Comwter Analvsis of Carotid and I Brachial /Pulse Waves Effects of Age in Normal Subjects* EDWARD D. FREIS, M.D., F.A.C.C. and MARY C. KYLE, B.S. Washington, D. C. I T HAS been known for some time that aging is associated with a reduction in the distensi- bility of the aorta and large arteries.`g2 Recent evidence suggests that these age-related changes may be detected in man by means of the exter- nally recorded carotid pulse wave.3-5 Further- more, pharmacologic agents that alter arterial distensibility change the configuration of the carotid wave in the expected direction; that is, during drug-induced reduction in distensibility the wave changes resemble those found with aging, whereas after acute increase in distensi- bility the reverse occurs.5m6 In addition to providing a possible index of arterial aging, external pulse wave recordings are of interest for the following reasons: (1) they can be recorded by a trained technician using painless, bloodless methods; (2) improve- ment in transducers has facilitated the obtaining of good quality tracings with rapidity and ease'; (3) magnetic tape recordings followed by computer analysis permit the processing of a large volume of recordings, an important consideration in population surveys or in large clinics. In the present investigation computer tech- nics were utilized in the analysis of the record- ings. The main purpose of the study was to assess the effects of aging on the carotid and brachial pulse waves of "normal" subjects. The brachial pulse was included because it has not been studied previously with respect to aging and is less subject to artifacts produced by respiratory and other movement or by venous pulsations than the carotid pulse. METHODS EQUIPMENT AND PROCEDURE Pulse wave recordings were made with a strain gauge transducer' consisting of a water-filled chamber surfaced on one side by a thin plastic membrane and on the opposite side by a metal diaphragm. The transducer was applied over the artery by means of a clamp adjusted to provide a loading pressure of 20 to 30 mm. Hg. The frequency response of the trans- ducer was flat (=t5 per cent) to 20 c.p.s. Linearity and hysteresis were within =I=1 per cent over the working range of the instrument.7 The transducer was connected to a Sanborn carrier-wave preamplifier and recordings made on magnetic tape. Three F.M. channels of information were recorded simultaneously, the carotid and bra- chial pulse waves and lead I of the electrocardio- gram. Blood pressure was taken by the auscultatory method at the beginning and end of the recording session. The subjects rested supine in an air-condi- tioned room (74 to 78O F.) during the procedure. They were instructed to hold their breath in expira- tion for approximately 5 seconds while the recordings were being taken. SUBJECTS Carotid and brachial pulse waves were recorded in 228 male subjects classified as normal without clinically evident atherosclerotic complications, heart disease, hypertension, or diabetes mellitus. They ranged in age from 15 to 90 years and had no evi- dence of anemia, recent weight loss, or fever. Ap- proximately a third were hospital personnel or stu- dents ; the remainder were patients admitted for elective surgical procedures or convalescing from acute illnesses. * From the Veterans Administration Hospital and the Department of Medicine, Georgetown University School of Medicine, Washington, D.C. This study was supported in part by Grant HE 09696 from the National Heart Institute, National Institutes of Health, U. S. Public Health Service. Address for reprints: Edward D. Freis, M.D., Veterans Administration Hospital, 50 Irving St., N.W., Washington, D.C. 20422: VOLUME 22, NOVEMBER 1968 691 692 Freis and Kyle METHOD OF ANALYSIS Conversion of the analog data to digital form was carried out by methods previousIy described.8 Com- puter plots were obtained from the digital tape of amplitudes against the times of the simultaneously recorded electrocardiogram and of the carotid and brachial pulse waves. The points used in the analysis were then identified on the waves by inspection as follows: the beginning of the QRS complex (Q) in the electrocardiogram, the beginning (FJ and end (F.J of each pulse wave, the incisura (I) and the dicrotic wave (D) (Fig. 1). Fr and Fz were identified as the intersection of a straight line fitted to the steep- est portion of the systolic upstroke of the pulse wave with a horizontal base line drawn through the points of minimal amplitude (foot points). The point se- Iected for the incisura was the minimum of that in- flection, and for the dicrotic wave the midpoint of the succeeding positive wave. In the brachial wave only one systolic peak (P) was identified, but in the carotid wave two systolic maxima (PI and P2) were identified. TABLE I Correlation Coefficients of Measurement Values Versus Age YBrachial Artery-- ,--Carotid Artery--. Correlation Correlation Measurement Coefficient Measurement Coefficient I/D 0.65 wp, 0.60 1-D % -0.59 I/PI 0.51 FL-P % 0.46 1-D 70 -0.50 I/P 0.35 I/D 0.38 Q-FL -0.32 FL-Pz % -0.29 