Introduction Babies with cardiac anomalies are often asymptomatic at birth, and many remain undetected despite routine newborn examination. We retrospectively assessed the effect of routine pulse oximetry in detection of such anomalies from a hospital birth population of 31 946 babies born between 1 April 1999 and 31 March 2009.
Method 29 925 babies who were not admitted to the neonatal unit at birth underwent postductal oxygen saturation measurement before discharge. If saturation was below 95% an examination was performed. If this was abnormal or saturation remained low, an echocardiogram was performed. All babies with cardiac anomaly diagnosed before 1-year were identified from the region's fetal abnormality database.
Results Critical anomalies affected 27 infants (1 in 1180); 10 identified prenatally, 2 after echocardiogram was performed because of other anomalies, 2 in preterm infants, 2 when symptomatic before screening, 5 by oximetry screening, 1 when symptomatic in hospital after a normal screen and 5 after discharge home.
Serious anomalies affected 50 infants (1 in 640); 8 identified antenatally, 7 because of other anomalies, 3 in the neonatal unit, 5 by pulse oximetry screening, 11 by routine newborn examination, and 16 after discharge home.
Conclusions Routine pulse oximetry aided detection of 5/27 of critical and 5/50 of serious anomalies in this sample, but did not prevent five babies with critical and 15 with serious anomalies being discharged undiagnosed. Results from screening over 250 000 babies have now been published, but this total includes only 49 babies with transposition, and even smaller numbers of rarer anomalies.
- Neonatal screening
- Pulse oximetry
- Congenital cardiac anomalies
Statistics from Altmetric.com
What is already known on this topic
Some critical and some serious cardiac anomalies are missed by routine clinical examination in the newborn period.
Early detection of such anomalies can improve outcome.
Pulse oximetry screening can improve detection in the newborn period.
What this study adds
The number of published cases subjected to oximetry screening is too small to estimate the effectiveness of screening specific cardiac anomalies with precision.
Review of published data suggests that transposition of the great arteries can give normal saturation levels on early screening.
Routine newborn examination fails to detect serious congenital cardiac anomalies in an appreciable proportion of babies.1–3 The most likely reason why such anomalies are missed is that the babies do not show any signs of a problem at the time of the examination.4 Some units have tried to improve detection by screening all newborn babies for mild degrees of cyanosis using pulse oximetry. We published results of a pilot study exploring this technique in 2002.5 Pulse oximetry screening has been our standard practice since, and we report our experience over the first 10 years, placing the results in the context of other studies published to date.
Since 1 April 1999, all babies born at Sunderland Royal Hospital, unless admitted directly to the neonatal unit after delivery, have had a postductal oxygen saturation routinely measured over a 2 min period, after the age of 2 h and before discharge. The result of this measurement is then manually recorded in the baby's notes. Babies achieving an oxygen saturation of at least 95% while breathing air are presumed normal. The midwife undertaking the routine measurement clinically evaluates any baby that fails to reach this level. If this examination is normal and the baby is not yet 24 h old the measurement is repeated an hour or two later. If this second measurement remains below 95% a further clinical examination along with an echocardiogram should be performed. Our aim is that no baby should be discharged home without either a postductal saturation measurement in air of at least 95%, or an echocardiogram.
We defined critical cardiac anomalies as all cases of hypoplastic left heart, pulmonary atresia with intact septum, transposition of the great arteries (TGA) or interruption of the aortic arch, as well as the following diagnoses if resulting in death or operation in the first 28 days of life—coarctation of the aorta, aortic valve stenosis, tetralogy of Fallot, pulmonary atresia with ventricular septal defect (VSD), or total anomalous pulmonary venous connection (TAPVC).6 Any cardiac anomaly causing death or receiving surgical intervention in the first year, if not critical, was defined as serious.7 Isolated patent arterial duct was excluded, as were cases of trisomy 13 or 18.
