Archives of Cardiovascular Disease (2009) 102, 519—524

CLINICAL RESEARCH

Accuracy of ankle-brachial index using an automaticblood pressure device to detect peripheral arterydisease in preventive medicine

Détection de l’artériopathie des membres inférieurs en médecine préventivepar la détermination de l’index de pression systolique à l’aide d’untensiomètre automatique

Daniel Benchimola, Xavier Pillois a, Alain Benchimolb,André Houittec, Pierre Sagardiluzd, Luc Tortelierc,Jacques Bonneta,∗

a Inserm U828, université Victor-Segalen de Bordeaux 2, hôpital Cardiologique,avenue du Haut-Lévêque, 33604 Pessac, Franceb 22, rue Daniel, Lormont, Francec Service AIMT, Rennes, Franced Les portes de Caudéran, Bordeaux, France

Received 17 October 2008; received in revised form 27 March 2009; accepted 30 March 2009Available online 21 June 2009

KEYWORDSAnkle brachial index;Peripheral arterydisease;Automatic bloodpressure device;Preventive medicine

SummaryBackground. — Previously, we validated determination of ankle-brachial index using an auto-matic blood pressure device.Aim. — To test the feasibility and accuracy of the automatic method in assessing pathologicalankle-brachial indexes in routine preventive examinations.Methods. — Two physicians enrolled 354 subjects (74% men) randomly for automatic ankle-brachial index measurements using an OMRON HM 722 device. Ankle-brachial index wascalculated by dividing the highest value obtained at each ankle by the highest arm value. Eachsubject with an abnormal (less than 0.90) automatic index and the six subsequent subjectsunderwent Doppler index determination.

Results. — Automatic ankle-brachial index determination was possible in both ankles in 350 sub-jects (99%; mean time 8.1 ± 2.1 minutes). The incidence of abnormal automatic ankle-brachialindex was 8% (n = 28). Correlations between the automatic and Doppler methods were goodin left and right legs (r = 0.84 and 0.78, respectively; p < 0.001). In subjects with an abnormal

∗ Corresponding author. Fax: +33 5 56 36 89 79.E-mail address: jacques.bonnet@inserm.fr (J. Bonnet).

1875-2136/$ — see front matter © 2009 Elsevier Masson SAS. All rights reserved.doi:10.1016/j.acvd.2009.03.011

520

MOTS CLÉSIndex de pressionssystoliques ;Artériopathie desmembres inférieurs ;Tensiomètreautomatique ;Médecine préventive

et de concordance (� = 0,87) sont obtenus chez les sujets ayant un index pathologique.Conclusions. — L’utilisation d’un tensiomètre automatique pour dépister les artériopathies des

xame. Tou

A

AP

B

Poatfocuas

neapa

Pa2[aiwmcoTtiw

M

S

membres inférieurs lors d’e© 2009 Elsevier Masson SAS

bbreviations

BI ankle-brachial indexAD peripheral artery disease

ackground

AD occurs frequently in the general population [1]. Likether atherosclerotic diseases, the risk of PAD increases withge, with the presence of classical cardiovascular risk fac-ors and with a Western lifestyle. Detection of PAD is usefulor the assessment of local ischaemic symptoms but it is alsof major value in evaluating the risk of total mortality andardiovascular morbidity and mortality in the general pop-lation [2—5]. The diagnosis of PAD identifies patients withhigh risk of cardiovascular events or death, and leads to

pecific follow-up and therapeutic interventions.The detection of PAD is often based on clinical exami-

ation — intermittent claudication being the most frequentxercise-induced ischaemic symptom [6—8] — together withlteration of peripheral pulse, especially posterior tibialulse [9]. However, the most valuable simple index is thenkle-arm index or the ABI of systolic blood pressure [9].

