Role of Pulse Oximetry in Examining Newborns for Congenital Heart Disease: A Scientific Statement from the AHA and AAP

Currently, children with CCHD are diagnosed by a variety of mechanisms. Neonates with CCHD may be diagnosed in the newborn nursery on the basis of physical examination findings, such as heart murmurs, tachypnea, or overt cyanosis. These findings are not always evident before hospital discharge, which may occur before 48 hours of life. A recent study from the United Kingdom suggested that 25% of infants with CCHD were not diagnosed with heart disease until after discharge from the newborn nursery.⁷  The median age of diagnosis in these cases was 6 weeks. A recent publication from the United States suggested that delayed or missed diagnosis occurs in 7 per 100 000 livebirths.⁸  However, because these data are derived from a birth defect surveillance program with passive and thus incomplete case ascertainment, this calculation most likely represents a minimum estimate.

Newborns with CCHD are susceptible to profound, sudden worsening in clinical status in the first days and weeks of life. These acute physiological changes correspond to changes in pulmonary vascular resistance and closure of the ductus arteriosus. In neonates with CCHD, the ductus arteriosus is often essential for maintaining either pulmonary or systemic blood flow. These CCHD defects are considered ductus arteriosus–dependent lesions (Table 2). The newborn hospitalization provides a critical window for caregivers to identify CCHD lesions in order to avoid hemodynamic embarrassment. The timing of constriction or closure of the ductus arteriosus also explains why children with CCHD may be particularly vulnerable to cardiovascular collapse soon after discharge from the newborn nursery.

With the advent of prostaglandin therapy for ductus arteriosus–dependent lesions, many previously lethal congenital heart conditions that present with severe hypoxemia, shock, and acidosis in the newborn period are now survivable. The severity of organ damage is a function of the extent of insult, differential flow to organs as the neonatal circulation responds to the hypoxic/ischemic insult, and the oxygen requirement of each organ.

Among sequelae of neonatal hemodynamic compromise, the most important long-term effects relate to the consequences of brain injury from ischemia and reperfusion, because the brain has the highest oxygen requirement of any organ. Cerebrovascular pressure autoregulation and reactivity to CO₂ are affected by hypoxic/ischemic injury, which renders the brain particularly vulnerable to hypotension and decreased cardiac output.⁹  Such hemodynamic instability is prevalent among neonates with CCHD who present with shock. Furthermore, preoperative events may interact with genetic mutations and both intraoperative and postoperative factors in determining later neurodevelopmental outcome.¹⁰ 

Using brain magnetic resonance imaging, a number of investigators have demonstrated acute brain injury in the newborn with CCHD before surgical intervention. Periventricular leukomalacia, which occurs secondary to vulnerability of the immature oligodendrocyte to hypoxia/ischemia, free radical attack, and excitotoxicity, and likely circulating cytokines, has been found on magnetic resonance imaging in up to 39% of neonates with CCHD.^11–14 

Children with CCHD are reported to experience more frequent impairments in motor function, speech and language, visual-motor-perceptual function, and executive function, as well as increased use of special services.^10,15–22  The greatest frequency of adverse outcomes is found among those with a single ventricle with obstruction to systemic outflow, such as hypoplastic left heart syndrome.²³  In this lesion, systemic perfusion occurs through the patent ductus arteriosus, and ductus closure results in shock and end-organ damage. Prenatal diagnosis of hypoplastic left heart syndrome has been reported in certain studies to reduce early neurological morbidity, with fewer adverse perioperative neurological events such as coma,²⁴  although earlier age at surgery has not been shown to result in better long-term neurodevelopmental outcomes.²³  One could infer that because delayed diagnosis is associated with damage to various end organs, it might also lead to hypoxic/ischemic brain injury; however, further studies are needed to demonstrate a true causal relationship.

