Effect of Diagnostic Criteria on the Prevalence of Peripheral Arterial Disease
The San Luis Valley Diabetes Study
Background The ankle/brachial systolic blood pressure index (ABI), a noninvasive measure of peripheral arterial disease (PAD), is widely used in epidemiological studies. However, the normal ranges of the ABI in healthy populations and ABI criteria for the diagnosis of PAD in large population studies have not been critically evaluated.
Methods and Results The San Luis Valley Diabetes Study (SLVDS) was designed to evaluate the prevalence and complications of non–insulin-dependent diabetes mellitus (NIDDM) in a biethnic population. The present study was conducted as part of the SLVDS to assess the prevalence of vascular disease in 1280 nondiabetic control subjects and 430 patients with NIDDM. The ABI criteria for PAD were developed in 403 healthy individuals with a low risk for cardiovascular disease. In these low-risk subjects, the average resting ABI value was 0.07 lower in women than in men. In both sexes, the dorsalis pedis ABI was 0.04 lower than in the posterior tibial artery, and the left leg ABI was 0.02 lower than the right leg ABI (all differences, P<.05). In the low-risk subjects, ABI values were lower after exercise than at rest and had similar differences by sex and leg as observed at rest. Using specific abnormal cutoff points for the ABI, we evaluated three criteria for PAD in the overall population: two abnormal vessels in the same leg at rest (both dorsalis pedis and posterior tibial arteries), one abnormal vessel per leg at rest, and an ABI abnormality only after exercise. Subjects classified with PAD by the two-vessel criterion had a higher frequency of claudication and the physical finding of an absent pulse compared with subjects without PAD or patients with PAD defined by the one-vessel or exercise criterion. Use of the two-vessel criterion identified an increased risk of PAD with increasing age, NIDDM, smoking, hypertension, and elevated cholesterol levels. In contrast, the one-vessel PAD criterion was associated only with increasing age and smoking, and exercise-diagnosed PAD was not associated with any cardiovascular risk factor except for male sex.
Conclusions In low-risk subjects, the normal distribution and lower abnormal cutoff point values of the ABI differed by type of test, sex, ankle vessel, and leg. When these specific abnormal cutoff points were applied to the SLVDS population, the two-vessel abnormal criterion described patients with typical clinical characteristics of PAD and the expected associations of PAD with cardiovascular risk factors. These clinical characteristics and cardiovascular risk factor associations were less evident with PAD diagnosed by the one-vessel or exercise criterion. Therefore, an abnormal dorsalis pedis and posterior tibial ABI in the same leg at rest should be used for the diagnosis of PAD in epidemiological studies.
Peripheral arterial disease (PAD) is a common manifestation of the atherosclerotic disease process, affecting from 12% to 14% of the general population and as many as 20% of individuals over the age of 75.1 2 3 The diagnosis of PAD solely from a questionnaire history of intermittent claudication or the finding of an absent pedal pulse underestimates the prevalence of the disease.4 5 The noninvasive laboratory diagnosis of PAD provides a higher estimate of disease prevalence, since these tests have a high sensitivity (as well as specificity) for detecting angiographically defined arterial occlusive disease.6 7
A commonly used noninvasive test for PAD is the measurement of systolic blood pressures in the ankles and arms with a Doppler ultrasonic instrument, from which the ankle/brachial index (ABI) is derived.6 A low ABI (or other abnormal hemodynamic test) is highly predictive not only of the presence of arterial occlusive disease but also of subsequent cardiovascular mortality.8 9 10 However, there is uncertainty regarding the lower normal limit of the ABI, with published abnormal cutoff points ranging from 0.80 to 0.98.1 2 6 7 Varying the value defining an abnormal ABI can markedly affect estimates of PAD prevalence,11 yet adequate studies have not been conducted in healthy populations to determine the normal ranges and lower abnormal cutoff point values of the ABI. Also, despite the presence of three vessels in each ankle, often only the pressure from one vessel is used to calculate the ABI. The choice of ankle vessel ranges from using the vessel with the higher pressure,7 12 the lower pressure,13 or the vessel with the best quality of the Doppler signal.14 Thus, there are no uniform abnormal ABI cutoff points or vessel-specific criteria for the diagnosis of PAD in population studies.
