Abstract
Background:
It remained unclear whether central blood pressures (BP) was more closely associated with cardiovascular disease (CVD) than brachial BP in different age groups.
Objectives:
To investigate the age-stratified association of CVD with brachial and central BPs, and to evaluate corresponding improvement in model performance.
Methods:
This cohort study included 34 289 adults without baseline CVD from the UK Biobank dataset. Participants were categorized into middle-aged and older aged groups using the cut-off of age 65 years. The primary endpoint was a composite cardiovascular outcome consisting of cardiovascular mortality combined with nonfatal coronary events, heart failure and stroke. Multivariable-adjusted hazard ratios expressed CVD risks associated with BP increments of 10 mmHg. Akaike Information Criteria (AIC) was used for model comparisons.
Results:
In both groups, CVD events were associated with brachial or central SBP (P ≤ 0.002). Model fit was better for central SBP in middle-aged adults (AIC 4427.2 vs. 4429.5), but model fit was better for brachial SBP in older adults (AIC 10 246.7 vs. 10 247.1). Central SBP remained significantly associated to CVD events [hazard ratio = 1.05; 95% confidence interval (CI) 1.0–1.1] and improved model fit (AIC = 4426.6) after adjustment of brachial SBP only in the middle-aged adults. These results were consistent for pulse pressure (PP).
Conclusion:
In middle-aged adults, higher central BPs were associated with greater risks of CVD events, even after adjusting for brachial BP indexes. For older adults, the superiority of central BP was not observed. Additional trials with adequate follow-up time will confirm the role of central BP in estimating CVD risk for middle-aged individuals.
Keywords: cardiovascular diseases, central blood pressure, hypertension
INTRODUCTION
Hypertension is the most common chronic condition affecting one-third of the world's adult population and is the leading cause of cardiovascular diseases (CVD) and deaths. Although brachial blood pressure (BP) measurement remains the standard to diagnose and manage hypertension, central BP measurement is superior in reflecting the stress imposed on arteries of internal vital organs (e.g. coronary and cerebral arteries) and may thereby provide superior prediction to cardiovascular events and deaths [1].
Although numerous studies supported superiority of central BP to predict CVD [2–9], results have been inconsistent and contradictory [10–13]. These inconsistencies can be because of small sample size in most studies, and differences in methodology and participants’ characteristics. In particular, pulse pressure (PP) amplification is attenuated with age because of increased vascular stiffness, secondary to increased reflected pressure wave bounded back to the aorta from stiff vasculature. This reduction in PP amplification minimizes the difference between brachial and central BP and may reduce the incremental predictive value of central BP (above that provided by brachial BP) in older people [14]. However, most relevant and existing literature did not conduct subgroup analysis according to age. Although an individual-level meta-analysis found a closer association between central SBP and stroke in middle-age participants than older participants [15], it remains unclear whether this similar age-dependent association applies to other CVD events and to total mortality. To date, central BP was not discussed or recommended in international guidelines because of inconclusive evidence. For instance, the 2023 European Society of Hypertension guideline stated that ‘the incremental prognostic value of central vs. conventional clinic BP measurement remains unclear’.
Therefore, we hypothesize that central BP is associated with cardiovascular outcomes even after controlling for brachial BP. We further hypothesize that this association of central BP is age-dependent (i.e. central BP is more closely associated with CVD in younger adults).
METHODS
Study population
This is a cohort study using the UK Biobank (UKB), which has recruited patients since 2006.
Detailed study design and methods of UKB has been published elsewhere [16]. We used the first central BP data, which were available since May 2014 until March 2020. This project was approved by the UK Biobank (application 64457 (PI: L.E.K.P.).
All participants with complete brachial and central BP and cardiovascular risk data (i.e. age, sex, smoking status, BMI, and history of hyperlipidemia, diabetes mellitus and chronic kidney disease) were included. Participants were excluded if they had any one of the following exclusion criteria: a past history of CVD, the implausible central BP measurement as defined by UKB dataset (i.e. if the raw data was not presented in the output data stream or the first pulse wave analysis did not contain central PP or augmentation index), extreme BP readings [5], that is, central SBP less than 70 mmHg or at least 230 mmHg; central DBP less than 40 mmHg or at least 150 mmHg; brachial PP at least 130 mmHg and that brachial PP smaller than central PP, or missing data on cardiovascular risk factors, and having only one central BP measurement record.
Blood pressure measurements
In a supine position, the brachial cuff was obtained using patients’ left arm. Trained staffs ensured the appropriate tightness of the brachial cuff. Although participants were not specifically instructed to rest, they did not engage in strenuous activities prior to the BP measurement. Brachial BP was measured in dual by digital oscillometric technique using the validated Vicorder system (SMT Medical, Bristol, UK) [17]. The average of the two readings was used for analyses, as recommended by American Heart Association [18]. Immediately after each BP measurement, the cuff was inflated again (to 70 mmHg) to capture the pulse wave using the same device using the volume displacement technique, and the waveform was digitally computed. Central BP parameters were estimated from brachial BP waveforms which were self-calibrated to brachial SBP/DBP by applying validated brachial-to-aortic transfer function [19]. The pulse wave analysis system was validated against invasive central blood pressure (BP) and the SphygmoCor applanation tonometry system (AtCor Medical, Sydney, Australia) in a study involving 140 participants. The validation study showed that Vicorder had noninferior accuracy to SphygmoCor and can serve as a reliable and simple noninvasive approach for assessing central BP [17]. More detailed descriptions of the Vicorder system can be found in Figure S1. Only brachial and central SBP and PP were compared in the current study because central and brachial DBP were similar throughout the arterial tree from the ascending aorta to small arterioles [20].