A number of children with CCHD are so severely compromised at presentation that they die before surgical intervention. For example, investigators have reported that between 3% and 6% of neonates with dextro-transposition of the great arteries died because of hemodynamic compromise before surgical intervention could be offered.^25,26  In a study from the Baltimore-Washington metropolitan area in the 1980s, Kuehl and colleagues²⁷  reported that among 4360 children with any form of congenital heart disease, 76 (1.7%) died before the identification of heart disease. Delayed or missed diagnosis of CCHD accounted for 1.4 deaths per 10 000 livebirths in that series. In 1994, Abu-Harb and colleagues²⁸  reported on death due to unrecognized CCHD in infancy over a 6-year period in a region of northern England. Fifty-six of 185 children died in infancy, and 27 (48%) of these deaths resulted from sequelae of undetected CCHD. The great majority of these subjects had CCHD lesions that might have manifested hypoxemia. In another study from the United Kingdom, Wren and colleagues²⁹  reported that 25% of CCHD lesions were not diagnosed until after hospital discharge, even in the most recent era. The data from these United Kingdom studies suggested that delayed or missed diagnosis of CCHD accounted for 0.4 to 2.0 deaths per 10 000 livebirths.

With the increased use of prenatal ultrasound and a better understanding of the presentation of CCHD in the past decade, the risk of death before diagnosis has undoubtedly declined, although it is still likely to be important.³⁰  Two recent studies have reported that the rate of mortality due to delayed diagnosis of CCHD is an order of magnitude lower than in the older studies discussed in the previous paragraph. First, a presentation from a study from metropolitan Atlanta, Ga, that used a population-based surveillance system reported that death due to delayed diagnosis of CCHD occurred in 1.0 of every 100 000 livebirths and may be decreasing with time³¹ ; however, this estimate could be understated, because in that study, only deaths that occurred before arrival at a hospital or before the child could be stabilized were attributed to delayed diagnosis. Another preliminary study from California reported 2.0 deaths per 100 000 livebirths related to delayed diagnosis of CCHD.³²  Presumably, earlier recognition of CCHD in these patients could have prevented death in at least some of these cases.

Impairment in cardiovascular function from delayed diagnosis may also adversely impact survival during neonatal cardiovascular surgery and recovery. Certain studies that compared outcomes in prenatal and postnatal diagnosis of CCHD have reported better short-term results for those who were diagnosed prenatally.^25,33  However, numerous other studies have failed to document any survival benefit of prenatal diagnosis among infants undergoing congenital heart surgery.^34,35 

In summary, delayed or missed diagnosis is associated with significant morbidity, the most significant being hypoxic/ischemic brain injury. In addition, delayed diagnosis appears to lead directly and indirectly to higher mortality in this population, although the number of deaths that might be prevented through pulse oximetry screening remains to be determined. Methods to improve early detection of CCHD appear warranted.

Children with CCHD are identified in a variety of ways. Since the late 1980s, prenatal ultrasound has been used to screen for congenital anomalies. An anatomic ultrasound is typically performed at 18 to 20 weeks’ gestation. During this process many, but not all, cases of CCHD can be identified by a methodical scan.³⁶  When CCHD is identified by this approach, the patient is often referred to a pediatric cardiologist for confirmatory imaging and counseling. With knowledge that the fetus has CCHD, the newborn can be delivered in a hospital capable of providing intensive care, including prostaglandin, as well as mechanical ventilation. The newborn can be stabilized and transferred to a congenital heart center.

Prenatal ultrasound, performed by those with specific training in congenital heart disease, can identify a variety of CCHD lesions; however, numerous studies have reported that even when fetal ultrasound is routinely performed during pregnancy, fewer than 50% of cases of CCHD are identified. Most of the published literature comes from European countries, which tend to have more centralized healthcare systems and uniform practices vis-à-vis prenatal ultrasound.^30,37–41  As such, these systems may represent the best-case scenario for population prenatal ultrasound screening. In the United States, many congenital surgery referral centers have reported prenatal detection rates >50% for functional single-ventricle lesions,^35,42  although the detection rate is generally <30% for CCHD lesions with 2-ventricle circulation.^43,44  These studies from referral centers may be biased toward higher detection rates, and population-based data on prenatal detection of CCHD in the United States are sparse.

