- Email: [email protected]
Electrocardiographic diagnosis of left ventricular hypertrophy: The effect of left ventricular wall thickness, size, and mass on the specific criteria for left ventricular hypertrophy
Electrocardiographic diagnosis of left ventricular hypertrophy: The effect of left ventricular wall thickness, size, and mass on the specific criteria for left ventricular hypertrophy Navin Budhwani, MD, Sanjeev Patel, MD, and Edward M. Dwyer, Jr, MD Newark, NJ
Background The purpose of our study was to determine the relative importance and effect of an increased left ventricle wall thickness, left ventricular diastolic diameter, and left ventricular mass (LVM) on the performance of the 4 major electrocardiogram (ECG) criteria of left ventricular hypertrophy (LVH) and to determine how these findings could be incorporated into the routine ECG interpretation of LVH. Methods
The ECG criteria of LVH that we chose to examine were voltage, repolarization abnormalities, left atrial abnormality, and ventricular conduction time. We analyzed data from 608 consecutive patients with left ventricular wall thickness of N13 mm on the echocardiogram and with a concurrent ECG. We arbitrarily divided patients into 3 groups (groups I-III) according to the calculated LVM. Group I had an LVM of b400 g; group II had an LVM from 400 to 600 g, and group III had an LVM of N600 g. We evaluated the effect of increasing LVM, wall thickness, and ventricular diameter on the performance of the 4 ECG criteria at different severity of thickness, diameter, and mass.
Results An increase in the echocardiogram-derived LVM had significant effect on all 4 ECG criteria. As LVM progressively increased from groups I to III, the frequency of voltage criteria for LVH increased from 52% to 83%; left atrial abnormality rose from 46% to 68%; ST-T wave changes increases from 55% to 95%, and QRS prolongation significantly increased from 42% to 70%. Conclusion Increased wall thickness and ventricular diameter failed to correlate with the overall ECG score or significantly influence the frequency of any of the 4 ECG criteria for LVH in patients when LVM was held relatively constant. We also demonstrated that an increasing number of criteria on the ECG are associated with a greater mean LVM. (Am Heart J 2005;149:709-14.) The electrocardiographic diagnosis of hypertrophy of the left ventricle (LVH) has been a source of active investigation since the initial observations by Lewis1 in 1914. The original reference standards, used in the development of specific electrocardiographic criteria for the presence of LVH, were either by autopsy measurements2 or clinical assessment.3 As a reference standard, autopsy data allowed for accurate determination of ventricular weight but had an inherent selection bias, whereas a diagnosis of LVH made by clinical criteria was widely applicable but was frequently inaccurate and could not provide a quantitative determination of LVH. Echocardiography has provided an excellent technique to measure left ventricular mass (LVM).4 This technique From the Department of Medicine, Division of Cardiovascular Diseases, New Jersey Medical School/UMDNJ, Newark, NJ. Submitted April 11, 2004; accepted July 6, 2004. Reprint requests: Edward M. Dwyer, MD, Division of Cardiovascular Diseases, NJ Medical School, MSB I-536, 185 S Orange Ave, Newark, NJ 07103. 0002-8703/$ - see front matter n 2005, Elsevier Inc. All rights reserved. doi:10.1016/j.ahj.2004.07.040
provides an excellent reference standard by which we can contemporaneously assess the validity of specific electrocardiographic criteria used for the diagnosis of LVH. Although the echocardiogram is extremely accurate in its ability to diagnose LVM, the electrocardiogram (ECG) is more widely used because of its easy performance and low cost. The clinical diagnosis of LVH, in disorders such as hypertension and valvular disease carries important prognostic implications.5,6 For these reasons, it is important to understand both the capabilities and limitations of the ECG in the diagnosis of LVH. Many excellent studies have been published which have attempted to refine or confirm the value of some of the ECG criteria widely used in the diagnosis of LVH.7-9 The electrocardiographic diagnosis of LVH has been shown to have reasonable specificity but lacks sensitivity. The degree of sensitivity described by many studies lies between 40% and 60%.10 Although LVH, at the most basic level, is defined as an enlargement of cardiac myocytes, the clinical representations of LVH that are most common are increased wall thickness or ventricular mass. An increased LVM
American Heart Journal April 2005
710 Budhwani, Patel, and Dwyer
can be characterized by different geometric forms. There is an increased wall thickness and normal left ventricular size at one end of the spectrum, whereas at the other end of the spectrum, there is increased left ventricular size and normal wall thickness. The performance of the several commonly used ECG criteria may vary as LVM increases. Each of the ECG criteria may also, individually, perform differently in the presence of concentric hypertrophy than they do in the presence of eccentric hypertrophy. There is limited understanding of the contribution and importance of left ventricular thickness, ventricular size, or overall LVM on the value and performance of the individual criteria such as ECG voltage, repolarization abnormalities, and alterations in atrial and ventricular conduction. The purpose of our study is to determine the relative importance and effect of an increased left ventricular wall thickness, left ventricular diastolic diameter, and LVM on the performance on each of the 4 major ECG criteria for LVH and to determine how the findings can be incorporated into the use of the routine ECG for the interpretation of LVH.
Methods Study group After exclusions, we analyzed data from 608 consecutive patients for study (obtained N2 years) from the echocardio graphic and electrocardiographic databases at UMDNJ University Hospital, who had a septal and posterior wall thickness z14 mm on the echocardiogram. We retrieved complete echocardiographic data along with the ECG for each patient. The ECG was obtained within a mean of 7 days of the echocardiogram. We excluded patients with inadequate technical quality on the echocardiogram and ECG evidence of a Q wave myocardial infarct, atrial fibrillation, complete left or right bundle-branch block, and a paced rhythm. Standard ECG criteria were adopted.11 Two physicians, who had no knowledge of the echocardiographic findings, separately performed an analysis of the ECG. All disagreements were resolved between the 2 readers.
Electrocardiographic measurements Our technical staff recorded a standard 12-lead ECG, on Marquette computerized recorders, at 25 mm/s and 1 mV/cm standardization with equipment having frequency response characteristics in accordance with American Heart Association recommendations.12 The tracings were interpreted for the presence or absence of the following 4 criteria for LVH: (1) (2) (3) (4)
Increased voltage, ST depression and/or T-wave inversion, Left atrial enlargement, and Prolonged QRS conduction time.
