Aims The aim of this study was to assess the prognostic importance of peripheral arterial disease (PAD) as evaluated by ankle blood pressure index (ABI), and the impact of ramipril on the prevention of major cardiovascular events in PAD patients included in the Heart Outcomes Prevention Evaluation (HOPE) study.
Methods and results Patients were randomized to treatment with ramipril or placebo and followed for 4.5 years. Ankle brachial blood pressure index was measured, mainly by digital palpation of the foot pulse, at baseline in 8986 patients. The ABI was subnormal (≤0.9) in 3099 patients and normal in 5887 patients. A low ABI was a strong predictor of morbidity and mortality during the follow-up even in patients with no clinical symptoms of PAD (n=6769). This was so for the primary outcome of the study; ABI>0.9:13.1%, 0.6–0.9: 18.2% and <0.6: 18.0% (P<0.0001) and for mortality from all causes: in those with a normal ABI it was 8.5%, in those with ABI >0.6–0.9, 12.4% and 14.2% in those with an ABI lower than 0.6 (P<0.0001). Ramipril reduced the risk of clinical outcomes in those with a clinical history of PAD as well as in the patients with subclinical PAD.
Conclusions The ABI even if measured simply by palpation of the foot arteries is a strong predictor for future cardiovascular events and for all-cause mortality. Ramipril prevented major cardiovascular events in patients with clinical as well as subclinical PAD.
Peripheral vascular disease
Ankle blood pressure
Ankle-brachial blood pressure index (ABI) is known to be a predictor for cardiovascular and total mortality when measured in both random samples from general practice1or in screening programmes.2–6In patients with intermittent claudication a low ABI is associated with progression of atherosclerotic disease.7In patients with a previous myocardial infarction or stroke8and in patients referred for coronary angiography9the occurrence of peripheral arterial disease (PAD) has been shown to increase the risk further.
In these epidemiological studies ankle blood pressure was measured in the foot arteries by ultrasound Doppler. This technique is not readily available in most general practices, so whether simpler measurement of the ABI using routine blood pressure measurement in the arm and palpation of the foot pulse is of predictive value isan important, but unanswered question. If a simple approach to measuring ankle and arm BP relation, utilizing no special equipment yields useful information, this method will have wide practical applicability.
In the Heart Outcomes Prevention Evaluation (HOPE) study patients at high risk for cardiovascular events were randomized in a 2×2 factorial design to treatment with ramipril or placebo and vitamin E or placebo. The main results of the study have been published.10,11The aim of the present study was (1) to evaluate this simpler measurement of ABI as a predictor of cardiovascular events, (2) to study the effect of ramipril treatment on prognosis in patients with symptomatic PAD and in groups of patients with different ABI but without symptomatic PAD.
2 Patients and methods
A detailed description of the design of HOPE study and the characteristics of the patients has been previously published.10A brief summary of the relevant details is provided below. HOPE study patients were aged 55 years or greater and had existing cardiovascular disease (coronary artery disease, previous stroke, peripheral vascular disease), or diabetes and an additional coronary risk factor (smoking, hypertension, hypercholesterolemia, low HDL or microalbuminuria) but no heart failure or evidence of left ventricular dysfunction. Patients were also excluded if they received an ACE-inhibitor or vitamin E or had uncontrolled hypertension. The definition of baseline PAD was either intermittent claudication with an ABI <0.9, or previous vascular intervention or limb amputation for PAD. A history of concomitant diseases was registered at baseline, together with the presence of known risk factors for cardiovascular disease.
