- Email: [email protected]
Reactive hyperemia test in a random sample of the general population
Reactive hyperemia test in a random sample of the general population Gillian C. Leng, M B C h B , F. G e r a l d R. Fowkes, F F P H M , Peter T. D o n n a n , MSc, and E d w a r d Housley, F R C P , Edinburgh, United Kingdom
Purpose: The purpose of this article is to determine the performance of a reactive hyperemia test in the general population in terms of validity, increase in case identification, and test refusal; and to identify differences between the two major ways of expressing the results (the postoccclusive ankle index and the percentage change in ankle systolic pressure). Methods: Changes in ankle pressures were measured during reactive hyperemia in 1460 men and women aged 55 to 74 years, who were randomly sampled in the Edinburgh Artery Study. The validity of the test was determined in 91 cases identified by the presence of intermittent claudication and a low resting ankle pressure and in 91 controls matched by age and sex. Results: The mean postocclusive ankle index was 0.98 (SD 0.20), and the mean drop in ankle pressure was 9.89% (SD 12.79%). In examining the validity of the test, differences occurred in the two ways of expressing the results: the postocelusive ankle index was 90% sensitive and 97% specific and the percentage drop in ankle pressure was 52% sensitive and 86% specific in detecting disease. The postocclusive ankle index increased the overall identification of cases in the Edinburgh Artery Study from 6.5% to 11.5%, but the percentage drop in ankle pressure increased identification to only 9.5%. Eight percent of subjects refused the test because of discomfort; this group contained more women, more elderly, and more obese subjects. Conclusions: The reactive hyperemia test is an appropriate test to use in the general population. The results are more accurate when expressed as the postocclusive ankle index rather than the percentage drop in ankle pressure. (J VAsc SURG 1993;17:479-86.)
The estimation o f resting ankle systolic blood pressure is probably the most widely used clinical test in the diagnosis o f peripheral atherosclerosis, but in some patients with substantial disease, ankle pressure falls only during exercise. Thus a treadmill exercise test is often used to enable ankle pressure to be measured after a defined period o f exercise. However, its use may, in some circumstances, be limited in patients with heart disease who are at risk o f developing ischemia or arrhythmias and by those who are physically unable to walk on a treadmill. T o ,~'gm the WolfsonUnit for the Preventionof PeripheralVascular ' Diseases, Department of Public Health Sciences,Universityof Edinburgh and The Royal Infirmary of Edinburgh (Dr. Housley). Supported by the British Heart Foundation and the Wolfson Foundation. Reprint requests: GillianC. Leng, MBChB,WolfsonUnit for the Prevention of Peripheral Vascular Diseases, Department of Public Health Sciences,Universityof Edinburgh, Teviot Place, Edinburgh, United Kingdom EH8 9AG. Copyright © 1993 by The Society for Vascular Surgery and International Societyfor CardiovascularSurgery,North American Chapter. 0741-5214/93/$1.00 + .10 24/1/38454
avoid these problems a reactive hyperemia test was developed in the early 1970s by Johnson 1 that was based on the changes in blood flow produced by obstruction to the circulation described by Dornhorst and Sharpey-Schafer2 in 1951. The reactive hyperemia test involves occlusion o f the arterial flow to the limb by use o f a pneumatic cuff, thereby inflicting stress only on the distal circulation. Reactive hyperemia tests used in clinical practice have been shown to have 89% to 92% sensitivity and 96% specificity in the detection o f angiogrampositive disease 3,4 and to correlate well with exercise testing. 4 However, the test has not been applied in epidemiologic surveys o f peripheral vascular disease, and the results o f this test in the "normal" population are not clearly known. Population studies o f peripheral atherosclerosis have assessed the presence o f vascular disease by a variety o f noninvasive tests, including pulse-wave oscillography, s the ankle/brachial pressure index (ABPI), 6,7 B-mode imaging o f the arterial wall, 8 and treadmill exercise testing. 7 The Lipid Research Clinics study in 19859 did, however, apply a reactive hyperamia test to a population 479
JOURNAL OF VASCULAR SURGERY
480 Leng et al.
sample, but this population was augmented by a hyperlipidemic group. Unfortunately the results were presented only in terms of large- and small-vessel disease, and no specific analysis was made of the hyperemic test. The Edinburgh Artery Study, I° begun in 1988, was the first truly cross-sectional population survey in which a standard reactive hyperemia test had been applied. This study has therefore provided the opporttmity to describe the results of a reactive hyperemia test applied to a "normal" population. The objectives were to determine the following: (1) the distribution of results throughout the population to identify any differences between the two ways of expressing the results; that is, (a) the postocclusive ankle index (i.e., the ankle/brachial systolic pressure ratio during hyperemia) and (b) the percentage change in ankle systolic pressure from the resting to the hyperemic state; (2) the accuracy of the test in identifying a group of patients with peripheral vascular disease and a group of control subjects; (3) the number and characteristics of those subjects refusing to undergo the test; (4) the proportion of cases that could be identified by the reactive hyperemia test, but that would otherwise have been missed.
METHODS Study population. This study was based on the Edinburgh Artery Study, a cross-sectional survey on a random sample of men and women aged 55 to 74 years who were selected from age-sex registers in 10 general practices in the city of Edinburgh (described in detail elsewherea°). Arterial disease was assessed in 1592 participants by means of the World Health Organization (WHO) questionnaire on intermittent claudication, 11measurement of the ABPI at rest, and the change in ankle systolic pressure during reactive hyperemia. Ethics Committee approval was given for this study, and informed consent was obtained from each patient. To determine the accuracy of the reactive hyperemia test, a group of cases and a group of control subjects were selected from the original 1592 participants. The cases included all those subjects with evidence of significant peripheral arterial disease identified by the following criteria: an ABPI _ 0.7; or an ABPI _< 0.9 and a positive history of intermittent claudication on the WHO/Rose questionnaire (n = 91). Ninety-one control subjects were selected at random from a total of approxirnately 800 subjects who had no evidence of either peripheral arterial disease (ABPI _> 0.9 and no intermittent claudication) or heart disease; these were frequency matched
March 1993
to the cases by age and sex. The extent of vascular disease was confirmed with duplex scanning in a subsample of 43 cases (representative of cases in terms of age, sex, and clinical findings). Disease was identified in all 43 subjects on the B-mode image, and this was moderate to severe in 93%; 87% had severe disease as defined by criteria including the image, wave form shape, and increase in velocity at the site of a stenosis. 12 Blood pressure measurements. After 10 minutes of rest in the supine position, the subjects' systolic and diastolic (phase V) blood pressures we,"7 taken in the right arm with a random zero sphygmomanometer. Ankle systolic blood pressures were taken first in the right and then in the left leg with the random zero sphygmomanometer and a Sonicaid Doppler probe (Sonicaid, Chichester, England). Where possible, blood flow was detected at the ankle in the posterior tibial artery. In the reactive hyperemia test that followed, ankle systolic pressure was measured in the right and left legs 15 seconds after the release of a cuff occluding arterial flow above the knee for 4 minutes at about 50 mm Hg above systolic pressure. The timing was standardized by use of an electronic timer. Statistical analysis. Information on the questionnaire and recording form was checked by the clinic staff and entered directly onto a DBASE III (Borland International, Inc., Scotts Valley, Calif,) data base. Error rates in the data entry were assessed by checking intermittent random samples of subjects and by carrying out logic checks. The data files were transferred to the Edinburgh University mainframe computer for analysis by use of the BMDP statistical package (BMDP Statistical Software, Inc., Los Angeles, Calif.). Is The results of the reactive hyperemia test were expressed in two ways: (1) the lower of the left and right ABPI during the hyperemic phase (the postocclusive ankle index) and (2) the higher of the left and right percentage drops in ankle pressure after occlusion. A third method of expressing the results is the absolute drop in ankle pressure, but this was not included because it takes no account of the restit.~" ankle pressure and was therefore considered to be less valid than the percentage drop.