In this region, details of all babies with a confirmed diagnosis of congenital heart disease are reported to the Northern Congenital Abnormality Survey (NorCAS).8 Freeman Hospital manages congenital heart disease in all babies in this region, and maintains a database with details of age at presentation, diagnosis and of cardiac investigation or operative interventions undertaken in these infants.9
Using these databases, we identified all babies born in Sunderland Royal Hospital between 1 April 1999 and 31 March 2009, and diagnosed with congenital heart disease under the age of 1 year. Congenital heart disease was taken to be ‘a gross structural abnormality of the heart or intrathoracic great vessels that is actually or potentially of functional significance’.10 The original baby notes for each case were retrieved and examined.
For 2 years, from 1 April 1999, all babies born at Sunderland Royal Hospital, unless admitted directly to the Neonatal Unit after delivery, had a postductal oxygen saturation measured over a 2 min period, after the age of 2 h and before discharge. These babies were part of a formal pilot study approved by the local research ethics committee which assessed the effectiveness of the technique in detecting congenital heart disease.5 This present paper reports details of all ‘critical’ and ‘serious’ cases from a birth cohort of 31 946 babies born in one hospital over a 10-year period, including 6166 babies born during the pilot study, and a further 25 780 babies born over the subsequent 8 years when the technique had been introduced into standard clinical practice. During the entire 10-year period, 1928 babies were admitted direct to the neonatal unit after delivery and were, thus, not subject to formal screening, and a further 93 babies were not formally screened during the pilot study for other reasons. The total population formally screened, thus comes to 29 925.
This study was unfunded. NorCAS is a member of the British Isles Network of Congenital Anomaly Registers. NorCAS has exemption from the UK National Information Governance Board from a requirement for consent for cases to be included in the register under section 251 of the National Health Service Act (2006), and has ethics approval (09/H0405/48) to undertake studies using its data.
Table 1 shows the timing of diagnosis of the 27 babies with critical and the 50 with serious cardiac anomalies. A further four babies with tetralogy of Fallot were identified in the first year of life but could not be formally defined as ‘serious’ because surgical intervention was not undertaken until shortly after the age of 1 year. Of these, 1 baby was diagnosed antenatally, 2 by routine newborn examination and 1 was preterm and detected in the neonatal unit. Because our study was retrospective, we cannot comment on false positive screening results.
Numbers of critical anomalies in table 1 are bold whereas those indicating serious anomalies are in italics. Five critical and five serious anomalies were diagnosed as a direct result of a response to a low saturation result. Two babies, both with total anomalous pulmonary venous connection (TAPVC), had low saturation results, but were not diagnosed before discharge. In one case, an echocardiogram was performed, but the diagnosis was missed, and in the other, the saturation when repeated at 72 h was normal.
Table 1 shows that 18 of the 77 critical and serious cases were identified antenatally, 10 by positive pulse oximetry screen, 11 by routine examination, 17 by other means before discharge, leaving 21 undiagnosed before discharge. Ten of the 18 antenatal diagnoses were critical rather than serious, as were 5 out of 10 screening diagnoses, but none of the routine examination successes was critical. Our data do not allow us to suggest whether any babies initially highlighted by saturation screening might have been diagnosed before discharge by some other means in the absence of oximetry screening.
Saturation measurements were often also obtained in the first 24 h in cases where attention was drawn to a baby for other reasons. We have assumed that these measurements were indicative of what would have been obtained by saturation screening (though babies less than 35 weeks’ gestation, or unwell for non-cardiological reasons, were excluded). Of the 81 babies mentioned above (77 critical and serious, plus four additional cases of tetralogy of Fallot) we can therefore suggest which of the 46 would have been detected by a routine postductal pulse oximetry test, as shown in table 2.
In 1999, when we first planned to explore the effect of pulse oximetry screening, we hoped that left ventricular outflow tract obstructive lesions might result in sufficient right to left flow through the arterial duct to affect postductal saturation. Unfortunately, this proved not to be the case as has been amply demonstrated in many subsequent studies (see table 2).