We calculated a point score according to the system proposed by Romhilt and Estes.9 In this system, each of the
Table I. Echocardiographic measurements in 608 patients with increasing LVM
Echocardiogram measurements Mean LV mass (g) Mean LV wall thickness (mm) Mean LV diastolic diameter (mm)
Group I
Group II
Group III
(b bb 400 g)
(400-600 g)
(N NN 600 g)
n = 290
n = 255
n = 63
318 15
480 17
718 18.5
41
51
60
LV, Left ventricle.
criteria is graded as 3 points except QRS duration which is awarded 1 point. The total score ranges from 0 to 10. QRS amplitude was measured to the nearest microvolt on a digitized ECG. The presence of increased voltage was calculated using the modified Cornell criteria. These more recent criteria are (for men: S in V3 plus R in aVL N2.8 mV [28 mm], and for women: S in V3 + R in aVL N2.0 mV [20 mm]).13 Left atrial abnormality was coded as present if the P-wave inversion in lead V1 was N1 mm in length and N1 mV in amplitude. ST depression N0.1 mV and/or T-wave inversion of any degree were considered present if found in any lead except aVR, and V1 to V3. QRS duration was measured to the nearest 10 milliseconds and was considered abnormal when the duration was z100 milliseconds.9
Echocardiographic measurements Comprehensive 2-dimensional and Doppler echocardiography was performed on all patients. Left ventricular dimensions (intraventricular septal thickness, posterior wall thickness, and left ventricular end-diastolic diameter) were measured at end of diastole with M-mode by using leading-edge–to–leadingedge convention. Left ventricular mass (g) was calculated using the Penn cube convention with the following formula: LVM = 1.04[(LVEDD + PW + IVS)3 (LVEDD)3] 13.6, where LVEDD is the left ventricular end diastolic diameter; PW is the posterior wall thickness, and IVS is the intraventricular septal thickness.
Data analysis Group subsets. We arbitrarily divided patients into 3 groups (groups I-III) according to the calculated LVM. We intended these groups to reflect mild, moderate, and severe hypertrophy. The groups represent approximate 200-g increments in LVM. All patients in group I had an LVM b400g; group II had an LVM from 400 to 600 g, and group III had an LVM N600 g. Left ventricular wall thickness. To evaluate the contribution of wall thickness to the performance of the 4 ECG criteria, we compared the frequency of each of the ECG criteria between 2 extreme groups (A and B) of wall thickness. In group A, all patients had both septal and posterior wall thickness between 14 and 15 mm. In group B, all patients had at least one wall thickness z17 mm. To eliminate the influence
American Heart Journal Volume 149, Number 4
Budhwani, Patel, and Dwyer 711
% Patients
Figure 1
Table II. Relationship between number of criteria for LVH present on the ECG and LVM
100 90 80 70 60 50 40 30 20 10 0 < 400 g (n = 290)
Romhilt and Estes ECG score
% ST and/or T abn % LAE % Voltage abn % QRS delay
0 1-3 4-6 7-9 10
Mean LV mass (g) 346 365 410 452 501
P value*
(0 vs 1-3) (1-3 vs 4-6) (4-6 vs 7-9) (7-9 vs 10)
= V V V
.11 .001 .001 .003
*t Test.
4-600 g (n = 255)
>600 g (n = 63)
LV Mass With increasing LVM, there was a significant increase in the number of patients demonstrating the presence of each of the 4 criteria on the ECG ( P b .001 for LAE and P b .0001 for the other 3 criteria). LV, left ventricle; LAE, left atrial abnormality; abn, abnormality.
of LVM, we restricted both groups to patients with an LVM b400 g so that both groups had a similarly low LVM. A m2 test was used to compare groups with respect to each of the ECG criteria frequency. In this subgroup of 147 patients, we also correlated the mean wall thickness with the overall ECG score. A P value of b.05 was considered significant. Left ventricular diastolic diameter. In a similar manner to the analysis of wall thickness, 2 groups of markedly different left ventricular diameters were also created from the subset with an LVM b400 g. Group C was selected from patients with a left ventricular diameter of b50 mm and group D from patients with a left ventricular diameter of N60 mm. A m2 test was used to compare groups with respect to each of the ECG criteria frequency. In this subgroup of 331 patients, we also correlated the left ventricular diameter with the overall ECG score. A P value of b.05 was considered significant. Left ventricular mass. To evaluate the effect of an increasing LVM on the 4 ECG criteria performance, we compared the frequency of each of the 4 ECG criteria, described above, as they occurred in each of the 3 LVM groups (see group subsets described above). A m2 test was used to compare the presence of each of the ECG criteria with respect to each of LVM groups. In the entire study cohort of 611 patients, we also correlated the continuous variables of LVM with the corresponding ECG score. A P value of b.05 was considered significant.
Results Patient characteristics We obtained ECGs and echocardiograms on 608 patients. The sex distribution of this cohort was 47% men and 53% women. The mean age of the study group was 56 years. The patients with the lowest LVM (b400 g)
had a mean age of 58 years. With increasing LVM, the mean age declined such that, at the highest level of LVM (N600 g), the mean age was 52 years ( P b .01).
Echocardiographic findings In the entire patient cohort, our measurements of the left ventricular septal and posterior wall thickness ranged between 13 and 30 mm with a mean value of 16 mm. All patients exhibited an abnormal wall thickness because this was the primary requirement for entry into the study. The left ventricular end-diastolic diameter ranged between 25 and 74 mm. with a mean value of 49 mm. The LVM ranged between 134 and 1006 g, with a mean value of 460 g. The relationship between left ventricular mass, wall thickness, and diastolic diameter Our patient cohort consisted of 3 groups, characterized by increasing LVM (see Table I). Group I (290 patients) had a mean LVM of 318 g; group II (255 patients) had a mean LVM of 480 g, and group III (63 patients) had a mean LVM of 718 g. From groups I to II and III (with increasing LVM), the mean left ventricular wall thickness progressively increased from 15 to 17 mm and 18.5 mm, respectively. The mean left ventricular diastolic diameter also progressively increased from 41 to 51 mm and then to 60 mm in those 3 groups with increasing LVM. Effect of increasing LVM on individual electrocardiogram criteria An increase in the echocardiogram-derived LVM had a significant effect on all 4 ECG criteria (see Figure 1). The differences in the frequency of each of the 4 criteria were statistically significant at a P value of b.001 or greater. Voltage criteria. The frequency of the occurrence of voltage criteria for LVH increased from 52% to 66% to 83% as LVM progressively increased from groups I to III ( P b .0001). Left atrial abnormality. As LVM increased through the 3 groups, the frequency of left atrial abnormality also rose from 46% to 59% and then to 68% ( P b .001).
