AlgorithmFieldAlgorithmFieldAlgorithmField
Abnormal       
 Positive screen 
 Negative screen 
Healthy       
 Positive screen 33 27 35 31 
 Negative screen 2763 2893 2832 2942 2893 2980 
Sensitivity, % (95% CI) 100.0 (15.8–100.0) 66.7 (9.4–99.2) 0.0 (0.0–97.5) 100.0 (2.5–100.0) 66.7 (9.4–99.2) 100.0 (29.2–100.0) 
Specificity, % (95% CI) 98.8 (98.3–99.2) 99.1 (98.7–99.4) 99.9 (99.7–100.0) 99.8 (99.6–99.9) 98.8 (98.3–99.2) 99.0 (98.5–99.3) 
PPV, % (95% CI) 5.7 (0.7–19.2) 6.9 (0.8–22.8) 0.0 (0.0–70.8) 14.3 (0.4–57.9) 5.4 (0.7–18.2) 8.8 (1.9–23.7) 
NPV, % (95% CI) 100.0 (99.9–100.0) 100.0 (99.8–100.0) 100.0 (99.8–100.0) 100.0 (99.9–100.0) 100.0 (99.8–100.0) 100.0 (99.9–100.0) 
ROC area (95% CI) 0.99 (0.99–1.00) 0.83 (0.50–1.00) 0.50 (a to 1.00) 1.00 (a to 1.00) 0.83 (0.50–1.00) 0.99 (0.99–1.00) 
False-positive rate, % 1.2 0.9 0.1 0.2 1.2 1.0 
False-negative rate, % 0.0 33.3 100.0 0.0 33.3 0.0 
Early ScreenLate ScreenCombined Screen
AlgorithmFieldAlgorithmFieldAlgorithmField
Abnormal       
 Positive screen 
 Negative screen 
Healthy       
 Positive screen 33 27 35 31 
 Negative screen 2763 2893 2832 2942 2893 2980 
Sensitivity, % (95% CI) 100.0 (15.8–100.0) 66.7 (9.4–99.2) 0.0 (0.0–97.5) 100.0 (2.5–100.0) 66.7 (9.4–99.2) 100.0 (29.2–100.0) 
Specificity, % (95% CI) 98.8 (98.3–99.2) 99.1 (98.7–99.4) 99.9 (99.7–100.0) 99.8 (99.6–99.9) 98.8 (98.3–99.2) 99.0 (98.5–99.3) 
PPV, % (95% CI) 5.7 (0.7–19.2) 6.9 (0.8–22.8) 0.0 (0.0–70.8) 14.3 (0.4–57.9) 5.4 (0.7–18.2) 8.8 (1.9–23.7) 
NPV, % (95% CI) 100.0 (99.9–100.0) 100.0 (99.8–100.0) 100.0 (99.8–100.0) 100.0 (99.9–100.0) 100.0 (99.8–100.0) 100.0 (99.9–100.0) 
ROC area (95% CI) 0.99 (0.99–1.00) 0.83 (0.50–1.00) 0.50 (a to 1.00) 1.00 (a to 1.00) 0.83 (0.50–1.00) 0.99 (0.99–1.00) 
False-positive rate, % 1.2 0.9 0.1 0.2 1.2 1.0 
False-negative rate, % 0.0 33.3 100.0 0.0 33.3 0.0 

Pulse oximetry readings were analyzed on the basis of strict algorithm interpretation (“algorithm”) and the midwife’s interpretation of the screen (“field”).

a

Unable to calculate values.

Most infants (2344; 78%) received the late pulse oximetry screen between 24 and 72 hours of life (Fig 1; bottom panel). After excluding infants who were not screened or had an incomplete screen, 3 (0.1%) infants had a positive screen on the basis of the algorithm and 7 (0.2%) on the basis of field interpretation (Fig 2; middle panel). Among infants who had a positive late screen, there were 7 instances in which the algorithm differed from field interpretation, most commonly because of an incomplete screen that was not repeated per algorithm guidelines (n = 3) or a positive screen result by using the algorithm deemed negative in the field (n = 1.) Three additional infants had a negative late screen result by algorithm interpretation but had positive results in the field; one was documented to have clinical concerns.

