- Email: [email protected]
The Poor Performance of RSR′ Pattern on Electrocardiogram Lead V1 for Detection of Secundum Atrial Septal Defects in Children
The Poor Performance of RSR0 Pattern on Electrocardiogram Lead V1 for Detection of Secundum Atrial Septal Defects in Children Ofer Schiller, MD1,2, E. Anne Greene, MD1,2, Jeffrey P. Moak, MD1,2, Marcin Gierdalski, PharmD, PhD1, and Charles I. Berul, MD1,2 Objective To assess the accuracy of RSR0 patterns in lead V1 (RSR0 -V1) in diagnosing atrial septal defects (ASDs) in children.
Study design Children who underwent an electrocardiogram (ECG) during 2010 were divided into 2 ECG groups:
RSR0 -V1 and normal (no RSR0 -V1). Children who underwent an echocardiogram during 2010 were also divided into an ASD group and a normal echocardiogram group. The 4 groups were matched in a 2 2 table format where the RSR0 -V1 was the “test” and ASD was the “disease.” Sensitivity, specificity, positive/negative predictive values, and pre/post-test probabilities were calculated. Results There were 4658 ECG studies included in the analysis: 836 had RSR0 -V1 and 3822 were normal without RSR0 -V1. Of 4935 echocardiographic studies analyzed, 329 had an ASD and 4606 were normal; 1363 patients had both studies done during the study period. The ECG sensitivity for diagnosing an ASD was 36.1%, specificity was 80%, positive predictive value was 14.7%, and negative predictive value was 92.9% with an overall accuracy of 76.2%. Patients with ASD and RSR0 -V1 were significantly older than patients with ASD and no RSR0 -V1 pattern. Conclusion RSR0 -V1 is a poor screening test for the detection of ASD. It should not change the clinical suspicion or the decision to obtain an echocardiogram. Older children without RSR0 -V1 on ECG are unlikely to have an ASD. (J Pediatr 2013;162:308-12).
A
trial septal defects (ASDs) are among the most common congenital heart defects (CHD) and comprise 6%-10% of all CHDs.1 Recent general population-based studies have documented an incidence of 0.15%-0.39%.2,3 Most children with an ASD are asymptomatic throughout childhood and young adulthood.4 Although some studies have documented that a significant proportion of the ASDs will become smaller with time and some will spontaneously close,5 others have argued that the majority of secundum ASDs become larger over time.6 Patients with an ASD may have subtle findings on physical examination, including flow-related systolic ejection murmurs, hyperactive precordium, or fixed splitting of the second heart sound. A similar pulmonary flow murmur is, however, often heard in children without an ASD or any CHD and is frequently referred to as an innocent murmur.7 The diagnostic challenge of differentiating an innocent murmur from an ASD-related flow murmur prompted clinicians to look at electrocardiographic (ECG) criteria for adjunctive data in diagnosing ASDs. A number of studies from the 1950s-1960s associated certain ECG patterns with the presence of hemodynamically significant ASDs.8-10 Most of the patients in these classic studies were adults who became symptomatic over time. One of the most studied ECG patterns that has been associated with the presence of an ASD is the RSR0 complex in lead V1 (RSR0 -V1), with or without mild QRS complex prolongation (incomplete right bundle branch block).9-11 The RSR0 -V1 complex has also been observed in the normal healthy pediatric population12 and has been speculated to be a mere expression of individual differences in the rate of depolarization of the crista supraventricularis9 or a physiologic variability in intraventricular conduction. The aim of the present study was to evaluate the predictive value of RSR0 -V1 pattern in the diagnosis of ASDs in children with an otherwise structurally normal heart.
Methods This is a retrospective chart review that was conducted in a large, university-affiliated pediatric hospital. The institutional review board approved the study and waived the need for written informed consent. The ECG acquisition and management system (MUSE ECG Management; GE Healthcare, Waukesha, Wisconsin) has a set of predefined diagnoses from which From the 1Division of Cardiology, Children’s National Medical Center; and 2Department of Pediatrics, George Washington University Medical School, Washington, DC
ASD CHD ECG RSR0 -V1 RVH
Atrial septal defect Congenital heart defects Electrocardiogram RSR0 pattern in lead V1 Right ventricular hypertrophy
The authors declare no conflicts of interest. This article was presented in abstract form at the American College of Cardiology Scientific Sessions in Chicago, IL March 24-27 2012. 0022-3476/$ - see front matter. Copyright ª 2013 Mosby Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpeds.2012.07.017
308
Vol. 162, No. 2 February 2013 the reading physician can choose. All ECGs are read by a pediatric cardiologist, with many of them interpreted by a pediatric electrophysiologist. The system was queried for the diagnoses as listed later. The echocardiographic system (Xcelera System, Philips, Amsterdam, The Netherlands) also uses predefined diagnoses. All of the echocardiographic studies are read by a pediatric cardiologist, with the vast majority of them interpreted by a specialist in pediatric noninvasive cardiac imaging. The system was queried for the diagnoses as listed later. The reading physicians for both the ECGs and the echocardiograms were blinded to the present study. The study group included patients who had an echocardiographic and/or ECG study in our institution between January 1, 2010, and December 31, 2010. Exclusion criteria were age <30 days or >18 years on the day of the ECG or echocardiogram study. In our center, most new patients