RESULTS Distribution o f results. Fig. 1 shows the distribution of results throughout the whole population for both the postocclusive ankle index and the percentage drop in ankle pressure. The distribution of the postocclusivc ankle index was slightly nega-
JOURNAL OF VASCULAR SURGERY V o l u m e 17, N u m b e r 3
Leng et al.
481
300 250
200 \\ \\ \\ \\ \\ \\ \\
160
100
\\ \\
50
o ....
H,m,..,,.~,,
0.12 0.24 0 . 3 6 0.48 0.60 0.72 0.84. 0.96 1.08
1.20 1.92 1.4-4 1.56 1.68
P o s t - o c c l u s i v e ankle index 600 N u m b e r
\\\ \\\ \\\ \\\ \\\ \\\ \k\ \\\
500
400
0
f
300 \\\ \\\ \\\ \\\ \\\ \\\ \\\ \\\
S U
b
J
e c
t
200
100
8
\\\ \\\ \\\ \\\ \\\ \\\ \\\ \\\
\\\ \\\ \\\ \\\ \\\ \\\ \\\ \\\
\\\ \\\
\\\ \\\
[~
-8 0 8 16 24 32 40 48 56 64 Percentage drop in ankle pressure
-40-32-24-16
72 80
Fig. 1. Postocclusiveankle index and percentage drop in ankle pressure in a population sample of 1460 men and women aged 55 to 74 years.
tively skewed (mean 0.98, SD 0.20), and the percentage drop was slightly positively skewed (mean 9.89%, SD 12.79%). The postocclusive ankle index decreased with age, but the expected equivalent increase in the percentage drop in ankle pressure was ~nuch less marked (Fig. 2); the age trend was significant at the 5 % level for the postocclusive ankle index but was not significant for the percentage drop. Accuracy o f the reactive hyperemia test. Fig. 3 shows the distribution of the postocclusive ankle index and the percentage drop in ankle pressure within the 91 patients with peripheral vascular disease and the 91 controls. Fig. 3 shows quite clearly that a better separation of cases from controls is achieved by use of the postocclusive ankle index.
With a cutoffpoint ___0.8, the postocclusive ankle index was 90% sensitive and 97% specific in detecting disease, whereas a c u t o f point ___20% for the percentage drop in ankle pressure produced a specificity of 86% and a sensitivity of only 52%; these points were selected by receiver operating characteristics curve analysis. Reducing the cutoff point in the percentage drop to >__10% increased sensitivity to 76% but decreased specificity to only 54%. Test refusal Only 1460 (92%) of the subjects recruited to the Edinburgh Artery Study underwent the reactive hyperemia test; the remaining 132 refused the test because they experienced discomfort. This group of 132 was atypical of the population as a whole: it contained significantly more women
JOURNAL
OF VASCULAR
482 Leng et al.
SURGERY
March 1993
P 0 S T 0 C C L U S I V E
1.03
0.98
A N K L E
0.93
I N D E X
0.88
55--59
60-64 6,5-69 AGE (YEARS)
70-74
% D R 0 P I N A N K L E P R E S S U R E
12
11
10
8
7
I
56-69
I
I
60-64 68-69 AGE (YEARS)
I
70-74
Fig. 2. Postocclusive ankle index and percentage drop in ankle pressure by age in a population sample of 1460 men and women aged 55 to 74 years (mean _ 95% confidence interval).
(2 < 0.001) and more elderly and obese subjects, but the resting ankle pressures were not significantly different from pressures in those who did undergo the test (Table I). Additional cases identified by the reactive hyperemia test. In the Edinburgh Artery Study 103 (6.5%) of 1592 subjects had either an ABPI __0.7 or an ABPI _<0.9 plus intermittent claudication, and they were thought likely to have significant peripheral vascular disease. However, several additional subjects had reactive hyperemia test results that suggested a strong possibility of substantial disease: 81 additional subjects had a postocclusive ankle index _<0.65, or a postocclusive ankle index -< 0.8 plus either intermit-
tent claudication or an ABPI _<0.9; whereas only 48 additional subjects had a percentage drop in ankle pressure ___35%, or a percentage drop _>20% plus either claudication or an ABPI _ 0.9. A postocclusive ankle index -<0.8 and percentage drop _>20% wee~ ~ selected by means of receiver operating characteristics curve analysis (see previous mention), and the more severe indicators of disease were chosen by reference to Fig. 3. With these criteria the inclusion of the postocclusive ankle index increased the detection of subjects with an abnormality from 6.5% to 11.5%, and the percentage drop increased detection from 6.5% to 9.5%. Notably, 20 of the 81 subjects identified by the postocclusive ankle index criteria
IOURNAL OF VASCULAR SURGERY Volume 17, Number 3
Leng et al.
t 0.1
483
I
0.25
ii]1 j 0.4.
0.55
0.7
0.85
1.0
1.15
1.3
1.45
66
77
P o s t - o c c l u s i v e ankle index Cases
N
~
Controls
40
u
3o r 2o
o -11
0
11 22 33 44 55 Percentage drop in ankle pressure Cases
~
Controls
Fig. 3. Postocclusive ankle index and percentage drop in ankle pressure in 91 patients with peripheral vascular disease and 91 control subjects. had a normal resting ABPI ( _ 0.9), and only two of these had intermittent dandication. To determine whether the additional cases identified by the reactive hyperemia test were likely to be true-positive cases, the percentages of certain risk factors for peripheral atherosderosis were compared in four distinct groups: (1) subjects with disease as detected by a low ABPI and intermittent claudication; (2) subjects with a low postocclusive ankle index; (3) subjects with a high percentage drop in ankle pressure [in these latter two groups, disease would not have been diagnosed on the basis of the ABPI or intermittent claudication, as in group 1]; (4) subjects who were "normal" with respect to the above criteria (Table II). The percentage of subjects with angina or a previous myocardial infarction in the group with a low postocclusive ankle index was
similar to the percentage in the group with low ABPI and claudication and was significantly different from the percentage in the normal group (p < 0.001). The group with a high percentage drop in ankle pressure also included a high proportion of subjects with a previous heart attack, but the percentages of patients with angina in this group and in the normal group were similar. The percentages of smokers and subjects with diabetes were similar in all diseased groups but were significantly different from the • percentages in the normal group (p < 0.001). The mean total cholesterol levels in both the ABPI plus claudication and the postocclusive ankle index groups were significantly higher than levels in the normal group (p < 0.001), but the mean cholesterol level was not significantly elevated in those with a high percentage drop in ankle pressure.
