HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:
Guest Access | Sign In via User Name/Password

This Article Abstract Full Text (PDF) Free Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Article Archive Download to citation manager Citing Articles Citing Articles via ISI Web of Science (9) Citing Articles via Google Scholar Google Scholar Articles by Madias, J. E. Search for Related Content PubMed PubMed Citation Articles by Madias, J. E.

A Comparison of 2-Lead, 6-Lead, and 12-Lead ECGs in Patients With Changing Edematous States^*

Implications for the Employment of Quantitative Electrocardiography in Research and Clinical Applications

^* From the Division of Cardiology, Elmhurst Hospital Center, Elmhurst, NY.

Correspondence to: John E. Madias, MD, Professor of Medicine (Cardiology), Division of Cardiology, Elmhurst Hospital Center, 79–01 Broadway, Elmhurst, NY 11373; e-mail: madiasj{at}nychhc.org

	Abstract

TOP Abstract Introduction Materials and Methods Results Discussion References

 
Objective: Precordial ECGs reveal significant intrasubject variability due to the inexact employment of the recommended V₁-V₆ chest landmarks. Also, as per the Einthoven law, the six limb leads can be derived from leads 1 and 2. The purpose of this study was to evaluate whether the 12-lead ECG could be substituted by ECG sets with a limited number of leads.

Materials and methods: The performance of three ECG systems (ie, the 12-lead ECG, a 6-lead ECG comprising the limb leads, and a 2-lead ECG comprising exclusively leads 1 and 2) was evaluated in data from 28 patients with anasarca (AN), 28 control patients, 10 patients who had undergone hemodialysis, and 3 patients with idiopathic dilated cardiomyopathy.

Results: Linear regression analyses of changes in ECG data with the weight gain of patients with AN and the intercorrelations of the three ECG systems in the various patient subgroups were found to be statistically significant at p = 0.0005 and r values ranging from 0.61 to 0.99, which are suggestive of good/excellent correlations. However, regression analyses of peak weight (PW) gain with changes in the 2-lead ECG (r = 0.43; p = 0.02) and 6-lead ECG (r = 0.48; p = 0.01), and half of PW gain and 12-lead ECG (r = 0.41; p = 0.03), 6-lead ECG (r= 0.18; p = 0.35), and 2-lead ECG (r = 0.43; p = 0.02) revealed poor correlations.

Conclusion: ECG systems, comprising 2 or 6 leads, can be substituted for the 12-lead ECG for certain clinical and research applications (pertaining to the amplitude of QRS complexes), attesting to the inherent redundancy of the information from the 12-lead ECG.

Key Words: anasarca • dilated cardiomyopathy • Einthoven law • electocardiography • hemodialysis • limb ECG leads • precordial ECG leads • quantitative electrocardiography • redundancy in electrocardiography

	Introduction

TOP Abstract Introduction Materials and Methods Results Discussion References

 
An ECG syndrome has been reported¹ that associated the development of peripheral edema of varying etiology with a decrease in the body surface QRS potentials involving all ECG leads. In that work, we employed the sum of QRS complexes of 12-lead ECGs (QRS₁₂) from routine daily tracings, which were obtained by our technicians without the aid of marking of the thoracic wall to ensure intrasubject reproducible recordings from the precordial ECG sites. It is common knowledge that precordial ECGs display considerable intraindividual variation stemming from an alteration in the use of thoracic landmarks for V₁-V₆ recording, and that even a slight change of the recording sites can alter significantly the ECG waveforms.^{2 3} This problem remains without solution, despite the clear treatment of the topic on the appropriate recording of the precordial ECG provided by the ECG textbooks.² In practice, this is of immense importance, since such inconsistencies frequently lead to erroneous diagnoses and a need for a cardiologic consultation.

