The effects of body mass index on surface electrocardiograms in young adults

Available online at www.sciencedirect.com

Journal of Electrocardiology 45 (2012) 646 – 651 www.jecgonline.com

The effects of body mass index on surface electrocardiograms in young adults☆ Javed M. Nasir, MD,⁎ Bernard J. Rubal, PhD, Samuel O. Jones, MD, MPH, Anand D. Shah, MD Department of Cardiology, SAMMC, San Antonio, TX, USA Received 10 April 2012

Abstract

Introduction: While BMI is known to affect ECG measurements, these effects have not been well characterized in young adults. Methods: We retrospectively reviewed all ECGs performed in adults 18 to 35 years old at a single institution over a 30 year period. ECG measurements were derived electronically and stratified by WHO BMI category. Results: A total of 55,218 ECGs were included. Increasing BMI led to increased P wave duration and decreasing P, R, and T wave axes. Additionally, while increasing BMI led to less R wave voltage, J point elevation, and T wave amplitude in patients with a BMI ≥18.5 kg/m 2, there was also a decrease in the measured parameters in patients with a BMI b18.5 kg/m 2. Discussion: BMI had significant effects on ECG measurements. For accurate assessment of ECGs, these data should be incorporated into established nomograms. Further investigation into the effects of BMI on the ECG is warranted. Published by Elsevier Inc.

Keywords:

Electrocardiogram; Body mass index; ECG; BMI

Introduction Since the invention of the galvanometer by Einthoven in 1901, the electrocardiogram (ECG) has become the most commonly performed cardiac procedure, with an estimated 20 million ECGs performed annually in the United States (US). 1 The ECG has become ubiquitous in modern medicine and is essential for the diagnosis, treatment, and screening for cardiovascular diseases. While most ECG readers consider the patient's baseline demographics (e.g., gender, age, ethnicity, and body composition) during ECG interpretation, the effects of these variables have not been fully characterized. For more than three decades, San Antonio Military Medical Center (SAMMC) has collected and electronically stored ECGs obtained from active duty and dependent personnel from diverse segments of the US population. This

☆ Disclaimer: The opinions expressed on this document are solely those of the authors and do not represent an endorsement by or the views of the United States Air Force, the Department of Defense, or the United States Government. ⁎ Corresponding author. Department of Cardiology, 81st MDOS/ SGOMC, Keesler AFB, MS 39564, USA. E-mail address: [email protected]

0022-0736/$ – see front matter. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jelectrocard.2012.07.022

large electronic repository provides a unique opportunity to assess for demographical differences in ECG measurements in a young adult population unlikely to have comorbid conditions. Here we analyze and report the effects of the body mass index (BMI) on ECG waveform measurements and assess the interaction of BMI with gender, age, and ethnicity. Although there are limitations in the accuracy of BMI for reporting adiposity, BMI is the most prevalent index of obesity and is used as an index of cardiovascular risk. 2

Methods Study design For this project we queried all the ECGs stored in SAMMC's MUSE (General Electric Healthcare) database between 1980 and 2010 in subjects 18 to 35 years old. We limited our search to ECGs that had been verified as a “normal ECG” after interpretation by a member of the Cardiology department, had heart rate of 60 to 100 bpm, a QRS duration b120 ms, a PR interval greater than 110 ms, and a QRS axis between −30° and 90°. Computerized analysis of all waveforms was performed with Marquette

J.M. Nasir et al. / Journal of Electrocardiology 45 (2012) 646–651

Statistical analysis

Table 1 Baseline demographics.

