Pulse wave velocity (PWV) is an accepted surrogate marker of arterial stiffness and may be a useful tool for assessing cardiovascular risk in hypertensive patients. The authors sought to compare a novel polyvinylidene fluoride (PVDF) piezoelectric–based sensing device for measuring PWV in the arm with a validated SphygmoCor device (AtCor Medical, West Ryde, Australia) in normal and hypertensive patients. They also sought to compare measured PWV in the forearm (brachial-radial PWV [BRPWV]) with values obtained in the carotid-radial segment (carotid-radial PWV [CRPWV]). Under standardized conditions, CRPWV in 108 normotensive patients with both devices was measured. BRPWV was measured with the PVDF device. Identical measurements were made in a group of 82 hypertensive patients before and after optimization of blood pressure control. Mean CRPWV was 8.7 m/s in the normotensive group and 9.4 m/s in the hypertensive group. Mean BRPWV was 9.2 m/s in the normotensive group and 10.3 m/s in the hypertensive group. There was excellent correlation between the 2 devices when comparing individual CRPWV values (normotensive group, R2=0.92; mean bias 0.04 m/s; hypertensive group, R2=0.89, mean bias 0.08 m/s). Correlation was also favorable when measuring changes in CRPWV in hypertensive patients undergoing pharmacotherapy (PVDF −0.52±0.90 m/s vs SphygmoCor −0.53±1.01 m/s; R2=0.81). Measured values for BRPWV were significantly higher than CRPWV values, and this discrepancy was more marked in the hypertensive group. The PVDF piezoelectric device has excellent correlation with the validated SphygmoCor device when measuring PWV. This novel device may have an important role in patients with conditions such as hypertension.
Left ventricular systole generates a pressure wave that is transmitted throughout the arterial tree and is manifested clinically as the pulse. The velocity of the pressure wave as it propagates via the arterial vessel wall is known as the pulse wave velocity (PWV) and can be measured in a number of vascular beds. The Moens-Korteweg equation states PWV=√(Eh/2 ρ R), where E is Young’s Modulus of the arterial wall, h is wall thickness, R is arterial radius at end-diastole, and ρ is blood density.1 The measurement provides useful information regarding the “stiffness” of the arterial wall, which, in turn, is related to structural and functional factors within the contents of the vessel and the vessel wall. Increased PWV in the aorta has become recognized as a marker of arterial disease in certain pathologic conditions such as essential hypertension,2 renal failure,3 and heart failure.4 In addition, PWV measurement in the arm is useful as a noninvasive assessment of endothelial function and vascular tone.5 However, many physiologic factors modify arterial stiffness, including increasing age,6 blood pressure (BP),7 eating,8 respiration,9 and psychological stress.10 Due to the influence of these variables, standardized guidelines for the measurement of PWV have been published.11
The distending BP within an artery is a major determinant of arterial PWV, and factors that influence BP modify pulse wave transmission times. Therefore, many antihypertensive medications reduce arterial PWV in patients prescribed these drugs, which include calcium channel blockers,12 angiotensin-converting enzyme inhibitors,13 angiotensin II receptor blockers,14α-blockers,15 aldosterone antagonists,16 and some newer generation β-blockers.17
Several devices are commercially available that allow immediate bedside measurement of arterial PWV. The SphygmoCor system (AtCor Medical, West Ryde, Australia) is a widely used modality that uses sequential or serial measurement of the pulse wave via applanation tonometry, gated to the R wave of an electrocardiogram (ECG). This system can measure PWV in the carotid-radial and carotid-femoral segments. The reproducibility of carotid-radial PWV (CRPWV) measurements using this device has been validated.18
A novel CE-approved system for measuring PWV has been developed by Intelesens and the University of Ulster, which uses piezoelectric technology.19 By exposing a polyvinylidene fluoride (PVDF) piezoelectric sensor to the undulations of a peripheral palpable pulse (eg, radial, carotid, brachial, femoral), the first derivative of the pulse wave will be generated. By simultaneous acquisition of two such signals a known distance apart, measurement of the time delay between the two derived pulse waveform peaks (ie, peak-to-peak time delay) will allow the PWV along the arterial segment to be measured (Figure 1). We hypothesized that PWV values derived by this novel method are comparable with values acquired via the SphygmoCor system. Unlike the SphygmoCor system, however, the PVDF device is small and portable, with dimensions of 10 cm × 6 cm × 3 cm, and a weight of approximately 200 g. The novel system is also advantageous in that it may be easily modified to measure PWV in a variety of vascular beds in addition to the carotid-radial and carotid-femoral arterial segments.