Outcome measures
The primary endpoint was a composite cardiovascular outcome consisting of cardiovascular mortality combined with nonfatal cardiovascular events (i.e. nonfatal myocardial infarction, coronary revascularization, heart failure and stroke). Secondary end-points included all-cause mortality, coronary events (which included sudden death, death from ischemic heart disease, nonfatal myocardial infarction and coronary revascularization) and stroke. Analysis using each type of coronary event is infeasible because of small number of participants having each outcome. The corresponding ICD-10 and OPCS4 codes for primary and secondary endpoints can be found in Table S1. In all outcome analyses, only first event within each category was included. Relevant data was retrieved by the UK biobank with linked death, inpatient diagnosis and operation records.
Covariates
Main covariates in this study included cardiovascular risk factors at baseline (sex, age, BMI, whether current smoker, history of diabetes mellitus, hyperlipidemia, chronic kidney disease, the use of antihypertensive agents) and factors affecting central BP measurement (sex, age, and heart rate during pulse wave analysis) [21]. Further co-variates were included in the sensitivity analyses to examine robustness of our results and included: race (i.e. Caucasian vs. non-Caucasian), Townsend deprivation index, being a current drinker, presence of family history of CVD, past history of comorbidities (e.g. migraine, severe mental illness, atrial fibrillation, chronic rheumatoid diseases, systemic lupus arthritis, erectile dysfunction, HIV or AIDs), the use of drugs associated with elevated CVD risks (i.e. atypical antipsychotic drugs and oral steroid). Detailed description of these covariates can be found in Table S2.
Statistical analysis
To investigate the age-dependent association of central BP, participants were categorized into middle-aged group (<65 years old) and older adult group (≥65 years old). Baseline characteristics between these two groups were compared by t test and chi-square test for continuous variables and for category variables, respectively. Statistical significance was defined by P values less than 0.05. Analyses were performed using Stata version 16.1 (Stata Corp, Texas) and R version 4.2.1 (R core team, Vienna, Australia).
Cox proportional-hazards models
All Cox proportional-hazards models in this study were controlled for all covariates listed above. Survival time was calculated from the date of central BP measurement to end-points or the end of follow-up [30 September 2021 (the last update by UKB)]. Hazard ratios were expressed as per increment of the 10 mmHg of BP.
To determine the association relating to endpoints with brachial and central SBP, four different models were constructed. The set of models were performed separately for middle-aged and older group, with the adjustment of covariates. The BP index was brachial SBP in model 1. In model 2, central SBP was added into model 1. To avoid multicollinearity, central SBP was uncorrelated by regressing central SBP on brachial SBP, and residuals were used to compute hazard ratios or to assess the improvement of model performance [22,23]. A positive residual identified an individual in whom central SBP was higher than would have been expected based on their level of brachial SBP, making the risk predicted by central SBP greater than predicted by brachial SBP. In contrast, a negative residual implied that the risk predicted by central SBP was lower than predicted by brachial SBP. Therefore, the residuals quantitated the discordance between brachial and central SBP. The BP index was central SBP in model 3. In model 4, brachial SBP was added into model 3 using the residual method. The same analyses were performed for PP. Proportional hazards assumption was checked by testing a nonzero slope between Schoenfeld residuals and survival times and by testing the interaction between follow-up duration and the BP variables.
Model performance
To evaluate the model performance, model fits and discrimination were assessed. For nonnested models, model fit was compared by Akaike Information Criteria (AIC). A lower value of AIC represented a better model fit. For nested models, log-likelihood ratio test was used to determine whether the model fit significantly improved when the second BP index was added. The model discrimination was measured by concordance (C) statistics. DeLong test was used to check whether the C-statistics was significantly different when the second BP index was added [24]. To evaluate the reclassification effects of adding the second BP index, net reclassification indices (NRI) were calculated for middle-aged and older groups, with a 95% CI obtained with bootstrapping (R package NRIcens, v1.6).
Sensitivity analyses
To check the robustness of the results, hazard ratios relating to the primary end point to the central and brachial BP were determined with the additional adjustment of brachial DBP or other indicator variables. Sensitivity analyses were also conducted by excluding people with the use of antihypertensive drugs, excluding people with the use of beta-blockers, and excluding people with baseline chronic kidney disease, excluding those with atrial fibrillation, excluding people with extreme central SBP/PP residuals, including those with higher central PP than brachial PP, using a different age cutoff (70 years of age).