There are several factors that might account for the relatively low prenatal CCHD detection rate. The quality of anatomic ultrasounds varies considerably.^45,46  A number of medical professionals, including radiologists, perinatologists, and general obstetricians with varying degrees of training, as well as technicians, perform these ultrasounds.⁴⁷  In addition to concerns about the quality, there may be limited access to prenatal ultrasound. In the United States, an anatomic ultrasound is not performed in all women.⁴⁸  The availability of anatomic ultrasound is likely to be particularly limited in certain racial/ethnic or low-socioeconomic-status groups.⁴⁹  Therefore, although prenatal ultrasound plays an important part in the timely identification of CCHD, population-based data demonstrate that this methodology by itself is insufficient to identify a high proportion of cases.

After birth, screening for congenital heart disease by primary care providers is currently accomplished by physical examination within the first 24 hours of life and on subsequent nursery visits. Supplemental tests, including electrocardiograms, pulse oximetry, and chest radiographs, are often obtained in suspicious cases. Echocardiograms can be done either with or without pediatric cardiology consultation. This strategy blends diagnostic assessment approaches from the 1950s to 1970s with the increasing availability of echocardiography. It results in substantial case identification but is regarded as inefficient and costly and misses a significant number of newborns with CCHD.⁵⁵ 

Skilled physical examination, a sensitive and specific screening tool in older children, does not always distinguish between neonates with and without congenital heart disease.^50,56  Hypoxemia is difficult to detect in newborns, and the transitional circulation masks important clinical findings such as absent femoral pulses while the ductus arteriosus remains patent. Reports of the late detection of coarctation of the aorta have been published since the 1960s.⁵¹  Perhaps most importantly, physical examination skills are on the decline in current trainees.⁵² 

Heart murmurs have a prevalence of between 0.6% and 4.2% in newborns and are mistakenly considered a hallmark of heart disease.^53,54  They often do not accompany critical heart defects, particularly those with valve atresia and transposition. Flow murmurs of the transitional circulation, transient tricuspid regurgitation, and small ventricular septal defects are common and of no clinical importance in newborns. Conversely, murmurs of many important complex heart defects, such as tricuspid atresia with ventricular septal defect, double-outlet right ventricle, and total anomalous pulmonary venous return, emerge only after the decline in pulmonary resistance and after neonatal discharge and are often heard but not considered pathological. Practicing pediatricians currently have limited experience in discriminating innocent from pathological murmurs. In a contemporary series in which echocardiography was performed to evaluate for possible heart disease based on suspicious physical examination, fewer than 15% of subjects were found to have significant congenital heart disease.⁵⁵ 

Clinical experience and epidemiological observations suggest that although physical examination, electrocardiogram, and chest radiograph are useful in identifying many cases of serious congenital heart disease postnatally, they do not have sufficient sensitivity and specificity to detect all cases. Echocardiography, although an essential diagnostic tool, has serious limitations as a universal screening tool, particularly its cost.⁵⁶  When used as a screening tool, echocardiography has a high frequency of either false-positive results (usually related to the transitional circulation) or recognition of clinically benign diagnoses (eg, small muscular ventricular septal defects). In addition, there may be an inadequate supply of trained personnel who could perform this screening with a reasonable degree of accuracy. Therefore, there is considerable interest in improving the detection of CCHD with novel diagnostic techniques.

A common feature of many forms of congenital heart disease is hypoxemia. Hypoxemia results from the mixing of systemic and venous circulations or parallel circulations as one might see in dextro-transposition of the great arteries. Hypoxemia may result in obvious cyanosis. However, generally, 4 to 5 g of deoxygenated hemoglobin is needed to produce visible central cyanosis, independent of hemoglobin concentration.⁵⁷  For the typical newborn with a hemoglobin concentration of 20 g/dL, cyanosis will only be visible when arterial oxygen saturation is <80%; if the infant only has a hemoglobin concentration of 10 g/dL, the saturation must be <60% before cyanosis is apparent.⁵⁸  Importantly, those children with mild hypoxemia, with arterial oxygen saturation of 80% to 95%, will not have visible cyanosis. Moreover, the identification of cyanosis is particularly problematic in black and Hispanic neonates because of skin pigmentation.⁵⁷ 

