Twenty‐Four‐Hour Central Ambulatory Blood Pressure: Relationship With Measures of Cardiovascular Subclinical Organ Damage in Healthy Individuals

Abstract

Background

In most, but not all, observational studies, ambulatory central systolic blood pressure (cSBP) is more closely related to left ventricular mass index than ambulatory brachial SBP. However, the association with other markers of subclinical organ damage is poorly understood. We investigated the association between ambulatory brachial SBP or cSBP and measures of subclinical organ damage, together with the influence of waveform calibration method, in a large community‐based population of untreated individuals.

Methods

In all, 1311 participants (mean age, 45±18 years; 589 women) had simultaneous measurements of ambulatory brachial and central pressure over 24 hours. Of these, left ventricular mass index was assessed in 675 individuals, carotid intima‐media thickness in 610 individuals and carotid–femoral (aortic) pulse‐wave velocity in 1091 individuals.

Results

Left ventricular mass index was most strongly associated with cSBP, calibrated using mean and diastolic blood pressure, with a β coefficient of 0.237 (95% CI, 0.225–0.453; P<0.001). The adjusted (partial) correlation coefficient for this association was also significantly higher versus both brachial SBP and cSBP derived from calibration using SBP and diastolic blood pressure (P<0.001 for both). The same was true for carotid intima‐media thickness (β coefficient=0.141 [95% CI, 0.777–2.502]; P<0.001), although the adjusted correlation coefficients did not differ significantly between ambulatory blood pressure indices. In contrast, aortic pulse‐wave velocity shared a similar association with both brachial and central ambulatory pressure, with no obvious effect of waveform calibration method.

Conclusions

Ambulatory cSBP may provide valuable additional information concerning cardiovascular risk, although the method of waveform calibration exerts a marked impact.

Nonstandard Abbreviations and Acronyms

aPWV

aortic pulse‐wave velocity

bSBP

brachial systolic blood pressure

CIMT

carotid intima‐media thickness

cPP

central pulse pressure

cSBP

central systolic blood pressure

LVMI

left ventricular mass index

PP

pulse pressure

PWV

pulse‐wave velocity

Clinical Perspective.

What Is New?

• Our study is the first to investigate the relationship between ambulatory central and brachial blood pressure and 3 distinct indices of subclinical organ damage within a large, healthy population at a single center.

What Are the Clinical Implications?

• Our findings provide insight into how noninvasive ambulatory central blood pressure relates to key markers of subclinical organ damage compared with ambulatory brachial blood pressure.

• This information may help clinicians select the most appropriate risk stratification approach when assessing patients with hypertension.

Ambulatory brachial blood pressure (BP) has superior prognostic power over brachial BP measured in the clinic. ¹ , ² , ³ Similarly, central BP measured in the clinic is superior to brachial BP in its prediction of cardiovascular events and death. ⁴ , ⁵ , ⁶ With recent advances in technology, central ambulatory BP can now be derived noninvasively, using a number of commercially available devices. However, limited data are available comparing ambulatory brachial and central BP values, in terms of clinical superiority. Moreover, the method by which brachial waveforms are calibrated to yield noninvasive estimates of central BP appears to exert a marked effect on central SBP (cSBP) and pulse pressure (PP) values, yet the clinical significance of this effect is uncertain.

Two existing studies suggest that left ventricular mass index (LVMI) has a stronger correlation with ambulatory cSBP than ambulatory brachial SBP (bSBP). ⁷ , ⁸ Interestingly, these studies also demonstrated that the method used to calibrate arterial pressure waveforms in the derivation of central BP values was important. Indeed, correlations between LVMI and cSBP were stronger when pressure waveforms were calibrated with brachial mean arterial pressure (MAP) and diastolic blood pressure (DBP; cSBP_MAP/DBP) versus those calibrated with brachial SBP and DBP (cSBP_SBP/DBP). However, further studies are required to confirm these findings and explore relationships with other surrogate markers of cardiovascular disease.

