For neonatal ECG screening there is no reason to relinquish old Bazett’s correction

For neonatal ECG screening there is no reason to relinquish old Bazett’s correction Abstract Aims There is an almost endless controversy regarding the choice of the QT correction formula to be used in electrocardiograms (ECG) in neonates for screening for long QT syndrome (LQTS). We compared the performance of four commonly used formulae and a new formula derived from neonates. Methods and results From a cohort of 44 596 healthy neonates prospectively studied in Italy between 2001 and 2006, 5000 ECGs including 17 with LQTS-causing mutation identified by genotyping were studied using four QT correction formulae [Bazett’s (QTcB), Fridericia’s (QTcF), Framingham (QTcL), and Hodges (QTcH)]. A neonate-specific exponential correction (QTcNeo) was derived using 2500 randomly selected ECGs and validated for accuracy in the remaining 2500 ECGs. Digital ECGs were recorded between the 15th and 25th day of life; QT interval was measured manually in leads II, V5, and V6. To assess the ability to provide heart rate (HR) independent QT correction, regression analysis of the QTc-HR plots for all 5000 ECGs with each correction formula was done. QTcB provided the most HR independent correction with a slope closest to zero (slope +0.086 ms/b.p.m.) followed by QTcF (slope −0.308 ms/b.p.m.), QTcL (slope −0.364 ms/b.p.m.), and QTcH (slope +0.962 ms/b.p.m.). The QTc-HR slope of QTcNeo (QT/RR0.467) was similar to QTcB. The ability to correctly identify neonates with LQTS was best with QTcB, QTcF, and QTcNeo (comparable areas under the receiver operating characteristic curves) with positive predictive value of 39–40% and sensitivity of 100%. Cut-off values were 460 ms for QTcB, 394 ms for QTcF, and 446 ms for QTcNeo. Conclusions The Bazett’s correction provides an effective HR independent QT correction and also accurately identifies the neonates affected by LQTS. It can be used with confidence in neonates, although other methods could also be used with appropriate cut-offs. View largeDownload slide View largeDownload slide Bazett’s formula, Long QT syndrome, Neonatal ECG, QT interval correction Introduction The long QT syndrome (LQTS) is a genetic disease with a prevalence of 1 in 2000 live births1 and is associated with a high risk for life-threatening arrhythmias among untreated patients.2,3 However, with appropriate treatment, mortality has now been reduced to below 1%.3 Sudden death is often the sentinel event, and cardiac events usually appear in childhood or adolescence. Long QT syndrome also contributes to almost 10% of cases of sudden infant death syndrome4 and early identification of LQTS through electrocardiogram (ECG) screening performed in the first month of life is likely to reduce mortality.5 The usefulness of neonatal ECG screening for the identification of LQTS has been demonstrated by two separate prospective studies involving almost 78 000 infants.1,6 Even though the number of clinical centres worldwide offering a service of neonatal electrocardiography is constantly growing, the question of which is the best method for the heart rate (HR) correction of the QT interval in neonates, given their high HRs remains unanswered. The method used in most studies in newborns1,6,7–9 is the one derived from the article by Bazett10 in 1920 (corrected QT or QTcB = QT/ RR). However, the endless criticism that this formula overcorrects the QT interval at high HRs has led many investigators to propose alternative formulae.11–13 Most of these publications suffer from several limitations: (i) many small cohorts, (ii) large cohorts but with automated measurements with various degrees of unreliability, and (iii) validation in children or adult populations but not in neonates. Moreover, no attempts were made to verify whether or not the other QT correction methods could reliably identify infants with an established diagnosis of LQTS which, after all, should be the main reason to measure their QT interval. This study was specifically designed to verify in a large population of infants which, among the many QT correction methods currently used or proposed, is superior and sufficiently reliable to identify those with LQTS. Methods This study was performed using ECGs recorded in healthy neonates, enrolled in a prospective study, which assessed and defined the prevalence of congenital LQTS.1 Study population The original study1 included 44 596 healthy neonates (43 080 Caucasians), 22 967 males (51%), and 21629 females (49%) consecutively enrolled by 18 Italian maternity hospitals between January 2001 and June 2006, and in whom an ECG was recorded between the 15th and the 25th day of life. All parents signed an informed consent for recording the ECG. Very premature and sick newborns who required transfer to neonatal intensive care units were excluded from the study. The protocol requested genetic analysis in all infants with a QTc >470 ms and in their parents, and recommended it in those with a QT >460 ms. All coding exons of KCNQ1, KCNH2, SCN5A, KCNE1, KCNE2, CAV3, and SCN4B (the genes screened at that time in our laboratory for the routine diagnosis of LQTS) were analysed for presence of disease-causing mutations. The details of the genetic analysis have been described.1 There were 59 neonates with a QTc >460 ms and 43 of them (90% of those above 470 ms and 50% of those above 460 ms) underwent genetic analysis. For this study, we analysed 5000 ECGs, which included all 43 ECGs with QTc >460 ms of the infants who underwent genotyping, and another 4957 ECGs all recorded at a single centre to reduce variability (this centre enrolled 5007 infants but we randomly excluded 50 of them to end up with a cohort of 5000). All these tracings were part of the 44 553 ECGs from the original dataset. These 5000 neonates included 2562 (51.2%) males and 2438 (48.8%) females. ECG recording and analysis The ECGs were recorded at a paper speed of 25 mm/s with a Marquette MAC 5000 recorder (digital sampling rate 500 Hz) and were transmitted from the participating centre via modem to the study co-ordinating centre and were read centrally. The QT intervals and the preceding RR intervals were measured on-screen manually in leads II, V5, and V6 from five non-consecutive beats. During ECG analysis, the QTc was calculated according to Bazett’s formula (QTcB) and the longest mean value of the five beats found in one of the three leads was considered. In neonates with a QTcB >450 ms, another ECG was recorded within 1 to 2 weeks to confirm the initial finding. ECGs were analysed as per the European Society of Cardiology guidelines for the interpretation of neonatal ECG and if QT prolongation or any other ECG abnormality was identified, the infants were managed and treated accordingly.14 Bradycardia was defined as HR <107 b.p.m. and tachycardia as HR >182 b.p.m.14 The QT and RR interval values from each ECG were used to calculate the corrected QT (QTc) interval using four commonly used HR correction formulae: Bazett10 QTcB = QT/RR Fridericia15 QTcF = QT/RR3 Hodges16 QTcH = QT + 1.75 × (Heart rate - 60) Framingham17 QTcL = QT + 0.154 × (1 - RR) Bazett10 QTcB = QT/RR Fridericia15 QTcF = QT/RR3 Hodges16 QTcH = QT + 1.75 × (Heart rate - 60) Framingham17 QTcL = QT + 0.154 × (1 - RR) Bazett10 QTcB = QT/RR Fridericia15 QTcF = QT/RR3 Hodges16 QTcH = QT + 1.75 × (Heart rate - 60) Framingham17 QTcL = QT + 0.154 × (1 - RR) Bazett10 QTcB = QT/RR Fridericia15 QTcF = QT/RR3 Hodges16 QTcH = QT + 1.75 × (Heart rate - 60) Framingham17 QTcL = QT + 0.154 × (1 - RR) Statistical analysis All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA). Differences in QTc and RR intervals between males and females were compared with the Student’s t-test. To identify which of the QT correction formulae was most appropriate for use in neonates, we plotted the QTc against HR values from each ECG and the slope with its 95% confidence interval (CI) of QTc-HR regression line was determined for each formula by linear mixed effects model using the PROC MIXED function in SAS. The QT correction formula with the slope of the regression line closest to zero was considered the best as it would provide a QTc value that was most independent of HR. Comparison of slopes between correction formulae was also performed by including the ‘correction method × concentration’ interaction term in the linear mixed effects model and the mean difference with 95% CI was reported. To determine if a neonate-specific QT correction (QTcNeo) formula was better than the other formulae studied, the dataset of 5000 ECGs was randomly split into two equal halves (a study dataset and a validation dataset of 2500 ECGs each), using a random number generator. For calculating the correction factor, the relationship between the HR and QT values in the study dataset was evaluated by the formula: Log QT = a + b*Log RR. Having obtained the value of b, this was then used to calculate QTcNeo using the formula QTcNeo = QT/RRb. The performance of the QTcNeo formula was evaluated using the remaining 2500 ECGs by comparing the slope of its QTc-HR regression line with the slope of the QTc-HR regression line of the best of the four other QT correction formulae studied. To assess the diagnostic performance of each correction method in identifying neonates with or without a LQTS disease-causing mutation, receiver operating characteristic (ROC) curves were generated. The area under the ROC (AuROC) was calculated with its 95% CI using the PROC LOGISTIC function in the SAS statistical software package. A χ2 test of significance was used to compare the areas under correlated ROC curves for different QTc methods,18 and the one with the largest AuROC was considered the best correction method for this purpose. The Youden’s J statistic was used to determine the cut-off values for each QT correction formula that best separated those with LQTS from those who did not. A two-sided P-value <0.05 was considered statistically significant. Results The mean (± standard deviation) HR for the 5000 ECGs was 154 ± 17 b.p.m. and mean QT interval was 252 ± 19 ms (Table 1). The QTc values by the four QT correction formulae (Bazett’s, Fridericia’s, Hodges’, and Framingham) did not differ significantly between males and females (Table 2). Of the 2438 females, five had bradycardia and 99 had tachycardia. Similarly, 3 of the 2562 males had bradycardia and 122 had tachycardia. Table 1 Mean (± standard deviation) and range of hazard ratio, QT interval, and corrected QT interval by four previously published correction formulae for 5000 neonates ECG parameters QT correction formula Mean ± SD Minimum Maximum Heart rate (b.p.m.) 154 ± 17 94 214 QT (ms) 252 ± 19 200 400 QTcB (ms) Bazett 402 ± 21 341 555 QTcF (ms) Fridericia 344 ± 19 286 497 QTcH (ms) Hodges 417 ± 21 340 497 QTcL (ms) Framingham 345 ± 15 302 474 ECG parameters QT correction formula Mean ± SD Minimum Maximum Heart rate (b.p.m.) 154 ± 17 94 214 QT (ms) 252 ± 19 200 400 QTcB (ms) Bazett 402 ± 21 341 555 QTcF (ms) Fridericia 344 ± 19 286 497 QTcH (ms) Hodges 417 ± 21 340 497 QTcL (ms) Framingham 345 ± 15 302 474 SD, standard deviation. Note: QTcB by Bazett's formula; QTcF by Fridericia's formula; QTcH by Hodges' formula; QTcL by the Framingham formula. Table 1 Mean (± standard deviation) and range of hazard ratio, QT interval, and corrected QT interval by four previously published correction formulae for 5000 neonates ECG parameters QT correction formula Mean ± SD Minimum Maximum Heart rate (b.p.m.) 154 ± 17 94 214 QT (ms) 252 ± 19 200 400 QTcB (ms) Bazett 402 ± 21 341 555 QTcF (ms) Fridericia 344 ± 19 286 497 QTcH (ms) Hodges 417 ± 21 340 497 QTcL (ms) Framingham 345 ± 15 302 474 ECG parameters QT correction formula Mean ± SD Minimum Maximum Heart rate (b.p.m.) 154 ± 17 94 214 QT (ms) 252 ± 19 200 400 QTcB (ms) Bazett 402 ± 21 341 555 QTcF (ms) Fridericia 344 ± 19 286 497 QTcH (ms) Hodges 417 ± 21 340 497 QTcL (ms) Framingham 345 ± 15 302 474 SD, standard deviation. Note: QTcB by Bazett's formula; QTcF by Fridericia's formula; QTcH by Hodges' formula; QTcL by the Framingham formula. Table 2 Heart rate, QT interval, and corrected QT interval (mean ± standard deviation) by four correction formulae in male and female neonates Females (n = 2438) Males (n = 2562) Heart rate (b.p.m) 154 ± 17 155 ± 17 QT (ms) 253 ± 19 251 ± 18 QTcB (ms) 403 ± 21 401 ± 21 QTcF (ms) 345 ± 19 343 ± 18 QTcH (ms) 417 ± 21 417 ± 21 QTcL (ms) 346 ± 15 344 ± 14 Females (n = 2438) Males (n = 2562) Heart rate (b.p.m) 154 ± 17 155 ± 17 QT (ms) 253 ± 19 251 ± 18 QTcB (ms) 403 ± 21 401 ± 21 QTcF (ms) 345 ± 19 343 ± 18 QTcH (ms) 417 ± 21 417 ± 21 QTcL (ms) 346 ± 15 344 ± 14 Note: QTcB by Bazett’s formula; QTcF by Fridericia’s formula; QTcH by Hodges’ formula; QTcL by the Framingham formula. Table 2 Heart rate, QT interval, and corrected QT interval (mean ± standard deviation) by four correction formulae in male and female neonates Females (n = 2438) Males (n = 2562) Heart rate (b.p.m) 154 ± 17 155 ± 17 QT (ms) 253 ± 19 251 ± 18 QTcB (ms) 403 ± 21 401 ± 21 QTcF (ms) 345 ± 19 343 ± 18 QTcH (ms) 417 ± 21 417 ± 21 QTcL (ms) 346 ± 15 344 ± 14 Females (n = 2438) Males (n = 2562) Heart rate (b.p.m) 154 ± 17 155 ± 17 QT (ms) 253 ± 19 251 ± 18 QTcB (ms) 403 ± 21 401 ± 21 QTcF (ms) 345 ± 19 343 ± 18 QTcH (ms) 417 ± 21 417 ± 21 QTcL (ms) 346 ± 15 344 ± 14 Note: QTcB by Bazett’s formula; QTcF by Fridericia’s formula; QTcH by Hodges’ formula; QTcL by the Framingham formula. Regression analysis of QTc and heart rate Linear regression analysis of the QTc values vs. HR showed that the Hodges’ formula (QTcH) grossly over-corrected for the effect of HR (Take home figure). QTcL (Framingham) and QTcF (Fridericia) slight under-corrected for the effect of HR, and had more or less similar slopes. QTcB minimally overcorrected for the effect of HR on the QT interval but had a slope that was the closest to the horizontal (Table 3), indicating that it provided a better HR independent estimate of QTc than the other three methods studied (Take home figure). Table 3 Regression parameters for QTc interval vs. heart rate correlation for 5000 neonatal ECGs QT correction method Intercept (ms) Slope (ms/b.p.m.) 95% CI for slope P-value for slope Difference from QTcB (ms/b.p.m.) 95% CI for difference P-value for difference Bazett’s method (QTcB) 388.7 0.086 0.059–0.113 <0.0001 NA NA NA Fridericia’s method (QTcF) 391.3 −0.308 −0.336 to −0.281 <0.0001 −0.3943 −0.432 to −0.356 <0.0001 Hodges’ method (QTcH) 268.5 0.962 0.935–0.989 <0.0001 0.876 0.838–0.915 <0.0001 Framingham method (QTcL) 401.3 −0.364 −0.391 to −0.337 <0.0001 −0.450 −0.488 to −0.411 <0.0001 QT correction method Intercept (ms) Slope (ms/b.p.m.) 95% CI for slope P-value for slope Difference from QTcB (ms/b.p.m.) 95% CI for difference P-value for difference Bazett’s method (QTcB) 388.7 0.086 0.059–0.113 <0.0001 NA NA NA Fridericia’s method (QTcF) 391.3 −0.308 −0.336 to −0.281 <0.0001 −0.3943 −0.432 to −0.356 <0.0001 Hodges’ method (QTcH) 268.5 0.962 0.935–0.989 <0.0001 0.876 0.838–0.915 <0.0001 Framingham method (QTcL) 401.3 −0.364 −0.391 to −0.337 <0.0001 −0.450 −0.488 to −0.411 <0.0001 NA, not applicable. Table 3 Regression parameters for QTc interval vs. heart rate correlation for 5000 neonatal ECGs QT correction method Intercept (ms) Slope (ms/b.p.m.) 95% CI for slope P-value for slope Difference from QTcB (ms/b.p.m.) 95% CI for difference P-value for difference Bazett’s method (QTcB) 388.7 0.086 0.059–0.113 <0.0001 NA NA NA Fridericia’s method (QTcF) 391.3 −0.308 −0.336 to −0.281 <0.0001 −0.3943 −0.432 to −0.356 <0.0001 Hodges’ method (QTcH) 268.5 0.962 0.935–0.989 <0.0001 0.876 0.838–0.915 <0.0001 Framingham method (QTcL) 401.3 −0.364 −0.391 to −0.337 <0.0001 −0.450 −0.488 to −0.411 <0.0001 QT correction method Intercept (ms) Slope (ms/b.p.m.) 95% CI for slope P-value for slope Difference from QTcB (ms/b.p.m.) 95% CI for difference P-value for difference Bazett’s method (QTcB) 388.7 0.086 0.059–0.113 <0.0001 NA NA NA Fridericia’s method (QTcF) 391.3 −0.308 −0.336 to −0.281 <0.0001 −0.3943 −0.432 to −0.356 <0.0001 Hodges’ method (QTcH) 268.5 0.962 0.935–0.989 <0.0001 0.876 0.838–0.915 <0.0001 Framingham method (QTcL) 401.3 −0.364 −0.391 to −0.337 <0.0001 −0.450 −0.488 to −0.411 <0.0001 NA, not applicable. Take home figure View largeDownload slide Individual QT/QTc values plotted vs. heart rate. The linear regression lines fitting the relationship between QT/QTc and heart rate are shown, along with the regression equation. The slope of QTcB was 0.086, which was the closest to zero. Take home figure View largeDownload slide Individual QT/QTc values plotted vs. heart rate. The linear regression lines fitting the relationship between QT/QTc and heart rate are shown, along with the regression equation. The slope of QTcB was 0.086, which was the closest to zero. Calculation of neonate-specific QT correction (QTcNeo) The relationship between HR and QT values of the study dataset of 2500 neonates was evaluated by the formula: Log QT = a + b*Log RR. The slope (b) in this equation was 0.467. Based on this analysis, the population-specific QT correction formula for neonates was found to be QTcNeo = QT/RR0.467. The performance of QTcNeo and QTcB were studied in the validation dataset by plotting QTc vs. HR. The R2- and P-values of their regression lines showed that QTcNeo is statistically better than QTcB. However, considering the very small difference in the slope values of QTcNeo (−0.0132 ms/b.p.m. in the study dataset) and QTcB (+0.0742 ms/b.p.m.), the difference was not significant clinically (Figure 1) as an increase in HR by 10 b.p.m. would result in a difference of only 0.9 ms in the values of QTcNeo and QTcB. Ability to identify neonates with long QT syndrome Of the 5000 neonates evaluated in this study, genetic analysis identified 17 who were carrying a LQTS-causing mutation (LQT1 = 9, LQT2 = 6, LQT5 = 1, and LQT6 = 1). Based on this identification of neonates with ‘true positive’ LQTS, sensitivity and specificity analyses were used to generate the ROC curves and the AuROC was calculated for each of the QT correction methodology and compared. The AuROC for QTcB, QTcF, QTcL, and QTcNeo were comparable (Table 4 and Figure 2). The AuROC for QTcH was significantly lower than the AuROC for the other 4 QT correction methods (Table 4). Cut-off values that provided the best sensitivity and specificity for the correction methods were 383 ms for QTcL, 394 ms for QTcF, 446 ms for QTcNeo, and 460 ms for QTcB; all these four correction formulae had negative predictive values of 100% and positive predictive values in the range of 32–40% using the above mentioned cut-off values (Table 5). The optimum cut-off value for QTcH was 434 ms and it had a specificity of only 80%; 978 of 5000 neonates had QTcH values above the cut-off and only 1.7% of these had LQTS (Table 5). Table 4 Area under receiver operating characteristic curve for the five QT correction formulae studied in 5000 neonates for their ability to correctly classify patients with the long QT syndrome QT correction formula AuROC (95% CI) Difference vs. AuROC of QTcB P-value Bazett’s method (QTcB) 0.9984 (0.997–1) Not applicable Fridericia’s method (QTcF) 0.9985 (0.997–1) 0.0001 (−0.0003 to 0.0006) 0.64 Hodges’ method (QTcH) 0.9523 (0.922–0.982) −0.0461 (−0.0754 to −0.0169) 0.002 Framingham method (QTcL) 0.9983 (0.997–0.999) −0.0001 (−0.0008 to 0.0006) 0.75 Neonate-specific method (QTcNeo)a 0.9985 (0.997–1) 0.0001 (−0.0001 to 0.0003) 0.36 QT correction formula AuROC (95% CI) Difference vs. AuROC of QTcB P-value Bazett’s method (QTcB) 0.9984 (0.997–1) Not applicable Fridericia’s method (QTcF) 0.9985 (0.997–1) 0.0001 (−0.0003 to 0.0006) 0.64 Hodges’ method (QTcH) 0.9523 (0.922–0.982) −0.0461 (−0.0754 to −0.0169) 0.002 Framingham method (QTcL) 0.9983 (0.997–0.999) −0.0001 (−0.0008 to 0.0006) 0.75 Neonate-specific method (QTcNeo)a 0.9985 (0.997–1) 0.0001 (−0.0001 to 0.0003) 0.36 a Neonate-specific correction QTcNeo = QT/RR0.467. Table 4 Area under receiver operating characteristic curve for the five QT correction formulae studied in 5000 neonates for their ability to correctly classify patients with the long QT syndrome QT correction formula AuROC (95% CI) Difference vs. AuROC of QTcB P-value Bazett’s method (QTcB) 0.9984 (0.997–1) Not applicable Fridericia’s method (QTcF) 0.9985 (0.997–1) 0.0001 (−0.0003 to 0.0006) 0.64 Hodges’ method (QTcH) 0.9523 (0.922–0.982) −0.0461 (−0.0754 to −0.0169) 0.002 Framingham method (QTcL) 0.9983 (0.997–0.999) −0.0001 (−0.0008 to 0.0006) 0.75 Neonate-specific method (QTcNeo)a 0.9985 (0.997–1) 0.0001 (−0.0001 to 0.0003) 0.36 QT correction formula AuROC (95% CI) Difference vs. AuROC of QTcB P-value Bazett’s method (QTcB) 0.9984 (0.997–1) Not applicable Fridericia’s method (QTcF) 0.9985 (0.997–1) 0.0001 (−0.0003 to 0.0006) 0.64 Hodges’ method (QTcH) 0.9523 (0.922–0.982) −0.0461 (−0.0754 to −0.0169) 0.002 Framingham method (QTcL) 0.9983 (0.997–0.999) −0.0001 (−0.0008 to 0.0006) 0.75 Neonate-specific method (QTcNeo)a 0.9985 (0.997–1) 0.0001 (−0.0001 to 0.0003) 0.36 a Neonate-specific correction QTcNeo = QT/RR0.467. Table 5 Specificity, sensitivity and predictive values for the five QT correction formulae for their ability to identify neonates with long QT syndrome QT correction formula Cut-off value (ms) Positive predictive value (%) Negative predictive value (%) Sensitivity (%) Specificity (%) Neonates with QTc ≤ cut-off Neonates with QTc > cut-off All LQTS All LQTS n (%)a n n (%)a n (%)b QTcB 460 39.5 100 100 99.5 4957 (99.1) 0 43 (0.9) 17 (39.5) QTcF 394 39.5 100 100 99.5 4956 (99.1) 0 44 (0.9) 17 (38.6) QTcH 434 1.7 100 100 80.1 4022 (80.4) 0 978 (19.6) 17 (1.7) QTcL 383 32.1 100 100 99.3 4947 (98.9) 0 53 (1.1) 17 (32.1) QTcNeo 446 40.5 100 100 99.5 4958 (99.2) 0 42 (0.8) 17 (40.5) QT correction formula Cut-off value (ms) Positive predictive value (%) Negative predictive value (%) Sensitivity (%) Specificity (%) Neonates with QTc ≤ cut-off Neonates with QTc > cut-off All LQTS All LQTS n (%)a n n (%)a n (%)b QTcB 460 39.5 100 100 99.5 4957 (99.1) 0 43 (0.9) 17 (39.5) QTcF 394 39.5 100 100 99.5 4956 (99.1) 0 44 (0.9) 17 (38.6) QTcH 434 1.7 100 100 80.1 4022 (80.4) 0 978 (19.6) 17 (1.7) QTcL 383 32.1 100 100 99.3 4947 (98.9) 0 53 (1.1) 17 (32.1) QTcNeo 446 40.5 100 100 99.5 4958 (99.2) 0 42 (0.8) 17 (40.5) Cut-off values were defined based on best Youden’s J index value such that sensitivity and specificity were close to 100%. QTcB by Bazett’s formula; QTcF by Fridericia’s formula; QTcH by Hodges’ formula; QTcL by the Framingham formula. a Percentage of all neonates studied (n = 5000). b Neonates with LQTS as percentage of all neonates with QTc > cut-off value. Table 5 Specificity, sensitivity and predictive values for the five QT correction formulae for their ability to identify neonates with long QT syndrome QT correction formula Cut-off value (ms) Positive predictive value (%) Negative predictive value (%) Sensitivity (%) Specificity (%) Neonates with QTc ≤ cut-off Neonates with QTc > cut-off All LQTS All LQTS n (%)a n n (%)a n (%)b QTcB 460 39.5 100 100 99.5 4957 (99.1) 0 43 (0.9) 17 (39.5) QTcF 394 39.5 100 100 99.5 4956 (99.1) 0 44 (0.9) 17 (38.6) QTcH 434 1.7 100 100 80.1 4022 (80.4) 0 978 (19.6) 17 (1.7) QTcL 383 32.1 100 100 99.3 4947 (98.9) 0 53 (1.1) 17 (32.1) QTcNeo 446 40.5 100 100 99.5 4958 (99.2) 0 42 (0.8) 17 (40.5) QT correction formula Cut-off value (ms) Positive predictive value (%) Negative predictive value (%) Sensitivity (%) Specificity (%) Neonates with QTc ≤ cut-off Neonates with QTc > cut-off All LQTS All LQTS n (%)a n n (%)a n (%)b QTcB 460 39.5 100 100 99.5 4957 (99.1) 0 43 (0.9) 17 (39.5) QTcF 394 39.5 100 100 99.5 4956 (99.1) 0 44 (0.9) 17 (38.6) QTcH 434 1.7 100 100 80.1 4022 (80.4) 0 978 (19.6) 17 (1.7) QTcL 383 32.1 100 100 99.3 4947 (98.9) 0 53 (1.1) 17 (32.1) QTcNeo 446 40.5 100 100 99.5 4958 (99.2) 0 42 (0.8) 17 (40.5) Cut-off values were defined based on best Youden’s J index value such that sensitivity and specificity were close to 100%. QTcB by Bazett’s formula; QTcF by Fridericia’s formula; QTcH by Hodges’ formula; QTcL by the Framingham formula. a Percentage of all neonates studied (n = 5000). b Neonates with LQTS as percentage of all neonates with QTc > cut-off value. Figure 1 View largeDownload slide Individual QTcB and QTcNeo values vs. heart rate and their linear regression estimates for 2500 ECGs in the study dataset (upper panel) used to derive QTcNeo and the 2500 ECGs in the validation dataset (lower panel). The difference between the slopes of QTcB and QTcNeo was −0.09 ms/b.p.m. (95% confidence interval −0.15 to −0.02; P = 0.0114) for the study dataset and −0.09 ms/b.p.m. (95% confidence interval −0.15 to −0.02; P = 0.0136) for the validation dataset. Figure 1 View largeDownload slide Individual QTcB and QTcNeo values vs. heart rate and their linear regression estimates for 2500 ECGs in the study dataset (upper panel) used to derive QTcNeo and the 2500 ECGs in the validation dataset (lower panel). The difference between the slopes of QTcB and QTcNeo was −0.09 ms/b.p.m. (95% confidence interval −0.15 to −0.02; P = 0.0114) for the study dataset and −0.09 ms/b.p.m. (95% confidence interval −0.15 to −0.02; P = 0.0136) for the validation dataset. Figure 2 View largeDownload slide Receiver operating characteristic curves for the five QT correction formulae. (Left panel) Extensive overlap of QTcB, QTcF, QTcL, and QTcNeo, when plotted on a scale of 0 to 1 for sensitivity and 1-specificity is shown. (Right panel) The shaded area in the left panel is expanded such that 1-specificity (x-axis) is plotted on a scale of 0 to 0.04 for visual differentiation of the receiver operating characteristic curves for QTcB, QTcF, QTcL, and QTcNeo. Figure 2 View largeDownload slide Receiver operating characteristic curves for the five QT correction formulae. (Left panel) Extensive overlap of QTcB, QTcF, QTcL, and QTcNeo, when plotted on a scale of 0 to 1 for sensitivity and 1-specificity is shown. (Right panel) The shaded area in the left panel is expanded such that 1-specificity (x-axis) is plotted on a scale of 0 to 0.04 for visual differentiation of the receiver operating characteristic curves for QTcB, QTcF, QTcL, and QTcNeo. Discussion The main finding of this study is that, with one exception, most QT interval correction formulae provide relatively similar results in terms of correction for HR and of the ability to identify infants with LQTS. The exception is represented by the Hodges correction formula which grossly overcorrected for the effect of HR and which had extremely poor positive predictive values for identification of LQTS infants. Among the four traditional formulae tested (Bazett, Fridericia, Hodges, and Framingham), the one proposed by Bazett10 in 1920 turned out to be the best. A vexed question It is almost 100 years since Bazett described his formula to measure the QT interval.10 Since then, several publications have strongly criticized the inability of his method to properly correct for HR changes leading to a search for better correction formulae.19–21 The excessive attention to the mathematical validity of the correction for HR might be understandable if the measurement of the actual QT interval were truly accurate, but this is seldom the case in clinical practice. We actually deal with a gross measurement and the end of the T-wave is often so indistinct that even expert cardiologists reach different conclusions.22–24 Therefore, there is an ‘intrinsic measurement error’ which voids the significance of a ‘perfect correction’ and makes its search a self-defeating purpose. It is also very important to keep in mind that clinical significance is usually associated with relatively gross prolongations (at least 20–30 ms above the upper limit of normal values). What really matters, therefore, is the ability of the formula used to distinguish between normal and abnormal values. In this regard, a group quite experienced with LQTS has recognized the consistent diagnostic value of the Bazett’s correction,25 a conclusion shared by others as well.26 There is one specific condition in which there is indeed the need for extremely accurate measurements of the QT interval, and this concerns the studies with new drugs when regulatory agencies wish to exclude a possible QT prolongation of a magnitude that is likely to produce drug-induced LQTS even in apparently normal individuals.27,28 For this important objective, a careful and individualized QT correction formula such as the one proposed by Malik et al.29 or a study population-specific correction formula is certainly appropriate.30,31 The priority here is to have a HR independent QT correction formula and this issue is quite different from the objective of identifying infants likely to be affected by LQTS. Similarly, accurate QT interval corrections are also needed in physiological studies that investigate other effects on myocardial repolarization. Present data In a set of ECGs recorded in 5000 healthy neonates between the 15th and the 25th day of life, we tested two commonly used linear formulae (Hodges’ and Framingham) and two non-linear exponential formulae (Bazett’s and Fridericia’s). We also derived, and then tested, a population-specific non-linear exponential QT correction formula (QTcNeo), where the QTc value was obtained by QTcNeo = QT/RRN, where the value of N was derived from 2500 randomly selected ECGs and was found to be 0.467. To decide which of these five formulae was the best or the most useful for assessing the QT interval in neonates, we used two criteria—the mathematical ability to provide a corrected QT value that was most HR independent, and the ability to correctly identify neonates with LQTS confirmed by genetic testing. The first criterion was based on plotting the QTc by each of the five correction formulae vs. the HR values from all ECGs. The formula for which the linear regression line fitting the QTc-HR data was closest to horizontal would be considered the best, as this would imply that the corrected QT value was now independent of the HR. The Hodges correction formula grossly overcorrected for HR (slope = +0.962 ms/b.p.m.); thus, for every 10 b.p.m. increase in HR, QTcH would overcorrect the QTc by 9.62 ms. The Framingham and Fridericia methods slightly undercorrected for the effect of HR (QTcL slope = −0.364 and QTcF slope = −0.308); i.e. the Framingham formula undercorrected the QTc by 3.64 ms and Fridericia’s by 3.08 ms for every 10 b.p.m. increase in HR. QTcB performed remarkably well, with a slope of +0.086 ms/b.p.m., implying that it would overcorrect by only 0.86 ms for every 10 b.p.m. increase in HR. QTcNeo performed the best, with a slope of −0.0132 ms/b.p.m., undercorrecting by 0.132 ms for every 10 b.p.m. increase in HR. However, the difference between the slopes of QTcB and QTcNeo were clinically not significant, as the difference between them was only 0.9 ms for an increase in HR of 10 b.p.m. The inescapable conclusion is that QTcB provided the most HR independent QT correction in this cohort of 5000 neonates and would give acceptable results, given the ease of calculation and the fact that the Bazett’s formula is familiar to most cardiologists worldwide. The second criterion used was to compare the predictive value of each of these formulae to correctly identify neonates with the congenital LQTS. We found that the AuROC curve (a measure of the diagnostic accuracy of a test) was significantly less for the Hodges correction formula than the other four methods. The other four QT correction formulae had comparable areas under the ROC curve, suggesting that any one of them would be equally good in identifying neonates with the LQTS. The optimum cut-off values used to identify neonates with LQTS, however, varied widely. The cut-off values were 394 ms for QTcF, 460 ms for QTcB, 383 ms for QTcL, and 446 ms for QTcNeo. Each of these had a 100% negative predictive value but the positive predictive value was 39–40% for QTcB, QTcF, and QTcNeo and 32% for QTcL (Framingham formula). Considering the ease of calculation and the wide availability of QTcB, we concluded that QTcB best served the purpose of identifying neonates with the LQTS, although other correction formulae were also acceptable alternatives using the cut-off value that we found, except QTcH. The present findings differ from what has been reported by a number of studies on QTc correction formulae in various populations. One possible explanation lies in the fact that most previous studies were performed in adults or older children who, for the most part, had HRs between 55 and 80 b.p.m. whereas newborns have HRs around 150 b.p.m. Limitations In the original study, initially only neonates with QTcB >470 ms and their families underwent genetic screening, but this was later extended to infants with a QTc between 460 and 470 ms. It is not possible to exclude that a few neonates with QTc >440 or 450 ms may have carried LQTS-causing mutations. In our large prospective screening study,1 one ECG was recorded in each neonate. Whenever a QTc >450 ms was found, the ECG was repeated within 2 weeks to confirm the initial finding, but only the first ECG was included in this study. Thus, the QTc vs. HR data analysed here contained only a single ECG from each neonate. If multiple ECGs are recorded in the same individual at different time points and various QTc methods are compared, the method that provides the best HR independent QTc value may differ from individual to individual.29 However, a population-specific QT correction formula is the one necessary in clinical practice where usually only a single ECG is obtained per individual to screen for LQTS. Lastly, although more complex methods for modelling the QT-RR relationship have been proposed which might more accurately define the QT-RR relationship,29 we limited our analysis to the development and validation of a single log-linear QT correction formula (QTcNeo) applicable over the entire range of HRs in neonates. Our study also has some important strengths: the study included a large number of neonates prospectively studied for LQTS, the diagnosis of LQTS was confirmed by genetic testing, and all ECGs were high-resolution digital ECGs that were interpreted in a central laboratory. Implications The main implication of our study is that the Bazett’s correction can be used with confidence when screening infants for the possible presence of LQTS. We also found that QTcB provided the most HR independent QT correction and could, therefore, also be used to compare QTc intervals in serial ECGs recorded in neonates at different HRs. In the context of the congenital LQTS, it is worthwhile to note that even in adults with LQTS, most clinical and research data come from QTcB values.32–34 Thus, in a clinical setting where what matters most is the ability to identify grossly abnormal values associated with increased risk for life-threatening arrhythmias, our study, taken in conjunction with the guidelines for diagnosis