The majority of CCHD lesions present with some degree of hypoxemia in the newborn period. Table 2 demonstrates the frequency of the most common forms of CCHD based on data from the Metropolitan Atlanta Congenital Birth Defects Surveillance Program⁵³  and the likelihood of having some degree of hypoxemia in the newborn period. To improve timely detection of CCHD, a number of investigators have proposed that pulse oximetry be considered as a complementary modality to the newborn physical examination.^59,60 

Pulse oximetry was developed in the early 1970s based on the different absorption spectra between oxygenated and deoxygenated hemoglobin.⁶¹  Deoxygenated hemoglobin absorbs light in the red band (600 to 750 nm), whereas oxygenated hemoglobin absorbs light in the infrared band (850 to 1000 nm). The ratio of light absorbance at these 2 wavelengths correlates with the saturation of hemoglobin in the capillaries.⁶²  Pulse oximetry has the potential to identify hypoxemia that might not otherwise produce visible cyanosis, especially among darkly pigmented newborns.

Pulse oximetry is used routinely in the assessment of young children in neonatal intensive care units and emergency departments and has been proposed as an adjunct to the assessment of the newborn in the delivery room.⁶³  As such, some have proposed that pulse oximetry be considered as a vital sign equivalent in importance to pulse, respirations, and blood pressure.⁶⁴  Contemporary use of pulse oximetry has thus already contributed to heightened recognition of congenital heart disease in neonates.

Pulse oximetry has gained wide acceptance as a noninvasive method to determine oxygen saturation (Spo₂). The method does not require calibration and is able to provide instantaneous data that correlate well with blood gas measurements. O'Brien and colleagues⁶⁵  have defined reference data for oxygen saturation in healthy full-term infants during their first 24 hours of life. The median value at 20 to 24 hours of life (97.8%) is similar to the results for healthy full-term infants between 2 and 7 days of age (97.6%).⁶⁶  Other investigators have reported similar results.^67,68  Beginning in the 1990s, investigators began to explore the possible role of neonatal oximetry in identifying CCHD that might otherwise go undetected. Initially, investigators demonstrated that in neonates with known CCHD, pulse oximetry measurements were significantly lower than in age-matched control subjects. Using a cutoff of 95% in lower-extremity saturation, Hoke and colleagues⁵⁹  suggested that 81% of neonates with CCHD could be identified. Given this association, the question arose as to whether oximetry can successfully identify CCHD in a population of newborns not otherwise suspected of having heart disease. To date, several published studies^{55,59,60,69–75}  have used newborn oximetry to screen for CCHD (Table 3). Most studies were relatively small, and screening protocols differed with respect to both age at screening and cutoff levels for an abnormal screen. Nonetheless, the cumulative experience of these investigations provides a framework for evaluation of the test characteristics of newborn oximetry screening. The results of these studies and differences in study protocols are described below.

Because newborns with CCHD may have clinical deterioration in the first 48 hours of life, one would ideally use oximetry screening soon after delivery. However, arterial oxygen saturation varies considerably in the first 24 hours, with many healthy newborns having arterial saturations of less than 95%. As such, oximetry screening before 24 hours of life can result in a significant number of false-positive results. A study from the United Kingdom reported that the false-positive rate was as high as 5% when oximetry screening was performed in the first 24 hours compared with 1% at the time of hospital discharge.⁷⁶  Therefore, to achieve an acceptable specificity, testing >24 hours after birth would appear to be the most reasonable strategy. This screening strategy assumes that the majority of newborns will not be discharged on the first day of life. With early discharge at less than 24 hours of age, many infants would not be screened.