A small study of 136 adolescents and young adults correlated LVMI, carotid intima‐media thickness (CIMT), and 24‐hour estimated pulse‐wave velocity (PWV) with 24‐hour brachial pressure and 24‐hour central pressure values obtained using both waveform calibration methods. Again, the results differed depending on which calibration method was used. For example, cSBP_MAP/DBP exhibited stronger correlations with LVMI and CIMT in adolescents but not in young adults. In contrast, 24‐hour estimated PWV showed a closer association with 24‐hour brachial than central SBP, derived from either calibration method. ⁹ Carotid IMT also correlated more closely with ambulatory cSBP than bSBP in 501 individuals in the SAFAR (Non‐Invasive Aortic Ambulatory Blood Pressure Monitoring for the Detection of Target Organ Damage) study, ¹⁰ whereas ECG voltages (indicating left ventricular hypertrophy) were similarly associated with bSBP and cSBP in 177 men, ¹¹ although the impact of calibration method was not examined in either of these studies.

Further investigation of the clinical utility of ambulatory central BP is clearly required, as is the impact of calibration method on relationships between ambulatory central BP and markers of subclinical organ damage. Therefore, we compared the associations between ambulatory brachial and central BP and markers of subclinical organ damage (LVMI, CIMT, aortic PWV [aPWV]) in a large, healthy, untreated population of adults covering a wide age span and recruited from a single center, ensuring homogeneous methods.

Methods

Subjects

Participants were drawn from the Anglo–Cardiff Collaborative Trial and from hypertension clinics within Addenbrooke’s Hospital (Cambridge, UK). Separate analyses were conducted in subsets of participants in whom LVMI, CIMT, and aPWV had been assessed at the time of ambulatory central BP measurement. All participants were free of cardiovascular disease and medication. Local research ethics committee approval was obtained, and written informed consent was given by all participants. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Twenty‐Four‐Hour Ambulatory BP Measurements

Clinic (seated) BP was assessed in duplicate (triplicate if >5 mm Hg difference), in the nondominant arm, using a validated oscillometric sphygmomanometer (HEM‐705CP; Omron Corporation, Kyoto, Japan). The average of the recorded readings was used as the final value in the analyses. Ambulatory BP monitoring was then undertaken for 24 hours, using the Mobil‐O‐Graph device (I.E.M. GmbH, Stolberg, Germany), which uses the ARCSolver algorithm to estimate cSBP, by transforming brachial pressure waveforms obtained by oscillometry. Two methods of brachial pressure waveform calibration were used to yield values of cSBP: calibration using brachial SBP and DBP (cSBP_SBP/DBP) and calibration using brachial mean arterial pressure (MAP) and DBP (cSBP_MAP/DBP). The device was programmed to record BP and arterial waveforms at 30‐minute intervals during the day and 60‐minute intervals overnight as per usual clinical practice in the United Kingdom. The final value for each time period (daytime, nighttime, and 24 hours) was calculated as the average of all recorded readings in that time period. Study participants were instructed to undertake normal activities but to refrain from strenuous exercise during the 24‐hour monitoring period and to remain still, with the arm relaxed, during cuff inflation and deflation. The ambulatory monitoring period was considered valid if it contained >70% of successful BP readings, according to the system software. If not, a second period of monitoring was undertaken.

Assessments of Subclinical Organ Damage

Left Ventricular Mass Index

A standardized transthoracic echocardiogram was performed using a commercially available vivid Q ultrasound system (GE Healthcare, Chicago, IL) with a 1.5‐ to 3.6‐MHz phased array transducer. Left ventricular internal dimension and wall thicknesses were measured at end‐diastole and end‐systole by American Society of Echocardiography recommendations using a computerized review station. End‐diastolic left ventricular septal and posterior wall thicknesses and internal dimensions were used to calculate left ventricular mass by a validated formula: left ventricular mass=1.04×0.8 [(left ventricular wall thicknesses+internal dimension)–(internal dimension)]+0.6 g. LVMI was calculated using the Mosteller formula. ¹²

Carotid Intima‐Media Thickness

The common carotid artery was identified and scanned using high‐resolution ultrasound (MyLab Gold, Esaote, Genoa, Italy) with a 10‐MHz linear probe. B‐mode images were obtained 1 cm from the bulb and far wall intima‐media thickness calculated using automated system software. The larger of the 2 values (left or right) was used in the analysis.

