Comparison of Non-Invasive Estimation of Central Aortic Blood Pressure and Biomarkers of Vascular Health Measured by Two Devices Using Brachial Cuff Waveforms

Abstract

Introduction

The pulse wave analysis technology in the specialist SphygmoCor XCEL device has now been implemented in the CONNEQT Pulse device, which is a conventional sphygmomanometer suitable for home blood pressure (BP) monitoring. The aim of this study was to compare the correlation and consistency of measurements between the two devices.

Methods

The study was conducted in participants admitted to the Department of Geriatrics, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine. Brachial BP (bBP), central BP (cBP), augmentation pressure (AP), augmentation index (AIx), and subendocardial viability ratio (SEVR) were measured with the CONNEQT Pulse and the SphygmoCor XCEL devices, both using a calibrated cuff-based brachial waveform.

Results

A total of 367 eligible subjects, aged 23–92 (57.6 ± 13.0) years, of whom 254 (69.2%) were male, was enrolled in this study. Paired t tests showed relatively low bias and elevated standard deviation: cSBP mm Hg ([121.7 ± 17.2] vs. [117.5 ± 14.5], p < 0.001), cDBP ([80.6 ± 11.0] vs. [78.3 ± 11.4], p < 0.001), cPP ([41.0 ± 12.7] vs. [39.2 ± 9.5], p < 0.001), AP mm Hg ([11.5 ± 6.2] vs. [12.8 ± 6.6], p < 0.001), AI ([26.4 ± 9.0]% vs. [31.9 ± 14.0]%, p < 0.01), and SEVR ([135.6 ± 25.5]% vs. [137.8 ± 26.4]%, p = 0.02). cDBP (r = 0.812) measured by both devices was strongly correlated, cSBP (r = 0.762), cPP (r = 0.702) and SEVR (r = 0.778) were strongly correlated, whereas AP (r = 0.587) and AIx (r = 0.522) were moderately correlated (all p < 0.001). For intraclass correlation coefficient (ICC) for both devices, cDBP (ICC = 0.794) and SEVR (ICC = 0.776) showed good reproducibility, while cSBP (ICC = 0.727), cPP (ICC = 0.666), AIx (ICC = 0.428), and AP (ICC = 0.575) showed good to modest reproducibility. The differences between cSBP, cDBP, cPP, AP, AIx, and SEVR measured by the two devices were 4.2 ± 11.2 mm Hg, 2.3 ± 6.9 mm Hg, −1.8 ± 9.0 mm Hg, −1.3 ± 5.8 mm Hg, −5.6 ± 12.0%, and 2.1 ± 17.3%, respectively.

Conclusion

Both devices show strong cBP-bBP correlations and low bias, suggesting that the new CONNEQT Pulse provides similar measurements of cBP to the original SphygmoCor XCEL. Although there is greater variability in waveform parameters, the magnitude of differences is within acceptable limits for meaningful use in conventional home BP monitoring. The relatively elevated variability in measurement precision (standard deviation), most likely due to sequential measurements with both devices in the same arm, needs to be further evaluated using simultaneous measurements in both arms.

Introduction

Central aortic blood pressure (cBP) reflects the pressure afterload on the ejecting left ventricle and is an improved predictor of subclinical cardiovascular disease compared to conventional brachial blood pressure (bBP). cBP also improves determination of risk of end-organ damage to critical organs such as the brain, kidneys and heart [1]. Augmentation of systolic pressure (AP), a pressure wave reflection leading to increased central aortic systolic pressure, is associated with increased cardiovascular risk and morbidity and mortality and can be a marker of arterial stiffness [2]. Augmentation index (AIx) is a marker of vascular damage and has been shown to be a potential independent predictor of cardiovascular risk and mortality [3]. Subendocardial viability ratio (SEVR), also known as the Buckberg index, is an important parameter for assessing myocardial perfusion as it reflects the balance between myocardial oxygen supply and demand, a non-invasive surrogate for coronary microvascular function [4]. It can effectively assess the function of coronary microcirculation and correlate with arterial stiffness and is also potentially valuable in predicting adverse cardiovascular events and mortality in different populations [5].

