The reproducibility of central aortic blood pressure measurements in healthy subjects using applanation tonometry and sphygmocardiography

Journal of Human Hypertension (1999) 13, 625 629 1999 Stockton Press. All rights reserved 0950-9240/99 $15.00 http://www.stockton-press.co.uk/jhh ORIGINAL ARTICLE The reproducibility of central aortic blood pressure measurements in healthy subjects using applanation tonometry and sphygmocardiography A Siebenhofer, CRW Kemp, AJ Sutton and B Williams University of Leicester, Cardiovascular Research Institute, Leicester, UK Aim: Sphygmocardiography via applanation tonometry is a non-invasive, bedside technology which utilises tonometric analysis of the radial artery pulse wave and measurement of peripheral arterial blood pressure (BP) to derive a central arterial pulse wave, central arterial BP and related indices. The present study was designed to determine: (1) the inter-operator variability in measurements obtained using this technique; (2) the relationship between measured peripheral arterial BP and derived central arterial BP. Method: Multiple measurements were made from 25 healthy subjects (15 male), mean age 33 (s.d. 10.3) years, mean arterial BP 90 (s.d. 12) mm Hg by two trained observers at the same time of day on three separate occasions. Results: The mean inter-operator difference was 0.1 (s.d. 1.7) mm Hg for derived systolic aortic BP and 0.1 (s.d. 0.7) mm Hg for derived diastolic aortic BP (Bland and Altman analysis). Pulse wave Augmentation Index (AIx) values, ranged from 22% to 40%, with the inter- operator measurement difference being only 0.4 (s.d. 6.4)%. Buckberg ratio measurements ranged from 119% to 254%, with the inter-operator measurement difference being only 2.7 (s.d. 15.4)%. The relationship between derived central systolic BP and peripheral systolic BP readings in individual patients was not constant and showed significant variance when compared on different days (ANOVA, P 0.03). This was not explained by any significant variance in heart rate (ANOVA, P 0.39). Conclusion: Applanation tonometry has excellent interobserver reproducibility when used by trained observers. Moreover, the inconsistency in the relationship between peripheral and central aortic BP suggests that the former is not a perfect surrogate for the latter. Further prospective studies are required to define whether derived central aortic BP may be a better predictor of cardiovascular morbidity and mortality and the impact of different antihypertensive therapies on the relationship between peripheral and central arterial BP. Keywords: applanation tonometry; reproducibility; blood pressure Introduction Sphygmomanometric measurements of peripheral blood pressure (BP) (usually brachial) have revealed peripheral BP to be a strong predictor of cardiovascular morbidity and mortality. 1 However, it is not the peripheral pressure per se that accounts for this increase in cardiovascular morbidity and mortality. On the contrary, BP is measured peripherally for convenience and simplicity and acts as a surrogate for central arterial BP. Nevertheless, the relationship between peripheral and central BP has not been measured in clinical trials and cannot be assumed to be constant because it is influenced by many factors, including the age and haemodynamic performance of the aorta. 2 6 Conventional sphygmomanometric peripheral BP measurements can be augmented by analysis of the Correspondence: Bryan Williams, Cardiovascular Research Institute, Sir Robert Kilpatrick Clinical Sciences Building, PO Box 65, Leicester Royal Infirmary, Leicester LE2 7LX, UK Received 16 December 1998; revised and accepted 28 May 1999 arterial pulse wave. It has been recognised (but largely ignored) for more than 150 years that the arterial pulse is of complex shape, varies with age and in different arteries and changes markedly with physiological and pharmacological interventions. 7 A new non-invasive method; applanation tonometry, widely used in ophthalmology, has been adapted to analyse the arterial pressure wave conveniently at the bedside. 7 9 The normal pulse wave in peripheral arteries comprises a systolic wave separated from a diastolic wave by the incisura, caused by aortic valve closure. 10 The diastolic wave is generated by reflection of the pulse wave within the circulation and is important for diastolic filling of coronary arteries. 11 Aging and/or arteriosclerosis, both lead to arterial stiffening and loss of arterial compliance, principally in the descending aorta. 9 The consequence is an increase in pulse wave velocity which causes the reflected wave to return earlier in the cardiac circle. This leads to diminution or disappearance of the diastolic wave and augmentation of the systolic wave, resulting in increased systolic work (and thus left ventricular hypertrophy and