Aortic PWV

aPWV was measured using the SphygmoCor device (Atcor Medical, Sydney, Australia) by sequentially recording ECG‐gated carotid and femoral artery waveforms, as previously described. ¹³ Path length for the determination of aPWV was measured, using a tape measure, as the surface distance between the suprasternal notch and femoral site minus the distance between the suprasternal notch and carotid site.

Statistical Analysis

All data were analyzed using SPSS software (version 27; SPSS, Chicago, IL). Normality of data distributions was checked using the Kolmogorov–Smirnov test. Univariable associations between measures of subclinical organ damage and ambulatory BP indices were assessed using Pearson correlation coefficients. Following this, Steiger’s method ¹⁴ was applied to compare, statistically, the relative strength of these univariable associations. Post hoc analyses indicated that the study sample for each measure of subclinical organ damage provided sufficient power for our univariable analyses. Multivariable regression models were then constructed using the “Enter” method, to assess independent associations between measures of subclinical organ damage and ambulatory BP indices. Other variables included in the models were age, sex, body mass index (BMI), and the averaged heart rate (HR) corresponding to each measurement period (24 hours, daytime, and nighttime). Variable selection was a priori on the basis of known or expected associations. Finally, the relative strength of the multivariable‐adjusted associations was assessed by deriving adjusted (partial) correlation coefficients (ie, correlations between measures of subclinical organ damage and ambulatory BP indices after adjusting for age, sex, BMI, and HR at the time of measurement) and comparing these following Steiger’s method. ¹⁴ Data are presented as mean±SD, unless otherwise stated, and a P value of <0.05 was considered significant.

Results

Twenty‐four‐hour ambulatory central BP data were available in 1311 individuals. Their demographic characteristics and office and ambulatory BP data are summarized in Table 1, together with data on measures of subclinical organ damage. Demographic characteristics and BP data are presented for each of the 3 participant subsets in Table S1. There were no significant differences observed between demographic subgroups and the overall study population. In the study population overall, ambulatory BP levels were lower than seated clinic‐based readings. However, the method of waveform calibration used to derive ambulatory cSBP readings had a marked effect on the values obtained. While cSBP_SBP/DBP was lower than corresponding brachial values over 24 hours and during daytime and nighttime, cSBP_MAP/DBP was actually higher than brachial values, especially during the night.

Table 1.

Demographic Characteristics, Measures of Cardiovascular Subclinical Organ Damage, and BP Values Within the Whole Study Population (N=1311)

Variable
Demographic parameters
Age, y	44.6±17.6
Sex, male; female	605 (46.1); 706 (53.8)
Height, m	1.70±0.09
Weight, kg	75.5±16.7
BMI, kg/m2	26.0±4.9
Smoker	282 (23.3)
Measures of subclinical organ damage
LVMI, g/m2 (n=675)	80.5±21.3
CIMT, mm (n=610)	615.9±117.3
aPWV, m/s (n=1091)	6.8±1.9
Office seated BP readings, mm Hg
bSBP	130±17
bDBP	81±12
HR, bpm	71±12
Ambulatory BP readings, mm Hg
24‐h
bSBP	123±13
bDBP	77±10
cSBPMAP/DBP	127±14
cSBPSBP/DBP	108±12
bPP	46±7
cPPMAP/DBP	49±11
cPPSBP/DBP	29±5
Daytime
bSBP	123±13
bDBP	80±10
cSBPMAP/DBP	128±14
cSBPSBP/DBP	110±12
bPP	45±8
cPPMAP/DBP	46±11
cPPSBP/DBP	28±11
Nighttime
bSBP	115±13
bDBP	69±11
cSBPMAP/DBP	125±16
cSBPSBP/DBP	101±12
bPP	45±8
cPPMAP/DBP	54±13
cPPSBP/DBP	30±6

Data are mean±SD except for sex and smoker status, which are n (%). aPWV indicates aortic pulse wave velocity; bDBP, brachial diastolic blood pressure; BMI, body mass index; BP, blood pressure; bPP, brachial pulse pressure; bSBP, brachial systolic blood pressure; cDBP, central diastolic blood pressure; cSBP, central systolic blood pressure; CIMT, carotid intima media thickness; cPP, central pulse pressure; DBP, diastolic blood pressure; HR, heart rate; LVMI, left ventricular mass index; and MAP, mean arterial pressure.