JOURNAL OF VASCULAR SURGERY
484 Leng et al.
March 1993
Table I. Characteristics of subjects who completed or did not complete the reactive hyperemia test, in a population sample of 1592 men and women aged 55 to 74 years Reactive hyperemia test Completed (687 women, 773 men)
Age (yr) BMI (kg/m 2) Height (m) ABPI ~
Not completed (96 women, 36 men)
Mean
SE
Mean
SE
t Test p value
64.6 25.4 1.66 1.03
0.1 0.1 0.002 0.005
67.4 27.9 1.60 1.00
0.4 0.5 0.007 0.02
< 0.0001 < 0.0001 < 0.0001 0.22~
~Ten subjects with no ABPI result in addition to a missing reactive hyperemia test result. ?Variances not assumed to be equal.
Table II. Risk factors for peripheral vascular disease in subjects with an abnormal reactive hyperemia test, subjects with a diagnosis of intermittent claudication and a low ABPI, and normal subjects Subjects with peripheral atherosclerosis Diagnosed by the hyperemia test
Characteristics
Normal subjects (n = 1288)
Diagnosed by the ABPI and claudication * (n = 103)
Postocclusive ankle index~ (n = 81)
% Drop in ankle pressure~ (n = 48)
Previous heart attack Angina .(WHO questionnaire) Total cholesterol (mmol/L) (mean + SE) Cumulative smoking (fpack yr) (mean + 5E) Known diabetic
7.0% 15.0% 7.0 (0.04) 4.7 (0.1) 1.9%
15.5% 25.0% 7.4 (0.2) 5.5 (0.2) 5.9%
16.0% 22.0% 7.3 (0.2) 5.9 (0.3) 6.3%
16.7% 14.6% 7.1 (0.2) 5.9 (0.3) 6.4%
~ABPI _ 0.9 plus intermittent claudication or ABPI _<7. ~Postocclusive ankle index _<0.65 or postocclusive ankle index < 0.8 plus ABPI _<0.9 or postocclusive ankle index _<0.8 plus claudication. SPercentage drop _>35% or percentage drop ->20% plus ABPI <_0.9 or percentage drop _>20% plus claudication.
DISCUSSION Distribution o f results. Fig. 1 shows the expected skewed distribution of reactive hyperemia test results. A significant increase in the prevalence of peripheral vascular disease with age was detected when the results were expressed as the postocclusive mff,le index, but the trend was much less clear when the results were expressed as the percentage drop in ankle pressure (Fig. 2). In the Edinburgh Artery Study 1° the prevalence of both intermittent claudication and a low ABPI was also found to increase with age, and previous population surveys have similarly demonstrated that the prevalence of disease increases with age.14 It was therefore unexpected that the percentage drop in ankle pressure did not reflect this. The 7.7% of patients who did not undertake the test might have affected the age trend because they tended to be older; however, this should also have affected the trend in the postocclusive ankle index, but it did not.
These results may be open to criticism because brachial pressure measurements were made in only the tight arm; this was done primarily for practical considerations, to restrict the number of observations in each subject, particularly because the prevalence of brachiocephalic disease in the general population was expected to be low. Most other population surveys of cardiovascular disease have also measured brachial pressures in only one arm. 6,15,16A l s o , ankle pressures were measured in only the posterior tibial arter), 'again for the same practical considerations. The posterior tibial artery was selected in preference to the dorsal artery of the foot, which is known to be congenitally absent in 4% to 12% of the general population. 17 Accuracy o f the reactive hyperemia test. To determine the accuracy of any test, some form of gold standard is required, and angiography has traditionally been the test for peripheral vascular disease. Unfortunately angiography is too hazardous to be
JOURNAL OF VASCULARSURGERY Volume 17, Number 3
applied tO a random sample of the general population, and therefore some other noninvasive test must be used. In this study the presence of intermittent claudication plus an abnormal ABPI was used to select the cases and controls; both are well-established measures of peripheral vascular disease. The criteria used were based on the generally accepted premises that an ABPI ---0.9 is suggestive of significant disease6'18 and that an ABPI -<0.7 indicates more severe disease. 19 To increase the probability that the subjects had significant disease, subjects with an • ~BPI _<0.9 but >0.7 were considered true-positive cases only when they also had intermittent claudication. In most of these subjects, subsequent duplex scanning indicated that this definition of disease was reasonable.12
Because the cases and controls were two separate g r o u p s , a b i m o d a l d i s t r i b u t i o n o f reactive h y p e r e m i a
test results would be expected. Fig. 3 clearly illustrates that the postocclusive ankle index separates most cases from controls but that the distinction is much less marked when the percentage drop is used to separate the groups. This difference is reflected in the greater sensitivity of the postocclusive ankle index (90%) compared to the percentage drop in ankle pressure (52%). It could be argued that direct comparison of the postocclusive ankle index with the ABPI is not valid, because few subjects will have a postocclusive ankle '~dex that is higher than the resting ABPI. However, the ABPI was not the only measure of disease but was used in conjunction with the presence of intermittent claudication and with duplex scanning in a representative sample of the cases. Thus there is little doubt that the cases had substantial disease as defined by criteria other than the ABPI and that the gold standard was reasonable, although not perfect. Few studies have attempted to determine the most accurate way of expressing the results of the reactive hyperemia test. In 1982 Ouriel et al. 3 carried out a large hospital-based study of 218 patients (372 limbs) and 25 normal subjects (50 limbs). They compared the results of the reactive ~;yperemia test with results of clinical examination plus arteriography in 33% of the patients. Again, the postocdusive ankle index was found to have greater sensitivity (89%) compared with the percentage drop in ankle pressure (64%), but both had identical specificity (96%). These results were therefore remarkably consistent with results from the Edinburgh Artery Study. An alternative method of performing the reacfve hyperemia test is by measurement of flow velocity changes in the femoral artery during and after the
Leng et al.