cBP measurements involve direct and indirect methods. The direct method, in which a pressure sensing device is inserted directly into the aortic root during cardiac catheterization, is the gold standard for measuring cBP, but is difficult to obtain in daily clinical practice given its invasive nature. The indirect method is the derivation of cBP and other indices by non-invasive peripheral artery (brachial or radial) pulse wave analysis using a transfer function [6]. Currently, the most commonly used device is the SphygmoCor pulse wave analysis system (Cardiex, AtCor Medical, Sydney, Australia). The original SphygmoCor device employed radial tonometry, and which formed the standard technique. For ease of use, the radial waveform was replaced by a brachial waveform obtained from the oscillations in the brachial cuff as employed in SphygmoCor XCEL. This technology has been applied to the latest device using the transfer function technique in the CONNEQT Pulse device (Cardiex) an easy-to-use, all-in-one device that measures bBP, cBP, and a variety of other pulse wave biomarkers of vascular health via pulse wave analysis, facilitating remote patient monitoring and at-home self-monitoring. It is a conventional sphygmomanometer augmented by pulse wave analysis and cloud connectivity. CONNEQT Pulse device has been cleared by the FDA in 2023, and validated by peer-reviewed arterial health and cardiovascular risk assessment studies [7]. The aim of this study was to investigate the consistency of two non-invasive measurement devices (CONNEQT vs. SphygmoCor XCEL) for measuring cBP and other biomarkers of vascular health. The original SphygmoCor’s radial tonometry device was followed by the more convenient SphygmoCor XCEL device employing a brachial cuff waveform. However, both are bulky items used mainly in specialist clinical research environments and population studies. The latest CONNEQT Pulse is a much lower cost device which is essentially a conventional sphygmomanometer which can be used as a home monitoring device, and with cloud communication features promises to collect large amounts of data on blood pressure (BP) and associated pulse waveform indices in the population at large. Since there is a legacy of a large amount of data available for the SphygmoCor XCEL device, it is important to perform a head-to-head comparison of both devices to establish comparison limits in a relatively large cohort to ensure continuity in potential longitudinal studies and clinical decision-making.

Methods

Subjects

This study has been reviewed and approved by the Medical Research Ethics Committee 2023-127 of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine. There were 396 inpatients in the Geriatrics Department of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine from April 2023 to February 2025. Written informed consent was obtained from all subjects. The inclusion criteria were age ≥18 years old, male and female ratio not less than 30% of the total number of subjects, who voluntarily took part in this clinical study, complied with the study protocol, and signed an informed consent form. The following were excluded: (i) arrhythmia, including atrial fibrillation, tachycardia/bradycardia, and arrhythmia; (ii) history of acute myocardial infarction within 3 months; (iii) history of stroke, including cerebral haemorrhage, cerebral infarction, and undifferentiated stroke, within 3 months; (iv) valvular disease; (v) pulselessness; (vi) recent injuries or deformity of the arm and wrist. A total of 29 subjects did not undergo SphygmoCor pulse wave analysis testing. Analysis of the present study was performed in 367 subjects.

Measurements

All subjects were asked to refrain from consuming alcohol, coffee and strong tea on the day of examination. Measurements were performed more than 2 h after a meal. Prior to measurements, medical history was collected and general information (name, gender, height, total body weight, history of hypertension) were recorded. After the subjects rested for 5–10 min in a quiet state, the test was completed by the same operator who was qualified in technical training and quality assessment. For each subject, the sequence of the measurement with the two devices was randomized to avoid any systematic bias. For a particular sequence (e.g., SphygmoCor XCEL followed by CONNEQT pulse), subjects were placed in the supine position and the SphygmoCor XCEL cuff placed around the left arm at heart level. BP was measured by the device during the deflation phase. The reading of SBP/DBP was used to calibrate the brachial cuff waveform obtained during a 5 s period of partial cuff inflation. The SphygmoCor technology uses a validated generalized transfer functions to estimate the aortic pulse waveform from a peripheral (radial or brachial) signal [6]. cSBP and cDBP are derived from the transformed central waveform. AP is calculated as the difference between the second and first systolic peaks. AIx is defined as AP divided by cPP and expressed as a percentage. SEVR is computed as the ratio of diastolic pressure-time integral to systolic pressure-time integral, reflecting myocardial oxygen supply-demand balance. In the study protocol, there was a rest period of 1.5 min before measurement by CONNEQT Pulse to allow the blood circulation in the arm to recover. The cuff was positioned on the left arm at the same level as the heart. At the completion of a BP measurement, CONNEQT Pulse automatically performs a cBP measurement by recording the brachial cuff pulse waves during a partial cuff inflation. The central aortic pressure waveform in the CONNEQT Pulse device is obtained using the same transfer function (generalized transfer function) as used in the SphygmoCor XCEL device. The SphygmoCor XCEL algorithm is an extension of the original transfer function developed for the radial artery, and the performance of the device using the brachial cuff waveform has been shown to be similar to that obtained using the radial waveform [8]. AP, AIx and SEVR were then calculated as described above.

