Automated pulse wave velocity assessment using a professional oscillometric office blood pressure monitor

Abstract

Carotid‐femoral pulse wave velocity (cfPWV) is the gold standard method for assessing arterial stiffness. This study evaluated automated brachial‐ankle PWV (baPWV) taken by a professional oscillometric blood pressure monitor (Microlife WatchBP Office Vascular) versus reference cfPWV (Complior device). Subjects recruited from a hypertension outpatient clinic had duplicate baPWV and cfPWV measurements (randomized crossover design) and carotid ultrasonography. Of 102 subjects recruited, 101 had valid baPWV measurements. Four subjects were excluded and 97 were analyzed (age 58.3 ± 11.4 years, men 70%, hypertensives 76%, diabetics 17%, cardiovascular disease 10%, smokers 23%). The mean difference between baPWV (13.1 ± 1.8 m/s) and cfPWV (9.1 ± 1.8 m/s) was 4.0 ± 1.4 m/s (P < .01) with close association between them (r = 0.70, P < .01). baPWV and cfPWV were correlated with age (r 0.54/0.49 respectively), systolic blood pressure (0.45/0.50), carotid intima‐media thickness (0.31/0.44), and carotid distensibility coefficient (−0.47/−0.34) (all P < .05; no difference between the two methods, z test). There was reasonable agreement (77%) between the two methods in identifying subjects at the top quartile of their distributions (kappa 0.39, P < .01). The areas under the receiver operating characteristic curves for the identification of carotid plaques were comparable for cfPWV and baPWV (0.79 and 0.74 respectively, P = NS). Automated baPWV measurement by a professional oscillometric blood pressure monitor is feasible and observer‐independent. baPWV values differ from those by cfPWV, yet they are closely correlated, have reasonable agreement in detecting increased arterial stiffness and give similar associations with carotid stiffness and atherosclerosis.

1. INTRODUCTION

Stiffness of the large arteries is closely associated with aging and hypertension and is an independent predictor of cardiovascular morbidity and mortality. ¹ , ² , ³ , ⁴ , ⁵ Carotid‐femoral pulse wave velocity (cfPWV) is the gold standard for measuring large artery stiffness ¹ and several studies have established the additive value of cfPWV in terms of risk stratification. ⁵ , ⁶ The European Society of Cardiology/European Society of Hypertension has endorsed cfPWV for assessing hypertension‐mediated organ damage. ¹ However, wide use of cfPWV measurement is impractical due to considerable technological and methodological heterogeneity in its assessment and other issues related to device cost, observer time, and experience, patient inconvenience in femoral pulse acquisition, etc, and thus, it is not recommended for routine use. ¹

In the last years, there is accumulating evidence regarding the automated measurement of brachial‐ankle pulse wave velocity (baPWV) which is based on an at least 2‐limb cuff oscillometry or plethysmography. ⁷ Specifically, baPWV is calculated as the ratio of the virtual arterial path length derived from the subject's height and the time difference between the commencements of systolic increases in brachial and ankle pressure waves. ⁷ The automated and observer‐independent nature of the baPWV measurement are particularly attractive for wide application. Moreover, accumulating evidence from Asian populations, mainly Japan, suggests that baPWV independently predicts the risk of developing cardiovascular disease. ⁸ However, head to head comparison studies of baPWV and cfPWV (the current reference method) in terms of diagnostic agreement and cardiovascular damage prediction are lacking.

This study evaluated automated baPWV measured using a professional oscillometric blood pressure (BP) monitor versus reference cfPWV in terms of: (a) difference in absolute values, (b) agreement in detecting high cardiovascular risk, (c) correlation with cardiovascular risk factors and damage.