referred with a diagnosis of heart murmur undergo an ECG study before being evaluated by the pediatric cardiologist. An echocardiogram is performed after the initial evaluation only if deemed clinically indicated. The patients who had an ECG study were divided into 2 groups: (1) the RSR0 -V1 group included all children with the following ECG diagnoses: “right ventricular conduction delay,” “incomplete right bundle branch block,” “RSR0 in V1,” “RSR or QR in V1 suggests right ventricular conduction delay,” “RSR0 pattern,” “RSR pattern in V1,” “RSR in V1,” or “RSR0 pattern in V1;” (2) The normal ECG group included all children with the following ECG diagnoses: “normal ECG,” “within normal limits,” “within normal limits for age,” “normal for age,” or “normal variant.” ECGs with preexcitation, poor acquisition quality, or pacing or those obtained during pharmacologic challenge testing were excluded. Patients who appeared in both groups 1 and 2 during the study period were excluded from the study. The patients who had an echocardiographic study were similarly divided into 2 groups: (3) The ASD group included all children with the following echocardiographic diagnoses: “small ASD, secundum type,” “small secundum ASD vs. patent foramen ovale,” “moderate ASD, secundum type,” “large ASD, secundum type,” or “ASD, fenestrated septum primum.” Those with minor findings such as physiologic branch pulmonary artery stenosis, mild pulmonary valve gradient (without evidence of obstruction or stenosis), bicuspid aortic valve, trivial tricuspid regurgitation or mitral regurgitation small aortopulmonary collaterals, dilated right ventricle, or enlarged right atrium were included in the study. Echocardiographic studies with all other diagnoses, including patent foramen ovale, were excluded; (4) The normal echocardiography group included all the children with the following echocardiographic diagnoses: “no cardiac disease identified,” “normal echo for age,” or “no structural cardiac disease identified.” Fetal studies were excluded. Patients who appeared in both groups 3 and 4 were excluded from the study.
Statistical Analyses The query results were exported to an Excel spreadsheet (Microsoft Corporation, Redmond, Washington). The 4 groups were matched by medical record number, and the numbers of matched records were entered into a 2 2 contingency table for analysis where the ECG is the “test” and the echocardiogram is the “gold standard.” Sensitivity, specificity, positive and negative predictive values, and accuracy were calculated. Comparison of the mean age across the 4 groups was performed with 1-way ANOVA. Comparison of the mean age of patients with different ASD sizes was performed with the Student t test. The 95% CIs were calculated using the Clopper-Pearson exact CI method with aid of R and PropCIs package. Correlation of ASD size with RSR presence was tested with Pearson c2 test, and the trend was tested with the Cochran-Armitage test.
Results A total of 6198 ECGs were retrieved from the ECG management using the query terms as listed earlier, of which 1540 were excluded (due to preexcitation, poor acquisition quality, pacing, or pharmacologic challenge), yielding 4658 ECG studies. Of those, 836 had RSR0 -V1 variants and 3822 were interpreted as normal. There were 7852 echocardiographic studies retrieved, of which 2717 studies were excluded (most due to major cardiac defects, other than secundum ASD), yielding 4935 echocardiographic studies; of those, 329 had secundum ASD and 4606 were normal. Therefore, the ASD prevalence in our cohort of children who underwent echocardiography during calendar year 2010 (excluding children with other cardiac defects) was 6.7%. The prevalence of RSR0 -V1 patterns was 17.9%. A total of 1363 patients matched between the ECG and echocardiogram groups during the study period and met the inclusion criteria. A 2 2 contingency table (Table I) revealed a sensitivity of 36%, specificity of 80%, positive predictive value of 14.7%, and negative predictive value of 92.9%. The false-discovery rate was 85.3% and the overall accuracy of ECG for detection of ASDs was 76.2%. Preand post-test probabilities were also calculated (Table II). Representative ECGs from the 4 groups are shown in the Figure. Forty children had a large ASD, 52 had a moderate ASD, and 224 had a small ASD. Children with a large ASD were significantly older (mean age of 6 years) than children with Table I. Contingency table showing the ECG and echocardiography results of 1363 children who underwent both studies Echocardiography
ECG Total
RSR0 -V1 Normal
ASD
Normal
Total
43 76 119
249 995 1244
292 1071 1363 309
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Table II. Performance of ECG for diagnosis of ASD Test
Result
Lower CI
Upper CI
Specificity Sensitivity Positive predictive value Negative predictive value False-positive rate False-discovery rate Positive likelihood ratio Pre-test probability Post-test probability
80.0% 36.1% 14.7% 92.9% 20.0% 85.3% 1.81 6.7% 11.4%
77.7% 27.5% 10.9% 91.1% 17.8% 80.7% 1.39
82.2% 45.4% 19.3% 94.3% 22.3% 89.1% 2.35
a moderate (2.4 years) or small (2.8 years) ASD with P values of <.001 and .001, respectively. There was a strong association between size of ASD and RSR pattern (P < .005), with RSR0 -V1 more prevalent among the larger ASDs; however, no linear relationship was found. The mean age of children with a normal echocardiogram, whether they had an RSR0 V1 pattern (8.6 years) or a normal ECG (8 years), was significantly higher than that of children with an ASD. Children with an ASD and RSR0 -V1 pattern were older (mean age 4.8 years) than children with an ASD and a normal ECG (mean age 2.3 years, P < .05).