The 2 true-positive cases with a positive early screen result did not receive a late screen because they were already receiving medical care. The newborn with heterotaxy and a single ventricle received a late screen result, which was negative by algorithm interpretation (false-negative) and positive by field interpretation (true-positive). Sensitivity and PPV were higher for the field interpretation of the late screen (100% [95% CI 2.5% to 100%] and 14.3% [95% CI 0.4% to 57.9%], respectively), compared with the algorithm (0% [95% CI 0% to 97.5%] and 0% [95% CI 0% to 70.8%], respectively; Table 2). Specificity, NPV, ROC area, and false-positive rates were similar between the algorithm and field interpretations. False-negative rate was higher for the algorithm (100%), compared with that of field interpretation (0%).

When screening results were combined, 37 of 2931 (1.3%) infants had a positive combined screen on the basis of algorithm and 34 of 3014 (1.1%) on the basis of field interpretation (Fig 2; bottom panel). Two true-positive cases were identified by both interpretations, and one additional true-positive case was identified by field interpretation. Sensitivity was higher for field interpretation (100%, 95% CI 29.2% to 100%) compared with the algorithm alone (66.7%, 95% CI 9.4-99.2, Table 2). PPV and false-positive rate was similar between the two interpretations. False-negative rate was higher for the algorithm (33.3%) compared with field interpretation (0%).

In this screening program, 3 infants were identified with critical congenital heart disease (common atrium; coarctation of the aorta with a moderate VSD and ASD; and heterotaxy with a single ventricle). Two were identified because of a pulse oximetry reading of <90% (Table 3). The remaining case had a false-negative screen result by using the algorithm alone but was positive because of clinical concerns in the field. Twelve of the newborns with a false-positive screen result had other pathology (prematurity, vocal cord paralysis, pulmonary hypertension, VSD with patent ductus arteriosus [PDA] and pulmonary hypertension, PDA with pulmonary hypertension, aortic insufficiency with PDA, sepsis, respiratory distress, or pneumothorax) that may have contributed to the low pulse oximetry at the time of screening. For these infants, the most common indication for a positive screen result was also pulse oximetry <90%.

TABLE 3

Summary of Abnormal Cases Detected by Pulse Oximetry Screening

OutcomeEarly ScreenLate ScreenCombined Screen
AlgorithmFieldAlgorithmFieldAlgorithmField
True-positive       
 Common atrium and EVC syndrome Positivea Positivea Incomplete Incomplete Positive Positive 
 Coarctation of the aorta, moderate VSD, ASD, and EVC syndrome Positivea Positivea Incomplete Incomplete Positive Positive 
 Heterotaxy and single ventricle Incomplete Negative Negative Positived Negative Positive 
False-positive with other pathology       
 Prematurity Positivea Positivea Incomplete Incomplete Positive Positive 
 Vocal cord paralysis Incomplete Positivee Negative Negative Negative Positive 
 Respiratory distress syndrome Positivea Positivea Incomplete Incomplete Positive Positive 
 Pulmonary hypertension Positivea Positivea Negative Negative Positive Positive 
 VSD, PDA, and pulmonary hypertension Positiveb Positiveb Negative Negative Positive Positive 
 PDA and pulmonary hypertension Positivec Positivec Negative Negative Positive Positive 
 Sepsis Positivea Positivea Incomplete Incomplete Positive Positive 
 PDA, PFO, and aortic insufficiency Negative Positivef Incomplete Incomplete Negative Positive 
 Respiratory distress syndrome, and pneumothorax Positivea Positivea Incomplete Incomplete Positive Positive 
 PDA, aortic arch abnormality, and persistent fetal circulation Positivea Positivea Negative Negative Positive Positive 
 PDA, PFO, and cyanosis Positivea Positivea Incomplete Incomplete Positive Positive 
 Respiratory distress syndrome Negative Negative Negative Negative Negative Negative 
OutcomeEarly ScreenLate ScreenCombined Screen
AlgorithmFieldAlgorithmFieldAlgorithmField
True-positive       
 Common atrium and EVC syndrome Positivea Positivea Incomplete Incomplete Positive Positive 
 Coarctation of the aorta, moderate VSD, ASD, and EVC syndrome Positivea Positivea Incomplete Incomplete Positive Positive 
 Heterotaxy and single ventricle Incomplete Negative Negative Positived Negative Positive 
False-positive with other pathology       
 Prematurity Positivea Positivea Incomplete Incomplete Positive Positive 
 Vocal cord paralysis Incomplete Positivee Negative Negative Negative Positive 
 Respiratory distress syndrome Positivea Positivea Incomplete Incomplete Positive Positive 
 Pulmonary hypertension Positivea Positivea Negative Negative Positive Positive 
 VSD, PDA, and pulmonary hypertension Positiveb Positiveb Negative Negative Positive Positive 
 PDA and pulmonary hypertension Positivec Positivec Negative Negative Positive Positive 
 Sepsis Positivea Positivea Incomplete Incomplete Positive Positive 
 PDA, PFO, and aortic insufficiency Negative Positivef Incomplete Incomplete Negative Positive 
 Respiratory distress syndrome, and pneumothorax Positivea Positivea Incomplete Incomplete Positive Positive 
 PDA, aortic arch abnormality, and persistent fetal circulation Positivea Positivea Negative Negative Positive Positive 
 PDA, PFO, and cyanosis Positivea Positivea Incomplete Incomplete Positive Positive 
 Respiratory distress syndrome Negative Negative Negative Negative Negative Negative 