Data Processing

Statistical analyses were performed using SPSS 26.0 and MedCalc, and the measured data were expressed as mean ± standard deviation (SD). Paired t tests, Pearson correlation analyses, intraclass correlation coefficient (ICC), and Bland-Altman plotting were used to evaluate the correlation and consistency of the indices measured by the two devices. For each parameter, the mean of the paired differences (CONNEQT minus SphygmoCor) was calculated to estimate the systematic bias. The SD of these paired differences was then computed. The 95% limits of agreement were defined as bias±1.96 × SD of the differences.

Results

General Information of Subjects

A cohort of 367 volunteers was enrolled, of whom 254 (69.2%) were male, aged 23–92 (57.6 ± 13.0) years, body weight of 70.8 ± 12.3 kg, and body mass index (BMI) of 25.5 ± 6.7 kg/m². A total of 228 cases had hypertension (62.1%), 107 cases (29.2%) had type 2 diabetes mellitus, and 119 cases (32.4%) had hyperlipidaemia. There were 216 cases (58.9%) taking antihypertensive drugs (including 175 cases of angiotensin-converting enzyme inhibitors or angiotensin receptor antagonists (81.0%), 88 cases of calcium channel antagonists (40.7%), 51 cases of β-blockers (23.6%), 31 cases of α-blockers (14.4%), and 22 cases of diuretics (10.2%). Their general characteristics are shown in Table 1.

Table 1.

Baseline characteristics

​	n = 367
Age, years	57.6±13.0
Male, n (%)	254 (69.2)
Body weight, kg	70.8±12.3
BMI, kg/m2	25.5±6.7
Hypertension, n (%)	228 (62.1)
Type 2 diabetes mellitus, n (%)	107 (29.2)
Hyperlipidaemia, n (%)	119 (32.4)
Antihypertensive drugs, n (%)	216 (58.9)
ACE inhibitors/ARBs, n (%)	175 (81.0)
Calcium channel antagonists, n (%)	88 (40.7)
β-Blockers, n (%)	51 (23.6)
α-Blockers, n (%)	31 (14.4)
Diuretics, n (%)	22 (10.2)

Correlation between bBP and cBP Measured by Both Devices

Paired sample t tests of brachial and central arterial pressures measured by the two devices showed that bSBP and bPP were all significantly higher than cSBP and cPP (p < 0.001), whereas bDBP was all slightly lower than cDBP (p < 0.001), as shown in Table 2. Pearson correlation analyses showed that CONNEQT Pulse and SphygmoCor XCEL measurements of cSBP, cDBP, and cPP were all correspondingly correlated with bSBP, bDBP, and bPP with r = 0.970, 0.991, 0.937 and r = 0.975, 0.996, 0.948 (p < 0.001), which were extremely strong correlations (Fig. 1). The ICC of SBP and DBP measured by the two devices were 0.861, 0.989 and 0.781, 0.991, respectively, showing significant reproducibility, while the reproducibility of PP was good (ICC was 0.740, 0.582, respectively) (Table 3).

Table 2.

BP, AP, AIx, and SEVR measured by both devices

​	CONNEQT pulse	SphygmoCor XCEL	p value
HR, beats/min	71.4±10.6	70.3±10.7	<0.001
bSBP, mm Hg	130.4±17.4	128.2±16.6	<0.001
bDBP, mm Hg	80.0±10.8	77.2±11.3	<0.001
bPP, mm Hg	50.4±13.0	51.0±12.2	0.22
cSBP, mm Hg	121.6±17.2	117.5±14.5	<0.001
cDBP, mm Hg	80.6±11.0	78.3±11.4	<0.001
cPP, mm Hg	41.0±12.7	39.2±9.5	<0.001
AP, mm Hg	11.5±6.2	12.8±6.6	<0.001
AIx, %	26.4±9.0	32.0±14.0	<0.001
SEVR, %	135.6±25.5	137.8±26.4	0.02

Fig. 1.

Scatterplot of cBP measured by two devices (n = 367): CONNEQT Pulse (a–c) and SphygmoCor XCEL (d–f).

