- FP —
- false positive
- POS —
- pulse oximetry screening
- CCHD —
- critical congenital heart defect
Pulse oximetry screening (POS) is an accepted test that improves detection of critical congenital heart defects (CCHD).1 Although outcome data are lacking, there is agreement among clinicians that POS identifies infants with CCHD before discharge.
After consideration by an expert workgroup, POS was adopted onto the US Recommended Uniform Screening Panel,2 and other countries have either introduced, or are considering introducing, POS.1 Despite this, there is considerable variation in screening, particularly the algorithm used.3 Differences include (1) pre- and postductal saturations (right hand and either foot) versus single postductal measurement (foot only) and (2) timing of screening (ie, before or after 24 hours). In algorithms using 2 limb measurements, there are also differences: inclusion of saturations <95% in 1 or both limbs and the absolute value of the differential between the 2 in determining positive results.1 So which algorithm is best?
When evaluating algorithms, it is important to consider sensitivity, specificity, false-positive (FP) and false-negative rate. It is also vital that screening leads to timely diagnosis (ie, before presentation with acute collapse). Meta-analysis of POS studies shows that overall, the test has moderate sensitivity (∼75%) and high specificity (99.8%), with no significant difference in sensitivity between pre/–post versus postductal testing or timing.3 However, analysis of raw saturation data from infants who had both limb measurements shows that some infants with CCHD would be missed by postductal testing alone.1 In addition, the FP rate is significantly higher with earlier testing (<24 hours).3 These factors were deemed important by the US workgroup considering the POS evidence, and their recommendation was that screening should include both pre and post measurements and be performed after 24 hours.2 This resulted in the algorithm introduced in the United States (Fig 1).2
A low FP rate is clearly important, but the strict definition of a FP is any test-positive infant who does not have CCHD. Interestingly, analysis of recent POS studies shows that many FPs (30%–80%) have alternative noncardiac conditions (eg, congenital pneumonia, early-onset sepsis, or pulmonary hypertension), which may be equally as life threatening as CCHD if diagnosed late.1,4–6 These conditions may benefit from earlier diagnosis and represent an important additional advantage. Also, important noncritical cardiac defects (eg, atrioventricular septal defects, ventricular septal defects) are identified as FPs,1,4–6
Timing of the Test
In published studies that adopted earlier screening,3,6 the FP rate was higher, but more noncardiac disease was identified; this is because such infants are more likely to develop hypoxemia within 24 hours and therefore be picked up by earlier screening.
Careful analysis of later screening studies4,5 reveals important additional findings. In Granelli’s4 and Riede’s5 studies, half of eligible infants with CCHD presented with symptoms before screening could take place; 28 of 57 infants with CCHD in Granelli’s and 18 of 36 in Riede’s. In Granelli’s,4 more than 10% of infants with CCHD (6 of 57) presented with acute collapse in hospital (the very situation that screening aims to prevent). It is well documented that infants with CCHD who collapse before surgery have worse outcomes and greater risk of neurodevelopmental complications,1 so these potentially avoidable collapses (ie, if screening was done earlier) may have significant consequences.