PFO, patent foramen ovale.

a

Indication for positive screen: <90% in 1 extremity (automatic positive screen).

b

Indication for positive screen: <95% in 1 extremity despite 2 attempts.

c

Indication for positive screen: >3% difference in extremity despite 2 attempts.

d

Pulse oximetry was normal but had a positive screen result on the basis of clinical suspicion.

e

Initial reading required repeat screen but was interpreted as positive screen result.

f

Negative screen initially but repeated at 14 h and had <95% in one extremity.

Several other types of neonatal pathology were reported among the infants with negative pulse oximetry screening, as listed in Supplemental Table 5. Briefly, these included arrhythmia, minor structural defects (isolated ASD or small VSD), culture negative sepsis, fever, urinary tract infection, viral meningitis, hypoglycemia, seizures, vagal episodes, or tracheomalacia/laryngomalacia. The timing of onset was not reported and may have been identified at any point before the final newborn visit (up to 6 weeks of age). The pathology would not necessarily cause abnormal pulse oximetry, especially if the screening occurred before the onset of symptoms.

Screening protocol error rates were highest during the first 2 months of study (32 errors among 69 total screens), most commonly because of pulse oximetry measurement in the right arm or leg but not both. Midwives were contacted to clarify misunderstandings in the protocol and error rates progressively decreased during the next 6 months of the study. By the ninth month of study, the error rate had decreased to 3.4% and remained at an average of 3.4% per month (range: 0.9% to 8.3%) for the remainder of the study (Supplemental Fig. 5).

We report the largest study in the United States of newborn pulse oximetry screening in infants born out-of-hospital. Screening was conducted by midwives of various training levels among a high-risk community with limited access to medical care. Despite these challenges, the screening proved to be an effective tool for detecting CCHD.

Our protocol was modeled after Narayen et al12  who aligned pulse oximetry screening with typical timing of midwifery care. We found a low prevalence (n = 3) of true-positive cases in the study, which resulted in variable sensitivities and wide CIs for many statistical measures and limited our ability to compare the algorithm and field interpretation to one another. Sensitivity and PPV are variable among other studies (42% to 84% and 0.6% to 52%, respectively)24,1318  and our early, late, and combined screen sensitivity (algorithm or field interpretation) was within or above this range, except for the late screen algorithm interpretation. Like previous studies, we found a high specificity and NPV, suggesting pulse oximetry screening is most effective at excluding CCHD among healthy newborns.

Twelve of the false-positive cases had other pathologies (noncritical congenital heart disease, pulmonary issues, or infections), which may have led to abnormal pulse oximetry at the time of screening. Notably, all were identified by using the early pulse oximetry screen. Although this was not the primary goal of the current study, it demonstrated that pulse oximetry screening can identify other infants who may benefit from early recognition.

The field interpretation demonstrates that human error, such as overlooking a positive screen or failing to repeat a screen as recommended, did not diminish the overall utility of the screen. Moreover, allowing midwives to use their clinical judgement and deviate from a structured algorithm enhanced their ability to identify at-risk newborns. We did not collect data regarding reasons for nonadherence to the algorithm or the steps that were taken after an abnormal screen, so we cannot say for certain what led to their decisions or which elements were most critical for the enhanced performance of the field interpretation in some instances.

The study population was mainly from the Plain community and largely uninsured, so minimizing extraneous costs was critical. In theory, introducing pulse oximetry before 24 hours, when the newborn is actively transitioning from intrauterine to extrauterine life, can lead to false-positive screen result because of shunting across the PDA. In our study, the false-positive rates for early screens were larger (1.2% for algorithm; 0.9% for field interpretation) than those from late screens (0.1% for algorithm; 0.2% for field interpretation). However, this false-positive rate (0.1% to 1.2%) is comparable to larger studies that screened closer to or after 24 hours (range: 0.17 to 0.8),2,3,18  suggesting the addition of an early screen did not prohibitively increase the cost of care or overuse of echocardiography, compared with other screening methods.