of LQTS in adults, children, and neonates,34 suggests that the Bazett’s correction can be used with confidence to make recommendations for diagnostic or therapeutic decisions.3,14,35 The previously identified cut-off value of 460 ms provides the best discriminative value for QTcB in neonates.34,36 The Fridericia’s or Framingham correction methods could also be employed, albeit using different cut-off values. Footnotes See page 2896 for the editorial comment on this article (doi: 10.1093/eurheartj/ehy386) Acknowledgements The authors are grateful to Pinuccia De Tomasi for expert editorial support and to Pramod Kadam for statistical support. Funding This work was supported by the Italian Ministry of Health and Regione Lombardia Ricerca Finalizzata 2001 grant ‘Studio sulla prevalenza, il significato clinico e l’evoluzione delle anomalie ECG neonatali associate ad aritmie nell’infanzia’. Conflict of interest: G.K.P. and S.K. are employees of IQVIA. D.R.K. and Y.Y.L. are consultants to IQVIA’s Cardiac Safety Services. IQVIA is a contract research organization that provides scientific and technical services for clinical trials conducted by pharmaceutical companies involved in new drug development. Other than this, the authors declare no professional, academic, competitive, or financial conflicts of interest. References 1 Schwartz PJ , Stramba-Badiale M , Crotti L , Pedrazzini M , Besana A , Bosi G , Gabbarini F , Goulene K , Insolia R , Mannarino S , Mosca F , Nespoli L , Rimini A , Rosati E , Salice P , Spazzolini C. Prevalence of the congenital long-QT syndrome . Circulation 2009 ; 120 : 1761 – 1767 . Google Scholar CrossRef Search ADS PubMed 2 Schwartz PJ , Periti M , Malliani A. The long Q-T syndrome . Am Heart J 1975 ; 89 : 378 – 390 . Google Scholar CrossRef Search ADS PubMed 3 Schwartz PJ , Ackerman MJ. The long QT syndrome: a transatlantic clinical approach to diagnosis and therapy . Eur Heart J 2013 ; 34 : 3109 – 3116 . Google Scholar CrossRef Search ADS PubMed 4 Arnestad M , Crotti L , Rognum TO , Insolia R , Pedrazzini M , Ferrandi C , Vege A , Wang DW , Rhodes TE , George AL Jr , Schwartz PJ. Prevalence of long QT syndrome gene variants in sudden infant death syndrome . Circulation 2007 ; 115 : 361 – 367 . Google Scholar CrossRef Search ADS PubMed 5 Saul JP , Schwartz PJ , Ackerman MJ , Triedman JK. Rationale and objectives for ECG screening in infancy . Heart Rhythm 2014 ; 11 : 2316 – 2321 . Google Scholar CrossRef Search ADS PubMed 6 Schwartz PJ , Stramba-Badiale M , Segantini A , Austoni P , Bosi G , Giorgetti R , Grancini F , Marni ED , Perticone F , Rosti D , Salice P. Prolongation of the QT interval and the sudden infant death syndrome . N Engl J Med 1998 ; 338 : 1709 – 1714 . Google Scholar CrossRef Search ADS PubMed 7 Chhina S , Peverini RL , Deming DD , Hopper AO , Hashmi A , Vyhmeister NR. QTc interval in infants receiving cisapride . J Perinatol 2002 ; 22 : 144 – 148 . Google Scholar CrossRef Search ADS PubMed 8 Lupoglazoff JM , Denjoy I , Villain E , Fressart V , Simon F , Bozio A , Berthet M , Benammar N , Hainque B , Guicheney P. Long QT syndrome in neonates: conduction disorders associated with HERG mutations and sinus bradycardia with KCNQ1 mutations . J Am Coll Cardiol 2004 ; 43 : 826 – 830 . Google Scholar CrossRef Search ADS PubMed 9 Gerosa M , Cimaz R , Stramba-Badiale M , Goulene K , Meregalli E , Trespidi L , Acaia B , Cattaneo R , Tincani A , Motta M , Doria A , Zulian F , Milanesi O , Brucato A , Riboldi P , Meroni PL. Electrocardiographic abnormalities in infants born from mothers with autoimmune diseases–a multicentre prospective study . Rheumatology 2007 ; 46 : 1285 – 1289 . Google Scholar CrossRef Search ADS PubMed 10 Bazett HC. An analysis of the time-relations of electrocardiograms . Heart 1997 ; 2 : 177 – 170 . 11 Luo S , Michler K , Johnston P , Macfarlane PW. A comparison of commonly used QT correction formulae: the effect of heart rate on the QTc of normal ECGs . J Electrocardiol 2004 ; 37 : 81 – 90 . Google Scholar CrossRef Search ADS PubMed 12 Chiladakis J , Kalogeropoulos A , Arvanitis P , Koutsogiannis N , Zagli F , Alexopoulos D. Heart rate-dependence of QTc intervals assessed by different correction methods in patients with normal or prolonged repolarization . Pacing Clin Electrophysiol 2010 ; 33 : 553 – 560 . Google Scholar CrossRef Search ADS PubMed 13 Iglesias-Álvarez D , Rodríguez-Mañero M , García-Seara FJ , Kreidieh O , Martínez-Sande JL , Álvarez-Álvarez B , Fernández-López XA , González-Melchor L , Lage-Fernández R , Moscoso-Galán I , González-Juanatey JR. Comparison and validation of recommended QT interval correction formulae for predicting cardiac arrhythmias in patients with advanced heart Failure and cardiac resynchronization devices . Am J Cardiol 2017 ; 120 : 959 – 965 . Google Scholar CrossRef Search ADS PubMed 14 Schwartz PJ , Garson A Jr , Paul T , Stramba BM , Vetter VL , Wren C. Guidelines for the interpretation of the neonatal electrocardiogram. A task force of the European Society of Cardiology . Eur Heart J 2002 ; 23 : 1329 – 1344 . Google Scholar CrossRef Search ADS PubMed 15 Fridericia LS. The duration of systole in an electrocardiogram in normal humans and in patients with heart disease . Acta Med Scand 2009 ; 53 : 469 – 486 . Google Scholar CrossRef Search ADS 16 Hodges M , Salerno D , Erlien D. Bazett's QT correction reviewed: evidence that a linear QT correction for heart rate is better . J Am Coll Cardiol 1983 ; 1 : 694. 17 Sagie A , Larson MG , Goldberg RJ , Bengtson JR , Levy D. An improved method for adjusting the QT interval for heart rate (the Framingham Heart Study) . Am J Cardiol 1992 ; 70 : 797 – 801 . Google Scholar CrossRef Search ADS PubMed 18 DeLong ER , DeLong DM , Clarke-Pearson DL. Comparing areas under two or more correlated receiver operating characteristics curves: a nonparametric approach . Biometrics 1988 ; 44 : 837 – 845 . Google Scholar CrossRef Search ADS PubMed 19 Schwartz PJ , Wolf S. QT interval prolongation as predictor of sudden death in patients with myocardial infarction . Circulation 1978 ; 57 : 1074 – 1077 . Google Scholar CrossRef Search ADS PubMed 20 Algra A , Tijssen JG , Roelandt JR , Pool J , Lubsen J. QTc prolongation measured by standard 12-lead electrocardiography is an independent risk factor for sudden death due to cardiac arrest . Circulation 1991 ; 83 : 1888 – 1894 . Google Scholar CrossRef Search ADS PubMed 21 Aro AL , Reinier K , Rusinaru C , Uy-Evanado A , Darouian N , Phan D , Mack WJ , Jui J , Soliman EZ , Tereshchenko LG , Chugh SS. Electrical risk score beyond the left ventricular ejection fraction: prediction of sudden cardiac death in the Oregon Sudden Unexpected Death Study and the Atherosclerosis Risk in Communities Study . Eur Heart J 2017 ; 38 : 3017 – 3025 . Google Scholar CrossRef Search ADS PubMed 22 Viskin S , Rosovski U , Sands AJ , Chen E , Kistler PM , Kalman JM , Rodriguez Chavez L , Iturralde Torres P , Cruz F FE , Centurión OA , Fujiki A , Maury P , Chen X , Krahn AD , Roithinger F , Zhang L , Vincent GM , Zeltser D. Inaccurate electrocardiographic interpretation of long QT: the majority of physicians cannot recognize a long QT when they see one . Heart Rhythm 2005 ; 2 : 569 – 574 . Google Scholar CrossRef Search ADS PubMed 23 Panicker GK , Karnad DR , Natekar M , Kothari S , Narula D , Lokhandwala Y. Intra-and inter-reader variability in QT interval measurement by tangent and threshold methods in a central electrocardiogram laboratory . J Electrocardiol 2009 ; 42 : 348 – 352 . Google Scholar CrossRef Search ADS PubMed 24 Panicker GK , Karnad DR , Joshi R , Shetty S , Vyas N , Kothari S , Narula D. Z-score for benchmarking reader competence in a central ECG laboratory . Ann Noninvasive Electrocardiol 2009 ; 14 : 19 – 25 . Google Scholar CrossRef Search ADS PubMed 25 Brouwer J , Van Den Berg MP , Grobbee DE , Haaksma J , Wilde AA. Diagnostic performance of various QTc interval formulas in a large family with long QT syndrome type 3: Bazett's formula not so bad after all… . Ann Noninvasive Electrocardiol 2003 ; 8 : 269 – 274 . Google Scholar CrossRef Search ADS PubMed 26 Phan DQ , Silka MJ , Lan YT , Chang RK. Comparison of formulas for calculation of the corrected QT interval in infants and young children . J Pediatr 2015 ; 166 : 960 – 964 . Google Scholar CrossRef Search ADS PubMed 27 Roden DM. Drug-induced prolongation of the QT Interval . N Engl J Med 2004 ; 350 : 1013 – 1022 . Google Scholar CrossRef Search ADS PubMed 28 Schwartz PJ , Woosley RL. Predicting the unpredictable: drug-induced QT prolongation and Torsades de Pointes . J Am Coll Cardiol 2016 ; 67 : 1639 – 1650 . Google Scholar CrossRef Search ADS PubMed 29 Malik M , Färbom P , Batchvarov V , Hnatkova K , Camm AJ. Relation between QT and RR intervals is highly individual among healthy subjects: implications for heart rate correction of the QT interval . Heart 2002 ; 87 : 220 – 228 . Google Scholar CrossRef Search ADS PubMed 30 US Food and Drug Administration . Guidance for Industry: E14 Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs. Food and Drug Administration; 2005 . http://www.fda.gov/downloads/drugs/guidancecomplianceregulatoryinformation/guidances/ucm073153.pdf (14 May 2018). 31 Salvi V , Karnad DR , Panicker GK , Kothari S. Update on the evaluation of a new drug for effects on cardiac repolarization in humans: issues in early drug development . Br J Pharmacol 2010 ; 159 : 34 – 48 . Google Scholar CrossRef Search ADS PubMed 32 Sauer AJ , Moss AJ , McNitt S , Peterson DR , Zareba W , Robinson JL , Qi M , Goldenberg I , Hobbs JB , Ackerman MJ , Benhorin J , Hall WJ , Kaufman ES , Locati EH , Napolitano C , Priori SG , Schwartz PJ , Towbin JA , Vincent GM , Zhang L. Long QT syndrome in adults . J Am Coll Cardiol 2007 ; 49 : 329 – 337 . Google Scholar CrossRef Search ADS PubMed 33 Schwartz PJ , Priori SG , Spazzolini C , Moss AJ , Vincent GM , Napolitano C , Denjoy I , Guicheney P , Breithardt G , Keating MT , Towbin JA , Beggs AH , Brink P , Wilde AAM , Toivonen L , Zareba W , Robinson JL , Timothy KW , Corfield V , Wattanasirichaigoon D , Corbett C , Haverkamp W , Schulze-Bahr E , Lehmann MH , Schwartz K , Coumel P , Bloise R. Genotype-phenotype correlation in the long QT syndrome. Gene-specific triggers for life-threatening arrhythmias . Circulation 2001 ; 103 : 89 – 95 . Google Scholar CrossRef Search ADS PubMed 34 Priori SG , Wilde AA , Horie M , Cho Y , Behr ER , Berul C , Blom N , Brugada J , Chiang CE , Huikuri H , Kannankeril P , Krahn A , Leenhardt A , Moss A , Schwartz PJ , Shimizu W , Tomaselli G , Tracy C. HRS/EHRA/APHRS Expert consensus statement on the diagnosis and management of patients with inherited primary arrhythmia syndromes . Heart Rhythm 2013 ; 10 : 1932 – 1963 . Google Scholar CrossRef Search ADS PubMed 35 Goldenberg I , Moss AJ , Peterson DR , McNitt S , Zareba W , Hong J , Andrews ML , Robinson JL , Locati EH , Ackerman MJ , Benhorin J , Kaufman ES , Napolitano C , Priori SG , Qi M , Schwartz PJ , Towbin JA , Vincent GM , Zhang L. Risk factors for aborted cardiac arrest and sudden cardiac death in children with the congenital Long-QT Syndrome . Circulation 2008 ; 117 : 2184 – 2191 . Google Scholar CrossRef Search ADS PubMed 36 Merri M , Benhorin J , Alberti M , Locati E , Moss AJ. Electrocardiographic quantitation of ventricular repolarization . Circulation 1989 ; 80 : 1301 – 1308 . Google Scholar CrossRef Search ADS PubMed Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Heart Journal Oxford University Press

For neonatal ECG screening there is no reason to relinquish old Bazett’s correction

European Heart Journal , Volume 39 (31) – Aug 14, 2018

Loading next page...
