Hubert V. Pipberger

Computer analysis of the orthogonal electrocardiogram and vectorcardiogram in 1,002 patients with myocardial infarction

Abstract Orthogonal ECGs (Frank system) were recorded from 1,002 patients with documented nyocardial infarction. In 39 per cent of this series, tracings were obtained during the acute phase of the disease and in 61 per cent at a later date. Records from 229 normal subjects above the age of 40 years served as control. A total of 333 different ECG measurements were computed from each record in order to search for optimal discriminators between the group of normal subjects and the patients with myocardial infarct. Using linear discriminant-function analysis with 15 different ECG variables, it was possible to identify correctly 84 per cent of the infarct series with 6 per cent false positives. When the total sample was subdivided on the basis of QRS morphology into anterior, postero-diaphragmatic, and lateral infarcts, the classification could be improved by 2 to 14 per cent. The results were tested on an independent record sample derived from 240 autopsy cases of myocardial infarcts. With the use of the multivariate classification procedure developed on the basis of the clinically diagnosed sample, 88 per cent of the autopsy cases could be diagnosed correctly. Computer classification results were also compared with conventional 12 lead ECG interpretations with the use of the Minnesota Code. When the false positive rate was kept at 6 per cent for both methods, 49 per cent of the standard ECGs were classified correctly as compared to 84 per cent for orthogonal ECGs analyzed by computer. In addition, optimal discriminators between normal and infarct records were determined for measurements which can be easily obtained by hand. Correct classifications decreased to 80 per cent with 8 per cent false positives. QR amplitude ratios proved most efficient. Results of the study emphasize the need for efficient control of false positive rates because increases in diagnostic sensitivity are frequently offset by concomitant losses in specificity. Application of multivariate analysis techniques proved very efficient for diagnostic ECG classification by computer.

The orthogonal electrocardiogram in normal women. Implications of sex differences in diagnostic electrocardiography

Abstract Normal limits of the orthogonal electrocardiogram and vectorcardiogram in adult women, ranging in age from 18 to 90 years, are presented. A comparison of results is made with those of normal age-matched men, and sex differences are analyzed from a total of 960 normal records (510 men and 450 women). For the majority of scalar and vectorial items, significant sex differences were found which in women included shorter QRS duration, smaller vector loops, and decreased P, Q, R, S, and T deflections. The upper normal limits of R x , R y , and R z amplitudes were 11 per cent, 20 per cent, and 30 per cent less, respectively, in women than in men. The sensitivity and specificity of electrocardiographic criteria, for high and low voltage, were significantly affected by these sex differences in amplitudes. For example, as a discriminator between normals and subjects with left ventricular hypertrophy, the upper normal limit of R x + R z amplitude sum was 3.10 millivolts in men but 2.50 millivolts in women. Hence, the use of the limit derived from males in a female population would decrease its sensitivity drasticially. Similar discrepancies existed in the sensitivity and specificity of electrocardiographic criteria for low voltage. Since the lower normal limit of R x amplitude was 0.51 millivolt in men but only 0.35 millivolt in women, a substantial number of normal women would be misclassified as having right ventricular hypertrophy or chronic obstructive pulmonary disease if the limit derived from males was used as a criterion. The absence of Q waves in Leads x and y was a common finding in each age and sex group and carries no diagnostic significance. While initial anterior QRS forces in Lead z were present in all normal men, they were smaller and even absent in 1 per cent of normal women. Hence, greater difficulties in electrocardiographic diagnosis of anteroseptal myocardial infarction in women may be encountered. Mean vectors at the end of QRS (point J) and early part of the ST segment were more inferiorly and anteriorly directed in men than in women. T waves in Lead z were always negative in men, but flat or positive T waves were observed in some of the normal women. Sex differences in the level of point J and the ST segment may have important bearings on the interpretation of exercise electrocardiograms. The shorter QRS duration in women signified the importance of sex-specific limits for ventricular conduction delays.

