Introduction

Arterial stiffness increases with advancing age and is known to lead to cardiovascular disease (CVD).1 It is important to assess arterial stiffness in older adults because CVDs, such as stroke and heart disease, are the major causes of death in these individuals. Pulse wave velocity (PWV) is known to be an indicator of arterial stiffness.2 Aortic PWV, as assessed by determining carotid-femoral PWV (cfPWV), has been reported to be associated with an increased risk of total or cardiovascular mortality in clinical populations, such as patients with end-stage renal disease,3 hypertension4, 5 or diabetes6 as well as in the general population.7, 8, 9, 10 Of the previous studies, only one examined this association in community-dwelling older adults.7

Brachial–ankle PWV (baPWV) can be measured with a simple and noninvasive device that estimates arterial stiffness. Cross-sectional studies have reported an association between high baPWV and prevalence of cardiovascular risk factors,11, 12 presence of coronary artery disease in male patients,13 presence of cerebrovascular disease in asymptomatic community-dwelling older persons14 and in hypertensive patients aged 50 years or older.15 However, there have been few longitudinal reports about the relationship between baPWV and mortality.16, 17, 18 One study suggested that increases in baPWV were associated with an elevated risk of cardiovascular mortality in 298 community-dwelling older adults, but the optimal cutoff value of baPWV was not shown.17

The purpose of this study was to evaluate the relationship between baPWV and 3-year mortality and to determine the optimal cutoff value of baPWV in terms of mortality among community-dwelling older adults.

Methods

Study population

All subjects were registered in the Kahoku longitudinal study, which was designed to prolong healthy life expectancy by providing health screening, counseling and educational services to older persons living in Kahoku, a rural town in Japan.19, 20 In this community of 5596 residents, 2092 (37.4%) were aged 65 years and older in 2000. Each year from 2000 to 2003, every individual aged 65 years or older was invited to the annual health checkup organized by the municipal health center. Without considering duplicates, the number of persons who participated in the checkup was 348 in 2000, 337 in 2001, 384 in 2002 and 361 in 2003. The first visit to the checkup service was defined as entry into the present longitudinal study. Of the 577 individuals who entered the study, 18 who had lumbar pain and gibbosity were excluded because measurement of baPWV was not applicable. The remaining 559 individuals (220 men and 339 women; mean age±s.d. at entry, 76.6±5.6 years)—26.7% of the number of Kahoku residents aged 65 years or older in 2000—were invited to have their baPWV measured. Of these 559 subjects, 29 were excluded; 15 had atrial fibrillation, 5 had a medical history of arteriosclerosis obliterans and 9 had an abnormal ankle/brachial pressure index of less than 0.9 as determined by plethysmography. Therefore, the effective sample size of this study was 530 subjects (207 men and 323 women; mean age at entry, 76.4±5.6 years).

Ethical consideration

Written informed consent was obtained from each person at entry into the study. The study protocol was approved by the research ethics committee of Kochi Medical School, Kochi University, Japan.

Study variables

Medical history and examinations

Medical history was self-reported and confirmed by a physician at the baseline health checkup. Participants were defined as having a medical history of CVD if they had a history of stroke or heart disease such as ischemic heart disease, heart failure or arrhythmia. Body height and weight were measured in the health checkup at entry into the study. Body mass index was defined as weight (kg) divided by the square of height (m2). Blood pressure (BP) measurements were performed in the morning in the health checkup at study entry. BP and pulse rate were recorded twice in the upper arm with an appropriately sized cuff at the level of the heart with the subject in a sitting position after a rest of at least 5 min, using an autosphygmomanometer (BP-203I; Colin Co., Komaki, Japan) according to the cuff-oscillometric method. For the analysis, we used an average of two measurements for BP and pulse rate.

Brachial–ankle pulse wave velocity measurements

baPWV was automatically measured using a form PWV/ABI instrument (Colin Co.), as previously described.11, 21 This device has four cuffs matched with oscillometric sensors, and the cuffs were wrapped around the upper arms and the ankles. The volume pulse forms of the bilateral brachial and tibial arteries were monitored during continuous deflation of the cuffs. The baPWV value was calculated by time-phase analysis of the right brachium and ankle and the left brachium and ankle, respectively. baPWV was measured after the subject had lain supine for at least 5 min. Measurements were performed twice, consecutively. An average of two baPWV measurements was used. Because there was a significant positive correlation between left and right baPWV (r=0.954, P<0.001), the right baPWV value was used in this study.

The participants were dichotomized according to a median value of baPWV. Those with baPWV <18.675 m s−1 (n=265) formed the low-baPWV group, and the subjects with baPWV 18.675 m s−1 (n=265) constituted the high-baPWV group.

Laboratory variables

All participants had blood drawn in the health checkup at entry into this study. The samples were placed in cold storage immediately after collection and were analyzed within 48 h. Blood sugar, hemoglobin A1c, total cholesterol, high-density lipoprotein cholesterol, triglyceride, hemoglobin and albumin levels were determined.