Table 3.

Paired sample t test and Pearson correlation analysis of BP, AP, AIx, and SEVR measured by the two devices

​	Mean of difference, mm Hg	SD of difference, mm Hg	Pearson’s r	ICC
bBP-cBP
CONNEQT pulse
SBP	8.7	4.3	0.970	0.861
DBP	−0.6	1.5	0.991	0.989
PP	9.4	4.6	0.937	0.740
SphygmoCor XCEL
SBP	10.7	4.1	0.975	0.781
DBP	−1.1	1	0.996	0.991
PP	11.8	4.4	0.948	0.582
CONNEQT pulse-SphygmoCor XCEL
cSBP	4.2	11.2	0.762	0.727
cDBP	2.3	6.9	0.812	0.794
cPP	1.8	9.0	0.702	0.666
AP	−1.3	5.8	0.522	0.428
AIx	−5.6%	12.0%	0.587	0.575
SEVR	−2.1%	17.3%	0.778	0.776

Evaluation of the Agreement of cBP Measured by the Two Devices

The cSBP [(121.7 ± 17.2) vs. (117.5 ± 14.5) mm Hg], cDBP [(80.6 ± 11.0) vs. (78.3 ± 11.4) mm Hg], and cPP [(41.0 ± 12.7) vs. (39.2 ± 9.5) mm Hg] obtained by CONNEQT Pulse and SphygmoCor XCEL were subjected to a paired sample t test, and all differences were found to be statistically significant (p < 0.001), although all were below 4.2 mm Hg.

Pearson’s correlation analysis showed that cDBP (r = 0.812) measured by the two devices was strongly correlated. cSBP (r = 0.762) and cPP (r = 0.702) were strongly correlated (shown in Fig. 2a). cDBP had an ICC of 0.794, which was very reproducible, and cSBP and cPP had ICCs of 0.727 and 0.666, respectively, which were good reproducibility, as shown in Table 3.

Fig. 2.

Correlation and consistency of cBP and biomarkers of vascular health measured by two devices (n = 367). a Scatterplot of cBP measured by both devices. b Bland-Altman plots of cSBP, cDBP, and cPP measured by two devices. c Scatterplot of arterial stiffness indices (AP, AIx, SEVR) measured by two devices. d Bland-Altman plots of AP, AIx, and SEVR measured by two devices.

Plotting the Bland-Altman plots of cSBP, cDBP, and cPP in 367 subjects showed that the mean differences in cSBP, cDBP, and cPP obtained by CONNEQT Pulse and SphygmoCor XCEL were 4.2, 2.3, and 1.7 mm Hg, with 95% CIs for the differences of (−17.8–26.1), (−11.1–15.8), and (−16.9–20.3) mm Hg, with 6.3% (23/367), 3.8% (14/367), and 4.1% (15/367) points outside the 95% CI range, respectively; within the limits of concordance, the maximum absolute value of the differences in cSBP, cDBP, and cPP measured by CONNEQT Pulse compared with SphygmoCor XCEL is 26, 17, and 20 mm Hg, respectively (shown in Fig. 2b).

Evaluation of the Agreement of AP, AIx, and SEVR Measured by Two Devices

Paired sample t tests revealed that the AP [(11.5 ± 6.2) mm Hg vs. (12.8 ± 6.6) mm Hg, p < 0.001], AIx [(26.4 ± 9.0)% vs. (31.9 ± 14.0)%, p < 0.001], and SEVR [(135.6 ± 25.5) mm Hg vs. (137.8 ± 26.4) mm Hg, p < 0.05] were statistically significant.

Pearson’s correlation analysis showed that AP (r = 0.587) and AIx (r = 0.522) measured by the two devices were moderately correlated while SEVR (r = 0.778) was strongly correlated (both p < 0.001) (shown in Fig. 2c). The ICCs of 0.428 and 0.575 for AP and AIx, respectively, were well reproduced, while the ICC for SEVR was 0.776, with good reproducibility (Table 3).

Bland-Altman plots of AP, AIx, and SEVR in 367 subjects showed that the mean differences in AP, AIx, and SEVR obtained by CONNEQT Pulse and SphygmoCor XCEL were −1.3 mm Hg, −5.6%, and −2.1%, respectively, and the 95% CIs for the differences were (−12.7–10.2) mm Hg, (−29.2–18.1)%, and (−36.1–31.8)%, with 5.2% (19/367), 2.7% (10/367), and 3.5% (13/367) of the points outside the 95% CI range, respectively; within the limits of consistency, the absolute value of the differences in AP, AIx, and SEVR measured by CONNEQT compared with SphygmoCor maximum of 12 mm Hg, 29%, and 36%, respectively (shown in Fig. 2d).