626 increased oxygen demand). This effect is compounded by the resulting decrease in diastolic pressure time interval, thus predisposing to cardiac ischaemia. 9,12 14 The data acquired from pulse wave analysis and simultaneous measurement of brachial arterial BP by conventional sphygmomanometry can be utilised to derive values for central aortic systolic and diastolic BPs using a validated mathematical transfer function. 3 It has been proposed that applanation tonometric pulse wave analysis may thus yield important additional information with regard to the haemodynamic status and therapy-induced haemodynamic changes in patients with cardiovascular disease, including hypertension. 14 Before this hypothesis can be tested in multicentre, prospective, randomised clinical trials, it is important to define whether the clinical use of applanation tonometry for sphygmocardiography by different operators, yields reproducible data that is not confounded by significant inter-observer variability. Moreover, it is important to define whether there is significant variance in the relationship between peripheral (brachial) BP and derived central arterial BP. If there is, then it is conceivable that the latter could be a more accurate predictor of cardiovascular outcome and response to antihypertensive therapy than the former. This study thus examines inter-observer variability in applanation tonometric-derived cardiovascular indices in repeated examinations of healthy subjects. with light pressure to flatten the vessel wall so that the transmural forces within the vessel are perpendicular to the arterial surface. 6,8 The arterial pressure wave form is modified during its transit from the ascending aorta to peripheral vessels, thus the central aortic wave form and pressure indices (the maximum and minimum pressures of the central waveform in mm Hg) are derived from radial tonometry and the peripheral brachial BP using a previously validated mathematical transfer function within the software package (Sphygmocor ). 3,6,17 Augmentation Index (AIx) reflects changes in pulse wave morphology as a result of the reflected wave and is calculated by the difference in pulse height between the second (reflected wave) and first shoulder (the primary wave) of the pulse wave divided by the maximum pulse height of the waveform (Figure 1). If the reflected wave is greater than the primary wave the AIx is positive. If the reflected wave is less than the primary wave then AIx is negative (Figure 1). The validity of these derived values have been confirmed by simultaneous direct central aortic measurements. 2,3,18 The BP and tonometric measurements were made by two trained observers (AS and CRWK). The tonometric assessment of each subject was completed by each observer within 5 min. Each observer studied each subject under identical conditions and the examinations were immediately consecutive. Each subject was examined by both observers at the same time of day on three separate occasions. Subjects and methods Subjects We assessed the reproducibility of applanation tonometry in 25 healthy volunteers (15 males), mean age 33 (s.d. 10.3) years, mean body mass index 24 (s.d. 3.2) kg/m 2, mean systolic BP 118 (s.d. 17.2) mm Hg, mean diastolic BP 75 (s.d. 10.1) mm Hg and mean arterial BP 90 (s.d. 12) mm Hg at the same time of day on three separate occasions. Statistical analysis The inter-observer reproducibility of the derived central aortic haemodynamic indices for each subject at each sitting (a total of 75 paired measurements) was determined using the method of Bland and Altman. 19 The variation between the peripheral (brachial) arterial BP, the derived central aortic BP and heart rate, by each observer, for each Blood pressure measurements Peripheral BP was measured using an Omron automatic oscillometric digital BP monitor (HEM- 705CP). The cuff was applied to the non-dominant arm with the subject in a sitting position. The subject was seated for 10 min prior to the first reading. Three measurements, 5 min apart, were performed. The final recorded BP was taken to represent the peripheral BP. Applanation tonometry Immediately after the final BP reading was recorded, the same arm was used for the applanation tonometry recording. A micromanometer-tipped probe (Sphygmocor, Pulse Wave Medical Ltd, Australia) was applied to the surface of the skin overlying the radial artery and the peripheral radial pulse wave form was continuously recorded. 6,15,16 For accurate recordings, the micromanometer must be applied Figure 1 The basic features of the arterial pulse. After the onset of ejection (T 0 ) the pressure wave rises to an initial shoulder (T 1 ) which relates to timing of peak flow, and then proceeds to the second shoulder (T 2 ) relating to the reflected waves. The end of ejection (T 3 ) is associated with closure of the aortic valve (incisura).