 
/lp/ou_press/for-neonatal-ecg-screening-there-is-no-reason-to-relinquish-old-bazett-VWW8e0w0xm
Publisher
Oxford University Press
Copyright
Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com.
ISSN
0195-668X
eISSN
1522-9645
D.O.I.
10.1093/eurheartj/ehy284
Publisher site
See Article on Publisher Site

Abstract

Abstract Aims There is an almost endless controversy regarding the choice of the QT correction formula to be used in electrocardiograms (ECG) in neonates for screening for long QT syndrome (LQTS). We compared the performance of four commonly used formulae and a new formula derived from neonates. Methods and results From a cohort of 44 596 healthy neonates prospectively studied in Italy between 2001 and 2006, 5000 ECGs including 17 with LQTS-causing mutation identified by genotyping were studied using four QT correction formulae [Bazett’s (QTcB), Fridericia’s (QTcF), Framingham (QTcL), and Hodges (QTcH)]. A neonate-specific exponential correction (QTcNeo) was derived using 2500 randomly selected ECGs and validated for accuracy in the remaining 2500 ECGs. Digital ECGs were recorded between the 15th and 25th day of life; QT interval was measured manually in leads II, V5, and V6. To assess the ability to provide heart rate (HR) independent QT correction, regression analysis of the QTc-HR plots for all 5000 ECGs with each correction formula was done. QTcB provided the most HR independent correction with a slope closest to zero (slope +0.086 ms/b.p.m.) followed by QTcF (slope −0.308 ms/b.p.m.), QTcL (slope −0.364 ms/b.p.m.), and QTcH (slope +0.962 ms/b.p.m.). The QTc-HR slope of QTcNeo (QT/RR0.467) was similar to QTcB. The ability to correctly identify neonates with LQTS was best with QTcB, QTcF, and QTcNeo (comparable areas under the receiver operating characteristic curves) with positive predictive value of 39–40% and sensitivity of 100%. Cut-off values were 460 ms for QTcB, 394 ms for QTcF, and 446 ms for QTcNeo. Conclusions The Bazett’s correction provides an effective HR independent QT correction and also accurately identifies the neonates affected by LQTS. It can be used with confidence in neonates, although other methods could also be used with appropriate cut-offs. View largeDownload slide View largeDownload slide Bazett’s formula, Long QT syndrome, Neonatal ECG, QT interval correction Introduction The long QT syndrome (LQTS) is a genetic disease with a prevalence of 1 in 2000 live births1 and is associated with a high risk for life-threatening arrhythmias among untreated patients.2,3 However, with appropriate treatment, mortality has now been reduced to below 1%.3 Sudden death is often the sentinel event, and cardiac events usually appear in childhood or adolescence. Long QT syndrome also contributes to almost 10% of cases of sudden infant death syndrome4 and early identification of LQTS through electrocardiogram (ECG) screening performed in the first month of life is likely to reduce mortality.5 The usefulness of neonatal ECG screening for the identification of LQTS has been demonstrated by two separate prospective studies involving almost 78 000 infants.1,6 Even though the number of clinical centres worldwide offering a service of neonatal electrocardiography is constantly growing, the question of which is the best method for the heart rate (HR) correction of the QT interval in neonates, given their high HRs remains unanswered. The method used in most studies in newborns1,6,7–9 is the one derived from the article by Bazett10 in 1920 (corrected QT or QTcB = QT/ RR). However, the endless criticism that this formula overcorrects the QT interval at high HRs has led many investigators to propose alternative formulae.11–13 Most of these publications suffer from several limitations: (i) many small cohorts, (ii) large cohorts but with automated measurements with various degrees of unreliability, and (iii) validation in children or adult populations but not in neonates. Moreover, no attempts were made to verify whether or not the other QT correction methods could reliably identify infants with an established diagnosis of LQTS which, after all, should be the main reason to measure their QT interval. This study was specifically designed to verify in a large population of infants which, among the many QT correction methods currently used or proposed, is superior and sufficiently reliable to identify those with LQTS. Methods This study was performed using ECGs recorded in healthy neonates, enrolled in a prospective study, which assessed and defined the prevalence of congenital LQTS.1 Study population The original study1 included 44 596 healthy neonates (43 080 Caucasians), 22 967 males (51%), and 21629 females (49%) consecutively enrolled by 18 Italian maternity hospitals between January 2001 and June 2006, and in whom an ECG was recorded between the 15th and the 25th day of life. All parents signed an informed consent for recording the ECG. Very premature and sick newborns who required transfer to neonatal intensive care units were excluded from the study. The protocol requested genetic analysis in all infants with a QTc >470 ms and in their parents, and recommended it in those with a QT >460 ms. All coding exons of KCNQ1, KCNH2, SCN5A, KCNE1, KCNE2, CAV3, and SCN4B (the genes screened at that time in our laboratory for the routine diagnosis of LQTS) were analysed for presence of disease-causing mutations. The details of the genetic analysis have been described.1 There were 59 neonates with a QTc >460 ms and 43 of them (90% of those above 470 ms and 50% of those above 460 ms) underwent genetic analysis. For this study, we analysed 5000 ECGs, which included all 43 ECGs with QTc >460 ms of the infants who underwent genotyping, and another 4957 ECGs all recorded at a single centre to reduce variability (this centre enrolled 5007 infants but we randomly excluded 50 of them to end up with a cohort of 5000). All these tracings were part of the 44 553 ECGs from the original dataset. These 5000 neonates included 2562 (51.2%) males and 2438 (48.8%) females. ECG recording and analysis The ECGs were recorded at a paper speed of 25 mm/s with a Marquette MAC 5000 recorder (digital sampling rate 500 Hz) and were transmitted from the participating centre via modem to the study co-ordinating centre and were read centrally. The QT intervals and the preceding RR intervals were measured on-screen manually in leads II, V5, and V6 from five non-consecutive beats. During ECG analysis, the QTc was calculated according to Bazett’s formula (QTcB) and the longest mean value of the five beats found in one of the three leads was considered. In neonates with a QTcB >450 ms, another ECG was recorded within 1 to 2 weeks to confirm the initial finding. ECGs were analysed as per the European Society of Cardiology guidelines for the interpretation of neonatal ECG and if QT prolongation or any other ECG abnormality was identified, the infants were managed and treated accordingly.14 Bradycardia was defined as HR <107 b.p.m. and tachycardia as HR >182 b.p.m.14 The QT and RR interval values from each ECG were used to calculate the corrected QT (QTc) interval using four commonly used HR correction formulae: Bazett10 QTcB = QT/RR Fridericia15 QTcF = QT/RR3 Hodges16 QTcH = QT + 1.75 × (Heart rate - 60) Framingham17 QTcL = QT + 0.154 × (1 - RR) Bazett10 QTcB = QT/RR Fridericia15 QTcF = QT/RR3 Hodges16 QTcH = QT + 1.75 × (Heart rate - 60) Framingham17 QTcL = QT + 0.154 × (1 - RR) Bazett10 QTcB = QT/RR Fridericia15 QTcF = QT/RR3 Hodges16 QTcH = QT + 1.75 × (Heart rate - 60) Framingham17 QTcL = QT + 0.154 × (1 - RR) Bazett10 QTcB = QT/RR Fridericia15 QTcF = QT/RR3 Hodges16 QTcH = QT + 1.75 × (Heart rate - 60) Framingham17 QTcL = QT + 0.154 × (1 - RR) Statistical analysis All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA). Differences in QTc and RR intervals between males and females were compared with the Student’s t-test. To identify which of the QT correction formulae was most appropriate for use in neonates, we plotted the QTc against HR values from each ECG and the slope with its 95% confidence interval (CI) of QTc-HR regression line was determined for each formula by linear mixed effects model using the PROC MIXED function in SAS. The QT correction formula with the slope of the regression line closest to zero was considered the best as it would provide a QTc value that was most independent of HR. Comparison of slopes between correction formulae was also performed by including the ‘correction method × concentration’ interaction term in the linear mixed effects model and the mean difference with 95% CI was reported. To determine if a neonate-specific QT correction (QTcNeo) formula was better than the other formulae studied, the dataset of 5000 ECGs was randomly split into two equal halves (a study dataset and a validation dataset of 2500 ECGs each), using a random number generator. For calculating the correction factor, the relationship between the HR and QT values in the study dataset was evaluated by the formula: Log QT = a + b*Log RR. Having obtained the value of b, this was then used to calculate QTcNeo using the formula QTcNeo = QT/RRb. The performance of the QTcNeo formula was evaluated using the remaining 2500 ECGs by comparing the slope of its QTc-HR regression line with the slope of the QTc-HR regression line of the best of the four other QT correction formulae studied. To assess the diagnostic performance of each correction method in identifying neonates with or without a LQTS disease-causing mutation, receiver operating characteristic (ROC) curves were generated. The area under the ROC (AuROC) was calculated with its 95% CI using the PROC LOGISTIC function in the SAS statistical software package. A χ2 test of significance was used to compare the areas under correlated ROC curves for different QTc methods,18 and the one with the largest AuROC was considered the best correction method for this purpose. The Youden’s J statistic was used to determine the cut-off values for each QT correction formula that best separated those with LQTS from those who did not. A two-sided P-value <0.05 was considered statistically significant. Results The mean (± standard deviation) HR for the 5000 ECGs was 154 ± 17 b.p.m. and mean QT interval was 252 ± 19 ms (Table 1). The QTc values by the four QT correction formulae (Bazett’s, Fridericia’s, Hodges’, and Framingham) did not differ significantly between males and females (Table 2). Of the 2438 females, five had bradycardia and 99 had tachycardia. Similarly, 3 of the 2562 males had bradycardia and 122 had tachycardia. Table 1 Mean (± standard deviation) and range of hazard ratio, QT interval, and corrected QT interval by four previously published correction formulae for 5000 neonates ECG parameters QT correction formula Mean ± SD Minimum Maximum Heart rate (b.p.m.) 154 ± 17 94 214 QT (ms) 252 ± 19 200 400 QTcB (ms) Bazett 402 ± 21 341 555 QTcF (ms) Fridericia 344 ± 19 286 497 QTcH (ms) Hodges 417 ± 21 340 497 QTcL (ms) Framingham 345 ± 15 302 474 ECG parameters QT correction formula Mean ± SD Minimum Maximum Heart rate (b.p.m.) 154 ± 17 94 214 QT (ms) 252 ± 19 200 400 QTcB (ms) Bazett 402 ± 21 341 555 QTcF (ms) Fridericia 344 ± 19 286 497 QTcH (ms) Hodges 417 ± 21 340 497 QTcL (ms) Framingham 345 ± 15 302 474 SD, standard deviation. Note: QTcB by Bazett's formula; QTcF by Fridericia's formula; QTcH by Hodges' formula; QTcL by the Framingham formula. Table 1 Mean (± standard deviation) and range of hazard ratio, QT interval, and corrected QT interval by four previously published correction formulae for 5000 neonates ECG parameters QT correction formula Mean ± SD Minimum Maximum Heart rate (b.p.m.) 154 ± 17 94 214 QT (ms) 