Arterial Stiffness and Gait Speed in Older Adults with and without Peripheral Arterial Disease

Abstract

BACKGROUND

Central arterial stiffness is increasingly recognized as an important predictor of cardiovascular events and mortality in older adults; however, few studies have evaluated the association of arterial stiffness with mobility decline, a common consequence of vascular disease.

METHODS

We analyzed the association of pulse wave velocity (PWV), a measure of aortic stiffness, with longitudinal gait speed over seven years in 2,172 participants in the Health, Aging and Body Composition (Health ABC) Study (mean age ± SD 73.6 ± 2.9 years, 48% men, 39% black).

RESULTS

In mixed-effects models adjusted for demographics, each SD (396 cm/s) higher PWV was associated with 0.015 (SE 0.004) m/s slower gait at baseline and throughout the study period in the full cohort (p < 0.001); this relationship was largely explained by hypertension and other vascular risk factors. Among participants with peripheral arterial disease (PAD) (n = 261; 12.7%) each SD higher PWV was independently associated with 0.028 (SE 0.010) m/s slower gait speed at baseline and throughout the study period (p < 0.01).

CONCLUSIONS

These findings suggest that aortic stiffness may be especially detrimental to mobility in older adults with already compromised arterial function.

INTRODUCTION

Central arterial stiffness is the primary determinant of hypertension in older adults and an important predictor of cardiovascular events and mortality independent of systolic blood pressure and other vascular risk factors ¹. Clinical consequences of arterial stiffness have been attributed to early wave reflection from the periphery and concurrent loss of cushioning capacity of the aorta ². These hallmarks of vascular aging increase cardiac afterload while failing to bolster diastolic filling, and allow transmission of highly pulsatile flow to the fragile small vessels of the brain and kidney ^2–5. Associated damaging central pressures may contribute to resistance vessel hypertrophy and increased vascular resistance ^{6, 7}, impeding blood flow to the peripheral microvasculature ⁸; however, the relationship of aortic stiffness with mobility decline, a common consequence of vascular disease, has not been well-characterized in older adults living in the community.

Several studies of older adults have identified obstructive peripheral arterial disease (PAD) as an important predictor of incident physical disability or gait speed decline ^9–11. Among patients with PAD, reductions in aortic stiffness by an angiotensin-converting enzyme (ACE) inhibitor were recently found to improve walking endurance beyond the extent expected by blood pressure lowering ¹², suggesting that aortic stiffness may contribute to compromised leg perfusion or diastolic function in PAD. To evaluate whether aortic stiffness may also in part explain mobility decline in initially well-functioning older adults, we analyzed the association of aortic pulse wave velocity (PWV) with longitudinal gait speed in the Health, Aging, and Body Composition Study and evaluated its role in participants with PAD. Independent associations of PWV with gait speed in this cohort may suggest functional consequences of diastolic dysfunction or reduced microvascular perfusion in the presence of aortic stiffness ¹².

METHODS

The Health ABC Study and Participants

From 1997 to 1998 the Health ABC study enrolled 3,075 Medicare-eligible nondisabled men and women aged 70–79 from Pittsburgh, Pa. and Memphis, Tn., USA. The population was 52% women and 42% black with a mean age of 73.6 years. Participants were recruited from Medicare-eligible adults with contact information provided by the Centers for Medicare & Medicaid Services (formerly the Health Care Financing Administration) on all black beneficiaries and a random sample of white beneficiaries in pre-designed zip code areas surrounding the study centers. Other household members aged 70–79 were also eligible for recruitment. Exclusion criteria included reported difficulty walking one quarter of a mile, climbing 10 steps without resting, or performing basic activities of daily living or need for a walking aid. The Institutional Review Boards of the University of Pittsburgh, Pa. as well as the University of Tennessee at Memphis approved the study and written, informed consent was obtained from each volunteer. PWV data were missing for 354 participants due to an equipment problem and an additional 233 participants who had unusable waveforms; those with missing PWV data were more often black (47% versus 40%) and had slightly lower blood pressures (134 versus 136 mmHg systolic blood pressure) and higher heart rates (67 versus 65 beats per minute). Of the remaining 2,488 participants, 2,172 had valid gait speed data at baseline and were included in the analysis (mean age ± SD 73.6 ± 2.9 years, 48% men, 39% black).