2. METHODS

2.1. Participants

Ambulatory subjects attending a Hypertension Clinic were recruited in a cross‐sectional study. Healthy subjects and others with cardiovascular risk factors and/or disease (treated and/or untreated) aged 25‐80 years were included. Exclusion criteria were conditions in which measurement of cfPWV is considered unreliable, such as arrhythmias, unstable clinical condition, carotid artery stenosis >70%, carotid sinus syndrome. ²

2.2. Test device

The Microlife WatchBP Office Vascular is a professional automated oscillometric upper‐arm cuff BP monitor for office use, which has been validated for BP measurement accuracy. ⁹ , ¹⁰ , ¹¹ In addition, the device provides automated measurements of ankle‐brachial index (ABI), which have been previously validated. ¹² The device has been further developed to assess baPWV by simultaneous non‐invasive assessment of brachial and ankle BP and respective pressure waveforms analysis. For baPWV measurement the ABI mode is selected, and the individual's body height is input before measurement. Then a regular automated BP measurement is taken with the cuff pressure remaining at 50‐70 mmHg for about 10 heartbeats. The time difference between the arm and the ankle is calculated by averaging data from 10 heartbeats. The device calculates baPWV as distance/Tba, with distance calculated from the distance between the brachial and ankle sampling points and Tba from the average time interval of the pressure waves between the sampling points.

2.3. Procedure

The study was performed in a quiet room with stable temperature and after the participants remained for 10 minutes in supine position. The following measurements were performed: (a) measurement of carotid intima‐media thickness (cIMT) and carotid distensibility coefficient, (b) duplicate baPWV and cfPWV in randomized order. In line with guidelines for cfPWV assessment, all measurements were made on the right side (right carotid artery ultrasonography; right common carotid, and right common femoral for cfPWV; right arm, and right foot for baPWV). ² Participants were instructed to avoid meal, caffeine or smoking for 3 hours before measurements. Talking and sleeping were not allowed during the entire procedure. Cuff with appropriate size according to the individuals’ right arm and ankle circumference was used (medium 22‐32 cm; large 32‐42 cm).

Carotid ultrasonography (GE Logiq C5 Premium) was performed for the following assessments at the right side: (a) cIMT was calculated using B‐mode ultrasonography and a 10 MHz probe by automated software allowing multiple measurements in an about 1‐cm segment across common carotid artery (1 cm proximal to the carotid bulb), and in an about 0.5‐cm segment of the carotid bulb (average of all values was used in the analysis); (b) The presence of carotid plaques was also recorded across common carotid, carotid bulb, internal and external carotid artery. Plaques were defined as focal structures encroaching into the arterial lumen of at least 0.5 mm or 50% of the surrounding IMT value, or demonstrating a thickness >1.5 mm as measured from the intima‐lumen interface to the media‐adventitia interface ¹³ ; (c) M‐mode tracing of the common carotid artery diameter during 3 cardiac cycles was performed for the calculation of the carotid distensibility coefficient ¹⁴ taking into account the central carotid BP assessed by the Complior Analyse device (Alam Medical, France).

Duplicate sequential cfPWV (Complior Analyse) and baPWV (Microlife WatchBP Office Vascular) measurements were performed in each participant in a randomized order. Before each cfPWV assessment, right brachial BP measurement was obtained by the Microlife BP monitor to be used for the Complior Analyse device (Alam Medical, France) calibration. The cfPWV measurements were performed with the Complior Analyse device using simultaneous recording of 2 pressure waves over the right common carotid artery and over the right femoral artery. The 80% of the straight distance between carotid and femoral arteries was used as pulse wave traveled distance. ² The average of the 2 PWV assessments for each method was used in the analysis. If the difference between the two cfPWV measurements was higher than 0.5 m/s, a third measurement was taken and the median value of all was used in line with current recommendations. ² Likewise, if the difference between the two baPWV measurements was higher than 0.7 m/s, a third measurement was taken and the median value of all was used.

2.4. Study power and sample size calculation

The primary endpoint of this study was to investigate the correlation between cfPWV and baPWV. In order to show a correlation coefficient r 0.7 between baPWV and cfPWV, being significantly stronger than the r 0.4 (threshold between weak and moderate coefficient), with study power 90% and alpha error probability .05, a total of 91 subjects would be required. Allowing for a drop out of 10%, a total of 100 subjects should be recruited.