In addition to hemodynamic measurements at rest, many studies use exercise testing to detect patients with mild forms of PAD who have normal resting ABI values.15 As for resting values, the normal ranges and test characteristics of the ABI after exercise have not been critically evaluated in large populations. Also, the additional diagnostic yield of an exercise test compared with only resting measurements has been questioned.7 Therefore, estimates of the prevalence and outcomes of PAD in epidemiological studies may be affected by a variety of factors, including the abnormal cutoff point of the ABI, the specific ankle vessels measured to form the ABI, and the type of test used (rest versus exercise).
The San Luis Valley Diabetes Study (SLVDS) is a geographically based study designed to evaluate the prevalence and complications of non–insulin-dependent diabetes mellitus (NIDDM) in a biethnic population of Hispanics and non-Hispanic whites.16 The present study was conducted to define the optimal diagnostic methods used to assess PAD prevalence in the SLVDS population. Normal ranges and lower limits of the distribution of ABI values were derived from a nondiabetic subset of the population with a very low risk for PAD. Sex-, vessel-, and leg-specific ABI criteria were used to describe patients with PAD by either rest or exercise testing. These methods were used to address the study hypothesis that patients with an abnormal ABI at rest in the dorsalis pedis (DP) and posterior tibial (PT) arteries of the same leg would have clinical characteristics and risk factor profiles typical of PAD compared with patients with only one abnormal vessel per leg or patients with an abnormal ABI only after exercise testing.
The present study, designed to evaluate tests used to assess the prevalence of PAD, was conducted as part of the SLVDS.16 The study area included Alamosa and Conejos counties, which are located in the San Luis Valley of southern Colorado at an elevation of 7100 to 7400 ft above sea level. These two rural counties had a population estimated to be 21 070 in 1990 that was composed of 46.1% Hispanic and 52.2% non-Hispanic white residents.17 Baseline assessments for prevalent and incident diabetic and control subjects were conducted between May 1984 and August 1988.
Patients with a history of diabetes and between the ages of 20 and 74 were identified by a review of the medical records from the two hospitals and five medical practices in the study area, as well as from public advertising. Of the 770 diabetics identified in the San Luis valley, 230 were ineligible for further evaluation based on their age or residency, and 100 refused to participate, leaving 440 presumed diabetic individuals who were available for further clinical evaluation. The diagnosis of diabetes was made or confirmed using World Health Organization criteria18 from an oral glucose tolerance test. After the oral glucose tolerance test, 337 individuals were confirmed to have NIDDM, and 71 control subjects were newly diagnosed as diabetic. An additional 22 individuals who did not meet criteria for diabetes from the oral glucose tolerance test but were receiving hypoglycemic medications (7 receiving insulin and 15 receiving oral hypoglycemic drugs) were also considered to have diabetes. Thus, a total of 430 individuals were classified as having NIDDM.
Control subjects without a history of diabetes and between the ages of 20 and 74 were identified from 3432 households interviewed in the two-county area. Control subjects were then randomly selected within age, sex, ethnic, and county strata to reflect the older age and ethnic distribution of the diabetic subjects. Of the 1991 eligible control subjects, 1351 participated in further clinical evaluation. After the oral glucose tolerance test, 71 potential control subjects were classified as having NIDDM as described above. The oral glucose tolerance test also identified 173 patients with impaired glucose tolerance who were still included in the control group of 1280 subjects. This study was approved by the University of Colorado Health Sciences Center Human Subjects Committee. Informed consent was obtained for an interview, a physical examination, laboratory tests, and a noninvasive evaluation of the peripheral circulation. In subjects undergoing exercise testing, additional informed consent was obtained.
