Low Ankle-Brachial Index Associated With Rise in Creatinine Level Over Time: Results From the Atherosclerosis Risk in Communities Study

Background  A low ankle-brachial index (ABI) predicts risk of cardiovascular death, myocardial infarction, peripheral arterial disease events, and stroke. However, it is unknown whether a low ABI also predicts a decline in renal function.

Methods  We examined the association between ABI and change in serum creatinine level over time among 13 655 participants in the Atherosclerosis Risk in Communities (ARIC) study who underwent serum creatinine and ABI measurement at baseline and also underwent serum creatinine measurement 3 years later at the second study visit. The study outcome was a 50% rise in serum creatinine level from baseline to the second study visit.

Results  Overall, 0.48% of participants with an ABI of 1 or higher, 0.9% of participants with an ABI between 0.9 and 0.99, and 2.16% of participants with an ABI lower than 0.9 experienced a 50% or greater increase in serum creatinine level. In multivariate analysis, participants with an ABI lower than 0.9 were still more than twice as likely as those in the referent category (ABI ≥1) to experience an increase in serum creatinine level (odds ratio 2.5; 95% confidence interval, 1.1-5.7) (P = .04), and a linear trend in the incidence of worsening renal function was noted across ABI categories (P = .02). Analyses excluding participants with renal insufficiency, diabetes, and hypertension at baseline all produced similar results.

Conclusion  In addition to known associations of the ABI with stroke, myocardial infarction, peripheral arterial disease events, and cardiovascular death, a low ABI also predicts an increase in serum creatinine level over time.

The prevalence and incidence of cardiovascular disease are both high among those with renal insufficiency,^1-10 and chronic kidney disease and cardiovascular disease share many of the same risk factors.^11-22 However, the contribution of preexisting cardiovascular disease to progression of renal disease has not been examined in detail.

We hypothesized that a low ankle-brachial index (ABI)—a marker for both lower extremity and more generalized atherosclerotic disease²³ and a known predictor of cardiovascular death, myocardial infarction, peripheral arterial disease events, and stroke^24-28—would also be associated with a decline in renal function over time. We tested this hypothesis using publicly available data from the first and second study visits of the Atherosclerosis Risk in Communities (ARIC) study.²⁹

The ARIC study²⁹ was a prospective cohort study of cardiovascular disease among a community sample of middle-aged residents of Jackson, Miss; Forsyth County, North Carolina; Washington County, Maryland; and several suburbs of Minneapolis, Minn. The study consisted of 15 792 men and women aged 45 to 64 years at baseline. Participant recruitment and baseline data collection were completed between 1987 and 1989, and participants subsequently returned for 3 visits, each spaced 3 years apart. The present analysis uses publicly available data from the first and second study visits and was approved by the institutional review board at the University of California, San Francisco.

The primary predictor variable for the present analysis was ABI at the first study visit. The ABI is the ratio of systolic blood pressure at the ankle to systolic blood pressure at the brachial artery. Ankle and brachial systolic blood pressure measurements were ascertained using the Dinamap 1846 SX (Critixon, Tampa, Fla). Ankle blood pressure was measured at the posterior tibial artery in one leg. Two measurements were taken 5 to 8 minutes apart while the participant was in the prone position. Brachial artery systolic blood pressure measurements were taken 5 minutes apart with the participant in the supine position. The ABI was computed by dividing the average of the 2 ankle systolic blood pressure measurements by the average of the first 2 brachial readings. For the purposes of this analysis, we divided ABI measurements into 3 categories at the clinically established cutoffs of 0.9 and 1.0.³⁰

The outcome measure for the present analysis was the occurrence of at least a 50% rise in serum creatinine level between baseline and the first follow-up visit approximately 3 years later. Too few participants experienced a doubling of serum creatinine level (a more conventional measure of renal functional decline) during follow-up for this to be used as an outcome. Regardless of initial creatinine level, a 50% increase in serum creatinine level translates into an approximately 25% decrease in creatinine clearance. This outcome is more easily interpreted than either a fixed increase in serum creatinine level or a fixed decrease in estimated creatinine clearance or glomerular filtration rate. The correlation between fixed increases in serum creatinine level and creatinine clearance is highly variable, depending on the initial creatinine level, and since values can be very inflated in patients with normal renal function, fixed decrements over time in calculated creatinine clearance or glomerular filtration rate may be misleading.

Our analysis was adjusted for the following covariates ascertained at visit 1: demographic characteristics (age, race [black vs nonblack], and sex), baseline Cockcroft-Gault–estimated creatinine clearance,³¹ comorbid conditions including diabetes (previously diagnosed or fasting glucose level ≥126 mg/dL [≥6.99 mmol/L]), smoking behavior (participants were classified as current smokers, former smokers, or nonsmokers), body mass index, fasting lipid levels (low-density lipoprotein [LDL], high-density lipoprotein [HDL], and triglyceride levels), alcohol use, hematocrit values, mean systolic and diastolic blood pressures, and educational level (college vs no college). To account for the slight differences in follow-up time from visit 1 to visit 2 between study participants, exact follow-up time was also included in the multivariable analysis.

