The Long-term Prognostic Value of the Resting and Postexercise Ankle-Brachial Index

Background  Peripheral arterial disease is associated with a high incidence of cardiovascular mortality. Peripheral arterial disease can be detected by using the ankle-brachial index (ABI). This study assessed the prognostic value of the postexercise ABI in addition to the resting ABI on long-term mortality in patients with suspected peripheral arterial disease.

Methods  In this prospective cohort study of 3209 patients (mean ± SD age, 63 ± 12 years; 71.1% male), resting and postexercise ABI values were measured and a reduction of postexercise ABI over baseline resting readings was calculated. The mean follow-up was 8 years (interquartile range, 4-11 years).

Results  During follow-up, 1321 patients (41.2%) died. After adjusting for clinical risk factors, lower resting ABI values (hazard ratio per 0.10 lower ABI, 1.08; 95% confidence interval [CI], 1.06-1.10), lower postexercise ABI values (hazard ratio per 0.10 lower ABI, 1.09; 95% CI, 1.08-1.11), and higher reductions of ABI values over baseline readings (hazard ratio per 10% lower ABI, 1.12; 95% CI, 1.09-1.14) were significantly associated with a higher incidence of mortality. In patients with a normal resting ABI (n =789), a reduction of the postexercise ABI by 6% to 24%, 25% to 55%, and greater than 55% was associated with a 1.6-fold (95% CI, 1.2-2.2), 3.5-fold (95% CI, 2.4-5.0), and 4.8-fold (95% CI, 2.5-9.1) increased risk of mortality, respectively.

Conclusions  Resting and postexercise ABI values are strong and independent predictors of mortality. A reduction of postexercise ABI over baseline readings can identify additional patients (who have normal ABI values at rest) at increased risk of subsequent mortality.

Peripheral arterial disease (PAD) is a manifestation of systemic atherosclerosis and is associated with a significantly increased risk of cardiovascular morbidity and mortality.^1-3 In the Netherlands, the combined prevalence of symptomatic and asymptomatic PAD in the population of those 55 years and older is 19%.⁴ In the United States, different prevalence values for PAD have been reported, ranging from 4% in patients 40 years and older to 29% in patients either older than 70 years or aged 50 to 59 years with a 10–pack year history of smoking or the presence of diabetes mellitus.^1,5-8

Management of PAD comprises walking exercise, modification of cardiovascular risk factors, and antiplatelet therapy.^2,9 Although the prevalence of PAD is high in industrialized countries, it is believed that PAD remains an underdiagnosed and undertreated disease in primary care.^3,6,8 Therefore, the identification of patients with suspected PAD who are at increased risk for cardiovascular events is necessary for disease control and appropriate application of treatment strategies.

The ankle-brachial index (ABI), a ratio of ankle systolic–brachial systolic blood pressure, is a simple, effective, and noninvasive test used for the assessment of lower extremity arterial obstruction and for the screening of patients with suspected PAD.⁸ Several studies^10-16 have demonstrated that low ABI values at rest predict cardiovascular and overall mortality. Most studies have used a cutoff of 0.90 to define a low ABI value. However, knowledge about the association of long-term mortality across the whole range of ABI values at rest and after exercise is limited.

The ABI is commonly measured at rest, but ABI measurements coupled with exercise testing may enhance the sensitivity of the test and may identify additional patients with PAD who have normal resting ABI values.^2,17 In this study, we assessed the association of long-term mortality across the whole range of resting and postexercise ABI values, and hypothesized that a reduction in postexercise ABI over baseline resting readings may identify patients at increased risk of long-term mortality.