485
same maneuver. Femoral artery flow is significantly higher after occlusion in control subjects than in patients with peripheral vascular disease, 2° but different grades of disease cannot be distinguished. 2~Its main disadvantage in an epidemiologic survey is the requirement for a duplex scanner, which is both expensive and cumbersome, to take velocity readings; however, smaller, portable scanners should be available in the future. In assessing the value of a diagnostic test, both reliability (repeatability) and accuracy are important. Few assessments of the reliability of the reactive hyperemia test have been reported, but in one recent study thc postocclusive ankle index was found, with respect to variability, to be a better measure than the drop in ankle pressure. 22 Test refusal. Of the participants in the Edinburgh Artery Study, 80% did not undergo the test because of the associated discomfort. The group who did not take the test was atypical because it contained older, more obese subjects; although there was no significant difference between ABPI in the two groups (Table I). Pain is an unavoidable side effect of tissue ischemia; therefore the duration of occlusion is a compromise between obtaining a significant change in blood flow and causing minimum discomfort to the patient. Dedichen and Myre23 suggest that blood flow increases very little with occlusion times longer than 3 minutes and that the cutoff pressure used to occlude the arterial flow is probably unimportant as long as distal pulses have been obliterated. Therefore, although future studies may be able to use slightly shorter occlusion times, it is likely that subjects with significant peripheral vascular disease will still experience discomfort. Several other studies have used much longer occlusion times, up to 7 minutes, 24 or occlusion pressures 100 mm Hg above systolic pressure.3,24, 25
Additional cases identified by the reactive hyperemia test. With our criteria for defining significant disease, the addition of the reactive hyperemia test to the standard tests for peripheral vascular disease increased the identification of cases from 6.5% to 9.5%, or to 11.5%, depending on whether the results were expressed as the percentage drop in ankle pressure or as the postocclusive ankle index. Unforttmately we were unable to make an additional and independent assessment of whether these were true-positive cases. However, the high prevalence of known risk factors in these subjects suggests that most were likely to have significant peripheral vascular disease (particularly the group identified by the postocclusive ankle index). The high
486
JOURNAL OF VASCULARSURGERY March 1993
Leng et al.
level of serum cholesterol found in those with no evidence of peripheral vascular disease is comparable to levels reported in other Scottish studies. 26 In a study of groups of hospitalized patients and control subjects, Ouriel et al.s concluded that stress tests should be applied only to subjects with symptoms of claudication who had normal resting ankle pressures. In this study, however, 2 0 additional subjects who had a normal resting ABPI were identified by the postocclusive ankle index to have peripheral vascular disease. Only two of these 20 patients had intermittent claudication. Therefore restricting the hyperemic test to only those patients with symptoms and a normal ABPI, as suggested by Ouriel et al., would not have been adequate in these circumstances. This probably reflects the different populations used in the two studies. In conclusion, the reactive hyperemia test would appear to be an accurate test for detecting significant peripheral vascular disease, and it increased the case detection rate in the recent Edinburgh Artery Stud),. Its main limitation was in poor patient participation, but some test refusal must be expected in population surveys, perhaps more so than in clinical practice where patients undergoing diagnosis and treatment may be more amenable to investigation. The postocelusive ankle index would appear to be a better method of expressing the results than the percentage drop in ankle pressure, because it has greater overall sensitivity and specificity, it identifies cases with a higher number of expected risk factors, it reflects the expected worsening of disease with age, and it is a less variable measure. REFERENCES
1. Johnson WC. Doppler ankle pressure and reactive hyperaemia in the diagnosis of arterial insufficiency. J Surg Res 1974;18: 177-80. 2. Dornhorst AC, Sharpey-Shafer EP. Collateral resistance in limbs with arterial obstruction: spontaneous changes and effects of sympathectomy. Clin Sci 1951;10:371-81. 3. Ouriel K, McDonnell A.E, Metz CE, Zarins CK. A critical evaluation of stress testing in the diagnosis of peripheral vascular disease. Surgery 1982;91:686-93. 4. Hummel BW, Hummd BA, Mowbry A, Maixner W, Barnes RW. Reactive hyperaemia versus treadmill exercise testing in arterial disease. Arch Surg 1978;113:95-8. 5. DaSilva A, Widmer LK. Arterienprojekt: Peripher arterieUe Verschusskrankheit. In: Widmer LK, ed. Venen-, Arterien-, Krankheiten, Koronare Herzkrankheit bei Bemfstatigen. Bern: Springer-Verlag, 1981:137-237. 6. Schroll M, Munck O. Estimation of peripheral arteriosclerotic disease by ankle blood pressure measurements in a population study of 60-year old men and women, l Chron Dis 1981;34: 261-9. 7. Gofin R, Kark ID, Friedlander Y, et al. Peripheral vascular disease in a middle-aged population sample: the Jerusalem
Lipid Research Clinic Prevalence Study. Int J Med Sci 1987;23:157-67. 8. The ARIC investigators, The Atherosclerosis Risk in Communities (ARIC) study: design and objectives. Am J Epidemiol 1989;129:687-702. 9. Criqui MH, Fronek A, Barrett-Connor E, et al. The prevalence of peripheral arterial disease in a defined population. Circulation 1985;71:510-5. 10. Fowkes FGR, Housley E, Cawood EHH, MacIntyre CCA, Ruckley CV, Prescott RT. Edinburgh Artery Study: prevalence of asympt0maticand symptomaticperipheral arterial disease in the general population. Int J Epidemiol 1991;20: 384-92. 11. Rose GA. The diagnosis of ischaemic heart pain a ~ intermittent clandication in field surveys. Bull World Health
Organ 1962;27:645-58. 12. Fowkes FGR, Allan PLA, Tsampoulas C, Smith FB, Dorman PT. Validity of duplex scanning in the detection of peripheral arterial disease in the general population. Eur 5 Vase Snrg 1992;6:31-5. 13. Dixon WJ, Brown MB, Engelman L, et al. BMDP statistical software manual. Berkeley: University of California Press, 1988. 14. Fowkes FGR. Epidemiologyofatheroscleroticarterialdisease in the lower limbs. Eur J Vase Surg 1988;2:283-91. 15. Holland WW, Raftery EB, McPearson P, Stone RW. A cardiovascular survey of American East Coast telephone workers. Am J Epidemiol 1967;85:61-71. 16. Reid DD, Brett GZ, Hamilton PJS, Jarret RJ, Keen H, Rose G. Cardiorespiratory disease and diabetes among middle-aged male civil servants. Lancet 1974;1:469-73. 17. Barnhorst DA, Barner HB. Prevalence of congenitally absent pedal pulses. N Engl J Med 1968;278:264-5. 18. Laing S, Greenhalgh RM. The detection and progression of asymptomatic peripheral arterial disease. Br J Surg 1983;7,O; 628-30. 19. Bernstein EF, Fronek A. Current status of non-invasive tests in the diagnosis of peripheral arterial disease. Surg Clin North
Am 1982;62:473-87. 20. Lewis P, Psaila J-V,Morgan Rt-I, Davies WT, Woodcock JP. Common femoral artery volume flow in peripheral vascular disease. Br J Surg 1990;77:183-7. 21. Horrocks M, Scott DJA. Non-invasive tests. In: Fowkes FGR, ed. Epidemiology of peripheral vascular disease. London: Springer-Verlag, 1991:17-27. 22. Fowkes FGR, Housley E, MacIntyre CCA, Prescott RJ, Rucldey CV. Reproducibility of reactive hyperaemia test in the measurement of peripheral vascular disease. Br J Surg 1988;75:743-6. 23. Dedichen H, Myre HO. Reactive hyperaemia of the human lower limb. Acta Clin Scand 1975;141:517-21. 24. Baker JD. Post-stress Doppler ankle pressures. Arch S u ~ 1978;113:1171-3. "~ 25. Verhagen PF, deJong TJ, van Vroomhoven T)'. Ankle pressure changes during reactive hyperaemia in peripheral vascular disease. Vasa 1983;12:29-34. 26. Bolton-Smith C, Woodward M, Smith WCS, Tunstall-Pedoe H. Dietary and non-dietary predictors of serum total and I-IDL-cholesterol in men and women: results from the Scottish Heart Health Study. Int J Epidemiol 1991;20:95104. Submitted Oct. 29, 1991; accepted Match 12, 1992.