Mean age (years) Ethnicity (%) - Caucasian (n=40,255) - African American (n=8666) - Hispanic (n=2281) - Other (n=4016) Height (cm) Weight (kg) Mean BMI (kg/m2) BMI category (%) - BMI b18.5 kg/m2 (n=1300) - BMI 18.5 to b25 kg/m2 (n=29,054) - BMI 25 to b30 kg/m2 (n=19,875) - BMI ≥30 kg/m2 (n=4989)

647

All (n=55,218)

Male (n=34,101)

Female (n=21,117)

25.9±5.4

25.8±5.3

26.1±5.4

72.9 15.7

77.5 11.7

65.5 22.1

4.13 7.27 171.5±9.9 73.3±13.5 24.9±3.9

3.9 6.9 176.6±7.4 78.5±11.1 25.2±3.2

4.5 7.9 163.2±7.6 65.0±12.5 24.4±4.7

2.4

1.0

4.5

52.6

47.7

60.5

36.0

44.3

22.6

9.0

7.0

12.4

This figure illustrates the baseline demographics of the study cohort. When males were compared to females, there were significant differences (p≤0.001) in all demographic data.

12SL (General Electric Healthcare), and demographic data were collected by ECG technicians when the ECG was acquired. BMI was calculated as weight/body surface area 2 (kg/m 2) and classified into four categories using the World Health Organization (WHO) BMI criteria: 3 underweight (b18.5 kg/m 2), normal (18.5 to b25 kg/m 2), overweight (25 to b30 kg/m 2), and obese (≥30 kg/m 2). For this analysis, age was classified into two groups, 18 to ≤25 years or N25 to 35 years, while ethnicity was categorized dichotomously as Caucasian or non-Caucasian. Waveform data were collected from all leads, but we have limited this analysis and report to select, clinically relevant measures. ECG measurements are reported and analyzed for the P wave duration, P wave amplitude (lead II), axes (P, R, and T wave), R wave amplitude, T wave amplitude, and J point elevation. More specifically, R wave amplitude was summed and examined in three categories: all leads, the precordial leads, and the frontal leads. T wave amplitude was treated as an absolute value and examined in four separate categories: maximal T wave amplitude, summed from all leads, summed from the precordial leads, and summed from the limb leads. J point elevation was also summed and examined in five categories: all leads, inferior leads (II, III, aVF), lateral leads (I, V4–6), high lateral leads (I, aVL) and anterior leads (V2–V4). Details of the derivation and validity of ECG waveform measurements from 12SL are available elsewhere. 4 In general, amplitudes are measured in each lead individually, and the interval measurements (e.g., P wave duration, PR interval, QRS interval, and QT interval) are measured simultaneously across all leads using a median complex. These measurements have been validated against electrocardiographic common standards, and many of the measurements have been validated independently in the scientific literature. 4

Data are presented as means±standard deviations (SD), percentages, or graphically as box-and-whisker plots to indicate the dispersion of data about the median. The whiskers of the box plot represent 1.5× the lower and upper inner quartiles limits or limits of the data if data are b1.5× quartile limit. To improve clarity, the box-and-whisker plots are displayed without outliers. A Student's t-test was employed to assess differences between genders in scalar demographic data, and a Z-test was used to assess difference in population frequencies. A fixed-effect linear model ANOVA (IBM SPSS Version 19.0) was employed to assess the difference in the ECG characteristic among WHO BMI categories and the significance of interaction between BMI and age, gender, and ethnicity. The Bonferroni post hoc test was applied for multiple group comparisons. All statistical tests were two-tailed with p values considered significant if b0.05. Results Our MUSE database contained 99,499 ECGs that met the inclusion criteria specified above. Due to limitations in MUSE's export function, completed data collection was achieved for 93,218 of these subjects. After exclusion of 14,586 duplicates and 23,414 subjects with incomplete or inaccurate demographical data, 55,218 subjects remained in our final cohort. Subject characteristics The demographics for the final cohort are presented in Table 1. The subjects were predominately Caucasian (72.9%) and male (61.8%), with the entire cohort having a mean age of 25.9±5.4 years. The mean BMI was 24.9±2.9 kg/m 2 with 52.6% of the subjects having a normal BMI, 36.0% were overweight, 2.4% were underweight, and 9.0% were obese. P wave In this cohort, the mean P wave duration was 98.3± 14.1 ms. As illustrated in Table 2, with increasing BMI there was a significant increase in the P wave duration (pb0.0001). Significant differences were observed between all BMI categories (p≤0.001), and no significant interactions were observed between BMI and gender, ethnicity or age. The mean P wave amplitude was 0.12±0.04 mV. While BMI had a statistically significant effect on the amplitude (pb0.0001), the differences in the amplitudes was minimal (b 0.004 mV) and unlikely to be clinically significant. PR interval The mean PR interval was 148± 19 ms. As illustrated in Table 2, increasing BMI was associated with a significant increase in the PR interval (pb0.0001). Comparison between groups showed a significant difference among all BMI categories (pb0.0001), except when overweight subjects were compared to obese subjects (p =0.24), and there were significant interactions of BMI with gender (p=0.0001) and ethnicity (p=0.02).