American Heart Journal April 2005
712 Budhwani, Patel, and Dwyer
Figure 2
Figure 3 GROUP A
GROUP B
80
70 62
68
% Patients
44 42
45
40
42
Voltage LAE
39 29
30
ST - T abn QRS Delay
% Patients
60
50
GROUP D
70
58
60
GROUP C
50
65*
55 49
45 40
40
43
46
30
Voltage LAE ST - T abn QRS delay
20
20
10
10 0
0 Wall Thickness 14 mm LVM = 313 g (n = 109)
Wall Thickness = 17-27 mm LVM = 345 g (n = 38)
LV diameter 50 mm LVM = 310 g (n = 263)
LV diameter 60 mm LVM = 276 g (n = 68)
* P ≤ .001
There is no significant effect of increasing left ventricular wall thickness on each of the 4 ECG criteria for LVH in the presence of similar LVM in each group.
There is no significant effect of increasing left ventricular wall diameter on 3 of the 4 ECG criteria for LVH in the presence of similar LVM in each group. Asterisk indicates that the effect of increasing left ventricular wall diameter significantly increased the frequency of a QRS delay ( P b .001).
ST-T changes. In a similar fashion, the ST-T wave changes also were more frequent with increasing LVM. We found a change from 55% in group I up to 95% in group III ( P b .0001). QRS duration. Finally, the presence of QRS prolongation was also significantly increased from 42% in group I up to 70% in group III ( P b .0001).
thickness, per se, did not influence ECG results. The correlation between wall thickness and the ECG score in this subgroup was poor (r = 0.07) and not statistically significant ( P = .34).
Relationship between electrocardiogram score and left ventricular mass When a zero score was present from the ECG, the mean LVM was 346 g (see Table II). When the greatest score (10) was present on the ECG, the mean LVM was 501 g. With each graduation above the 0 -to -3 ECG score, the difference was significant at a P value of b.001, b.001, and b.003, respectively. In addition, in all 608 patients, we demonstrated a significant correlation (r = 0.35) of the Romhilt-Estes ECG scores with the LVMs ( P b .0001). Effect of left ventricular wall thickness on electrocardiogram criteria From our blowQ LVM patients (b400 g), we created 2 subgroups. Group A included 109 patients with a septal and posterior wall thickness between 14 and 15 mm, and group B had 38 patients with 1 wall thickness N17 mm (see Figure 2). The 2 subgroups had a similar and statistically not different mean LVM (313 vs 345 g). In the analysis of frequencies of each of the 4 ECG criteria between the group with the least increase in wall thickness and the group with the greatest increase in wall thickness, we observed that there were no significant differences in the frequency of any of the 4 ECG criteria for LVH between the 2 groups. Wall
Effect of left ventricular diastolic diameter on electrocardiogram criteria In a similar manner to the description above, we selected from the blow Q LVM (b400 g) patients and created 2 groups (see Figure 3). Group C included 263 patients who had a left ventricular end-diastolic diameter of b50 mm, whereas the 68 patients in group D patients all had a diameter of N60 mm. These 2 subgroups had a similar mean LVM (310 vs 276 g) that was not statistically different. In our analysis, there were no significant differences between the 2 groups, in the frequency of any of the 4 ECG criteria except for the QRS prolongation. The frequency of QRS prolongation (64% vs 40%) was significantly ( P b .001) greater in the group with increased diameter despite a somewhat lower mean LVM. Except for QRS duration, left ventricular diameter did not influence ECG results. The correlation between left ventricular diameter and the ECG score in this subgroup was also poor (r = 0.07) and not statistically significant ( P = .23).
Discussion Despite the unquestioned demonstration that the M-mode and 2-dimensional echocardiogram provide us with a satisfactory bgold standardQ for the determi-
American Heart Journal Volume 149, Number 4
nation of LVH, there have been continuous efforts to improve both the performance of the ECG and our understanding of its limitations. The ECG is relatively inexpensive and easy to perform, insuring its role as an excellent screening tool for the diagnosis of LVH. In this vein, we carried out this analysis to improve the understanding of which anatomic features of the remodeled left ventricle influence the ECG criteria that are used in the diagnosis of LVH. Anatomically, LVH is characterized by an increase in muscle mass, and mass is a function of both left ventricular chamber size and left ventricular wall thickness. The relative importance of wall thickness and ventricular diameter, vis-a`-vis LVM, in the determination of the ECG abnormalities has not been previously examined. This study was designed to examine the relation between these 3 anatomic characteristics of LVH and the 4 ECG criteria used for the diagnosis of LVH. The ECG criteria we used were those in common usage and included voltage, left atrial abnormality, prolonged QRS conduction time, and repolarization abnormalities. Echocardiography is the procedure of choice for the quantitation of left ventricular wall thickness, diameter, and mass.4 We used this technique as our reference standard for these measurements. Our patient population was selected based on the presence of 1 definitive aspect of LVH on the echocardiogram, that is, an abnormal left ventricular wall thickness. As a result of this inclusion criterion, we anticipated that all patients would have an increased LVM. In fact, our cohort demonstrated a wide range of abnormal echocardiographic changes such as a left ventricular wall thickness between 13 and 30 mm, enddiastolic dimensions from 25 to 74 mm, and an LVM between 134 and 1006 g. Although we lacked the data (height and weight) to normalize the LVM, the purpose of the study was to examine the effect of the mass on the ECG rather than to establish normal versus abnormal, which is the primary value of normalizing such data. The interpretation of LVH on the ECG has usually been approached as a dichotomous interpretation (either present or not present). Of all the ECG criteria for LVH, increased voltage has always been a necessity for the consideration of a diagnosis of LVH. It is the only ECG abnormality that has been uniquely linked with LVH. As a result, the sensitivity for the diagnosis of LVH, by the ECG, rests with the presence of increased voltage. The sensitivity for detection of LVH in our patient population, based on the overall frequency of abnormal voltage, was 61%. This percentage compares favorably with previously published studies.10,14 A number of efforts have been published that attempt to improve sensitivity through differing methods of estimating QRS voltage.14 -16 Also, the effect of external factors on voltage, such as body habitus, has also been examined.17
Budhwani, Patel, and Dwyer 713