Purpose: The purpose of this article is to determine the performance of a reactive hyperemia test in the general population in terms of validity, increase in case identification, and test refusal; and to identify differences between the two major ways of expressing the results (the postoccclusive ankle index and the percentage change in ankle systolic pressure). Methods: Changes in ankle pressures were measured during reactive hyperemia in 1460 men and women aged 55 to 74 years, who were randomly sampled in the Edinburgh Artery Study. The validity of the test was determined in 91 cases identified by the presence of intermittent claudication and a low resting ankle pressure and in 91 controls matched by age and sex. Results: The mean postocclusive ankle index was 0.98 (SD 0.20), and the mean drop in ankle pressure was 9.89% (SD 12.79%). In examining the validity of the test, differences occurred in the two ways of expressing the results: the postocelusive ankle index was 90% sensitive and 97% specific and the percentage drop in ankle pressure was 52% sensitive and 86% specific in detecting disease. The postocclusive ankle index increased the overall identification of cases in the Edinburgh Artery Study from 6.5% to 11.5%, but the percentage drop in ankle pressure increased identification to only 9.5%. Eight percent of subjects refused the test because of discomfort; this group contained more women, more elderly, and more obese subjects. Conclusions: The reactive hyperemia test is an appropriate test to use in the general population. The results are more accurate when expressed as the postocclusive ankle index rather than the percentage drop in ankle pressure. (J VAsc SURG 1993;17:479-86.)
The estimation o f resting ankle systolic blood pressure is probably the most widely used clinical test in the diagnosis o f peripheral atherosclerosis, but in some patients with substantial disease, ankle pressure falls only during exercise. Thus a treadmill exercise test is often used to enable ankle pressure to be measured after a defined period o f exercise. However, its use may, in some circumstances, be limited in patients with heart disease who are at risk o f developing ischemia or arrhythmias and by those who are physically unable to walk on a treadmill. T o ,~'gm the WolfsonUnit for the Preventionof PeripheralVascular ' Diseases, Department of Public Health Sciences,Universityof Edinburgh and The Royal Infirmary of Edinburgh (Dr. Housley). Supported by the British Heart Foundation and the Wolfson Foundation. Reprint requests: GillianC. Leng, MBChB,WolfsonUnit for the Prevention of Peripheral Vascular Diseases, Department of Public Health Sciences,Universityof Edinburgh, Teviot Place, Edinburgh, United Kingdom EH8 9AG. Copyright © 1993 by The Society for Vascular Surgery and International Societyfor CardiovascularSurgery,North American Chapter. 0741-5214/93/$1.00 + .10 24/1/38454
avoid these problems a reactive hyperemia test was developed in the early 1970s by Johnson 1 that was based on the changes in blood flow produced by obstruction to the circulation described by Dornhorst and Sharpey-Schafer2 in 1951. The reactive hyperemia test involves occlusion o f the arterial flow to the limb by use o f a pneumatic cuff, thereby inflicting stress only on the distal circulation. Reactive hyperemia tests used in clinical practice have been shown to have 89% to 92% sensitivity and 96% specificity in the detection o f angiogrampositive disease 3,4 and to correlate well with exercise testing. 4 However, the test has not been applied in epidemiologic surveys o f peripheral vascular disease, and the results o f this test in the "normal" population are not clearly known. Population studies o f peripheral atherosclerosis have assessed the presence o f vascular disease by a variety o f noninvasive tests, including pulse-wave oscillography, s the ankle/brachial pressure index (ABPI), 6,7 B-mode imaging o f the arterial wall, 8 and treadmill exercise testing. 7 The Lipid Research Clinics study in 19859 did, however, apply a reactive hyperamia test to a population 479
JOURNAL OF VASCULAR SURGERY
480 Leng et al.
sample, but this population was augmented by a hyperlipidemic group. Unfortunately the results were presented only in terms of large- and small-vessel disease, and no specific analysis was made of the hyperemic test. The Edinburgh Artery Study, I° begun in 1988, was the first truly cross-sectional population survey in which a standard reactive hyperemia test had been applied. This study has therefore provided the opporttmity to describe the results of a reactive hyperemia test applied to a "normal" population. The objectives were to determine the following: (1) the distribution of results throughout the population to identify any differences between the two ways of expressing the results; that is, (a) the postocclusive ankle index (i.e., the ankle/brachial systolic pressure ratio during hyperemia) and (b) the percentage change in ankle systolic pressure from the resting to the hyperemic state; (2) the accuracy of the test in identifying a group of patients with peripheral vascular disease and a group of control subjects; (3) the number and characteristics of those subjects refusing to undergo the test; (4) the proportion of cases that could be identified by the reactive hyperemia test, but that would otherwise have been missed.
METHODS Study population. This study was based on the Edinburgh Artery Study, a cross-sectional survey on a random sample of men and women aged 55 to 74 years who were selected from age-sex registers in 10 general practices in the city of Edinburgh (described in detail elsewherea°). Arterial disease was assessed in 1592 participants by means of the World Health Organization (WHO) questionnaire on intermittent claudication, 11measurement of the ABPI at rest, and the change in ankle systolic pressure during reactive hyperemia. Ethics Committee approval was given for this study, and informed consent was obtained from each patient. To determine the accuracy of the reactive hyperemia test, a group of cases and a group of control subjects were selected from the original 1592 participants. The cases included all those subjects with evidence of significant peripheral arterial disease identified by the following criteria: an ABPI _ 0.7; or an ABPI _< 0.9 and a positive history of intermittent claudication on the WHO/Rose questionnaire (n = 91). Ninety-one control subjects were selected at random from a total of approxirnately 800 subjects who had no evidence of either peripheral arterial disease (ABPI _> 0.9 and no intermittent claudication) or heart disease; these were frequency matched
March 1993
to the cases by age and sex. The extent of vascular disease was confirmed with duplex scanning in a subsample of 43 cases (representative of cases in terms of age, sex, and clinical findings). Disease was identified in all 43 subjects on the B-mode image, and this was moderate to severe in 93%; 87% had severe disease as defined by criteria including the image, wave form shape, and increase in velocity at the site of a stenosis. 12 Blood pressure measurements. After 10 minutes of rest in the supine position, the subjects' systolic and diastolic (phase V) blood pressures we,"7 taken in the right arm with a random zero sphygmomanometer. Ankle systolic blood pressures were taken first in the right and then in the left leg with the random zero sphygmomanometer and a Sonicaid Doppler probe (Sonicaid, Chichester, England). Where possible, blood flow was detected at the ankle in the posterior tibial artery. In the reactive hyperemia test that followed, ankle systolic pressure was measured in the right and left legs 15 seconds after the release of a cuff occluding arterial flow above the knee for 4 minutes at about 50 mm Hg above systolic pressure. The timing was standardized by use of an electronic timer. Statistical analysis. Information on the questionnaire and recording form was checked by the clinic staff and entered directly onto a DBASE III (Borland International, Inc., Scotts Valley, Calif,) data base. Error rates in the data entry were assessed by checking intermittent random samples of subjects and by carrying out logic checks. The data files were transferred to the Edinburgh University mainframe computer for analysis by use of the BMDP statistical package (BMDP Statistical Software, Inc., Los Angeles, Calif.). Is The results of the reactive hyperemia test were expressed in two ways: (1) the lower of the left and right ABPI during the hyperemic phase (the postocclusive ankle index) and (2) the higher of the left and right percentage drops in ankle pressure after occlusion. A third method of expressing the results is the absolute drop in ankle pressure, but this was not included because it takes no account of the restit.~" ankle pressure and was therefore considered to be less valid than the percentage drop.