RESULTS
Participants
Of 40 367 UKB participants who received brachial and central BP measurements, a total of 34 289 participants (17 383 in middle-age group and 16 906 in older adult group) were included (Fig. 1). Demographics and diseases prevalence of participants in the middle-age group were significantly different from that of the older adult group. For example, as expected, younger adults were less likely to have diabetes mellitus and chronic kidney disease (Table 1). Similarly, participants in the middle-age group had lower brachial BP (134.8/69.9 vs. 143.2/70.2 mmHg), central BP (130.5/69.9 vs. 139.4/70.2 mmHg) but higher PP amplification (4.3 vs. 3.8 mmHg; P < 0.001) (Table 1).
FIGURE 1.
Flow chart for eligible participants.
TABLE 1.
Baseline characteristics of participants by whether aged less than 65 years
| Characteristic | Middle-aged (N = 17 383) | Older adults (N = 16 906) | P value |
| Number (%) with characteristic | |||
| Male | 7723 (44.42%) | 8602 (50.88%) | <0.001 |
| Current smokera | 756 (4.35%) | 425 (2.51%) | <0.001 |
| Hypertensionb | 8032 (46.20%) | 11730 (69.38%) | <0.001 |
| Use of antihypertensive drugc | 2710 (15.59%) | 5323 (31.49%) | <0.001 |
| Use of beta-blockers | 440 (2.53%) | 893 (5.28%) | <0.001 |
| High cholesterol leveld | 7488 (43.08%) | 11025 (65.21%) | <0.001 |
| Diabetes mellituse | 871 (5.01%) | 1388 (8.21%) | <0.001 |
| Chronic kidney diseasef | 220 (1.27%) | 804 (4.76%) | <0.001 |
| Mean (SD) of characteristic | |||
| Age, mean (SD) | 57.54 (4.48) | 70.43 (3.87) | <0.001 |
| BMI, mean (SD) | 26.57 (4.58) | 26.23 (4.03) | <0.001 |
| Brachial SBP | 134.77 (17.54) | 143.23 (18.43) | <0.001 |
| Brachial DBP | 69.90 (11.05) | 70.19 (10.87) | 0.015 |
| Brachial PP | 64.87 (16.35) | 73.04 (17.53) | <0.001 |
| Central SBP | 130.46 (17.53) | 139.40 (18.32) | <0.001 |
| Central DBP | 69.90 (11.05) | 70.19 (10.87) | 0.014 |
| Central PP | 60.57 (16.05) | 69.21 (17.29) | <0.001 |
| Brachial PP–central PP | 4.30 (3.28) | 3.83 (3.03) | <0.001 |
| Heart rate | 62.43 (11.24) | 63.09 (11.00) | <0.001 |
BP, blood pressure; PP, pulse pressure.
Whether currently smoking was assessed by self-reported at the baseline.
Hypertension was measured by brachial SBP of 140 mmHg or higher or/and brachial DBP of 90 mmHg or higher, self-reported diagnosis of hypertension, self-reported use of antihypertensive drugs, diagnosis in the hospital records, diagnosis in the general practice and drug prescription records in the general practice.
Use of antihypertensive drug was measured by self-reported drug use and drug prescription records in the general practice.
High cholesterol level was measured by self-reported diagnosis of high cholesterol level, self-reported use of lipid-lowering drugs, diagnosis in the hospital records, diagnosis in the general practice, and biochemistry tests at previous UK Biobank visit, including LDL-C at least 4.1 mmol/l or total cholesterol at least 6.2 mmol/l or triglycerides at least 2.3 mmol/l.
Diabetes mellitus was measured by self-reported diagnosis of diabetes, self-reported use of insulin, diagnosis in the hospital records, diagnosis in the general practice and biochemistry tests at previous UK Biobank visit, including random blood glucose of 11.1 mmol/l or higher or HAb1c of 6.5% or higher.
Chronic kidney disease was defined as diagnosis of chronic disease (stage 3–5) and the major chronic renal disease including nephrotic syndrome, chronic glomerulonephritis, chronic pyelonephritis, renal dialysis and renal transplant. It was obtained from the diagnosis in the general practice or hospital records.
Participants were followed up for 42.4 and 36.9 months in the middle-age group and older adult group, respectively. In the middle-aged group, 243 experienced CVD (129 coronary events, 89 strokes) and 93 participants died. Among the older adults, 566 participants experienced CVD (248 coronary events and 211 strokes) and 208 died.
Primary outcome
Association between blood pressure indices and cardiovascular outcomes
Brachial and central SBP and PP were both associated with composite CVD to a similar extent in models 1 and 3 (Table 2). Only in middle-age group, the central SBP (hazard ratio = 1.05) and PP (hazard ratio = 1.06) were more associated with CVD outcomes than brachial indices because the residuals of central BP indices remained significantly associated with CVD outcomes in model 2, but residuals of brachial BP indices were not associated with CVD outcomes in model 4 (Tables 2 and 3).
TABLE 2.