The establishment of a cutoff threshold for an abnormal Spo₂ is important. Other factors being constant, a higher threshold will increase sensitivity and at the same time decrease specificity. Setting the Spo₂ cutoff value closer to the normal level will decrease the number of false-negative screening results at the cost of increasing the number of false-positive screening results. Conversely, a lower Spo₂ threshold will lower sensitivity and raise specificity. Although a number of Spo₂ thresholds have been proposed, many investigators believe that an Spo₂ of ≤95% is appropriate. In studies of healthy populations, the distribution of Spo₂ measured in a lower extremity at 24 hours was reported to be 97.3±1.3%.⁶⁸  One study suggested that Spo₂ <92% be considered a positive sign of hypoxemia; however, others have argued that a low threshold is likely to result in a number of infants with CCHD being misclassified as normal without markedly improving specificity.⁷⁷ 

Most published studies of oximetry screening for CCHD have been performed at relatively low altitude. It is known, however, that arterial saturation in children and adults is lower at high altitudes, especially above 5000 ft. Several investigators have reported the normal Spo₂ values for neonates at high altitude.^77,78  Bakr and colleagues⁷⁸  reported a mean Spo₂ of 95.4% at 24 hours of life from a population evaluated at 5300 ft. Presumably, one would need to establish a different threshold for high-altitude populations to maintain a reasonable balance between sensitivity and specificity of oximetry screening. Pilot projects are currently under way to examine how high altitude impacts newborn screening with oximetry.

Another variation among published oximetry screening studies has been the position of the pulse oximetry probe with respect to the upper or lower extremity. Previous investigators have demonstrated slightly lower Spo₂ measurements in the lower extremity than in the upper extremity in the newborn at 24 hours of life due to shunting at the level of the ductus arteriosus.⁶⁸  In general, the mean difference between the Spo₂ in the upper and lower extremities is <1%; however, some newborns with CCHD may have a more profound difference in saturation between the upper and lower body. For example, neonates with some forms of left obstructive heart lesions, such as critical coarctation of the aorta, in which the ductus arteriosus supplies a portion of the systemic flow, may have lower Spo₂ readings in legs than in the arm.⁵⁹ 

Some investigators proposed that oximetry screening should include measurements of both upper and lower extremities and that differences in Spo₂ of more than 3% or 4% be used to identify newborns with CCHD who might otherwise be missed by measuring lower-extremity Spo₂ alone.^55,79  One study that examined newborns with known CCHD suggested that the addition of upper and lower measurements would increase sensitivity from 89.4% to 92.4% without a decrease in specificity.⁷⁹  However, these data were not obtained in the setting of a screening birth cohort but rather among those with known CCHD. It is possible that the inclusion of both upper and lower Spo₂ measurements would result in a significantly higher false-positive rate. Moreover, screening both upper and lower extremities would increase the time required to screen a single newborn. Therefore, a single lower-extremity reading would appear to be the most appropriate for the purposes of large-scale screening.

The results of published studies using oximetry screening to detect CCHD in a representative birth population are shown in Table 3. Ten studies with a total of 123 846 infants screened reported a mean of 0.87% of infants with false-positive screens but a false-positive rate of 0.035% when screening was done after 24 hours; however, there was remarkable dispersion in reported screening performance. Five studies reported a low false-positive rate (≤0.1%) when measurements were made after 24 hours of life. The low false-positive rate is somewhat surprising given the reported variation of Spo₂ reported in normal newborn populations. It is not known whether there might be a publication bias in that only studies with favorable specificity might be published. A low false-positive rate would reduce the number of unnecessary echocardiograms. Nine of 10 studies listed in Table 3 reported sensitivity of <90%, ranging from 0% to 87%. This is explained in part by the fact that hypoxemia is not present in some forms of CCHD (Table 2).

False-positive results can be a cause for concern in public health newborn screening programs that are based on the laboratory analysis of dried blood spot specimens collected on filter paper cards. These false-positive results typically require families to be notified to bring their child in for further testing, and there can be a delay of several days before the results of such testing become available. False-positive newborn screening results have been reported to sometimes result in lasting parental anxiety and possibly elevated use of healthcare services.⁸⁰  In the case of pulse oximetry, this type of psychosocial risk of harm is very unlikely to be a problem in the typical hospital setting for infants not subject to early discharge. A positive test result leads to an immediate referral for an echocardiogram, and the results are reported before discharge. However, when the birth center does not have ready access to cardiac consultation, delay in hospital discharge or transfer to another facility may result in anxiety and added stress.