Vascular History and Physical Examination
Patients were evaluated by the Rose Questionnaire for intermittent claudication.19 Claudication was defined as pain or discomfort in the calf of either leg that was not present at rest and began with walking exercise, regardless of whether the remaining Rose criteria for claudication were met. All physical examinations were consistently performed by one nurse practitioner. Training in the physical examination and monitoring of quality control was done by one of the investigators (W.R.H.) every 6 months during the first 2 years and yearly thereafter during the data-gathering phase of the study. The DP and PT pulse of each foot was recorded as normal (distinct) and considered abnormal if either diminished (attenuated but palpable) or absent. Details of the methods and performance of the vascular history and physical examination in this population have been previously reported.2
The vascular laboratory studies were performed by two trained technicians as previously described.2 Subjects were evaluated while supine, after a 5-minute rest. A cuff (D.E. Hokanson) was placed on each arm and ankle, and a Doppler ultrasonic instrument (model 801, Parks Electronics) was used to detect each pulse. The cuff was inflated to 10 mm Hg above systolic pressure and deflated at 2 mm Hg/s. The first reappearance of the pulse was taken as the systolic pressure. The systolic pressure was taken a second time, and the two values were averaged. If any pair of values differed by more than 6 mm Hg, repeat pressures were taken, and the average of the most consistent pair was used for subsequent analysis. Pressures were obtained in the following order: right arm, right DP, right PT, left DP, left PT, and then left arm.
Subjects were next considered for additional diagnostic evaluation with an exercise test. Individuals who had resting ratios of less than 0.80 did not perform the exercise test because ratios below this value have been strongly associated with the presence of angiographically defined PAD.15 Subjects were also excluded from the exercise test if they had symptoms of chest pain or shortness of breath on minimal activity, severe arthritis, amputation, or any other non-PAD medical condition that limited their ability to exercise. In the SLVDS study population of 1710 individuals, 466 (27%) were excluded from the exercise test based on the above criteria. The exclusion rate was greater (P<.05) for women (30%) than for men (24%).
Exercise testing was performed on a motor-driven treadmill (model 300M, Trotter) at 1.5 mph, 8% grade for 30 seconds; then the speed was advanced to 2 mph, 8% grade for the final 4.5 minutes. At the termination of exercise, subjects returned to the supine position for the recording of arm and ankle pressures at 1 minute after exercise. Given the need to rapidly obtain pressure measurements after exercise, only the arm with the highest resting pressure and the artery in each ankle with the highest resting pressure were used for the postexercise measurements.
Calculation of Ratios
The calculation of the ankle-to-arm systolic blood pressure ratios has been previously described.2 Briefly, in an individual if the difference in blood pressure between the right and left arms was in the range of −9 mm Hg to +8 mm Hg (the 95% confidence interval), then the two pressures were averaged. If the difference between arm pressures was more than this range (occurring in 5% of the subjects), it was assumed that there may be an occlusion in the arterial circulation to the arm with the lower pressure. In this situation, the highest arm pressure was used in the subsequent calculation of ankle-to-arm ratios.
Once the four ABI values were obtained (right and left DP and PT ratios), two different patterns of ABI abnormalities at rest were evaluated in the population. Patients with only one abnormal ABI in a leg were considered to have one-vessel disease. If they had one abnormal vessel in both legs, they were considered to have bilateral one-vessel disease but were still classified as having one-vessel disease. Patients with both vessels (DP and PT) abnormal in the same leg (either right leg, left leg, or both legs) were classified as having two-vessel disease.
After exercise, pressures were measured in one arm and in one ankle vessel, resulting in one postexercise ABI per leg. Subjects who had normal resting ABI values (above the percentile cutoff points defined in Table 2⇓) but an abnormal postexercise value in either leg were classified as having exercise-diagnosed PAD.
Definition of Subgroups
From the control (nondiabetic) group, a subset of individuals at a very low risk for cardiovascular disease (the “low-risk group”) was selected to establish normal ranges for the vascular laboratory tests. This low-risk group consisted of 403 individuals: 231 women and 172 men. Criteria for inclusion in the low-risk group included normal glucose tolerance, no symptoms of angina or claudication, and no previous history of cardiovascular disease. In addition, low-risk subjects were either nonsmokers (less than 100 lifetime cigarettes) or former smokers (more than 100 lifetime cigarettes but quit an average of 14 years before enrollment in the study). Current smokers were excluded from the low-risk group. To qualify for inclusion in the low-risk group, the LDL cholesterol concentration in low-risk subjects was less than 164 mg/dL in men and less than 156 mg/dL in women (the lower 75th percentile of the values in the control group). Similarly, the triglyceride concentration was less than 254 mg/dL in men and less than 216 mg/dL in women (the lower 90th percentile of the values in the control group). Low-risk subjects were included who had a blood pressure below 140/90 mm Hg, but 5.2% were taking hypertensive medications. Hypertension was defined as taking antihypertensive medications, systolic blood pressure of more than 140 mm Hg, or diastolic blood pressure of more than 90 mm Hg.