252 ± 19 200 400 QTcB (ms) Bazett 402 ± 21 341 555 QTcF (ms) Fridericia 344 ± 19 286 497 QTcH (ms) Hodges 417 ± 21 340 497 QTcL (ms) Framingham 345 ± 15 302 474 ECG parameters QT correction formula Mean ± SD Minimum Maximum Heart rate (b.p.m.) 154 ± 17 94 214 QT (ms) 252 ± 19 200 400 QTcB (ms) Bazett 402 ± 21 341 555 QTcF (ms) Fridericia 344 ± 19 286 497 QTcH (ms) Hodges 417 ± 21 340 497 QTcL (ms) Framingham 345 ± 15 302 474 SD, standard deviation. Note: QTcB by Bazett's formula; QTcF by Fridericia's formula; QTcH by Hodges' formula; QTcL by the Framingham formula. Table 2 Heart rate, QT interval, and corrected QT interval (mean ± standard deviation) by four correction formulae in male and female neonates Females (n = 2438) Males (n = 2562) Heart rate (b.p.m) 154 ± 17 155 ± 17 QT (ms) 253 ± 19 251 ± 18 QTcB (ms) 403 ± 21 401 ± 21 QTcF (ms) 345 ± 19 343 ± 18 QTcH (ms) 417 ± 21 417 ± 21 QTcL (ms) 346 ± 15 344 ± 14 Females (n = 2438) Males (n = 2562) Heart rate (b.p.m) 154 ± 17 155 ± 17 QT (ms) 253 ± 19 251 ± 18 QTcB (ms) 403 ± 21 401 ± 21 QTcF (ms) 345 ± 19 343 ± 18 QTcH (ms) 417 ± 21 417 ± 21 QTcL (ms) 346 ± 15 344 ± 14 Note: QTcB by Bazett’s formula; QTcF by Fridericia’s formula; QTcH by Hodges’ formula; QTcL by the Framingham formula. Table 2 Heart rate, QT interval, and corrected QT interval (mean ± standard deviation) by four correction formulae in male and female neonates Females (n = 2438) Males (n = 2562) Heart rate (b.p.m) 154 ± 17 155 ± 17 QT (ms) 253 ± 19 251 ± 18 QTcB (ms) 403 ± 21 401 ± 21 QTcF (ms) 345 ± 19 343 ± 18 QTcH (ms) 417 ± 21 417 ± 21 QTcL (ms) 346 ± 15 344 ± 14 Females (n = 2438) Males (n = 2562) Heart rate (b.p.m) 154 ± 17 155 ± 17 QT (ms) 253 ± 19 251 ± 18 QTcB (ms) 403 ± 21 401 ± 21 QTcF (ms) 345 ± 19 343 ± 18 QTcH (ms) 417 ± 21 417 ± 21 QTcL (ms) 346 ± 15 344 ± 14 Note: QTcB by Bazett’s formula; QTcF by Fridericia’s formula; QTcH by Hodges’ formula; QTcL by the Framingham formula. Regression analysis of QTc and heart rate Linear regression analysis of the QTc values vs. HR showed that the Hodges’ formula (QTcH) grossly over-corrected for the effect of HR (Take home figure). QTcL (Framingham) and QTcF (Fridericia) slight under-corrected for the effect of HR, and had more or less similar slopes. QTcB minimally overcorrected for the effect of HR on the QT interval but had a slope that was the closest to the horizontal (Table 3), indicating that it provided a better HR independent estimate of QTc than the other three methods studied (Take home figure). Table 3 Regression parameters for QTc interval vs. heart rate correlation for 5000 neonatal ECGs QT correction method Intercept (ms) Slope (ms/b.p.m.) 95% CI for slope P-value for slope Difference from QTcB (ms/b.p.m.) 95% CI for difference P-value for difference Bazett’s method (QTcB) 388.7 0.086 0.059–0.113 <0.0001 NA NA NA Fridericia’s method (QTcF) 391.3 −0.308 −0.336 to −0.281 <0.0001 −0.3943 −0.432 to −0.356 <0.0001 Hodges’ method (QTcH) 268.5 0.962 0.935–0.989 <0.0001 0.876 0.838–0.915 <0.0001 Framingham method (QTcL) 401.3 −0.364 −0.391 to −0.337 <0.0001 −0.450 −0.488 to −0.411 <0.0001 QT correction method Intercept (ms) Slope (ms/b.p.m.) 95% CI for slope P-value for slope Difference from QTcB (ms/b.p.m.) 95% CI for difference P-value for difference Bazett’s method (QTcB) 388.7 0.086 0.059–0.113 <0.0001 NA NA NA Fridericia’s method (QTcF) 391.3 −0.308 −0.336 to −0.281 <0.0001 −0.3943 −0.432 to −0.356 <0.0001 Hodges’ method (QTcH) 268.5 0.962 0.935–0.989 <0.0001 0.876 0.838–0.915 <0.0001 Framingham method (QTcL) 401.3 −0.364 −0.391 to −0.337 <0.0001 −0.450 −0.488 to −0.411 <0.0001 NA, not applicable. Table 3 Regression parameters for QTc interval vs. heart rate correlation for 5000 neonatal ECGs QT correction method Intercept (ms) Slope (ms/b.p.m.) 95% CI for slope P-value for slope Difference from QTcB (ms/b.p.m.) 95% CI for difference P-value for difference Bazett’s method (QTcB) 388.7 0.086 0.059–0.113 <0.0001 NA NA NA Fridericia’s method (QTcF) 391.3 −0.308 −0.336 to −0.281 <0.0001 −0.3943 −0.432 to −0.356 <0.0001 Hodges’ method (QTcH) 268.5 0.962 0.935–0.989 <0.0001 0.876 0.838–0.915 <0.0001 Framingham method (QTcL) 401.3 −0.364 −0.391 to −0.337 <0.0001 −0.450 −0.488 to −0.411 <0.0001 QT correction method Intercept (ms) Slope (ms/b.p.m.) 95% CI for slope P-value for slope Difference from QTcB (ms/b.p.m.) 95% CI for difference P-value for difference Bazett’s method (QTcB) 388.7 0.086 0.059–0.113 <0.0001 NA NA NA Fridericia’s method (QTcF) 391.3 −0.308 −0.336 to −0.281 <0.0001 −0.3943 −0.432 to −0.356 <0.0001 Hodges’ method (QTcH) 268.5 0.962 0.935–0.989 <0.0001 0.876 0.838–0.915 <0.0001 Framingham method (QTcL) 401.3 −0.364 −0.391 to −0.337 <0.0001 −0.450 −0.488 to −0.411 <0.0001 NA, not applicable. Take home figure View largeDownload slide Individual QT/QTc values plotted vs. heart rate. The linear regression lines fitting the relationship between QT/QTc and heart rate are shown, along with the regression equation. The slope of QTcB was 0.086, which was the closest to zero. Take home figure View largeDownload slide Individual QT/QTc values plotted vs. heart rate. The linear regression lines fitting the relationship between QT/QTc and heart rate are shown, along with the regression equation. The slope of QTcB was 0.086, which was the closest to zero. Calculation of neonate-specific QT correction (QTcNeo) The relationship between HR and QT values of the study dataset of 2500 neonates was evaluated by the formula: Log QT = a + b*Log RR. The slope (b) in this equation was 0.467. Based on this analysis, the population-specific QT correction formula for neonates was found to be QTcNeo = QT/RR0.467. The performance of QTcNeo and QTcB were studied in the validation dataset by plotting QTc vs. HR. The R2- and P-values of their regression lines showed that QTcNeo is statistically better than QTcB. However, considering the very small difference in the slope values of QTcNeo (−0.0132 ms/b.p.m. in the study dataset) and QTcB (+0.0742 ms/b.p.m.), the difference was not significant clinically (Figure 1) as an increase in HR by 10 b.p.m. would result in a difference of only 0.9 ms in the values of QTcNeo and QTcB. Ability to identify neonates with long QT syndrome Of the 5000 neonates evaluated in this study, genetic analysis identified 17 who were carrying a LQTS-causing mutation (LQT1 = 9, LQT2 = 6, LQT5 = 1, and LQT6 = 1). Based on this identification of neonates with ‘true positive’ LQTS, sensitivity and specificity analyses were used to generate the ROC curves and the AuROC was calculated for each of the QT correction methodology and compared. The AuROC for QTcB, QTcF, QTcL, and QTcNeo were comparable (Table 4 and Figure 2). The AuROC for QTcH was significantly lower than the AuROC for the other 4 QT correction methods (Table 4). Cut-off values that provided the best sensitivity and specificity for the correction methods were 383 ms for QTcL, 394 ms for QTcF, 446 ms for QTcNeo, and 460 ms for QTcB; all these four correction formulae had negative predictive values of 100% and positive predictive values in the range of 32–40% using the above mentioned cut-off values (Table 5). The optimum cut-off value for QTcH was 434 ms and it had a specificity of only 80%; 978 of 5000 neonates had QTcH values above the cut-off and only 1.7% of these had LQTS (Table 5). Table 4 Area under receiver operating characteristic curve for the five QT correction formulae studied in 5000 neonates for their ability to correctly classify patients with the long QT syndrome QT correction formula AuROC (95% CI) Difference vs. AuROC of QTcB P-value Bazett’s method (QTcB) 0.9984 (0.997–1) Not applicable Fridericia’s method (QTcF) 0.9985 (0.997–1) 0.0001 (−0.0003 to 0.0006) 0.64 Hodges’ method (QTcH) 0.9523 (0.922–0.982) −0.0461 (−0.0754 to −0.0169) 0.002 Framingham method (QTcL) 0.9983 (0.997–0.999) −0.0001 (−0.0008 to 0.0006) 0.75 Neonate-specific method (QTcNeo)a 0.9985 (0.997–1) 0.0001 (−0.0001 to 0.0003) 0.36 QT correction formula AuROC (95% CI) Difference vs. AuROC of QTcB P-value Bazett’s method (QTcB) 0.9984 (0.997–1) Not applicable Fridericia’s method (QTcF) 0.9985 (0.997–1) 0.0001 (−0.0003 to 0.0006) 0.64 Hodges’ method (QTcH) 0.9523 (0.922–0.982) −0.0461 (−0.0754 to −0.0169) 0.002 Framingham method (QTcL) 0.9983 (0.997–0.999) −0.0001 (−0.0008 to 0.0006) 0.75 Neonate-specific method (QTcNeo)a 0.9985 (0.997–1) 0.0001 (−0.0001 to 0.0003) 0.36 a Neonate-specific correction QTcNeo = QT/RR0.467. Table 4 Area under receiver operating characteristic curve for the five QT correction formulae studied in 5000 neonates for their ability to correctly classify patients with the long QT syndrome QT correction formula AuROC (95% CI) Difference vs. AuROC of QTcB P-value Bazett’s method (QTcB) 0.9984 (0.997–1) Not applicable Fridericia’s method (QTcF) 0.9985 (0.997–1) 0.0001 (−0.0003 to 0.0006) 0.64 Hodges’ method (QTcH) 0.9523 (0.922–0.982) −0.0461 (−0.0754 to −0.0169) 0.002 Framingham method (QTcL) 0.9983 (0.997–0.999) −0.0001 (−0.0008 to 0.0006) 0.75 Neonate-specific method (QTcNeo)a 0.9985 (0.997–1) 0.0001 (−0.0001 to 0.0003) 0.36 QT correction formula AuROC (95% CI) Difference vs. AuROC of QTcB P-value Bazett’s method (QTcB) 0.9984 (0.997–1) Not applicable Fridericia’s method (QTcF) 0.9985 (0.997–1) 0.0001 (−0.0003 to 0.0006) 0.64 Hodges’ method (QTcH) 0.9523 (0.922–0.982) −0.0461 (−0.0754 to −0.0169) 0.002 Framingham method (QTcL) 0.9983 (0.997–0.999) −0.0001 (−0.0008 to 0.0006) 0.75 Neonate-specific method (QTcNeo)a 0.9985 (0.997–1) 0.0001 (−0.0001 to 0.0003) 0.36 a Neonate-specific correction QTcNeo = QT/RR0.467. Table 5 Specificity, sensitivity and predictive values for the five QT correction formulae for their ability to identify neonates with long QT syndrome QT correction formula Cut-off value (ms) Positive predictive value (%) Negative predictive value (%) Sensitivity (%) Specificity (%) Neonates with QTc ≤ cut-off Neonates with QTc > cut-off All LQTS All LQTS n (%)a n n (%)a n (%)b QTcB 460 39.5 100 100 99.5 4957 (99.1) 0 43 (0.9) 17 (39.5) QTcF 394 39.5 100 100 99.5 4956 (99.1) 0 44 (0.9) 17 (38.6) QTcH 434 1.7 100 100 80.1 4022 (80.4) 0 978 (19.6) 17 (1.7) QTcL 383 32.1 100 100 99.3 4947 (98.9) 0 53 (1.1) 17 (32.1) QTcNeo 446 40.5 100 100 99.5 4958 (99.2) 0 42 (0.8) 17 (40.5) QT correction formula Cut-off value (ms) Positive predictive value (%) Negative predictive value (%) Sensitivity (%) Specificity (%) Neonates with QTc ≤ cut-off Neonates with QTc > cut-off All LQTS All LQTS n (%)a n n (%)a n (%)b QTcB 460 39.5 100 100 99.5 4957 (99.1) 0 43 (0.9) 17 (39.5) QTcF 394 39.5 100 100 99.5 4956 (99.1) 0 44 (0.9) 17 (38.6) QTcH 434 1.7 100 100 80.1 4022 (80.4) 0 978 (19.6) 17 (1.7) QTcL 383 32.1 100 100 99.3 4947 (98.9) 0 53 (1.1) 17 (32.1) QTcNeo 446 40.5 100 100 99.5 4958 (99.2) 0 42 (0.8) 17 (40.5) Cut-off values were defined based on best Youden’s J index value such that sensitivity and specificity were close to 100%. QTcB by Bazett’s formula; QTcF by Fridericia’s formula; QTcH by Hodges’ formula; QTcL by the Framingham formula. a Percentage of all neonates studied (n = 5000). b Neonates with LQTS as percentage of all neonates with QTc > cut-off value. Table 5 Specificity, sensitivity and predictive values for the five QT correction formulae for their ability to identify neonates with long QT syndrome QT correction formula Cut-off value (ms) Positive predictive value (%) Negative predictive value (%) Sensitivity (%) Specificity (%) Neonates with QTc ≤ cut-off Neonates with QTc > cut-off All LQTS All LQTS n (%)a n n (%)a n (%)b QTcB 460 39.5 100 100 99.5 4957 (99.1) 0 43 (0.9) 17 (39.5) QTcF 394 39.5 100 100 99.5 4956 (99.1) 0 44 (0.9) 17 (38.6) QTcH 434 1.7 100 100 80.1 4022 (80.4) 0 