Unadjusted (univariable) associations between ambulatory BP indices and measures of subclinical organ damage are presented in Table 2 for 24‐hour BP data, and Table S2 for daytime and nighttime data. Table 2 and Table S2 also show the results (P values) of comparing correlation coefficients for the associations between ambulatory BP indices and measures of subclinical organ damage, following Steiger’s method. ¹⁴

Table 2.

Univariable Correlations and Comparison of Correlation Coefficients Between 24‐Hour Ambulatory BP Indices and Measures of Subclinical Organ Damage

Variable	Pearson correlation		Comparison of correlation coefficients
	r	P value	P value* for comparison with brachial BP	P value* for comparison with central BP(SBP/DBP)
LVMI
bSBP	0.303	<0.001	…	0.004
cSBP(SBP/DBP)	0.274	<0.001	0.004	…
cSBP(MAP/DBP)	0.403	<0.001	<0.001	<0.001
bPP	0.187	<0.001	…	<0.001
cPP(SBP/DBP)	0.114	0.003	<0.001	…
cPP(MAP/DBP)	0.273	<0.001	<0.001	<0.001
CIMT
bSBP	0.261	<0.001	…	0.001
cSBP(SBP/DBP)	0.305	<0.001	0.001	…
cSBP(MAP/DBP)	0.292	<0.001	0.138	0.611
bPP	0.185	<0.001	…	<0.001
cPP(SBP/DBP)	0.287	<0.001	<0.001	…
cPP(MAP/DBP)	0.2	<0.001	0.514	0.001
aPWV
bSBP	0.398	<0.001	…	<0.001
cSBP(SBP/DBP)	0.441	<0.001	<0.001	…
cSBP(MAP/DBP)	0.313	<0.001	<0.001	<0.001
bPP	0.218	<0.001	…	<0.001
cPP(SBP/DBP)	0.311	<0.001	<0.001	…
cPP(MAP/DBP)	0.093	0.002	<0.001	<0.001

aPWV indicates aortic pulse wave velocity; BP, blood pressure; bDBP, brachial diastolic blood pressure; bPP, brachial pulse pressure; bSBP, brachial systolic blood pressure; cDBP, central diastolic blood pressure; cPP, central pulse pressure; cSBP, central systolic blood pressure; CIMT, carotid intima media thickness; LVMI, left ventricular mass index; and MAP, mean arterial pressure.

^* P values in the final 2 columns result from comparisons of correlation coefficients, following Steiger’s method. ¹⁴

Multivariable regression models were then constructed, including age, sex, BMI, and the averaged HR corresponding to each time period (24 hours, daytime, or nighttime). These models, along with a comparison of the partial correlation coefficients for the associations between measures of subclinical organ damage and ambulatory BP indices, are shown in Table 3 (24‐hour BP indices) and Table S3 (daytime and nighttime BP indices).

Table 3.

Multivariable Associations Between 24‐Hour Ambulatory BP Indices and Measures of Subclinical Organ Damage