Compared with the non–low-risk subjects in the control group or patients with NIDDM, low-risk subjects had a similar distribution of men and women but were younger and had a lower percentage of Hispanic subjects (Table 1⇓). Total pack-years of smoking (packs per day multiplied by years of smoking) in the low-risk group was significantly less than in the control group or in patients with NIDDM. Low-risk subjects also had lower resting arm blood pressures and lipid levels than the control or NIDDM population.
The glucose oxidase method was used to measure glucose concentrations from venous plasma samples.20 Total serum cholesterol concentration was measured with an esterase-oxidase method,21 and HDL cholesterol concentration was measured by an enzymatic method using magnesium precipitation.22 Serum triglyceride concentration was measured with an enzymatic assay.23 The LDL cholesterol concentration was calculated as the total cholesterol concentration minus one fifth the triglyceride concentration minus the HDL cholesterol concentration.
Data analyses were conducted using the Statistical Analysis System (version 5, SAS Institute, Inc). Descriptive data are presented as mean±SD or as percentiles for the distribution of the ABI values. Because there was an overselection of patients with diabetes in the SLVDS, the prevalence of PAD (Table 3⇓) was adjusted for the prevalence of NIDDM. Odds ratios are presented with their 95% confidence intervals. Logistic regression was used to calculate the odds ratios for PAD, adjusted for specific patient characteristics such as age, sex, and diabetes status. There was no interaction between such variables as ethnicity with diabetes, so ethnicity was not included in the logistic models presented. Differences between patients with and without PAD were determined by unpaired t tests, between-subject ANOVA, or χ2 analysis. Multiple linear regression techniques were used to evaluate the effects of height on the ABI. In this analysis, ankle blood pressure was regressed on height and adjusted for arm blood pressure, age, diabetes status, and sex.
Criteria for PAD
Noninvasive vascular measurements from the SLVDS low-cardiovascular-risk group (n=403, see “Methods” for group definition) were used to define normal ranges for the ABI. Analysis of the distribution of the ABI values from the low-risk subset revealed that men had higher average resting ratios in both the DP and PT arteries than women (Table 2⇓). The average ABI from all vessels was 0.07 greater in men than in women. In both sexes, PT arteries had higher ratios than DP arteries, with an average difference in all legs between the two arteries of 0.04. In addition, ABI ratios in the right leg were higher than in the left leg by 0.02. Because all of these differences between mean ABI values were statistically significant, separate lower 1.0, 2.5, and 5.0 percentile cutoff points for the ABI were determined by sex, vessel, and leg (Table 2⇓). These differentiated cutoff point values for an abnormal ABI had a large range. For example, the lower 2.5 percentile value for the right PT artery in a man was 1.03, whereas the lower 2.5 percentile value for the left DP artery in a woman was 0.87.
The low-risk subjects also underwent exercise testing. After exercise, ABI values were reduced relative to the resting measurements (Table 2⇑). Similar to the findings at rest, men had higher postexercise ABI values compared with women, and the right leg ABI was higher than in the left leg (both P<.05). Therefore, after exercise, separate lower limits of the normal range of ABI values were also determined for sex and leg. These differences in ABI ratios between healthy men and women were consistently observed in both Hispanics and non-Hispanic whites (data not shown).
Previous studies have suggested that height may explain a portion of the difference in ABI values between men and women.13 In the low-risk subset and in the remaining population (n=1307), height had a significant but modest effect on ankle pressure when controlling for arm pressure, sex, diabetes status, and age. In all subjects, for each 10 cm in height, there was an approximate 1 mm Hg increase in ankle pressure. After adjusting for height, the difference in ABI values between men and women was reduced but remained statistically significant.