A simple physical examination was performed comprising weight, height, waist/hip ratio and heart rate. ECG was recorded and blood pressure was measured in the arm by auscultation after 5min of supine rest. Ankle blood pressures in both ankles were measured after inflation of a cuff around the ankle, by detection of either the posterior tibial or the dorsalis-pedis artery during deflation with the ankle cuff pressure read at appearance of the foot pulse. Two measurements of the ankle pressure were performed and the average recorded. The pressure in the foot with the lowest pressure was recorded. The ABI was derived from the ratio of arm/foot systolic pressure. Pulse detection was either by manual palpation of the artery or by Doppler ultrasound. The method for pulse detection was only recorded at the final visit (manual palpation of the artery in 85.5% and by Doppler ultrasound in 14.5%).
Eligible patients entered a run-in period receiving ramipril 2.5mg daily for 7–10 days in a single blind fashion. S-creatinine and s-potassium as well as blood pressure were measured after this period. The patients who tolerated the medication entered a 10–14 day course of placebo and were then randomized using a central telephone system to treatment with either ramipril (target dose 10mg once daily) or placebo and vitamin E (400IU once daily) or placebo. The patients were followed for a mean duration of 4.5 years. The ABI was recorded again after 2 years and at the final visit. Ten thousand five hundred and seventy-six patients entered the run-in phase. 1035 of these were not randomized because of non-adherence, side-effects or withdrawal of consent. Of the 9541 patients randomized, 4645 were assigned to receive ramipril and 4652 were assigned to the placebo group. 244 patients entered a substudy of 2.5mgramipril12and the results in these patients are not included in this paper. Ankle brachial blood pressure was measured at baseline in 8986 patients. The first subject was recruited in December 1993. The study was scheduled to close in November 1999 but due to a clear treatment benefit in the ramipril arm of the study was terminated early (April 1999).
The primary end-point for the HOPE-study as well as for this analysis was the combined end-point of cardiovascular death, non-fatal myocardial infarction or non-fatal stroke. In addition to this the individual components of the composite end-point, as well as all-cause death, hospitalizations for heart failure and diabetic complications were end-points.
2.1 Statistical analysis
Results are presented as means and standard deviations for continuous variables, and frequency and proportion for categorical variables. ABI level was categorized into three groups. An ABI above 0.9 was considered normal, an ABI of 0.9–0.6 as moderately pathologic, and below 0.6 as severely pathologic. Patients without a history of PAD falling into each of the above categories and patients with known PAD were compared with regard to baseline characteristics and clinical outcomes. Trend for each of baseline characteristics with ABI level was examined by fitting a logistic model for a categorical variable, and a linear model for a continuous variable. Kaplan–Meier curves were presented to show unadjusted impact of ABI level on clinical outcomes, and log-rank tests were performed to test the impact. Trend for each of the clinical outcomes with ABI level was tested by fitting a Cox regression model. Multivariate Cox models were further fitted to find adjusted significance of the trend by adding all the baseline variables with a P-value of ≤0.20 in the univariate analysis. The effect of ABI on the clinical outcome was also examined by fitting Cox regression model with ABI value as a continuous variable and clinical PAD was ignored (data are not presented). Stratified analyses were performed to assess whether the impact of ramipril on each of the clinical outcomes varies with the ABI level, which was tested by fitting a Cox model with the second order interaction term. The model was used to estimate the relative risk and the 95% confidence intervals associated with ramipril treatment. In the Cox model analyses, hypothesis tests and 95% confidence intervals were based on Wald test statistics. The assumption of proportional hazard of a variable was assessed by examining the Kaplan–Meier curves or fitting a model with an interaction term between the survival time and a continuous variable. The assumption was satisfied if no crossover of the survival curves exists between categories, or the interaction term is not significant. Since the multiplicity of testing, 0.01 was considered as a significant level.
2.2 Study organization
The study was conducted in 267 centres in 19 countries in North and South America as well as Europe. HOPE was organized, conducted and analysed at the Canadian Cardiovascular Collaboration in Hamilton, Canada. The study was performed in accordance with the Declaration of Helsinki and the local ethics committees approved the protocol. Informed consent wasobtained from all subjects prior to study enrolment.