Pulse wave velocity

Considered the gold standard measure of central arterial stiffness ¹³, aortic PWV was evaluated noninvasively via simultaneous Doppler-recorded carotid and femoral pulse waveforms (model 810A, 9.0- to 10-MHz probes, Parks Medical Electronics, Inc). A minimum of ten beats were recorded for each simultaneous recording site. Three separate runs were recorded for each participant, and all usable runs were averaged to calculate the final PWV measure. The distance between the carotid and femoral recording sites was measured above the surface of the body with a tape measure. The time delay between the feet of the pressure waves at each site was divided by the associated distance to calculate PWV in cm/s; higher values represent greater aortic stiffness. Replicate measures of PWV in 14 subjects revealed intraclass correlations of 0.88 between sonographers and 0.84 between readers ¹.

Peripheral arterial disease

Blood pressures were measured in the right arm and both ankles (posterior tibial artery) using standard cuffs and a pencil Doppler flow probe (Parks Medical Electronics, Inc., Aloha, Oregon). The systolic blood pressure of the ankle was divided by the systolic blood pressure of the arm to calculate the ankle-arm index (AAI), or ankle-brachial index. Measures were performed twice and the results averaged; the lower of the values between the two legs was used to define an individual’s AAI. Peripheral arterial disease (PAD) was defined as an AAI less than 0.9, according to traditional diagnostic criteria ¹⁴.

Gait speed

Usual gait speed was assessed over a 20-meter straight course at baseline and annually through Year 8, except Year 7. Participants were instructed to walk at their usual pace from the starting point to the end of the course. Timing began at the first footfall over the starting line and ended with the first footfall over the finishing line.

Covariates

We considered as potential confounders variables that were identified in the literature as confounders of the relationship between vascular disease and physical function, or were associated with arterial stiffness and gait speed in this cohort with p < 0.15. Selected covariates included smoking status, body-mass index (BMI), physical activity, resting heart rate, systolic blood pressure, total cholesterol, prevalent coronary heart disease, diabetes and hypertension. Presence of chronic conditions was determined from participant reports from the baseline visit and confirmed by use of specific medications or procedures ¹⁵. BMI was calculated as measured weight in kilograms divided by measured height in meters squared. Systolic blood pressure was measured twice and the results averaged. Total cholesterol was determined from fasting blood samples collected at the baseline clinic visit. Physical activity was evaluated using a standardized questionnaire designed specifically for the Health ABC study, modeled from commonly used physical activity assessments including the leisure-time physical activity questionnaire ¹⁶. Physical activity data were calculated by multiplying the metabolic equivalent of each task (kcal) by the amount of time spent during the week doing the activity ¹⁶; participants with at least 1,000 kcal expended were defined as active ¹⁷.

Because aortic stiffness is associated with both overt and subclinical cerebrovascular disease ^{1, 18}, we also evaluated whether an association of PWV with gait speed may be explained by prevalent cerebrovascular disease or performance on the Digit Symbol Substitution Test (DSST), a test of attention and psychomotor speed ¹⁹ that is sensitive to cerebral small-vessel disease ²⁰ and predicts incident stroke in older adults ²¹.

Statistical Analysis

Differences in baseline characteristics across quartiles of PWV and gait speed were evaluated using Chi-square tests for categorical variables and Kruskal-Wallis tests or analysis of variance for continuous variables. Mixed-effects models were used to evaluate the association of PWV with longitudinal gait speed over seven years. Coefficients for fixed effect variables represented the association of the variable with gait speed (m/s) at baseline and throughout the study period; interactions of fixed effect variables with time represented the contribution of the variable to the rate of gait speed decline (m/s) per year. Simple models included a random intercept for subject, a random slope with time, clinic site and PWV as fixed effects, and the interaction of PWV with time. A second model additionally adjusted for demographics (age, sex, and race) and the interaction of each covariate with time. A backward procedure (p-out = 0.05) was used to select a final model (Table 3; Model 3) following additional adjustment for vascular risk factors and chronic conditions, the interaction of each covariate with time and the interaction of PWV with PAD. Analyses were then repeated in subgroups stratified by PAD and additionally adjusted for prevalent cerebrovascular disease, DSST score, and AAI. To evaluate the sensitivity of estimates to missing data, we repeated the analysis of the full sample after limiting followup to four years. No structure was imposed on the covariance matrix of the random effects. Continuous covariates were centered to simplify interpretation of model coefficients and to reduce multicollinearity. Finally, associations with longitudinal gait speed were compared among PWV and pulse pressure, an indirect measure of arterial stiffness calculated as systolic blood pressure – diastolic blood pressure. Statistical significance for all tests was set to the conventional level of p < 0.05. Analyses were performed using Stata 10 (STATA, Houston, Texas).