978 (19.6) 17 (1.7) QTcL 383 32.1 100 100 99.3 4947 (98.9) 0 53 (1.1) 17 (32.1) QTcNeo 446 40.5 100 100 99.5 4958 (99.2) 0 42 (0.8) 17 (40.5) QT correction formula Cut-off value (ms) Positive predictive value (%) Negative predictive value (%) Sensitivity (%) Specificity (%) Neonates with QTc ≤ cut-off Neonates with QTc > cut-off All LQTS All LQTS n (%)a n n (%)a n (%)b QTcB 460 39.5 100 100 99.5 4957 (99.1) 0 43 (0.9) 17 (39.5) QTcF 394 39.5 100 100 99.5 4956 (99.1) 0 44 (0.9) 17 (38.6) QTcH 434 1.7 100 100 80.1 4022 (80.4) 0 978 (19.6) 17 (1.7) QTcL 383 32.1 100 100 99.3 4947 (98.9) 0 53 (1.1) 17 (32.1) QTcNeo 446 40.5 100 100 99.5 4958 (99.2) 0 42 (0.8) 17 (40.5) Cut-off values were defined based on best Youden’s J index value such that sensitivity and specificity were close to 100%. QTcB by Bazett’s formula; QTcF by Fridericia’s formula; QTcH by Hodges’ formula; QTcL by the Framingham formula. a Percentage of all neonates studied (n = 5000). b Neonates with LQTS as percentage of all neonates with QTc > cut-off value. Figure 1 View largeDownload slide Individual QTcB and QTcNeo values vs. heart rate and their linear regression estimates for 2500 ECGs in the study dataset (upper panel) used to derive QTcNeo and the 2500 ECGs in the validation dataset (lower panel). The difference between the slopes of QTcB and QTcNeo was −0.09 ms/b.p.m. (95% confidence interval −0.15 to −0.02; P = 0.0114) for the study dataset and −0.09 ms/b.p.m. (95% confidence interval −0.15 to −0.02; P = 0.0136) for the validation dataset. Figure 1 View largeDownload slide Individual QTcB and QTcNeo values vs. heart rate and their linear regression estimates for 2500 ECGs in the study dataset (upper panel) used to derive QTcNeo and the 2500 ECGs in the validation dataset (lower panel). The difference between the slopes of QTcB and QTcNeo was −0.09 ms/b.p.m. (95% confidence interval −0.15 to −0.02; P = 0.0114) for the study dataset and −0.09 ms/b.p.m. (95% confidence interval −0.15 to −0.02; P = 0.0136) for the validation dataset. Figure 2 View largeDownload slide Receiver operating characteristic curves for the five QT correction formulae. (Left panel) Extensive overlap of QTcB, QTcF, QTcL, and QTcNeo, when plotted on a scale of 0 to 1 for sensitivity and 1-specificity is shown. (Right panel) The shaded area in the left panel is expanded such that 1-specificity (x-axis) is plotted on a scale of 0 to 0.04 for visual differentiation of the receiver operating characteristic curves for QTcB, QTcF, QTcL, and QTcNeo. Figure 2 View largeDownload slide Receiver operating characteristic curves for the five QT correction formulae. (Left panel) Extensive overlap of QTcB, QTcF, QTcL, and QTcNeo, when plotted on a scale of 0 to 1 for sensitivity and 1-specificity is shown. (Right panel) The shaded area in the left panel is expanded such that 1-specificity (x-axis) is plotted on a scale of 0 to 0.04 for visual differentiation of the receiver operating characteristic curves for QTcB, QTcF, QTcL, and QTcNeo. Discussion The main finding of this study is that, with one exception, most QT interval correction formulae provide relatively similar results in terms of correction for HR and of the ability to identify infants with LQTS. The exception is represented by the Hodges correction formula which grossly overcorrected for the effect of HR and which had extremely poor positive predictive values for identification of LQTS infants. Among the four traditional formulae tested (Bazett, Fridericia, Hodges, and Framingham), the one proposed by Bazett10 in 1920 turned out to be the best. A vexed question It is almost 100 years since Bazett described his formula to measure the QT interval.10 Since then, several publications have strongly criticized the inability of his method to properly correct for HR changes leading to a search for better correction formulae.19–21 The excessive attention to the mathematical validity of the correction for HR might be understandable if the measurement of the actual QT interval were truly accurate, but this is seldom the case in clinical practice. We actually deal with a gross measurement and the end of the T-wave is often so indistinct that even expert cardiologists reach different conclusions.22–24 Therefore, there is an ‘intrinsic measurement error’ which voids the significance of a ‘perfect correction’ and makes its search a self-defeating purpose. It is also very important to keep in mind that clinical significance is usually associated with relatively gross prolongations (at least 20–30 ms above the upper limit of normal values). What really matters, therefore, is the ability of the formula used to distinguish between normal and abnormal values. In this regard, a group quite experienced with LQTS has recognized the consistent diagnostic value of the Bazett’s correction,25 a conclusion shared by others as well.26 There is one specific condition in which there is indeed the need for extremely accurate measurements of the QT interval, and this concerns the studies with new drugs when regulatory agencies wish to exclude a possible QT prolongation of a magnitude that is likely to produce drug-induced LQTS even in apparently normal individuals.27,28 For this important objective, a careful and individualized QT correction formula such as the one proposed by Malik et al.29 or a study population-specific correction formula is certainly appropriate.30,31 The priority here is to have a HR independent QT correction formula and this issue is quite different from the objective of identifying infants likely to be affected by LQTS. Similarly, accurate QT interval corrections are also needed in physiological studies that investigate other effects on myocardial repolarization. Present data In a set of ECGs recorded in 5000 healthy neonates between the 15th and the 25th day of life, we tested two commonly used linear formulae (Hodges’ and Framingham) and two non-linear exponential formulae (Bazett’s and Fridericia’s). We also derived, and then tested, a population-specific non-linear exponential QT correction formula (QTcNeo), where the QTc value was obtained by QTcNeo = QT/RRN, where the value of N was derived from 2500 randomly selected ECGs and was found to be 0.467. To decide which of these five formulae was the best or the most useful for assessing the QT interval in neonates, we used two criteria—the mathematical ability to provide a corrected QT value that was most HR independent, and the ability to correctly identify neonates with LQTS confirmed by genetic testing. The first criterion was based on plotting the QTc by each of the five correction formulae vs. the HR values from all ECGs. The formula for which the linear regression line fitting the QTc-HR data was closest to horizontal would be considered the best, as this would imply that the corrected QT value was now independent of the HR. The Hodges correction formula grossly overcorrected for HR (slope = +0.962 ms/b.p.m.); thus, for every 10 b.p.m. increase in HR, QTcH would overcorrect the QTc by 9.62 ms. The Framingham and Fridericia methods slightly undercorrected for the effect of HR (QTcL slope = −0.364 and QTcF slope = −0.308); i.e. the Framingham formula undercorrected the QTc by 3.64 ms and Fridericia’s by 3.08 ms for every 10 b.p.m. increase in HR. QTcB performed remarkably well, with a slope of +0.086 ms/b.p.m., implying that it would overcorrect by only 0.86 ms for every 10 b.p.m. increase in HR. QTcNeo performed the best, with a slope of −0.0132 ms/b.p.m., undercorrecting by 0.132 ms for every 10 b.p.m. increase in HR. However, the difference between the slopes of QTcB and QTcNeo were clinically not significant, as the difference between them was only 0.9 ms for an increase in HR of 10 b.p.m. The inescapable conclusion is that QTcB provided the most HR independent QT correction in this cohort of 5000 neonates and would give acceptable results, given the ease of calculation and the fact that the Bazett’s formula is familiar to most cardiologists worldwide. The second criterion used was to compare the predictive value of each of these formulae to correctly identify neonates with the congenital LQTS. We found that the AuROC curve (a measure of the diagnostic accuracy of a test) was significantly less for the Hodges correction formula than the other four methods. The other four QT correction formulae had comparable areas under the ROC curve, suggesting that any one of them would be equally good in identifying neonates with the LQTS. The optimum cut-off values used to identify neonates with LQTS, however, varied widely. The cut-off values were 394 ms for QTcF, 460 ms for QTcB, 383 ms for QTcL, and 446 ms for QTcNeo. Each of these had a 100% negative predictive value but the positive predictive value was 39–40% for QTcB, QTcF, and QTcNeo and 32% for QTcL (Framingham formula). Considering the ease of calculation and the wide availability of QTcB, we concluded that QTcB best served the purpose of identifying neonates with the LQTS, although other correction formulae were also acceptable alternatives using the cut-off value that we found, except QTcH. The present findings differ from what has been reported by a number of studies on QTc correction formulae in various populations. One possible explanation lies in the fact that most previous studies were performed in adults or older children who, for the most part, had HRs between 55 and 80 b.p.m. whereas newborns have HRs around 150 b.p.m. Limitations In the original study, initially only neonates with QTcB >470 ms and their families underwent genetic screening, but this was later extended to infants with a QTc between 460 and 470 ms. It is not possible to exclude that a few neonates with QTc >440 or 450 ms may have carried LQTS-causing mutations. In our large prospective screening study,1 one ECG was recorded in each neonate. Whenever a QTc >450 ms was found, the ECG was repeated within 2 weeks to confirm the initial finding, but only the first ECG was included in this study. Thus, the QTc vs. HR data analysed here contained only a single ECG from each neonate. If multiple ECGs are recorded in the same individual at different time points and various QTc methods are compared, the method that provides the best HR independent QTc value may differ from individual to individual.29 However, a population-specific QT correction formula is the one necessary in clinical practice where usually only a single ECG is obtained per individual to screen for LQTS. Lastly, although more complex methods for modelling the QT-RR relationship have been proposed which might more accurately define the QT-RR relationship,29 we limited our analysis to the development and validation of a single log-linear QT correction formula (QTcNeo) applicable over the entire range of HRs in neonates. Our study also has some important strengths: the study included a large number of neonates prospectively studied for LQTS, the diagnosis of LQTS was confirmed by genetic testing, and all ECGs were high-resolution digital ECGs that were interpreted in a central laboratory. Implications The main implication of our study is that the Bazett’s correction can be used with confidence when screening infants for the possible presence of LQTS. We also found that QTcB provided the most HR independent QT correction and could, therefore, also be used to compare QTc intervals in serial ECGs recorded in neonates at different HRs. In the context of the congenital LQTS, it is worthwhile to note that even in adults with LQTS, most clinical and research data come from QTcB values.32–34 Thus, in a clinical setting where what matters most is the ability to identify grossly abnormal values associated with increased risk for