Since the six standard (bipolar and unipolar) limb ECG leads are not subject to the vagaries of inconsistent recording that is inherent with the six precordial leads, it was hypothesized that the correlation of the weight (W) gain with a reduction in the sum of the QRS complexes from the six standard ECG leads (QRS₆) in our patients with anasarca (AN) would be better than the one we found with the employment of the QRS₁₂ (r = 0.61; p = 0.0005) in our previous article.¹

Most modern ECG machines record only leads 1 and 2, and they calculate in real time the remaining limb leads. Accordingly, it was hypothesized that using merely the sum of QRS complexes of leads 1 and 2 (QRS₂) from a recorded 12-lead ECG will suffice to provide at a glance a quick index of body fluid retention. Such an index could be used at bedside for the evaluation of patients with an edematous state, thus obviating the inconvenience of calculating QRS₆ or QRS₁₂.

Consequently, the database of our previous study¹ in conjunction with QRS₆ and QRS₂, in addition to the corresponding QRS₁₂ used previously, was employed to investigate the two hypotheses cited above. In addition, similar ECG data from another recent study⁴ of three patients with idiopathic dilated cardiomyopathy who were admitted to the hospital with congestive heart failure (CHF) were included.

	Materials and Methods

TOP Abstract Introduction Materials and Methods Results Discussion References

 
Study Subjects
A detailed description of the study populations and methods have been published elsewhere.^{1 4} Briefly for the purposes of the present investigation, data from 69 patients, 28 with AN, 28 control subjects without peripheral edema who had been admitted to the hospital contemporaneously with the patients with AN, 10 patients who had undergone hemodialysis (13 sessions), and 3 patients with peripheral edema due to CHF, were employed. The patients with AN had a variety of illnesses and significant comorbidity, and 21 of the 28 patients had underlying sepsis. Eventually, only four of these patients survived to be discharged from the hospital. Pericardial effusion was excluded by serial echocardiograms.¹ The electrolyte status of these patients (28 patients) was followed daily, and the mean (± SD) K⁺ values on hospital admission and at the time of peak W (PW) were 4.4 ± 0.8 and 4.4 ± 0.8 mEq/L (p = 0.98), respectively, the corrected Ca++ values were 10.0 ± 1.1 and 9.6 ± 0.8 mg/dL (p = 0.12), respectively, the Mg++ values were 1.9 ± 0.4 and 2.1 ± 0.4 mg/dL (p = 0.12), respectively, and the bicarbonate values were 22.4 ± 4.7 and 19.8 ± 6.1 mmol/L (p = 0.04), respectively.

The 28 control subjects had a variety of cardiovascular illnesses necessitating admission to the coronary care unit (CCU), but none of these patients had an acute myocardial infarction. The 10 patients who underwent hemodialysis had a mixture of cardiovascular and pulmonary illnesses, and established or newly diagnosed advanced renal failure for which they required hemodialysis. Finally, the three patients with CHF showed nonischemic dilated cardiomyopathy and intraventricular conduction delay on their ECGs. All patients had been admitted to our CCU in 1999 with a critical illness.