Association between brachial and central SBP with the composite cardiovascular outcome, stratified by whether aged 65 years and older
| bSBPb | cSBPb | ebSBPb | ecSBPb | |||||||||||||||||
| Modela | HR (95% CI)e | P value | HR (95% CI)e | P value | HR (95% CI) | P value | HR (95% CI) | P value | AIC | P valuec | c-statistic | P valued | ||||||||
| Middle-aged group, age <65 years (N = 243/17 383) | ||||||||||||||||||||
| Model 1 | 1.12 (1.04, 1.21) | 0.002 | – | – | – | – | – | – | 4429.50 | – | 0.704 | – | ||||||||
| Model 2 | 1.13 (1.05, 1.21) | 0.002 | – | – | – | – | 1.05 (1.00, 1.09) | 0.03 | 4426.61 | 0.03 | 0.707 | 0.32 | ||||||||
| Model 3 | – | – | 1.14 (1.06, 1.23) | <0.001 | – | – | – | – | 4427.23 | – | 0.705 | – | ||||||||
| Model 4 | – | – | 1.14 (1.06, 1.23) | <0.001 | 0.97 (0.93, 1.01) | 0.11 | – | – | 4426.61 | 0.11 | 0.707 | 0.39 | ||||||||
| Older adult group, age> = 65 (N = 566/16906) | ||||||||||||||||||||
| Model 1 | 1.11 (1.06, 1.17) | <0.001 | – | – | – | – | – | – | 10246.66 | – | 0.675 | – | ||||||||
| Model 2 | 1.11 (1.06, 1.17) | <0.001 | – | – | – | – | 1.00 (0.97, 1.03) | 0.95 | 10248.65 | 0.94 | 0.675 | 0.57 | ||||||||
| Model 3 | – | – | 1.11 (1.06, 1.17) | <0.001 | – | – | – | – | 10247.08 | – | 0.675 | – | ||||||||
| Model 4 | – | – | 1.11 (1.06, 1.17) | <0.001 | 1.01 (0.98, 1.04) | 0.51 | – | – | 10248.65 | 0.51 | 0.675 | 0.89 | ||||||||
All models were adjusted for age, sex, current smoker, diabetes, use of antihypertensive drug, BMI, high cholesterol level, heart rate and chronic kidney disease.
bSBP indicates brachial systolic blood pressure; cSBP indicates central systolic blood pressure; ebSBP represented the residual generated from bSBP∼cSBP (without the adjustment of other risk factors); and ecSBP represented the residual generated from cSBP∼bSBP (without the adjustment of other risk factors).
The P value was obtained from log-likelihood ratio test. Model 2 was compared with model 1 whereas model 4 was compared with model 3.
The P value was for the comparison of c-statistics. Model 2 was compared with model 1 whereas model 4 was compared with model 3.
Hazard ratio for blood pressure was expressed per increment of 10 mmHg.
TABLE 3.
Association between brachial and central pulse pressure with the composite cardiovascular outcome, stratified by whether aged 65 years and older
| bPPb | cPPb | ebPPb | ecPPb | |||||||||||||||||
| Modela | HR (95% CI)e | P value | HR (95% CI)e | P value | HR (95% CI) | P value | HR (95% CI) | P value | AIC | P valuec | c-statistic | P valued | ||||||||
| Middle-aged group, age <65 years (N = 243/17 383) | ||||||||||||||||||||
| Model 1 | 1.14 (1.05, 1.23) | 0.002 | – | – | – | – | – | – | 4429.16 | – | 0.704 | – | ||||||||
| Model 2 | 1.14 (1.05, 1.23) | 0.001 | – | – | – | – | 1.06 (1.01, 1.10) | 0.01 | 4424.58 | 0.01 | 0.709 | 0.27 | ||||||||
| Model 3 | – | – | 1.16 (1.07, 1.26) | <0.001 | – | – | – | – | 4426.18 | – | 0.706 | – | ||||||||
| Model 4 | – | – | 1.16 (1.07, 1.26) | <0.001 | 0.96 (0.92, 1.00) | 0.06 | – | – | 4424.58 | 0.06 | 0.709 | 0.37 | ||||||||
| Older adult group, age at least 65 years (N = 566/16 906) | ||||||||||||||||||||
| Model 1 | 1.09 (1.03, 1.14) | 0.001 | – | – | – | – | – | – | 10255.95 | – | 0.670 | – | ||||||||
| Model 2 | 1.09 (1.03, 1.14) | 0.001 | – | – | – | – | 1.00 (0.97, 1.03) | 0.84 | 10257.91 | 0.84 | 0.670 | 0.80 | ||||||||
| Model 3 | – | – | 1.09 (1.03, 1.15) | 0.001 | – | – | – | – | 10256.05 | – | 0.670 | – | ||||||||
| Model 4 | – | – | 1.09 (1.03, 1.14) | 0.001 | 1.01 (0.98, 1.03) | 0.71 | – | – | 10257.91 | 0.71 | 0.670 | 0.89 | ||||||||
All models were adjusted for age, sex, current smoker, diabetes, use of antihypertensive drug, BMI, high cholesterol level, heart rate and chronic kidney disease.
bPP indicates brachial pulse pressure; cPP indicates central pulse pressure; ebPP represented the residual generated from brachial PP∼central PP (without the adjustment of other risk factors); and ecPP represented the residual generated from central PP∼brachial PP (without the adjustment of other risk factors).
The P value was obtained from log-likelihood ratio test. Model 2 was compared with model 1 whereas model 4 was compared with model 3.
The P value was for the comparison of c-statistics. Model 2 was compared with model 1 whereas model 4 was compared with model 3.