Given the low frequency of prenatal ultrasound and an expected increased prevalence of genetic disease among our study population, we anticipated a higher prevalence of congenital heart disease compared with that of other studies. However, 3 (0.1%) infants with critical congenital heart disease were identified, similar to or below the rate reported by researchers in previous studies of populations with standard ultrasounds rates.2,3,13,14,1618  Prenatal ultrasounds detect only 36% to 39% of congenital heart disease,19,20  so limited use of this tool may not have impacted the study to a significant degree. Some infants with CCHD may have been brought to medical attention immediately after birth and thus were not available for screening. Additionally, several genetic disorders are more common among Plain community members, but most are not associated congenital heart disease.

Mothers in the study had GBS colonization rates comparable with that of the general population (15% to 35%)21  with inconsistent treatment of GBS colonization, so we expected several cases of GBS disease. Although 2 infants were reported to have sepsis (1 with a positive early screen result), no infants in the study had documented GBS disease. Before the widespread use of peripartum antibiotics, early-onset GBS disease occurred in 1 to 2 infants per 100 colonized women.21  Our study included 420 mothers with documented GBS colonization, and only 25% received intravenous antibiotics, so the study size should have been large enough to detect several cases of GBS disease. Several of the mothers colonized with GBS used oral antibiotics, antiseptic washes, or other remedies, which may have impacted GBS colonization or transmission rates more than we anticipated.

This study adds to a growing body of literature demonstrating that pulse oximetry screening protocols can be adapted to the out-of-hospital setting and administered by community midwives without sacrificing the effectiveness of the tool.4,14  This has strong implications for the global community, particularly in areas where home birth is the norm. Numerous pulse oximetry screening algorithms have been successful despite differences in screening methods,22  suggesting an opportunity to simplify the tool for populations with limited resources.

We thank the community midwives and expectant mothers who participated in the study. Your collaborative spirit and commitment to providing quality care for all mothers and newborns, regardless of place of delivery, inspires us.

Dr Williams contributed to the study design, data acquisition, data analysis, and interpretation of the data, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Horst led the data analysis and critically reviewed the manuscript; Ms Hollinger, Mr Freedman, and Dr Demczko contributed significantly to data collection and critically reviewed the manuscript; Dr Chowdhury contributed to the study design, data analysis, and interpretation of the data and critically reviewed the manuscript; all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: Supported by The Abigail L. Longenecker Memorial Foundation, Ronald McDonald House Charities, Hershey Rotary, and Pennsylvania Department of Health. Funders provided pulse oximeters for the project. The funders were not involved in the design, conduct, or analysis of the study.