Discussion In this retrospective cross-sectional single-center study, we found that RSR0 -V1 patterns on the ECG of children with otherwise normal hearts have low sensitivity and positive predictive value for detection of ASDs. Although some consider RSR0 -V1 pattern a normal variant of the pediatric ECG,
Vol. 162, No. 2 we have defined it as abnormal for the interest of the sensitivity analysis. The relatively high false discovery rate indicates that an RSR0 -V1 pattern will falsely predict an ASD in 85.3%, because the majority of patients with RSR0 -V1 ECG have no atrial defect. Interestingly, our data show that a normal ECG can predict an intact atrial septum in almost 93% of the patients. In addition, we have observed that children with an ASD were referred earlier for cardiac evaluation and that children with an ASD and RSR0 -V1 pattern are significantly older than children with an ASD and normal ECG. Inconsistency still exists regarding the nomenclature of conduction abnormalities in the right precordial leads. A definition of “abnormality” was lacking until Menendez and Marriott10 defined normal and abnormal R and R’ waves and described the differential diagnosis of narrow (ie, <0.12 second) RSR0 patterns. A narrow RSR0 pattern can appear in normal subjects and in patients with pectus excavatum, Wolff-Parkinson-White syndrome, right ventricular hypertrophy (RVH), diastolic overloading of the right ventricle, and other conditions. Due to the inconsistent association of subtle ECG changes and ASDs, a number of studies have looked at other ECG patterns and technologies to diagnose ASDs, among them, artificial neural network,13 the Crochetage pattern,14 and QRST time-integral,15 with varying results. Pediatric studies from the recent era are scant. Zufelt et al published in 1998 a retrospective evaluation of ECG patterns in children with an ASD compared with healthy children. The authors found that 68.7% of the children with ASD had both an rsR0 pattern and RVH criteria and only 9% had rsR0 without RVH on their ECGs. The rsR0 pattern was documented in
Figure. A, ECG of an 11-year-old boy with small ASD and mild right atrial enlargement showing RSR0 -V1 pattern. B, ECG of a 17-year-old girl with sickle cell disease and normal cardiac anatomy showing RSR0 -V1 pattern. C, Normal ECG of a 6-month-old girl evaluated for respiratory distress and found to have a moderate ASD. D, Normal ECG of a 17 year-old adolescent evaluated for cardiac thrombi and found to have normal cardiac anatomy and no intra-cardiac lesions. 310
Schiller et al
February 2013 2.6% of control group.16 Similarly, Arrington et al found that the standard ECG criteria for right ventricular enlargement, as defined by Garson,17 had only a 57% sensitivity for diagnosing large ASDs that subsequently required surgical or catheter-based closure. The authors concluded that the ECG cannot be relied on as a screening tool for ASDs.18 Christensen et al showed that 78% of the children with secundum ASDs who required intervention had right ventricular conduction delay on ECG.19 Pulmonary flow murmur that is caused by an ASD is often hard to discriminate from an innocent (functional) murmur.20 In our study, children with normal echocardiograms, often referred for a heart murmur, presented at an older age for cardiac evaluation than did children with an ASD. This may indicate the ability of the primary pediatrician to identify patients with ASD earlier, although other indications for referral (ie, syncope or palpitations) are more common in older children. Our data also show that children with ASD and right ventricular conduction delay are 2.5 years older at presentation for evaluation than are children with an ASD and a normal ECG. As right ventricular volume overload progresses over time, the relatively late development of RSR-V1 pattern is expected. This may explain the finding of a strong association between ASD size and the presence of RSR0 -V1 pattern in children who had both an ASD and RSR0 -V1 pattern. Our study has several significant limitations. As this is a retrospective chart review of patients referred to the cardiology clinic, selection bias could have occurred and affected the results. The ECGs were read by several physicians who, without commonly accepted criteria for right ventricular conduction delay patterns, may have classified minor ECG changes or RSR0 -V1 pattern as normal. On the other hand, this is likely representative of the situation in many larger pediatric cardiology centers where the ECGs are read by the primary cardiologist. A similar problem may have occurred with the echocardiogram study readings as the differentiation between a small secundum ASD and patent foramen ovale is sometimes difficult and interpreter-dependent. As most of our echocardiography reading physicians are trained in pediatric noninvasive cardiac imaging, we suspect that this has not affected our conclusions. The inclusion of small, non–hemodynamically significant ASDs may have contributed to the high false-negative rate, as these patients may not be expected to have ECG abnormalities. As not all the children had an echocardiographic study, some children with normal ECG and an ASD may have been missed, leading to a falsely elevated negative predictive value. One hundred seventy-seven patients with multiple ECGs had at different times a normal ECG and an ECG with the RSR0 V1 pattern during the study period and were excluded from the study. Although this decreased the number of patients in the final contingency table, we preferred stricter inclusion criteria with less ambiguity and more unique groups, reducing the interpretation bias. A post-hoc analysis that included these subjects revealed that excluding these patients did not change the study results.