In this paper, we have opted to explore the effect of pulse oximetry screening in our hospital on the timely diagnosis of ‘critical’ and ‘serious’ congenital heart disease as we have defined them. Both definitions include some specific cardiac anomalies that one would not expect to be susceptible to pulse oximetry screening. For example, the critical group includes left ventricular outflow tract obstructions, such as interrupted aortic arch, coarctation and critical aortic valve stenosis which, in one large population-based study, together constituted about 35% of all critical cases.6
Any definition of severity which depends on the timing of intervention is problematic in that delay in diagnosis caused by failure to detect the problem will reduce the severity classification of the anomaly if the baby survives with the anomaly undetected for a sufficient period. In such a case, one might argue that had the anomaly been detected earlier, intervention would have occurred earlier, and a higher classification might have resulted. Such a reclassification affected one of our cases of TAPVC. Similarly, cases of serious anomaly where the age at operation exceeds 1 year by however small a margin will fail to be classified as ‘serious’, as happened with four cases of Fallot's tetralogy in this study.
Attention is now focussing on oximetry in the detection of critical cardiac anomalies.11 Congenital cardiac anomalies occur in a little over six per 1000 livebirths but only about 15% of these are critical. Thus, critical anomalies affect a little less than one per 1000 livebirths. This means that even studies involving much larger birth populations than our study are best evaluated when data is combined with that from other studies.
Opinion is now strengthening in favour of routine pulse oximetry screening despite a number of differences in technique, in saturation thresholds and the very small number of cases of specific cardiac anomalies included in published studies.12–,15 We have identified 13 studies of pulse oximetry screening for congenital cardiac anomalies in the newborn period.5 ,7 ,15–25 When interpreting this data, it is important to bear in mind that some studies involve oximetry screening performed after 24 h of age, whereas in others, screening takes place during the first day of life (see table 2). Also, 10 of the 13 studies required action if the postductal saturation failed to reach 95%, whereas, three used a threshold of 96%.15 ,23 ,25
A second issue concerns the benefit of measuring both preductal and postductal saturation which has been recommended for use in the USA.14 Two studies measured both preductal and postductal saturation, and also required action if the difference in saturation between these measurements exceeded 3%21 or 2%,7 even if both measurements were above 94%. (A third study also measured both preductal and postductal saturation, but only demanded action if either was below 94%.17) Though this double measurement may improve detection, the published evidence probably involves only five cases. The PulseOx study found 26 cases with actionable results, but this included only four cases where action was indicated purely on the basis of the pre-/postductal saturation percentage difference (in three cases, the difference was 4% or more). Two of these cases, both hypoplastic left heart, had been diagnosed antenatally.7 The de-Wahl Granelli study detected 19 cases by oximetry screening, but only one of these had both a postductal measurement above the trigger level as well as a pre-/postductal percentage difference of more than 3%. Unfortunately, this baby was discharged undiagnosed, a protocol violation presumably due to human error.21 Bakr's study used both preductal and postductal measurements, and all five cases detected by oximetry recorded saturations below 95%, but it is not clear which, if any, of these low measurements were preductal rather than postductal.17
A third issue is the reliability of saturation measurements. By comparing electronic records from the saturation monitors used for screening, and the nursing logs used for recording the results, Reich et al24 judged the reliability of the readings. Pulse oximetry did not assist diagnosis of any of the 12 affected babies in their study, and the only baby to be discharged home undiagnosed after normal oximetry was judged to have an unreliable and unverifiable reading, as well as a normal routine examination. Liske et al26 suggested that equipment used for pulse oximetry screening should have ‘the capacity to correlate the oxygen saturation value with the quality of the reading and also the capability to store or print a permanent record’. This would certainly help in auditing the effectiveness of the technique, as well as providing an improved record of the screen if faced with complaints when anomalies are not detected. Liske's suggestion has not been included in the recommendations from the American Association of Pediatrics, but development of equipment incorporating this facility deserves serious consideration.15
Many would suggest that TGA would be an anomaly that could not be missed if pulse oximetry was used. However, Meberg reported a screening saturation of 96% in a baby with transposition and VSD discharged undiagnosed.22 Ewer also reported a similar case (case 21) as did Riede.7 ,25 One of the babies in our study, with an antenatal diagnosis of transposition (with VSD) postnatally confirmed, recorded a postductal saturation of 96% at 3 h of age. In addition, in the month following the end of our study, we had experience of a baby with transposition and a large arterial duct where the screening result was normal at 3 h of age, and subsequent routine newborn examination failed to detect a problem. Though we cannot exclude faulty technique, the fact that these five instances arise from four separate studies suggests that in some cases of transposition, surprisingly high postductal saturations can be achieved sufficient to affect the sensitivity of pulse oximetry screening. It is possible that fewer such cases would be missed if preductal and postductal measurements were the norm, however, one of these cases was missed despite such combined screening.