Hazard ratio for blood pressure was expressed per increment of 10 mmHg.
Comparison of model performance by central and brachial blood pressure indices
In the middle-age group, central PP (AIC: 4426.2 vs. 4429.2) and SBP (AIC: 4427.2 vs. 4429.5) provided better model fit than respective brachial BPs in models that included a single BP index (i.e. models 1 and 3) (Tables 2 and 3). Model fit was improved when residuals of central PP (P = 0.01) and SBP (P = 0.03) were added to models that already included brachial PP and SBP, respectively (Tables 2 and 3).
In the older adult group, brachial PP (AIC: 10256.0 vs. 10256.1) and SBP (AIC: 10246.7 versus 10247.1) provided better model fit than respective central BPs in models that included a single BP index (i.e. models 1 and 3) (Tables 2 and 3). In contrast to findings in the middle-age group, adding residual of central or brachial PP/SBP did not improve model fit that already included brachial or central BP indices (model 2 and 4) (Tables 2 and 3).
The discriminative ability evaluated by c-statistic was similar in each model for both middle-aged and older adult group (Tables 2 and 3). In both groups, the reclassification effect was not significant when a second BP indices was added (Table S6).
Secondary outcomes
Adding residuals of central PP improved risk prediction to coronary events only in the middle-age group; however, this was not observed for all-cause mortality and stroke (Table S5). Only in the middle-age group, central PP provided incremental predictive value above brachial PP for coronary events because model fit improved when residuals of central PP (P = 0.02) was added to models that already included brachial PP (Table S5). Adding residuals of brachial SBP indices did not improve the model fit for secondary outcomes (Table S4). The comparison of c-statistics and reclassification effects did not support any significant difference for secondary outcomes (Tables 2 and 3, Tables S4–S6).
Sensitivity analyses
Similar results were obtained in our sensitivity analyses (Tables S7–S15). In particular, central BP indices (especially PP) only consistently had stronger association with CVD than brachial BP indices in the middle-age group.
DISCUSSION
Main findings
Pulse wave analysis data obtained in middle-aged and older participants in UK Biobank were used to examine the relationship between central BP and CVD in different age groups. Only in middle-aged population, higher central BP were associated with increased CVD risks, even after adjusting for brachial BP and other established risk factors. Central BP residuals were utilized in our analysis to capture variations in central BP not explained by brachial BP, providing additional prognostic information beyond brachial BP. Over half of middle-aged participants had positive central PP residuals, indicating that their risk predicted by central PP greater than predicted by the brachial counterpart. Around 29% middle-aged participants had central PP residuals exceeding 2 mmHg, corresponding to 12% higher CVD risks, and 12% had residuals exceeding 3 mmHg, corresponding to 19% higher risks, compared with those whose central PP can be accurately predicted by brachial PP.
Compared with models with brachial BP, models with central BP had better model fit, as measured by AIC. The relative predictive value of central and brachial BP was further assessed by entering one BP measure and then the other to a base model that incorporated established risk factors. For the middle-aged group, adding central BP in a model already including brachial BP improved the model fit. Conversely, adding brachial BP after central BP did not significantly enhance the model fit. Among older adults, the superiority of central BP was not observed.
Our findings of age discrepancy can be explained by the observation that older adults tend to have more similar central and brachial BP readings (Fig. S2). This trend can be attributed to vascular aging, large artery stiffening, and increased wave reflections, which collectively result in reduced pressure amplification as individuals age [25]. Our results showing age discrepancy is consistent with existing literature. First, a meta-analysis supported that there is a closer association between central SBP and stroke in younger adults (hazard ratio = 1.87) than in older adults (hazard ratio = 1.09, P interaction <0.01) [15]. Similarly, majority of studies involving younger participants (when mean age <60) showed a superior predictive value of central SBP than brachial BP to cardiovascular events and deaths [4–7,9,26]. In the contrary, several studies that recruited only geriatric patients did not show a superiority of central BP when compared with brachial BP [12,27].
Clinical implications
Our results support clinicians to measure and monitor central BP in the younger population (age <65), in addition to traditional monitoring of brachial BP. On the contrary, central BP is similar to brachial BP in older adults (age ≥65) and does not provide additional information concerning CVD prediction.
In fact, using a commonly used definition of central hypertension (i.e. 130/90 mmHg), 18.2% of middle-age population (age <65) in the UKB had central hypertension but a normal brachial BP [4,5]. These patients had, therefore, elevated and untreated CVD risks according to our results. Due to the high prevalence of hypertension, even a small reduction in CVD risk caused by elevated central BP in these populations will translate into enormous health benefits at the population level. Although older methods (e.g. tonometry) to measure central BP require additional equipment and training, cuff-based devices that are validated according to international recommendations are becoming available and may be feasible for routine clinical practice. Furthermore, multiple randomized-controlled trials (RCT) have shown a differential effect of antihypertensive drugs on central BP [28–30]. For instance, despite similar brachial BP reduction, central BP reduction was more prominent using amlodipine-based regimen than atenolol-based regimen in the landmark CAFÉ study [30]. This differential reduction in central BP has translated to reduction in cardiovascular events [30]. Other RCTs have also shown that the use of central BP to guide antihypertension treatments could reduce the number of antihypertension drugs, and increase exercise capacity [31,32].