Table 3.

Coefficients estimated by mixed models of PWV as a predictor of longitudinal gait speed (m/s)

	Model 1		Model 2		Model 3
Full cohort (n = 2172)	Coefficient (95% CI)	p-value	Coefficient (95% CI)	p-value	Coefficient (95% CI)	p-value
PWV (SD)	−0.021 (−0.029, −0.013)	<0.001	−0.015 (−0.023, −0.008)	<0.001	−0.005 (−0.012, 0.002)	0.155
Time (yr)	−0.041 (−0.042, −0.039)	<0.001	−0.039 (−0.041, −0.037)	<0.001	−0.037 (−0.040, −0.035)	<0.001
AAI < 0.9 (n = 261)
PWV (SD)	−0.032 (−0.052, −0.011)	0.002	−0.030 (−0.049, −0.011)	0.002	−0.028 (−0.047, −0.010)	0.003
Time (yr)	−0.044 (−0.048, −0.040)	<0.001	−0.046 (−0.054, −0.037)	<0.001	−0.048 (−0.058, −0.038)	<0.001
AAI ≥ 0.9 (n = 1802)
PWV (SD)	−0.015 (−0.024, −0.006)	0.001	−0.011 (−0.019, −0.003)	0.009	−0.001 (−0.009, 0.007)	0.804
Time (yr)	−0.040 (−0.042, −0.039)	<0.001	−0.038 (−0.041, −0.036)	<0.001	−0.037 (−0.039, −0.034)	<0.001

Model 1 adjusted for Memphis versus Pittsburgh site

Model 2 additionally adjusted for age, sex, race, and interaction of each covariate with time

Model 3 additionally adjusted for BMI, systolic blood pressure, heart rate, smoking, physical activity, coronary heart disease, diabetes, hypertension and interaction of hypertension with time

RESULTS

Mean age (SD) of the cohort was 73.6 (2.9) years; 48% were men and 39% were black (Table 1). Participants in higher quartiles of PWV (mean (SD) = 898 (396) cm/s) were older, more likely to be male, black, less active, and have higher BMI, systolic blood pressure, resting heart rate, a history of smoking, and prevalent diabetes and hypertension (Table 1). Higher PWV quartiles were also associated with PAD, lower DSST score and slower baseline gait speed (Table 2).

Table 1.

Baseline characteristics of the cohort by quartiles (range in parentheses) of PWV (cm/s)

Mean ± SD or %
	Total (312–2998) n=2172	Q1 (312–636) n=543	Q2 (636–800) n=545	Q3 (801–1046) n=541	Q4 (1047–2998) n=543	p-value*
Age, years	73.6 ± 2.9	73.3 ± 2.8	73.5 ± 2.8	73.9 ± 2.9	73.9 ± 2.8	<0.01
Men	48.1	44.9	44.2	50.1	53.2	<0.01
Black	38.5	33.5	37.3	39.9	43.5	<0.01
BMI, kg/m2	27.3 ± 4.7	26.1 ± 4.4	27.4 ± 4.5	28.1 ± 4.8	27.4 ± 4.8	<0.01
Systolic blood pressure, mmHg	135.5 ± 19.3	129.8 ± 19.3	133.9 ± 18.6	137.5 ± 19.0	140.6 ± 19.5	<0.01
Diastolic blood pressure, mmHg	71.6 ± 10.9	70.5 ± 10.4	71.6 ± 10.6	71.8 ± 11.2	72.6 ± 11.4	0.02
Pulse pressure, mmHg	63.9 ± 16.8	59.3 ± 15.6	62.3 ± 15.9	65.7 ± 16.9	68.0 ± 17.5	<0.01
Heart rate, bpm	64.5 ± 10.6	62.0 ± 9.9	64.4 ± 9.9	65.4 ± 10.5	66.4 ± 11.5	<0.01
Cholesterol, mg/dL	203.7 ± 37.8	203 ± 35.0	204.7 ± 36.3	204.5 ± 40.8	202.7 ± 38.9	0.74
Current or former smoker	55.4	49.5	54.1	59.7	58.2	<0.01
Physical activity ≥ 1000 kcal/wk	32.9	37.6	33.0	31.1	29.8	0.04
Coronary heart disease	18.4	15.9	17.2	21.3	19.3	0.11
Diabetes	14.0	8.3	12.0	17.0	19.6	<0.01
Hypertension	49.0	36.9	46.6	52.2	60.5	<0.01

^*Analysis of variance

Table 2.