The Erasmus Medical Centre serves a population of approximately 3 million people and acts as a tertiary referral center for approximately 30 affiliated hospitals. This cohort study prospectively included consecutive patients with suspected or known PAD referred to our university clinic of vascular surgery between January 14, 1983, and January 1, 2005, for the evaluation and management of their disease. Patients with suspected PAD had a typical history of intermittent claudication or other symptoms of chronic arterial insufficiency, including ulceration of the foot, hair loss, or reduced capillary refill. Patients with known PAD had a resting ABI of 0.90 or less. Excluded were patients who were unable to perform exercise and those who underwent previous vascular surgery. The hospital's Medical Ethical Committee approved the study protocol, and patients who fulfilled the inclusion criteria agreed to participate in the study. Based on hospital records and personal interviews at the visit, a medical history was recorded. Diabetes mellitus was recorded if patients presented with a fasting glucose level of 126 mg/dL or more (≥7.0 mmol/L) or in those who required treatment. Hypertension was recorded if patients presented with a blood pressure of 140/90 mm Hg or higher or if patients were medically treated for hypertension. Hypercholesterolemia was recorded if patients presented with a plasma cholesterol level of 212 mg/dL or more (≥5.5 mmol/L) or if patients were taking lipid-lowering agents. Renal dysfunction was recorded if patients presented with a serum creatinine level of 2.0 mg/dL or more (≥177 μmol/L) or in those who required dialysis. Cigarette smoking included only current smoking. Patients were assessed for cardiac medication use. A baseline 12-lead electrocardiogram was obtained, and the ABI at rest and after exercise was measured in each patient.

Trained technicians, using a Doppler ultrasonic instrument with an 8-MHz vascular probe (Imexdop CT+ Vascular Doppler; Miami Medical, Glen Allen, Va), measured systolic blood pressure readings in the right and left brachial arteries, right and left dorsalis pedis arteries, and right and left posterior tibial arteries. The ABI in the right and left legs was calculated by dividing the right and the left ankle pressure by the brachial pressure. The higher of the 2 brachial blood pressure readings was used if a discrepancy in systolic blood pressure was present. Again, the higher of the dorsalis pedis and posterior tibial artery pressures was used when a discrepancy in systolic blood pressure readings between the 2 arteries was measured. If no pressure in the dorsalis pedis artery was obtained because of an absent dorsalis pedis artery, the pressure in the posterior tibial artery was used. The ABI at rest was measured after the participants had been resting in the supine position for at least 10 minutes. Measurements were then repeated at both sides with the patient in the supine position, after 5 minutes of walking on a treadmill at 4.0 km/h. No inclining plane or graded inclines were used with treadmill testing, and the treadmill tests were performed without continuous electrocardiographic monitoring before, during, and after the testing. Of the ABI values obtained in each leg, the lower was used in all analyses. Interobserver and intraobserver agreement for resting ABI was 97% and 98%, respectively; and for postexercise ABI, 96% and 97%, respectively. Although an ABI of greater than 1.30 has been used as a marker of calcified atherosclerosis, we considered patients with values greater than 1.50 to have calcified atherosclerosis, resulting in high ABI readings. These patients were excluded from the study.

We included 3209 patients who were examined during a median follow-up of 8 years (interquartile range, 4-11 years). No patients were lost to follow-up. Follow-up ended at the date of the last visit or the date of death. Information about the patient's vital status was obtained at the Office of Civil Registry. For patients who died at our hospital during follow-up, hospital records and autopsy results were reviewed. For patients who died elsewhere, general practitioners were approached to ascertain the cause of death. A cardiac cause of death was defined as death caused by acute myocardial infarction (postmortem evidence of acute myocardial infarction or definite criteria for myocardial infarction within the 4 weeks before death), cardiac arrhythmias, congestive heart failure, or sudden death.

Continuous data are expressed as mean and SD or median (interquartile range), and compared using the t test or Mann-Whitney test when appropriate. Categorical data are presented as percentage frequencies, and differences between proportions were compared using the χ² test with Yates correction. The primary end points were overall mortality and cardiac death. We used univariate and multivariate Cox proportional hazards models to analyze the effect of clinical characteristics and ABI values on survival. Hazard ratios are given with 95% confidence intervals (CIs). In multivariate analyses, all clinical variables, including medication, were entered, irrespective of the significance level in univariate analysis. The Kaplan-Meier method with the log-rank test was used for comparing survival curves in 2 or more groups. The reduction of the postexercise ABI compared with the resting ABI was calculated and expressed as a percentage.