Left atrial overload: A hemodynamic, echocardiographic, electrocardiographic and vectorcardiographic study

Combining electrocardiography with current techniques for continuous bedside hemodynamic monitoring and echocardiography permits analysis of P wave morphology in light of concurrent and accurate measurements of left atrial pressure and chamber size. In order to determine a noninvasive means of estimating left ventricular filling pressure during changing hemodynamics, and to evaluate contributions of left atrial size and pressure to P-wave morphology, 144 pulmonary capillary wedge pressures (PCW, mm. Hg) with 12-lead electrocardiograms, vectorcardiograms, and left atrial echocardiograms were obtained in 61 cardiac care patients. n nNoninvasive predictors of the PCW were sought initially and after changes in the PCW. Three electrocardiographic and seven vectorcardiographic P-wave indices and three echocardiographic left atrial dimension indices were evaluated. This study found that the P-terminal negativity, PTF V1, (abnormal defined as greater than −.30 mm.-sec. negativity) was the best electrocardiographic predictor of PCW; PTF V1 correlated moderately well with PCW (r = .67, P 14 (sensitivity 86 per cent and specificity 79 per cent). Despite the lack of a high correlation, PTF V1 was helpful in that a normal PTF V1 excluded patients with PCW pressures in the pulmonary edema range (PCW > 24 mm. Hg). No patient with PCW > 24 had a normal PTF V1. Similarly, presence of an abnormal PTF V1 excluded patients with low left ventricular filling pressures (PCW < 10 mm. Hg). Neither P-terminal positivity in orthogonal lead z nor a multiple coefficient regression program of computer measured vectorcardiographic parameters was superior to other individual electrocardiographic and vectorcardiographic measurements in theri correlation with PCW. Despite the widespread use of PTF V1 for the electrocardiographic diagnosis of left atrial enlargement, echocardiographic left atrial dimension showed a significant but low correlation with PTF V1 (r = .49), suggesting that atrial size is a weaker determinant of PTF V1 than atrial hypertension. Left atrial wall tension, expressed as the product of PCW pressure and left atrial dimension, also showed a moderate though better correlation (r = .72) with PTF V1 than did PCW pressure or left atrial dimension individually. Conclusions of this study are that additional factors are importantly related to the ECG pattern of left atrial overload. This conclusion is supported by the rather modest correlation coefficient between PTF V1 and PCW pressure initially and an important second finding that in 51 patients studied serially, simultaneous changes in the electrocardiogram, vectorcardiogram, and left atrial dimension were unreliable predictors of acute changes in PCW pressure. n nThe study design utilized offers potential for testing and improving current electrocardiographic criteria and the results obtained caution against inferences made about hemodynamics and cardiac anatomy from the electrocardiogram, particularly when non-dynamic comparisons between the electrocardiogram and these functions are made.

Distortions in infant electrocardiograms caused by inadequate high-frequency response

Frank lead ECGs from infants were studied for frequency content by introducing low-pass filters of 50, 75, 100, and 150 150 Hz bandwidths before obtaining computer measurements. Results indicated that a minimum bandwidth of 100 Hz is required to avoid amplitude error of 10 per cent or greater. This bandwidth requierement is essentially the same as that required for adult ECGs despite the fact that infant QRS durations are usually about half those of adults. Although the average infant ECG spectrum is likely to contain higher frequencies than the average adult ECG spectrum, duration values for Q, R, and S waves overlap in these populations to such an extent that bandwidth requirements are practically identical.

Computer analysis of changes in frank vectorcardiograms of 666 normal infants in the first 72 hours of life

Frank vectorcardiograms (VCGs) were collected on magnetic tape for 666 normal newborn infants at 1, 6, 12, 24, 36, 48, 60, and 72 hours after birth and analyzed by computers. The final total included 1,337 VCGs for white babies and 413 for blacks. No previous report has been made for the normal neonate with such a large sample, and no previous substantiation exists of possible age or sex differences at this early age. This study establishes a statistically significant change in vectorcardiographic patterns over the first three days, specifically in the measurements P duration, QRS duration, maximal spatial QRS amplitude, S in lead x, and T in lead z, as well as for several time-normalized QRS vectors. (P less than or equal to .005.) Racial differences were significant for T waves in lead z. This study supports the use of vectorcardiographic standards sensitive to the age of the newborn as well as to race.