The presence or increased frequency of the other ECG criteria for LVH, beyond voltage, usually has the effect of improving the specificity of the diagnosis when combined with increased voltage. Therefore, any factors that influence the frequency of the ECG criteria will logically also affect either sensitivity or specificity. It is in this light that we analyzed the impact that anatomic features of LVH would have on the ECG criteria. There is minimal information in the literature on the independent effect that left ventricular wall thickness, size, or mass has on the frequency of the ECG criteria. This in turn could influence the sensitivity and specificity for the ECG diagnosis of LVH.18,19 Unexpectedly, our analysis demonstrated that the extent of increase in LVM was the only anatomic abnormality that influenced the frequency of all 4 ECG abnormalities and correlated with the total ECG score. In contrast, neither wall thickness or ventricular diameter significantly affected the frequency of the appearance of the ECG criteria. We found that, as LVM increased, the frequency of each of the 4 ECG criteria correspondingly increased. By example, at the lowest level of LVM (b400 g), abnormal voltage occurred in only 52%, whereas at the highest tertile of LVM (N 600 g), the frequency of abnormal voltage was 83%. Therefore, the sensitivity of detecting increased LVM improved proportionally to the degree of increased mass. In the same manner and to the same degree, we found that an increased occurrence of left atrial abnormality, QRS conduction time, and ST-T repolarization abnormalities was associated with an increasing LVM. Therefore, one would anticipate an improved specificity, as well, as the mean LVM increased. In the highest tertile of LVM, the frequency of the appearance of left atrial abnormality (68%), QRS conduction time (70%), and ST-T changes (95%) was significantly greater than in the patients in the lowest tertile of LVM, who had mean rates of 46%, 42%, and 55%, respectively. In support of the above observations, we also observed that 19% had a normal ECG in the lowest tertile group, whereas a normal ECG was not present in any patient in the upper tertile group for LVM. These findings can explain the variability in the sensitivity and specificity of the ECG in the detection of increased LVM in previous studies. Based on our data, we would anticipate that a population under study with a large increase in LVM (renal insufficiency, idiopathic cardiomyopathies, and severe aortic valvular disease) would be expected to show both a higher sensitivity and specificity than a population consisting of disorders associated with lower LVM changes. We suggest that expectations, for the performance of the ECG in the detection of increased LVM (LVH), should be moderated accordingly. One might anticipate that clinical disorders, commonly associated with large increases in LVM, will reflect LVH on the ECG, on average, in 70% to 80% of the
American Heart Journal April 2005
714 Budhwani, Patel, and Dwyer
patients. On the other hand, patients with the lowest increases in LVM will often go undetected. Detection of small increases in LVM would be best approached with echocardiographic studies. In our study, we also examined the effect that increasing severity of wall thickness and ventricular diameter has on the frequency of the 4 ECG criteria when LVM is held relatively constant. We found, in both analyses, that neither the increase in wall thickness or ventricular diameter correlated with the ECG score nor resulted in a significant change in the frequency of any of the 4 ECG criteria. There was 1 exception to that general conclusion. We did observe that a prolonged QRS conduction time occurred significantly more often with increased left ventricular size, although LVM was similar. In addition to our observations of the effect that LVM has on the ECG, we also demonstrated that as the number of criteria increases on the ECG, the LVM significantly increases. For example, when voltage criterion is the only criterion present on the ECG, the average LVM is 365 g, whereas the presence of all 4 criteria on the ECG is associated with a 37% greater mean LVM at 501 g. From this, we conclude the ECG indicators of LVH are reasonable representations of the severity of LVM. Because of the wide range of LVM within each ECG grouping, these findings are only reflective of the group and cannot be applied with any assurance to the individual patient.
6.
7.
8.
9.
10.
11.
12.
13.
14.
References 1. Lewis T. Observations upon ventricular hypertrophy with special reference to preponderance of one or other chamber. Heart 1914;5:367. 2. Scott RC, Seiwert VJ, Simon DL, et al. Left ventricular hypertrophy — a study of the accuracy of current electrocardiographic criteria when compared with autopsy findings in one hundred cases. Circulation 1955;11:89 - 96. 3. Rosenfeld L, Goodrich C, Kassenbaum G, et al. The electrocardiographic recognition of left ventricular hypertrophy. Am Heart J 1962;63:731 - 42. 4. Devereux RB, Casale PN, Eisenberg RR, et al. Electrocardiographic detection of left ventricular hypertrophy using echocardiographic determination of left ventricular mass as the reference standard. J Am Coll Cardiol 1984;3:82 - 7. ¨ rnlo¨v J, et al. Echocardiographic and 5. Sundstro¨m J, Lind L, A electrocardiographic diagnoses of left ventricular hypertrophy
15.
16.
17.
18.
19.
predict mortality independently of each other in a population of elderly men. Circulation 2001;103:2346 - 51. Levy D, Garrison RJ, Savage DD, et al. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Heart Study. N Engl J Med 1990;322: 1561 - 6. Casale PN, Devereux RB, Kligfield P, et al. Electrocardiographic detection of left ventricular hypertrophy: development and prospective validation of improved criteria. J Am Coll Cardiol 1985; 6:572 - 80. Romhilt DW, Bove KE, Norris RJ, et al. A critical appraisal of the electrocardiographic criteria for the diagnosis of left ventricular hypertrophy. Circulation 1969;40:185 - 95. Romhilt DW, Estes Jr EH. A point score system for the electrocardiographic diagnosis of left ventricular hypertrophy. Am Heart J 1968;75:752 - 8. Levy D, Labib SB, Anderson KM, et al. Determinants of sensitivity and specificity of electrocardiographic criteria for left ventricular hypertrophy. Circulation 1990;81:815 - 20. Sgarbossa EB, Wagner G. Electrocardiography. In: Topol E J, editor. Textbook of cardiovascular medicine. Philadelphia: Lippincott-Raven Publishers Philadelphia. p.1545-89. Pipberger HV, Arzbaecher RC, Berson AS, et al. Recommendations for standardization of leads and specifications for instruments in electrocardiography and vectorcardiography. Report of the committee on electrocardiography, American Heart Association. Circulation 1975;AHA(Suppl):I -1 - 31. Casale PN, Devereux RB, Kligfield P, et al. Electrocardiographic detection of left ventricular hypertrophy: development and prospective validation of improved criteria. J Am Coll Cardiol 1985;6: 572 - 580. Murphy ML, Thenabadu PN, de Spyza N, et al. Sensitivity of electrocardiographic criteria for left ventricular hypertrophy according to type of cardiac disease. Am J Cardiol 1985;55:545 - 53. Sokolow M, Lyon TP. Ventricular complex in left ventricular hypertrophy as obtained by unipolar precordial and limb leads. Am Heart J 1949;37:161 - 86. Casale PN, Devereux RB, Alonso DR, et al. Improved sexspecific criteria of left ventricular hypertrophy for clinical and computer interpretation of electrocardiograms: validation with autopsy findings. Circulation 1987;75:565 - 7. Okin PM, Roman MJ, Devereux RB, et al. Electrocardiographic identification of left ventricular hypertrophy: test performance in relation to definition of hypertrophy and presence of obesity. J Am Coll Cardiol 1996;27:124 - 31. Verdecchia P, Dovellini EV, Gorini M, et al. Comparison of electrocardiographic criteria for diagnosis of left ventricular hypertrophy in hypertension: the MAVI study. Ital Heart J 2000;1:207 - 15. Schillaci G, Verdecchia P, Borgioni C, et al. Improved electrocardiographic diagnosis of left ventricular hypertrophy. Am J Cardiol 1994;74:714 - 9.