The prevalence of clinical risk factors, medication use, and mortality rate might have changed during the 22-year period in which the study was conducted. Therefore, we evaluated differences in baseline characteristics and mortality results between patients enrolled from January 14, 1983, to December 31, 1993, and those enrolled from January 1, 1994, to January 1, 2005. In multivariate analyses, the relation between ABI values (resting ABI, postexercise ABI, and reduction of postexercise ABI) and survival was evaluated separately for the periods from 1983 to 1993 and from 1994 to 2005, and an interaction term was evaluated to reveal possible heterogeneity between the different periods. In addition, tests for heterogeneity were used to evaluate the effect of a reduction of postexercise ABI in patients with different resting ABI values. For all tests, P<.05 (2-sided) was considered significant. All analyses were performed using a commercially available software program (SPSS-11.0 statistical software; SPSS Inc, Chicago, Ill).

The mean age of the patients was 63 ± 12 years, and 71.1% were male. The baseline characteristics of the 3209 patients are presented in Table 1. Patients included from 1994 to 2005 compared with those included from 1983 to 1993 more commonly presented with a history of myocardial infarction, percutaneous transluminal coronary angiography, and hypercholesterolemia, and less commonly smoked. Aspirin, angiotensin-converting enzyme inhibitors, β-blockers, digoxin, diuretics, nitrates, and statins were more frequently prescribed in patients who were included from 1994 to 2005, compared with patients included from 1983 to 1993.

The median ABI measured at rest was 0.65 (interquartile range, 0.50-0.90) (P = .80). A resting ABI of greater than 0.90 was measured in 789 patients (24.6%). The median ABI measured after exercise was 0.45 (interquartile range, 0.25-0.75) (P = .90). A postexercise ABI of greater than 0.90 was observed in 443 patients (13.8%). Baseline characteristics of the 443 patients with a postexercise ABI of greater than 0.90 were different compared with those of patients with a postexercise ABI of 0.90 or less. Patients with an ABI of greater than 0.90 were younger (mean age, 60 ± 13 vs 64 ± 12 years; P<.001) and had a lower prevalence of coronary artery disease (35% vs 43%; P<.001), hypertension (29% vs 36%; P = .002), and smoking (27% vs 35%; P<.001). A reduction of the postexercise ABI over baseline resting readings was observed in 2559 patients (79.7%), and was observed in 444 (56.3%) of 789 patients who had a normal resting ABI. The median reduction of ABI over baseline resting readings was 25% (interquartile range, 6%-55%). The postexercise ABI was similar to the resting ABI in 552 patients (17.2%), and was higher in only 98 patients (3.1%).

During a median follow-up of 8 years (range, 0.1-22.2 years), death was recorded in 1321 patients (41.2%); 671 (50.8%) of these deaths were because of cardiac causes. Figure 1 shows the incidence of overall mortality and cardiac death in patients included from 1983 to 1993, from 1994 to 2005, and from 1983 to 2005, across the whole range of resting ABI values, postexercise ABI values, and reductions of ABI values over baseline resting readings. Tests for heterogeneity revealed no significant difference in unadjusted risk of resting ABI values, postexercise ABI values, and reductions of ABI over baseline readings for predicting overall (P=.60, .34, and .15, respectively) and cardiac mortality (P=.40, .15, and .09, respectively) between patients included from 1983 to 1993 and those included from 1994 to 2005. The unadjusted risk of overall mortality and cardiac death increased with decreasing resting ABI values (hazard ratio per 0.10 decrease, 1.11 [95% CI, 1.09-1.13] and 1.15 [95% CI, 1.13-1.18], respectively), with decreasing postexercise ABI values (hazard ratio per 0.10 decrease, 1.16 [95% CI, 1.13-1.18] and 1.10 [95% CI, 1.09-1.13], respectively), and with increasing reductions of ABI over baseline resting readings (hazard ratio per 10% decrease, 1.11 [95% CI, 1.09-1.14] and 1.17 [95% CI, 1.14-1.21], respectively) (P<.001 for all differences). In patients who died and who had normal resting ABI values and no reduction in the ABI after exercise, a cardiac cause of death was noted in 38.2%; in patients who died and who had normal resting ABI values and reductions in the ABI after exercise, a cardiac cause of death was noted in 55.1% (P<.001).

