Kidney Function and Risk of Peripheral Arterial Disease: Results from the Atherosclerosis Risk in Communities (ARIC) Study

Abstract

Chronic kidney disease (CKD) is associated with an increased risk for cardiovascular disease, but its association with peripheral arterial disease (PAD) is unclear. With the use of data from the Atherosclerosis Risk in Communities (ARIC) Study, 14,280 middle-aged adults were categorized on the basis of estimated GFR >/=90, 60 to 89, and 15 to 59 ml/min per 1.73 m(2) for normal kidney function, mildly decreased kidney function, and stages 3 to 4 CKD, respectively. Incident PAD was defined as a new onset of ankle-brachial index <0.9 assessed at regular examinations, new intermittent claudication assessed by annual surveillance, or PAD-related hospital discharges. Incidence rates and relative risks (RR) for PAD were compared across these categories. During a mean follow-up time of 13.1 yr (186,616 person-years), 1016 participants developed PAD. The incidence rates per 1000 person-years were 4.7, 4.9, and 8.6 for the normal kidney function, mildly decreased kidney function, and CKD groups, respectively. Compared with participants with normal kidney function, the age-, gender-, race-, and ARIC field center-adjusted RR for PAD was 1.04 (95% confidence interval [CI] 0.91 to 1.18) for those with mildly decreased kidney function and 1.82 (95% CI 1.34 to 2.47) for those with CKD. After additional adjustment for cardiovascular disease risk factors, an increase in risk for incident PAD still was observed in participants with CKD, with a multivariable adjusted RR of 1.56 (95% CI 1.13 to 2.14). Patients with CKD are at increased risk for incident PAD. Development of strategies for screening and prevention of PAD in this high-risk population seems warranted.

Full text

Kidney Function and Risk of Peripheral Arterial Disease:
Results from the Atherosclerosis Risk in Communities
(ARIC) Study
Keattiyoat Wattanakit,* Aaron R. Folsom,* Elizabeth Selvin,
†
Josef Coresh,
†
Alan T. Hirsch,*
and Beth D. Weatherley
‡
*Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis,
Minnesota;
†
Welch Center for Prevention, Epidemiology and Clinical Research and the Department of Epidemiology,
Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland; and
‡
Duke Clinical Research Institute, Duke
University Medical Center, Durham, North Carolina
Chronic kidney disease (CKD) is associated with an increased risk for cardiovascular disease, but its association with
peripheral arterial disease (PAD) is unclear. With the use of data from the Atherosclerosis Risk in Communities (ARIC) Study,
14,280 middle-aged adults were categorized on the basis of estimated GFR >90, 60 to 89, and 15 to 59 ml/min per 1.73 m
2
for
normal kidney function, mildly decreased kidney function, and stages 3 to 4 CKD, respectively. Incident PAD was defined as
a new onset of ankle-brachial index <0.9 assessed at regular examinations, new intermittent claudication assessed by annual
surveillance, or PAD-related hospital discharges. Incidence rates and relative risks (RR) for PAD were compared across these
categories. During a mean follow-up time of 13.1 yr (186,616 person-years), 1016 participants developed PAD. The incidence
rates per 1000 person-years were 4.7, 4.9, and 8.6 for the normal kidney function, mildly decreased kidney function, and CKD
groups, respectively. Compared with participants with normal kidney function, the age-, gender-, race-, and ARIC field
center–adjusted RR for PAD was 1.04 (95% confidence interval [CI] 0.91 to 1.18) for those with mildly decreased kidney
function and 1.82 (95% CI 1.34 to 2.47) for those with CKD. After additional adjustment for cardiovascular disease risk factors,
an increase in risk for incident PAD still was observed in participants with CKD, with a multivariable adjusted RR of 1.56
(95% CI 1.13 to 2.14). Patients with CKD are at increased risk for incident PAD. Development of strategies for screening and
prevention of PAD in this high-risk population seems warranted.
J Am Soc Nephrol 18: 629–636, 2007. doi: 10.1681/ASN.2005111204
Estimates from the Third National Health and Nutrition
Examination Survey (NHANES III) suggest that 8 mil-
lion people (4.5% of the US population) have chronic
kidney disease (CKD), defined by estimated GFR (eGFR) be-
tween 15 and 59 ml/min per 1.73 m
2
, and approximately
300,000 people (0.1%) have ESRD, defined by eGFR ⬍15 ml/min
per 1.73 m
2
or dialysis treatment (1). It now is recognized that
the clinical course of CKD often is complicated by cardiovas-
cular disease (CVD) (2,3) and death (4,5), independent of es-
tablished risk factors.
Patients with non–dialysis-dependent CKD also might be at
increased risk for developing peripheral arterial disease (PAD).
Most previous studies that investigated this relationship fo-
cused only on patients with dialysis-dependent ESRD, and
studies of populations with non–dialysis-dependent CKD
largely have been examined in cross-sectional surveys (6,7) or
have either not included PAD as an outcome (2,3) or included
PAD in a composite CVD outcome (5,8). To our knowledge,
only one previous study prospectively examined the associa-
tion between CKD and incident PAD (9). This single study
cohort included only postmenopausal women with docu-
mented coronary heart disease (CHD). We therefore conducted
a prospective study to investigate whether level of kidney
function is inversely related to risk for PAD.