648

J.M. Nasir et al. / Journal of Electrocardiology 45 (2012) 646–651

Table 2 Results from BMI univariate analysis.

P wave duration (ms) P wave amplitude (mV) P axis (degrees) R axis (degrees) T axis (degrees) PR interval (ms) QRS duration (ms) Heart rate (bpm) QT duration (ms) QTc duration (ms) Maximal T wave amplitude (mV) Summed T wave amplitude—limb leads (mV) Summed T wave amplitude—precordial leads (mV) Summed T wave amplitude (mV) Summed R wave amplitude (mV) Summed R wave amplitude—limb leads (mV) Summed R wave amplitude—precordial leads (mV) Total J point elevation (mV) Anterior J point elevation V2–V4 (mV) Lateral J point elevation I, V4–6 (mV) Inferior J point elevation II, III, aVF (mV)

Overall (n=55,218)

BMI b18.5 kg/m2 (n=1300)

BMI 18.5 to b25 kg/m2 (n=29,054)

BMI 25 to b30 kg/m2 (n=19,875)

BMI ≥30 kg/m2 (n=4989)

ANOVA, p value

98±14 0.12±0.43 48.1±21.1 61.6±21.7 44.6±16.5 148±19 89±9 71.8±9.2 381±22 413±18 0.63±0.26 1.42±0.46

92±16 0.12±0.50 54.0±22.0 70.5±17.5 53.6±15.4 143±19 85±8 73.8±9.8 379±23 415±18 0.56±0.24 1.27±0.46

96±15 0.12±0.46 50.2±21.6 66.2±19.4 48.4±15.4 147±19 88±9 71.6±9.0 382±22 412±18 0.64±0.27 1.42±0.47

101±14 0.12±0.40 46.0±20.4 57.3±22.6 40.5±16.1 151±19 90±8 71.3±9.0 382±22 411±18 0.64±0.24 1.44±0.44

102±13 0.12±0.39 43.0±18.9 49.6±23.1 35.8±16.4 152±18 88±9 74.6±10.0 378±23 417±19 0.52±0.21 1.34±0.41

b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001 b0.0001

2.44±0.97

2.13±0.87

2.50±0.99

2.49±0.94

1.97±0.82

b0.0001

3.85±1.31 10.3±2.8 4.0±1.3

3.40±1.23 9.6±2.8 3.9±1.3

3.92±1.35 10.5±2.8 4.1±1.3

3.93±1.27 10.3±2.6 3.9±1.2

3.30±1.12 9.2±2.4 3.7±1.1

b0.0001 b0.0001 b0.0001

6.3±1.9

5.7±1.8

6.4±1.9

6.4±1.8

5.5±1.6

b0.0001

0.31±0.22 0.13±0.12 0.12±0.10 0.08±0.08

0.26±0.20 0.10±0.10 0.10±0.09 0.08±0.08

0.32±0.22 0.13±0.12 0.12±0.10 0.09±0.08

0.32±0.22 0.13±0.11 0.13±0.11 0.08±0.08

0.26±0.19 0.09±0.09 0.10±0.09 0.07±0.07

b0.0001 b0.0001 b0.0001 b0.0001

This table illustrates the results stratified by BMI. P values are reported for the fixed-effect ANOVA analysis. See the text for comparison between individual groups.