Discussion

Both devices showed the expected differences in central and peripheral systolic pressure with CONNEQT Pulse (8.7 ± 4.3 mm Hg) showing a slight (non-significant) difference compared to the SphygmoCor XCEL (10.7 ± 4.1 mm Hg), associated with a significant difference in cPP (9.4 ± 4.6 vs. 11.8 ± 4.4 mm Hg). These differences were comparable but slightly lower than differences found between cSBP and bSBP measured in a previous study by Bui et al. [9] using radial tonometry (12 ± 4 mm Hg). This is generally consistent with previous findings by direct manometry that arterial pressure is progressively amplified from the central aorta to the periphery by SBP and PP, whereas DBP remains essentially unchanged [10].

Pulse pressure amplification describes the progressive widening of PP as the pressure wave propagates from the central aorta to peripheral muscular arteries along the arterial tree [11]. It is worth noting that some studies have shown that bSBP overestimates cSBP mostly in young and middle-aged, low-risk populations, while different relationships may be found in the elderly and high-risk patients [12]. A systematic review of 1,598 patients included in 24 studies comparing central and peripheral BP gradients obtained by arterial cannulation found that cSBP was significantly higher than bSBP, with a mean gradient of 8.0 mm Hg [13]. Significant heterogeneity in the sensitivity analyses of SBP suggests that SBP gradients may vary considerably in different patient populations and clinical settings, and that more research is necessary to identify the specific factors that contribute to SBP variability.

The cBP measured by the CONNEQT and SphygmoCor devices correlated well. The results of correlation analysis showed that cSBP, cDBP, and cPP measured by the two devices were strongly and positively correlated, and the ICC was 0.727, 0.794, and 0.666, respectively, and it is generally believed that the ICC <0.40 indicates poor reproducibility, the ICC 0.40∼<0.75 indicates good reproducibility, and the ICC ≥0.75 indicates very good reproducibility, so it can be concluded that central aortic pressures measured by the two devices are comparable. Bland-Altman analysis revealed clinically larger differences in our study (cSBP: 4.2 ± 11.2 mm Hg; cDBP: 2.3 ± 6.9 mm Hg) than the study by Butlin et al. [8] (cSBP: 0.5 ± 1.8 mm Hg; cDBP: −0.01 ± 0.5 mm Hg), which were obtained by SphygmoCor XCEL and SphygmoCor using radial tonometry. While ICC values indicated moderate-to-good reproducibility, the wide limits of agreement suggest substantial inter-device variability at the individual level. A potential explanation for this greater variability and difference in the precisions of the measurement (reflected by the SD), is the sequential nature of the measurements with the two devices on the same arm. It is conceivable that although the randomization of measurement in the same arm may reduce any underlying systematic bias in the measurement, including potential effects of inter-arm difference, the variation in physiological change in BP may contribute to the range of differences found between the two devices, even though there was a period of rest between the measurements for equilibration. It would be of interest to compare similar evaluation studies done with device randomization in both arms, with the expectation that the effects of time-dependent physiological variability would be reduced. In addition, our study had a higher proportion of elderly patients, whereas the cohort in the study of Butlin et al. [8] spanned a broader age range (22–79 years). Advanced arterial stiffness in older adults may amplify waveform reflection complexity, affecting inter-device agreement.

The present study did not use an invasive method to directly measure cBP in subject pairs, and it was not possible to directly compare the accuracy of each of the two non-invasive measures. Based on the results of previous validation studies of the SphygmoCor device, there was no statistically significant difference in cDBP when compared with direct measurement by the invasive catheter method, and non-invasive cDBP can be a good substitute for invasive cDBP [14]. The underestimation of cSBP by SphygmoCor XCEL may be due to various factors, mainly being the underestimation of the calibration brachial measured as measured by the conventional sphygmomanometer [15]. Zuo et al. [16] showed that the original SphygmoCor (using radial tonometry) underestimated cSBP 1.5–13.3 mm Hg approximately. Similar to other studies, this range of variation is attributed to the non-invasive calibration effects of the brachial sphygmomanometer, which generally underestimates intra-arterial systolic pressure.