Kaplan-Meier survival curves for patients with varying resting ABI values, postexercise ABI values, and reductions of ABI are presented in Figure 2. Worse survival was observed in patients with lower resting ABI values, lower postexercise ABI values, and higher reductions of ABI values over baseline resting readings (P<.001 for all).

We found that decreasing values of resting ABI, decreasing values of postexercise ABI, and higher reductions of ABI were significantly associated with an increased risk of overall mortality and cardiac death (Table 2). In a final multivariate model, we assessed the value of a reduction of the ABI with adjustment for clinical risk factors and resting ABI values, and observed that an increased reduction remained significantly associated with overall mortality and cardiac death (Table 2). Tests for heterogeneity revealed no significant difference in adjusted risk of resting ABI values, postexercise ABI values, and reductions of ABI over baseline readings for predicting overall and cardiac mortality between patients included from 1983 to 1993 and patients included from 1994 to 2005.

Table 3 shows the relative risk ratios of a reduction of ABI in patients with resting ABI values of greater than 0.90 and those with resting ABI values of 0.90 or less, with adjustment for clinical risk factors and age. In both subgroups of patients, a 6% to 24%, a 25% to 55%, and a greater than 55% reduction of ABI over baseline resting readings was associated with an increased risk for late mortality, using a reduction of less than 6% as the comparator. More important, in patients with a resting ABI of greater than 0.90, the relative risk ratio for late mortality was 1.6 times greater for a reduction of 6% to 24%, 3.5 times greater for a reduction of 25% to 55%, and 4.8 times greater for a reduction of greater than 55% when a reduction of less than 6% was used as the comparator. Tests for heterogeneity revealed a significant interaction between resting ABI values and reductions of ABI over baseline readings (P = .03). Kaplan-Meier survival curves for patients with resting ABI values greater than 0.90 show a significantly worse outcome for those with a 6% to 24% and a greater than 25% reduction of the ABI (P<.001) (Figure 3).

The present study shows that lower resting and postexercise ABI values and an increased reduction of ABI over baseline readings are significantly associated with an increased incidence of long-term overall mortality and cardiac death. In addition, in patients with resting ABI values of more than 0.90, a reduction of ABI over baseline readings identified those at increased risk for late mortality. These findings were independent of the presence of established clinical risk factors.

In the United States, different prevalence values for PAD have been reported, ranging from 4% in patients 40 years and older to more than 20% in patients 70 years and older.^1,5-8 PAD is a manifestation of atherosclerosis, and its presence should be regarded as a marker for atherosclerosis in other vascular beds.¹⁷ Important risk factors include cigarette smoking, hypercholesterolemia, and hypertension, and modification of these risk factors is a substantial part in managing PAD. In our study, patients included from 1994 to 2005 compared with those included from 1983 to 1993 more commonly presented with a history of myocardial infarction and hypercholesterolemia, less commonly smoked, and more commonly received cardiovascular medication. This may reflect the trend in increasing awareness of detecting and treating clinical risk factors.