FFFz -0.03 FL-PI So 0.10 FrI % 0 02 Q-F, -0.08 FI-I y0 -0.03 FI-FP -0.01 Fl-F2 = time interval between the beginning and end of the pulse wave cycle; Fl-I 2: = time interval from beginning of pulse wave to incisura as a percentage of total cycle length; F,-P yG = foot to peak time as a per cent of total cycle length in the brachial pulse wave; I/D = amplitude ratio of the incisura relative to the midpoint of the dicrotic wave: I-D %, = time interval as a per cent of the total cycle length from the incisura to the midpoint of ;he dicrotic *ax; I/P = amplitude ratio of incisura relative to peak in the brachixl pulse wave; P,/P, = amplitude ratio of the second systolic maximum relative to the first in the carotid pulse wave; Q-FL = interval between beginning of QRS complex in electrocardiogram and beginning of pulse wave cycle. Figure 1. Printouts of average .imlse waves calculated by digital computer for the carotid artery (left) and the brachial artery (right). The average waves for the subjects under age 30 are shown above, those for ages 3045 inclusive are in the middle and the average waves for subjects over 45 are shown below. A line has been drawn connecting the average values. The shaded areas indi- cate f 1 standard deviation from the mean. THE AMERICAN JOURNAL OF CARDIOLOGY Computer Analysis of Carotid and Brachial Pulse 693 The times of occurrence of these identified points were punched on cards. The latter plus the digital tape comprised the data input of the computer pro- grams. From these data the various time and ampli- tude measurements indicated in Table I were calcu- lated. Amplitude ratios were indicated by fractions, such as I/D, the amplitude ratio of the incisura rela- tive to the midpoint of the dicrotic wave. Absolute time intervals were indicated by a hyphen separating the points of measurement, such as Fi-Fz, the time interval between the beginning and end of the pulse wave cycle. Relative time intervals were indicated by a hyphen and per cent sign, such as I-D $!& the time interval as a per cent of the total cycle length from the incisura to the midpoint of the dicrotic wave. wave also increased with age, the correlation coefficient being 0.46. In the carotid pulse wave the value for the correlation coefficient between P.JPr and age was 0.60 (Table I, Fig. 2) and 0.51 between I/Pi and age. The I-D y0 interval decreased with age, the correlation coefficient being -0.50. No significant correlations were found be- tween age and cycle length (Fi-FJ or in ejection period as a percentage of cycle length (Fr-I %) in either the brachial or the carotid pulse waves. To illustrate the pulse wave changes which occur with aging, average curves were constructed from the ca- rotid and brachial waves for each of three subgroups: subjects under 30, subjects from 30 to 45 and subjects more than 45 years of age. Utilizing the computer, average curves were obtained by normalizing the amplitudes of the pulse cycles with respect to the base line and then dividing each cycle into 100 intervals. The mean and standard deviation of the amplitudes at each division point were then computed and plotted (Fig. 1). Multiple correlation coeficients were determined by using measurement values from both the carotid and brachial pulse waves. The multi- ple correlation coefficient for carotid P2/P1 and brachial I/D with age was 0.71. Addi- tional multiple correlation coefficients with age were as follows: brachial I/D and carotid I/P1 (r = 0.68), brachial I-D yc and carotid PJPr (r = 0.66) and brachial I-D `% and carotid I/P1 (Y = 0.62). The correlation coejkients between the various mea- surement values and age were calculated. Scatter diagrams were also plotted (Fig. 2 and 3). RESULTS The reconstructed average pulse waves in the normal subjects revealed characteristic changes in shape with advancing age (Fig. 1). In the carotid waves the most prominent change was in the amplitude of the second systolic maximum relative to the first (PJPr). In the subjects under age 30 the height of the first was of greater amplitude than the second, whereas with increasing age, the second maximum rose relative to the first. The height of the incisura relative to the first systolic maximum (I/Pi) also rose with increasing age. In the averaged brachial pulse waves the prominent changes were a decrease in amplitude of the dicrotic wave relative to the incisura (I/D) and a shortening of I-D with increasing age (Fig. 1). DISCUSSION Carotid Pulse Wave: The principal age- related changes in the carotid pulse waves of male subjects were an increase in the amplitude of both the incisura and the second systolic maximum relative to the first systolic maxi- mum. These carotid pulse wave changes with increasing age have been noted previously by Dontas et a1.3 and Friart.4 Although the mecha- nism of these changes has not been completely elucidated, they probably reflect at least in part the decrease in distensibility of the central arterial system which occurs with aging.6.6 Brachial -Pulse Wave: In addition to these alterations in the carotid wave