Materials and Methods
Study Population
The Atherosclerosis Risk in Communities (ARIC) Study is a large,
prospective, community-based study of the etiology and natural his-
tory of atherosclerosis and CVD. The study cohort comprised 15,792
participants who were aged 45 to 64 yr at baseline in 1987 to 1989 and
were recruited from four US communities: Forsyth County, NC; Jack-
son, MS; suburbs of Minneapolis, MN; and Washington County, MD.
The cohort underwent reexamination visits at roughly 3-yr intervals,
with a 93% return rate for visit 2 (1990 to 1992), 86% for visit 3 (1993 to
1995), and 81% for visit 4 (1996 to 1998). Detailed descriptions of the
ARIC study design and objectives have been published elsewhere (10).
Of the 15,792 ARIC participants, we included 14,390 who had serum
creatinine measured and no history of PAD (ankle-brachial index [ABI]
ⱖ0.9) or intermittent claudication at baseline. We excluded 97 partici-
pants whose race was other than black or white and an additional 13
Received November 19, 2005. Accepted December 4, 2006.
Published online ahead of print. Publication date available at www.jasn.org.
Address correspondence to: Dr. Aaron R. Folsom, Division of Epidemiology and
Community Health, School of Public Health, University of Minnesota, Suite 300,
1300 South Second Street, Minneapolis, MN 55454-1015. Phone: 612-626-8862; Fax:
612-624-0315; E-mail: folsom@epi.umn.edu
Copyright © 2007 by the American Society of Nephrology ISSN: 1046-6673/1802-0629

participants whose eGFR was ⬍15 ml/min per 1.73 m
2
, leaving a total
of 14,280 participants for the final analyses. Most participants had
nearly complete data.
Ascertainment of the Level of Kidney Function
Serum creatinine was measured using the modified kinetic Jaffe
method. Because a number of factors, such as age, ethnicity, and
gender, can influence serum creatinine, the level of kidney function was
ascertained by eGFR, which was calculated using the formula that was
developed and validated in the Modification of Diet in Renal Disease
(MDRD) study (11). The MDRD formula is as follows:
eGFR ⫽186.3 ⫻(serum creatinine
⫺1.154
)⫻(age
⫺0.203
)⫻1.212 (if
black) ⫻0.742 (if female)
For use in this formula, serum creatinine was calibrated by subtrac-
tion of 0.24 (12). We assigned participants with a physiologically im-
plausible high eGFR (n⫽3) to a maximum of 200 ml/min per 1.73 m
2
.
eGFR was divided into the following categories on the basis of the
National Kidney Foundation guidelines: eGFR ⬎90 ml/min per 1.73 m
2
for normal kidney function, eGFR between 60 and 89 ml/min per 1.73
m
2
for mildly decreased kidney function, and eGFR between 15 and 59
ml/min per 1.73 m
2
for stages 3 to 4 CKD, which hereafter is referred
to as CKD.
Ascertainment of Incident PAD
In individuals without prevalent PAD, PAD incidence was charac-
terized by one of the following criteria: (1) A new ABI ⬍0.9 at either
visit 3 or 4; (2) new intermittent claudication based on Rose Question-
naire (13); or (3) a hospital discharge diagnosis of PAD, leg amputation,
or leg revascularization procedures (leg endarterectomy, aorto-iliac-
femoral bypass surgery, or leg bypass surgery).
ABI was measured on nearly all participants at ARIC visit 1 (96.4%),
but only a random sample of participants were invited for ABI mea-
surement during visits 3 (n⫽4197) and 4 (n⫽5882). At visit 1, ABI was
computed by dividing the average of ankle systolic BP (SBP) measure-
ments by the average of brachial SBP measurements (14). Using the
Dinamap 1846 SX automated oscillometric device (Criticon, Tampa,
FL), trained technicians measured two ankle BP, taken 5 to 8 min apart,
at the posterior tibial artery in a randomly selected leg while the
participant was prone. This automated BP measurement device has
high validity compared with the standard Doppler ultrasound mea-
surement and high repeatability (15). Two brachial artery BP were
measured, usually in the right arm, with the participant supine. At
visits 3 and 4, the ABI was defined as the ratio of a single ankle SBP to
a single brachial BP, both measured with the participant supine (16).
Interviewers contacted participants annually by telephone to identify
intermittent claudication symptoms and all hospitalizations. The Rose
Questionnaire (13) was used to evaluate whether participants had
developed intermittent claudication, which was defined as exertional
leg pain relieved within 10 min by resting. Depending on the study, it
has 9 to 92% sensitivity and 95.9 to 100% specificity in identifying PAD
(17). When a hospitalization had occurred, a trained abstractor obtained
and recorded all International Classification of Disease, Ninth Revision
hospital discharge diagnoses. All records with an International Classifi-
cation of Disease, Ninth Revision code of 443.9 (claudication, peripheral
arterial disease not otherwise specified, peripheral angiopathy not oth-
erwise specified, spasm of artery), 84.11 (toe amputation), 84.12 (foot
amputation), 84.15 (below-knee amputation), 84.17 (above-knee ampu-
tation), 38.18 (leg endarterectomy), 39.25 (aorto-iliac-femoral bypass),
and 39.29 (leg bypass surgery) were considered to be hospitalized PAD.