life-threatening arrhythmias, our study, taken in conjunction with the guidelines for diagnosis of LQTS in adults, children, and neonates,34 suggests that the Bazett’s correction can be used with confidence to make recommendations for diagnostic or therapeutic decisions.3,14,35 The previously identified cut-off value of 460 ms provides the best discriminative value for QTcB in neonates.34,36 The Fridericia’s or Framingham correction methods could also be employed, albeit using different cut-off values. Footnotes See page 2896 for the editorial comment on this article (doi: 10.1093/eurheartj/ehy386) Acknowledgements The authors are grateful to Pinuccia De Tomasi for expert editorial support and to Pramod Kadam for statistical support. Funding This work was supported by the Italian Ministry of Health and Regione Lombardia Ricerca Finalizzata 2001 grant ‘Studio sulla prevalenza, il significato clinico e l’evoluzione delle anomalie ECG neonatali associate ad aritmie nell’infanzia’. Conflict of interest: G.K.P. and S.K. are employees of IQVIA. D.R.K. and Y.Y.L. are consultants to IQVIA’s Cardiac Safety Services. IQVIA is a contract research organization that provides scientific and technical services for clinical trials conducted by pharmaceutical companies involved in new drug development. Other than this, the authors declare no professional, academic, competitive, or financial conflicts of interest. References 1 Schwartz PJ , Stramba-Badiale M , Crotti L , Pedrazzini M , Besana A , Bosi G , Gabbarini F , Goulene K , Insolia R , Mannarino S , Mosca F , Nespoli L , Rimini A , Rosati E , Salice P , Spazzolini C. Prevalence of the congenital long-QT syndrome . Circulation 2009 ; 120 : 1761 – 1767 . Google Scholar CrossRef Search ADS PubMed 2 Schwartz PJ , Periti M , Malliani A. The long Q-T syndrome . Am Heart J 1975 ; 89 : 378 – 390 . Google Scholar CrossRef Search ADS PubMed 3 Schwartz PJ , Ackerman MJ. The long QT syndrome: a transatlantic clinical approach to diagnosis and therapy . Eur Heart J 2013 ; 34 : 3109 – 3116 . Google Scholar CrossRef Search ADS PubMed 4 Arnestad M , Crotti L , Rognum TO , Insolia R , Pedrazzini M , Ferrandi C , Vege A , Wang DW , Rhodes TE , George AL Jr , Schwartz PJ. Prevalence of long QT syndrome gene variants in sudden infant death syndrome . Circulation 2007 ; 115 : 361 – 367 . Google Scholar CrossRef Search ADS PubMed 5 Saul JP , Schwartz PJ , Ackerman MJ , Triedman JK. Rationale and objectives for ECG screening in infancy . Heart Rhythm 2014 ; 11 : 2316 – 2321 . Google Scholar CrossRef Search ADS PubMed 6 Schwartz PJ , Stramba-Badiale M , Segantini A , Austoni P , Bosi G , Giorgetti R , Grancini F , Marni ED , Perticone F , Rosti D , Salice P. Prolongation of the QT interval and the sudden infant death syndrome . N Engl J Med 1998 ; 338 : 1709 – 1714 . Google Scholar CrossRef Search ADS PubMed 7 Chhina S , Peverini RL , Deming DD , Hopper AO , Hashmi A , Vyhmeister NR. QTc interval in infants receiving cisapride . J Perinatol 2002 ; 22 : 144 – 148 . Google Scholar CrossRef Search ADS PubMed 8 Lupoglazoff JM , Denjoy I , Villain E , Fressart V , Simon F , Bozio A , Berthet M , Benammar N , Hainque B , Guicheney P. Long QT syndrome in neonates: conduction disorders associated with HERG mutations and sinus bradycardia with KCNQ1 mutations . J Am Coll Cardiol 2004 ; 43 : 826 – 830 . Google Scholar CrossRef Search ADS PubMed 9 Gerosa M , Cimaz R , Stramba-Badiale M , Goulene K , Meregalli E , Trespidi L , Acaia B , Cattaneo R , Tincani A , Motta M , Doria A , Zulian F , Milanesi O , Brucato A , Riboldi P , Meroni PL. Electrocardiographic abnormalities in infants born from mothers with autoimmune diseases–a multicentre prospective study . Rheumatology 2007 ; 46 : 1285 – 1289 . Google Scholar CrossRef Search ADS PubMed 10 Bazett HC. An analysis of the time-relations of electrocardiograms . Heart 1997 ; 2 : 177 – 170 . 11 Luo S , Michler K , Johnston P , Macfarlane PW. A comparison of commonly used QT correction formulae: the effect of heart rate on the QTc of normal ECGs . J Electrocardiol 2004 ; 37 : 81 – 90 . Google Scholar CrossRef Search ADS PubMed 12 Chiladakis J , Kalogeropoulos A , Arvanitis P , Koutsogiannis N , Zagli F , Alexopoulos D. Heart rate-dependence of QTc intervals assessed by different correction methods in patients with normal or prolonged repolarization . Pacing Clin Electrophysiol 2010 ; 33 : 553 – 560 . Google Scholar CrossRef Search ADS PubMed 13 Iglesias-Álvarez D , Rodríguez-Mañero M , García-Seara FJ , Kreidieh O , Martínez-Sande JL , Álvarez-Álvarez B , Fernández-López XA , González-Melchor L , Lage-Fernández R , Moscoso-Galán I , González-Juanatey JR. Comparison and validation of recommended QT interval correction formulae for predicting cardiac arrhythmias in patients with advanced heart Failure and cardiac resynchronization devices . Am J Cardiol 2017 ; 120 : 959 – 965 . Google Scholar CrossRef Search ADS PubMed 14 Schwartz PJ , Garson A Jr , Paul T , Stramba BM , Vetter VL , Wren C. Guidelines for the interpretation of the neonatal electrocardiogram. A task force of the European Society of Cardiology . Eur Heart J 2002 ; 23 : 1329 – 1344 . Google Scholar CrossRef Search ADS PubMed 15 Fridericia LS. The duration of systole in an electrocardiogram in normal humans and in patients with heart disease . Acta Med Scand 2009 ; 53 : 469 – 486 . Google Scholar CrossRef Search ADS 16 Hodges M , Salerno D , Erlien D. Bazett's QT correction reviewed: evidence that a linear QT correction for heart rate is better . J Am Coll Cardiol 1983 ; 1 : 694. 17 Sagie A , Larson MG , Goldberg RJ , Bengtson JR , Levy D. An improved method for adjusting the QT interval for heart rate (the Framingham Heart Study) . Am J Cardiol 1992 ; 70 : 797 – 801 . Google Scholar CrossRef Search ADS PubMed 18 DeLong ER , DeLong DM , Clarke-Pearson DL. Comparing areas under two or more correlated receiver operating characteristics curves: a nonparametric approach . Biometrics 1988 ; 44 : 837 – 845 . Google Scholar CrossRef Search ADS PubMed 19 Schwartz PJ , Wolf S. QT interval prolongation as predictor of sudden death in patients with myocardial infarction . Circulation 1978 ; 57 : 1074 – 1077 . Google Scholar CrossRef Search ADS PubMed 20 Algra A , Tijssen JG , Roelandt JR , Pool J , Lubsen J. QTc prolongation measured by standard 12-lead electrocardiography is an independent risk factor for sudden death due to cardiac arrest . Circulation 1991 ; 83 : 1888 – 1894 . Google Scholar CrossRef Search ADS PubMed 21 Aro AL , Reinier K , Rusinaru C , Uy-Evanado A , Darouian N , Phan D , Mack WJ , Jui J , Soliman EZ , Tereshchenko LG , Chugh SS. Electrical risk score beyond the left ventricular ejection fraction: prediction of sudden cardiac death in the Oregon Sudden Unexpected Death Study and the Atherosclerosis Risk in Communities Study . Eur Heart J 2017 ; 38 : 3017 – 3025 . Google Scholar CrossRef Search ADS PubMed 22 Viskin S , Rosovski U , Sands AJ , Chen E , Kistler PM , Kalman JM , Rodriguez Chavez L , Iturralde Torres P , Cruz F FE , Centurión OA , Fujiki A , Maury P , Chen X , Krahn AD , Roithinger F , Zhang L , Vincent GM , Zeltser D. Inaccurate electrocardiographic interpretation of long QT: the majority of physicians cannot recognize a long QT when they see one . Heart Rhythm 2005 ; 2 : 569 – 574 . Google Scholar CrossRef Search ADS PubMed 23 Panicker GK , Karnad DR , Natekar M , Kothari S , Narula D , Lokhandwala Y. Intra-and inter-reader variability in QT interval measurement by tangent and threshold methods in a central electrocardiogram laboratory . J Electrocardiol 2009 ; 42 : 348 – 352 . Google Scholar CrossRef Search ADS PubMed 24 Panicker GK , Karnad DR , Joshi R , Shetty S , Vyas N , Kothari S , Narula D. Z-score for benchmarking reader competence in a central ECG laboratory . Ann Noninvasive Electrocardiol 2009 ; 14 : 19 – 25 . Google Scholar CrossRef Search ADS PubMed 25 Brouwer J , Van Den Berg MP , Grobbee DE , Haaksma J , Wilde AA. Diagnostic performance of various QTc interval formulas in a large family with long QT syndrome type 3: Bazett's formula not so bad after all… . Ann Noninvasive Electrocardiol 2003 ; 8 : 269 – 274 . Google Scholar CrossRef Search ADS PubMed 26 Phan DQ , Silka MJ , Lan YT , Chang RK. Comparison of formulas for calculation of the corrected QT interval in infants and young children . J Pediatr 2015 ; 166 : 960 – 964 . Google Scholar CrossRef Search ADS PubMed 27 Roden DM. Drug-induced prolongation of the QT Interval . N Engl J Med 2004 ; 350 : 1013 – 1022 . Google Scholar CrossRef Search ADS PubMed 28 Schwartz PJ , Woosley RL. Predicting the unpredictable: drug-induced QT prolongation and Torsades de Pointes . J Am Coll Cardiol 2016 ; 67 : 1639 – 1650 . Google Scholar CrossRef Search ADS PubMed 29 Malik M , Färbom P , Batchvarov V , Hnatkova K , Camm AJ. Relation between QT and RR intervals is highly individual among healthy subjects: implications for heart rate correction of the QT interval . Heart 2002 ; 87 : 220 – 228 . Google Scholar CrossRef Search ADS PubMed 30 US Food and Drug Administration . Guidance for Industry: E14 Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs. Food and Drug Administration; 2005 . http://www.fda.gov/downloads/drugs/guidancecomplianceregulatoryinformation/guidances/ucm073153.pdf (14 May 2018). 31 Salvi V , Karnad DR , Panicker GK , Kothari S. Update on the evaluation of a new drug for effects on cardiac repolarization in humans: issues in early drug development . Br J Pharmacol 2010 ; 159 : 34 – 48 . Google Scholar CrossRef Search ADS PubMed 32 Sauer AJ , Moss AJ , McNitt S , Peterson DR , Zareba W , Robinson JL , Qi M , Goldenberg I , Hobbs JB , Ackerman MJ , Benhorin J , Hall WJ , Kaufman ES , Locati EH , Napolitano C , Priori SG , Schwartz PJ , Towbin JA , Vincent GM , Zhang L. Long QT syndrome in adults . J Am Coll Cardiol 2007 ; 49 : 329 – 337 . Google Scholar CrossRef Search ADS PubMed 33 Schwartz PJ , Priori SG , Spazzolini C , Moss AJ , Vincent GM , Napolitano C , Denjoy I , Guicheney P , Breithardt G , Keating MT , Towbin JA , Beggs AH , Brink P , Wilde AAM , Toivonen L , Zareba W , Robinson JL , Timothy KW , Corfield V , Wattanasirichaigoon D , Corbett C , Haverkamp W , Schulze-Bahr E , Lehmann MH , Schwartz K , Coumel P , Bloise R. Genotype-phenotype correlation in the long QT syndrome. Gene-specific triggers for life-threatening arrhythmias . Circulation 2001 ; 103 : 89 – 95 . Google Scholar CrossRef Search ADS PubMed 34 Priori SG , Wilde AA , Horie M , Cho Y , Behr ER , Berul C , Blom N , Brugada J , Chiang CE , Huikuri H , Kannankeril P , Krahn A , Leenhardt A , Moss A , Schwartz PJ , Shimizu W , Tomaselli G , Tracy C. HRS/EHRA/APHRS Expert consensus statement on the diagnosis and management of patients with inherited primary arrhythmia syndromes . Heart Rhythm 2013 ; 10 : 1932 – 1963 . Google Scholar CrossRef Search ADS PubMed 35 Goldenberg I , Moss AJ , Peterson DR , McNitt S , Zareba W , Hong J , Andrews ML , Robinson JL , Locati EH , Ackerman MJ , Benhorin J , Kaufman ES , Napolitano C , Priori SG , Qi M , Schwartz PJ , Towbin JA , Vincent GM , Zhang L. Risk factors for aborted cardiac arrest and sudden cardiac death in children with the congenital Long-QT Syndrome . Circulation 2008 ; 117 : 2184 – 2191 . Google Scholar CrossRef Search ADS PubMed 36 Merri M , Benhorin J , Alberti M , Locati E , Moss AJ. Electrocardiographic quantitation of ventricular repolarization . Circulation 1989 ; 80 : 1301 – 1308 . Google Scholar CrossRef Search ADS PubMed Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

European Heart JournalOxford University Press

Published: Aug 14, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off