The ABI is commonly used for the assessment of lower extremity arterial obstruction and for the screening of patients with suspected PAD.⁸ The ABI correlates with the extent of angiographic coronary artery disease, reflecting the concept that PAD is a marker of generalized atherosclerosis.¹⁸ It is well established that low ABI values at rest predict cardiac and overall mortality. A resting ABI of 0.90 or less has been associated with an increased risk of 2 to 7 for overall mortality and 2 to 4 for cardiovascular mortality, compared with a resting ABI of more than 0.90.^10-16 Knowledge about the risk of mortality across the whole range of ABI values is limited. The Honolulu Heart Program study¹⁹ compared the adjusted risk for the composite end point of cardiac death and nonfatal myocardial infarction in patients with an ABI of less than 0.80 and 0.80 to 1.00 vs those with an ABI of 1.00 or more, and found a higher risk in patients with an ABI of less than 0.80 compared with patients with an ABI of 0.80 to 1.00. Another study¹⁰ used resting ABI values of 0.40 and 0.85 as cutoffs to divide its population into subpopulations; the highest risk-adjusted relative risk for mortality was observed in patients with an ABI of less than 0.40. Leng et al¹⁶ used resting ABI cutoff values of 1.10, 1.00, 0.90, and 0.70, and found a significant linear increase in mortality across decreasing ABI categories. In the present study, we found that the adjusted risk for overall mortality increased by 8% for every 0.10 decrease of the resting ABI, and by 9% for every 0.10 decrease of the postexercise ABI. The adjusted risk for cardiac death increased by 12% and 15% for every 0.10 decrease of the resting ABI and postexercise ABI, respectively.

A healthy person can maintain ankle systolic pressures at normal levels during modest workloads. In patients with PAD, however, a decrease in systolic pressures is often measured during low levels of workload.^20,21 In a previously published study, Ouriel et al²² investigated whether the diagnostic accuracy for PAD could be improved through the use of postexercise ABI measurements, and they suggested that postexercise ABI values may especially be useful in patients with normal ABI values at rest. To our knowledge, no studies have been published about the predictive value of postexercise testing on long-term mortality. Our results demonstrate that reductions of the postexercise ABI offered additional mortality information in patients with abnormal resting ABI values. More important, in symptomatic and asymptomatic patients with resting ABI values above 0.90, a reduction of ABI over baseline resting readings can also identify those at increased risk for mortality.

Although the prevalence of PAD is high in industrialized countries, PAD remains an underdiagnosed and undertreated disease in primary care.^3,6,8 Therefore, the identification of patients with suspected PAD who are at increased risk for late events is necessary for disease control and selection of appropriate treatment strategies.²³ During recent years, more attention has focused on the identification and treatment of underlying cardiovascular risk factors in patients with PAD, which may have resulted in an increased prescription of statins, angiotensin-converting enzyme inhibitors, aspirin, and β-blockers. Measurements of resting and postexercise ABI values are simple, inexpensive, and noninvasive, and can be performed when patients are assessed for their risk factors. A recent study¹³ showed that the measurement of resting ABI in addition to conventional risk factors significantly improved the prediction of fatal myocardial infarction. Results from this study suggest that measurement of ABI at rest may be incorporated among other tools for identifying patients at increased risk for cardiovascular events. These patients may subsequently benefit from preventive pharmacologic and nonpharmacologic interventions, to reduce their risk of complications of systemic atherosclerosis.

In conclusion, this long-term prospective cohort study shows that resting and postexercise ABI values are independently associated with late overall mortality and cardiac death. In addition, in patients with normal resting ABI values, a reduction of the postexercise ABI over baseline readings can identify those at increased risk of long-term mortality.

Correspondence: Don Poldermans, MD, PhD, Erasmus Medical Centre, Dr Molewaterplein 40, 3015 GD Rotterdam, the Netherlands (d.poldermans@erasmusmc.nl).

Accepted for Publication: September 12, 2005.

Author Contributions: Dr Poldermans had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Financial Disclosure: None.

Funding/Support: This study was supported by an unrestricted educational grant from Sanofi-Aventis/BMS, Maassluis/Woerden, the Netherlands (Dr Feringa); and an unrestricted educational grant from Lijf en Leven, Rotterdam (Dr Schouten).

Role of the Sponsors: The funding bodies had no role in data extraction and analyses, in the writing of the manuscript, or in the decision to submit the manuscript for publication.