RESULTS Distribution o f results. Fig. 1 shows the distribution of results throughout the whole population for both the postocclusive ankle index and the percentage drop in ankle pressure. The distribution of the postocclusivc ankle index was slightly nega-
JOURNAL OF VASCULAR SURGERY V o l u m e 17, N u m b e r 3
Leng et al.
481
300 250
200 \\ \\ \\ \\ \\ \\ \\
160
100
\\ \\
50
o ....
H,m,..,,.~,,
0.12 0.24 0 . 3 6 0.48 0.60 0.72 0.84. 0.96 1.08
1.20 1.92 1.4-4 1.56 1.68
P o s t - o c c l u s i v e ankle index 600 N u m b e r
\\\ \\\ \\\ \\\ \\\ \\\ \k\ \\\
500
400
0
f
300 \\\ \\\ \\\ \\\ \\\ \\\ \\\ \\\
S U
b
J
e c
t
200
100
8
\\\ \\\ \\\ \\\ \\\ \\\ \\\ \\\
\\\ \\\ \\\ \\\ \\\ \\\ \\\ \\\
\\\ \\\
\\\ \\\
[~
-8 0 8 16 24 32 40 48 56 64 Percentage drop in ankle pressure
-40-32-24-16
72 80
Fig. 1. Postocclusiveankle index and percentage drop in ankle pressure in a population sample of 1460 men and women aged 55 to 74 years.
tively skewed (mean 0.98, SD 0.20), and the percentage drop was slightly positively skewed (mean 9.89%, SD 12.79%). The postocclusive ankle index decreased with age, but the expected equivalent increase in the percentage drop in ankle pressure was ~nuch less marked (Fig. 2); the age trend was significant at the 5 % level for the postocclusive ankle index but was not significant for the percentage drop. Accuracy o f the reactive hyperemia test. Fig. 3 shows the distribution of the postocclusive ankle index and the percentage drop in ankle pressure within the 91 patients with peripheral vascular disease and the 91 controls. Fig. 3 shows quite clearly that a better separation of cases from controls is achieved by use of the postocclusive ankle index.
With a cutoffpoint ___0.8, the postocclusive ankle index was 90% sensitive and 97% specific in detecting disease, whereas a c u t o f point ___20% for the percentage drop in ankle pressure produced a specificity of 86% and a sensitivity of only 52%; these points were selected by receiver operating characteristics curve analysis. Reducing the cutoff point in the percentage drop to >__10% increased sensitivity to 76% but decreased specificity to only 54%. Test refusal Only 1460 (92%) of the subjects recruited to the Edinburgh Artery Study underwent the reactive hyperemia test; the remaining 132 refused the test because they experienced discomfort. This group of 132 was atypical of the population as a whole: it contained significantly more women
JOURNAL
OF VASCULAR
482 Leng et al.
SURGERY
March 1993
P 0 S T 0 C C L U S I V E
1.03
0.98
A N K L E
0.93
I N D E X
0.88
55--59
60-64 6,5-69 AGE (YEARS)
70-74
% D R 0 P I N A N K L E P R E S S U R E
12
11
10
8
7
I
56-69
I
I
60-64 68-69 AGE (YEARS)
I
70-74
Fig. 2. Postocclusive ankle index and percentage drop in ankle pressure by age in a population sample of 1460 men and women aged 55 to 74 years (mean _ 95% confidence interval).
(2 < 0.001) and more elderly and obese subjects, but the resting ankle pressures were not significantly different from pressures in those who did undergo the test (Table I). Additional cases identified by the reactive hyperemia test. In the Edinburgh Artery Study 103 (6.5%) of 1592 subjects had either an ABPI __0.7 or an ABPI _<0.9 plus intermittent claudication, and they were thought likely to have significant peripheral vascular disease. However, several additional subjects had reactive hyperemia test results that suggested a strong possibility of substantial disease: 81 additional subjects had a postocclusive ankle index _<0.65, or a postocclusive ankle index -< 0.8 plus either intermit-
tent claudication or an ABPI _<0.9; whereas only 48 additional subjects had a percentage drop in ankle pressure ___35%, or a percentage drop _>20% plus either claudication or an ABPI _ 0.9. A postocclusive ankle index -<0.8 and percentage drop _>20% wee~ ~ selected by means of receiver operating characteristics curve analysis (see previous mention), and the more severe indicators of disease were chosen by reference to Fig. 3. With these criteria the inclusion of the postocclusive ankle index increased the detection of subjects with an abnormality from 6.5% to 11.5%, and the percentage drop increased detection from 6.5% to 9.5%. Notably, 20 of the 81 subjects identified by the postocclusive ankle index criteria
IOURNAL OF VASCULAR SURGERY Volume 17, Number 3
Leng et al.
t 0.1
483
I
0.25
ii]1 j 0.4.
0.55
0.7
0.85
1.0
1.15
1.3
1.45
66
77
P o s t - o c c l u s i v e ankle index Cases
N
~
Controls
40
u
3o r 2o
o -11
0
11 22 33 44 55 Percentage drop in ankle pressure Cases
~
Controls
Fig. 3. Postocclusive ankle index and percentage drop in ankle pressure in 91 patients with peripheral vascular disease and 91 control subjects. had a normal resting ABPI ( _ 0.9), and only two of these had intermittent dandication. To determine whether the additional cases identified by the reactive hyperemia test were likely to be true-positive cases, the percentages of certain risk factors for peripheral atherosderosis were compared in four distinct groups: (1) subjects with disease as detected by a low ABPI and intermittent claudication; (2) subjects with a low postocclusive ankle index; (3) subjects with a high percentage drop in ankle pressure [in these latter two groups, disease would not have been diagnosed on the basis of the ABPI or intermittent claudication, as in group 1]; (4) subjects who were "normal" with respect to the above criteria (Table II). The percentage of subjects with angina or a previous myocardial infarction in the group with a low postocclusive ankle index was
similar to the percentage in the group with low ABPI and claudication and was significantly different from the percentage in the normal group (p < 0.001). The group with a high percentage drop in ankle pressure also included a high proportion of subjects with a previous heart attack, but the percentages of patients with angina in this group and in the normal group were similar. The percentages of smokers and subjects with diabetes were similar in all diseased groups but were significantly different from the • percentages in the normal group (p < 0.001). The mean total cholesterol levels in both the ABPI plus claudication and the postocclusive ankle index groups were significantly higher than levels in the normal group (p < 0.001), but the mean cholesterol level was not significantly elevated in those with a high percentage drop in ankle pressure.