Or5u

ns de médecine préventive apparaît faisable et rapide.s droits réservés.

revalence of PAD is usually estimated at 2 to 3% whenssessed on the basis of claudication and at up to 10 to0% when assessed by ABI in adults [10]. In a previous study11], we validated the use of a simple, commercially avail-ble, automatic blood pressure device for determining ABIn patients consulting in a cardiology department, comparedith intra-arterial measurements and the standard Dopplerethod. As mentioned by Simon et al. [12], it can be diffi-

ult to implement ABI measurement by general practitionersr preventive medicine physicians in routine examinations.his study was performed to assess the automatic method inhe detection of pathological indexes in subjects undergo-ng their routine annual preventive medical examination atork, in terms of feasibility, time constraint and accuracy.

ethods

tudy protocol

D. Benchimol et al.

automatic index, correlations with Doppler indexes were good in both legs (r = 0.67, p < 0.001).In terms of detecting an abnormal index in a routine preventive examination, the automaticmethod had good sensitivity (92%), specificity (98%), positive predictive value (86%), negativepredictive value (99%) and accuracy (97% compared with the Doppler method). Good resultswere obtained in subjects with an abnormal index in terms of agreement and concordance withthe Doppler method (� = 0.87).Conclusions. — The use of a commercially available automatic blood pressure device to detectperipheral artery disease appears feasible and quick in routine medical examinations.© 2009 Elsevier Masson SAS. All rights reserved.

RésuméContexte. — Nous avons précédemment validé le calcul des index de pressions systoliquesutilisant un tensiomètre automatique chez des patients.Objectifs. — L’étude teste la faisabilité et l’exactitude de cette méthode pour la détectiond’indexes pathologiques lors d’examens systématiques de médecine préventive.Méthodes. — Deux médecins du travail ont inclus 354 sujets, pour la mesure de leur index àl’aide d’un appareil OMRON HM 722 lors de leur visite annuelle. L’index automatisé est calculéen divisant la plus forte valeur obtenue à chaque cheville par la plus forte valeur obtenue àl’un des bras. Les sujets ayant un index automatique pathologique et les six sujets suivants ontune détermination de leur index selon la méthode Doppler.Résultats. — L’index automatisé est mesurable aux deux membres inférieurs chez 350 sujets(99 %) en 8,1 ± 2,1 minutes. Un index pathologique (inférieur à 0,90) est noté dans 8 % des cas(n = 28). Les sujets ayant un index pathologique inférieur à 0,90 et les six sujets suivants ont euune determination de l’index par la méthode Doppler. Les correlations entre les deux méthodessont bonnes à gauche (r = 0,84 ; p < 0,001) et à droite (r = 0,78 ; p < 0,001). Les corrélations desindexes pathologiques obtenus par la méthode automatique avec les index Doppler sont bonnes(r = 0,67 ; p < 0,001) aux deux membres inférieurs. En termes de diagnostic d’une artérite lorsd’un examen systématique de médecine préventive, la méthode automatique a une bonnesensibilité : 92 %, spécificité : 98 %, valeur prédictive positive : 86 %, valeur prédictive négative :99 % et exactitude : 97 % par rapport à la méthode Doppler. De bons résultats en termes d’accord

ver a 1-year period, two physicians of preventive medicineecruited one in every 15 referrals from a total cohort of402 subjects attending for an annual systematic check-p. Subjects aged 40 to 60 years who gave their informed

odstibt

R

S

A5ws6a(cp(sceb

A

Twpsis

1Piba

Ca

M(tiof

ftTs

ABI using automatic blood pressure device

consent for the study were included; those with severe PAD(severe claudication or previous arterial procedure) wereexcluded.

Each subject first underwent a clinical examination. Clin-ical history was analysed in terms of previous cardiovasculardiseases (such as hypertension, PAD, coronary disease, cere-brovascular disease, angioplasty or arterial surgery) andknown diseases or classical risk factors predisposing to PAD(diabetes, hyperlipidaemia and smoking). Clinical symp-toms (essentially intermittent claudication) were analysedaccording to the Rose questionnaire [13]. Physical examina-tion detailed pulses in legs, especially posterior tibial pulse.