Baseline characteristics of patients in the respective groups of symptomatic PAD and of no clinical PAD but with different ABI levels. Values are given as numbers and frequencies or as means and standard deviations for the continuous variablesa
One thousand seven hundred and fifteen patients were included in the HOPE-study on the basis of having symptomatic PAD. In the overall study population 3099 patients had a subnormal ABI (≤0.9) and 5887 an ABI more than 0.9. In the total HOPE population the mean systolic blood pressure at the ankle was 119.7mmHg (range 0–283mmHg) and at the arm was 138.1mmHg (range 80–230mmHg).
In the patients with clinical PAD ABI was 0.87±0.22mmHg and in the patients with no clinical PAD it was 1.00±0.18 (P<0.0001). At the last visit in the study 4360 patients had their ABI measured by manual palpation of the pulse and 850 by Doppler measurement. There was no significant difference between these two groups in recorded ABI (manual 0.97±0.22 and Doppler 1.00±0.25; P=0.12).
3.1 Baseline characteristics
The overall baseline characteristics of the HOPE population have been published elsewhere.10The characteristics of patients with PAD and without known PAD but normal, moderately or severely abnormal ABI respectively are given in Table 1. For the group of patients with low ABI, a higher proportion of females was observed as was current smoking, hypertension, diabetes and previous cardiovascular disease. Inversely a lower proportion of patients with previous myocardial infarction was found. A low ABI was further related to higher age, higher systolic arm blood pressure and an increased heart rate. Diastolic blood pressure was the same in all four categories of PAD. Thus a low ABI or presence of clinical PAD was strongly related to the pulse pressure.
3.2 Prediction of events
Clinical PAD as well as a low ABI at baseline in the absence of clinical PAD were strong predictors for cardiovascular morbidity and mortality during the follow-up period (Table 2; Fig. 1a). This was seen both for the primary end-point in the study (cardiovascular mortality, myocardial infarction or stroke) and for each individual component. ABI also predicted all cause mortality (Fig. 1b). In those with a normal ABI and no clinical PAD the mortality was 8.5%, in those with a moderately pathologic ABI, 12.4%, in those with an ABI lower than 0.6, 14.2% and in those with clinical PAD 17.5% (P<0.0001). This association persisted also after adjustment for age, gender, former and current smoking, existence of hypertension, hypercholesterolaemia, diabetes, previous stroke, previous myocardial infarction, and systolic blood pressure in a Cox regression model.
Low ABI also predicted the development of heart failure and diabetic complications. Future vascular procedures (mainly CABG) were not more frequent in the group with a lower ABI but were more common in those with clinically symptomatic PAD. We also performed an analysis with ABI as a continuous variable which did not alter these main results (data not shown).
3.3 Impact of ramipril treatment on clinical outcomes subdivided by ABI or clinical PAD
Ramipril reduced the risk of cardiovascular events in all groups of patients with clinical PAD and with subclinical PAD regardless of ABI (Table 3; Fig. 2). There was no heterogeneity in treatment effects across the various subcategories of individuals with different levels of ABI on any of the outcomes examined indicating the value of ramipril in patients across a range of risk levels (Table 4).