JOURNAL OF VASCULAR SURGERY
484 Leng et al.
March 1993
Table I. Characteristics of subjects who completed or did not complete the reactive hyperemia test, in a population sample of 1592 men and women aged 55 to 74 years Reactive hyperemia test Completed (687 women, 773 men)
Age (yr) BMI (kg/m 2) Height (m) ABPI ~
Not completed (96 women, 36 men)
Mean
SE
Mean
SE
t Test p value
64.6 25.4 1.66 1.03
0.1 0.1 0.002 0.005
67.4 27.9 1.60 1.00
0.4 0.5 0.007 0.02
< 0.0001 < 0.0001 < 0.0001 0.22~
~Ten subjects with no ABPI result in addition to a missing reactive hyperemia test result. ?Variances not assumed to be equal.
Table II. Risk factors for peripheral vascular disease in subjects with an abnormal reactive hyperemia test, subjects with a diagnosis of intermittent claudication and a low ABPI, and normal subjects Subjects with peripheral atherosclerosis Diagnosed by the hyperemia test
Characteristics
Normal subjects (n = 1288)
Diagnosed by the ABPI and claudication * (n = 103)
Postocclusive ankle index~ (n = 81)
% Drop in ankle pressure~ (n = 48)
Previous heart attack Angina .(WHO questionnaire) Total cholesterol (mmol/L) (mean + SE) Cumulative smoking (fpack yr) (mean + 5E) Known diabetic
7.0% 15.0% 7.0 (0.04) 4.7 (0.1) 1.9%
15.5% 25.0% 7.4 (0.2) 5.5 (0.2) 5.9%
16.0% 22.0% 7.3 (0.2) 5.9 (0.3) 6.3%
16.7% 14.6% 7.1 (0.2) 5.9 (0.3) 6.4%
~ABPI _ 0.9 plus intermittent claudication or ABPI _<7. ~Postocclusive ankle index _<0.65 or postocclusive ankle index < 0.8 plus ABPI _<0.9 or postocclusive ankle index _<0.8 plus claudication. SPercentage drop _>35% or percentage drop ->20% plus ABPI <_0.9 or percentage drop _>20% plus claudication.
DISCUSSION Distribution o f results. Fig. 1 shows the expected skewed distribution of reactive hyperemia test results. A significant increase in the prevalence of peripheral vascular disease with age was detected when the results were expressed as the postocclusive mff,le index, but the trend was much less clear when the results were expressed as the percentage drop in ankle pressure (Fig. 2). In the Edinburgh Artery Study 1° the prevalence of both intermittent claudication and a low ABPI was also found to increase with age, and previous population surveys have similarly demonstrated that the prevalence of disease increases with age.14 It was therefore unexpected that the percentage drop in ankle pressure did not reflect this. The 7.7% of patients who did not undertake the test might have affected the age trend because they tended to be older; however, this should also have affected the trend in the postocclusive ankle index, but it did not.
These results may be open to criticism because brachial pressure measurements were made in only the tight arm; this was done primarily for practical considerations, to restrict the number of observations in each subject, particularly because the prevalence of brachiocephalic disease in the general population was expected to be low. Most other population surveys of cardiovascular disease have also measured brachial pressures in only one arm. 6,15,16A l s o , ankle pressures were measured in only the posterior tibial arter), 'again for the same practical considerations. The posterior tibial artery was selected in preference to the dorsal artery of the foot, which is known to be congenitally absent in 4% to 12% of the general population. 17 Accuracy o f the reactive hyperemia test. To determine the accuracy of any test, some form of gold standard is required, and angiography has traditionally been the test for peripheral vascular disease. Unfortunately angiography is too hazardous to be
JOURNAL OF VASCULARSURGERY Volume 17, Number 3
applied tO a random sample of the general population, and therefore some other noninvasive test must be used. In this study the presence of intermittent claudication plus an abnormal ABPI was used to select the cases and controls; both are well-established measures of peripheral vascular disease. The criteria used were based on the generally accepted premises that an ABPI ---0.9 is suggestive of significant disease6'18 and that an ABPI -<0.7 indicates more severe disease. 19 To increase the probability that the subjects had significant disease, subjects with an • ~BPI _<0.9 but >0.7 were considered true-positive cases only when they also had intermittent claudication. In most of these subjects, subsequent duplex scanning indicated that this definition of disease was reasonable.12
Because the cases and controls were two separate g r o u p s , a b i m o d a l d i s t r i b u t i o n o f reactive h y p e r e m i a
test results would be expected. Fig. 3 clearly illustrates that the postocclusive ankle index separates most cases from controls but that the distinction is much less marked when the percentage drop is used to separate the groups. This difference is reflected in the greater sensitivity of the postocclusive ankle index (90%) compared to the percentage drop in ankle pressure (52%). It could be argued that direct comparison of the postocclusive ankle index with the ABPI is not valid, because few subjects will have a postocclusive ankle '~dex that is higher than the resting ABPI. However, the ABPI was not the only measure of disease but was used in conjunction with the presence of intermittent claudication and with duplex scanning in a representative sample of the cases. Thus there is little doubt that the cases had substantial disease as defined by criteria other than the ABPI and that the gold standard was reasonable, although not perfect. Few studies have attempted to determine the most accurate way of expressing the results of the reactive hyperemia test. In 1982 Ouriel et al. 3 carried out a large hospital-based study of 218 patients (372 limbs) and 25 normal subjects (50 limbs). They compared the results of the reactive ~;yperemia test with results of clinical examination plus arteriography in 33% of the patients. Again, the postocdusive ankle index was found to have greater sensitivity (89%) compared with the percentage drop in ankle pressure (64%), but both had identical specificity (96%). These results were therefore remarkably consistent with results from the Edinburgh Artery Study. An alternative method of performing the reacfve hyperemia test is by measurement of flow velocity changes in the femoral artery during and after the
Leng et al.
485
same maneuver. Femoral artery flow is significantly higher after occlusion in control subjects than in patients with peripheral vascular disease, 2° but different grades of disease cannot be distinguished. 2~Its main disadvantage in an epidemiologic survey is the requirement for a duplex scanner, which is both expensive and cumbersome, to take velocity readings; however, smaller, portable scanners should be available in the future. In assessing the value of a diagnostic test, both reliability (repeatability) and accuracy are important. Few assessments of the reliability of the reactive hyperemia test have been reported, but in one recent study thc postocclusive ankle index was found, with respect to variability, to be a better measure than the drop in ankle pressure. 22 Test refusal. Of the participants in the Edinburgh Artery Study, 80% did not undergo the test because of the associated discomfort. The group who did not take the test was atypical because it contained older, more obese subjects; although there was no significant difference between ABPI in the two groups (Table I). Pain is an unavoidable side effect of tissue ischemia; therefore the duration of occlusion is a compromise between obtaining a significant change in blood flow and causing minimum discomfort to the patient. Dedichen and Myre23 suggest that blood flow increases very little with occlusion times longer than 3 minutes and that the cutoff pressure used to occlude the arterial flow is probably unimportant as long as distal pulses have been obliterated. Therefore, although future studies may be able to use slightly shorter occlusion times, it is likely that subjects with significant peripheral vascular disease will still experience discomfort. Several other studies have used much longer occlusion times, up to 7 minutes, 24 or occlusion pressures 100 mm Hg above systolic pressure.3,24, 25
Additional cases identified by the reactive hyperemia test. With our criteria for defining significant disease, the addition of the reactive hyperemia test to the standard tests for peripheral vascular disease increased the identification of cases from 6.5% to 9.5%, or to 11.5%, depending on whether the results were expressed as the percentage drop in ankle pressure or as the postocclusive ankle index. Unforttmately we were unable to make an additional and independent assessment of whether these were true-positive cases. However, the high prevalence of known risk factors in these subjects suggests that most were likely to have significant peripheral vascular disease (particularly the group identified by the postocclusive ankle index). The high
486
JOURNAL OF VASCULARSURGERY March 1993
Leng et al.
