Journal of Human (1999) 13, 105–109  1999 Stockton Press. All rights reserved 0950-9240/99 $12.00 http://www.stockton-press.co.uk/jhh ORIGINAL ARTICLE Arm position is important for measurement

RT Netea, JWM Lenders, P Smits and T Thien Department of Internal Medicine, University Hospital Nijmegen, The Netherlands

Aim: To test the effect of positioning the arm on the arm- nificantly higher (P Ͻ 0.0001) when the arm was placed rest of a common chair, below the officially rec- on the arm-rest of the chair than at the right atrial level. ommended right atrial level, on the blood pressure (BP) The same differences ± s.d. in BP between the two pos- readings in a group of out-patients. itions were obtained with both measurement tech- Patients and methods: A group of 69 patients (58 hyper- niques: 9.7 ± 9.4 mm Hg (systolic) and 10.8 ± 5.8 mm Hg tensives; 39 males; mean ± s.d. age 54.1 ± 16.0 years) (diastolic) with the mercury sphygmomanometer and participated in the present study. BP and rate respectively 7.3 ± 8.9 mm Hg and 8.3 ± 6.0 mm Hg with values obtained in each of the following two positions the oscillometric device. No difference in the were compared: (1) sitting with the arms supported on was found between the two positions. the arm-rests of the chair and (2) sitting with the arms Conclusions: Placing the patient’s arms on the arm-rest supported at the level of the mid-sternum (the approxi- of the chair instead of at the reference right atrial level, mation of the right atrial level). BP was measured simul- BP measurement will result in spuriously elevated BP taneously at both arms, with a mercury sphygmoman- values. This may be of great importance for the diag- ometer at the right arm and with an automatic nosis and the subsequent treatment decisions for oscillometric device at the left arm. patients with hypertension. Results: Both the systolic and diastolic BPs were sig-

Keywords: blood pressure measurement; arm position

Introduction rest of a common chair. The aim of the present study was to examine the effect of such a deviation from The diagnosis and treatment of hypertension is the recommended position on the BP readings in a mostly based on the indirect blood pressure (BP) group of out-patients. The indirect BP readings measurement. Therefore this clinical procedure can taken in two positions were compared: (1) seated have far-reaching consequences. One of the most with the arms supported at the level of the right important factors that affects the BP measurement is atrium and (2) seated with the arms supported on the patient’s position in which the BP is measured. the arm-rests of a common chair. The reference point for the measurement of BP is the right atrium,1 the so called heart level. Recent guidelines recommend that the arm should be Patients and methods placed at the ‘heart level’ during BP measurement, A group of 69 out-patients (30 females; 39 males; irrespective of the body position.2–6 However, it is mean ± s.d. age 54.1 ± 16.0) years; body mass index only vaguely suggested how the ‘heart level’ can be 26.7 ± 4.1 kg/m2; right arm circumference 29.5 ± practically estimated. Moreover it is not always indi- 2.7 cm; left arm circumference 29.2 ± 2.7 cm) partici- cated which anatomical segment of the arm should pated to our study. Fifty-eight of them were hyper- be at heart level.4–6 tensives and out of these 43 were taking antihyper- Previous studies have shown large differences in tensive medication at the time when they were the sitting BP between readings taken with the arm involved in the study. Patients with diabetes mel- at the heart level and with the arm vertically pen- litus, pregnancy, heart failure, autonomic neuro- dent by the patient’s side.7–11 However, in daily pathy or those taking ␣-adrenoceptor blockers or tri- practice as well as in research the arm position dur- cyclic antidepressants were excluded from the ing BP measurement is often neglected.12,13 As a present study. deviation from the recommended position, the BP is All the subjects were familiar with the measure- often measured by having the patient seated with ment of BP and a detailed explanation of the pro- the forearm supported on the desk or on the arm- cedure was given to the subjects before starting the measurements. Speaking with the subjects was avoided during both the resting time and the Correspondence: Professor Theo Thien, Department of Internal measurements. Medicine, University Hospital Nijmegen, Geert Grooteplein 8, 6500 HB Nijmegen, The Netherlands The whole procedure was carried out in a quiet Received 16 April 1998; revised 10 August 1998; accepted 17 room, at a comfortable temperature (21–23°C), by a August 1998 single trained observer (RTN). The right atrium level Blood pressure measurement and arm position RT Netea 106 was located according to British Hypertension Table 1 Systolic and diastolic BPs (mm Hg) and heart rate Society recommendations5 at the level of mid-ster- (beats/min) in 69 volunteers with the arm at right atrial level (arm up) and on the arm-rest of the chair (arm low). Results are given num (precisely determined in each subject at half- as mean ± s.e. way between manubrium at fossa jugularis and xyphoid at fossa epigastrica). In all subjects the BP Right arm Left arm was measured after 5 min of rest sitting in a comfort- (sphygmomanometer) (oscillometric) able ordinary chair with the back supported and the feet on the floor. Three BP readings 1-min apart were Systolic blood pressure taken, the patients having their arms supported arm up 133.3 ± 2.5 131.8 ± 2.3 arm low 143.0 ± 2.4 138.9 ± 2.5 passively at each of the two levels: the level of the difference 9.7 ± 1.1* 7.3 ± 1.0* arm-rest of the chair and with the cubital fossa at the level of mid-sternum.13 The sequence of the pos- Diastolic blood pressure itions was randomised. Two minutes of pause were arm up 77.7 ± 1.2 78.0 ± 1.2 ± ± allowed between the last reading in the first position arm low 88.6 1.1 86.2 1.2 difference 10.8 ± 0.7* 8.3 ± 0.7* and the first reading in the second position. The BP was measured simultaneously at both arms using a Heart rate Hawksley random-zero sphygmomanometer arm up 66.2 ± 1.3 66.0 ± 1.3 × arm low 66.7 ± 1.3 66.1 ± 1.3 (bladder size 36 13 cm) together with a Rappaport ± ± Sprague stethoscope at the right arm and an auto- difference 0.5 0.3 0.1 1.1 matic oscillometric device (Bosomat) (bladder size P Ͻ 0.0001 for arm low vs arm up. 22 × 12.5 cm), previously tested for accuracy,14 at the left arm. With the random-zero sphygmoman- ometer, phase I of was used to Table 2 Systolic and diastolic BP values (mm Hg) and the differ- determine the systolic BP and phase V for the dia- ence (mm Hg) between the readings with the arm on the arm-rest of the chair (low) and with the arm at right atrial level (up) in the stolic BP to the nearest 2 mm Hg, with a maximal ± deflation rate of 2 mm Hg/sec. Corrected BP values two sequences of these positions. Results are given as mean s.e. for random zero readings were calculated after all Sequence low/up Sequence up/low measurements were done. After each BP reading a (n = 34 subjects) (n = 35 subjects) 20-sec rate was counted at the right radial . The automatic device also registered the systolic diastolic systolic diastolic heart rate. Finally, each subjects body weight, height and arm circumference (halfway between the acro- Right mion and olecranon) were determined. (sphygmomanometer) The three readings taken in each position were arm up 135.5 ± 3.9 78.2 ± 1.8 131.2 ± 3.4 77.3 ± 1.7 arm low 148.3 ± 3.9 90.2 ± 1.8 137.9 ± 2.7 87.1 ± 1.5 averaged to obtain a representative BP for that pos- difference 12.9 ± 1.5**11.9 ± 0.7** 6.7 ± 1.5** 9.8 ± 1.2** ition. Student’s t-test was used to test for differences in BP and heart rate between the two positions. Left Spearman’s rank correlations were computed (oscillometric) ± ± ± ± between the individual differences obtained with arm up 133.5 3.7 78.1 1.8 129.8 3.0 77.6 1.6 arm low 144.6 ± 3.9 89.0 ± 1.8 133.4 ± 2.9 83.5 ± 1.7 the two measurement techniques with respect to the difference 11.1 ± 1.6**10.8 ± 1.1** 3.6 ± 1.2* 5.8 ± 0.8** systolic and diastolic BPs. The statistical signifi- Ͻ cance was reached when P 0.05 (two-sided). *P Ͻ 0.01; **P Ͻ 0.001.