Prevalence of PAD
The estimated prevalence of PAD in the SLVDS population (n=1710) was determined using the different abnormal cutoff points of the ABI as well as with either the one-vessel, two-vessel, or exercise assessments (Table 3⇓). The one-vessel abnormality at rest yielded a disease prevalence (adjusted for diabetes status) ranging from 3.4% to 12.1% at the 1.0 and 5.0 percentile cutoff points, respectively. Within the one-vessel disease category, an abnormal DP ABI was less common than an abnormal PT ABI. The finding of one abnormal DP ratio in one leg and one abnormal PT ratio in the other leg was uncommon. The two-vessel disease category gave intermediate adjusted prevalence estimates, ranging from 1.6% to 4.3% at the 1.0 and 5.0 percentiles, respectively. The exercise PAD category gave the lowest adjusted prevalence estimates, ranging from 0.5% at the 1.0 percentile to 4.1% at the 5.0 percentile.
Clinical Characterization of Subjects With ABI-Defined PAD
Based on the different diagnostic criteria for PAD, the adjusted prevalence of the disease may be as low as 0.5% (exercise criterion at the 1.0 percentile) or as high as 12.1% (one-vessel criterion at the 5.0 percentile). Given the lack of an historical precedent for optimal diagnostic criteria for PAD in epidemiological studies,24 the clinical characteristics of subjects in the different diagnostic groups were compared. At the 1.0 percentile cutoff point of the ABI, patients with two-vessel PAD had a frequency of claudication by the Rose questionnaire of 17.4%, which was greater than the 4.0% frequency of claudication in subjects without PAD by any criterion (Fig 1⇓). The frequency of claudication symptoms in individuals with two-vessel disease remained significantly greater than in nondiseased subjects at the 2.5 (12.3%) and 5.0 (11.3%) percentile cutoff points. In contrast, either one-vessel– or exercise-diagnosed PAD was not associated with an increased frequency of claudication at any cutoff point (Fig 1⇓).
The frequency of an abnormal pulse by physical examination in individuals with two-vessel disease was 45.7% at the 1.0 percentile, 37.0% at the 2.5 percentile, and 28.2% at the 5.0 percentile (Fig 2⇓). All of these values were greater than the 11.5% frequency of an abnormal pulse in subjects without PAD (P<.05). At the 1.0 percentile, the one-vessel disease frequency of an abnormal pulse was 20.5%, which was increased compared with subjects without PAD but less than the frequency of an abnormal pulse in patients with two-vessel PAD. In patients with one-vessel disease, the frequency of a pulse abnormality was not increased compared with subjects without PAD at the 2.5 or 5.0 percentile cutoff points. Patients with exercise-diagnosed PAD did not have an increased frequency of a pulse abnormality at any cutoff point value. Therefore, only the two-vessel criterion for PAD was consistently associated with the expected history and physical examination findings of vascular disease.
PAD Cardiovascular Risk Factor Associations
The relationships between cardiovascular risk factors and PAD were contrasted among the one-vessel, two-vessel, and exercise assessments of disease prevalence. The following characteristics were included in the analysis because of significant univariate associations with PAD: age, sex, diabetes status, smoking history, hypertension, and cholesterol and triglyceride levels (Table 4⇓). Smoking was associated with an increased prevalence of one-vessel disease at all three cutoff points, and increasing age was associated with one-vessel disease at the 2.5 and 5.0 percentile cutoff points. However, diabetes, hypertension, and elevated lipid levels were not associated with one-vessel disease. In contrast, patients with two-vessel PAD had significant associations with increasing age, NIDDM, smoking, hypertension, and cholesterol levels at all three cutoff points. Male sex increased the risk of exercise-diagnosed PAD at the 2.5 and 5.0 percentiles, but no other cardiovascular risk factors were associated with exercise-diagnosed PAD.
The ABI is a noninvasive, hemodynamic measure used to estimate the prevalence of individuals with atherosclerotic PAD. The test characteristics of the ABI are robust when patients with angiographically defined PAD are compared with healthy subjects.6 7 14 25 However, when these hemodynamic measures are applied to population studies, there are no uniform criteria regarding which ankle vessels should be measured (DP versus PT), what value of ABI should be considered abnormal, and the role of exercise testing.