Determination of ankle-brachial index

In subjects resting in the supine position for 10 minutes,the ABI was determined by the automatic method using anOMRON HM 722 device (Omron Matsusaka Co. Ltd., Japan) ateach limb and in a random order. The blood pressure cuff ofthe automatic instrument was placed just above the anklesand on the upper arms, thus measuring blood pressure auto-matically. A first measurement at any site served only toassess the systolic blood pressure level and reduce any stresseffect, and was excluded from the calculation. ABI was cal-culated by dividing the highest value obtained at each ankleby the highest of the arm values. An index less than 0.90 wasregarded as being pathological. Patients in whom systolicblood pressure could not be measured with the automaticdevice at one ankle but who had measurements in both armswere also classified as having a pathological index.

To facilitate comparison of the automatic and Dopplermethods, systolic blood pressures were determined at eacharm and ankle by the standard Doppler method in all sub-jects with a pathological automatic index and in a controlgroup comprising the six subjects who followed each subjectwith a pathological index, who fulfilled the inclusion crite-ria but had a normal automatic ABI in both lower limbs. This6:1 ratio was chosen to ensure that a very large group ofsubjects with normal automatic ABI also had determinationof Doppler ABI, to guarantee a good assessment of negativepredictive value, taking into consideration the number ofpatients seen in a day by a physician and the time neededfor both determination methods.

The brachial and posterior tibial systolic pressures weremeasured using appropriately-sized blood pressure cuffslinked to a mercury sphygmomanometer placed successivelyon the upper arms and just above the ankles. Using a hand-held, continuous-wave Doppler probe (8 MHz Microdop,SonoMed, Versailles, France), the systolic pressure in eachartery was measured when the flow resumed after gradualand slow cuff deflation. Doppler ABI was calculated in thesame way as automatic ABI and the index was classified inthe same manner as normal or abnormal.

Statistical analysis

The mean values for systolic blood pressure and ABI mea-

surements were compared using a paired Student’s t test.The series of values of systolic blood pressure and calculatedindexes were correlated using linear regression. Incidencedifferences were tested with the �2 test. The inter-method concordance on the binary determination (presence

iVnaa

521

r absence) of a pathological index (less than 0.9) wasetermined by the kappa coefficient (�). The sensitivity,pecificity, and positive and negative predictive values ofhe new calculated indexes were determined in compar-son with abnormal standard Doppler indexes. Agreementetween the two methods was tested by the Bland & Altmanest.

esults

ubject characteristics

total of 354 subjects were selected: mean age0.5 ± 6 years (range 40 to 60); 261 (74%) men; 93 (26%)omen. Of these, 164 (46%) subjects were smokers or former

mokers, 87 (25%) had hypertension, 20 (6%) had diabetes,3 (18%) had hypercholesterolaemia, 13 (4%) had coronaryrtery disease (eight had myocardial infarction) and eight2%) had PAD. Nine patients complained of an intermittentlaudication (two with known PAD and seven without). Tibialosterior pulse was diminished or abolished in four patientsthe two with known PAD and claudication and two of theeven with claudication but without known PAD). The meanholesterol level was 5.69 ± 1.03 mmol/L, the mean triglyc-ride concentration was 1.43 ± 0.99 mmol/L and the meanody mass index was 25.8 ± 3.5.

utomatic ankle-brachial index measurements

he mean time for measurements with the automatic deviceas 8.1 ± 2.1 minutes. Automatic determination of bloodressure at the ankle was possible in both legs in 350 (99%)ubjects and not possible in four (1%) subjects (three timesn both ankles and once in only the left ankle). None of theubjects complained of any side effects.

