Evaluation of a QRS Scoring System for Estimating Myocardial Infarct Size VIII. Specificity in a Control Group With Left Ventricular Hypertrophy and Proposal of a New Scoring System for Use With This Confounding Factor
Christopher J. Freye, MD,* Nancy B. Wagner, BA,* Catherine M. Howe, BA,* Nancy C. Stack, BA,* Raymond E. Ideker, MD, PhD,* Ronald H. Selvester, MD,+ and Galen S. Wagner, MD*
Abstract: Electrocardiographic differentiation between left ventricular hypertrophy (LVH) and myocardial infarction (MI) is often difficult because both diagnoses are based primarily on QRS changes on the electrocardiogram (ECG). The specific goal of this study was the development of ECG criteria that could be used with the complete Selvester QRS scoring system for MI size in patients with LVH. A study population of 127 patients had significant aortic valve disease verified by cardiac catheterization. Inclusion in the study required no significant coronary artery disease, no focal contraction abnormality on the left ventriculogram, and no documented MI. Quantitative criteria for LVH developed by Bonner (IBM) and also those developed by the Cornell group were used to determine the ECG evidence for LVH in each patient. One or both sets of criteria were met in 110 (87%) of the 127 patients. This group was compared to a previously evaluated control population of 500 normal subjects. The complete 54-criteria, 32-point QRS MI size scoring system was applied to the I2-lead ECG of both groups. The score was 98% specific in the normal controls and 73% specific in the LVH group using a score of >3 points as diagnostic for MI. Of the 54 individual QRS criteria, 16 failed to achieve 95% specificity in the LVH population: 13 were for anterior (and apical), 2 for inferior, and 1 for posterior locations. Of these 16, minor modifications to 11 were sufficient to achieve the 95% specificity standard. The resultant 49-criteria, 31-point QRS score for estimating MI size in patients with LVH was 95% specific using a level of >3 points as diagnostic for MI. The sensitivity of this score and its ability to predict MI size in the presence of LVH must now be tested in prospective studies of patients with documented LVH, with and without MI. Key words: left ventricular hypertrophy, electrocardiogram, Selvester QRS infarct size score.
From the *Division of Cardiology. Department of Medicine, and the Department of Pathology, Duke University Medical Center, Durham, North Carolina and fMemoria1 Heart Institute of Long Beach, California.
Supported in part by Research Grants HL- 17670 and HL-00546 from the National Heart, Lung and Blood Institute. Reprint University
19
requests: Galen S. Wagner, MD, Box Medical Center, Durham, NC 27710.
3636,
Duke
20
Journal of Electrocardiology
Vol. 25 No. 1 January 1992
Recent studies from this institution have evaluated a simplified version of the QRS scoring system of Selvester for estimating the size of myocardial infarcts (MI) based on the QRS complex of the standard 12-lead electrocardiogram (ECG). This infarct scoring system was shown to be specific in normal subjects’ and sensitive for quantifying single anterior,2 inferior,3 and posterolatera14 MIS. The complete QRS scoring system has also exhibited high specificity (98%) in 500 normal subjects5 when a score of >3 points was considered diagnostic for MI. False positive ECG diagnosis of MI in patients with left ventricular hypertrophy (LVH) is well recognized6,‘; consequently, patients meeting ECG criteria for LVH were excluded from all previous studies of the QRS scoring system. This study of patients with presumed LVH due to aortic valve disease and previous studies of patients with presumed right ventricular hypertrophy due to mitral stenosis’ or car pulmonale’ are the first attempts to evaluate the application of the QRS scoring system in patients with these potentially confounding cardiac problems. The specific goal of this study is the development of ECG criteria that can be utilized with the complete QRS scoring system for MI size in patients with an ECG diagnosis of LVH. To accomplish this goal we will ( 1) evaluate each of the 54 individual scoring criteria to determine those that retain high specificity in the presence of LVH and (2) propose modifications of nonspecific criteria that will enable them to achieve the ~95% specificity standard.