QRS interval The mean QRS interval was 89±9 ms. As illustrated in Table 2, BMI had a statistically significant effect on the QRS duration (pb0.0001). However, this effect was minimal, and when subjects with a normal BMI were compared to the other WHO groups, the differences in the mean values were ≤3 ms.

BMI categories (pb0.0001), and there was a significant interaction between BMI and ethnicity (pb0.0001). R wave axis The mean frontal axis of the R wave was 61.6° ±21.7°. As illustrated in Table 2 and Fig. 1, increasing BMI led to a significant decrease in the axis (pb0.0001). Comparison

QT interval The mean QT interval was 381±22 ms. As illustrated in Table 2, a significant variation in QT interval was noted with BMI (pb0.0001). However, this effect was minimal, and when subjects with a normal BMI were compared to the other groups, the difference in means was ≤4 ms. In this cohort the mean heart rate HR was 71.8±9.7 bpm, and there was a significant difference between BMI groups (pb0.0001) with a higher HR in the underweight (73.8±9.8 bpm) and obese groups (74.6±10.0 bpm) when compared to the normal (71.6± 9.0 bpm) and overweight groups (71.3±9.0 bpm). When the QT interval was adjusted for the heart rate using the Bazett's formula, the QTc results are similar to the QT interval with minimal differences between groups (≤5 ms) when subjects with a normal BMI were compared to the other WHO groups. P wave axis The mean frontal axis of the P wave was 48.1°± 21.1°. As illustrated in Table 2 and Fig. 1, increasing BMI led to a significant decrease in the axis (pb0.0001). Comparison between groups showed a significant difference among all

Fig. 1. Box-and-whisker plot of P, R, T wave axes by BMI category. This figure illustrates the affect of BMI on the P, R, and T wave axes. For all, with increasing BMI, there was a leftward shift in the axis (pb0.0001).

J.M. Nasir et al. / Journal of Electrocardiology 45 (2012) 646–651

649

between groups showed a significant difference among all BMI categories (pb0.0001), and there were significant interactions of BMI with gender (p= 0.002), age (p=0.008), and ethnicity (pb0.0001). T wave axis The mean frontal axis of the T wave was 44.6° ±16.5°. As illustrated in Table 2 and Fig. 1, increasing BMI led to a significant decrease in the axis (pb0.0001). Comparison between groups showed a significant difference among all BMI categories (pb0.0001), and there were significant interactions between BMI and both gender (pb0.0001) and age (p=0.03). R wave amplitude The mean total R wave amplitude was 10.3±2.8 mV. As illustrated in Table 2 and Fig. 2, BMI significantly affected R wave amplitude (pb0.0001). When compared to normal subjects (10.5±2.8 mV), increasing adiposity led to decreased voltage in both the overweight (10.3±2.6 mV) and obese (9.2±2.4 mV) subjects; however, a decreased voltage was also noted in the underweight subjects (9.6±2.8 mV). Comparison between groups showed significant differences between all BMI categories (pb0.0001), and there were significant interactions of BMI with gender (pb0.0001) and age (p= 0.002). Summed R wave amplitudes from the limb and precordial leads were examined separately with similar results. T wave amplitude The maximal T wave amplitude was most frequently found in lead V2 (36.8%) and was present in V2–V4 in 80.1% of the subjects. The mean maximal T wave amplitude was 0.63 ± 0.26 mV. As illustrated in Table 2, BMI significantly affected the maximal T wave amplitude (pb 0.0001). There were significant differences (pb0.0001) between groups with a decreased amplitude in the under-

Fig. 3. Box-and-whisker plot of the J point elevation by BMI category. This figure illustrates the affect of BMI on the J point elevation. When compared to subjects with a normal BMI, obese and underweight subjects had less elevation.