ORIGINAL ARTICLES Interestingly, the specificity of the ECG findings was moderately good with 80% of the children with normal echocardiographic studies having a normal ECG. This yielded an ECG negative predictive value of 92.9% in ruling out an ASD and might allow the clinician to consider not obtaining further testing such as an echocardiogram to rule out an ASD when the pre-test probability is low. Our finding that children with an ASD and RSR0 -V1 pattern are older than those with an ASD and normal ECG is consistent with the concept that right ventricular volume overload is a progressive process. ECG performs poorly as a screening test for ASDs in children presenting for cardiac evaluation and should not change the clinical decision as to whether to obtain an echocardiogram. n Submitted for publication Feb 7, 2012; last revision received May 16, 2012; accepted Jul 10, 2012. Reprint requests: Charles I. Berul, MD, Division of Cardiology, Children’s National Medical Center, 111 Michigan Ave NW, Washington, DC 20010-2970. E-mail: [email protected]
References 1. Triedman JK. Methodologic issues for database development: Trends. In: Keane JF, Lock JE, Fyler DC, eds. Nadas’ pediatric cardiology. 2nd ed. Philadelphia: Saunders Elsevier; 2006. p. 323-36. 2. Ishikawa T, Iwashima S, Ohishi A, Nakagawa Y, Ohzeki T. Prevalence of congenital heart disease assessed by echocardiography in 2067 consecutive newborns. Acta Paediatr 2011;100:e55-60. 3. Moons P, Sluysmans T, De Wolf D, Massin M, Suys B, Benatar A, et al. Congenital heart disease in 111225 births in Belgium: birth prevalence, treatment and survival in the 21st century. Acta Paediatr 2009;98:472-7. 4. Verheugt CL, Uiterwaal CS, Grobbee DE, Mulder BJ. Long-term prognosis of congenital heart defects: a systematic review. Int J Cardiol 2008;131:25-32. 5. Saxena A, Divekar A, Soni NR. Natural history of secundum atrial septal defect revisited in the era of transcatheter closure. Indian Heart J 2005; 57:35-8. 6. McMahon CJ, Feltes TF, Fraley JK, Bricker JT, Grifka RG, Tortoriello TA, et al. Natural history of growth of secundum atrial septal defects and implications for transcatheter closure. Heart 2002;87:256-9. 7. Newburger JW, Alexander ME, Fulton DR. Innocent murmurs, syncope, and chest pain. In: Keane JF, Lock JE, Fyler DC, eds. Nadas’ pediatric cardiology. 2nd ed. Philadelphia: Saunders Elsevier; 2006. p. 357-71. 8. Barber JM, Magidson O. Atrial septal defect with special reference to the electrocardiogram. Br Heart J 1950;12:203-4. 9. Depasquale NP, Burch GE. Analysis of the RSR’ complex in lead V1. Circulation 1963;28:362-7. 10. Menendez MM, Marriott HJ. Differential diagnosis of RSR0 and dominant R wave patterns in right chest leads. JAMA 1966;198:843. 11. Sung RJ, Tamer DM, Agha AS, Castellanos A, Myerburg RJ, Gelband H. Etiology of the electrocardiographic pattern of “incomplete right bundle branch block” in atrial septal defect: an electrophysiologic study. J Pediatr 1975;87:1182-6. 12. Surawicz B, Childers R, Deal BJ, Gettes LS, Bailey JJ, Gorgels A, et al. AHA/ACCF/HRS recommendations for the standardization and interpretation of the electrocardiogram: part III: intraventricular conduction disturbances: a scientific statement from the American Heart Association Electrocardiography and Arrhythmias Committee, Council on Clinical Cardiology; the American College of Cardiology Foundation; and the Heart Rhythm Society: endorsed by the International Society for Computerized Electrocardiology. Circulation 2009;119:e235-40. 13. Yang S, Yamauchi K, Nonokawa M, Ikeda M. Use of an artificial neural network to differentiate between ECGs with IRBBB patterns of atrial
The Poor Performance of RSR0 Pattern on Electrocardiogram Lead V1 for Detection of Secundum Atrial Septal Defects in Children
311
THE JOURNAL OF PEDIATRICS
www.jpeds.com
septal defect and healthy subjects. Med Inform Internet Med 2002;27: 49-58. 14. Cohen JS, Patton DJ, Giuffre RM. The crochetage pattern in electrocardiograms of pediatric atrial septal defect patients. Can J Cardiol 2000;16: 1241-7. 15. Izumida N, Asano Y, Kiyohara K, Doi S, Wakimoto H, Tsuchiya S, et al. Precordial leads QRST time integrals for evaluation of right ventricular overload in children with congenital heart diseases. J Electrocardiol 1997;30:257-64. 16. Zufelt K, Rosenberg HC, Li MD, Joubert GI. The electrocardiogram and the secundum atrial septal defect: a reexamination in the era of echocardiography. Can J Cardiol 1998;14:227-32.
312
Vol. 162, No. 2 17. Garson A. Chamber enlargement and hypertrophy. In: Garson A, ed. The electrocardiogram in infants and children. Philadelphia: Lea & Febiger; 1983. p. 99. 18. Arrington CB, Tani LY, Minich LL, Bradley DJ. An assessment of the electrocardiogram as a screening test for large atrial septal defects in children. J Electrocardiol 2007;40:484-8. 19. Christensen DD, Vincent RN, Campbell RM. Presentation of atrial septal defect in the pediatric population. Pediatr Cardiol 2005;26:812-4. 20. Rudolph AM. Chapter 8: Atrial septal defect and partial anomalous drainage of the pulmonary veins. In: Rudolph AM, ed. Congenital diseases of the heart: clinical-physiological considerations. 3rd ed. Chichester (UK)/Hoboken (NJ): Wiley-Blackwell; 2009. p. 179-202.