Measurement of Baseline Risk Factors
After informed consent, the ARIC participants underwent a stan-
dardized medical history and examination that included interviews
and a fasting venipuncture. Participants were classified as never,
former, or current alcohol drinkers. Pack-years of smoking were calcu-
lated by multiplying the average number of cigarettes per day by the
number of years smoked and dividing by 20. Physical activity was
assessed using the Baecke sports questionnaire, with scores ranging
from 1 (low) to 5 (high), and participants were categorized as low (⬍2)
moderate (2 to 4), or high (ⱖ4) (18). Participants were asked to bring all
current medications to each ARIC study visit. Medication types were
recorded, including cholesterol-lowering medications,
␤
blockers, an-
giotensin-converting enzyme inhibitors, or other antihypertensive
medications. Anthropometrics, including weight and height, were ob-
tained while the participant was wearing a scrub suit. Body mass index
(BMI) was calculated as weight in kilograms divided by the square of
height in meters.
Fasting blood samples were drawn from an antecubital vein for
measurement of hemostatic factors, lipids, glucose, and chemistries.
Laboratory assays were performed in standardized research laborato-
ries except for white blood cell count, which was measured in local
laboratories. The ARIC Central Lipid Laboratory’s methods for triglyc-
erides, HDL cholesterol, and calculated LDL cholesterol have been
reported (19). The ARIC Central Hemostasis Laboratory measured
fibrinogen, factor VII, and factor VIII using published methods (20).
Prevalent CHD was defined as a self-reported history of physician-
diagnosed myocardial infarction (MI), coronary artery bypass surgery,
coronary angioplasty, or a previous MI by electrocardiogram. SBP and
diastolic BP (DBP) were measured three times using a random-zero
sphygmomanometer, and the average of the last two measurements
was used for analysis. Prevalent hypertension was defined as seated
DBP ⱖ90 mmHg, SBP ⱖ140 mmHg, or use of antihypertensive medi-
cations within the past 2 wk. Prevalent diabetes was defined as a fasting
serum glucose level ⱖ7.0 mmol/L (126 mg/dl), nonfasting glucose
level ⱖ11.1 mmol/L (200 mg/L), participant report of a physician
diagnosis of diabetes, or current use of any diabetes medication.
Statistical Analyses
We analyzed incident PAD through 2002. For those who developed
PAD, we calculated the length of follow-up from the baseline exami-
nation to the time of first PAD diagnosis. The date of event was based
on the earliest date of visit 3 or 4 (when ABI was ⬍0.9), date when
intermittent claudication first was classified, or hospitalization dis-
charge date. We did not exclude participants with missing ABI mea-
surement at visit 3 or 4 or both, as long as they continued annual
telephone contacts to allow intermittent claudication and PAD-related
revascularization procedures to be ascertained. For participants with-
out a PAD event, follow-up ended on the date of death, date of last
known contact, or December 31, 2002, whichever came first.
We compared baseline CVD risk factors of the participants who did
and did not develop PAD during follow-up using ANOVA, adjusted
for age, gender, race, and ARIC field center. Similarly adjusted inci-
dence rates per 1000 person-years were estimated for the three catego-
ries of kidney function using Poisson regression. Kaplan-Meier curves
were created to compare the cumulative probability of remaining free
of PAD events for each category of CKD. With normal kidney function
as a reference group, proportional hazards regression was used to
calculate relative risks (RR) and 95% confidence intervals (CI) for
incident PAD, adjusting for age, gender, race, and ARIC field center.
Multiple models were constructed for further exploration of pathways
that mediate CKD and PAD. Model 2 was adjusted for traditional CVD
risk factors, including diabetes, LDL and HDL cholesterol, triglycer-
630 Journal of the American Society of Nephrology J Am Soc Nephrol 18: 629–636, 2007

ides, prevalent CHD, pack-years of cigarettes, BMI, physical activity,
use of alcohol, and use of cholesterol medication; model 3 was adjusted
for all covariates in model 2 plus SBP and DBP and use of antihyper-
tensive medication; and model 4 was adjusted for all covariates in
model 2 plus inflammatory and procoagulant markers (fibrinogen,
factor VII, factor VIII, and white blood cell count). To examine the
dosage-response relation between the risk for PAD and eGFR, we fitted
a restricted cubic spline regression model to the data. This approach has
advantages over the categorical analysis because it does not assume the
form of the exposure–disease relation and avoids the problem of plac-
ing eGFR levels with different degrees of PAD risk in the same category
(21,22). All statistical analyses were conducted using SAS software
version 8.2 (SAS Institute, Cary, NC).