weight (0.56 ± 0.24 mV) and obese groups (0.52 ± 0.21 mV) when they were compared to the normal BMI group (0.65 ± 0.27 mV). The amplitude in the overweight group (0.65 ± 0.24 mV) did not differ from the normal group (p=1.00). There were significant interactions of BMI with gender (pb0.0001), age (pb0.0001), and ethnicity (pb0.01). When the summed T waves were examined in the limb leads, precordial leads, or all leads, the results were similar. J point elevation The mean summed J point elevation was 0.31± 0.22 mV. As illustrated in Fig. 3 and Table 2, BMI has a significant effect (pb0.0001) on J point elevation. When compared to normal subjects (0.32 ±0.22), increasing adiposity led to decreased voltage in the obese (0.26 ± 0.19 mV) and underweight subjects (0.26 ± 0.20 mV), but not in the overweight subjects (0.32 ±0.22 mV, p =0.19). There were significant interactions of BMI with gender (pb0.0001), age (p=0.03), and ethnicity (p=0.043). Summed J points in the inferior, lateral, high lateral, and anterior leads were examined separately with similar results.

Discussion

Fig. 2. Box-and-whisker plot of R wave amplitude by BMI category. This figure illustrates the affect of BMI on the R wave amplitude. When compared to subjects with a normal BMI, an increasing or decreasing BMI led to a decreased voltage.

While the ECG is the most commonly performed cardiac procedure, 1 and adiposity is known to affect the ECG, the effects of adiposity on ECG measurement have not been previously evaluated in large cohorts. Here we report the largest study to date on the effects of BMI on the ECG. The most marked electrocardiographic manifestation of BMI was in the P, R, and T wave axes. With increasing BMI there was a leftward shift of both atrial and ventricular vectors with the P wave axis affected most. This leftward change in the axes with increasing adiposity and rightward change in the axes with weight loss has been previously described; 5-9 however, this is the first large study to show

650

J.M. Nasir et al. / Journal of Electrocardiology 45 (2012) 646–651

incremental change in the axes with increasing BMI. While the etiology of this axis shift is not fully understood, it may be due to a more horizontal and leftward cardiac orientation with obesity. In this study, we demonstrated an increase in the P wave duration with increasing BMI while the amplitude was relatively unchanged. This increase in duration with increasing adiposity is consistent with prior smaller data sets, 7,10-12 and others have reported increased P wave durations and left atrial sizes by echocardiography when obese subjects were compared to age- and sex-matched controls. 10,11 It is possible the increased duration demonstrated here may be a manifestation of left atrial enlargement directly related to obesity, as secondarily associated disease states (e.g., systemic hypertension, diabetes mellitus, and coronary artery disease) are unlikely to have developed yet in this young cohort. There was also a significant increase in the PR interval with increasing BMI, with a mean PR interval of 143 ms in patients with a BMI b18.5 kg/m 2 and 152 ms in subjects with a BMI ≥ 30 kg/m 2. The prolongation of the PR interval was related to the increased P wave duration and not intra-atrial or AV nodal conduction delay. As can be seen in Table 2, the magnitude of the changes between the PR interval and P wave duration was similar between the BMI categories, and subsequent stratification of the PR interval minus the P wave duration by BMI category revealed minimal variation across all categories (≤0.6 ms). There are conflicting data in the literature regarding the effect of BMI on QRS voltage, with studies suggesting obesity is associated with both lower voltages and increased voltages. 6,7 In the present study, we have shown decreasing R wave voltage when overweight and obese subjects are compared to subjects with a normal BMI and decreased voltage when underweight subjects are compared to subjects with a normal BMI. While the etiology of the lower voltages with increasing adiposity has been reported to be a manifestation of increased tissue interposed between the myocardial surface and the surface electrode, this would not explain the lower amplitudes observed when underweight subjects were compared to normal subjects. To further evaluate this, we examined the total QRS voltage by BMI category and saw similar trends (underweight: 15.2±3.7 kg/m 2, normal BMI: 16.3±3.8 kg/m 2, overweight: 16.0±3.4 kg/m 2, and obese: 14.3±3.0 kg/m 2) suggesting the decreased R wave amplitude was not offset by increased S or Q wave amplitudes. One possible explanation for the decreased voltage in the underweight subjects would be that this is a manifestation of decreased left ventricular mass. This is speculative, as we were unable to find data correlating BMI and left ventricular mass in healthy adults; however, there are data showing patients with anorexia nervosa have decreased left ventricular mass. 13,14 There are also data in Japanese adolescents showing a positive correlation of the R wave amplitude in V5 with the LV mass (correlation coefficient: boys, 0.32; girls, 0.27, pb0.001 for each) with better correlations in subjects with less fat. 15 Future echocardiographic assessment of left ventricular mass in healthy subjects over a range of BMIs would be useful to elucidate the mechanism responsible for these voltage drops.