Methods Patient Selection We reviewed the records of all 1,43 5 patients from Duke University Medical Center ( 1971- 1983) who were documented to have hemodynamically significant aortic valve disease, as defined by a peak-topeak systolic aortic gradient of at least 30 mmHg or at least 3 + aortic insufficiency by angiography, to maximize the probability of LVH. MI was excluded by requiring that each patient have ( 1) no significant coronary artery disease (525% reduction in crosssectional diameter of any vessel), (2) no focal contraction abnormality on the left ventriculogram, and (3) no previously documented MI, including history, positive cardiac enzymes, or serial ECG changes in the acute clinical setting. Patients with complete left bundle branch-block (QRS 20.14 ms), left anterior fascicular block (frontal plane axis 5 - 45’), or right bundle branch block were also excluded. Furthermore, all patients with valve disease other than aortic
were excluded to minimize the likelihood of coexisting RV hypertrophy. Mean pulmonary arterial pressures in the study ranged from 6 to 40 mmHg. Since there was no relationship between the mean pulmonary artery pressure and the QRS score, patients with elevated pressures were not excluded from the study. The final study group consisted of 127 patients.
Study Design Standard 12-lead ECGs were recorded in the supine position using a Hewlett-Packard automatic cardiograph (15 15B). All ECGs were recorded within 1 week prior to catheterization. The cineangiograms, taken at a speed of 60 frames per second, were evaluated for abnormalities in left ventricular wall motion, the degree of occlusive coronary artery disease, and the amount of mitral and aortic insufficiency. The ECGs of the 127 study patients were evaluated for evidence of LVH using the quantitative criteria of both the IBM automated ECG analysis program developed by BoMer and co-workers” and the Cornell group.’ ’ A “Banner” ECG-LVH score was determined for each patient (Table 1). A score of at least 5 points has exhibited 92% specificity and 66% sensitivity in test populations” and was, therefore, considered indicative of LVH in this study. The simplified “Cornell Voltage” criteria of R amplitude in aVL + S amplitude in V3 > 2.0 mV for women and > 2.8 mV for men were also used for diagnosis of LVH. l1 The 54 criteria from the complete Selvester QRS scoring system5 were used to determine an MI score for each patient. A total score of > 3 points has achieved 98% specificity in control populations without ventricular hypertrophy.5 The level of QRS MI score that achieved 95% specificity in the study Table 1. Bonner ECG-LV Hypertrophy
Criteria Points
A
1. 2. 3. 4. 5.
(SinV2) + (RinV6)54.3mV,or S in VI 2 2.4, or R in Vz 2 2.8. or Any QRS deflection in VI-V6 2 3.0 mV. or Maximum positive + negative in V1-V6 2 5.0 mV B If A not met: RinaVLZ 1.3mV C If A or B not met: 1. R in aVL 2 1.1 mV, or 2. (S in V,) + (R in V,) 2 3.0 mV D T positive in V2, T < - 1 mV in V6, and QRS - T angle 2 100” E QRS frontal plane axis negative F Intrinsicoid deflection 2 50 ms in Va with total QRs< 130ms
5 4
3 3 1 1
Evaluation of QRS Scoring System population with LVH was determined. The specificity of each MI criterion was evaluated, and any criterion exhibiting less than the 95% level was considered “nonspecific.” Attempts were then made to modify all nonspecific criteria suffkiently to achieve this standard. The QRS screening subset identified by Anderson et al. I2 (Q z 30 ms in aVF; R 2 40 ms in VI; or any Q or R I 0.1 mV and 5 10 ms in V,) were also tested to determine if they achieved the 95% specificity standard in the presence of LVH.