Results
Among the 14,280 participants (mean age 54 yr), the mean
visit 1 eGFR was 93.1 ml/min per 1.73 m
2
(SD 20.5). Among
those with CKD, 366 (97.3%) had eGFR between 30 and 59
ml/min per 1.73 m
2
, and 10 (2.7%) participants had eGFR
between 15 and 29 ml/min per 1.73 m
2
. The mean eGFR among
participants who met National Kidney Foundation criteria for
CKD was 52.6 ml/min per 1.73 m
2
(SD 7.7).
During a mean follow-up time of 13.1 yr (186,616 person-
years), 1016 participants developed PAD. Of these, 576 were
detected only by low ABI at visit 3 or 4 (271 [47.0%], 290
[50.3%], and 15 [2.6%] had normal kidney function, mildly
decreased kidney function, and CKD at baseline, respectively).
Of those with clinically recognized disease, 308 had Rose Ques-
tionnaire intermittent claudication (130, 163, and 15 had normal
kidney function, mildly decreased kidney function, and CKD at
baseline, respectively), and 242 had PAD-related leg amputa-
tions or revascularization procedures (96, 124, and 22 had nor-
mal kidney function, mildly decreased kidney function, and
CKD at baseline, respectively). Compared with participants
without incident PAD, those who developed PAD were more
likely to be older and have higher prevalences of diabetes,
cholesterol medication use, and CHD as well as greater mean
pack-years of cigarette use, alcohol intake, LDL cholesterol,
triglycerides, fibrinogen, and BMI and lower mean values of
HDL cholesterol, ABI, and physical activity score (Table 1). This
worse risk factor profile also was present for all subgroups of
incident PAD (ABI ⬍0.9, intermittent claudication, or PAD-
related revascularization procedures) compared with PAD-free
counterparts (data not shown).
The age-, gender-, race-, and ARIC field center–adjusted PAD
incidence rates per 1000 person-years were 4.7, 4.9, and 8.6 for
the normal kidney function, mildly decreased kidney function,
and CKD groups, respectively (Table 2). Kaplan-Meier curves
confirmed a greater probability of remaining free of PAD in
participants with normal kidney function than in those with
CKD (Figure 1). The probability of remaining free of PAD was
not statistically significantly different between those with
mildly decreased kidney function and those with normal kid-
ney function (P⫽0.17).
Table 1. Adjusted baseline characteristics of study population by incident PAD status: The ARIC Study, 1987
a
Baseline Risk Factor
Incident PAD
P
Yes (n⫽1016) No (n⫽13,264)
Age (yr) 56 54 ⬍0.0001
Male (%) 45 45 0.73
White race (%) 76 74 0.20
Diabetes (%) 22.0 10.6 ⬍0.0001
Prevalent hypertension (%) 43.0 33.2 ⬍0.0001
Prevalent CHD (%) 10.7 4.2 ⬍0.0001
Pack-years of cigarette smoking among current
smokers
23.1 15.0 ⬍0.0001
Mean LDL cholesterol (mg/dl) 142 137 ⬍0.0001
Mean HDL cholesterol (mg/dl) 48 52 ⬍0.0001
SBP (mmHg) 125 121 ⬍0.0001
DBP (mmHg) 74 74 0.92
Mean triglycerides (mg/dl) 146 129 ⬍0.0001
Fibrinogen (mg/dl) 321 300 ⬍0.0001
BMI (kg/m
2
)28.4 27.6 ⬍0.0001
Mean ankle-brachial index 1.10 1.15 ⬍0.001
Mean eGFR (ml/min per 1.73 m
2
)92 93 0.17
Sport index score (out of 5) 2.37 2.45 0.0023
Current alcohol drinkers (%) 35.1 39.1 0.007
Use of cholesterol medication (%) 4.8 2.6 ⬍0.0001
Use of anti-hypertensive medication (%) 39 29 ⬍0.0001
a
Adjusted for age, gender, race, and Atherosclerosis Risk in Communities (ARIC) field center. BMI, body mass index; CHD,
coronary heart disease; DBP, diastolic BP; eGFR, estimated GFR; PAD, peripheral arterial disease; SBP, systolic BP.
J Am Soc Nephrol 18: 629–636, 2007 PAD Risk in Chronic Kidney Disease 631

Compared with participants with normal kidney function,
the age-, gender-, race-, and ARIC field center–adjusted RR for
PAD was 1.04 (95% CI 0.91 to 1.18) for those with mildly
decreased kidney function and 1.82 (95% CI 1.34 to 2.47) for
those with CKD. After additionally adjusting for CVD risk
factors (model 2), an increase in risk for incident PAD still was
observed, with multivariable adjusted RR of 1.07 (95% CI 0.94
to 1.23) for those with mildly decreased kidney function and
1.56 (95% CI 1.13 to 2.14) for those with CKD. The multivariable
adjusted RR (model 2) for incident PAD is presented in Table 3.
Adjusting for covariates in models 3 and 4 did not appreciably
alter the RR for PAD in relation to kidney function. Results
from the categorical analyses largely were supported by a
restricted cubic spline regression, which demonstrated that the
multivariable adjusted log hazard of PAD started to increase
linearly at eGFR ⬍75 ml/min per 1.73 m
2
(Figure 2).