Research implications
Although awaiting results from the targeted LOWering of Central Blood Pressure in patients with hypertension (LOW CBP) RCT (which investigates whether addition of antihypertension medications in patients with high central BP and normal brachial BP can reduce left ventricular mass and pulse wave velocity), more RCTs are needed to determine whether central BP-guided management can reduce cardiovascular events and deaths. Furthermore, although out-of-office BP has been consistently shown to be more reproducible, reliable, and associated with CVD than traditional office BP, there is paucity of studies to investigate the role of central BP in home BP or 24 h ambulatory BP monitoring [23,33]. Although benefits of central BP measurement may be appealing, future cost-effectiveness study is warranted to evaluate the benefits and corresponding trade-offs. Finally, if central BP is recommended for routine clinical practice, implementation studies need to be conducted because clinicians often have poor adherence to clinical BP measurement recommendations. For instance, despite emphasized by multiple international guidelines, out-of-office BP is underused to diagnose and manage hypertension by clinicians [34].
Strengths and limitations
This is one of the largest cohort studies (n = 34 289; with 40-month follow-up). For instance, the meta-analysis that contained 11 longitudinal studies only contained 5648 participants with a mean follow-up of 45 months [35]; Another meta-analysis that contained 15 studies also had less number of participants (n = 22 433) than the current study [15]. Standard data collection, careful ascertainment of outcomes and diversity of covariates (to allow multiple sensitivity analyses) were also key strengths of the study. In addition, the residual method eased the strong correlation between brachial and central BP and allowed us to determine whether central BP provided additional prognostic information to cardiovascular risk prediction. Furthermore, our secondary and sensitivity analyses showed similar results, supporting that our conclusions are robust.
Several limitations of this study can be discussed. First, although similar to previous studies [15,35], the median of 40-month follow-up is relative short for ‘hard’ cardiovascular outcomes such as stroke and death. Our analysis did not find an additional prognostic value of central BP on stroke, nor did it identify an association between central BP and total mortality in middle-aged group. These findings are likely attributed to the limited number of events of stroke (N = 89) and deaths (N = 93). Second, although Vicorder device and its related generalized transfer functions are adequately validated to estimate central BP, the reference standard to measure central BP values is direct measurement from the aorta using invasive methods. We would argue that invasive measurements are infeasible for large cohort and routine practice. On the contrary, the use of noninvasive measurement methods better reflect true clinical practice [17,36]. The current study also used Vicorder instead of SphygmoCor (AtCor Medical) devices, which were more commonly used in previous studies [4,5,7,11,37]. Although both Vicorder and SphygmoCor apply global transfer function to derive central BP, Vicorder devices may be preferred because they used brachial BP waveforms instead of radical BP wavesforms (which are used by SphygmoCor devices) to estimate central BP [17]. SphygmoCor devices assume that brachial and radial pressures are equal, leading to underestimation of central BP caused by brachial-to-radial amplification [38]. In fact, when radial waveforms were calibrated on brachial mean arterial pressure (MAP)/DBP, inter-device agreement between Vicorder and SphygmoCor devices were excellent [17]. Another limitation is that participants did not have a period of rest prior to BP measurements, but no participants had strenuous activities prior to BP measurements. A recent randomized controlled trial also showed that different rest periods (ranging from 0 to 5 min) did not significantly impact on BP measurements [39]. Fourth, our population compose of a predominantly older Caucasian population, and the generalizability to people younger than 45 years or other ethnicities is unknown. In addition, previous cardiovascular prediction models included blood lipids (total, high-density lipoprotein, low-density lipoprotein cholesterol level and triglycerides) and other blood variables (blood glucose, C reactive protein, creatinine) as predictors [21], but these blood variables were not available in the UKB at the time when central BP was collected. Although high cholesterol level was used as a binary covariate in the current study, it may lead to the less accurate risk estimation. Lastly, although endpoints were identified from death registry, operation records and inpatient diagnoses, these records did not contain comprehensive imaging profiles. Therefore, the identification of certain disease subtypes, such as the heart failure with preserved or reduced ejection fraction, could not be achieved.
In conclusion, in this population-based cohort study, higher central BP were associated with greater risks of CVD events, even after adjusting for brachial BP indexes in middle-aged participants. For older adults, the superiority of central BP was not observed. Additional trials with adequate follow-up time will confirm the role of central blood pressure in estimating disease risk for middle-aged individuals. This will provide a clearer understanding and more effective guidance for preventing and managing CVD.
ACKNOWLEDGEMENTS
Conflicts of interest
There are no conflicts of interest.
Supplementary Material
B.H.K.Y. and E.K.P.L. are co-corresponding authors.
Abbreviations: AIC, Akaike Information Criteria; BP, blood pressure; C-statistics, concordance statistics; CVD, cardiovascular diseases; IDCARS, International Database of Central Arterial Properties for Risk Stratification; MAP, mean arterial pressure; NRI, net reclassification indices; PP, pulse pressure; RCT, randomized-controlled trials; UKB, UK biobank
Supplemental digital content is available for this article.