Schiller et al
Study design Children who underwent an electrocardiogram (ECG) during 2010 were divided into 2 ECG groups:
RSR0 -V1 and normal (no RSR0 -V1). Children who underwent an echocardiogram during 2010 were also divided into an ASD group and a normal echocardiogram group. The 4 groups were matched in a 2 2 table format where the RSR0 -V1 was the “test” and ASD was the “disease.” Sensitivity, specificity, positive/negative predictive values, and pre/post-test probabilities were calculated. Results There were 4658 ECG studies included in the analysis: 836 had RSR0 -V1 and 3822 were normal without RSR0 -V1. Of 4935 echocardiographic studies analyzed, 329 had an ASD and 4606 were normal; 1363 patients had both studies done during the study period. The ECG sensitivity for diagnosing an ASD was 36.1%, specificity was 80%, positive predictive value was 14.7%, and negative predictive value was 92.9% with an overall accuracy of 76.2%. Patients with ASD and RSR0 -V1 were significantly older than patients with ASD and no RSR0 -V1 pattern. Conclusion RSR0 -V1 is a poor screening test for the detection of ASD. It should not change the clinical suspicion or the decision to obtain an echocardiogram. Older children without RSR0 -V1 on ECG are unlikely to have an ASD. (J Pediatr 2013;162:308-12).
A
trial septal defects (ASDs) are among the most common congenital heart defects (CHD) and comprise 6%-10% of all CHDs.1 Recent general population-based studies have documented an incidence of 0.15%-0.39%.2,3 Most children with an ASD are asymptomatic throughout childhood and young adulthood.4 Although some studies have documented that a significant proportion of the ASDs will become smaller with time and some will spontaneously close,5 others have argued that the majority of secundum ASDs become larger over time.6 Patients with an ASD may have subtle findings on physical examination, including flow-related systolic ejection murmurs, hyperactive precordium, or fixed splitting of the second heart sound. A similar pulmonary flow murmur is, however, often heard in children without an ASD or any CHD and is frequently referred to as an innocent murmur.7 The diagnostic challenge of differentiating an innocent murmur from an ASD-related flow murmur prompted clinicians to look at electrocardiographic (ECG) criteria for adjunctive data in diagnosing ASDs. A number of studies from the 1950s-1960s associated certain ECG patterns with the presence of hemodynamically significant ASDs.8-10 Most of the patients in these classic studies were adults who became symptomatic over time. One of the most studied ECG patterns that has been associated with the presence of an ASD is the RSR0 complex in lead V1 (RSR0 -V1), with or without mild QRS complex prolongation (incomplete right bundle branch block).9-11 The RSR0 -V1 complex has also been observed in the normal healthy pediatric population12 and has been speculated to be a mere expression of individual differences in the rate of depolarization of the crista supraventricularis9 or a physiologic variability in intraventricular conduction. The aim of the present study was to evaluate the predictive value of RSR0 -V1 pattern in the diagnosis of ASDs in children with an otherwise structurally normal heart.
Methods This is a retrospective chart review that was conducted in a large, university-affiliated pediatric hospital. The institutional review board approved the study and waived the need for written informed consent. The ECG acquisition and management system (MUSE ECG Management; GE Healthcare, Waukesha, Wisconsin) has a set of predefined diagnoses from which From the 1Division of Cardiology, Children’s National Medical Center; and 2Department of Pediatrics, George Washington University Medical School, Washington, DC
ASD CHD ECG RSR0 -V1 RVH
Atrial septal defect Congenital heart defects Electrocardiogram RSR0 pattern in lead V1 Right ventricular hypertrophy
The authors declare no conflicts of interest. This article was presented in abstract form at the American College of Cardiology Scientific Sessions in Chicago, IL March 24-27 2012. 0022-3476/$ - see front matter. Copyright ª 2013 Mosby Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpeds.2012.07.017
308
Vol. 162, No. 2 February 2013 the reading physician can choose. All ECGs are read by a pediatric cardiologist, with many of them interpreted by a pediatric electrophysiologist. The system was queried for the diagnoses as listed later. The echocardiographic system (Xcelera System, Philips, Amsterdam, The Netherlands) also uses predefined diagnoses. All of the echocardiographic studies are read by a pediatric cardiologist, with the vast majority of them interpreted by a specialist in pediatric noninvasive cardiac imaging. The system was queried for the diagnoses as listed later. The reading physicians for both the ECGs and the echocardiograms were blinded to the present study. The study group included patients who had an echocardiographic and/or ECG study in our institution between January 1, 2010, and December 31, 2010. Exclusion criteria were age <30 days or >18 years on the day of the ECG or echocardiogram study. In our center, most new patients referred with a diagnosis of heart murmur undergo an ECG study before being evaluated by the pediatric cardiologist. An echocardiogram is performed after the initial evaluation only if deemed clinically indicated. The patients who had an ECG study were divided into 2 groups: (1) the RSR0 -V1 group included all children with the following ECG diagnoses: “right ventricular conduction delay,” “incomplete right bundle branch block,” “RSR0 in V1,” “RSR or QR in V1 suggests right ventricular conduction delay,” “RSR0 pattern,” “RSR pattern in V1,” “RSR in V1,” or “RSR0 pattern in V1;” (2) The normal ECG group included all children with the following ECG diagnoses: “normal ECG,” “within normal limits,” “within normal limits for age,” “normal for age,” or “normal variant.” ECGs with preexcitation, poor acquisition quality, or pacing or those obtained during pharmacologic challenge testing were excluded. Patients who appeared in both groups 1 and 2 during the study period were excluded from the study. The patients who had an echocardiographic study were similarly divided into 2 groups: (3) The ASD group included all children with the following echocardiographic diagnoses: “small ASD, secundum type,” “small secundum ASD vs. patent foramen ovale,” “moderate ASD, secundum type,” “large ASD, secundum type,” or “ASD, fenestrated septum primum.” Those with minor findings such as physiologic branch pulmonary artery stenosis, mild pulmonary valve gradient (without evidence of obstruction or stenosis), bicuspid aortic valve, trivial tricuspid regurgitation or mitral regurgitation small aortopulmonary collaterals, dilated right ventricle, or enlarged right atrium were included in the study. Echocardiographic studies with all other diagnoses, including patent foramen ovale, were excluded; (4) The normal echocardiography group included all the children with the following echocardiographic diagnoses: “no cardiac disease identified,” “normal echo for age,” or “no structural cardiac disease identified.” Fetal studies were excluded. Patients who appeared in both groups 3 and 4 were excluded from the study.