Formal testing for effect modification showed no statistically
significant interaction between any of the variables in Table 1
and the level of kidney function in relation to incident PAD,
except for gender (P⫽0.02). For men, the multivariable ad-
justed RR for PAD were 1.00 (95% CI 0.81 to 1.24) for those with
mildly decreased kidney function and 2.41 (95% CI 1.53 to 3.80)
for those with CKD. For women, the multivariable adjusted RR
for PAD were 1.15 (95% CI 0.96 to 1.38) for those with mildly
decreased kidney function and 1.19 (95% CI 0.76 to 1.86) for
those with CKD.
Table 2. Rates and relative risks (95% confidence intervals) of incident PAD events by level of eGFR: The ARIC
Study, 1987 to 2002
Parameter eGFR ⱖ90
(ml/min per 1.73 m
2
)
(n⫽6825)
eGFR 60 to 89
(ml/min per 1.73 m
2
)
(n⫽7079)
eGFR 15 to 59
(ml/min per 1.73 m
2
)
(n⫽376)
No. of PAD cases 453 516 47
Incidence rate per 1000 person-years 4.7 4.9 8.6
Model 1
a
1.0 1.04 (0.91 to 1.18) 1.82 (1.34 to 2.47)
Model 2
b
1.0 1.07 (0.94 to 1.23) 1.56 (1.13 to 2.14)
Model 3
c
1.0 1.08 (0.94 to 1.24) 1.54 (1.19 to 2.12)
Model 4
d
1.0 1.10 (0.96 to 1.26) 1.58 (1.14 to 2.17)
a
Adjusted for age, gender, race, and ARIC field center.
b
Additionally adjusted for diabetes, LDL and HDL cholesterol, triglycerides, prevalent coronary heart disease, pack-years of
cigarette smoking, BMI, physical activity, use of alcohol, and use of cholesterol medication.
c
Model 2 plus SBP, DBP, and use of antihypertensive medication.
d
Model 2 plus fibrinogen, factor VII, factor VIII, and white blood cell count.
Figure 1. Kaplan-Meier curves for remaining free of peripheral arterial disease (PAD) by level of estimated GFR (eGFR): The
Atherosclerosis Risk in Communities (ARIC) Study, 1987 to 2002.
632 Journal of the American Society of Nephrology J Am Soc Nephrol 18: 629–636, 2007

Supplemental Analysis
We conducted two supplemental analyses. First, because
participants with baseline ABI close to 0.9 might have a greater
burden of atherosclerosis, we performed an analysis with ad-
ditional adjustment for baseline ABI. Adjusted also for the same
covariates as listed in the model 2, the RR for PAD was 1.04
(95% CI 0.83 to 1.89) for participants with mildly decreased
kidney function and 1.46 (95% CI 0.98 to 2.19) for those with
CKD. Second, we also separately examined the associations of
CKD with asymptomatic PAD (low ABI ⬍0.9) and symptom-
atic PAD (intermittent claudication or hospital discharge of
PAD-related leg amputation or leg revascularization proce-
dures). Mildly decreased kidney function and CKD were not
statistically significantly associated with asymptomatic PAD by
low ABI. For symptomatic PAD, having CKD but not mildly
decreased kidney function increased the risk for PAD, with a
multivariable adjusted RR of 2.28 (95% CI 1.53 to 3.38).
Discussion
We report that individuals with CKD, defined by eGFR be-
tween 15 and 59 ml/min per 1.73 m
2
, also are at increased risk
for developing PAD. Compared with individuals with normal
kidney function, the demographic adjusted rate was approxi-
mately 80% higher in those with CKD. After adjustment for
CVD risk factors, individuals with CKD had a 1.5-fold higher
risk for developing PAD than those with normal kidney func-
tion.
On the basis of this RR and 4.5% of the US population with
CKD (eGFR 15 to 59 ml/min per 1.73 m
2
), the population
attributable risk for PAD was estimated to be 2.2%. This means
that 2.2% of the incidence of PAD in the general population
might be attributed to CKD.
The association of CKD with PAD has been examined mostly
in cross-sectional studies (6,7,23–25). For example, in NHANES
III, individuals with eGFR ⬍60 mg/ml per 1.73 m
2
were more
than twice as likely to have prevalent PAD, defined by ABI
⬍0.9, compared with those with eGFR ⱖ60 mg/ml per 1.73 m
2
(25). This magnitude of association was comparable to that of
hypertension, hypercholesterolemia, and a self-reported his-
tory of CHD. Nevertheless, prospective studies are needed to
understand the temporal relation between CKD and PAD. We
are aware of only one other study that specifically evaluated
this association longitudinally (9). Using either revasculariza-
tion procedures or lower extremity amputation as a composite
outcome for PAD, the Heart and Estrogen/Progestin Replace-
ment Study (HERS) reported that individuals with eGFR be-
tween 30 and 59 ml/min per 1.73 m
2
and individuals with
eGFR ⬍30 ml/min per 1.73 m
2
had an increased risk for de-
veloping a lower extremity PAD event compared with those
with eGFR ⱖ60 ml/min per 1.73 m
2
, with multivariable RR of
1.63 (95% CI 1.04 to 2.54) and 3.24 (95% CI 1.20 to 8.78),
respectively. That study derived its cohort from postmeno-
pausal women with documented CHD and did not exclude
prevalent PAD at baseline, thus limiting its generalizability and
raising questions regarding temporality and reverse causality.