Results The specificity of the complete Selvester QRS scoring system is illustrated in Figure 1. A level of > 3 points achieved a specificity of 73%. A level of > 6 points was required to achieve the 2 95% specificity standard. At least 1 of the MI screening criteria of Table
Location Lateral
Inferior
Lead I
2 30 ms WQ 5 1 R 5 0.2 mV
No. of Times Criterion Met
Q
Criterion Specificity (%I
5
96 100 100
-
-
aVL
Q z 30 ms R/Q I 1
19
86 98
II
Q z 30 ms Q z 40 ms
8 0
94 100
11 3 1 6
87 98 99 95 96
Any Q S 2 1.8 mV
21 58
83 54
Any Q
12 26 17 13
91 80 87 90
8 15 11
94 88 87
aVF
Anterior
Criterion
2. Specificity
VI
VI
lQ
2 Q z Q P R/Q R/Q
30 ms 40 ms 50ms 5 2 5 1
*R< 10ms *R 5 0.1 mV RVZ 5 RV,
V3
Any Q R 5 20 ms R 5 0.2 mV
*MI screening
criteria.
Freye et al.
l
21
Anderson et al.‘* was falsely present in 28 patients (22%), yielding 78% specificity for the subset. Of the 127 patients, 110 (87%) met Bonner and/ or Cornell voltage criteria for LVH. Thirty-seven met the Bonner criteria alone, 12 met the Cornell voltage criteria alone, and 61 met both. We determined the specificity of each of the 54 MI criteria (Table 2). Thirty-eight (70%) of the criteria achieved at least 95% specificity. Twelve of the 16 nonspecific criteria are associated with the anterior MI location, including 9 related to abnormal Q waves or diminished initial R waves in aVL and V, to Vs and 2 related to R amplitude or R to S amplitude ratios in Vq. The anterior criterion S 2 1.8 mV in V1 was present in 58 (46%) of the patients. The single point R/S ratio criterion in apically oriented lead VT, both of the 30 ms Q wave duration criteria in inferiorly oriented leads, and the 50 ms R wave duration criterion (for posterior MI) in Vz also failed to meet the specificity standard. of Each of the MI Criteria Criterion Specificity W)
Location
Lead
Criterion
No. of Times Criterion Met
Posterior
VI
R/s 2 1 *R 2 40 ms R z 50 ms R 2 0.6 mV RrlmV S 5 0.3 mV
1 2 0 2 0 0
99 98 100 98 100 100
VZ
Rls 2 1.5 R > 50 ms R 2 60 ms R? 1.5mV Rz2mV S 5 0.4 mV
3 8
98 94 99 97 99 99
Apical
1 4 1
1
5
Q 2 20 ms R/Q 5 1 WQ C 0.5 R 5 0.7 mV R/s 5 1 R/s 5 0.5 Notched R
0 7 23 6 6
96 99 100 94 82 95 95
VS
Q 2 30 ms WQ 5 2 WQ 5 1 R 5 0.7 mV R/s 5 2 R/s 5 1 Notched R
3 1 0 0 13 1 2
98 99 100 100 90 99 98
Vh
Q 2 30 ms R/Q53 R/Q5 1 R 5 0.6 mV R/s 5 3 R/s ‘; 1 Notched R
6 0 0 3 6 1 0
95 100 100 98 95 99 100
V4
1
22
Journal of Electrocardiology
Vol. 25 No. 1 January 1992
Fig. 1. Specificity of the QRS infarct scoring system in study patients. The numbers of patients achieving that score is indicated above each bar. The broken vertical line indicates the QRS scoring level of > 6 points required to reach the 2 95% specificity standard.
Table 3 presents the modifications
of 11 of these
16 criteria that enabled them to achieve the ~95%
specificity standard. No modifications were possible in the remaining five criteria. An increase of Q duration in the three limb leads to ~40 ms and consideration of a Q wave in V, -V3 only if followed by an R wave are examples of modifications required for identification of MI in the presence of LVH. The resultant 49-criteria, 3 1-point QRS score for MI size was 95% specific in this LVH population using a score of >3 points as diagnostic for MI.