Statistical Analyses The query results were exported to an Excel spreadsheet (Microsoft Corporation, Redmond, Washington). The 4 groups were matched by medical record number, and the numbers of matched records were entered into a 2 2 contingency table for analysis where the ECG is the “test” and the echocardiogram is the “gold standard.” Sensitivity, specificity, positive and negative predictive values, and accuracy were calculated. Comparison of the mean age across the 4 groups was performed with 1-way ANOVA. Comparison of the mean age of patients with different ASD sizes was performed with the Student t test. The 95% CIs were calculated using the Clopper-Pearson exact CI method with aid of R and PropCIs package. Correlation of ASD size with RSR presence was tested with Pearson c2 test, and the trend was tested with the Cochran-Armitage test.
Results A total of 6198 ECGs were retrieved from the ECG management using the query terms as listed earlier, of which 1540 were excluded (due to preexcitation, poor acquisition quality, pacing, or pharmacologic challenge), yielding 4658 ECG studies. Of those, 836 had RSR0 -V1 variants and 3822 were interpreted as normal. There were 7852 echocardiographic studies retrieved, of which 2717 studies were excluded (most due to major cardiac defects, other than secundum ASD), yielding 4935 echocardiographic studies; of those, 329 had secundum ASD and 4606 were normal. Therefore, the ASD prevalence in our cohort of children who underwent echocardiography during calendar year 2010 (excluding children with other cardiac defects) was 6.7%. The prevalence of RSR0 -V1 patterns was 17.9%. A total of 1363 patients matched between the ECG and echocardiogram groups during the study period and met the inclusion criteria. A 2 2 contingency table (Table I) revealed a sensitivity of 36%, specificity of 80%, positive predictive value of 14.7%, and negative predictive value of 92.9%. The false-discovery rate was 85.3% and the overall accuracy of ECG for detection of ASDs was 76.2%. Preand post-test probabilities were also calculated (Table II). Representative ECGs from the 4 groups are shown in the Figure. Forty children had a large ASD, 52 had a moderate ASD, and 224 had a small ASD. Children with a large ASD were significantly older (mean age of 6 years) than children with Table I. Contingency table showing the ECG and echocardiography results of 1363 children who underwent both studies Echocardiography
ECG Total
RSR0 -V1 Normal
ASD
Normal
Total
43 76 119
249 995 1244
292 1071 1363 309
THE JOURNAL OF PEDIATRICS
www.jpeds.com
Table II. Performance of ECG for diagnosis of ASD Test
Result
Lower CI
Upper CI
Specificity Sensitivity Positive predictive value Negative predictive value False-positive rate False-discovery rate Positive likelihood ratio Pre-test probability Post-test probability
80.0% 36.1% 14.7% 92.9% 20.0% 85.3% 1.81 6.7% 11.4%
77.7% 27.5% 10.9% 91.1% 17.8% 80.7% 1.39
82.2% 45.4% 19.3% 94.3% 22.3% 89.1% 2.35
a moderate (2.4 years) or small (2.8 years) ASD with P values of <.001 and .001, respectively. There was a strong association between size of ASD and RSR pattern (P < .005), with RSR0 -V1 more prevalent among the larger ASDs; however, no linear relationship was found. The mean age of children with a normal echocardiogram, whether they had an RSR0 V1 pattern (8.6 years) or a normal ECG (8 years), was significantly higher than that of children with an ASD. Children with an ASD and RSR0 -V1 pattern were older (mean age 4.8 years) than children with an ASD and a normal ECG (mean age 2.3 years, P < .05).