Conversely, our cohort was composed of a large, diverse, com-
munity-based population of men and women. The HERS and
Table 3. Multivariable adjusted relative risks for
incident PAD: The ARIC Study, 1987 to 2002
a
Variables Predictors of
Incident PAD
(RR 关95% CI兴)
Age 1.05 (1.04 to 1.07)
Gender
female 1.00
male 0.67 (0.56 to 0.80)
Race
black 1.00
white 0.55 (0.35 to 0.88)
ARIC field center
Minneapolis 1.00
Washington 2.18 (1.75 to 2.72)
Forsyth 0.99 (0.76 to 1.29)
Jackson 0.88 (0.53 to 1.45)
Diabetes
no 1.00
yes 1.85 (1.50 to 2.29)
LDL cholesterol
⬍100 mg/dl 1.00
100 to 130 mg/dl 1.42 (1.05 to 1.90)
131 to 160 mg/dl 1.46 (1.09 to 1.96)
⬎160 mg/dl 1.55 (1.15 to 2.08)
HDL cholesterol
⬍40 mg/dl 1.00
40 to 60 mg/dl 0.78 (0.65 to 0.94)
⬎60 mg/dl 0.49 (0.36 to 0.66)
Triglycerides
⬍78 mg/dl 1.00
79 to 109 mg/dl 0.93 (0.72 to 1.21)
110 to 156 mg/dl 0.99 (0.77 to 1.28)
⬎157 mg/dl 0.88 (0.67 to 1.15)
Prevalent coronary heart disease
no 1.00
yes 2.25 (1.78 to 2.85)
Pack-years of cigarette smoking
⬍3.9 1.00
4.0 to 5.4 0.46 (0.36 to 0.59)
5.5 to 7.8 0.46 (0.36 to 0.58)
⬎7.8 0.88 (0.67 to 1.00)
BMI
⬍25 1.00
25 to 30 0.97 (0.80 to 1.18)
⬎30 0.97 (0.77 to 1.21)
Physical activity
high 1.00
moderate 1.10 (0.88 to 1.37)
low 1.23 (1.03 to 1.48)
Use of alcohol
no 1.00
yes 0.90 (0.76 to 1.06)
Use of cholesterol medication
no 1.00
yes 1.15 (0.79 to 1.69)
a
CI, confidence interval; RR, relative risk.
J Am Soc Nephrol 18: 629–636, 2007 PAD Risk in Chronic Kidney Disease 633

ARIC findings together support the conclusion of previous
cross-sectional studies that CKD should be considered a risk
marker for PAD.
Possible pathways that link CKD and PAD are unclear. Tra-
ditional CVD risk factors, which are highly prevalent in pa-
tients with CKD, could contribute etiologically to the develop-
ment of PAD. In our study, the strength of association was
attenuated only modestly after adjustment for traditional CVD
risk factors, suggesting that these risk factors may explain only
partly an increased risk for PAD in the CKD population. After
additional adjustment for SBP and DBP and use of antihyper-
tensive medication (model 3) and inflammatory and procoagu-
lant markers (model 4), the RR for kidney dysfunction essen-
tially were unchanged compared with those in model 2,
suggesting that these covariates may not be in a causal pathway
that mediates CKD and PAD. Alternatively, the effect of the
covariates in models 3 and 4 may be accounted for fully by
these traditional CVD risk factors, which are highly prevalent
in the CKD population. Other potential pathways that are
unique to CKD but that are less well investigated include
hyperhomocysteinemia and abnormal calcium and phosphate
metabolism. For example, prospective studies have demon-
strated that an elevated level of homocysteine is a risk factor for
a composite CVD outcome (including PAD) among dialysis-
dependent patients with ESRD (26,27). Using a similar study
population, Goldsmith et al. (28) reported that higher levels of
serum phosphate and vitamin D concentrations both were sta-
tistically significantly associated with the severity and the rapid
progression of vascular calcification, which was hypothesized
to lead to arterial stiffness and impaired tissue perfusion.
The findings of our study have clinical implications. Because
both CKD and PAD share a number of common risk factors,
including hypertension, diabetes, and smoking, it is probable
that modification of these risk factors to slow progression of
CKD also might beneficially decrease PAD incidence. More
important, the findings of our study call for an increased
awareness and early detection of PAD in the CKD population.
Recognition of an increased risk for PAD in this population,
particularly in individuals with no typical ischemic symptoms,
potentially could avert adverse limb as well as CVD events if
modification of risk factors were intensified. To identify indi-
viduals who are at high risk for PAD, the American Diabetes
Association recently recommended that a screening ABI be
performed in individuals who have diabetes and are older than
50 yr and in individuals who are diabetes and are younger than
50 yr and have other PAD risk factors. Our findings similarly
highlight and support development of a PAD screening strat-
egy to identify patients who have CKD and are at high risk for
PAD.