Baseline characteristics of the cohort by quartiles (range in parentheses) of PWV (cm/s)

Mean ± SD or %
	Total (312–2998) n=2172	Q1 (312–636) n=543	Q2 (636–800) n=545	Q3 (801–1046) n=541	Q4 (1047–2998) n=543	p-value*
Gait speed, m/s	1.34 ± 0.25	1.38 ± 0.25	1.35 ± 0.25	1.31 ± 0.26	1.32 ± 0.24	<0.01
DSST score	36.5 ± 14.5	39.2 ± 15.1	37.6 ± 14.1	34.7 ± 14.3	34.7 ± 13.8	<0.01
Cerebrovascular disease, %	7.4	6.9	6.3	6.9	9.3	0.25
AAI < 0.9, %	12.7	8.1	10.3	14.3	18.2	<0.01

^*Analysis of variance

DSST = Digit Symbol Substitution Test

In mixed models adjusted for demographics, each SD higher PWV was associated with 0.015 m/s slower gait speed at baseline and throughout the study period, represented by the significant main effect of PWV (p<0.001) (Table 3). The effect was attenuated after additional adjustment for vascular risk factors and chronic conditions (Model 3); the covariates BMI, systolic blood pressure, heart rate, diabetes and hypertension were particularly important to explain the association of PWV with gait speed. No significant interaction was identified for PWV with time (Beta (SE) = 0.000 (0.001); p = 0.66), indicating no significant contribution of PWV to the longitudinal decrease in gait speed.

Although the interaction of PWV with PAD was not statistically significant (p = 0.21), in analyses restricted to participants with PAD (n = 261; 12.7%), each SD higher PWV was associated with 0.028 (SE 0.010) m/s slower gait speed at baseline and through the study period after adjustment for demographics, risk factors and chronic conditions (p<0.01) (Table 3). This effect did not substantially change after additional adjustment for prevalent cerebrovascular disease, DSST score, or AAI. As in the full cohort, the interaction of PWV with time was nonsignificant (Beta (SE) = 0.000 (0.002); p = 0.96).

In the full cohort, significant interactions with time were identified for the covariates age, sex, race, hypertension and cerebrovascular disease in a full model of gait speed decline. Calculated contributions to annual gait speed decline, represented by interaction terms for each covariate with time, were modest although statistically significant: −0.001 m/s per year older age at baseline; −0.006 m/s each for male sex and white race; −0.003 m/s for prevalent hypertension; and −0.007 m/s for cerebrovascular disease (p<0.05 for each). Pulse pressure was not independently associated with gait speed in the full cohort or among participants with PAD.

Finally, 1673 (77%) of participants completed gait speed testing at Year 5 and 1219 (56%) at Year 8. In analyses restricted to followup through Year 5, estimates for PWV and the interaction of PWV with time did not substantially change, suggesting reported estimates were not sensitive to missing observations.

DISCUSSION

Although PWV was not independently associated with gait speed in this initially well-functioning cohort, among those with evidence of PAD, higher PWV was associated with slower gait independent of hypertension and other vascular risk factors. These data suggest that aortic stiffness may be especially detrimental to mobility in older adults with already compromised arterial function.

Cardiovascular consequences of arterial stiffening may be explained by gradual loss of the cushioning capacity of the aorta and early wave reflection from the periphery, resulting in small-vessel injury of target organs ^{2–4, 22} and impaired diastolic function ²³. Current findings are consistent with those of a previous publication that reported an independent association of arterial stiffness with impaired functional capacity in patients evaluated for peripheral arterial disease (PAD) ²⁴, and further suggest a clinically important synergistic relationship of aortic stiffness and PAD. In fact, among PAD patients, reductions of PWV by an ACE inhibitor were recently found associated with improved performance on a treadmill walk test, consistent with the hypothesis that arterial stiffness may in part mediate the relationship of PAD with functional impairment by contributing to resistance vessel hypertrophy and increased vascular resistance. These data suggest a potential to improve peripheral perfusion and walking distance through the reduction of arterial stiffness in individuals with PAD ¹².