Study Variables
The patients had daily Ws and ECGs recorded and printed routinely with the placement of the limb electrodes on the wrists and ankles, however, for the purposes of this study, the ECGs and Ws from hospital admission and at the point of half-PW gain (HF-W), and PW for the patients with AN, the hospital admission and discharge data for the control subjects, the prehemodialysis and posthemodialysis data for the patients undergoing this procedure, and the data from hospital admission and discharge for the patients with CHF were employed. An ECG unit (PageWriter XLi ECG, model M1700A; Philips; Amsterdam, the Netherlands) was used that acquires data simultaneously and digitally from all 12 ECG leads, instead of acquiring data from a few leads and calculating the rest of them online.⁵ Calibration of the unit was 1.0 mV = 1.0 cm. Measurements of the sum of the highest positive plus the lowest negative deflections of the QRS complex in all 12 ECG leads of all study ECGs were made to the nearest 0.5 mm, employing hard copies and using manual calipers and a magnifying glass (Fig 1 ). For ECGs with atrial fibrillation, the average of measurements of three consecutive heart beats was used. Since the six limb leads can be calculated from leads 1 and 2, as per Einthoven law (leads 1 + 3 = 2), and the aV leads, as per formulas (lead aVR = 1/2[1 + 2], aVL = 1 - 1/2, [2], lead aVF = 2 - 1/2, [1], and lead aVR + aVL + aVF = 0),^{2 6 7} the QRS₂ was employed as a variable, along with the QRS₁₂ and QRS₆. It should be emphasized here that the use of QRS₂ as an ECG system is employed with the full realization that for the derivation of the remaining four limb leads, through the above-described formulas, the values of leads 1 and 2 are treated algebraically in serial online measurements from the entire duration of the QRS complex in ECG machines,⁵ while in the present work the sum of the peak-to-peak (ie, positive-to-negative) amplitude values of the QRS complex of these two leads is implemented. Moreover, it should be understood that the same measurement approach was used for all three ECG systems. QRS₁₂, QRS₆, and QRS₂ were calculated from the day of admission, the HF-W point, and the PW point for each of the 28 patients (Fig 1) . Obviously, the HF-W point was determined after the PW was attained, and it represented a W value that was closest to one half of the PW gained. The mean (± SD) PW gain of the 28 patients with AN was 23.9 ± 14.8 lbs, and the mean HF-W gain was 16.9 ± 10.7 lbs. The corresponding ECG for the day when the HF-W was reached was used in the analysis. Similar calculations were employed for the hospital admission and discharge ECGs of the control subjects, the preprocedural and postprocedural tracings of patients who underwent hemodialysis, and the pretreatment and posttreatment tracings of the patients with CHF. The mean intraobserver variability of QRS₁₂ in 10 random ECGs has been found previously to be 0.41 ± 3.34%.¹ The mean frontal QRS complex axis (in degrees), as calculated by the automated ECG interpretation program,⁵ also was employed as a variable at all study time points in all patients.

View larger version (47K): [in this window] [in a new window] [Download PPT slide]   Figure 1.. ECGs of patient 9.1 Values at the bottom represent QRS12 on hospital admission, at the HF-W gain, and at the PW. Data from the subsequent loss of W, which was noted in a few patients (fourth column), were not considered in this study. Values for the QRS6 for this patient were 41.5 mm, 15.0 mm, and 15.0 mm, respectively, for the admission, HF-W, and PW. Similarly, values for the QRS2 were 13.0 mm, 5.0 mm, and 7.0 mm, respectively, for the admission, HF-W, and PW.

Statistical Analysis
Continuous data are reported as the mean ± SD. The relationship between %W at HF-W and PW, and the corresponding %QRS₁₂, %QRS₆, and %QRS₂ were evaluated by regression analysis, considering the %W as the independent variables and the % in the ECG data as the dependent variables.^{8 9} Intercorrelations of the QRS₁₂, QRS₆, and QRS₂ from hospital admission, at HF-W, and at PW of the patients with AN, the hospital admission and discharge tracings of the control subjects, the preprocedure and postprocedure hemodialysis patients, and the patients with CHF were carried out. In the last group, the data recorded before and after therapy for CHF were analyzed together due to the small number of data points (six data points), since for these patients the primary objective of the study was the correlation of the three ECG systems. The intercorrelations among the three ECG systems are employed here with the understanding that when a subsystem (data from leads 1 and 2, or from the six standard ECG leads) is correlated with a system (data from the 12 ECG leads), part of which is the subsystem, intuitively one should expect r values 0.95.⁸ All in all, 10 sets of data groups (data groups A to J) were considered, with a total of 30 correlations/regressions employed using a statistical software package (SPSS/PC+, version 4.0.1; SPSS; Chicago, IL).⁹ Since in each of the above 10 sets of data groups three correlations/regressions were carried out, a Bonferroni correction for multiple correlations was employed by dividing a p value < 0.05 by 3. Thus, a p value of < 0.017 was taken here as being statistically significant.¹⁰ Data on the frontal QRS axis were compared by paired t test for the patients with AN (at hospital admission, HF-W point, and PW point), for control subjects (hospital admission vs discharge), and for patients undergoing hemodialysis (before vs after the procedure). Since univariate analysis for all group B variables (Table 1 ) showed p values > 0.017 (as per the Bonferroni correction), and since the same was true for the %QRS₂ and %W in data group B (Table 1) , it was thought that it would be inadvisable to enter the dependent variables (ie, ECG systems) with the dependent variable (ie, W) in a multivariate analysis model. Besides, when such models are implemented, the recommendation is that the independent variables should be really such, which is not the case with the ECG systems employed here.