subject on three separate days (a total of 150 measurements) was analysed using analysis of variance (ANOVA). Results The inter-operator difference in the derived central aortic BP indices was statistically analysed using the method of Bland and Altman. 19 The differences between the two operators for each subject at each sitting were plotted against the average of the indices recorded by both operators. The mean of the difference between the two operator measurements was 0.1 (s.d. 1.7) mm Hg for derived systolic central aortic BP and 93% of the readings were within 2 standard deviations (Figure 2a). For derived diastolic central aortic BP, the mean of the difference was also 0.1 (s.d. 0.7) mm Hg and 97% of the readings were within 2 standard deviations (Figure 2b). Augmentation index (AIx), the difference between the height of the first and second derived central aortic systolic peaks expressed as a percentage of the pulse pressure, can also be computed from the tonometry data (Figure 1). AIx reflects the haemody- namic performance of large conduit blood vessels and is inversely related to vascular compliance. The range of AIx from all measurements was 22 to +40%, the mean of difference between two operator s readings was 0.4 (s.d. 6.4)% and 94% of the readings were within 2 standard deviations (Figure 3a). When expressed as absolute derived pressure, values ranged from 6 to 15 mm Hg. The mean of difference in augmentation in mm Hg (AG, difference between first and second systolic peak in mm Hg) was 0.1 (s.d. 2.1) mm Hg with 94% of the readings within 2 standard deviations (Figure 3b). The Buckberg ratio can also be calculated from the tonometry data. This ratio is dependent on the diastolic pressure time interval/(dpti) systolic pressure time interval (SPTI). This ratio DPTI/SPTI is an index of myocardial oxygen supply and demand and is a sensitive marker of subendocardial ischaemia. 11,20 Buckberg ratio measurements ranged from 119% to 254% with a mean operator difference of 2.7 (s.d. 15.4) %, with 93% of readings within 2 standard deviations (Figure 4). 627 Figure 2 Bland and Altman plot for the comparison of two operator measurements of derived systolic aortic BP (a), mean of difference 0.1 (s.d. 1.7) mm Hg, and derived diastolic aortic BP (b), mean of difference 0.1 (s.d. 0.7) mm Hg. These values indicate excellent inter-observer reproducibility. Figure 3 Bland and Altman plot for the comparison of two operator measurements of calculated augmentation index expressed as either a percentage (AIx) (a), mean of difference 0.4 (s.d. 6.4) %, or in mm Hg (b), mean of difference 0.1 (s.d. 2.1) mm Hg. These values indicate excellent inter-observer reproducibility.

628 in heart rate on the three different days of study (Table 1) and the significant variance in the peripheral BP:central aortic pressure ratio was not explained by significant variance in heart rate on different days (P = 0.39) (Table 1). This observation confirms that the relationship between peripheral and central arterial BP is not constant and that the former does not uniformly predict the latter. Figure 4 Bland and Altman plot for the comparison of two operator measurements of calculated Buckberg ratio, mean of difference 2.7 (s.d. 15.4) %. These values indicate excellent interobserver reproducibility. Since approximately 95% of all readings are within the error range defined by the coefficient of repeatability, we can conclude that the procedure is reproducible as defined by the British Standards Institute. 21 We next examined whether the ratio of peripheral BP: central aortic pressure remained constant in subjects analysed on the three separate days. The variance in this ratio was analysed using ANOVA which took into account repeated measurements. No significant difference between readings on subjects taken on the same day were observed (Table 1); however, there was statistically significant variation between readings taken from the same subject on different days, ie, the ratio of central aortic and peripheral systolic BP (P = 0.03), AI% (P = 0.02) and AG (P = 0.04) (Table 1). The ratio of central aortic diastolic and peripheral diastolic BP did not differ significantly. Changes in heart rate can influence the morphology of the derived central arterial pulse wave, 22 however, there were only