The mean calculated ABI for the 354 subjects was.06 ± 0.12 for the right leg and 1.06 ± 0.13 for the left leg.athological automatic ABI was detected in 28 (8%) subjects,ncluding the four subjects with a non-measurable automaticlood pressure at one ankle, who were therefore classifieds having an abnormal index.

orrelations and agreements between Dopplernd automatic methods

ean ABI was measured by the two methods in 196 subjectsall those with a pathological automatic index [n = 28] andhe control group who had a normal index [n = 168]). Thencidences of different values of automatic and Doppler ABIf clinical relevance are shown in Fig. 1; no significant dif-erences were observed for ABI less than 1.

The correlation between the two methods was very goodor systolic blood pressure in the arms: r = 0.95, p < 0.001 inhe right arms and r = 0.96, p < 0.001 in the left arms (Fig. 2).he correlation between the two methods was very good forystolic blood pressure in the lower limbs: r = 0.91, p < 0.001

n the right legs and r = 0.93, p < 0.001 in the left legs (Fig. 3).ery low systolic blood pressure less than 80 mmHg couldot be measured by the automatic blood pressure device,s noted during the previous validation study versus intra-rterial measurements [11]. The correlation between the

522 D. Benchimol et al.

Figure 1. Incidences of clinical relevance by ABI values.

Figure 2. Correlation between Doppler and automatic methodsin terms of systolic blood pressures in arms.

Figure 3. Correlation between Doppler and automatic methodsin terms of systolic blood pressures in lower limbs.

Fi

tlvtFe(

ctiwFlm

S

Ambbm

Fa

igure 4. Correlation between Doppler and automatic methodsn terms of ankle-brachial indexes.

wo methods was good for ABI: r = 0.85, p < 0.001 in theeft legs and r = 0.81, p < 0.001 in the right legs (Fig. 4) Thealue-to-value comparison assessing the agreement of thewo methods for all 196 patients in this subgroup is shown inig. 5; the comparison revealed several discrepancies thatxplain the significant statistical difference in ABI valuesp < 0.001; two-tailed paired t test).

In the 28 patients with a pathological ABI, no statisti-al differences could be observed in ABI values (p = 0.40;wo-tailed paired t test). Correlations between patholog-cal automatic indexes and pathological Doppler indexesere good (r = 0.68, p < 0.001) in left and right legs (Fig. 6).urthermore, the value-to-value comparison in this patho-ogical subgroup showed good agreement between the twoethods (Fig. 7).

ensitivity and specificity

mong the eight patients with known PAD, four had nor-

al ABI with both methods and four had abnormal ABI withoth methods; among the seven patients with claudicationut without known PAD, five had pathological ABI with bothethods and two had normal ABI with both methods.

igure 5. Comparability of the Doppler and automatic methodsccording to the Bland & Altman test.

ABI using automatic blood pressure device

Figure 6. Correlation between Doppler and automatic methodsin terms of pathological ankle-brachial indexes.

tjslflIdTi

ttp

maua(cmean

Am

Tmvttlpto

beAtaappp[[apva

P

T

Figure 7. Comparability of the Doppler and automatic methodsaccording to the Bland & Altman test in patients with a pathologicalankle-brachial index.

The sensitivity, specificity, positive predictive value, neg-ative predictive value and accuracy of a pathological indexdetermined using the automatic device were tested to pre-dict an abnormal index in at least one leg with the standardDoppler measurement (analysis in terms of patients ratherthan limbs in the 196 patients with determination of ABI byboth methods).

There were 166 true negatives, 24 true positives, twofalse negatives and four false positives. The four patientswith an automatic index that was not measurable wereclassified correctly as true positives. Sensitivity was 92%,specificity was 98%, positive predictive value was 86% andnegative predictive value was 99%. Accuracy (patients classi-fied correctly by the automatic ABI in reference to standardDoppler indexes) was 97%. Analysis of concordance wasexcellent (� = 0.87).