Clinical events in the different ranges of ABI for patients randomized to ramipril and placebo respectively (percentage incidence). Relative risk with ramipril treatment and 95% confidence intervals are given below percentages
No clinical PAD
ABI >0.9 (n=5231)
Clinical PAD (n=1725)
P-value for trend
Primary outcome (cardiovascular
mortality, MI, stroke)
0.83 (0.71, 0.96)
0.72 (0.56, 0.92)
0.77 (0.55, 1.09)
0.75 (0.61, 0.92)
0.83 (0.70, 0.99)
0.81 (0.60, 1.09)
0.73 (0.48, 1.11)
0.75 (0.58, 0.98)
0.72 (0.53, 0.98)
0.44 (0.26, 0.77)
0.99 (0.52, 1.89)
0.72 (0.50, 1.05)
0.83 (0.65, 1.05)
0.62 (0.42, 0.90)
0.76 (0.46, 1.25)
0.75 (0.56, 0.99)
All cause death
0.99 (0.83, 1.20)
0.58 (0.42, 0.79)
0.81 (0.55, 1.19)
0.85 (0.68, 1.07)
0.87 (0.76, 0.99)
0.82 (0.64, 1.07)
0.91 (0.61, 1.37)
0.89 (0.74, 1.09)
0.89 (0.70, 1.13)
0.80 (0.53, 1.21)
0.83 (0.50, 1.39)
0.87 (0.62, 1.21)
Hospitalizations for CHF
1.13 (0.80, 1.60)
0.69 (0.38, 1.23)
0.66 (0.34, 1.28)
0.81 (0.53, 1.24)
P-values are for trend of effect on ramipril on each of the outcomes with PAD category. Primary outcome=cardiovascular mortality, MI, stroke., MI=Myocardial infarction, CHF=Congestive heart failure.
↵a In those with an ABI ≤0.9 and no clinical PAD the results in the primary outcome is RR 0.73 (95% CI=0.60–0.90).
↵b After adjustment for all baseline variables (in Table 1) which are significantly different.
In the HOPE study a low ankle brachial blood pressure index was a strong predictor for cardiovascular morbidity and mortality in patients without clinical peripheral arterial disease. It has previously been shown that the ABI is a predictor for cardiovascular and all cause mortality but in these epidemiological studies the ankle pressure was measured with the help of an ultrasound Doppler device. This is a more precise but more expensive method than simple palpation of the arteries, and is not readily available in most physicians’ offices; but the predictive value of the ABI using palpation required evaluation. In the HOPE-study the majority of the measurements were performed by palpation of the foot arteries, with no difference in ABI between the group measured by manual palpation compared with Doppler, indicating no systematic bias from the differing methods used. Thus, even a relatively crude method of assessing a diminished circulation to the foot has great predictive value and is widely applicable in clinical practice to identify individuals at high risk.
The prevalence of a low ABI in this study may seem high. However, PAD was one of the inclusion criteria in the HOPE-study and thus a selection bias exists. Further, the prevalence of asymptomatic or symptomatic PAD has been reported to be very high (29%) in older patients and patients with other risk factors such as those evaluated in the PARTNERS-study.13Many of our patients with ABI less than 0.6 did not have symptomatic PAD but this may be due to other limiting symptoms from coronary artery disease or previous stroke. Further, McDermott et al. reported that only 33% of adults with ABI indicative of PAD had intermittent claudication.14
4.1 Relationship of ABI to risk factors and to future outcomes
In our study, this simple approach to measuring ABI was related to risk factors for atherosclerosis. An ABI of 0.9 or less has been the usual cut-off level to indicate peripheral vascular disease both in clinical practice and in epidemiological studies.15–19Since an ABI above 0.9 means that the ankle pressure is almost equal to the arm blood pressure, foot pulses are usually easily palpated when ABI is at this level.
A low ABI, as expected, was related to higher age, higher systolic blood pressure, higher heart rate and to higher pulse pressure. There were also more current smokers and patients with diabetes mellitus in the groups with a low ABI, again in accordance with the risk factor patterns for peripheral arterial disease.20,21All these variables are associated with more generalized vascular disease.
We observed a higher rate of CV events in patients with clinical PAD, which is consistent with previous studies.2–6Interestingly, in those without clinical PAD, lower ABI levels (<0.9) increased the risk of future events (from 13.1% to 18.2%) suggesting that even a simple marker of peripheral arterial disease has substantial prognostic value. Given that digital palpation of ankle pulse to record BP is relatively crude, our data are probably underestimating the true relationship between ABI and outcome had we used more sensitive methods such as ultrasound.