small differences Discussion Applanation tonometry adapted for sphygmocardiography is a method designed for use in a clinician s office, in association with conventional BP recordings by sphygmomanometer. The examination is quick, non-invasive and the radial artery is easily accessible. The present study demonstrates that operators can be easily trained in the use of this technique and that the technique yields excellent inter-observer reproducibility when assessing multiple subjects on different days. In this regard, the technique appears to be well suited for the study of derived central aortic BP indices in multicentre clinical trials. Our finding of significant variation in the relationship between peripheral BP pressure measurements and derived central aortic BP indices in addition to other calculated central haemodynamic parameters is potentially important. It suggests that even in healthy, relatively young subjects, without overt evidence of cardiovascular disease, conventional peripheral BP measurements may be an imperfect surrogate for central aortic BP. This is consistent with previous studies which have shown that the relationship between derived central aortic BP and peripheral BP can vary markedly during exercise 23 and in patients exposed to vasodilator therapy. 14 These conclusions are based on the important assumption that the derived central aortic pressure indices are an accurate reflection of central aortic pressures. Previous studies have examined and validated this relationship by simultaneous direct measurements of central aortic pressures. 3,18 Table 1 Variance of derived central aortic:peripheral systolic and diastolic BPs (%), augmentation index (AIx as %), augmentation (AG in mm Hg), Buckberg ratio (%) and heart rate/min in 25 subjects examined on three different days by two independent observers (AS and CRWK). The significance of variance in day-day values was determined by ANOVA Observer Day 1 Day 2 Day 3 P-value between P-value on 3 two observers on days same day CASBP:PSBP (%) AS 90.0 (4.2) 89.2 (4.6) 88.6 (4.3) NS P = 0.03 CRWK 89.9 (4.3) 88.9 (4.4) 88.7 (4.3) CADBP:PDBP (%) AS 101.8 (1.0) 101.9 (1.0) 101.5 (1.0) NS NS CRWK 101.5 (1.0) 101.7 (1.0) 101.6 (1.1) AIx (%) AS 14.8 (12.7) 12.7 (14.0) 9.3 (14.0) NS P = 0.02 CRWK 12.6 (14.3) 12.8 (14.1) 12.2 (12.0) AG (mm Hg) AS 4.4 (3.9) 4.0 (4.5) 2.8 (4.1) NS P = 0.04 CRWK 3.8 (4.1) 4.1 (4.6) 3.6 (3.3) Buckberg ratio (%) AS 176.3 (24.7) 176.7 (32.5) 172.6 (25.2) NS NS CRWK 172.4 (32.1) 173.5 (29.7) 171.7 (31.0) Heart rate/min AS 67.6 (8.9) 67.7 (9.2) 69.7 (10.2) NS NS CRWK 66.7 (7.6) 67.1 (9.1) 68.3 (8.5) Values are mean and s.d., significant P-value 0.05. CASBP, derived central aortic systolic blood pressure; PSBP, peripheral systolic blood pressure; CADBP, derived central aortic diastolic blood pressure; PDBP, peripheral diastolic blood pressure.

The central aortic wave form synthesised from peripheral pulse wave tonometry and brachial BP provides not only systolic and diastolic central aortic BP but also diastolic and systolic pressure time intervals. Diastolic pressure time interval (DPTI) determines myocardial blood supply and systolic pressure time interval (SPTI) is a determinant of left ventricular load and thus myocardial demand. Thus, the ratio of DPTI/SPTI is an index of myocardial oxygen supply and demand. This viability ratio, the Buckberg ratio, is a sensitive marker of subendocardial ischaemia. 11,20 Hence, additional data regarding the status of the cardiovascular system can be derived from the peripheral arterial pulse wave and our study suggests that these important indices can be reproducibly measured by different observers on repeated occasions. These results have important implications for clinical assessment of patients with hypertension and those predisposed to cardiovascular disease. The significant variance between peripheral and central arterial BPs suggests that the latter may be a better predictor of cardiovascular outcome in patients with hypertension than conventional BP measurement alone. Moreover, the potential for different classes of antihypertensive therapy to differentially effect the relationship between treated peripheral and central BP could provide a basis for differences in efficacy between drug classes in protecting against cardiovascular events in