The J point has recently been an area of intense research given its relationship with SCD and there have been recent proposals to classify various patterns of J point elevation. 16 Given the current interest and research in J point elevation, it is important that other factors affecting the J point are identified and understood. To our knowledge, no prior studies that have specifically evaluated the effect of BMI on the J point, but patients in the ARIC trial without J point elevation had significantly higher BMI 17 suggesting that obesity may decrease J point elevation. T wave abnormalities have also been associated with SCD and the effects of obesity on the T wave are incompletely understood. While there are small studies that have reported T wave flattening of the inferior and inferolateral leads with obesity, 5,7,9 there have been no large studies examining the effects of obesity on the T wave. In the present study, we have demonstrated changes in both the T wave amplitude and J point elevation that appear to mirror the changes noted in R wave amplitude (with the exceptions of the T wave amplitude not differing between the subjects with a normal BMI and the overweight subjects). It is possible the non-linear relationship between these ventricular repolarization metrics and BMI is related to both the effects of increased adipose tissue with obesity and a relationship between BMI and myocardial mass. Echocardiographic correlation would be helpful to clarify these findings. Finally, in this cohort, it is worth reporting that we failed to see several ECG manifestations of adiposity that have been reported by others. Frank et al. 6 has reported increases in the QRS duration and heart rate with obesity which we did not see in this cohort. In regard to the QTc interval, several articles have reported an increased QTc interval with obesity 6,18 and others have reported a decreased QTc interval with weight loss. 19,20 However, there are also data showing that obesity does not affect the QTc interval. 21 Our data support the latter, with only small differences between groups (≤5 ms) without a clear pattern or correlation with adiposity as others have reported. These discrepancies between the established literature and our large study population may be a reflection of the differences in methodology and/or patient populations, but the large size of the cohort and the utilization of uniform computed analysis of waveforms in the current study suggest that prior ECG findings may need to be reconsidered.

Limitations Several limitations to our study should be considered when evaluating the clinical implications of our findings. The demographic data, from which BMI was computed, were obtained by ECG technicians at the time of acquisition. It is possible that these data were obtained by questioning, and not measurement, in a significant portion of our subjects, and the resultant BMIs would be less accurate than if the data were directly measured. However, examination of the frequency histogram our cohort's BMI did not show significant skew and is unlikely to significantly affect the trends reported here. Additionally, we applied very strict inclusion criterion to this cohort. While this minimized pathology in the cohort,

J.M. Nasir et al. / Journal of Electrocardiology 45 (2012) 646–651

undoubtedly some normal subjects were excluded (e.g., subjects with heart rates b60 bpm or R wave axis N90°) and may limit the generalizability of our results.