	Results

TOP Abstract Introduction Materials and Methods Results Discussion References

Regression in data group A suggested that W gain in patients with AN, at the PW point, was better associated with changes noted in the QRS₁₂, than with those at the QRS₆, while no association was found with the QRS₂. Regression in data group B suggested that W gain in the patients with AN, at the HF-W point, was not associated with changes noted in any of the three ECG systems.

Intercorrelations at data groups C, D, and E in the three ECG systems at hospital admission, at the point of HF-W and at PW, correspondingly in the patients with AN indicated that the QRS₂ and QRS₆ equally reflect the QRS₁₂, and that the strongest correlation is the QRS₂ with the QRS₆.

Intercorrelations at data groups F and G in the three ECG systems at hospital admission, and hospital discharge correspondingly in the control subjects, mirrored the findings at data groups C, D, and E, and indicated that the QRS₂ and QRS₆ reflect well the QRS₁₂ and that the strongest correlation is the one of the QRS₂ with the QRS₆.

Intercorrelations at data group H and I in the three ECG systems at the preprocedure level, and correspondingly at the postprocedure level in the patients who underwent hemodialysis, indicated that the degree of association among the three ECG systems was strong.

Finally, intercorrelations at data group J for the patients with CHF before and after the management of their condition indicated that the degree of association among the three ECG systems was strong.

The mean QRS axis in the patients with AN was not different between hospital admission and the HF-W point (15.5 ± 79.9° vs 18.3 ± 59.4°, respectively; p = 0.78), between hospital admission and the P-W point (15.5 ± 79.7° vs 24.9 ± 54.7°, respectively; p = 0.24), and between the HF-W point and the PW point (18.3 ± 59.4° vs 24.9 ± 54.7°, respectively; p = 0.23). The mean QRS axis in the control subjects was as follows: on hospital admission and at hospital discharge, 25.8 ± 68.2° vs 5.6 ± 54.6°, respectively (p = 014); and before and after hemodialysis, 59.2 ± 95.2° vs 47.7 ± 83.4°, respectively (p = 0.24). Finally, the QRS axes for the three patients with CHF on hospital admission and at hospital discharge were as follows: -29° vs 22°, respectively; -26° vs –28°, respectively; and -56° vs -54°, respectively.

	Discussion

TOP Abstract Introduction Materials and Methods Results Discussion References

 
The following conclusions can be drawn from this work:

1. The QRS₁₂ reflects better the changes in W in patients with AN than the QRS₂, at the PW point, and all three ECG systems are not predictive of the HF-W gain. This indicates that there is a nonlinear relationship between changes in W and the ECG, and thus it takes significant W gain for ECG detection. When this stage is reached, the precordial leads are essential for such detection. This may be a reflection of the accumulation of fluid in the dorsal/lateral body plane in patients with AN who are kept supine in critical care units (ie, the gravity effect). Accordingly, we have shown previously a better correlation of the %W and the sum of the QRS complexes from leads V₅ and V₆ (reflecting the lateral body plane) [ r = 0.65] than the one with the QRS₁₂ (0.61). Moreover, it is revealing that the %W and the sum of the QRS complexes from leads V₁ and V₂ (reflecting the ventral body plane) [ r = 0.22] did not reach statistical significance (p = 0.26).¹ This overexpression of AN on leads V₅ and V₆ may be due to the sequestration of fluid in the region of the V₅ and V₆ leads, leading to further decrease in body resistance. Moreover, such regional fluid accumulation increases the distance of the heart from V₅ and V₆ recording sites (ie, the Wilson proximity effect).¹¹ Both of these mechanisms would cause attenuation of surface ECG voltage. The almost complete lack of orthogonality inherent in the limb leads, and the fact that their vectors are expressed in one plane, previously has been commented on.^{6 7}
   