A fourth issue concerns the interpretation of cases where the postductal saturation gradually reverts to normal. A review in 2007 suggested that, in the continued absence of symptoms, if the postductal saturation is between 90% and 94%, one might repeat the measurements until about 72 h of age in the hope of obtaining a normal result.27 However, our experience of missing a case of total anomalous pulmonary venous connection whose saturation ‘normalised’ on day 3 has led us to insist on an echocardiogram if the saturation is abnormal after 24 h of age. Bearing in mind that Koppel et al15 encountered only one false positive among over 11 000 screened babies when screening was limited to babies over 24 h of age.
A fifth issue is the paucity of data on screening in specific important cardiac anomalies. Adding our data to that from all previously published studies gives an apparently impressive total of over 250 000 babies screened. Despite this, table 2 reveals that the number of cases of specific cardiac anomalies subjected to oximetry screening is pitifully small. If one includes all babies where a saturation result was obtained on a similar timescale to screened babies, even if the possibility of cardiac anomaly had already been raised for other reasons, such measurements are available in only 50 cases of coarctation/aortic arch interruption, 49 cases of transposition, 28 cases of Fallot's tetralogy, 22 cases of hypoplastic left heart, and correspondingly smaller numbers of rarer conditions. As a consequence, the 95% CIs on published potential oximetry screening success by cardiac diagnosis are very wide. However, even allowing for these small numbers, it is pretty clear that pulse oximetry is relatively insensitive in the detection of coarctation of the aorta/aortic arch interruption (95% CI 24 to 50%) and tetralogy of Fallot (24–58%).
If pulse oximetry screening is to be introduced more widely, it pays to be aware of the many shortcomings of the data so far published, and the fact that anomalies one might assume would always show some cyanosis, such as transposition, can have a normal percentage saturation in some cases. Before introducing such screening, it is vital that a timely and robust system for accurate diagnosis of any infants with positive results is established, bearing in mind that a positive result has a significant chance of indicating a non-cardiac problem, usually a respiratory one. The de-Wahl Granelli study identified 31 non-cardiac problems amongst their 69 ‘false-positives’.21 We should also remember that a number of important cardiac anomalies are not amenable to pulse oximetry screening.
The authors would like to thank Mary Bythell and the staff of NorCAS, as well as countless obstetricians, paediatricians, radiologists, ultrasonographers, geneticists, paediatric surgeons, paediatric cardiologists, paediatric pathologists whose commitment maintains the data. We are also grateful to the midwives and obstetric staff in Sunderland hospital.
Author note Since this paper was published Online First, one of the authors, Sam Richmond has died.
Contributors MAH and SR devised the original study and planned the 10-year review. SP gathered the original data, performed the first analysis and produced the first draft. CW and MAH classified the cases and CW contributed details of timing and type of interventions. All the authors worked on the subsequent drafts, and have contributed to the final draft.
Competing interests None.
Ethics approval Sunderland local research ethics committee.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.