Conclusion Our study demonstrates the independent effects of BMI on ECG parameters in patients 18 to 35 years of age and also significant interactions between BMI, gender, age, and ethnicity. For accurate assessment of ECGs, we recommend incorporation of the effects of BMI into published normograms. Additionally, further research into the effects of BMI on the ECG is needed and these results should be validated in other large ECG databases. References 1. Mirvis DM, Goldberger AL. Electrocardiography. In: Libby P, Bonow RO, Mann DL, Zipes DP, editors. Braunwald's heart disease. A textbook of cardiovascular medicine. 8th ed. St Louis, MO: Saunders: Elsevier; 2008. 2. Prospective Studies Collaboration. Body-mass index and cause-specific mortality in 900 000 adults: collaborative analyses of 57 prospective studies. Lancet 2009;373:1083. 3. World Health Organization. BMI classification website. (Last updated: 08/04/2012), http://apps.who.int/bmi/index.jsp?introPage=intro_3.html [Accessed 6 April 2012]. 4. Marquette 12SL ECG analysis, physician guide, Revision E, General Electric Healthcare, 2008. 5. Eisenstein J, Sarma R, San Marco M, Selvester RH. The electrocardiogram in obesity. J Electrocardiol 1982;5:115. 6. Frank S, Colliver JA, Frank A. The electrocardiogram in obesity: statistical analysis of 1,029 patients. J Am Coll Cardiol 1986;7:295. 7. Alpert Martin A, Terry Boyd E, Cohen Michael V, et al. The electrocardiogram in morbid obesity. Am J Cardiol 2000;85:908.

651

8. Zack PM, Wiens RD, Kennedy HL. Left-axis deviation and adiposity: the United States Health and Nutrition Examination Survey. Am J Cardiol 1984;53:1129. 9. Alpert MA, Terry BE, Hamm CR, et al. Effect of weight loss on the ECG of normotensive morbidly obese patients. Chest 2001;119:507. 10. Liu T, Fu Z, Korantzopoulos P, Zhang X, Wang S, Li G. Effect of obesity on p-wave parameters in a Chinese population. Ann Noninvasive Electrocardiol 2010;15:259. 11. Kosar F, Aksoy Y, Ari F, Keskin L, Sahin I. P-wave duration and dispersion in obese subjects. Ann Noninvasive Electrocardiol 2008;13:3. 12. Duru M, Seyfeli E, Kuvandik G, Kaya H, Yalcin F. Effect of weight loss on P wave dispersion in obese subjects. Obesity (Silver Spring) 2006; 14:1378. 13. Romano C, Chinali M, Pasanisi F, et al. Reduced hemodynamic load and cardiac hypotrophy in patients with anorexia nervosa. Am J Clin Nutr 2003;77:308. 14. Ulger Z, Gürses D, Ozyurek AR, Arikan C, Levent E, Aydoğdu S. Follow-up of cardiac abnormalities in female adolescents with anorexia nervosa after refeeding. Acta Cardiol 2006;61:43. 15. Sugita S, Takada K, Takada H, Nagashima M, Hayano J. The relationship between R amplitude in lead V5 (RV5) and left ventricular mass in the groups of adolescent subjects classified by body composition. Jpn Circ J 1998;62:893. 16. Antzelevitch C, Yan GX. J wave syndromes. Heart Rhythm 2010; 7:549. 17. Olson KA, Viera AJ, Soliman EZ, Crow RS, Rosamond WD. Longterm prognosis associated with J-point elevation in a large middle-aged biracial cohort: the ARIC study. Eur Heart J 2011;32:3098. 18. Braschi A, Abrignani MG, Francavilla VC, Francavilla G. Novel electrocardiographic parameters of altered repolarization in uncomplicated overweight and obesity. Obesity (Silver Spring) 2011;19:875. 19. Papaioannou A, Michaloudis D, Fraidakis O, et al. Effects of weight loss on QT interval in morbidly obese patients. Obes Surg 2003;13:869. 20. Mshui ME, Saikawa T, Ito K, Hara M, Sakata T. QT interval and QT dispersion before and after diet therapy in patients with simple obesity. Proc Soc Exp Biol Med 1999;220:133. 21. Nomura A, Zareba W, Moss AJ. Obesity does not influence electrocardiographic parameters in coronary patients. Am J Cardiol. 2000;85(1):106, A9.