hypertensive subjects. With the availability of simple bedside technology with which central aortic pressures can be accurately and reproducibly measured, it will now be important to define whether central aortic pressure, related derived indices, and their response to antihypertensive therapy are better predictors of cardiovascular outcome than conventional BP measurements. The application of this technology to large scale, randomised clinical trials of antihypertensive therapy and other cardiovascular interventions is now practical and essential to address this important question. Acknowledgements Andrea Siebenhofer was partly supported by a grant of the Styrian government, Department of Research and Technology (Austria). Christian Kemp and the applanation tonometry/sphygmocardiography research programme is supported by a grant from the UK National Health Service Research and Development (Trent Region) programme. References 1 Collins R, MacMahon S. Blood pressure, antihypertensive drug treatment and the risks of coronary heart disease. Br Med Bull 1994; 50: 272 298. 2 Gallagher D. Analysis of pressure wave propagation in the human upper limb: physical determinants and clinical application. [MD]. University of New South Wales, 1994. 3 Karamanoglu M, O Rourke MF, Avolio AP, Kelly RP. An analysis of the relationship between central aortic and peripheral upper limb pressure waves in man. Eur Heart J 1993; 14: 160 167. 4 Messerli FH, Frohlich ED, Ventura HO. Arterial compliance in essential hypertension. Cardiovasc Pharmacol 1985; 7 (Suppl 2): 533 535. 5 O Rourke MF. The arterial pulse in health and disease. Am Heart J 1971; 82: 687 702. 6 Remington JW, Wood EH. Formation of peripheral pulse contour in man. J Applied Physiol 1956; 9: 433 442. 7 O Rourke MF, Gallagher DE. Pulse wave analysis. J Hypertens 1996; 14 (Suppl 5): S147 157. 8 Drzewiecki GM, Melbin J, Noordergraaf A. Arterial tonometry: review and analysis. J Biomech 1983; 16: 141 153. 9 O Rourke MF, Kelly RP. Wave reflection in the systemic circulation and its implications in ventricular function [editorial]. J Hypertens 1993; 11: 327 337. 10 Nichols WW, Avolio AP, O Rourke MF. Ascending aortic impedence patterns in the kangeroo: their explanation and relation to pressure wave forms. Circulation 1986; 59: 247 255. 11 Buckberg GD, Fixler DE, Archi JP, Hoffman JIE. Experimental subendocardioal ischemia in dogs with normal coronary arteries. Circ Res 1972; 30: 67 81. 12 Ferro G et al. Relation between diastolic perfusion time and coronary artery stenosis during stressinduced myocardial ischemia. Circulation 1995; 92: 342 347. 13 Kelly RP, Ting CT, Yang TM. Effective arterial elastance as index of arterial vascular load in humans. Circulation 1992; 86: 513 521. 14 Kelly RP. The pharmocological potential for reversing the ill-effects of aging and arterial hypertension on the central arterial pulse. J Hypertens 1992; 10 (Suppl 6): 97 100. 15 Kelly R et al. Noninvasive registration of the arterial pressure pulse waveform using high-fidelity applanation tonometry. J Vasc Med Biol 1989; 1: 142 149. 16 Kelly R et al. Noninvasive carotid pressure wave registration as an indicator of ascending aortic pressure. J Vasc Med Biol 1989; 1: 241 247. 17 Pauca AL, Wallenhaupt SL, Kon ND, Tucker WY. Does radial artery pressure accurately reflect aortic pressure? Chest 1992; 102: 1193 1198. 18 Chen-Huan C et al. Estimation of central aortic pressure waveform by mathmatical transformation of radial tonometry pressure. Validation of generalized transfer function. Circulation 1997; 95: 1827 1836. 19 Bland JM, Altman DG. Statistical methods for assesing agreement between two methods of clinical measurement. Lancet 1986; 1: 307 310. 20 Buckberg GD et al. Subendocardial ischaemia after cardiopulmonary bypass. J Thorac Cardiovasc Surgery 1972; 64: 669 685. 21 British Standards Institution. Precision of test methods 1: Guide for the determination and reproducability for a standard test method (BS 5497). BSI: London, 1979. 22 Wilkinson IB et al. The effect of heart rate on augmentation index and central arterial pressure. Circulation 1998; 98: 4565. 23 Kroecker EJ, Wood EH. Comparison of simultaneously recorded central and peripheral arterial pressure pulses during rest, exercise and tilted position in man. Circ Res 1955; 3: 623 632. 629