Neonatal ECG screening and QT correction: the march towards consistency and accuracy

This editorial refers to ‘For neonatal ECG screening there is no reason to relinquish old Bazett’s correction’†, by M. Stramba-Badiale et al., on page 2888. The long QT syndrome (LQTS) is common, affecting between 1/1500 and 1/3000 persons.1 , 2 Debate exists regarding the utility of newborn universal ECG screening for the condition, but there are many arguments to support this approach.3–5 These include the relative frequency of the disorder, the fact that it probably accounts for at least 10% of sudden infant death syndrome deaths (as well as many later sudden deaths), feasibility of screening with a relatively inexpensive test (the ECG), and, perhaps most importantly, the condition is largely treatable. For all these reasons, though universal screening is not presently performed in most of the world and remains somewhat controversial, such screening has been routinely employed in some countries, most notably in Italy. There are many potential technical issues associated with assessing an ECG in the newborn period, including the timing of the test in relation to birth, the actual measurement of the QT interval, and, finally, the correction of the QT interval for heart rate which is naturally elevated. In this issue of the European Heart Journal, Stramba-Badiale and colleagues take advantage of the large data set of newborns who are screened in Italy in order to help determine how best to deal with the final of these three issues, specifically the issue of correction of the QT interval for the elevated heart rates of the newborn.6 At present, there are many correction formulae that have been previously validated and used, but this work reviewed the validity of two commonly used linear formulae (Hodges and Framingham), two non-linear exponential formulae (Bazett’s and Fridericia’s), and, finally, a newly created, non-linear exponential QT correction (QTc) formula (‘QTcNeo’) where the QTc was equal to QT/RRN in which N was derived to be 0.467 from evaluation of >2500 neonatal ECGs.7–10 These formulae were assessed in two ways. First, the authors sought to determine which approach provided a corrected QT interval that was most heart rate independent. Secondly, the authors used genetic testing in the ‘positive’ patients to determine which formula most correctly identified neonates with genetically confirmed LQTS. Using multiple statistical tests, the authors were able to demonstrate that all except the Hodges correction provided acceptable results in regards to correction of the QT interval for heart rate, with the most heart rate independent being the Bazett’s formula as well as the author’s novel QTcNeo formula. In terms of identifying patients with genetically positive LQTS, the formulae were comparable in regards to area under the receiver operating characteristic (ROC) curve, though the actual ‘cut-off’ values for an abnormal QTc were different between the different correction approaches. For Bazett’s formula, the ‘cut-off’ value of 460 ms for ‘prolonged’ was most appropriate. Though the QTcNeo formula was marginally better at providing heart rate-independent values, given the ubiquity of Bazett’s formula in clinical practice, its excellent negative predictive value, and fair positive predictive value, the authors ultimately suggest that the general reliability of this most common approach to QT correction means that ‘there is no reason to relinquish’ its use (Figure 1). They suggest that it is an appropriate solution to the problem of QT interval correction in this setting of newborns with naturally elevated sinus heart rates. Neonatal heart rate corrected QT (QTc) interval calculation. Bazett,7 Fridericia,8 Hodges,9 Framingham,10 and Neo,6 the latter from the present paper. RR = heart rate expressed as mean cycle length in milliseconds. Bazett’s formula remains unsurpassed for a combination of accuracy, utility, and simplicity. These findings are important in that they validate the use of the most commonly used correction formula (Bazett’s) for neonatal screening, and their analysis and conclusions are quite similar to those of Phan et al. who previously reported similar findings in a somewhat smaller cohort.11 The implications of the findings are clear. Bazett’s commonly available correction formula is appropriate for use in the newborn screening ECG setting of elevated sinus heart rates. For patients of all ages, Bazett’s formula also offers the value of being easy to use at the bedside with minimal mental arithmetic, deleting the final ‘0’. For example, if the RR = 500 ms, imagine that as 50, the square root of which is ∼7. If the QT = 260 ms, ‘think’ 26. Then 26/7 = a bit under 4, specifically 3.7, or QTc = 370 ms. This can be handy in clinical situations. Though Stramba-Badiale et al. present an important and convincing report, the larger issue of accuracy of the QT measurement is not addressed in this work, and it is this issue that may ultimately be the most challenging to address in any ECG screening assessment. As any practising electrophysiologist can attest, accurate measurement of the QT interval, even by an experienced cardiologist, can be challenging and error prone.12 , 13 Over 10 years ago, Taggart and colleagues reviewed the records of 176 consecutive patients referred to a large LQTS referral centre with a prior diagnosis of LQTS. Of these patient referrals with prior diagnosis of LQTS, 40% were determined not to have LQTS when assessed by this very experienced group, and one of the most common reasons for misdiagnosis was simple inaccurate measurement of the QT interval.13 Computer programs associated with ECG machines are programmed to assess and identify long QT intervals, but anyone with experience in ECG interpretation knows that these are plagued with inaccuracy and this has been borne out in multiple studies.14 Thus, though the findings of Stramba-Badiale and colleagues are critically important in demonstrating the reliability of the Bazett’s correction for this screening purpose, more efforts in determining mechanisms for proper and accurate QT measurement are needed. This may prove an area of ‘low hanging fruit’ for the medical artificial intelligence community.15 Use of such artificial intelligence might even result in the development of more accurate heart rate correction formulae. With machine learning and proper rules and techniques of QT measurements fed to the computers by experts such as Dr Stramba-Badiale and colleagues, it would seem that the exciting day of a computer that is able accurately to measure the QT interval better than humans is likely to be rapidly approaching. Until then, Bazett’s correction may be the best or most practical method to correct the QT value for heart rate properly and thus identify at-risk newborns. Conflict of interest: none declared.