Year-to-year variation of the orthogonal electrocardiogram and vectorcardiogram among 243 normal white males

When assessing patients serial ECG changes, the clinician implicitly compares those changes to the limits of change expected in a healthy population. Prospective epidemiological studies, too, develop their criteria from the limits of normal serial ECG changes. Surprisingly then, few studies have reported normal limits for changes between serial ECGs taken six months or longer apart, and all are based on small samples. The present study has a large sample size: 243 white middle-aged and older males, after exclusions for heart disease. Each had at least four consecutive annual examinations with ECGs. Limits of serial variability were computed for 52 measurements. The ECG measurements included durations, amplitudes, ratios, angles and spatial magnitudes. Clinical measurements included blood pressure, cholesterol relative weight and hemoglobin. Year-to-year ECG variabilities were compared to day-to-day variabilities of the same measurements reported earlier. Year-to-year variation was virtually identical to the reported day-to-day variation in most measurements. In only two measurements was year-to-year variation over 25% greater than the reported day-to-day variation.

The impact of an ECG computer analysis program on the cardiologist's interpretation. A cooperative study

Nine experienced electrocardiographers and the ECG computer program developed in the Veterans Administration (AVA 4.0) were evaluated against ECG-independent evidence of 180 patients true diagnoses. A cross section of cardiac abnormalities was included. Each reader was given the 12-lead and orthogonal 3-lead ECG. The impact of ECG computer reports on the interpretations by the nine readers was evaluated by comparing their interpretations before and after the addition of a computer report. Using only high probability statements, the average accuracy of ECG diagnosis by the nine readers was 54%. It increased to 62% when the computer report was added. Computer interpretation was correct in 76%. It was shown that the Bayesian classification method together with multivariate analysis, used in the VA program, are mainly responsible for the improvement in diagnostic accuracy.

Studies on the nature of the repolarization process. XIX. Studies on the mechanism of ventricular activity.

Abstract The clinical significance of the T wave in representing the electrical recovery process of the ventricular myocardium is well known. This electrical phenomenon, however, has been the subject of much less basic research than other parts of the electrocardiogram. With the development, in recent years, of more reliable recording techniques and the introduction of an incubator in which animal experiments can be performed in a more physiologic medium, this study of the repolarization process of the heart muscle seemed justified. Abundant controversy and confusion have developed over the direction of repolarization through the ventricular wall on the basis of polarity of the T wave. T polarity was studied by means of minute plunge electrodes recording simultaneously at selected depths in the ventricular walls. Moreover, the time course of the electrical recovery process was examined by comparing analogous points of the T wave in simultaneous records from the myocardial surface and the under-lying subendocardium and from different surface points, and these results compared with the polarity of the T waves. Experiments were designed also to study (a) the effect of local thermal changes upon adjacent and remote regions of the heart, (b) the effect of primary T-wave changes upon secondary ones by superimposing thermal changes upon left bundle branch block, and (c) the effect upon the T wave of ischemia due to chronic obstruction of coronary arteries.

Studies on the nature of the repolarization process.

The Institute for Medical Research, Cedars of Lebanon Hospital, Los Angeles, Calif.; and the Department of Medicine, University of California School of Medicine, Los Angeles, Calif. The process of repolarization of the ventricular musculature, represented by the T wave in the electrocardiogram, has been studied relatively little as compared to that of depolarization. This is due mainly to the fact that the T wave is very unstable in the usual experimental preparation.

The orthogonal electrocardiogram as risk indicator for the prediction of myocardial infarction and/or cardiac death

In a prospective study on Coronary Heart Disease (CHD) orthogonal electrocardiograms (Frank) were recorded annually for ten years from 1,444 asymptomatic, middle-aged males with a mean age of 57.4 +/- 10.6 years. Cases with overt or suspected CHD were excluded. The purpose of the study was to identify risk indicators in electrocardiograms and to compare them with other known risk factors used for prediction of acute CHD events such as myocardial infarction (MI) and/or cardiac death (CD). Such acute events occurred in 88 cases. Pre-event ECGs of these acute events were compared with all others without events, using logistic regression analysis. Identified ECG risk indicators were then compared with other known risk factors such as smoking, blood pressure, cholesterol, age, weight, etc. The predictive power of the ECG, derived mainly from the ST-T complex, exceeded all others by a wide margin. The amplitude of the first 1/8 of the ST-T complex in lead x (similar to V5-V6) together with relative body weight proved best when one pre-event record was available. Prediction improved when ECG changes between two pre-event recordings were included. Precision of measurements by computer appeared essential for improvements in CHD prediction.