Variable	Adjusted R 2	β coefficient	95% CI	P value	Comparison of adjusted (partial) correlation coefficients
					r *	P value† for comparison with brachial BP	P value† for comparison with central BP(SBP/DBP)
LVMI
bSBP	0.170	0.175	0.154 to 0.390	<0.001	0.172	…	0.005
cSBP(SBP/DBP)	0.162	0.143	0.114 to 0.372	<0.001	0.142	0.005	…
cSBP(MAP/DBP)	0.186	0.237	0.225 to 0.453	<0.001	0.220	0.008	<0.001
bPP	0.152	0.083	0.034 to 0.439	0.022	0.088	…	0.001
cPP(SBP/DBP)	0.146	0.027	−0.208 to 0.434	0.490	0.027	0.001	…
cPP(MAP/DBP)	0.161	0.160	0.143 to 0.484	<0.001	0.138	0.022	<0.001
CIMT
bSBP	0.390	0.115	0.612 to 2.448	0.001	0.132	…	0.045
cSBP(SBP/DBP)	0.387	0.096	0.385 to 2.455	0.007	0.109	0.045	…
cSBP(MAP/DBP)	0.394	0.141	0.777 to 2.502	<0.001	0.150	0.353	0.081
bPP	0.388	0.091	0.588 to 3.248	0.005	0.114	…	0.331
cPP(SBP/DBP)	0.385	0.080	0.391 to 4.605	0.020	0.094	0.331	…
cPP(MAP/DBP)	0.389	0.120	0.633 to 2.855	0.002	0.124	0.665	0.260
aPWV
bSBP	0.543	0.279	0.036 to 0.050	<0.001	0.346	…	0.805
cSBP(SBP/DBP)	0.543	0.277	0.058 to 0.067	<0.001	0.344	0.805	…
cSBP(MAP/DBP)	0.535	0.277	0.061 to 0.069	<0.001	0.323	0.094	0.207
bPP	0.503	0.150	0.028 to 0.050	<0.001	0.205	…	0.145
cPP(SBP/DBP)	0.508	0.177	0.050 to 0.084	<0.001	0.227	0.145	…
cPP(MAP/DBP)	0.494	0.138	0.016 to 0.034	<0.001	0.156	0.004	<0.001

All analyses are adjusted for age, sex, BMI, and HR at time of measurement. β coefficients represent the change in the dependent variable for a 1 SD change in the exposure variable. aPWV indicates aortic pulse wave velocity; BP, blood pressure; bDBP, brachial diastolic blood pressure; bPP, brachial pulse pressure; bSBP, brachial systolic blood pressure; cDBP, central diastolic blood pressure; CIMT, carotid intima media thickness; cPP, central pulse pressure; cSBP, central systolic blood pressure; HR, heart rate; LVMI, left ventricular mass index; and MAP, mean arterial pressure.

^*Correlation coefficient derived from adjusted (partial) correlation analyses.

^† P values in the final 2 columns result from comparisons of adjusted (partial) correlation coefficients, following Steiger’s method. ¹⁴

Left Ventricular Mass Index

Pearson’s correlation coefficients describing the associations between LVMI and ambulatory cSBP and PP indices were highest overall when calibration with MAP/DBP was used and were significantly higher than the corresponding coefficients describing associations between LVMI and ambulatory brachial systolic pressure and PP and ambulatory central systolic pressure and PP calibrated using SBP/DBP (P<0.001 for all; Table 2 and Table S2).

In linear regression models adjusted for age, sex, BMI, and HR, LVMI was positively and independently associated with 24‐hour bSBP and cSBP, irrespective of calibration method, and 24‐hour brachial and central PP (cPP) when calibrated using MAP/DBP (Table 3). In keeping with the results of the unadjusted analyses, the highest β coefficients were observed for the associations between LVMI and 24‐hour ambulatory cSBP_(MAP/DBP) (β=0.237 [95% CI, 0.225–0.453]; P<0.001) and 24‐hour ambulatory cPP_(MAP/DBP) (β=0.160 [95% CI, 0.143–0.484]; P<0.001). Moreover, the partial correlation coefficients describing these associations were also highest overall and were significantly higher than those describing associations between LVMI and 24‐hour brachial systolic pressure (P=0.008) and PP (P=0.022) and 24‐hour cSBP and cPP calibrated using SBP/DBP (P<0.001 for both; Table 3). These patterns tended to be consistent across 24‐hour, daytime, and nighttime periods (Table S3).