Oximetry screening may be less effective at identifying some CCHD lesions at greatest risk for acute cardiovascular compromise, namely, obstructive left heart lesions. A published analysis of oximetry has suggested that the difficulty in detecting hypoplastic left heart syndrome, aortic stenosis, and coarctation of the aorta limits the usefulness of this screening tool.⁸¹  However, it should be noted that nearly all forms of CCHD—even those unrelated to left heart obstruction—can result in serious morbidity and even death when diagnosis is delayed.^25,31  Moreover, oximetry can detect a significant number of newborns with obstructive left heart lesions and right-to-left shunting at the ductus arteriosus (Table 4). In published series, hypoplastic left heart syndrome was detected in all cases, and coarctation of the aorta was detected in just over half the cases. Studies that have obtained Spo₂ measurements on newborns with known CCHD have similarly reported that a lower-extremity Spo₂ of ≤95% detected hypoplastic left heart syndrome in all cases and critical coarctation of the aorta in the majority of cases.^59,79 

Several studies of screening oximetry have reported incidental findings of persistent fetal circulation, defined as elevated pulmonary vascular resistance and right-to-left shunting at the ductus arteriosus. In some reports, these cases have been reported as false-negative findings. In other studies, the investigators have emphasized the benefits of identifying these patients.⁶⁹  The finding of persistent fetal circulation in otherwise healthy newborns may be of benefit to medical care.⁸³  An understanding of the outcome of newborns who are asymptomatic with a decreased lower-extremity Spo₂ will be needed to understand whether identification of this population is a true benefit of oximetry screening.

There are technical limitations to oximetry measurement in the newborn. As noted above, the mean Spo₂ in the newborn at >24 hours of age is 97% to 98%; however, when continuous pulse oximetry is used, multiple investigators have demonstrated periodic and/or sustained desaturation below 95% during sleep, feeding, and crying.^65,68,84,85  Sustained rather than variable hypoxemia is consistent with the diagnosis of cyanotic congenital heart disease. Low oximetry readings in the setting of normal arterial oxygen saturation have been reported by multiple investigators.^76,86,87  In fact, falsely low oximetry readings in the newborn population are known to be associated with low peripheral perfusion and motion artifact,^88,89  probe placement site and partial probe detachment,⁹⁰  and hyperbilirubinemia or dyshemoglobinemias. It is known that technical differences between the various types of oximeters in general use include measurement of functional or fractional oxygen saturation, preset signal-averaging times, and methods for the exclusion of motion artifact.⁹¹  There has been some research into the variability among various commercially available pulse oximeters; however, most of the variability occurs in the cyanotic range (<90%) or at the highest saturations (99% to 100%). The peak performance of the commercially available oximeters occurs in the range of 92% to 97%.⁹²  Therefore, in the critical range for oximetry screening (94% to 97%), the variability of the most commonly used oximeters should be negligible.

There has also been concern that pulse oximeters may not be as accurate in darkly pigmented adults and children. At low Spo₂ levels (<70%), commercially available oximeters appear to overestimate arterial saturation by 3% in darkly pigmented subjects.⁹³  However, when Spo₂ is >90%, measurement bias related to skin pigmentation appears negligible (<0.2%). Lastly, the quality of oximetry measurements may be lower when performed in a screening setting.⁹⁴ 

When neonates are identified as having hypoxemia (Spo₂ ≤95%), it is necessary to evaluate them for CCHD. Although physical examination, chest radiography, and electrocardiography can assist in this process, echocardiography is now considered the definitive diagnostic modality. Whenever possible, the echocardiograms should be interpreted by pediatric cardiologists; major errors in the interpretation of a newborn echocardiogram by trained pediatric cardiologists are rare.⁹⁵ 