Although we did not identify an association of PWV with rate of gait speed decline, this analysis may be limited by the initially high mobility status of the sample, in which no participants reported difficulty walking and only 7% had gait speed below 1.0 m/s, a clinically relevant cut point found predictive of lower extremity limitation and hospitalization ²⁵. It is also possible that arterial stiffness may have contributed to gait decline prior to our observation, resulting in slower gait observed throughout the study period in participants with PAD who also had elevated PWV. This finding is consistent with an influence of the oscillatory component of blood pressure, or pulse pressure, on the structure and function of the peripheral vasculature in PAD ²⁴. With widening of the pulse pressure, primarily determined by altered wave reflections with aortic stiffness, lower diastolic pressure may be inadequate to overcome increased peripheral vascular resistance associated with both obstructive atherosclerosis and arterial stiffness ²⁴. Resulting insufficient leg perfusion during muscle contraction may account for slower walking speed even in the absence of intermittent claudication. Because the association of aortic stiffness with slower gait in PAD was not explained by AAI, these data may suggest a value for PWV in identifying individuals with PAD who are at highest risk of functional decline.

The independent association of hypertension with accelerated gait decline in the full cohort is consistent with previous data ²⁶ including an apparent contribution of cerebral microvascular disease to functional decline in aging ^27–29. Considered evidence of cerebral small vessel injury ³⁰, hyperintensities of white matter tracts on brain MRI commonly manifest with elevated systolic blood pressure ³¹ and predict gait speed decline and incident physical impairment in older adults ²⁷. The relationship of hypertension with functional decline may be further mediated by chronic inflammation ^32–34, oxidative stress ³⁵, alterations in the renin-angiotensin aldosterone system ³⁶, or deficits in executive cognitive function ³⁷. Although AAI was previously found not associated with C-reactive protein, interleukin-6 or tumor necrosis factor-α in Health ABC ³⁸, the independent association of aortic stiffness with gait speed in PAD may reflect underlying chronic inflammation not captured by these available markers or traditional vascular risk factors. It is also possible that subtle cognitive deficits among participants with PAD may exacerbate physical functional declines ³⁹; a detailed cognitive assessment is warranted to further evaluate cognitive change as mediator of gait slowing in PAD.

Although 0.1 m/s is generally considered clinically meaningful gait speed decline ⁴⁰, the relevance of PWV among participants with PAD may be interpreted in relative terms; the estimated association of one standard deviation higher PWV with 0.028 m/s slower gait was equal to that of 2 years older age at baseline. This effect size is likely underestimated due to the restriction of the analysis to those who were initially free of functional limitation, in addition to the potential nonrandom withdrawal of participants with chronic conditions.

Study Limitations

Gait speed and other lower-extremity performance measures may not capture subtle functional changes in healthier older adults; rather, extended walking tests ^{41, 42} may be more sensitive to diastolic dysfunction associated with aortic stiffness. Measures of brain integrity were not available to evaluate the hypothesis that cerebrovascular abnormalities may mediate the association of arterial stiffness with gait speed in PAD. Finally, it is important to acknowledge that the subgroup analysis was performed following the apriori evaluation of arterial stiffness and gait speed in the full Health ABC cohort; the association identified among participants with PAD warrants replication in other older adult populations.

Strengths of this study include the large, community-based population, ability to account for several potential confounders, and repeated assessment of gait speed, a reliable, valid measure of physical function that predicts incident disability in older adults ⁴³.

In summary, arterial stiffness was inversely related to gait speed in participants with PAD, but was not independently associated with gait speed or the rate of gait speed decline in the full cohort. The association of hypertension with accelerated gait speed decline in this cohort reinforces the value of blood pressure control in maintaining function in aging. Current evidence implicates arterial stiffness as a potential target of interventions to improve peripheral perfusion and walking performance in PAD. More work is needed to evaluate whether reduction of aortic stiffness may reduce risk of functional decline and disability in PAD independent of blood pressure lowering.