Carotid Intima‐Media Thickness

In contrast with the findings with LVMI, higher univariable correlation coefficients tended to be observed between CIMT and ambulatory cSBP_(SBP/DBP) and cPP_(SBP/DBP) indices, compared with correlations with either brachial or central BP indices calibrated using MAP/DBP (Table 2 and Table S2). In models adjusted for age, sex, BMI, and HR, CIMT was positively and independently associated with 24‐hour bSBP and central SBP and PP indices, irrespective of calibration method. However, and in contrast with the results of univariable analyses, the highest β coefficients were observed between CIMT and 24‐hour cSBP_(MAP/DBP) (β=0.141 [95% CI, 0.777–2.502]; P<0.001) and 24‐hour cPP_(MAP/DBP) (β=0.120 [95% CI, 0.633–2.855]; P=0.002). Nevertheless, the partial correlation coefficients for CIMT versus 24‐hour cSBP_(MAP/DBP) and CIMT versus 24‐hour cPP_(MAP/DBP) were only slightly and nonsignificantly higher than corresponding coefficients for 24‐hour brachial or central BP_(SBP/DBP) indices (Table 3). Again, these patterns were consistent across all time periods (Table S3).

Aortic PWV

Similar to our observations with CIMT, higher univariable correlation coefficients tended to be observed for associations between aPWV and ambulatory cSBP_(SBP/DBP) and cPP_(SBP/DBP) indices, compared with those for either brachial BP or central BP indices calibrated using MAP/DBP (Table 2). In models adjusted for age, sex, BMI, and HR, aPWV was also positively and independently associated with 24‐hour brachial and central systolic pressure and cPP, irrespective of calibration method. However, in contrast with the findings with LVMI and CIMT, the β coefficients were broadly similar across 24‐hour brachial and central ambulatory BP indices and between calibration methods, as were the partial correlation coefficients for these associations. The exception to this trend was a significantly lower partial correlation coefficient for the association between aPWV and 24‐hour cPP_(MAP/DBP) (Table 3), which tended to be consistent across all time periods (Table S3).

Discussion

We have observed that LVMI was more strongly associated with central, rather than brachial, 24‐hour ambulatory BP in analyses adjusted for age, sex, BMI, and HR. Moreover, the associations with central ambulatory pressure were strongest when waveform calibration with brachial MAP and DBP were performed. In contrast, CIMT and aPWV, a robust measure of aortic stiffness, were similarly associated with both brachial and central ambulatory BP, when factors such as age, sex, BMI, and HR were taken into account. Moreover, their associations with ambulatory central pressure did not depend on method of waveform calibration.

In a recent meta‐analysis, resting cSBP measured in the clinic was more closely related to surrogate end points, including LVMI, CIMT, and aPWV, when compared with clinic bSBP. ¹⁵ Several studies have also evaluated the relationship between ambulatory central or brachial BP with left ventricular mass. ⁷ , ⁸ To the best of our knowledge, this is the first study to examine the relationship between ambulatory central and brachial BP with 3 different indices (LVMI, CIMT, and aPWV) in a single center, thus extending on previously published observations. In our study of a large population of untreated individuals, ambulatory cSBP_MAP/DBP was more closely associated with LVMI compared with cSBP_SBP/DBP and ambulatory bSBP on the basis of unadjusted univariable correlations and in analyses adjusted for age, sex, BMI, and HR. Our findings are consistent with 2 previous studies ⁷ , ⁸ demonstrating similar univariable correlation coefficients between LVMI and 24‐hour cSBP_(MAP/DBP) of 0.41 ⁸ and 0.49 ⁷ and a stronger βcoefficient for the same association in multivariable analyses. ⁷ Furthermore, unlike in previous studies, we have demonstrated that the univariable association between ambulatory cSBP_MAP/DBP and LVMI remained significantly stronger compared with cSBP_SBP/DBP and ambulatory bSBP, after adjustment for age, sex, BMI, and HR. Our findings also extend to cPP, where similar patterns were observed.

In our study, CIMT was also more strongly associated with cSBP_(MAP/DBP) and cPP_(MAP/DBP), but only after multivariable adjustment. It should be noted, however, that statistical comparison of the adjusted correlations between CIMT and ambulatory BP indices revealed no significant differences overall. Indeed, on the basis of simple, univariate correlations, it appeared that CIMT was significantly more strongly associated with cSBP_(SBP/DBP) and cPP_(SBP/DBP) compared with either brachial pressures or central pressures using the MAP/DBP calibration method, emphasizing the importance of considering additional factors that may confound univariable associations. In a previous study of 501 individuals at low to intermediate cardiovascular risk, ambulatory cSBP was superior to ambulatory bSBP in the association with CIMT and atheromatosis, although only waveform calibration with MAP/DBP was used. ¹⁰ Another small study of 136 adolescents and young adults examined the relationship of ambulatory central BP, derived from different calibration methods, with preclinical organ damage. The authors showed that the correlation between CIMT and ambulatory cSBP_MAP/DBP was not significantly stronger than between CIMT and ambulatory cSBP_SBP/DBP or bSBP in young adults, but a stronger correlation was observed between CIMT and cSBP_MAP/DBP in adolescents. ⁹