Follow-up survey for prognosis

Survivorship of the participants for 3 years from entry into the study was followed. No study participant moved out of the municipality during the period. For those who died, we recorded the cause of death from the death certificate submitted to the municipality. Causes of death were coded according to the International Classification of Diseases, ninth edition.

Statistical analysis

All continuous variables were expressed as the mean±s.d. The mean values between the two groups were compared using the Student's t-test. The χ2-test was used to compare the proportions of specific categories for categorical variables. The Kaplan–Meier method was used to estimate unadjusted survival curves of the two baPWV groups. The log-rank test was used to compare the unadjusted survival curves. A Cox proportional hazards model was used to describe the association between baPWV and mortality while adjusting for age, sex and BP measures. A receiver-operating characteristic (ROC) curve analysis was performed to identify the optimal cutoff value of baPWV for predicting total mortality and cardiovascular mortality. The value with the highest sum of sensitivity and specificity was used as the optimal cutoff value. P-values <0.05 were considered statistically significant. All analyses were performed using SPSS 15.0J for Windows (SPSS Japan Inc., Tokyo, Japan).

Results

Baseline characteristics

The mean value of baPWV was 18.90±3.95 m s−1 in all subjects. There was no significant difference in the means of baPWV between men and women (men, 18.81±4.01 m s−1; women, 18.97±3.92 m s−1, P=0.647). Table 1 shows the baseline characteristics according to the baPWV groups. The high-baPWV group was older and had a higher proportion of subjects taking antihypertensive medication than the low-baPWV group. There were no significant differences in body mass index, the proportion of current smokers, the use of medications for diabetes mellitus and hyperlipidemia or history of CVDs between the two groups. Systolic BP (SBP), diastolic BP, pulse pressure and pulse rate were higher in the high-baPWV group as compared with the low-baPWV group. No significant difference was found in the blood test results between the two groups.

Table 1 Baseline characteristics according to the level of baPWV

Three-year mortality according to baPWV

In the 3 years after the onset of the study, 30 persons died, 10 of whom died as a result of CVDs. Table 2 shows the number and percentage of all-cause and cause-specific deaths during the 3 years. The high-baPWV group had higher proportions of both total deaths and deaths due to CVDs. There were no significant differences in the proportions of deaths due to other causes.

Table 2 Three-year cause-specific mortality according to the level of baPWV

Kaplan–Meier curves showed that the high-baPWV group had a significantly higher probability of overall mortality (Figure 1a) and also of cardiovascular mortality (Figure 1b). After adjustment for age, sex and SBP, a high-baPWV level was significantly associated with an increased risk of 3-year total mortality (adjusted HR=2.98, 95% CI=1.25–7.07) and cardiovascular mortality (adjusted HR=10.01, 95% CI=1.21–82.49) (Table 3). When baPWV was treated as a continuous variable, a 1 m s−1 increase in baPWV was significantly associated with an increased risk of 3-year total mortality (adjusted HR=1.09, 95% CI=1.00–1.18) and cardiovascular mortality (adjusted HR=1.12, 95% CI=1.01–1.25) after adjustment for age, sex and SBP. After adjustment for antihypertensive medication use in place of SBP, a high-baPWV level was still associated with an increased risk of 3-year total mortality (adjusted HR=2.66, 95% CI=1.17–6.05) and cardiovascular mortality (adjusted HR=8.53, 95% CI=1.07–67.76). This association remained after adjustment for pulse pressure in place of SBP (total mortality: adjusted HR=2.68, 95% CI=1.16–6.15; cardiovascular mortality: adjusted HR=9.09, 95% CI=1.14–72.65). After adjustment for pulse rate in place of SBP, this association again remained (total mortality: adjusted HR=2.45, 95% CI=1.07–5.62; cardiovascular mortality: adjusted HR=8.48, 95% CI=1.07–67.45), as well as after adjustment for medical history of CVD in addition to age, sex, and SBP (total mortality: adjusted HR=2.97, 95% CI=1.25–7.09; cardiovascular mortality: adjusted HR=9.01, 95% CI=1.08–74.90).

Figure 1
figure 1

(a) Kaplan–Meier curves of total mortality. Survival probabilities according to brachial–ankle pulse wave velocity. (b) Kaplan–Meier curves of cardiovascular mortality. Survival probabilities according to brachial–ankle pulse wave velocity.

Table 3 The relation between baPWV and 3-year all-cause/cardiovascular mortality

Figure 2 shows the ROC curves used to define the optimal cutoff value of baPWV in relation to 3-year total/cardiovascular mortality. The area under the ROC curve for total mortality was 0.673 (95% CI=0.586–0.760), and that for cardiovascular mortality was 0.795 (95% CI=0.701–0.890). The optimal cutoff value of baPWV for total mortality was 19.63 m s−1 (73% sensitivity and 63% specificity) and that for cardiovascular mortality was 19.63 m s−1 (91% sensitivity and 62% specificity). A Cox proportional hazards model showed that a baPWV level 19.63 m s−1 was significantly associated with an increased risk of 3-year total mortality (adjusted HR=5.3, 95% CI=2.2–12.7) and cardiovascular mortality (adjusted HR=18.7, 95% CI=2.2–157.6) after adjustment for age, sex and SBP. The optimal cutoff value produced higher point estimates of HRs than the cutoff at the median value.