REFERENCES
- 1.Laurent S, Boutouyrie P. Recent advances in arterial stiffness and wave reflection in human hypertension. Hypertension 2007; 49:1202–1206. [DOI] [PubMed] [Google Scholar]
- 2.Eguchi K, Miyashita H, Takenaka T, Tabara Y, Tomiyama H, Dohi Y, et al. High central blood pressure is associated with incident cardiovascular events in treated hypertensives: the ABC-J II Study. Hypertens Res 2018; 41:947–956. [DOI] [PubMed] [Google Scholar]
- 3.Roman MJ, Devereux RB, Kizer JR, Lee ET, Galloway JM, Ali T, et al. Central pressure more strongly relates to vascular disease and outcome than does brachial pressure: the Strong Heart Study. Hypertension 2007; 50:197–203. [DOI] [PubMed] [Google Scholar]
- 4.Cheng HM, Chuang SY, Sung SH, Yu WC, Pearson A, Lakatta EG, et al. Derivation and validation of diagnostic thresholds for central blood pressure measurements based on long-term cardiovascular risks. J Am Coll Cardiol 2013; 62:1780–1787. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Cheng YB, Thijs L, Aparicio LS, Huang QF, Wei FF, Yu YL, et al. Risk stratification by cross-classification of central and brachial systolic blood pressure. Hypertension 2022; 79:1101–1111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Dong Y, Jiang L, Wang X, Chen Z, Zhang L, Zhang Z, et al. Central rather than brachial pressures are stronger predictors of cardiovascular outcomes: A longitudinal prospective study in a Chinese population. J Clin Hypertens (Greenwich) 2020; 22:623–630. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lamarche F, Agharazii M, Madore F, Goupil R. Prediction of cardiovascular events by type I central systolic blood pressure: a prospective study. Hypertension 2021; 77:319–327. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Pini R, Cavallini MC, Palmieri V, Marchionni N, Di Bari M, Devereux RB, et al. Central but not brachial blood pressure predicts cardiovascular events in an unselected geriatric population: the ICARe Dicomano Study. J Am Coll Cardiol 2008; 51:2432–2439. [DOI] [PubMed] [Google Scholar]
- 9.Wang KL, Cheng HM, Chuang SY, Spurgeon HA, Ting CT, Lakatta EG, et al. Central or peripheral systolic or pulse pressure: which best relates to target organs and future mortality? J Hypertens 2009; 27:461–467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Huang QF, Aparicio LS, Thijs L, Wei FF, Melgarejo JD, Cheng YB, et al. IDCARS (International Database of Central Arterial Properties for Risk Stratification) Investigators. Cardiovascular end points and mortality are not closer associated with central than peripheral pulsatile blood pressure components. Hypertension 2020; 76:350–358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Mitchell GF, Hwang SJ, Larson MG, Hamburg NM, Benjamin EJ, Vasan RS, et al. Transfer function-derived central pressure and cardiovascular disease events: the Framingham Heart Study. J Hypertens 2016; 34:1528–1534. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Dart AM, Gatzka CD, Kingwell BA, Willson K, Cameron JD, Liang YL, et al. Brachial blood pressure but not carotid arterial waveforms predict cardiovascular events in elderly female hypertensives. Hypertension 2006; 47:785–790. [DOI] [PubMed] [Google Scholar]
- 13.Laugesen E, Knudsen ST, Hansen KW, Rossen NB, Jensen LO, Hansen MS, et al. Invasive aortic pulse pressure is not superior to cuff pulse pressure in cardiovascular risk prediction. J Hypertens 2021; 39:607–613. [DOI] [PubMed] [Google Scholar]
- 14.Murgo JP, Westerhof N, Giolma JP, Altobelli SA. Aortic input impedance in normal man: relationship to pressure wave forms. Circulation 1980; 62:105–116. [DOI] [PubMed] [Google Scholar]
- 15.McEniery CM. Central blood pressure and cardiovascular risk: an individual participant meta-analysis of prospective observational data from 22,433 subjects. J Am Coll Cardiol 2015; 65 (10S):A1464. [Google Scholar]
- 16.Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med 2015; 12:e1001779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Pucci G, Cheriyan J, Hubsch A, Hickson SS, Gajendragadkar PR, Watson T, et al. Evaluation of the Vicorder, a novel cuff-based device for the noninvasive estimation of central blood pressure. J Hypertens 2013; 31:77–85. [DOI] [PubMed] [Google Scholar]
- 18.Muntner P, Shimbo D, Carey RM, Charleston JB, Gaillard T, Misra S, et al. Measurement of blood pressure in humans: a scientific statement from the American Heart Association. Hypertension 2019; 73:e35–e66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.ORourke MF. Influence of ventricular ejection on the relationship between central aortic and brachial pressure pulse in man. Cardiovasc Res 1970; (4):291–300. [DOI] [PubMed] [Google Scholar]
- 20.Remington JW, Wood EH. Formation of peripheral pulse contour in man. J Appl Physiol 1956; (9):433–442. [DOI] [PubMed] [Google Scholar]