CCHD

critical congenital heart defect

CI

confidence interval

EVC

Ellis–van Creveld

GBS

group B streptococcal

NPV

negative predictive value

PDA

patent ductus arteriosus

PPV

positive predictive value

ROC

receiver operating curve

VSD

ventricular septal defect

1
Oster
ME
,
Lee
KA
,
Honein
MA
,
Riehle-Colarusso
T
,
Shin
M
,
Correa
A
.
Temporal trends in survival among infants with critical congenital heart defects
.
Pediatrics
.
2013
;
131
(
5
):
e1502
e1508
2
de-Wahl Granelli
A
,
Wennergren
M
,
Sandberg
K
, et al
.
Impact of pulse oximetry screening on the detection of duct dependent congenital heart disease: a Swedish prospective screening study in 39,821 newborns
.
BMJ
.
2009
;
338
:
a3037
3
Ewer
AK
,
Middleton
LJ
,
Furmston
AT
, et al;
PulseOx Study Group
.
Pulse oximetry screening for congenital heart defects in newborn infants (PulseOx): a test accuracy study
.
Lancet
.
2011
;
378
(
9793
):
785
794
4
Lhost
JJ
,
Goetz
EM
,
Belling
JD
,
van Roojen
WM
,
Spicer
G
,
Hokanson
JS
.
Pulse oximetry screening for critical congenital heart disease in planned out-of-hospital births
.
J Pediatr
.
2014
;
165
(
3
):
485
489
5
Centers for Disease Control and Prevention
.
Annual Natality Information; Live Births. 2004-2019
.
Atlanta, GA
:
Centers for Disease Control and Prevention
:
Available at: https://wonder.cdc.gov/natality.html. Accessed January 13, 2021
6
Commonwealth of Pennsylvania Department of Health Division of Health Informatics
.
Live Births by Hospital and Method of Delivery
.
Harrisburg, PA
:
Commonwealth of Pennsylvania Department of Health
;
2018
7
Kraybill
DB
,
Johnson-Weiner
K
,
Nolt
SM
.
The Amish
.
Baltimore, MD
:
Johns Hopkins University Press
;
2013
8
Strauss
KA
,
Puffenberger
EG
.
Genetics, medicine, and the Plain people
.
Annu Rev Genomics Hum Genet
.
2009
;
10
:
513
536
9
Narayen
IC
,
Blom
NA
,
Bourgonje
MS
, et al
.
Pulse oximetry screening for critical congenital heart disease after home birth and early discharge
.
J Pediatr
.
2016
;
170
:
188
92.e1
10
Kemper
AR
,
Mahle
WT
,
Martin
GR
, et al
.
Strategies for implementing screening for critical congenital heart disease
.
Pediatrics
.
2011
;
128
(
5
):
e1259
e1267
11
American College of Obstetricians and Gynecologists Committee on Obstetric Practice
.
ACOG Committee Opinion No. 485: prevention of early-onset group B streptococcal disease in newborns
.
Obstet Gynecol
.
2011
;
117
(
4
):
1019
1027
12
Narayen
IC
,
Blom
NA
,
Verhart
MS
, et al
.
Adapted protocol for pulse oximetry screening for congenital heart defects in a country with homebirths
.
Eur J Pediatr
.
2015
;
174
(
1
):
129
132
13
Jawin
V
,
Ang
HL
,
Omar
A
,
Thong
MK
.
Beyond critical congenital heart disease: newborn screening using pulse oximetry for neonatal sepsis and respiratory diseases in a middle-income country
.
PLoS One
.
2015
;
10
(
9
):
e0137580
14
Narayen
IC
,
Blom
NA
,
van Geloven
N
, et al;
POLAR study group
.
Accuracy of pulse oximetry screening for critical congenital heart defects after home birth and early postnatal discharge
.
J Pediatr
.
2018
;
197
:
29
35.e1
15
Plana
MN
,
Zamora
J
,
Suresh
G
,
Fernandez-Pineda
L
,
Thangaratinam
S
,
Ewer
AK
.
Pulse oximetry screening for critical congenital heart defects
.
Cochrane Database Syst Rev
.
2018
;
3
:
CD011912
16
Saxena
A
,
Mehta
A
,
Ramakrishnan
S
, et al
.
Pulse oximetry as a screening tool for detecting major congenital heart defects in Indian newborns
.
Arch Dis Child Fetal Neonatal Ed
.
2015
;
100
(
5
):
F416
F421
17
Turska Kmieć
A
,
Borszewska Kornacka
MK
,
Błaż
W
,
Kawalec
W
,
Zuk
M
.
Early screening for critical congenital heart defects in asymptomatic newborns in Mazovia province: experience of the POLKARD pulse oximetry programme 2006-2008 in Poland
.
Kardiol Pol
.
2012
;
70
(
4
):
370
376
18
Zhao
QM
,
Ma
XJ
,
Ge
XL
, et al;
Neonatal Congenital Heart Disease screening group
.
Pulse oximetry with clinical assessment to screen for congenital heart disease in neonates in China: a prospective study
.
Lancet
.
2014
;
384
(
9945
):
747
754
19
Friedberg
MK
,
Silverman
NH
,
Moon-Grady
AJ
,
Tong
E
,
Nourse
J
,
Sorenson
B
, et al
.
Prenatal detection of congenital heart disease
.
J Pediatr
.
2009
;
155
(
1
):
26
31
20
Pinto
NM
,
Keenan
HT
,
Minich
LL
,
Puchalski
MD
,
Heywood
M
,
Botto
LD
.
Barriers to prenatal detection of congenital heart disease: a population-based study
.
Ultrasound Obstet Gynecol
.
2012
;
40
(
4
):
418
425
21
Kimberlin
DM
,
Brady
MT
,
Jackson
MA
,
Long
SS
.
Redbook
. 31st ed.
Itasca, IL
:
American Academy of Pediatrics
;
2018
22
Martin
GR
,
Ewer
AK
,
Gaviglio
A
, et al
.
Updated strategies for pulse oximetry screening for critical congenital heart disease
.
Pediatrics
.
2020
;
146
(
1
):
146

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest relevant to this article to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data