2. The correlations of the QRS₆ and QRS₂ with the QRS₁₂ are such that the first two can be considered reflective of the latter. In the control subjects at hospital discharge and in the hemodialysis patients (Table 1 , data groups G to I), the QRS₆ and QRS₂ reflected better the QRS₁₂ than in the patients with AN. This should be expected since in the patients with AN the fluid accumulates especially in places other than the frontal plane, and thus the correlations among ECG systems (Table 1 , data groups C to E) are attenuated.
   
3. The similarity in the correlations of the QRS₂ and QRS₆ with the QRS₁₂, and the correlation with r = 0.86 to 0.93, respectively, between the QRS₂ and the QRS₆ are undoubtedly due to the redundancy in the ECG information contained in the frontal ECG leads, an issue that was well-researched and discussed by Rautaharju and colleagues.^{6 7} This is not surprising, since by design such correlations would be expected to be expressed by r values of approximately 1. If such values were not reached in our calculations, it was perhaps due to the error in our crude (to the nearest 50 µV) measurements (the machine captures waveforms at 5-µV resolution),⁵ and to the fact that ECG machines electronically calculate on-line the nonmeasured ECG leads from data sampled many hundreds of times per second, which is in contrast to our measurements at points of the QRS complexes (ie, peak to peak). Considering that the limb leads reflect the cardiac electrical field in the frontal plane, while the 12-lead ECG reflects the field in 3-D space, it is puzzling that such good intercorrelations among the three ECG systems are found. Perhaps this is due to an expected good match when a part is correlated with its whole,⁸ or the probability that whatever affects the generation/conduction of ECG potentials has a proportional effect in all three planes.
   

Electrocardiography is employed in qualitative (pattern-based) and quantitative formats, for the diagnosis and follow-up of patients. Measuring the amplitude of various parts of the ECG waveform provides data suitable for comparison and statistical analysis. Measurements of the ST segment have been employed in exercise testing and in the CCU for the purposes of monitoring myocardial ischemia. Similar treatment of the QRS complexes can be used for the purposes of diagnosing and monitoring patients who have developed or are recovering from edematous states of various etiologies.^{1 4 12}

Daily experience particularly in critical care units reveals the vagaries of obtaining patients’ Ws using sling scales. The calculation of QRS₁₂ may be useful in monitoring changes in extent of peripheral edema. Although this also could be time-consuming, it could be probably handled electronically by automated ECG interpretation programs.⁵ Another application of the ECG/W correlations could be in setting up changing voltage-dependent ECG criteria for various diagnoses (ie, left ventricular hypertrophy) in patients with extreme fluid overload states. Thus, for patients with AN or CHF, or for those undergoing hemodialysis, the diagnosis of left ventricular hypertrophy should be made with reference to the stage (ie, compensated and uncompensated) of their edematous state.