the present study demonstrated that the brachial pulse also exhibits characteristic changes with age. The brachial wave is technically somewhat easier to record and is less subject to respiratory or other motion artifacts than is the carotid wave. These technical considerations may be of some importance in large scale surveys in which it is often necessary to minimize the time spent on individual recordings. The correlation coefficients calculated be- The most significant age-related changes in tween the various measurement values and age the bra&al arterial pulse were a decrease in are presented in Table I. The most significant the amplitude and duration of the dicrotic changes in the brachial recordings were indi- wave. Lax and Feinberg et a1.gv10 noted a cated by the correlation coefficient of 0.65 be- similar diminution in the dicrotic wave of the tween I/D and age (Table I, Fig. 2) and of digital pulse in subjects with atherosclerosis -0.59 between I-D yo and age. The foot and adults with diabetes as well as in children to peak time (Fr-P %) in the brachial pulse with diabetes mellitus. Strandness and Bell." VOLUME 22, NOVEMBER 1968 694 Freis and Kyle 1.66L 1.39 - . . 15 25 35 45 55 65 75 85 AGE IN YE4RS Figure 2. Scatter diagram showing the relation between the amplitude ratio of the second to the first systolic maximum of the carotid pulse wave and age. 1.23 . 15 25 35 45 55 65 75 85 AGE IN YEARS Figure 3. Scatter diagram showing the relation between the amplitude ratio of the incisura to the dicrotic wave of the brachial arterial pulse and age. using a mercury resistance gauge plethysmo- graph, also observed diminished to absent dicrotic waves in the presence of clinically ap- parent atherosclerosis. Thus, alterations in the dicrotic wave of the peripheral pulse may prove to be a useful index of arterial changes associated with aging and vascular disease. Clinical Implications: The present study indi- cates that the shape of the arterial pulse wave changes progressively with age from the third decade onward. Does the young individual with a pulse wave configuration "older" than his chronologic age already have the beginnings of vascular disease? The observations of Fein- berg and Lax10 in children with diabetes suggest that this may be the case. Long-term prospec- THE AMERICAN JOURNAL OF CARDIOLOGY Computer Analysis of Carotid and Brachial Pulse 695 tive studies will he needed to answer this im- portant question. It is hoped that as a result of the present study and others made earlier, pulse wave recordings will be incorporated in longitudinal surveys currently being carried out on large population groups. SUMMARY Externally recorded carotid and brachial arterial pulse waves were analyzed with the aid of a digital computer in a series of 228 male sub- jects free of clinically detectable cardiovascular disease. Alterations in pulse wave contours were found to be significantly correlated with age. In the carotid wave the predominant changes with age involved the relative ampli- tudes of the incisura and the two systolic max- ima ; in the brachial pulse the principal altera- tion with age was a diminution in the ampli- tude and duration of the dicrotic wave. The degree of correlation of pulse wave changes with age was approximately the same in the brachial and carotid tracings. REFERENCES 1. HALLOCK, P. and BENSON, I. C. Studies on the elastic properties of the human isolated aorta. J. Clin. Znvest., 16: 595, 1937. 2. ROACH, M. R. and BURTON, A. C. The effect of age on the elasticity of human iliac arteries. Canad. J. Biochem., 37: 557, 1959. 3. DONTAS, A. S., TAYLOR, H. L. and KEYS, A. Ca- rotid pressure plethysmograms. Effects of age, diastolic blood pressure, relative body weight and physical activity. Arch. Kreirlaufforsch., 36: 49, 1961. 4. FRIART, J. La morphologie du sphygmogramme carotidien dam l'arteriosclerose. Acta cardiol., 15: 557, 1960. 5. FREIS, E. D., HEATH, W. C., LUCHSINGER, P. C. and SNELL, R. E. Changes in the carotid pulse which occur with age and hypertension. Am. Heart J., 71: 757, 1966. 6. DADDARIO, R. C. and FREIS, E. D. Kinetocardio- gram, phonocardiogram, and arterial pulse waves during acute hemodynamic changes. Circulation, 34: 423, 1966. 7. DAVIS, M., GILMORE, B. and FREIS, E. Improved transducer for external recording of arterial pulse waves. IEEE Trans. Bio-Med. Electronics, 10: 173, 1963. 8. PIPBERGER, H. V., FREIS, E. D., TABACK, L. and MASON, H. L. Preparation of electrocardio- graphic data for analysis by digital electronic com- puter. Circulation, 21: 413, 1960. 9. LAX, H., FEINBERG, A. W. and COHEN, B. M. Studies of the arterial pulse wave. I. The normal pulse wave and its modification in the presence of human arteriosclerosis. J. Chum. Dis., 3: 618, 1956. 10. FEINBERG, A. W. and LAX, H. Vascular abnormali- ties in children with diabetes mellitus. J.A.M.A., 201: 105, 1967. 11. STRANDNESS, D.E., JR. and BELL, J.W. Peripheral vascular disease : Diagnosis and objective evalua- tion using a mercury strain gauge. Ann. Surg., 161 (Suppl.): 3, 1965. VOLUME 22, NOVEMBER 1968