Background The purpose of our study was to determine the relative importance and effect of an increased left ventricle wall thickness, left ventricular diastolic diameter, and left ventricular mass (LVM) on the performance of the 4 major electrocardiogram (ECG) criteria of left ventricular hypertrophy (LVH) and to determine how these findings could be incorporated into the routine ECG interpretation of LVH. Methods
The ECG criteria of LVH that we chose to examine were voltage, repolarization abnormalities, left atrial abnormality, and ventricular conduction time. We analyzed data from 608 consecutive patients with left ventricular wall thickness of N13 mm on the echocardiogram and with a concurrent ECG. We arbitrarily divided patients into 3 groups (groups I-III) according to the calculated LVM. Group I had an LVM of b400 g; group II had an LVM from 400 to 600 g, and group III had an LVM of N600 g. We evaluated the effect of increasing LVM, wall thickness, and ventricular diameter on the performance of the 4 ECG criteria at different severity of thickness, diameter, and mass.
Results An increase in the echocardiogram-derived LVM had significant effect on all 4 ECG criteria. As LVM progressively increased from groups I to III, the frequency of voltage criteria for LVH increased from 52% to 83%; left atrial abnormality rose from 46% to 68%; ST-T wave changes increases from 55% to 95%, and QRS prolongation significantly increased from 42% to 70%. Conclusion Increased wall thickness and ventricular diameter failed to correlate with the overall ECG score or significantly influence the frequency of any of the 4 ECG criteria for LVH in patients when LVM was held relatively constant. We also demonstrated that an increasing number of criteria on the ECG are associated with a greater mean LVM. (Am Heart J 2005;149:709-14.) The electrocardiographic diagnosis of hypertrophy of the left ventricle (LVH) has been a source of active investigation since the initial observations by Lewis1 in 1914. The original reference standards, used in the development of specific electrocardiographic criteria for the presence of LVH, were either by autopsy measurements2 or clinical assessment.3 As a reference standard, autopsy data allowed for accurate determination of ventricular weight but had an inherent selection bias, whereas a diagnosis of LVH made by clinical criteria was widely applicable but was frequently inaccurate and could not provide a quantitative determination of LVH. Echocardiography has provided an excellent technique to measure left ventricular mass (LVM).4 This technique From the Department of Medicine, Division of Cardiovascular Diseases, New Jersey Medical School/UMDNJ, Newark, NJ. Submitted April 11, 2004; accepted July 6, 2004. Reprint requests: Edward M. Dwyer, MD, Division of Cardiovascular Diseases, NJ Medical School, MSB I-536, 185 S Orange Ave, Newark, NJ 07103. 0002-8703/$ - see front matter n 2005, Elsevier Inc. All rights reserved. doi:10.1016/j.ahj.2004.07.040
provides an excellent reference standard by which we can contemporaneously assess the validity of specific electrocardiographic criteria used for the diagnosis of LVH. Although the echocardiogram is extremely accurate in its ability to diagnose LVM, the electrocardiogram (ECG) is more widely used because of its easy performance and low cost. The clinical diagnosis of LVH, in disorders such as hypertension and valvular disease carries important prognostic implications.5,6 For these reasons, it is important to understand both the capabilities and limitations of the ECG in the diagnosis of LVH. Many excellent studies have been published which have attempted to refine or confirm the value of some of the ECG criteria widely used in the diagnosis of LVH.7-9 The electrocardiographic diagnosis of LVH has been shown to have reasonable specificity but lacks sensitivity. The degree of sensitivity described by many studies lies between 40% and 60%.10 Although LVH, at the most basic level, is defined as an enlargement of cardiac myocytes, the clinical representations of LVH that are most common are increased wall thickness or ventricular mass. An increased LVM
American Heart Journal April 2005
710 Budhwani, Patel, and Dwyer
can be characterized by different geometric forms. There is an increased wall thickness and normal left ventricular size at one end of the spectrum, whereas at the other end of the spectrum, there is increased left ventricular size and normal wall thickness. The performance of the several commonly used ECG criteria may vary as LVM increases. Each of the ECG criteria may also, individually, perform differently in the presence of concentric hypertrophy than they do in the presence of eccentric hypertrophy. There is limited understanding of the contribution and importance of left ventricular thickness, ventricular size, or overall LVM on the value and performance of the individual criteria such as ECG voltage, repolarization abnormalities, and alterations in atrial and ventricular conduction. The purpose of our study is to determine the relative importance and effect of an increased left ventricular wall thickness, left ventricular diastolic diameter, and LVM on the performance on each of the 4 major ECG criteria for LVH and to determine how the findings can be incorporated into the use of the routine ECG for the interpretation of LVH.
Methods Study group After exclusions, we analyzed data from 608 consecutive patients for study (obtained N2 years) from the echocardio graphic and electrocardiographic databases at UMDNJ University Hospital, who had a septal and posterior wall thickness z14 mm on the echocardiogram. We retrieved complete echocardiographic data along with the ECG for each patient. The ECG was obtained within a mean of 7 days of the echocardiogram. We excluded patients with inadequate technical quality on the echocardiogram and ECG evidence of a Q wave myocardial infarct, atrial fibrillation, complete left or right bundle-branch block, and a paced rhythm. Standard ECG criteria were adopted.11 Two physicians, who had no knowledge of the echocardiographic findings, separately performed an analysis of the ECG. All disagreements were resolved between the 2 readers.
Electrocardiographic measurements Our technical staff recorded a standard 12-lead ECG, on Marquette computerized recorders, at 25 mm/s and 1 mV/cm standardization with equipment having frequency response characteristics in accordance with American Heart Association recommendations.12 The tracings were interpreted for the presence or absence of the following 4 criteria for LVH: (1) (2) (3) (4)
Increased voltage, ST depression and/or T-wave inversion, Left atrial enlargement, and Prolonged QRS conduction time.