We sought to compare measurement of PWV as assessed by the novel PVDF device with that derived from the validated SphygmoCor system. We chose carotid-radial segment for comparison purposes in a normal and a hypertensive cohort. We also sought to assess the validity of measuring PWV along the more easily accessible brachial-radial segment. The rationale for this was to assess whether these values are comparable and therefore a surrogate for PWV measurements in the longer carotid-radial segment.
A cohort of normal patients was recruited. Patients were identified as normal if they had no pre-existing cardiovascular disease, diabetes, or hypertension and were taking no regular medication. Normal patients attended on a single occasion. Written informed consent was obtained. They had been asked to fast for a minimum of 6 hours and abstain from smoking for a minimum of 3 hours prior to attendance. Demographic details were collated and physiologic parameters were measured by a single investigator. After a 15-minute period of relaxation in a quiet room, BP was measured in the right arm by a calibrated semi-automatic device (Omron M5-I; Omron Healthcare, Hoofddorp, the Netherlands). The mean of the second and third readings was accepted as the measured BP. Thereafter, PWV was measured in the right arm of the supine patient.
Applanation Tonometry (SphygmoCor)
Measurements of CRPWV were recorded with the patient lying supine on a bed with the head slightly extended and the right arm relaxed with the palm facing up. The strongest foci of the right carotid and right radial pulsations were identified and marked. The distance from the suprasternal notch to both these pulsations were measured20 and entered into the SphygmoCor software. The difference between these two measurements corresponds to the distance covered by the carotid-radial pulse wave time delay. Applanation tonometry was carried out sequentially on the carotid artery and thereafter on the radial artery, with the derived pulse waveforms gated to the R wave of an ECG. A CRPWV value was automatically generated by the SphygmoCor device.
The SphygmoCor system is not validated for BRPWV assessments and the available software does not allow this measurement to be calculated. Therefore, pulse wave transit time was not assessed in the brachial-radial arterial segment using the SphygmoCor system.
PVDF Piezoelectric Device
Measurements of CRPWV were recorded with the patient lying supine on a bed with the head slightly extended and the right arm relaxed with the palm facing up. The strongest foci of the right carotid and right radial pulsations were identified. A piezoelectric sensor was placed over these peripheral pulsations and fixed in place with adhesive tape. These piezoelectric sensors were attached to the PVDF device by low weight electrical leads. The distances from the suprasternal notch to the piezoelectric sensors were measured and entered into the PVDF piezoelectric software. The difference between these two measurements corresponds to the distance covered by the carotid-radial pulse wave time delay. Simultaneous acquisition of carotid and radial pulse waveforms were recorded during a 10-second period and the mean CRPWV was calculated via automated software.
Measurement of BRPWV via the PVDF piezoelectric device was recorded in a similar manner to CRPWV. However, in this instance, the piezoelectric sensors were placed on the strongest foci of the brachial and radial pulsations. The direct distance between the two sensors was measured and entered into the PVDF piezoelectric software. Simultaneous acquisition of brachial and radial pulse waveforms were recorded during a 10-second period and the mean BRPWV was calculated via automated software (Figure 1).
CRPWV was measured via the PVDF piezoelectric device and applanation tonometry with a mean of two sequential measurements taken as the accepted CRPWV, unless the two values differed by ≥10%, in which case a third measurement was taken. In these instances, the mean of the 3 readings was accepted as the measured CRPWV.11 Brachial-radial PWV (BRPWV) was measured via the PVDF piezoelectric device with a mean of 5 sequential readings accepted as the measured BRPWV.