Discussion The ability of the QRS complex to indicate either the diagnosis or size of an MI may be limited by the presence of confounding factors such as bundle Table
branch block or ventricular hypertrophy. Previous studies have investigated the specificity of both the simplified and complete QRS MI scoring systems in 500 normal subjects and in patients with presumed right ventricular hypertrophy due to either mitral stenosis or car pulmonale.8,9 It is important to determine the precise ways that these factors (left or right bundle branch block and left and right ventricular hypertrophy) either mimic or obscure the QRS changes due to MI. The goal of this study was to investigate the specificity of the complete QRS scoring system in patients with LVH. The presence of an MI in the LVH population is reduced to a very low probability, according to published reports,‘3-15 by the absence of a clinical history of MI and the absence of any abnormality on selective coronary angiograms and left ventriculograms. The presence of LVH in aortic valve disease patients was assumed using the requirement that only those with significant stenosis and/or regurgitation be included in this study. In addition, the Bonner criteria for LVH were 77% sensitive in this study versus 66% in a previous study of 302 patients with LVH proven by autopsy or angiography. lo The addition of the Cornell voltage criteria identified 12 more patients, increasing the sensitivity to 87%. The specificity of the QRS scoring system at the level of >3 points in the 127 patients studied with aortic valve disease was 73% versus 98% in the 500 normal subjects.5 These results indicate that significant LVH due to aortic valve disease markedly decreases the specificity of the QRS scoring system. The anterior MI location is that most falsely indicated by the electrocardiographic manifestations of LVH. Wilson et al. r6 first suggested that a QRS pattern in right precordial leads was common in the setting of LVH. 0thers6,17,18 have also reported that the loss of an-
3. Proposed Tightening of Nonspecific Criteria Proposed
Current
Specificity
Specificity Location Lateral Inferior Anterior
Lead aVL II aVF VI VZ
V3
Posterior Apical
VZ V4 VS
Criterion Q 2 30 ms Q 2 30 ms Q 2 30 ms S 2 1.8 mV any Q any Q Rs 10ms R 5 0.1 mV RV;! 5 RV, any Q R 5 20 ms R 5 0.2 mV R 2 50 ms R I 0.7 mV Rls 5 1 R/s 5 2
W)
Criterion
WI
86 94 87 54 83 91 80 87 90 94 88 87 94 94 81 90
Q 2 40 ms Q 2 40 ms Q 2 40 ms
96 100 97
qrS qrS
98 98
qrS
96
R 5 R 2 R 5 R/s R/s
0.1 mV 60 ms 0.6 mV s 0.5 5 1.5
95 99 97 95 98
Evaluation of QRS Scoring System
terior forces on the ECG can occur in patients with LVH uncomplicated by MI, leading to the false positive ECG diagnosis of anterior MI. Selvester et al. l9 previously proposed that the QRS MI scoring criterion of a QRS complex in leads V, and VZ and S 2 1.8 mV in V, would not be applicable in patients with LVH. Relatively minor modifications to 11 of the nonspecific criteria were sufftcient to achieve the 95% specificity standard (Table 3). The addition of these criteria to the 38 that originally met the standard creates a new proposed 49-criteria, 3 l-point QRS scoring system for estimating MI size in patients with LVH that is 95% specific using the >3 point threshold. The sensitivity of these criteria to MI and their ability to predict infarct size in the presence of LVH must now be tested in prospective studies of patients with documented LVH, with and without MI.
Acknowledgments The authors wish to express their appreciation to Gail McKinnis, Judy Berry, and Sylvester Cherry for their expert preparation of the manuscript.