Normal Ranges of the ABI
Previous descriptions of the range of ABI values in healthy subjects revealed mean values of approximately 1.10, with a lower 95% confidence interval of 0.97 to 0.98.6 7 15 However, these studies did not investigate the potential confounders of sex, vessel, and leg in defining the normal range of ABIs. In the SLVDS, 403 subjects with a low risk of vascular disease were selected to develop normal ranges and cutoff values for the ABI. The ABI values in the present study were higher in healthy men than in healthy women for each vessel tested. Higher ABI values in men may be explained, in part, by the effects of height.13 26 Similar, but modest, associations of height with increasing ankle pressure were observed in the present study, but other unexplained factors may also relate to the sex effects on the ABI.
The differences in ABI between PT and DP ankle vessels have not been previously described. In a study of 23 healthy subjects, it was noted that the two ankle pressures did not differ by more than 10 mm Hg.15 A smaller-diameter DP vessel (relative to the PT) may be more difficult to locate with the Doppler probe, leading to an underestimation of the “true” arterial pressure. The difference in pressures between the right and left legs has been previously observed13 and may relate to the order in which the pressures were taken. Even though subjects rested supine for 5 minutes before measurements of systolic pressures, a decrease in central arterial pressure over time would result in the higher pressure occurring in the leg first measured.13
In healthy subjects, it has been observed that ABI values are lower after exercise than at rest, necessitating lower cutoff point values for the postexercise ABI.7 The present study confirmed these findings, but it is important that the effects of sex and order of measurement observed with the resting measurements persisted after exercise. Therefore, a critical finding of the present study was that there is no one ABI cutoff value that can be applied to population studies. Rather, abnormal cutoff points of the ABI should be based on the type of test (rest or exercise), sex, vessel, and order of measurements.
Diagnostic Criteria for PAD
Despite the presence of three vessels at the ankle (PD, PT, and peroneal), most investigators choose only one to form the ABI at rest, and there is no consistency among studies as to which vessel is evaluated. Some investigators report the higher ankle pressure,7 12 others report the lower pressure,13 and in others the choice is based on the quality of the Doppler signal.14 PAD is a diffuse process in both diabetic and nondiabetic subjects, involving the iliac, femoral, and popliteal arteries as well as the tibial vessels.27 28 The finding of an isolated, single tibial artery occlusion, with the remainder of the circulation unaffected, is extremely uncommon.27 28 These observations suggest that the majority of patients with hemodynamically significant occlusive disease should have abnormal ABIs in both the DP and PT arteries of the same leg (two-vessel disease). Large differences in pressure between the DP and PT suggest additional occlusive disease in one of the tibial arteries.15 Thus, the two-vessel criterion is the expected pattern of ABI abnormality in the majority of patients with hemodynamically significant PAD. In contrast, the one-vessel criterion of PAD, with only one abnormal vessel per leg, defines either subjects with isolated occlusive disease in a tibial artery or subjects classified as abnormal because of measurement error or alterations in physiology (eg, vasospasm in one vessel).
In patients with mild forms of occlusive disease, the ankle pressure may be normal at rest because of adequate collateral flow around the arterial occlusions.15 In these subjects, an exercise test was proposed as a method of detecting mild forms of PAD.15 This is because with exercise the decrease in peripheral resistance distal to the arterial obstruction (vasodilation) is not matched by an appropriate increase in arterial flow, resulting in a pressure drop at the ankle. In contrast, healthy subjects will have an increase in ankle systolic pressure with exercise but not to the same degree as the arm pressure, resulting in a lower ABI after exercise than at rest.7 This normal hemodynamic response to exercise necessitates lower abnormal ABI cutoff points after exercise than at rest, as demonstrated in the present study. However, in epidemiological studies it was not known if an abnormal exercise test classified patients with typical clinical features of PAD.