Although the majority of metropolitan areas in the United States have access to pediatric subspecialists, such as pediatric cardiologists, availability in rural areas can be limited. Approximately 15% of births in the United States occur in nonmetropolitan areas.⁹⁶  In these settings, echocardiograms are often performed by sonographers without formal pediatric training and are interpreted by adult cardiologists. Several investigators have found that the accuracy of pediatric echocardiograms interpreted by adult cardiologists is low.^95,97  One alternative is to use telemedicine, in which echocardiograms are interpreted distantly at a pediatric referral center.^98,99  The accuracy may be improved by direct guidance of the sonographers by a pediatric cardiologist via videoconferencing. This approach, which has been shown to be efficient and accurate, may be required to enhance detection of CCHD in rural or underserved areas. Another option is for newborns with suspected CCHD to be transported to a tertiary center. This strategy, however, would be expensive and impractical in many cases.

The cost of routine pulse oximetry performed on asymptomatic newborns after 24 hours of age includes both the direct cost of the pulse oximetry and the follow-up costs of any additional examinations and transfers. The largest direct cost component is staff time. At experienced centers, it may take a technician only 45 seconds on average to perform pulse oximetry on a newborn infant. The cost of diagnostic evaluation of infants who are referred for further examination after pulse oximetry depends on the frequency of referral, the duration of the diagnostic evaluation, and the ability for the evaluation to be performed without transfer to another center. A detailed cost accounting, to be reported elsewhere, indicates an average cost of approximately $1 per asymptomatic newborn infant, which includes the cost of diagnostic evaluations, in hospitals with moderate obstetric volume and ready access to pediatric echocardiography. Further work is needed to assess the cost and yield of routine pulse oximetry examination of newborns in a wider range of settings.

Oximetry to enhance the detection of CCHD has been considered previously in an evidence review sponsored by the United Kingdom's National Health Service Health Technology Assessment program.¹⁰⁰  The investigators observed that pulse oximetry is much more effective than current clinical practice in identifying infants with CCHD and more accurate and much less expensive than screening all newborns with echocardiography. The incremental cost per timely diagnosis of life-threatening congenital heart defects was calculated to be approximately $10 000 for pulse oximetry and $10 million for screening echocardiography. Although pulse oximetry was regarded as more promising than either the current practice or other options, the report called for further research to improve estimates of test performance and to inform timing, diagnostic, and management strategies and to “investigate the psychosocial effects of newborn screening for congenital heart disease” (p 127).¹⁰⁰  Another report has suggested families were quite receptive to newborn screening with pulse oximetry, with 99.8% of a sample of parents in Poland reported to approve of the screening technique.⁷⁰ 

The association of delayed diagnosis of CCHD with mortality, morbidity, and disability provides a rationale for strategies such as pulse oximetry assessment to improve early detection. Some studies have reported a reasonable detection rate with pulse oximetry; however, the usefulness of oximetry in clinical practice is not well established (Class IIb, Level of Evidence C; Level of Evidence C corresponds to observational studies [case-control and cohort design]). Additional studies in larger populations and across a broad range of newborn delivery systems are needed to determine whether this practice should become the standard of care in the routine assessment of the neonate.

Currently, pulse oximetry is being performed routinely in some delivery centers in the United States and elsewhere.¹⁰¹  Because pulse oximetry cannot detect all cases of CCHD, the diagnoses in some infants will be missed until after discharge from the newborn nursery. Such cases will provoke the question of whether the newborn oximetry screen was performed accurately. Therefore, it is reasonable for centers that routinely use pulse oximetry to ensure the fidelity of oximetry measurements through periodic quality assessment. Parents and caretakers should also be informed that pulse oximetry cannot detect all cases of CCHD, and hence, a negative test result does not exclude the possibility of heart disease.

Collaborative studies among hospitals conducting routine pulse oximetry should analyze pooled data and report detection, false-positive rates, and false-negative rates of CCHD. A pilot study of pulse oximetry screening has recently completed enrollment at 6 English hospitals by the National Institute for Health Research.¹⁰²  In addition, a comprehensive assessment of the impact of pulse oximetry assessment and early detection of CCHD on morbidity, postoperative survival, and hospital costs will allow a more critical evaluation of the economic impact of efforts to improve timely diagnosis of CCHD.