This study has several limitations. First, the MDRD formula
that was used to estimate GFR was based on individuals with
CKD and did not include healthy individuals. Using the MDRD
formula in healthy individuals has been shown to underesti-
mate systemically GFR by much as 29% (29). A direct measure-
ment from iothalamate or creatinine clearance using a 24-h
urine collection would yield more accurate estimation of renal
function. However, direct measurement of GFR is not feasible
in a large epidemiologic study. Second, a random sample of
participants was invited for the ABI measurements at visits 3
and 4, and these measurements were performed on one poste-
rior tibial artery of a randomly selected leg. This means that we
would have missed some participants with asymptomatic PAD.
Figure 2. Spline regression of the log hazard of PAD on eGFR: The ARIC Study, adjusted for age, gender, race, ARIC field center,
LDL and HDL cholesterol, triglycerides, prevalent coronary heart disease, number of pack-years of cigarette smoking, body mass
index, physical activity, use of alcohol, and use of cholesterol medication.
634 Journal of the American Society of Nephrology J Am Soc Nephrol 18: 629–636, 2007

Furthermore, we would have identified all participants with
bilateral PAD and potentially misclassified half of participants
with unilateral PAD as non-PAD cases. Hence, the true inci-
dence rate of PAD by low ABI and the magnitude of association
with CKD in our study are likely to be underestimated. In
addition, this misclassification as a result of ABI measurement
of a single leg and inherent measurement error of ABI may
explain the null association between CKD and asymptomatic
PAD. Third, a few patients who had had amputation and were
counted as having PAD may have had amputation for infec-
tion. However, the total number of patients with amputations
was only 16, so the impact of any misclassification would have
been small. Last, C-reactive protein and D-dimer may be im-
portant predictors of PAD incidence in the CKD population
and unfortunately were not available on the full cohort of the
ARIC study cohort.
Conclusion
We found that individuals with CKD are at moderately in-
creased risk for developing PAD. Because early diagnosis of
PAD is critical to prevent lower extremity revascularization
procedures and amputations, further studies should evaluate
whether PAD screening strategies in patients with CKD, par-
ticularly those with eGFR ⬍60 ml/min per 1.73 m
2
, are effec-
tive.
Acknowledgments
The ARIC study was supported by National Heart, Lung, and Blood
Institute contracts N01-HC-55015, N01-HC-55016, N01-HC-55018, N01-
HC-55019, N01-HC-55020, N01-HC-55021, and N01-HC-55022. K.W.
was supported by National Heart, Lung, and Blood Institute training
grant T32-HL-07779, and E.S. was supported by T32-HL-07024.
We thank the participants and staff of the ARIC study for important
contributions.
Disclosures
None.
References
1. US Renal Data System: USRDS 2000 Annual Data Report,
Bethesda, National Institutes of Health, National Institute of
Diabetes and Digestive and Kidney Diseases, 2000
2. Manjunath G, Tighiouart H, Coresh J, Macleod B, Salem
DN, Griffith JL, Levey AS, Sarnak MJ: Level of kidney
function as a risk factor for cardiovascular outcomes in the
elderly. Kidney Int 63: 1121–1129, 2003
3. Manjunath G, Tighiouart H, Ibrahim H, MacLeod B, Salem
DN, Griffith JL, Coresh J, Levey AS, Sarnak MJ: Level of
kidney function as a risk factor for atherosclerotic cardio-
vascular outcomes in the community. J Am Coll Cardiol 41:
47–55, 2003
4. Muntner P, He J, Hamm L, Loria C, Whelton PK: Renal
insufficiency and subsequent death resulting from cardio-
vascular disease in the United States. J Am Soc Nephrol 13:
745–753, 2002
5. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY:
Chronic kidney disease and the risks of death, cardiovas-
cular events, and hospitalization. N Engl J Med 351: 1296–
1305, 2004
6. O’Hare AM, Glidden DV, Fox CS, Hsu CY: High preva-
lence of peripheral arterial disease in persons with renal
insufficiency: Results from the National Health and Nutri-
tion Examination Survey 1999–2000. Circulation 109: 320–
323, 2004
7. de Vinuesa SG, Ortega M, Martinez P, Goicoechea M,
Campdera FG, Luno J: Subclinical peripheral arterial dis-
ease in patients with chronic kidney disease: Prevalence
and related risk factors. Kidney Int Suppl 67: S44–S47, 2005
8. Fried LF, Shlipak MG, Crump C, Bleyer AJ, Gottdiener JS,
Kronmal RA, Kuller LH, Newman AB: Renal insufficiency
as a predictor of cardiovascular outcomes and mortality in
elderly individuals. J Am Coll Cardiol 41: 1364–1372, 2003
9. O’Hare AM, Vittinghoff E, Hsia J, Shlipak MG: Renal in-
sufficiency and the risk of lower extremity peripheral ar-
terial disease: Results from the heart and estrogen/proges-
tin replacement study (HERS). J Am Soc Nephrol 15: 1046–
1051, 2004
10. The Atherosclerosis Risk in Communities (ARIC) Study:
Design and objectives. The ARIC Investigators. Am J Epi-
demiol 129: 687–702, 1989
11. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D:
A more accurate method to estimate glomerular filtration
rate from serum creatinine: A new prediction equation.