All statistical analyses were performed using STATA statistical software (College Station, Tex). We compared baseline characteristics (ascertained at visit 1) across ABI categories. Normally distributed continuous variables were compared using the t test; nonnormally distributed variables (eg, triglyceride levels) were compared using the Mann-Whitney U test; and categorical variables were compared using the χ² test. For all analyses, participants with an ABI of 1 or higher served as the referent category with which the other groups were compared. The association of ABI with worsening renal function was measured using bivariate and multivariate logistic regression analysis. We checked the linearity assumption for numeric predictors by including log-transformed and quadratic terms and breaking the predictor into quartiles. To satisfy the linearity assumption, triglyceride levels, systolic blood pressure, and hematocrit values were log transformed, and estimated creatinine clearance was modeled by quartile. We tested for the presence of a linear trend in the odds of a 50% increase in creatinine level across ABI categories by incorporating the categorical ABI variable into the logistic regression analysis as a numeric variable (scored 0-1-2).

We confirmed the presence of an association between low ABI and the study outcome after excluding, one group at a time, participants with an estimated creatinine clearance lower than 60 mL/min, diabetes, and hypertension (defined as either a mean systolic blood pressure of ≥140 mm Hg, a mean diastolic blood pressure of ≥90 mm Hg, or use of antihypertensive medication). To evaluate the impact of possible measurement error for serum creatinine on our results, we performed a companion analysis in which we defined the outcome as a 50% rise in serum creatinine level with an absolute rise in creatinine level of at least 0.4 mg/dL (35.4 μmol/L).¹⁴ This analysis was designed to ignore small increases in serum creatinine level occurring among cohort patients with low baseline serum creatinine levels that might be due solely to measurement error.

A total of 13 655 ARIC participants²⁹ underwent creatinine and ABI measurement at visit 1 and creatinine measurement again at visit 2. Among these, 12 179 participants (89%) had an ABI of 1 or higher; 1105 (8%) had an ABI between 0.9 and 0.99; and 371 (3%) had an ABI lower than 0.9. Compared with the referent category of participants with an ABI of 1 or higher, the mean serum creatinine level was similar, but mean creatinine clearance was slightly lower among those with an ABI lower than 0.9 (Table 1). The relationship between baseline creatinine clearance and ABI was nonlinear: a disproportionate number of participants in the lowest and highest quartiles of creatinine clearance had low ABIs (Figure).

Compared with participants with an ABI of 1 or higher, those with an ABI lower than 0.9 were older; included a higher percentage of women, blacks, and persons with diabetes; had lower HDL and higher LDL and triglyceride levels; had a slightly lower mean BMI; included a smaller percentage of alcohol users and college graduates and a higher percentage of current smokers; had a higher mean systolic blood pressure and lower mean diastolic blood pressure; and included a higher percentage of participants using antihypertensive medications (Table 1). Overall, serum creatinine level increased in 0.56% of participants (n = 76): 0.48% of those with an ABI of 1 or higher (n = 58), 0.9% of those with an ABI between 0.9 and 0.99 (n = 10), and 2.16% of those with an ABI lower than 0.9 (n = 8). Baseline serum creatinine levels were not significantly different among patients who did and did not experience a 50% rise in serum creatinine level (median creatinine level, 1 mg/dL [88.4 μmol/L] with 25th-75th percentile range, 0.8-1.4 mg/dL [123.8 μmol/L] vs 1.1 mg/dL [97.2 μmol/L] with 25th-75th percentile range, 1.1-1.2 mg/dL [97.2-106.1 μmol/L]) (P = .36). In both bivariate and multivariate analyses, an ABI lower than 0.9 was associated with an increase in serum creatinine level during follow-up, and there was a linear trend across ABI categories (Table 2).

Results were similar after exclusion, one group at a time, of those with an estimated clearance lower than 60 mL/min (odds ratio [OR], 2.6; 95% confidence interval [CI], 1.04-6.61) (P = .04), diabetes (OR, 3.5; 95% CI, 1.2-10.5) (P = .03), and hypertension or use of antihypertensive medications (OR, 5.5; 95% CI, 1.4-21.4) (P = .02) at baseline. When we defined the outcome as a 50% rise in serum creatinine level with at least a 0.4 mg/dL (35.4 μmol/L) absolute increase in serum creatinine level (n = 66), the results did not differ substantially from the primary analysis: patients with an ABI lower than 0.9 still had a more than 2-fold adjusted risk of this outcome compared with the referent category (OR, 2.3; 95% CI, 0.92-5.8) (P = .07), and there was a trend across ABI categories (P = .03).