In the present study, the validity of the different diagnostic modalities for PAD (one-vessel, two-vessel, or exercise testing at the different cutoff points) was challenged by several different criteria. First, although the presence of claudication and an abnormal pulse may underestimate disease prevalence,4 these clinical findings should be more common in patients with PAD than in those without the disease. Second, there is a well-described association of PAD with increasing age,26 29 diabetes,29 30 smoking,31 32 hypertension,31 32 and elevated lipid levels.33 34 However, sex is not a consistent risk factor, with some studies31 35 showing a higher PAD prevalence in men and others26 29 32 showing an equal sex prevalence.
The one-vessel ABI abnormality had the highest prevalence in the SLVDS population. An isolated abnormal PT ABI was more common than an isolated abnormal DP ABI, perhaps because the PT artery had a higher cutoff point value. Although a tibial artery occlusion may not be associated with claudication (assuming that the proximal circulation and the other tibial and peroneal vessels are not occluded), it is surprising that these subjects did not have an increased frequency of a pulse abnormality at all abnormal cutoff points of the ABI. Patients with the one-vessel ABI abnormality also had few of the expected risk factor associations with PAD, in that smoking was the only risk factor statistically associated at the first to the 5.0 percentile cutoff points. Therefore, the one-vessel criterion may describe individuals with only modest clinical and risk factor characteristics of PAD.
Subjects who had normal ABI values at rest but abnormal values with exercise were uncommon in the population. At all ABI cutoff points, patients with PAD only by the exercise test did not have an increased frequency of the symptom of claudication or a pulse abnormality on examination. Exercise-diagnosed subjects had no risk factor associations with PAD except for male sex, which is not a consistently observed risk factor for PAD.26 29 32 The increased prevalence of men with exercise-diagnosed PAD at the 2.5 and 5.0 percentile cutoff points may be due to selection bias, with more men than women tested on the treadmill. Thus, there was no cutoff point of the postexercise ABI that described a group of individuals with the expected characteristics of PAD.
As lower extremity atherosclerosis often affects major proximal arteries or is diffuse, it was anticipated that the two-vessel disease criterion, with the DP and PT ABI abnormality in the same leg, would be most consistent with the disease biology. Patients with two-vessel PAD had an increased frequency of claudication and an abnormal pulse at all cutoff points of the ABI (in contrast to the clinical findings in the exercise and one-vessel disease groups). Therefore, the two-vessel PAD criterion is both biologically plausible and associated with the clinical characteristics of PAD. Because patients with two-vessel disease also had the expected cardiovascular risk factor associations, an abnormal DP and PT artery ABI in the same leg is the most robust criterion for PAD in population studies. The prevalence of PAD in the SLVDS, corrected for NIDDM prevalence, may thus be estimated by the two-vessel criterion to fall within the range of 1.6% (1.0 percentile cutoff point) to 4.3% (5.0 percentile cutoff point).
The present study evaluated the effects of different diagnostic criteria for PAD on the estimates of disease prevalence in a population-based study. Different diagnostic criteria for PAD by the ABI not only yield markedly different estimates of disease prevalence but also describe groups of patients with different clinical and risk factor associations for PAD. The two-vessel disease criterion was strongly associated with the expected clinical characteristics of patients with PAD and therefore is the recommended ABI diagnostic strategy for detecting PAD in epidemiological studies.
This work was supported by National Institutes of Health (NIH) grants DK-30747 and CRC-RR00051. The authors wish to thank Judith Baxter, MA; Julie Marshall, PhD; Marion Rewers, MD, PhD; Judith Regensteiner, PhD; and Eric P. Brass, MD, PhD, for their helpful comments on the manuscript. Dr Hiatt is the recipient of an NIH Academic Award in Vascular Disease.
- Received September 29, 1994.
- Accepted October 14, 1994.
- Copyright © 1995 by American Heart Association
Criqui MH, Fronek A, Barrett-Connor E, Klauber MR, Gabriel S, Goodman D. The prevalence of peripheral arterial disease in a defined population. Circulation. 1985;71:510-515.
Criqui MH, Fronek A, Klauber MR, Barrett-Connor E, Gabriel S. The sensitivity, specificity, and predictive value of traditional clinical evaluation of peripheral arterial disease: results from noninvasive testing in a defined population. Circulation. 1985; 71:516-522.
Carter SA. Indirect systolic pressures and pulse waves in arterial occlusive disease of the lower extremities. Circulation. 1968; 37:624-637.