Modification of Diet in Renal Disease Study Group. Ann
Intern Med 130: 461–470, 1999
12. Coresh J, Astor BC, McQuillan G, Kusek J, Greene T, Van
Lente F, Levey AS: Calibration and random variation of the
serum creatinine assay as critical elements of using equa-
tions to estimate glomerular filtration rate. Am J Kidney Dis
39: 920–929, 2002
13. Rose GA, Blackburn H, Gillum RF, Prineas RJ: Cardiovas-
cular Survey Methods, Geneva, World Health Organization,
1982
14. ARIC: Atherosclerosis Risk in Communities Study: Manual 11:
Sitting Blood Pressure and Postural Changes in Blood Pressure
and Heart Rate, Chapel Hill, NC, National Heart, Lung, and
Blood Institute, 1987
15. Mundt KA, Chambless LE, Burnham CB, Heiss G: Measur-
ing ankle systolic blood pressure: Validation of the Di-
namap 1846 SX. Angiology 43: 555–566, 1992
16. ARIC: Atherosclerosis Risk in Communities Study: Manual 6A:
Ultrasound Assessment, Chapel Hill, National Heart, Lung,
and Blood Institute, 1987
17. Dormandy J, Heeck L, Vig S: Intermittent claudication: A
condition with underrated risks. Semin Vasc Surg 12: 96–
108, 1999
18. Baecke JA, Burema J, Frijters JE: A short questionnaire for
the measurement of habitual physical activity in epidemi-
ological studies. Am J Clin Nutr 36: 936–942, 1982
19. Chambless LE, Heiss G, Folsom AR, Rosamond W, Szklo
M, Sharrett AR, Clegg LX: Association of coronary heart
disease incidence with carotid arterial wall thickness and
major risk factors: The Atherosclerosis Risk in Communi-
ties (ARIC) Study, 1987–1993. Am J Epidemiol 146: 483–494,
1997
20. Folsom AR, Wu KK, Rosamond WD, Sharrett AR, Chamb-
less LE: Prospective study of hemostatic factors and inci-
dence of coronary heart disease: The Atherosclerosis Risk
J Am Soc Nephrol 18: 629–636, 2007 PAD Risk in Chronic Kidney Disease 635

in Communities (ARIC) Study. Circulation 96: 1102–1108,
1997
21. Witte JS, Greenland S: A nested approach to evaluating
dose-response and trend. Ann Epidemiol 7: 188–193, 1997
22. Greenland S: Dose-response and trend analysis in epide-
miology: Alternatives to categorical analysis. Epidemiology
6: 356–365, 1995
23. Shlipak MG, Fried LF, Crump C, Bleyer AJ, Manolio TA,
Tracy RP, Furberg CD, Psaty BM: Cardiovascular disease
risk status in elderly persons with renal insufficiency. Kid-
ney Int 62: 997–1004, 2002
24. Leskinen Y, Salenius JP, Lehtimaki T, Huhtala H, Saha H: The
prevalence of peripheral arterial disease and medial arterial cal-
cification in patients with chronic renal failure: Requirements for
diagnostics. Am J Kidney Dis 40: 472–479, 2002
25. Selvin E, Erlinger TP: Prevalence of and risk factors for
peripheral arterial disease in the United States: Results
from the National Health and Nutrition Examination Sur-
vey, 1999–2000. Circulation 110: 738–743, 2004
26. Moustapha A, Naso A, Nahlawi M, Gupta A, Arheart KL,
Jacobsen DW, Robinson K, Dennis VW: Prospective study
of hyperhomocysteinemia as an adverse cardiovascular
risk factor in end-stage renal disease. Circulation 97: 138–
141, 1998
27. Bostom AG, Shemin D, Verhoef P, Nadeau MR, Jacques
PF, Selhub J, Dworkin L, Rosenberg IH: Elevated fasting
total plasma homocysteine levels and cardiovascular
disease outcomes in maintenance dialysis patients. A
prospective study. Arterioscler Thromb Vasc Biol 17: 2554–
2558, 1997
28. Goldsmith DJ, Covic A, Sambrook PA, Ackrill P: Vascu-
lar calcification in long-term haemodialysis patients in a
single unit: A retrospective analysis. Nephron 77: 37–43,
1997
29. Rule AD, Larson TS, Bergstralh EJ, Slezak JM, Jacobsen SJ,
Cosio FG: Using serum creatinine to estimate glomerular
filtration rate: Accuracy in good health and in chronic
kidney disease. Ann Intern Med 141: 929–937, 2004
Access to UpToDate on-line is available for additional clinical information
at http://www.jasn.org/
636 Journal of the American Society of Nephrology J Am Soc Nephrol 18: 629–636, 2007