The correlation between AIx (r = 0.522) measured by both devices was lower than the consistency between CONNEQT Pulse and SphygmoCor XCEL (r² = 0.82), suggesting that waveform feature extraction sensitivity may vary by device type [8], or the accuracy of placement of the cuff, which may affect the fidelity of the brachial waveform used for analysis. From the Bland-Altman plot, the difference in AP was −1.3 ± 5.8 mm Hg. However, compared to the findings of Butlin et al. [8], the AIx difference in this study (−5.6 ± 12.0%) was larger than their reported difference (1.8 ± 7.0%). One possible reason is that heart rate correction was not performed in this study and CONNEQT Pulse did not provide correction for an AIx@75 for a heart rate of 75 beats/min. Although small, a statistically significant difference (p < 0.001) was observed between heart rates measured by SphygmoCor (71.4 ± 10.6 bpm) and CONNEQT (70.3 ± 10.7 bpm). AIx is negatively correlated with heart rate, and an increase in heart rate reduces the absolute duration of cardiac contraction, effectively shifting the reflected wave to diastole and thus reducing AIx.

A strong correlation (r = 0.778) was found between the SEVR measured by the two devices. The Bland-Altman plot shows the difference in SEVR was −2.1 ± 17.3%, possibly due to heart rate. Heart rate has some negative effects on SEVR, with increased heart rate resulting in decreased SEVR values. Cardiac diastolic time has a high dependence on heart rate, and the faster the heart rate, the longer the relative cardiac systolic time and the shorter the cardiac diastolic time, which leads to a shortening of coronary artery perfusion time, and a reduction in the SEVR value reflecting myocardial perfusion. The correlation coefficients of the SEVR measured by CONNEQT Pulse and SphygmoCor XCEL with their respective HRs were −0.684 (p < 0.001) and −0.712 (p < 0.001), respectively, showing a significant negative correlation. Currently, SphygmoCor is the most commonly used method to detect SEVR. Butlin et al. [8] demonstrated high consistency in SEVR between SphygmoCor XCEL and radial tonometry SphygmoCor (R² = 0.82). Previous studies have shown that SEVR, estimated by the systolic pressure-time integral and diastolic pressure-time integral is more accurate than traditional methods that do not take intracardiac diastolic pressure into account [17].

The findings of the study show generally good agreement between the devices, although they need to be interpreted in the context of the study limitations. The cohort studied was mainly elderly; hence, there may have been potential measurement errors related to placement of arm cuffs in frail patients. This would contribute to increased variability in the differences. The sequential measurements in the same arm, notwithstanding the device randomization and rest period, may have also contributed to intra-device variability. Moreover, BP was only measured once with CONNEQT Pulse, not assessing intra-device repeatability.

In conclusion, the new CONNEQT Pulse device demonstrates similar performance to SphygmoCor XCEL, with good correlation and low bias for central aortic systolic and diastolic pressure measurements from the brachial cuff pulse waveform. There is also good agreement for waveform indices, although with elevated variability, which in this study may be partly related to methodological issues.

Statement of Ethics

The study protocol was reviewed and approved by the Ethics Committee of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine (2023-127). Written informed consent was obtained from participants to participate in the study.

Conflict of Interest Statement

Alberto Avolio was a member of the journal’s Editorial Board at the time of submission. The authors have no other conflicts of interest to declare.

Funding Sources

The study was funded by the Shanghai Municipal Health Commission (20234Y2139) and Shanghai Municipal Commission of Science and Technology (23015820100).

Author Contributions

Xinyue Zhou, Xibao Shi, and Yueliang Hu contributed equally to this work. They were involved in the study design, data collection, and analysis. Biwen Tang and Yakun Li assisted with data collection and manuscript preparation. Mark Butlin and Alberto Avolio provided expertise in pulse wave velocity measurements and contributed to the interpretation of results. Junli Zuo supervised the study, provided critical insights, and revised the manuscript. All authors reviewed and approved the final version of the manuscript.

Funding Statement

The study was funded by the Shanghai Municipal Health Commission (20234Y2139) and Shanghai Municipal Commission of Science and Technology (23015820100).

Data Availability Statement

The data that support the findings of this study are not publicly available due to containing information that could compromise the privacy of research participants but are available from the corresponding author, Junli Zuo (zuo-junli@163.com), upon reasonable request.