2.5. Statistical analysis

The Kolmogorov‐Smirnov statistic was performed to test continuous variables for normality. For the assessment of the bivariate relationship between the examined variables, Pearson correlations coefficients (r) were determined provided that variables had normal distribution. Comparison of correlation coefficients was performed using Steiger's z test. ¹⁵ Linear regression analysis was performed for the estimation of the regression equation between baPWV (dependent variable) and cfPWV (independent). The kappa statistic was used to evaluate the diagnostic agreement between the two methods in the identification of subjects at the top quartile of their distributions. Chi‐square test was used for comparison of the prevalence of carotid plaques across quartiles. Receiver Operating Characteristic (ROC) curves for both cfPWV and baPWV in terms of carotid plaques detection were constructed and areas under the ROC curves were compared using Medcal.net with the method by Hanley & McNeil. ¹⁶ Results are expressed as mean values with SD. Statistical analysis was performed using the IBM SPSS Statistics (Version 21.0. Armonk, NY IBM Corp). A P‐value of .05 was considered statistically significant.

3. RESULTS

Of 102 subjects recruited, valid automated baPWV measurements were obtained in 101. Four had peripheral artery disease (ankle‐brachial index < 0.9). These subjects were excluded from the analysis as the posterior‐tibial artery pressure waveform analysis might be faulty. Indeed, these subjects had very low difference between baPWV and cfPWV (1.2 ± 2 m/s). A total of 97 subjects were analyzed and their characteristics are shown in Table 1.

TABLE 1.

Characteristics of the study participants

Characteristics
N	97
Age (years)	58.3 ± 11.4
Males (%)	68 (70)
Body mass index (kg/m2)	28.0 ± 3.9
Hypertension (%)	74 (76)
Treated for hypertension (%)	64 (66)
Cardiovascular disease (%)	10 (10)
Diabetes (%)	16 (17)
Smoking (%)	22 (23)
Systolic blood pressure (mmHg)	134.9 ± 16.2
Diastolic blood pressure (mmHg)	79.8 ± 10.4
Carotid IMT (μm)	834 ± 147
Carotid plaque (%)	56 (58)
Carotid distensibility coefficient (10‐3*kPa‐1)	29.3 ± 12.0
Carotid‐femoral PWV (m/s)	9.1 ± 1.8
Brachial‐ankle PWV (m/s)	13.1 ± 1.8
Ankle‐brachial index	1.25 ± 0.1

Abbreviations: IMT, intima‐media thickness; PWV, pulse wave velocity.

Average baPWV (13.1 ± 1.8) was higher than cfPWV (9.1 ± 1.8 m/s) by 4.0 ± 1.4 m/s (P < .01), yet with close association between them (r = 0.70, P < .01) (Figure 1). Linear regression analysis provided the following formula for the estimation of baPWV from cfPWV: baPWV = 6.64 + 0.71 * cfPWV (R ² = 0.50). Bland‐Altman plot is shown in Figure 2.

FIGURE 1.

Correlation between brachial‐ankle and carotid‐femoral pulse wave velocity

FIGURE 2.

Bland‐Altman scatterplots of the brachial‐ankle versus carotid‐femoral pulse wave velocity differences against their average value

The correlation coefficient between the 2 cfPWV assessments was r = 0.96 and for baPWV r = 0.85 (all P < .05; P < .05 for comparison). A total of 9 errors occurred with baPWV measurement in 8 subjects. A third measurement for cfPWV was required in 19 subjects and for baPWV in 18 subjects (4 subjects required both). Both baPWV and cfPWV were closely correlated with age, systolic blood pressure, carotid intima‐media thickness and carotid distensibility coefficient (all P < .05; no differences between the two methods, z test) (Figure 3).

FIGURE 3.

Correlation of brachial‐ankle and carotid‐femoral pulse wave velocity with age, brachial systolic blood pressure, and indices of carotid stiffness and atherosclerosis. BP, blood pressure; DC, distensibility coefficient; IMT, intima‐media thickness

The agreement between the two methods in the identification of subjects at the top quartile of their distributions (cfPWV ≥ 10.3 m/s; baPWV ≥ 14.1 m/s) was 77% (kappa 0.39, P < .01). The areas under the ROC curves for cfPWV and baPWV were comparable for identification of carotid plaques (0.79; 95% confidence intervals [CI]: 0.70, 0.88, and 0.74; 95% CI: 0.64, 0.84 respectively, z‐statistic: 0.88, P = NS) (Figure 4). The prevalence of subjects with carotid plaques increased across both cfPWV and baPWV quartiles in a similar pattern (P < .01 for both; P = NS for comparison of prevalence between the 2 methods within each quartile; Figure 5).