Discussion

Feasibility of use and incidence ofpathological indexes

The automatic sphygmomanometer appeared to give veryprecise measurements of systolic pressure in lower limbs in

toeua

523

he normal range of blood pressures in 99% of tested sub-ects, but seemed to be unsuitable for measuring very lowystolic blood pressure. Very low systolic pressure in theower limbs is a marker of severe deterioration of arterialow when blood pressure is normal (or elevated) in the arms.

n fact, this kind of automatic blood pressure device waseveloped to assess hypertension rather than hypotension.hese observations are in agreement with the results of our

ntra-arterial validation in a previous study [11].Patients did not complain of any severe side effects;

wo found the automatic method slightly more painful thanhe standard Doppler method, perhaps because of a longereriod of cuff inflation/deflation.

The 8% incidence of PAD according to pathological auto-atic ABI seems to be in agreement with the incidences of

round 10% reported at this age (40 to 60 years) in studiessing Doppler measurements of ABI [10]. This new methodppears to be very easy, does not need special equipmentsuch as a Doppler probe) and requires only minimal physi-ian training. It is less costly than the standard Dopplerethod. The use of a single device rather than two differ-

nt devices (Doppler probe and sphygmomanometer) seemspractical advantage and would probably lead a greater

umber of physicians to assess ABI routinely.

ccuracy and comparability with the Dopplerethod

he global comparison of the values obtained by the twoethods showed good correlation. However, the value-to-

alue comparison showed some small differences, mainly inhe normal values; this explains the weak agreement in theotal range of values, which was not found in the patho-ogical values. These differences could be due to technicalrinciples (oscillometric versus auscultatory). Nevertheless,hese differences in normal values do not affect the abilityf the automatic method to detect pathological subjects.

This simple new method for determining ABI seems toe in agreement with the standard Doppler method and isffective in the detection of PAD defined by a pathologicalBI less than 0.90 in routine clinical examinations in asymp-omatic subjects. Subjects with a pathological Doppler testre classified accurately by the new method (97% accuracynd 87% concordance). This result is in line with a studyerformed to compare an automatic oscillometric bloodressure device with the standard Doppler method in theostoperative surveillance of infra-inguinal bypasses by ABI14] and with our previous study in cardiovascular patients11]. Sensitivity (92%), specificity (98%) and accuracy (97%)re higher here than in our previous study in cardiovascularatients (76, 95 and 89%, respectively). This underlines thealue of this method for detecting unknown PAD in middle-ged subjects not consulting for cardiovascular reasons.

erspectives

his new method for determining ABI in the routine detec-

ion of moderate PAD appears to be effective. It seemsf particular value in ruling out significant PAD, given itsxcellent specificity and negative predictive value. It seemsseful for general or preventive medicine physicians whore not trained in using the Doppler probe. Furthermore, it

5

wiluoddabfei

L

AfimspdDotDsmdnpr

C

Tsaoepmam

R

[

[

[

[

[

[

24

ould be of value if automatic devices devoted to measur-ng blood pressure (particularly low blood pressure) in theower limbs could be developed by manufacturers, partic-larly in terms of cuffs and algorithms. In fact, the valuef automatic ABI measurement lies in its capacity to easilyetect patients with PAD who would probably not have beeniagnosed in real medical practice. The subjects who aret high risk of cardiovascular events (also identified as suchy other methods, such as risk factor scores) [15] will needurther appropriate investigations (Doppler, Echo-Doppler,tc.) to assess the focal lesions precisely and appropriatenterventions.

imitations

s the aim of this study was to test the accuracy andeasibility of the automatic method for detecting patholog-cal indexes in preventive medicine, the two measurementethods (automatic and Doppler) were not used in all

ubjects but only in a large subgroup (n = 196). A recentaper [16] pointed out somewhat less inter-observer repro-ucibility between automatic measures of ABI than betweenoppler measures (in an analysis of limbs) in a short seriesf 50 tests done by vascular physicians. This result showshat the automatic method is perhaps less reproducible thanoppler method and somewhat less precise in assessing theeverity of PAD, but in practice, for general physicians, theost important factor is whether the method is able toetect patients with an abnormal index in routine exami-ations. Our results in routine examinations performed byreventive medicine physicians at work support the accu-acy of the automatic method for this purpose.