The present study demonstrates that, in addition to its previously demonstrated associations with stroke, myocardial infarction, peripheral arterial disease events, and cardiovascular death,^24-28 a low ABI is also associated with an increase in serum creatinine level over time. Participants with an ABI lower than 0.9 had more than 4-fold odds of experiencing a 50% rise in creatinine level compared with those with an ABI of 1 or higher, and the association persisted after adjustment for known predictors of renal functional decline. Because a low ABI is a specific measure of overall atherosclerotic burden,^23,32-34 the existence of an association between baseline ABI and subsequent increase in serum creatinine supports the notion that preexisting atherosclerosis may be a risk factor for deterioration of renal function over time. Furthermore, the strong association of ABI with a rise in serum creatinine level in both unadjusted and adjusted analysis highlights the possible importance of ABI measurement as a means to identify patients at greatest risk for deterioration of renal function over time.

Many studies have reported both cross-sectional and longitudinal associations between baseline renal insufficiency and coronary, cerebral, and peripheral arterial disease.^1-9,35 There is also extensive research exploring possible causal mechanisms for an association between renal insufficiency and cardiovascular disease. For example, hyperhomocysteinemia, oxidant stress, elevated levels of lipoprotein(a) and inflammatory markers, and disordered calcium phosphorous metabolism all represent potential pathways for accelerated atherosclerosis in patients with renal insufficiency.³⁶

There has been much less interest in the reverse association: the contribution of atherosclerosis to progression of renal disease. Using baseline data on screenees for the Multiple Risk Factor Intervention Trial (MRFIT), Klag et al¹² reported that a history of myocardial infarction was independently associated with an increased risk of end-stage renal disease, though this was not the primary focus of the analysis. Levin et al³⁷ found that among patients with existing chronic kidney disease, the presence of cardiovascular disease predicted time to onset of end-stage renal disease. Bleyer et al²² described a positive independent association between maximum carotid intimal thickness and subsequent rise in serum creatinine level of at least 0.3 mg/dL (26.5 μmol/L) over a 3- to 4-year period among subjects without diabetes enrolled in the Cardiovascular Health Study. Finally, a recent study by McClellan et al³⁸ demonstrated a high rate of progression to end-stage renal disease among Medicare beneficiaries hospitalized for congestive heart failure and for acute myocardial infarction.

Data from several autopsy series also lend some support to the notion that systemic atherosclerosis may predispose to renal insufficiency. Among persons free of clinical renal disease and renal artery stenosis, the prevalence of glomerulosclerosis was noted to be higher among those with moderate to severe intrarenal atherosclerosis than among those with mild atherosclerosis.³⁹ In a second autopsy series of persons aged 25 to 54 years, renal arteriolar hyalinization was correlated with raised lesions in the coronary arteries and aorta.⁴⁰

While the association reported herein between low ABI and rising creatinine level may reflect a shared risk factor profile between renal disease and atherosclerosis,⁴¹ we were not able to explain away this association by controlling for many of the known risk factors for renal disease. Because the kidney is a highly vascular organ whose function is dependent on an intact circulatory system, it makes intuitive sense that systemic atherosclerotic disease could be causally associated with declining renal function. Based on our findings and the aforementioned studies, we would argue that systemic atherosclerosis may be a risk factor for renal insufficiency as it is for myocardial infarction, stroke, and peripheral arterial disease events.

A limitation of the present study is that our definition of worsening renal function may be inaccurate owing to changes in patient age and muscle mass over time. While measured changes in creatinine clearance would probably be the best way to identify those with declining renal function, these data were not collected in ARIC.²⁹ Alternate measures of declining renal function such as fixed changes in serum creatinine level, calculated clearance, or glomerular filtration rate are less clearly interpretable than a percentage change in serum creatinine level. Finally, we were unable to use incidence of new onset of renal insufficiency (defined as incidence of a calculated creatinine clearance of <60 mL/min) among patients experiencing a 50% increase in serum creatinine level because of the rarity of this outcome (n = 41) over the short follow-up period. Thus, in the absence of measured creatinine clearance, we believe that our definition of worsening renal function as a percentage rise in serum creatinine level represents the most appropriate choice.

A second limitation is that our analysis did not include detailed information on the severity of important confounding comorbid conditions such as diabetes and hypertension. However, it is somewhat reassuring that even after excluding participants with diabetes or hypertension at baseline, a low ABI was still strongly associated with the study outcome.

Finally, declining renal function was a relatively uncommon outcome in this sample of middle-aged community-dwelling adults. Therefore, the clinical importance of ABI measurement as a predictor of declining renal function would be best assessed in populations where worsening renal function is a more common outcome.

In conclusion, a low ABI is predictive not just of recognized atherosclerotic events such as stroke, myocardial infarction, peripheral arterial disease events, and cardiovascular death, but also of rising creatinine level over time. These findings support the notion of systemic atherosclerosis as a risk factor for decline in renal function and also highlight the possible importance of ABI measurement as a means to identify patients at greatest risk for deterioration of renal function over time.

Correspondence: Ann M. O'Hare, MA, MD, Veterans Affairs Medical Center, Box 111J (Nephrology), 4150 Clement St, San Francisco, CA 94121 (Ann.O'Hare@med.va.gov).

Accepted for Publication: November 3, 2004.

Financial Disclosure: None.

Funding/Support: Dr O'Hare is supported by a Career Development Award from the Department of Veterans Affairs Health Services Research and Development Service, Washington, DC.