We calculated a point score according to the system proposed by Romhilt and Estes.9 In this system, each of the
Table I. Echocardiographic measurements in 608 patients with increasing LVM
Echocardiogram measurements Mean LV mass (g) Mean LV wall thickness (mm) Mean LV diastolic diameter (mm)
Group I
Group II
Group III
(b bb 400 g)
(400-600 g)
(N NN 600 g)
n = 290
n = 255
n = 63
318 15
480 17
718 18.5
41
51
60
LV, Left ventricle.
criteria is graded as 3 points except QRS duration which is awarded 1 point. The total score ranges from 0 to 10. QRS amplitude was measured to the nearest microvolt on a digitized ECG. The presence of increased voltage was calculated using the modified Cornell criteria. These more recent criteria are (for men: S in V3 plus R in aVL N2.8 mV [28 mm], and for women: S in V3 + R in aVL N2.0 mV [20 mm]).13 Left atrial abnormality was coded as present if the P-wave inversion in lead V1 was N1 mm in length and N1 mV in amplitude. ST depression N0.1 mV and/or T-wave inversion of any degree were considered present if found in any lead except aVR, and V1 to V3. QRS duration was measured to the nearest 10 milliseconds and was considered abnormal when the duration was z100 milliseconds.9
Echocardiographic measurements Comprehensive 2-dimensional and Doppler echocardiography was performed on all patients. Left ventricular dimensions (intraventricular septal thickness, posterior wall thickness, and left ventricular end-diastolic diameter) were measured at end of diastole with M-mode by using leading-edge–to–leadingedge convention. Left ventricular mass (g) was calculated using the Penn cube convention with the following formula: LVM = 1.04[(LVEDD + PW + IVS)3 (LVEDD)3] 13.6, where LVEDD is the left ventricular end diastolic diameter; PW is the posterior wall thickness, and IVS is the intraventricular septal thickness.
Data analysis Group subsets. We arbitrarily divided patients into 3 groups (groups I-III) according to the calculated LVM. We intended these groups to reflect mild, moderate, and severe hypertrophy. The groups represent approximate 200-g increments in LVM. All patients in group I had an LVM b400g; group II had an LVM from 400 to 600 g, and group III had an LVM N600 g. Left ventricular wall thickness. To evaluate the contribution of wall thickness to the performance of the 4 ECG criteria, we compared the frequency of each of the ECG criteria between 2 extreme groups (A and B) of wall thickness. In group A, all patients had both septal and posterior wall thickness between 14 and 15 mm. In group B, all patients had at least one wall thickness z17 mm. To eliminate the influence
American Heart Journal Volume 149, Number 4
Budhwani, Patel, and Dwyer 711
% Patients
Figure 1
Table II. Relationship between number of criteria for LVH present on the ECG and LVM
100 90 80 70 60 50 40 30 20 10 0 < 400 g (n = 290)
Romhilt and Estes ECG score
% ST and/or T abn % LAE % Voltage abn % QRS delay
0 1-3 4-6 7-9 10
Mean LV mass (g) 346 365 410 452 501
P value*
(0 vs 1-3) (1-3 vs 4-6) (4-6 vs 7-9) (7-9 vs 10)
= V V V
.11 .001 .001 .003
*t Test.
4-600 g (n = 255)
>600 g (n = 63)
LV Mass With increasing LVM, there was a significant increase in the number of patients demonstrating the presence of each of the 4 criteria on the ECG ( P b .001 for LAE and P b .0001 for the other 3 criteria). LV, left ventricle; LAE, left atrial abnormality; abn, abnormality.
of LVM, we restricted both groups to patients with an LVM b400 g so that both groups had a similarly low LVM. A m2 test was used to compare groups with respect to each of the ECG criteria frequency. In this subgroup of 147 patients, we also correlated the mean wall thickness with the overall ECG score. A P value of b.05 was considered significant. Left ventricular diastolic diameter. In a similar manner to the analysis of wall thickness, 2 groups of markedly different left ventricular diameters were also created from the subset with an LVM b400 g. Group C was selected from patients with a left ventricular diameter of b50 mm and group D from patients with a left ventricular diameter of N60 mm. A m2 test was used to compare groups with respect to each of the ECG criteria frequency. In this subgroup of 331 patients, we also correlated the left ventricular diameter with the overall ECG score. A P value of b.05 was considered significant. Left ventricular mass. To evaluate the effect of an increasing LVM on the 4 ECG criteria performance, we compared the frequency of each of the 4 ECG criteria, described above, as they occurred in each of the 3 LVM groups (see group subsets described above). A m2 test was used to compare the presence of each of the ECG criteria with respect to each of LVM groups. In the entire study cohort of 611 patients, we also correlated the continuous variables of LVM with the corresponding ECG score. A P value of b.05 was considered significant.
Results Patient characteristics We obtained ECGs and echocardiograms on 608 patients. The sex distribution of this cohort was 47% men and 53% women. The mean age of the study group was 56 years. The patients with the lowest LVM (b400 g)
had a mean age of 58 years. With increasing LVM, the mean age declined such that, at the highest level of LVM (N600 g), the mean age was 52 years ( P b .01).
Echocardiographic findings In the entire patient cohort, our measurements of the left ventricular septal and posterior wall thickness ranged between 13 and 30 mm with a mean value of 16 mm. All patients exhibited an abnormal wall thickness because this was the primary requirement for entry into the study. The left ventricular end-diastolic diameter ranged between 25 and 74 mm. with a mean value of 49 mm. The LVM ranged between 134 and 1006 g, with a mean value of 460 g. The relationship between left ventricular mass, wall thickness, and diastolic diameter Our patient cohort consisted of 3 groups, characterized by increasing LVM (see Table I). Group I (290 patients) had a mean LVM of 318 g; group II (255 patients) had a mean LVM of 480 g, and group III (63 patients) had a mean LVM of 718 g. From groups I to II and III (with increasing LVM), the mean left ventricular wall thickness progressively increased from 15 to 17 mm and 18.5 mm, respectively. The mean left ventricular diastolic diameter also progressively increased from 41 to 51 mm and then to 60 mm in those 3 groups with increasing LVM. Effect of increasing LVM on individual electrocardiogram criteria An increase in the echocardiogram-derived LVM had a significant effect on all 4 ECG criteria (see Figure 1). The differences in the frequency of each of the 4 criteria were statistically significant at a P value of b.001 or greater. Voltage criteria. The frequency of the occurrence of voltage criteria for LVH increased from 52% to 66% to 83% as LVM progressively increased from groups I to III ( P b .0001). Left atrial abnormality. As LVM increased through the 3 groups, the frequency of left atrial abnormality also rose from 46% to 59% and then to 68% ( P b .001).
American Heart Journal April 2005
712 Budhwani, Patel, and Dwyer
Figure 2
Figure 3 GROUP A
GROUP B
80
70 62
68
% Patients
44 42
45
40
42
Voltage LAE
39 29
30
ST - T abn QRS Delay
% Patients
60
50
GROUP D
70
58
60
GROUP C
50
65*
55 49
45 40
40
43
46
30
Voltage LAE ST - T abn QRS delay
20
20
10
10 0
0 Wall Thickness 14 mm LVM = 313 g (n = 109)
Wall Thickness = 17-27 mm LVM = 345 g (n = 38)
LV diameter 50 mm LVM = 310 g (n = 263)
LV diameter 60 mm LVM = 276 g (n = 68)
* P ≤ .001
There is no significant effect of increasing left ventricular wall thickness on each of the 4 ECG criteria for LVH in the presence of similar LVM in each group.