In contrast with the findings with LVMI and CIMT, aPWV shared a similar association with brachial and central ambulatory pressures in the current study, irrespective of calibration method. Moreover, in the study of adolescents and young adults mentioned above, ⁹ ambulatory bSBP, rather than cSBP, was more closely related to estimated PWV. A further, small Japanese study of 24 older women receiving treatment for hypertension also showed no correlation between PWV and ambulatory cSBP, where ambulatory central BP readings were derived from radial pulse waveforms. ¹⁶ Differences between studies in terms of participant demographics and methods of assessing PWV might explain, in part, the diverse findings underlying the association between PWV and ambulatory BP. Indeed, in the study by Ntineri et al, ⁹ PWV was estimated using an algorithm‐based approach rather than from carotid and femoral artery waveforms, as in the current study.

Further work is clearly required to understand the dependency of the observed associations between surrogate risk markers and ambulatory central blood pressure on the method of waveform calibration. As discussed previously, ¹⁷ , ¹⁸ although waveform calibration with brachial MAP and DBP can yield central pressure values that appear higher than measured brachial BP, they may better reflect actual intra‐arterial aortic BP. As such, these central pressure values may relate more closely to hypertension‐associated end‐organ damage and clinical end points. Our data also demonstrate the importance of not relying solely on univariable correlation coefficients to describe what are likely to be complex associations between variables. Until definitive studies based on ambulatory BP are published, demonstrating superiority of one calibration method over another in terms of clinical outcomes, both calibration methods should continue to be considered, especially when multiple surrogate risk markers are evaluated.

Our study has several limitations. Our study population involved predominantly White individuals, and our results might not be applicable to other ethnic groups. Moreover, not all surrogate cardiovascular risk measures were assessed in all individuals, limiting our ability to compare “head‐to‐head” the associations between surrogate risk markers and ambulatory BP indices across all individuals. Our study may have introduced a selection bias given that healthy study participants were required to attend a hospital setting for their examinations and the study was conducted in a single center, limiting the generalizability of our findings. Finally, the cross‐sectional nature of our study precludes any conclusions being drawn about cause and effect. In spite of these limitations, we believe that our study has a number of strengths. The inclusion of several different measures of subclinical organ damage, in relatively large numbers of participants covering the adult age span, confirms and extends recent findings and makes a robust contribution to the developing body of evidence surrounding central ambulatory BP. Moreover, while a source of potential bias, the single‐center nature of our study ensured uniformity of methods and reduced heterogeneity in the data.

In conclusion, ambulatory central blood pressure had a stronger association with LVMI than ambulatory brachial BP, particularly when the MAP/DBP waveform calibration method was used. However, CIMT and aPWV shared a similar association with both brachial and central ambulatory pressures, with no obvious effect of the waveform calibration method, once factors such as age, sex, BMI, and HR were considered. Calibration method clearly has an important impact on the observed associations between central ambulatory pressure and surrogate end points, and further studies are required to clarify this important issue.

Sources of Funding

This work was supported by grants received from British Heart Foundation (PG/11/35/28879) and the National Institute for Health Research Biomedical Research Centre (BRC‐1215‐20 014).

Disclosures

C.M. McEniery is supported by the National Institute for Health Research Cambridge Biomedical Research Centre (BRC‐1215‐20 014). K.M. Mäki‐Petäjä was employed by the University of Cambridge when the study was conducted but now works at AstraZeneca (Cambridge, UK). The remaining authors have no disclosures to report.

Supporting information

Tables S1–S3

Checklist S1