onclusion

he use of a commercially available automatic blood pres-ure device for determining ABI in preventive medicineppears feasible, easy, quick and precise in detecting normalr abnormal ratios. It will provide a practical and cost-ffective tool that enables preventive medicine or generalhysicians not trained in the Doppler method to assess ABIore systematically in any routine medical consultation in

dults, to detect PAD, to make a more complete local assess-ent and to assess global cardiovascular risk.

eferences

[1] Leng GC, Lee AJ, Fowkes FG, et al. Incidence, natural historyand cardiovascular events in symptomatic and asymptomatic

[

D. Benchimol et al.

peripheral arterial disease in the general population. Int J Epi-demiol 1996;25:1172—81.

[2] Criqui MH, Coughlin SS, Fronek A. Noninvasively diagnosedperipheral arterial disease as a predictor of mortality:results from a prospective study. Circulation 1985;72:768—73.

[3] Criqui MH, Langer RD, Fronek A, et al. Mortality over a periodof 10 years in patients with peripheral arterial disease. N EnglJ Med 1992;326:381—6.

[4] McKenna M, Wolfson S, Kuller L. The ratio of ankle and armarterial pressure as an independent predictor of mortality.Atherosclerosis 1991;87:119—28.

[5] Newman AB, Siscovick DS, Manolio TA, et al. Ankle-arm indexas a marker of atherosclerosis in the Cardiovascular HealthStudy. Cardiovascular Heart Study (CHS) Collaborative ResearchGroup. Circulation 1993;88:837—45.

[6] Murabito JM, D’Agostino RB, Silbershatz H, Wilson WF. Inter-mittent claudication. A risk profile from The Framingham HeartStudy. Circulation 1997;96:44—9.

[7] Schmieder FA, Comerota AJ. Intermittent claudication: mag-nitude of the problem, patient evaluation, and therapeuticstrategies. Am J Cardiol 2001;87:3D—13D.

[8] Smith GD, Shipley MJ, Rose G. Intermittent claudication, heartdisease risk factors, and mortality. The Whitehall Study. Circu-lation 1990;82:1925—31.

[9] Criqui MH, Fronek A, Barrett-Connor E, Klauber MR, Gabriel S,Goodman D. The prevalence of peripheral arterial disease in adefined population. Circulation 1985;71:510—5.

10] Meijer WT, Hoes AW, Rutgers D, Bots ML, Hofman A, GrobbeeDE. Peripheral arterial disease in the elderly: The Rot-terdam Study. Arterioscler Thromb Vasc Biol 1998;18:185—92.

11] Benchimol A, Bernard V, Pillois X, Hong NT, Benchimol D,Bonnet J. Validation of a new method of detecting periph-eral artery disease by determination of ankle-brachial indexusing an automatic blood pressure device. Angiology 2004;55:127—34.

12] Simon A, Papoz L, Ponton A, et al. Feasibility and reliabil-ity of ankle/arm blood pressure index in preventive medicine.Angiology 2000;51:463—71.

13] Rose GA. The diagnosis of ischaemic heart pain and intermit-tent claudication in field surveys. Bull World Health Organ1962;27:645—58.

14] Blebea J, Ali MK, Love M, Bodenham R, Bacik B. Automaticpostoperative monitoring of infrainguinal bypass procedures.Arch Surg 1997;132:286—91.

15] Fowkes FG, Murray GD, Butcher I, et al. Ankle brachialindex combined with Framingham Risk Score to predict car-diovascular events and mortality: a meta-analysis. JAMA2008;300:197—208.

16] Aboyans V, Lacroix P, Doucet S, Preux PM, Criqui MH, LaskarM. Diagnosis of peripheral arterial disease in general practice:can the ankle-brachial index be measured either by pulse pal-pation or an automatic blood pressure device? Int J Clin Pract2008;62:1001—7.