4.2 Impact of ramipril
There are two important findings related to the use of ramipril in this report. First, ramipril prevented clinical events in patients with clinical evidence of PAD as well as in those without it. In particular the relative benefit was similar in patients subdivided by levels of ABI (RR 0.83 with ABI >0.9, RR of 0.75 with ABI between 0.6 to 0.9 and RR of 0.79 in those with ABI <0.6, (P for heterogeneity was not significant). However, given that the event rates were higher in those with an ABI <0.9, the absolute benefits are about twice as large in this group (50 per 1000 events prevented) compared to those with an ABI >0.9 (24 per 1000 events prevented). This suggests that in patients with CAD and no clinical evidence of PAD, utilizing the ABI is a simple method to further identify high risk patients who can benefit from preventive strategies.
The primary goal of this study was to evaluate the effects of ramipril in a broad group of high risk patients who were easily identifiable. This goal was clearly achieved. An additional goal was to use simple and commonly available markers that could be utilized with little or no additional costs to further stratify patients into risk categories who may benefit to a greater extent by preventive treatments. Our study was not designed to compare digital versus ultrasound approaches to measuring ABI and our data do not shed light on this issue. Nor do our data examine the intra or inter observation variabilities in obtaining these recordings. However, our approach has ‘face’ validity as ABI (as measured in this study) relates to risk factors for atherosclerosis in the expected manner and also predicts future cardiovascular events. So although we have not formally validated our approach against Doppler studies (the ‘gold’ standard), the relationship to risk factors and predictive value for events suggest that even this simple approach to measure ABI has clinical value. It is possible that utilization of Doppler methods with more precise measurements may have shown even clearer relationships. Nevertheless our data indicate that measuring ankle-arm BP ratio either by the simple approach that we have used (or when available by more precise methods) should be considered in the planning of future large cardiovascular prevention trials and epidemiological studies.
In conclusion, the ABI even if measured with simple palpation of the foot arteries is a strong and independent predictor for future cardiovascular morbidity and mortality as well as for all-cause mortality. Ramipril reduced the risk of clinical outcomes in those with a clinical history of PAD as well as in the patients with subclinical PAD as manifested by a reduced ABI to ≤0.9.
The HOPE-study was funded by the Medical Research Council of Canada (now Canadian Institutes for Health Research), Hoechst-Marion Rousell (now Aventis), AstraZeneca, King Pharmaceuticals, Natural Source Vitamin E Association, Negma and the Heart and Stroke Foundation of Ontario.
LengGC, Lee AJ, Fowkes GR et al. Incidence, natural history and cardiovascular events in symptomatic and asymptomatic peripheral artery disease in the general population. Int J Epidemiol. 1996;25:1172–1179.
KornitzerM, Dramaix M, Sobolski J et al. Ankle/Arm blood pressure index in asymptomatic middle-aged males: An independent predictor of ten-year coronary heart disease mortality. Angiology. 1995;46:211–219.
VogtMT, McKenna M, Wolfson SK et al. The relationship between ankle brachial index, other atherosclerotic disease, diabetes, smoking and mortality in older men and women. Atherosclerosis. 1993;101:191–202.
PapamicaelCM, Lekakis JP, Stamatelopoulos KS et al. Ankle-Brachial index as a predictor of the extent of coronary atherosclerosis and cardiovascular events in patients with coronary artery disease. Am J Cardiol. 2000;86:615–618.
LonnEM, Yusuf S, Dzavik V et al. Effects of ramipril and vitamin E on Atherosclerosis: The Study to Evaluate Carotid Ultrasound Changes in Patients Treated With Ramipril and Vitamin E (SECURE). Circulation. 2001;103(7):919–925.
ZhengZJ, Sharett AR, Chambless LE et al. Association of ankle-brachial index with clinical coronary heart disease, stroke and preclinical carotid and popliteal atherosclerosis: the Atherosclerosis Risk in Communities (ARIC) study. Atherosclerosis. 1997;131:111–125.