Discussion In this retrospective cross-sectional single-center study, we found that RSR0 -V1 patterns on the ECG of children with otherwise normal hearts have low sensitivity and positive predictive value for detection of ASDs. Although some consider RSR0 -V1 pattern a normal variant of the pediatric ECG,
Vol. 162, No. 2 we have defined it as abnormal for the interest of the sensitivity analysis. The relatively high false discovery rate indicates that an RSR0 -V1 pattern will falsely predict an ASD in 85.3%, because the majority of patients with RSR0 -V1 ECG have no atrial defect. Interestingly, our data show that a normal ECG can predict an intact atrial septum in almost 93% of the patients. In addition, we have observed that children with an ASD were referred earlier for cardiac evaluation and that children with an ASD and RSR0 -V1 pattern are significantly older than children with an ASD and normal ECG. Inconsistency still exists regarding the nomenclature of conduction abnormalities in the right precordial leads. A definition of “abnormality” was lacking until Menendez and Marriott10 defined normal and abnormal R and R’ waves and described the differential diagnosis of narrow (ie, <0.12 second) RSR0 patterns. A narrow RSR0 pattern can appear in normal subjects and in patients with pectus excavatum, Wolff-Parkinson-White syndrome, right ventricular hypertrophy (RVH), diastolic overloading of the right ventricle, and other conditions. Due to the inconsistent association of subtle ECG changes and ASDs, a number of studies have looked at other ECG patterns and technologies to diagnose ASDs, among them, artificial neural network,13 the Crochetage pattern,14 and QRST time-integral,15 with varying results. Pediatric studies from the recent era are scant. Zufelt et al published in 1998 a retrospective evaluation of ECG patterns in children with an ASD compared with healthy children. The authors found that 68.7% of the children with ASD had both an rsR0 pattern and RVH criteria and only 9% had rsR0 without RVH on their ECGs. The rsR0 pattern was documented in
Figure. A, ECG of an 11-year-old boy with small ASD and mild right atrial enlargement showing RSR0 -V1 pattern. B, ECG of a 17-year-old girl with sickle cell disease and normal cardiac anatomy showing RSR0 -V1 pattern. C, Normal ECG of a 6-month-old girl evaluated for respiratory distress and found to have a moderate ASD. D, Normal ECG of a 17 year-old adolescent evaluated for cardiac thrombi and found to have normal cardiac anatomy and no intra-cardiac lesions. 310
Schiller et al
February 2013 2.6% of control group.16 Similarly, Arrington et al found that the standard ECG criteria for right ventricular enlargement, as defined by Garson,17 had only a 57% sensitivity for diagnosing large ASDs that subsequently required surgical or catheter-based closure. The authors concluded that the ECG cannot be relied on as a screening tool for ASDs.18 Christensen et al showed that 78% of the children with secundum ASDs who required intervention had right ventricular conduction delay on ECG.19 Pulmonary flow murmur that is caused by an ASD is often hard to discriminate from an innocent (functional) murmur.20 In our study, children with normal echocardiograms, often referred for a heart murmur, presented at an older age for cardiac evaluation than did children with an ASD. This may indicate the ability of the primary pediatrician to identify patients with ASD earlier, although other indications for referral (ie, syncope or palpitations) are more common in older children. Our data also show that children with ASD and right ventricular conduction delay are 2.5 years older at presentation for evaluation than are children with an ASD and a normal ECG. As right ventricular volume overload progresses over time, the relatively late development of RSR-V1 pattern is expected. This may explain the finding of a strong association between ASD size and the presence of RSR0 -V1 pattern in children who had both an ASD and RSR0 -V1 pattern. Our study has several significant limitations. As this is a retrospective chart review of patients referred to the cardiology clinic, selection bias could have occurred and affected the results. The ECGs were read by several physicians who, without commonly accepted criteria for right ventricular conduction delay patterns, may have classified minor ECG changes or RSR0 -V1 pattern as normal. On the other hand, this is likely representative of the situation in many larger pediatric cardiology centers where the ECGs are read by the primary cardiologist. A similar problem may have occurred with the echocardiogram study readings as the differentiation between a small secundum ASD and patent foramen ovale is sometimes difficult and interpreter-dependent. As most of our echocardiography reading physicians are trained in pediatric noninvasive cardiac imaging, we suspect that this has not affected our conclusions. The inclusion of small, non–hemodynamically significant ASDs may have contributed to the high false-negative rate, as these patients may not be expected to have ECG abnormalities. As not all the children had an echocardiographic study, some children with normal ECG and an ASD may have been missed, leading to a falsely elevated negative predictive value. One hundred seventy-seven patients with multiple ECGs had at different times a normal ECG and an ECG with the RSR0 V1 pattern during the study period and were excluded from the study. Although this decreased the number of patients in the final contingency table, we preferred stricter inclusion criteria with less ambiguity and more unique groups, reducing the interpretation bias. A post-hoc analysis that included these subjects revealed that excluding these patients did not change the study results.