Patients with a new diagnosis of essential hypertension or patients with existing hypertension and suboptimal BP control were recruited.
Patients who were already taking antihypertensive medications were adjudged suitable for inclusion if their summary BP was in the range >140/85 mm Hg (or >130/80 mm Hg if diabetic). Individuals who were not taking medication were assessed on the basis of their BP. If BP ≥180/110 mm Hg on two occasions 7 to 14 days apart, they were considered suitable for inclusion. If BP >160/100 mm Hg on two occasions, ≥4 weeks apart, they were adjudged suitable for inclusion. If BP 140 mm Hg to 159/90 to 99 mm Hg was confirmed on two occasions 4 weeks apart and there was evidence of target organ damage, patients were considered suitable for recruitment.21
Hypertensive patients attended for PWV measurements on two occasions. On the first occasion, written informed consent was obtained. Demographic details and physiologic, BP, and PWV measurements were taken as outlined above in the healthy population. After this index meeting, all patients received lifestyle advice, and antihypertensive medications were prescribed or optimized. British Hypertension Society guidelines were used for choice of antihypertensive medication.7 Further appointments to assess response to therapy were organized and, if applicable, further up-titration of medications was carried out. Target BP was <140/85 mm Hg in nondiabetic patients. In diabetic patients, the target BP was <130/80 mm Hg.
After a minimum of 3 months, the patients were asked to return for study measurements on a second occasion. Measurement techniques at this meeting were identical to those used at the index assessment.
Ethical approval for this study was obtained from Queens University Belfast ethics committee.
Paired Student t test was used for comparison of means. Stepwise multivariate regression was used for assessment of independent predictors of CRPWV. Scatterplot analysis and Bland-Altman plots were used for method comparison. Mean bias values derived by Bland-Altman plots are reported in the text as mean ± standard deviation. P<.05 was accepted as statistically significant.
A total of 108 patients were recruited for the study. All patients were Caucasian. Details are outlined in Table I. Fifty-seven (52.8%) participants were women and 51 (47.2%) were men. The mean age was 44.2±13.0 years, with a range of 18.2 to 79.4 years. Fourteen (13.0%) were current smokers and 57 (52.8%) had a family history of premature cardiovascular disease. None had diabetes or a history of cardiovascular disease. Mean systolic BP (SBP) within the group was 133.8±20.4 mm Hg. Mean diastolic BP (DBP) was 82.1±12.4 mm Hg. SBP was seen to increase modestly with advancing age (r=0.3, P=.002). DBP also increased with age (r=0.36, P<.001).
Table I. Demographic Details and Physiologic and PWV Measurements of Normal and Hypertensive Cohorts
Distance covered by the carotid-radial pulse wave time delay, cm
61.3±4.2 (range 50.0–71.0)
61.4±3.7 (range 54.0–71.0)
Brachial-radial distance, cm
24.5±1.6 (range 20.5–28.5)
24.3±1.8 (range 20.5–28.5)
CRPWV SphygmoCor, m/s
CRPWV PVDF device, m/s
BRPWV PVDF device, m/s
Mean CRPWV in the normal patients, as measured by the SphygmoCor system was 8.70±1.16 m/s (range: 6.50–12.45 m/s). Mean CRPWV as assessed by the piezoelectric device was 8.70±1.20 m/s (range: 6.24–12.40 m/s). There was a strong correlation of the values obtained between the two systems (R2=0.92, P<.001). On Bland-Altman analysis, the mean bias between the PVDF and SphygmoCor systems was +0.04±0.35 m/s (Figure 2A).
Stepwise multivariate regression was performed separately on CRPWV as derived by the SphygmoCor device and the PVDF piezoelectric device. Variables assessed were age, sex, smoking history, history of premature cardiovascular disease, height, weight, body mass index, abdominal girth, pulse, DBP, SBP, pulse pressure, and distance over which PWV was measured (CR distance). Using measurements derived via the SphygmoCor only, age and DBP were identified as independent predictors of CRPWV in this cohort (R2 adjusted=0.75). Similarly, age and DBP were identified as the only independent predictors of CRPWV when assessing measurements derived by only the PVDF piezoelectric device (R2 adjusted=0.81).