Although earlier screening results in more test-positive infants, it is important to balance a low FP rate with timely diagnosis. Some FP infants will be healthy, having transitional circulation; but others have life-threatening noncardiac conditions, and the earlier these are identified, the better. In some countries, mothers and infants are discharged from hospital within 24 hours after birth, and an increasing proportion is born at home. In these circumstances, later screening in hospital is not practical. UK evidence (screening at a mean age of 7 hours) reported a test-positive rate of 0.8%7 (similar to PulseOx study6). With ∼26 000 infants screened, 9 CCHDs were identified and, within the FPs, 79% had a significant medical condition. One of the major concerns regarding a high FP rate is the increased need for specialist assessment, particularly echocardiography, which can be challenging in some areas. Only 29% of test-positive infants in the UK study underwent echocardiography (mainly because an alternative non-cardiac diagnosis was established), and the echocardiography was positive in 48%.7 This compared favorably with infants in the same unit undergoing echocardiography for asymptomatic murmur.7
The experience after the introduction of POS in New Jersey was recently reported.8 Almost 73 000 babies were screened (after 24 hours), and the FP rate was 0.04%. However, only 3 infants with CCHD and only 12 infants with noncardiac conditions were detected. Although the FP rate is admirably low, the number of infants with CCHD is also low. In the United Kingdom, it took 2873 screens to detect 1 CCHD versus 24 231 screens in the United States.7 The likelihood is that in the US cohort, many infants with CCHD presented before screening took place. These important considerations led to the Nordic countries recommending screening at <24 hours.9
Differences in Definition of a Test-Positive Result
The UK study used the PulseOx6 algorithm, which, in addition to a difference in timing, has subtle differences in the definition of test positivity (Fig 2). In the US algorithm, test positivity is defined as saturation <90% in either limb or saturations between 90% to 94% in both limbs, or a difference of >3% between the 2, on 3 separate occasions (ie, 2 retests each after 1 hour, before clinical assessment). In the United Kingdom, a test-positive saturation is less than 95% in either limb or a difference of >2% on 2 occasions (ie, 1 retest after 2 hours, which is preceded by clinical assessment).
Do these minor differences matter? Examining the raw pre/–post saturation data and applying the US protocol to the PulseOx patients would have missed 1 CCHD (detected prenatally) and 2 serious CHDs. These numbers are small but may be important when scaled up nationally; additional evidence is required before a precise estimate of the difference can be stated with conviction. The application of a second retest almost certainly reduces the FP rate (infants with transitional circulation improve between screens) but because the majority (up to 80%) of infants who test positive after 1 retest have a significant condition,7 the second retest before clinical assessment potentially introduces a delay in diagnosis and treatment, which may result in a worse outcome.
So should the US screening algorithm remain as it is, or should a change be considered? Countries wishing to introduce screening may, quite rightly, wish to follow tried and tested practice but unfortunately there are limited data reporting outcomes of US screening. Additional research is probably unnecessary; however, collection and analysis of saturation data from populations already being screened is required to refine the minor differences in the algorithm.
The evidence to support a change in timing is perhaps more convincing, but it is important to accept that test positives will increase (to ∼0.8%) with earlier screening and many noncardiac conditions are identified in addition. Of course, POS uses hypoxemia as a proxy for CCHD and does not detect CCHD directly. Given this, should we screen for all hypoxemic conditions rather than just CCHD? The concern is that hypoxemia is not a condition as such and newborns may have “physiologic” hypoxemia as the cardiorespiratory systems adapt to extrauterine life. Once again, the earlier screening takes place, the more likely this “transitional circulation” is identified in test-positive infants. The other major concern is that although we have good data for CCHD, there are no robust data on the accuracy of POS for noncardiac conditions. This presents difficulties for public health decision makers sanctioning POS as a valid test for these conditions. Perhaps the best compromise is to continue POS for CCHD and accept that detection of noncardiac conditions (technically FPs but could be considered secondary targets) is an important additional benefit. Clinical staff and parents should be made aware of this.
Until these issues are resolved and more data are forthcoming, is it worth considering an algorithm that has a consistent slightly higher FP rate but will potentially identify more infants with life-threatening disease?
- Accepted July 11, 2016.
- Address for correspondence Andrew K. Ewer, Neonatal Unit, Birmingham Women’s Hospital, Edgbaston, Birmingham UK. B15 2TG. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- de-Wahl Granelli A,
- Wennergren M,
- Sandberg K, et al
- Riede FT,
- Wörner C,
- Dähnert I,
- Möckel A,
- Kostelka M,
- Schneider P
- Singh A,
- Rasiah SV,
- Ewer AK
- Copyright © 2016 by the American Academy of Pediatrics