The reliability of the QRS₁₂ in detecting fluid accumulation or loss may improve when ECGs are recorded from fixed chest wall landmarks, ensuring comparable precordial ECG recordings. In our previous studies,^{1 4} we used routine daily ECGs recorded by our technicians, without the above provision. It appears from the present study that for clinical and research applications focusing on the QRS amplitude (and excluding patients with AN), one can employ measurements from the 2-lead ECG and 6-lead ECG instead of the 12-lead ECG. This was highlighted by the data from the three patients treated for CHF (Table 1 , data group J),⁴ for whom the correlations among the three ECG systems were very strong. However, this might have been a chance finding, since only three patients were studied, and data from hospital admission and discharge were lumped together. A large study with patients with CHF using data analysis that considers separately the ECG data from hospital admission and discharge is needed. Nevertheless, recent data from monitoring patients treated for CHF in the CCU (unpublished data) have suggested that employment of the QRS₂ visually calculated at the time of routine rounds, is a useful index of the effectiveness of diuresis, and such daily values correlate well with daily Ws. Also, the usefulness of employing sums of limited leads or all 12 ECG leads in monitoring patients with edematous states of various degree, distribution, and etiology (eg, AN, CHF, and chronic renal failure) needs to be investigated further, taking great care in measuring the QRS complexes, obtaining the patients’ W, and maintaining fluid intake/output records.

	Footnotes

 
Abbreviations: AN = anasarca; CHF = congestive heart failure; CCU = coronary care unit; % = percentage change; HF-W = half peak weight gain; PW = peak weight; QRS₂ = sum of QRS complexes of limb leads 1 and 2; QRS₆ = sum of the QRS complexes from the 6-limb leads; QRS₁₂ = sum of QRS complexes of 12-lead ECGs; W = weight

Received for publication October 24, 2002. Accepted for publication May 15, 2003.

	References

TOP Abstract Introduction Materials and Methods Results Discussion References

 

1. Madias, JE, Bazaz, R, Agarwal, H, et al (2001) Anasarca-mediated attenuation of the amplitude of ECG complexes: a description of a heretofore unrecognized phenomenon. J Am Coll Cardiol 38,756-764[Abstract/Free Full Text]
2. Wagner, GS Marriott’s practical electrocardiography 9th ed. 1994,23, 25, 26 Williams & Wilkins. Baltimore, MD:
3. Madias, JE, Hood, WB, Jr Precordial ST-segment mapping 3: stability of maps in the early phase of acute myocardial infarction. Am Heart J 1977;93,603-609[CrossRef][ISI][Medline]
4. Madias, JE, Agarwal, H, Win, M, et al Effect of weight loss in congestive heart failure from idiopathic dilated cardiomyopathy on electrocardiographic QRS voltage. Am J Cardiol 2002;89,86-88[CrossRef][ISI][Medline]
5. Hewlett-Packard interpretive cardiograph: physician’s guide. 4th ed. Palo Alto, CA: Hewlett-Packard, 1994; 2–2, 3–3, 7–4; Hewlett-Packard part No. M1700–92908
6. Rautaharju, PM, Warren, J, Seale, D, et al Exploitation of the redundancy of the conventional limb lead electrocardiograms for prolongation of the record length. J Electrocardiol 1981;14,39-41[CrossRef][ISI][Medline]
7. Rautaharju, PM The inappropriateness of the commonly used augmentation and lead recording sequence for ECG analysis. Pract Cardiol 1982;8,120-139
8. Colton, T Statistics in medicine. 1974,17, 33, 120, 131, 189, 211 Little, Brown and Company. Boston, MA:
9. Norusis, MJ The SPSS guide to data analysis for PDD/PC+ 2nd ed. 1991,244, 362, 374, 390 SPSS. Chicago IL:
10. Curtis, F, Schultz, P Multiple correlations and Bonferroni’s correction. Biol Psychiatry 1998;44,775-777[CrossRef][ISI][Medline]
11. Wilson, FN, Johnston, FD, Rosenbaum, F, et al The precordial electrogram. Am Heart J 1944;27,1953
12. Madias, JE, Attanti, S, Narayan, V Relationship among electrocardiographic potential amplitude, weight, and resistance/reactance/impedance in a patient with peripheral edema treated for congestive heart failure. J Electrocardiol 2003;36,167-171[ISI][Medline]

This Article Abstract Full Text (PDF) Free Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Article Archive Download to citation manager Citing Articles Citing Articles via ISI Web of Science (9) Citing Articles via Google Scholar Google Scholar Articles by Madias, J. E. Search for Related Content PubMed PubMed Citation Articles by Madias, J. E.