Mean SphygmoCor CRPWV value was 8.70±1.16 m/s (range: 6.50–12.45 m/s). Mean BRPWV as derived by the PVDF device was 9.22±1.22 (range: 6.59–13.27). A strong correlation existed between the two devices. (R2=0.72, P<.001), although the BRPWV values were significantly faster than the CRPWV values (Bland-Altman mean bias +0.53±0.66 m/s) (Figure 2B).
Eighty-two patients were recruited for the study, of which 42 (51.2%) were women and 40 (48.8%) were men. The mean age was 57.5±12.5 years. Other demographic values are outlined in Table I.
Mean SBP was 159.8±22.1 mm Hg and mean DBP was 86.5±14.1 mm Hg. There were 64 patients already established on antihypertensive therapy at the start of the study, with 20 taking 1, 22 taking 2, 17 taking 3, and 5 taking 4 medications. The remaining 18 patients were not on regular antihypertensive therapy.
All patients received education on lifestyle measures and the importance of compliance with their medication. Thirty-one patients were prescribed ≥1 additional antihypertensive medication, with subsequent up-titration of doses if required. Twenty-four patients had up-titration of existing drug dosages. Six patients had a drug discontinued and replaced with a more appropriate drug class. There were 18 patients in whom no change in drug regimen was made during the study period. This was due to adequate BP response to lifestyle measures and education (n=13) or patients declining to comply with changes to their drug regimen (n=5). Three patients were noted to have discontinued ≥1 of their medications at follow-up visit. These changes were contrary to medical advice.
At the end of the study period, the 82 patients were established on a mean of 2.1 antihypertensive agents. Thirty-two patients were taking an ACE inhibitor (perindopril, 14; ramipril, 12; lisinopril, 3; trandolapril, 2; captopril, 1), 32 were taking an angiotensin receptor blocker (irbesartan, 17; candesartan, 8; losartan, 2; olmesartan, 3; valsartan, 2), 29 were taking a β-blocker (bisoprolol, 11; atenolol, 6; nebivolol, 10; propanolol, 2), 40 were taking a calcium channel blocker (amlodipine, 23; diltiazem, 9; lercanidipine, 6; felodipine, 2), 30 were taking a diuretic (bendroflumethiazide, 21; hydrochlorothiazide, 9), and 8 were taking other agents (doxazocin, 7; monoxidine, 1).
Changes in measured parameters between visit 1 and visit 2 are outlined in Table II. There were significant changes noted in SBP, DBP, CRPWV, and BRPWV. The mean change in BP during the study period was −15.9/−5.9 mm Hg.
Table II. Change in Physiologic Measurements in the Hypertensive Cohort After Optimization of Blood Pressure Control
Distance covered by the carotid-radial pulse wave time delay, cm
Brachial-radial distance, cm
At visit 1, mean CRPWV in this hypertensive group as measured by the SphygmoCor system was 9.4±1.2 m/s (range, 7.3–13.2 m/s). Mean CRPWV as assessed by the piezoelectric device was 9.4±1.1 m/s (range, 7.7–12.8 m/s). At the second visit, mean CRPWV in the hypertensive group as measured by the SphygmoCor system was 8.8±0.9 m/s (range, 5.4–10.9 m/s). Mean CRPWV as assessed by the piezoelectric device was 8.9±0.8 m/s (range, 6.6–11.1 m/s). When comparing all measurements taken, a strong correlation between the two systems was observed (R2=0.89, P<.001). On Bland-Altman analysis, the mean bias between the PVDF and SphygmoCor systems was +0.08±0.38 m/s. (Figure 3A).