level of serum cholesterol found in those with no evidence of peripheral vascular disease is comparable to levels reported in other Scottish studies. 26 In a study of groups of hospitalized patients and control subjects, Ouriel et al.s concluded that stress tests should be applied only to subjects with symptoms of claudication who had normal resting ankle pressures. In this study, however, 2 0 additional subjects who had a normal resting ABPI were identified by the postocclusive ankle index to have peripheral vascular disease. Only two of these 20 patients had intermittent claudication. Therefore restricting the hyperemic test to only those patients with symptoms and a normal ABPI, as suggested by Ouriel et al., would not have been adequate in these circumstances. This probably reflects the different populations used in the two studies. In conclusion, the reactive hyperemia test would appear to be an accurate test for detecting significant peripheral vascular disease, and it increased the case detection rate in the recent Edinburgh Artery Stud),. Its main limitation was in poor patient participation, but some test refusal must be expected in population surveys, perhaps more so than in clinical practice where patients undergoing diagnosis and treatment may be more amenable to investigation. The postocelusive ankle index would appear to be a better method of expressing the results than the percentage drop in ankle pressure, because it has greater overall sensitivity and specificity, it identifies cases with a higher number of expected risk factors, it reflects the expected worsening of disease with age, and it is a less variable measure. REFERENCES
1. Johnson WC. Doppler ankle pressure and reactive hyperaemia in the diagnosis of arterial insufficiency. J Surg Res 1974;18: 177-80. 2. Dornhorst AC, Sharpey-Shafer EP. Collateral resistance in limbs with arterial obstruction: spontaneous changes and effects of sympathectomy. Clin Sci 1951;10:371-81. 3. Ouriel K, McDonnell A.E, Metz CE, Zarins CK. A critical evaluation of stress testing in the diagnosis of peripheral vascular disease. Surgery 1982;91:686-93. 4. Hummel BW, Hummd BA, Mowbry A, Maixner W, Barnes RW. Reactive hyperaemia versus treadmill exercise testing in arterial disease. Arch Surg 1978;113:95-8. 5. DaSilva A, Widmer LK. Arterienprojekt: Peripher arterieUe Verschusskrankheit. In: Widmer LK, ed. Venen-, Arterien-, Krankheiten, Koronare Herzkrankheit bei Bemfstatigen. Bern: Springer-Verlag, 1981:137-237. 6. Schroll M, Munck O. Estimation of peripheral arteriosclerotic disease by ankle blood pressure measurements in a population study of 60-year old men and women, l Chron Dis 1981;34: 261-9. 7. Gofin R, Kark ID, Friedlander Y, et al. Peripheral vascular disease in a middle-aged population sample: the Jerusalem
Lipid Research Clinic Prevalence Study. Int J Med Sci 1987;23:157-67. 8. The ARIC investigators, The Atherosclerosis Risk in Communities (ARIC) study: design and objectives. Am J Epidemiol 1989;129:687-702. 9. Criqui MH, Fronek A, Barrett-Connor E, et al. The prevalence of peripheral arterial disease in a defined population. Circulation 1985;71:510-5. 10. Fowkes FGR, Housley E, Cawood EHH, MacIntyre CCA, Ruckley CV, Prescott RT. Edinburgh Artery Study: prevalence of asympt0maticand symptomaticperipheral arterial disease in the general population. Int J Epidemiol 1991;20: 384-92. 11. Rose GA. The diagnosis of ischaemic heart pain a ~ intermittent clandication in field surveys. Bull World Health
Organ 1962;27:645-58. 12. Fowkes FGR, Allan PLA, Tsampoulas C, Smith FB, Dorman PT. Validity of duplex scanning in the detection of peripheral arterial disease in the general population. Eur 5 Vase Snrg 1992;6:31-5. 13. Dixon WJ, Brown MB, Engelman L, et al. BMDP statistical software manual. Berkeley: University of California Press, 1988. 14. Fowkes FGR. Epidemiologyofatheroscleroticarterialdisease in the lower limbs. Eur J Vase Surg 1988;2:283-91. 15. Holland WW, Raftery EB, McPearson P, Stone RW. A cardiovascular survey of American East Coast telephone workers. Am J Epidemiol 1967;85:61-71. 16. Reid DD, Brett GZ, Hamilton PJS, Jarret RJ, Keen H, Rose G. Cardiorespiratory disease and diabetes among middle-aged male civil servants. Lancet 1974;1:469-73. 17. Barnhorst DA, Barner HB. Prevalence of congenitally absent pedal pulses. N Engl J Med 1968;278:264-5. 18. Laing S, Greenhalgh RM. The detection and progression of asymptomatic peripheral arterial disease. Br J Surg 1983;7,O; 628-30. 19. Bernstein EF, Fronek A. Current status of non-invasive tests in the diagnosis of peripheral arterial disease. Surg Clin North
Am 1982;62:473-87. 20. Lewis P, Psaila J-V,Morgan Rt-I, Davies WT, Woodcock JP. Common femoral artery volume flow in peripheral vascular disease. Br J Surg 1990;77:183-7. 21. Horrocks M, Scott DJA. Non-invasive tests. In: Fowkes FGR, ed. Epidemiology of peripheral vascular disease. London: Springer-Verlag, 1991:17-27. 22. Fowkes FGR, Housley E, MacIntyre CCA, Prescott RJ, Rucldey CV. Reproducibility of reactive hyperaemia test in the measurement of peripheral vascular disease. Br J Surg 1988;75:743-6. 23. Dedichen H, Myre HO. Reactive hyperaemia of the human lower limb. Acta Clin Scand 1975;141:517-21. 24. Baker JD. Post-stress Doppler ankle pressures. Arch S u ~ 1978;113:1171-3. "~ 25. Verhagen PF, deJong TJ, van Vroomhoven T)'. Ankle pressure changes during reactive hyperaemia in peripheral vascular disease. Vasa 1983;12:29-34. 26. Bolton-Smith C, Woodward M, Smith WCS, Tunstall-Pedoe H. Dietary and non-dietary predictors of serum total and I-IDL-cholesterol in men and women: results from the Scottish Heart Health Study. Int J Epidemiol 1991;20:95104. Submitted Oct. 29, 1991; accepted Match 12, 1992.