Figure 2
figure 2

Receiver-operating characteristic curves to define the optimal cutoff value of baPWV in relation to 3-year total/cardiovascular mortality. Total mortality: area under the curve=0.673 (95% CI=0.586–0.760); optimal cutoff of baPWV=19.63 m s−1 (73% sensitivity and 63% specificity). Cardiovascular mortality: area under the curve=0.795 (95% CI=0.701–0.890); optimal cutoff of baPWV=19.63 m s−1 (91% sensitivity and 62% specificity).

Discussion

This study showed that high baPWV was predictive of increased mortality, especially increased cardiovascular mortality, in community-dwelling older adults. This finding indicates that baPWV might be a useful marker of arterial stiffness in older adults. Arterial stiffness increases with advancing age and is accompanied by structural changes to the arterial system, including fragmentation and degeneration of elastin and increases in collagen.1 An increase in arterial stiffness results in premature return of reflected waves in late systole and increases central SBP and pulse pressure. Therefore, systolic hypertension and increased pulse pressure are known to be markers of arterial stiffness and have been reported to be related to higher total/cardiovascular mortality in older adults.22, 23 In this study, SBP and pulse pressure had positive correlations with baPWV (SBP: r=0.346, P<0.01; pulse pressure: r=0.211, P<0.01). Moreover, high baPWV was associated with higher total/cardiovascular mortality, adjusted for SBP or pulse pressure. These results suggest that baPWV is a predictor of total/cardiovascular mortality independently of SBP and pulse pressure.

Aortic stiffness is closely related to total and cardiovascular mortality.3, 4, 5, 6, 7, 8, 9, 10 cfPWV is regarded to be the gold standard of aortic PWV measurement, and the association between cfPWV and cardiovascular mortality has been reported in older adults.7, 24 In contrast to cfPWV measurements, which require a specialized technique, baPWV is a new method for measuring PWV that does not require a specialized technique. Four cuffs matched with oscillometric sensors are wrapped around the upper arms and the ankles. A high reproducibility of baPWV measurements obtained with this method has been reported (correlation coefficient: interobserver reproducibility=0.98, intraobserver reproducibility=0.87).21 Another feature of baPWV measurements is that they yield data for both the central and peripheral arteries. baPWV has been reported to be closely correlated with aortic PWV obtained by an invasive measurement using a catheter tip manometer (correlation coefficient=0.87)21 and with cfPWV (correlation coefficient=0.73).25 baPWV was comparable to cfPWV in predicting the presence of both stroke and coronary artery disease.25

Although 12.00 m s−1 of cfPWV was recommended as a cutoff value in hypertensive patients,26 the optimal cutoff value of cfPWV in older adults has not been reported. With regard to the cutoff value of baPWV, 17.00 m s−1 was the optimal cutoff value for predicting recurrence of cardiovascular events, including coronary reintervention, reinfarction and readmission for congestive heart failure, stroke and cardiac death, in patients with history of hospitalization as a result of acute coronary syndrome.16 In addition, 18.00 m s−1 was the optimal cutoff value for predicting recurrence of cardiovascular events, including reinfarction and readmission for congestive heart failure, stroke and cardiac death in patients with acute coronary syndrome.16 This study showed the optimal cutoff value of baPWV in relation to 3-year total/cardiovascular mortality, and it suggested that the cutoff value of baPWV in community-dwelling older adults was higher than that in clinical populations.

A potential intervention for the improvement of arterial compliance might be targeted to community-dwelling older adults with high baPWV. It was reported that 3 months of aerobic exercise and brisk walking increased central arterial compliance in middle-aged and older healthy men.27 In addition, it was reported that aortic PWV was reduced by dietary sodium restriction in healthy postmenopausal women.28 Therefore, education for changing a sedentary lifestyle and reducing sodium intake might be considered for community-dwelling older adults with high baPWV. Among the limitations of this study were the small number of participants and the short follow-up period. Consequently, the number of total deaths, including cardiovascular deaths, was small. Therefore, the analysis of the association between baPWV and prognosis, especially between baPWV and cause-specific mortality, may not be reliable. This study should thus be considered preliminary; a longitudinal study with a large cohort is necessary to confirm the relationship with long-term mortality. Second, it has been reported that antihypertensive drugs decrease baPWV, and the degree of the decrease varies according to the class of antihypertensive drug.29 Although in our study the presence or absence of on medication use was self-reported and confirmed by a physician, on the classes of antihypertensive drugs was not confirmed. A study of a population including a larger number of subjects taking distinct classes of antihypertensive drugs is necessary.

In conclusion, high baPWV was associated with higher mortality in community-dwelling older adults, indicating that noninvasive assessment of arterial stiffness by baPWV may be useful for predicting prognosis.