References Wagner GS, Freye CJ, Palmeri ST et al: Evaluation of a QRS scoring system for estimating myocardial infarct size. I. Specificity and observer agreement. Circulation 65:342, 1982 Ideker RE, Wagner GS, Ruth WK et al: Evaluation of a QRS scoring system for estimating myocardial infarct size. Il. Correlation with quantitative anatomic findings for anterior infarcts. Am J Cardio149: 1604, 1982 Roark SF, ldeker RE, Wagner GS et al: Evaluation of a QRS scoring system for estimating myocardial infarct size. Ill. Correlation with quantitative anatomic fmdings for inferior infarcts. Am J Cardiol 51:382, 1983 Ward RM, White RD, ldeker RE et al: Evaluation of a QRS scoring system for estimating myocardial infarct size. IV. Correlation with quantitative anatomic findings for posterofateral infarcts. Am J Cardiol 53:706, 1984 Hindman NB, Schocken DD, Widmann M et al: Evaluation of a QRS scoring system for estimating myocardial infarct size. V. Specificity and method of application of the complete system. Am J Cardiol 55:1485, 1985
l
Freye et al.
23
6. Myers GB: QRS-T patterns in multiple precordial leads that may be mistaken for myocardial infarction. I. Left ventricular hypertrophy and dilatation. Circulation 1:844, 1950 7. Phillips HR, Starr JW, Behar VS et al: Evaluation of vectorcardiographic criteria for the diagnosis of myocardial infarction in the presence of left ventricular hypertrophy. Circulation 53:235, 1976 8. Howe CM, Freye CJ, Wagner NB et al: Evaluation of a QRS scoring system for estimating myocardial infarct size. VII. Specificity in a control group with right ventricular hypertrophy due to mitral stenosis. Am .J Cardiol 62:322, 1988 9. Behar JV, Howe CM, Wagner NB et al: The performances of new criteria for right ventricular hypertrophy and myocardial infarction in patients with car pulmonale and with mitral stenosis. J Electrocardiol 24:231, 1991 10. Bonner RE, Crevasse JL. Ferrer Ml, Greenfield JC Jr: A new complete program for comparative analysis of serial scalar electrocardiograms and performance of the 1976 IBM program. Comput Biomed Res 11: 103, 1978 11. Casale PN, Devereux RB, Kligtield P et al: Electrocardiographic detection of left ventricular hypertrophy: development and prospective validation of improved criteria. J Am Co11Cardiol 6:572, 1985 12. Anderson WD, Wagner NB, Lee KL et al: Evaluation of a QRS scoring system for estimating myocardial infarct size. VI. Identification of screening criteria for non-acute myocardial infarcts. Am J Cardiol 6 1: 729, 1988 13. ldeker RE, Behar VS, Wagner GS et al: Evaluation of asynergy as an indicator of myocardial fibrosis. Circulation 57:715, 1978 14. Miller RA, Amsterdam EA, Bogren HG et al: Electrocardiographic and cineangiographic correlations in assessment of the location, nature and extent of abnormal left ventricular segmental contraction in coronary artery disease. Circulation 49:447, 1974 15. Field BJ, Russell RO, Dowling JT, Rackley CT: Regional left ventricular performance in the year following myocardial infarction. Circulation 46:679, 1972 16. Wilson FV, Johnston FD, Rosenbaum FF et al: The precordial electrocardiogram. Am Heart J 27: 19, 1944 17. Benchimol AB, Schlesinger P: Electrocardiographic changes in a case of left ventricular hypertrophy and septal hypertrophy resembling anterior myocardial infarction. Circulation 1:970, 1950 18. Master AM, Silverblatt MC, Rosenfeld I, Donoso E: Electrocardiographic patterns simulating coronary occlusion in patients with chronic rheumatic cardiovascular disease. Am Heart J 54:50, 1957 19. Selvester RH, Sanmarco ME, Solomon JC, Wagner GS: The ECG: QRS change. p. 23. In Wagner GS (ed): Myocardial infarction: measurement and intervention. Martinus Nijhoff, The Hague, 1982