FIGURE 4.

Receiver operating characteristic curves of brachial‐ankle and carotid‐femoral pulse wave velocity for the detection of carotid plaques. AUC, area under the curve

FIGURE 5.

Prevalence of subjects with carotid plaques across quartiles of brachial‐ankle versus carotid‐femoral pulse wave velocity. baPWV, brachial‐ankle pulse wave velocity; cfPWV, carotid‐femoral pulse wave velocity

One observer was needed to perform baPWV measurements and two observers for cfPWV. The average time required for duplicate automated baPWV measurement was 6.9 ± 0.8 minutes (measurement of arm and ankle circumference; placement of cuffs; input of body height value in the device; 2 measurements) versus 6.7 ± 1.2 minutes for cfPWV (marking common carotid and femoral artery sites; measurement of carotid‐femoral distance; BP measurement; input of variables in Complior software; 2 measurements; P = NS).

4. DISCUSSION

This study compared PWV measurements taken by a professional automated oscillometric BP measuring device (baPWV) versus the reference cfPWV method. The main findings of this study are: (a) baPWV measurement obtained by the Microlife WatchBP Office Vascular device was feasible, observer‐independent, and easy to perform with single observer; (b) baPWV values were at about 40% higher than cfPWV; (c) baPWV and cfPWV were closely associated and presented reasonable agreement in detecting individuals with increased arterial stiffness; (d) baPWV, as much as cfPWV, was associated with cardiovascular risk factors and indices of carotid stiffness and atherosclerosis and presented similar predictive value for the detection of carotid plaques, which demonstrates the clinical relevance of baPWV measurements in risk stratification.

The measurement of baPWV presented feasibility rate at 99%. The measurement is taken by a single observer, it is fully automated (observer‐independent) and the only requirement is the arm and ankle cuff placement without need to expose the inguinal area for femoral pulse palpation which is inconvenient for the patient. The time required for performing duplicate measurements using the two methods was comparable, yet 2 observers are required for cfPWV and one of them should have experience in using the method which is observer dependent. The ease of performance and the automated and observer‐independent nature of measurement makes baPWV assessment appropriate for widespread use in primary care settings.

In this study sample, which mainly included middle‐aged treated hypertensive patients, baPWV values were about 40% higher than cfPWV. This difference is not unexpected, as it is known that baPWV incorporates a longer arterial distance and contains components of both peripheral (muscular type) and central (aortic) arterial stiffness. The average difference between baPWV and cfPWV in this study was about 4 m/s.

Most of the published studies comparing cfPWV with baPWV have been conducted in Asia and USA and reported differences between cfPWV and baPWV ranging from 2 m/s to 9 m/s. ¹⁷ , ¹⁸ This is probably due to heterogeneity in technology and methodology applied for cfPWV measurement, as well as the characteristics of the studied populations. It is well recognized that the arterial distance taken into account in the calculation of the cfPWV plays a major role in defining the cfPWV values. ² , ¹⁹ The distance recommended for cfPWV (0.8*[common carotid artery‐common femoral artery]) has been used in only three comparative studies. Strasser et al compared baPWV and cfPWV in a general population sample and reported a difference of 2.6 m/s (cfPWV 10.4 m/s; baPWV measured by VP‐1000 plus, Omron Healthcare, Bannockburn, IL). ²⁰ Tanaka et al reported a difference of 4.6 m/s in a sample of healthy Japanese and USA individuals (cfPWV 7.8 m/s; baPWV measured by VP‐2000; Omron Healthcare, Kyoto, Japan). ²¹ Last, Kumagai et al reported a 5.6 m/s difference (cfPWV 9.1 m/s; baPWV measured by form PWV/ABI; Colin Medical Technology, Komaki, Japan) in a study including only men. ²² Thus, it appears that baPWV is consistently higher than the cfPWV, yet the evidence on the absolute baPWV value and the magnitude of the cfPWV‐baPWV difference presents considerable heterogeneity. Reference baPWV values have been reported to be higher ²³ or slightly lower ²⁴ than the values reported in this study. Differences in the equations used in the different devices, as well as differences in the sample characteristics (age, ethnicity, cardiovascular risk factors, arterial wall properties, drug treatment, etc), might have accounted for these differences. This observed heterogeneity mandates that each such automated baPWV device should be compared versus the current reference method and at least until a reference baPWV device is developed and agreed to be regarded as reference.