Mean BRPWV on the first attendance as derived by the PVDF device was 10.3±1.4 m/s (range, 8.1–15.8 m/s). Mean BRPWV on the second attendance as derived by the PVDF device was 9.6±1.1 m/s (range, 6.4–13.1 m/s). When comparing all BRPWV measurements with all SphygmoCor CRPWV measurements, a modest correlation existed between the two devices (R2=0.38, P<.001), with the BRPWV values significantly faster than the CRPWV values (Bland-Altman mean bias, +0.9±1.1 m/s) (Figure 3B).
When comparing changes in CRPWV between visit 1 and visit 2, there was an excellent correlation observed between the two devices (mean change SphygmoCor CRPWV, −0.53±1.01 m/s; mean change PVDF CRPWV, −0.52±0.90 m/s; R2=0.81, P<.001, Bland-Altman mean bias, +0.01±0.44 m/s) (Figure 4A). Comparison between the two devices weakened when comparing changes in BRPWV with changes in CRPWV (mean change SphygmoCor CRPWV, −0.53±1.01 m/s; mean change PVDF BRPWV, −0.73±1.35 m/s; R2=0.49, P<.001, Bland-Altman mean bias, −0.2±1.0 m/s) (Figure 4B).
Increasing interest in assessing pulse wave hemodynamics as a potential risk stratification tool in patients with hypertension and other disease states has led to the development of a number of different modalities for measuring PWV. The novel PVDF piezoelectric device is advantageous in that it is a small, portable, and user-friendly appliance that can be used to measure PWV in a number of different arterial segments. This study now also demonstrates that calculated PWV values using the novel PVDF device has excellent correlation with the popular, validated SphygmoCor system.
The range of CRPWV values within our normal cohort is similar to previous published data on PWV in the arm. The average CRPWV of 8.7 m/s compares favorably with other investigators.7,18 We found age and DBP to be the independent predictors of CRPWV in a Northern Irish population. This is similar to previous published studies regarding the major determinants of PWV in the arm.2 The range of BRPWV as determined by the PVDF piezoelectric device in our normal population also compares favorably with previous published values for PWV in this vascular bed.2,22
Within the hypertensive cohort, mean CRPWV as assessed by both the SphygmoCor system and the PVDF piezoelectric device was 9.4 m/s, and this value reflects well with other published data on this arterial segment in hypertensive patients.7 Using the PVDF system, we obtained a value for BRPWV of 10.3 m/s in our hypertensive cohort. This value is slightly lower than the previously published data for this arterial segment,2,22 although this may reflect the lower BP values when compared with the previous studies and the number of patients taking antihypertensive agents in our study group.
When comparing CRPWV values derived by the PVDF piezoelectric device with SphygmoCor values, an excellent correlation was noted in both the normotensive (R2=0.92) and hypertensive (R2=0.89) cohorts. On Bland-Altman plots, the mean bias in the normotensive group was +0.04 m/s and in the hypertensive cohort was +0.08 m/s, with the PVDF system generating slightly higher values (Figure 2A and Figure 3A). This difference may be considered trivial. The reason for the marginally increased values may be due to the mode of acquiring the carotid pulse wave form. Whereas the SphygmoCor system directly applanates the carotid artery, the piezoelectric sensor records the carotid pulsation as it travels through the subcutaneous and cutaneous tissues. This mechanoelectrical delay may be extremely short, but it may be significant enough to generate slightly faster PWV, particularly as there is negligible mechanoelectrical delay at the more superficial radial artery. It is noteworthy that the mean bias in the normotensive group was +0.04 m/s and +0.08 m/s in the hypertensive group. The hypertensive cohort had higher recorded body mass indices than the normotensive group and therefore was likely to have significantly more subcutaneous tissue in the neck. This arguably may have led to relatively larger mechanoelectrical delays.
The PVDF piezoelectric device also compared well with the SphygmoCor system when assessing for changes in CRPWV on serial measurements in hypertensive patients undergoing optimization of BP control. There was strong correlation on scatterplot analysis (R2=0.81), with a mean bias of 0.01 m/s, demonstrating excellent concordance between the two devices (Figure 4A).