There is no significant effect of increasing left ventricular wall diameter on 3 of the 4 ECG criteria for LVH in the presence of similar LVM in each group. Asterisk indicates that the effect of increasing left ventricular wall diameter significantly increased the frequency of a QRS delay ( P b .001).
ST-T changes. In a similar fashion, the ST-T wave changes also were more frequent with increasing LVM. We found a change from 55% in group I up to 95% in group III ( P b .0001). QRS duration. Finally, the presence of QRS prolongation was also significantly increased from 42% in group I up to 70% in group III ( P b .0001).
thickness, per se, did not influence ECG results. The correlation between wall thickness and the ECG score in this subgroup was poor (r = 0.07) and not statistically significant ( P = .34).
Relationship between electrocardiogram score and left ventricular mass When a zero score was present from the ECG, the mean LVM was 346 g (see Table II). When the greatest score (10) was present on the ECG, the mean LVM was 501 g. With each graduation above the 0 -to -3 ECG score, the difference was significant at a P value of b.001, b.001, and b.003, respectively. In addition, in all 608 patients, we demonstrated a significant correlation (r = 0.35) of the Romhilt-Estes ECG scores with the LVMs ( P b .0001). Effect of left ventricular wall thickness on electrocardiogram criteria From our blowQ LVM patients (b400 g), we created 2 subgroups. Group A included 109 patients with a septal and posterior wall thickness between 14 and 15 mm, and group B had 38 patients with 1 wall thickness N17 mm (see Figure 2). The 2 subgroups had a similar and statistically not different mean LVM (313 vs 345 g). In the analysis of frequencies of each of the 4 ECG criteria between the group with the least increase in wall thickness and the group with the greatest increase in wall thickness, we observed that there were no significant differences in the frequency of any of the 4 ECG criteria for LVH between the 2 groups. Wall
Effect of left ventricular diastolic diameter on electrocardiogram criteria In a similar manner to the description above, we selected from the blow Q LVM (b400 g) patients and created 2 groups (see Figure 3). Group C included 263 patients who had a left ventricular end-diastolic diameter of b50 mm, whereas the 68 patients in group D patients all had a diameter of N60 mm. These 2 subgroups had a similar mean LVM (310 vs 276 g) that was not statistically different. In our analysis, there were no significant differences between the 2 groups, in the frequency of any of the 4 ECG criteria except for the QRS prolongation. The frequency of QRS prolongation (64% vs 40%) was significantly ( P b .001) greater in the group with increased diameter despite a somewhat lower mean LVM. Except for QRS duration, left ventricular diameter did not influence ECG results. The correlation between left ventricular diameter and the ECG score in this subgroup was also poor (r = 0.07) and not statistically significant ( P = .23).
Discussion Despite the unquestioned demonstration that the M-mode and 2-dimensional echocardiogram provide us with a satisfactory bgold standardQ for the determi-
American Heart Journal Volume 149, Number 4
nation of LVH, there have been continuous efforts to improve both the performance of the ECG and our understanding of its limitations. The ECG is relatively inexpensive and easy to perform, insuring its role as an excellent screening tool for the diagnosis of LVH. In this vein, we carried out this analysis to improve the understanding of which anatomic features of the remodeled left ventricle influence the ECG criteria that are used in the diagnosis of LVH. Anatomically, LVH is characterized by an increase in muscle mass, and mass is a function of both left ventricular chamber size and left ventricular wall thickness. The relative importance of wall thickness and ventricular diameter, vis-a`-vis LVM, in the determination of the ECG abnormalities has not been previously examined. This study was designed to examine the relation between these 3 anatomic characteristics of LVH and the 4 ECG criteria used for the diagnosis of LVH. The ECG criteria we used were those in common usage and included voltage, left atrial abnormality, prolonged QRS conduction time, and repolarization abnormalities. Echocardiography is the procedure of choice for the quantitation of left ventricular wall thickness, diameter, and mass.4 We used this technique as our reference standard for these measurements. Our patient population was selected based on the presence of 1 definitive aspect of LVH on the echocardiogram, that is, an abnormal left ventricular wall thickness. As a result of this inclusion criterion, we anticipated that all patients would have an increased LVM. In fact, our cohort demonstrated a wide range of abnormal echocardiographic changes such as a left ventricular wall thickness between 13 and 30 mm, enddiastolic dimensions from 25 to 74 mm, and an LVM between 134 and 1006 g. Although we lacked the data (height and weight) to normalize the LVM, the purpose of the study was to examine the effect of the mass on the ECG rather than to establish normal versus abnormal, which is the primary value of normalizing such data. The interpretation of LVH on the ECG has usually been approached as a dichotomous interpretation (either present or not present). Of all the ECG criteria for LVH, increased voltage has always been a necessity for the consideration of a diagnosis of LVH. It is the only ECG abnormality that has been uniquely linked with LVH. As a result, the sensitivity for the diagnosis of LVH, by the ECG, rests with the presence of increased voltage. The sensitivity for detection of LVH in our patient population, based on the overall frequency of abnormal voltage, was 61%. This percentage compares favorably with previously published studies.10,14 A number of efforts have been published that attempt to improve sensitivity through differing methods of estimating QRS voltage.14 -16 Also, the effect of external factors on voltage, such as body habitus, has also been examined.17
Budhwani, Patel, and Dwyer 713