Criqui MH, Langer RD, Fronek A, Feigelson HS, Klauber MR, McCann TJ, Browner D. Mortality over a period of 10 years in patients with peripheral arterial disease. N Engl J Med. 1992; 326:381-386.
Fowkes FGR, Housley E, Cawood EHH, MacIntyre CCA, Ruckley CV, Prescott RJ. Edinburgh Artery Study: prevalence of asymptomatic and symptomatic peripheral arterial disease in the general population. Int J Epidemiol. 1991;20:384-392.
Carter SA. Response of ankle systolic pressure to leg exercise in mild or questionable arterial disease. N Engl J Med. 1972; 287:578-582.
Hamman RF, Marshall JA, Baxter J, Kahn LB, Mayer EJ, Orleans M, Murphy JR, Lezotte DC. Methods and prevalence of non-insulin-dependent diabetes mellitus in a biethnic Colorado population: the San Luis Valley Diabetes Study. Am J Epidemiol. 1989;129:295-311.
US Bureau of the Census. 1990 Census of the Population: General Population Characteristics of Colorado. Washington, DC: US Government Printing Office; 1990 CP-1-7, 1992:Table 5.
World Health Organization. Diabetes Mellitus: Report of a WHO Study Group. Geneva, Switzerland: WHO; Technical Report Series No. 727, 1985.
Rose GA, Blackburn H. Cardiovascular Survey Methods. Geneva, Switzerland: WHO; Monograph Series No. 56, 1968:172-175.
Beckman Instruments. Glucose Analyzer 2 Operating Manual. Fullerton, Calif: Beckman Instruments; 1988.
Richmond W. Preparation and properties of a cholesterol oxidase from Nocardia sp and its application to the enzymatic assay of total cholesterol in serum. Clin Chem. 1973;19:1350-1356.
Warnick GR, Benderson J, Albers JJ. Dextran sulfate-Mg precipitation procedure for quantitation of high-density-lipoprotein cholesterol. Clin Chem. 1982;23:1379-1388.
Stavropoulos WS, Crouch RD. A new colorimetric procedure for the determination of serum triglycerides. Clin Chem. 1974;20:857.
Fowkes FGR. The measurement of atherosclerotic peripheral arterial disease in epidemiological surveys. Int J Epidemiol. 1988;17:248-254.
Donnan PT, Thomson M, Fowkes FGR, Prescott RJ, Housley E. Diet as a risk factor for peripheral arterial disease in the general population: the Edinburgh Artery Study. Am J Clin Nutr. 1993;57:917-921.
Strandness DE, Priest RE, Gibbons GE. Combined clinical and pathologic study of diabetic and nondiabetic peripheral arterial disease. Diabetes. 1964;13:366-372.
Guggenheim W, Koch G, Adams AP, Hoar CS, Wheelock FC. Femoral and popliteal occlusive vascular disease: a report on 143 diabetic patients. Diabetes. 1969;18:428-433.
Newman AB, Siscovick DS, Manolio TA, Polak J, Fried LP, Borhani NO, Wolfson SK, Cardiovascular Health Study Collaborative Research Group. Ankle-arm index as a marker of atherosclerosis in the Cardiovascular Health Study. Circulation. 1993;88:837-845.
Brand FN, Abbott RD, Kannel WB. Diabetes, intermittent claudication, and risk of cardiovascular events. Diabetes. 1989;38:504-509.
Reunanen A, Takkunen H, Aromaa A. Prevalence of intermittent claudication and its effect on mortality. Acta Med Scand. 1991;211:249-256.
Pomrehn P, Duncan B, Weissfeld L, Wallace RB, Barnes R, Heiss G, Ekelund LG, Criqui MH, Johnson N, Chambless LE. The association of dyslipoproteinemia with symptoms and signs of peripheral arterial disease: the Lipids Research Clinics Program Prevalence Study. Circulation. 1986;73(suppl I):I-100-I-107.
Juergens JL, Barker NW, Hines EA. Arteriosclerosis obliterans: a review of 520 cases with special reference to pathogenic and prognostic factors. Circulation. 1960;21:188-195.