In this study, when comparing BRPWV as assessed by the PVDF piezoelectric system with CRPWV as assessed by the SphygmoCor or piezoelectric systems, the correlation between the two arterial segments was less favorable. This may have a physiologic basis. In the normotensive cohort, the correlation on scatterplot between the two methods was good (R2=0.72) (Figure 2B). However, when performing the same analysis in the hypertensive cohort, the correlation was seen to become less robust (R2=0.38) (Figure 3B). This discrepancy between the two cohorts was also replicated in the Bland-Altman analysis (mean bias, 0.53 m/s vs 0.90 m/s). The finding that the PWV is faster in the brachioradial segment when compared with the carotid-radial segment is perhaps not surprising. The Moens-Korteweg equation states that PWV is inversely proportional to the radius of the artery in diastole. The radial artery is known to be of smaller diameter than the axillary and brachial vessels. It is also well established that PWV is not uniform throughout the arterial tree, tending to increase with increasing distance from the heart. This has been demonstrated in both the aorta and lower limb.23,24 Our study demonstrates that this relationship has now also been established in the arm.
The larger variation in BRPWV values relative to the CRPWV in the hypertensive group may also have a physiologic basis. Unrecognized tortuosity in this arterial segment, along with the inherent difficulty in measuring PWV over a short distance may both be important in this regard. Measuring PWV over a relatively short distance will invariably increase the risk of error. BRPWV was assessed over a distance of 20.5 cm to 28.5 cm, compared with CRPWV, which, in our study, was measured over a distance of 50.0 cm to 71.0 cm. Attempts were made to negate this effect by increasing the number of readings in the brachial-radial segment (5 vs 2 in the carotid-radial segment), although it is arguable whether this was effective in significantly reducing the risk of error.
Arterial tortuosity is caused by arteriosclerosis, with increasing age and hypertension predisposing to the phenomenon. The hypertensive cohort in the study had higher BP than the normotensive group, but they were also appreciably older. These two factors therefore increased the likelihood of tortuosity in the hypertensive group. Tortuosity is not uniform throughout the length of an artery, and therefore tortuosity in the axillary artery in an individual may not be reflective of the degree of tortuosity in the radial artery, which may be normal. This may be an additional factor that explains the degree of variation in the BRPWV when compared with CRPWV.
The larger discrepancy between BRPWV and CRPWV in the hypertensive cohort compared with the normotensive cohort may also be due to the significantly heavier patients in the hypertensive cohort when compared with the normotensive group. With increased subcutaneous tissue between the brachial artery and the brachial sensor in the hypertensive cohort, this may have caused a degree of mechanoelectrical delay. This would have had significant implications for measured BRPWV values, which were calculated over a relatively short distance.
It should be noted that although the BRPWV measurement did not correlate well with CRPWV, particularly in the hypertensive cohort, this does not automatically render the measurement as valueless. It could be argued that as the radial artery demonstrates little or no tortuosity, the BRPWV may be a better measure of muscular artery stiffness than CRPWV, which may be more vulnerable to measurement inaccuracies related to unrecognized tortuosity.
Although we have discussed the possible physiologic reasons for the differences noted in values derived by the two devices, we acknowledge that we cannot exclude the possibility that the difference noted between the two measurement techniques may be simply due to technical issues related to the novel piezoelectric device.
The novel PVDF piezoelectric device compares favorably with the validated SphygmoCor system. Values obtained for BRPWV and CRPWV via the PVDF are consistent with previously published data. There is excellent correlation between the two devices when measuring CRPWV in both a normal and hypertensive population. BRPWV can be accurately measured using the novel piezoelectric device but does not appear to be an acceptable surrogate of CRPWV, particularly in hypertensive patients. This is likely due to pathophysiologic changes throughout this muscular arterial segment in the arm, coupled with the inherent difficulties in measuring velocities over short distances. We have shown that the novel PVDF piezoelectric device accurately measures PWV in the arm. The adaptability of the sensors to be easily modified and sited on any peripheral pulsation is innovative and should prove advantageous to researchers studying PWV in a number of different vascular beds.
Disclosures: The authors have no conflicts of interest to declare. Funding for this study was provided by the Heart Trust Fund.