The presence or increased frequency of the other ECG criteria for LVH, beyond voltage, usually has the effect of improving the specificity of the diagnosis when combined with increased voltage. Therefore, any factors that influence the frequency of the ECG criteria will logically also affect either sensitivity or specificity. It is in this light that we analyzed the impact that anatomic features of LVH would have on the ECG criteria. There is minimal information in the literature on the independent effect that left ventricular wall thickness, size, or mass has on the frequency of the ECG criteria. This in turn could influence the sensitivity and specificity for the ECG diagnosis of LVH.18,19 Unexpectedly, our analysis demonstrated that the extent of increase in LVM was the only anatomic abnormality that influenced the frequency of all 4 ECG abnormalities and correlated with the total ECG score. In contrast, neither wall thickness or ventricular diameter significantly affected the frequency of the appearance of the ECG criteria. We found that, as LVM increased, the frequency of each of the 4 ECG criteria correspondingly increased. By example, at the lowest level of LVM (b400 g), abnormal voltage occurred in only 52%, whereas at the highest tertile of LVM (N 600 g), the frequency of abnormal voltage was 83%. Therefore, the sensitivity of detecting increased LVM improved proportionally to the degree of increased mass. In the same manner and to the same degree, we found that an increased occurrence of left atrial abnormality, QRS conduction time, and ST-T repolarization abnormalities was associated with an increasing LVM. Therefore, one would anticipate an improved specificity, as well, as the mean LVM increased. In the highest tertile of LVM, the frequency of the appearance of left atrial abnormality (68%), QRS conduction time (70%), and ST-T changes (95%) was significantly greater than in the patients in the lowest tertile of LVM, who had mean rates of 46%, 42%, and 55%, respectively. In support of the above observations, we also observed that 19% had a normal ECG in the lowest tertile group, whereas a normal ECG was not present in any patient in the upper tertile group for LVM. These findings can explain the variability in the sensitivity and specificity of the ECG in the detection of increased LVM in previous studies. Based on our data, we would anticipate that a population under study with a large increase in LVM (renal insufficiency, idiopathic cardiomyopathies, and severe aortic valvular disease) would be expected to show both a higher sensitivity and specificity than a population consisting of disorders associated with lower LVM changes. We suggest that expectations, for the performance of the ECG in the detection of increased LVM (LVH), should be moderated accordingly. One might anticipate that clinical disorders, commonly associated with large increases in LVM, will reflect LVH on the ECG, on average, in 70% to 80% of the
American Heart Journal April 2005
714 Budhwani, Patel, and Dwyer
patients. On the other hand, patients with the lowest increases in LVM will often go undetected. Detection of small increases in LVM would be best approached with echocardiographic studies. In our study, we also examined the effect that increasing severity of wall thickness and ventricular diameter has on the frequency of the 4 ECG criteria when LVM is held relatively constant. We found, in both analyses, that neither the increase in wall thickness or ventricular diameter correlated with the ECG score nor resulted in a significant change in the frequency of any of the 4 ECG criteria. There was 1 exception to that general conclusion. We did observe that a prolonged QRS conduction time occurred significantly more often with increased left ventricular size, although LVM was similar. In addition to our observations of the effect that LVM has on the ECG, we also demonstrated that as the number of criteria increases on the ECG, the LVM significantly increases. For example, when voltage criterion is the only criterion present on the ECG, the average LVM is 365 g, whereas the presence of all 4 criteria on the ECG is associated with a 37% greater mean LVM at 501 g. From this, we conclude the ECG indicators of LVH are reasonable representations of the severity of LVM. Because of the wide range of LVM within each ECG grouping, these findings are only reflective of the group and cannot be applied with any assurance to the individual patient.
6.
7.
8.
9.
10.
11.
12.
13.
14.
References 1. Lewis T. Observations upon ventricular hypertrophy with special reference to preponderance of one or other chamber. Heart 1914;5:367. 2. Scott RC, Seiwert VJ, Simon DL, et al. Left ventricular hypertrophy — a study of the accuracy of current electrocardiographic criteria when compared with autopsy findings in one hundred cases. Circulation 1955;11:89 - 96. 3. Rosenfeld L, Goodrich C, Kassenbaum G, et al. The electrocardiographic recognition of left ventricular hypertrophy. Am Heart J 1962;63:731 - 42. 4. Devereux RB, Casale PN, Eisenberg RR, et al. Electrocardiographic detection of left ventricular hypertrophy using echocardiographic determination of left ventricular mass as the reference standard. J Am Coll Cardiol 1984;3:82 - 7. ¨ rnlo¨v J, et al. Echocardiographic and 5. Sundstro¨m J, Lind L, A electrocardiographic diagnoses of left ventricular hypertrophy
15.
16.
17.
18.
19.
predict mortality independently of each other in a population of elderly men. Circulation 2001;103:2346 - 51. Levy D, Garrison RJ, Savage DD, et al. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Heart Study. N Engl J Med 1990;322: 1561 - 6. Casale PN, Devereux RB, Kligfield P, et al. Electrocardiographic detection of left ventricular hypertrophy: development and prospective validation of improved criteria. J Am Coll Cardiol 1985; 6:572 - 80. Romhilt DW, Bove KE, Norris RJ, et al. A critical appraisal of the electrocardiographic criteria for the diagnosis of left ventricular hypertrophy. Circulation 1969;40:185 - 95. Romhilt DW, Estes Jr EH. A point score system for the electrocardiographic diagnosis of left ventricular hypertrophy. Am Heart J 1968;75:752 - 8. Levy D, Labib SB, Anderson KM, et al. Determinants of sensitivity and specificity of electrocardiographic criteria for left ventricular hypertrophy. Circulation 1990;81:815 - 20. Sgarbossa EB, Wagner G. Electrocardiography. In: Topol E J, editor. Textbook of cardiovascular medicine. Philadelphia: Lippincott-Raven Publishers Philadelphia. p.1545-89. Pipberger HV, Arzbaecher RC, Berson AS, et al. Recommendations for standardization of leads and specifications for instruments in electrocardiography and vectorcardiography. Report of the committee on electrocardiography, American Heart Association. Circulation 1975;AHA(Suppl):I -1 - 31. Casale PN, Devereux RB, Kligfield P, et al. Electrocardiographic detection of left ventricular hypertrophy: development and prospective validation of improved criteria. J Am Coll Cardiol 1985;6: 572 - 580. Murphy ML, Thenabadu PN, de Spyza N, et al. Sensitivity of electrocardiographic criteria for left ventricular hypertrophy according to type of cardiac disease. Am J Cardiol 1985;55:545 - 53. Sokolow M, Lyon TP. Ventricular complex in left ventricular hypertrophy as obtained by unipolar precordial and limb leads. Am Heart J 1949;37:161 - 86. Casale PN, Devereux RB, Alonso DR, et al. Improved sexspecific criteria of left ventricular hypertrophy for clinical and computer interpretation of electrocardiograms: validation with autopsy findings. Circulation 1987;75:565 - 7. Okin PM, Roman MJ, Devereux RB, et al. Electrocardiographic identification of left ventricular hypertrophy: test performance in relation to definition of hypertrophy and presence of obesity. J Am Coll Cardiol 1996;27:124 - 31. Verdecchia P, Dovellini EV, Gorini M, et al. Comparison of electrocardiographic criteria for diagnosis of left ventricular hypertrophy in hypertension: the MAVI study. Ital Heart J 2000;1:207 - 15. Schillaci G, Verdecchia P, Borgioni C, et al. Improved electrocardiographic diagnosis of left ventricular hypertrophy. Am J Cardiol 1994;74:714 - 9.