ORIGINAL ARTICLES Interestingly, the specificity of the ECG findings was moderately good with 80% of the children with normal echocardiographic studies having a normal ECG. This yielded an ECG negative predictive value of 92.9% in ruling out an ASD and might allow the clinician to consider not obtaining further testing such as an echocardiogram to rule out an ASD when the pre-test probability is low. Our finding that children with an ASD and RSR0 -V1 pattern are older than those with an ASD and normal ECG is consistent with the concept that right ventricular volume overload is a progressive process. ECG performs poorly as a screening test for ASDs in children presenting for cardiac evaluation and should not change the clinical decision as to whether to obtain an echocardiogram. n Submitted for publication Feb 7, 2012; last revision received May 16, 2012; accepted Jul 10, 2012. Reprint requests: Charles I. Berul, MD, Division of Cardiology, Children’s National Medical Center, 111 Michigan Ave NW, Washington, DC 20010-2970. E-mail: [email protected]
References 1. Triedman JK. Methodologic issues for database development: Trends. In: Keane JF, Lock JE, Fyler DC, eds. Nadas’ pediatric cardiology. 2nd ed. Philadelphia: Saunders Elsevier; 2006. p. 323-36. 2. Ishikawa T, Iwashima S, Ohishi A, Nakagawa Y, Ohzeki T. Prevalence of congenital heart disease assessed by echocardiography in 2067 consecutive newborns. Acta Paediatr 2011;100:e55-60. 3. Moons P, Sluysmans T, De Wolf D, Massin M, Suys B, Benatar A, et al. Congenital heart disease in 111225 births in Belgium: birth prevalence, treatment and survival in the 21st century. Acta Paediatr 2009;98:472-7. 4. Verheugt CL, Uiterwaal CS, Grobbee DE, Mulder BJ. Long-term prognosis of congenital heart defects: a systematic review. Int J Cardiol 2008;131:25-32. 5. Saxena A, Divekar A, Soni NR. Natural history of secundum atrial septal defect revisited in the era of transcatheter closure. Indian Heart J 2005; 57:35-8. 6. McMahon CJ, Feltes TF, Fraley JK, Bricker JT, Grifka RG, Tortoriello TA, et al. Natural history of growth of secundum atrial septal defects and implications for transcatheter closure. Heart 2002;87:256-9. 7. Newburger JW, Alexander ME, Fulton DR. Innocent murmurs, syncope, and chest pain. In: Keane JF, Lock JE, Fyler DC, eds. Nadas’ pediatric cardiology. 2nd ed. Philadelphia: Saunders Elsevier; 2006. p. 357-71. 8. Barber JM, Magidson O. Atrial septal defect with special reference to the electrocardiogram. Br Heart J 1950;12:203-4. 9. Depasquale NP, Burch GE. Analysis of the RSR’ complex in lead V1. Circulation 1963;28:362-7. 10. Menendez MM, Marriott HJ. Differential diagnosis of RSR0 and dominant R wave patterns in right chest leads. JAMA 1966;198:843. 11. Sung RJ, Tamer DM, Agha AS, Castellanos A, Myerburg RJ, Gelband H. Etiology of the electrocardiographic pattern of “incomplete right bundle branch block” in atrial septal defect: an electrophysiologic study. J Pediatr 1975;87:1182-6. 12. Surawicz B, Childers R, Deal BJ, Gettes LS, Bailey JJ, Gorgels A, et al. AHA/ACCF/HRS recommendations for the standardization and interpretation of the electrocardiogram: part III: intraventricular conduction disturbances: a scientific statement from the American Heart Association Electrocardiography and Arrhythmias Committee, Council on Clinical Cardiology; the American College of Cardiology Foundation; and the Heart Rhythm Society: endorsed by the International Society for Computerized Electrocardiology. Circulation 2009;119:e235-40. 13. Yang S, Yamauchi K, Nonokawa M, Ikeda M. Use of an artificial neural network to differentiate between ECGs with IRBBB patterns of atrial
The Poor Performance of RSR0 Pattern on Electrocardiogram Lead V1 for Detection of Secundum Atrial Septal Defects in Children
311
THE JOURNAL OF PEDIATRICS
www.jpeds.com
septal defect and healthy subjects. Med Inform Internet Med 2002;27: 49-58. 14. Cohen JS, Patton DJ, Giuffre RM. The crochetage pattern in electrocardiograms of pediatric atrial septal defect patients. Can J Cardiol 2000;16: 1241-7. 15. Izumida N, Asano Y, Kiyohara K, Doi S, Wakimoto H, Tsuchiya S, et al. Precordial leads QRST time integrals for evaluation of right ventricular overload in children with congenital heart diseases. J Electrocardiol 1997;30:257-64. 16. Zufelt K, Rosenberg HC, Li MD, Joubert GI. The electrocardiogram and the secundum atrial septal defect: a reexamination in the era of echocardiography. Can J Cardiol 1998;14:227-32.
312
Vol. 162, No. 2 17. Garson A. Chamber enlargement and hypertrophy. In: Garson A, ed. The electrocardiogram in infants and children. Philadelphia: Lea & Febiger; 1983. p. 99. 18. Arrington CB, Tani LY, Minich LL, Bradley DJ. An assessment of the electrocardiogram as a screening test for large atrial septal defects in children. J Electrocardiol 2007;40:484-8. 19. Christensen DD, Vincent RN, Campbell RM. Presentation of atrial septal defect in the pediatric population. Pediatr Cardiol 2005;26:812-4. 20. Rudolph AM. Chapter 8: Atrial septal defect and partial anomalous drainage of the pulmonary veins. In: Rudolph AM, ed. Congenital diseases of the heart: clinical-physiological considerations. 3rd ed. Chichester (UK)/Hoboken (NJ): Wiley-Blackwell; 2009. p. 179-202.
Schiller et al