- 21.Damen JA, Hooft L, Schuit E, Debray TP, Collins GS, Tzoulaki I, et al. Prediction models for cardiovascular disease risk in the general population: systematic review. BMJ 2016; 353:i2416. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Karlson KB, Holm A, Breen R. Comparing regression coefficients between same-sample nested models using Logit and Probit. Sociol Methodol 2012; 42:286–313. [Google Scholar]
- 23.Yang WY, Melgarejo JD, Thijs L, Zhang ZY, Boggia J, Wei FF, et al. International Database on Ambulatory Blood Pressure in Relation to Cardiovascular Outcomes (IDACO) Investigators. Association of office and ambulatory blood pressure with mortality and cardiovascular outcomes. JAMA 2019; 322:409–420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Elizabeth R, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44:837–945. [PubMed] [Google Scholar]
- 25.Avolio AP, Van Bortel LM, Boutouyrie P, Cockcroft JR, McEniery CM, Protogerou AD, et al. Role of pulse pressure amplification in arterial hypertension: experts’ opinion and review of the data. Hypertension 2009; 54:375–383. [DOI] [PubMed] [Google Scholar]
- 26.Jankowski P, Kawecka-Jaszcz K, Czarnecka D, Brzozowska-Kiszka M, Styczkiewicz K, Loster M, et al. Pulsatile but not steady component of blood pressure predicts cardiovascular events in coronary patients. Hypertension 2008; 51:848–855. [DOI] [PubMed] [Google Scholar]
- 27.Zhang Y, Protogerou AD, Iaria P, Safar ME, Xu Y, Blacher J. Prognosis in the hospitalized very elderly: the PROTEGER study. Int J Cardiol 2013; 168:2714–2719. [DOI] [PubMed] [Google Scholar]
- 28.Mackenzie IS, McEniery CM, Dhakam Z, Brown MJ, Cockcroft JR, Wilkinson IB. Comparison of the effects of antihypertensive agents on central blood pressure and arterial stiffness in isolated systolic hypertension. Hypertension 2009; 54:409–413. [DOI] [PubMed] [Google Scholar]
- 29.Morgan T, Lauri J, Bertram D, Anderson A. Effect of different antihypertensive drug classes on central aortic pressure. Am J Hypertens 2004; 17:118–123. [DOI] [PubMed] [Google Scholar]
- 30.Williams B, Lacy PS, Thom SM, Cruickshank K, Stanton A, Collier D, et al. CAFE Investigators, Anglo-Scandinavian Cardiac Outcomes Trial Investigators, CAFE Steering Committee and Writing Committee. Differential impact of blood pressure-lowering drugs on central aortic pressure and clinical outcomes: principal results of the Conduit Artery Function Evaluation (CAFE) study. Circulation 2006; 113:1213–1225. [DOI] [PubMed] [Google Scholar]
- 31.Sharman JE, Marwick TH, Gilroy D, Otahal P, Abhayaratna WP, Stowasser M. Value of Central Blood Pressure for GUIDing ManagEment of Hypertension Study Investigators. Randomized trial of guiding hypertension management using central aortic blood pressure compared with best-practice care: principal findings of the BP GUIDE study. Hypertension 2013; 62:1138–1145. [DOI] [PubMed] [Google Scholar]
- 32.Borlaug BA, Olson T, Abdelmoneim Mohamed S, Melenovsky V, Sorrell VL, Noonan K, et al. A randomized pilot study of aortic waveform guided therapy in chronic heart failure. J Am Heart Assoc 2014; 3:e000745. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Bo Y, Kwok KO, Chung VC, Yu CP, Tsoi KK, Wong SY, Lee EK. Short-term reproducibility of ambulatory blood pressure measurements: a systematic review and meta-analysis of 35 observational studies. J Hypertens 2020; 38:2095-2109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Lee EK, Choi RC, Liu L, Gao T, Yip BH, SY W. Preference of blood pressure measurement methods by primary care doctors in Hong Kong: a cross-sectional survey. BMC Fam Pract 2020; 21:1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Vlachopoulos C, Aznaouridis K, O’Rourke MF, Safar ME, Baou K, Stefanadis C. Prediction of cardiovascular events and all-cause mortality with central haemodynamics: a systematic review and meta-analysis. Eur Heart J 2010; 31:1865–1871. [DOI] [PubMed] [Google Scholar]
- 36.Hope SA, Meredith IT, Cameron JD. Arterial transfer functions and the reconstruction of central aortic waveforms: myths, controversies and misconceptions. J Hypertens 2008; 26:4–7. [DOI] [PubMed] [Google Scholar]
- 37.Rahman M, Hsu JY, Desai N, Hsu CY, Anderson AH, Appel LJ, et al. Central blood pressure and cardiovascular outcomes in chronic kidney disease. Clin J Am Soc Nephrol 2018; 13:585–595. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.McEniery CM, Cockcroft JR, Roman MJ, Franklin SS, Wilkinson IB. Central blood pressure: current evidence and clinical importance. Eur Heart J 2014; 35:1719–1725. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Brady TM, Charleston J, Ishigami J, Miller ER, 3rd, Matsushita K, Appel LJ. Effects of different rest period durations prior to blood pressure measurement: the Best Rest Trial. Hypertension 2021; 78:1511–1519. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.