There are several findings in the present study which are reassuring that baPWV reflects mainly central aortic stiffness and might be a useful alternative to cfPWV. First, baPWV was closely associated with cfPWV (r 0.70). Second, baPWV and cfPWV presented reasonable agreement (77%) in detecting subjects at the top quartiles of their distribution (subjects with increased arterial stiffness). Third, baPWV and cfPWV assess different arterial segments but are associated with the same ongoing systematic pathophysiology; thus, these indices were similarly and closely associated with age and systolic BP, which are major determinants of arterial stiffness (Figure 3). Fourth, baPWV and cfPWV were similarly and closely associated with carotid damage (Figure 3) and had similar predictive value in detecting carotid plaques (Figures 4 and 5). These findings are reassuring for the clinical utility and relevance of the baPWV measurement.

Interestingly, the threshold for the top quartile of the cfPWV distribution in this sample was at 10.3 m/s, which is close to the recommended PWV threshold for hypertension‐mediated arterial damage. ¹ The corresponding threshold for the top quartile in the baPWV distribution was at 14.1 m/s. Thus, this value might represent the threshold for increased arterial stiffness using this method. However, these preliminary findings require verification in larger studies in the community.

In patients with peripheral artery disease (defined by ABI < 0.9), baPWV measurement appeared to provide low values, which were close to those by cfPWV measurements. However, in these patients with considerable deterioration in peripheral artery perfusion, the analysis of the tibial pressure waveform is problematic. Thus, when the Microlife device provides ABI values of less than 0.9, baPWV measurements should be ignored. Indeed, PWV measurement is unnecessary in these patients with advanced arterial damage, which are already classified at very high cardiovascular risk.

These findings should be interpreted in light of several limitations. First, the results are derived from subjects attending a BP clinic and may not be generalized to other populations. However, these results are in line with those reported by community‐based studies examining other baPWV methodologies. ¹⁷ , ¹⁸ Second, although the correlation and the agreement between the two methods are satisfactory, they are still imperfect. This is mainly attributed to the different methodologies and technologies used, but also to the imperfect reproducibility of the reference cfPWV assessed by either the Complior or the SphygmoCor device. ²⁵ Third, the associations between cfPWV, baPWV, and carotid damage are at least in part driven by their strong association with age as a part of pathophysiology.

In conclusion, these data suggest that automated baPWV measurement using a professional oscillometric BP monitor is feasible, observer‐independent, and strongly associated with reference cfPWV, risk factors for arterial stiffness and the presence of arterial damage. Thus, the test device appears to be an attractive tool in clinical practice for the evaluation of apparently low or moderate risk patients. Future community‐based studies are needed to confirm the present findings and to define normalcy thresholds.

CONFLICT OF INTEREST

GSS received consultation fees by Microlife for the development of blood pressure measurement technologies. For other authors nothing to declare.

AUTHOR CONTRIBUTIONS

Anastasios Kollias and George S Stergiou: designed and conceptualized the study, acquired the data, analyzed and interpreted the data, drafted the manuscript, and revised the manuscript. Konstantinos G Kyriakoulis, Areti Gravvani, and Ioannis Anagnostopoulos: acquired the data, analyzed and interpreted the data, and revised the manuscript.

Kollias A, Kyriakoulis KG, Gravvani A, Anagnostopoulos I, Stergiou GS. Automated pulse wave velocity assessment using a professional oscillometric office blood pressure monitor. J Clin Hypertens. 2020;22:1817–1823. 10.1111/jch.13966