Results were significantly lower when the patient’s arms Both systolic and diastolic BPs were significantly were placed at the right atrial level than when the higher (P Ͻ 0.001) when the arm was placed on the arms were on the arm-rests of the chair. For both arm-rest of the chair (Table 1). This was the case systolic and diastolic BPs, the differences between for both arms. A significant correlation was found the two positions were larger when the BP was mea- between the individual differences obtained with sured with the arms on the arm-rests of the chair the random zero sphygmomanometer and with the first than when it was measured with the arm at the automatic device (r2 = 0.422, P Ͻ 0.001 for systolic right atrial level first. BP and r2=0.127, P Ͻ 0.01 for diastolic BP). There Figure 1 shows the individual differences between was no significant difference in the heart rates the BP readings with the arms on the arm-rests of between the two positions. the chair and the corresponding BP readings with In Table 2 the BP values in the two positions are the arms at right atrial level vs the average BP in given for the 34 patients in which the measurements those two positions. The differences between the BP were performed first with the arms on the arm-rest measured in the two positions were not related to of the chair and then with the arms placed at the the average BP level. right atrial level, and for the 35 patients in which the measurements were performed in the reversed Discussion sequence of the two positions. The two subgroups did not significantly differ with respect to age, body The present study examined the influence of the arm mass index, sex, number of patients with hyperten- position, at right atrial level vs at the level of a com- sion and medication. Both systolic and diastolic BPs mon arm-rest of a chair, on BP readings in a group Blood pressure measurement and arm position RT Netea 107 lel to the body in the sitting position have been reported.7–10 However, in some of these studies7,9,10 ‘the heart level’ was not precisely defined. Recent guidelines2–6 recommend that the patient should be seated relaxed with the arm-rested passively at the ‘heart level’ during BP measurement. However, in these guidelines it is not always exactly specified what particular anatomical segment of the arm should be at ‘heart level’ and how the heart level should be estimated. The WHO guidelines re- commend that the cubital fossa should be placed at the heart level as approximated by the fourth inter- costal space.2 The British Hypertension Society re- commends that the arm should be placed horizontal at the level of mid-sternum.5 The American Heart Association recommends that the elbow should be flexed at the heart level.3 Most of the physicians measure the BP in a sitting position with the arm-rested on a table or on the common arm-rest of the chair, ignoring the relation between the subject’s height and the level of the arm-rest. Our findings reveal that even such a small deviation from the officially recommended position might result in falsely elevated BP readings. The dif- ferences of both systolic and diastolic BPs between the two positions were consistent and are clini- cally relevant. The difference in the BP between the two pos- itions was higher when the BP measurement was started with the arm on the arm-rest of the chair, so there appeared to be a sequence effect too. This might be due to differences in the resting time before the BP measurement in each position. In this study BP measurements in the first position were perfor- med after 5-min of quiet rest whereas the BP measurements in the second position were perfor- med after at least 10 min of quiet sitting. It has been 15 Figure 1 Differences between the BP readings with the arm at the reported previously that the BP may decrease over two levels (BParm up − BParm low) vs the mean BP in the two time during quiet rest. Therefore the factor time positions ((BParm up + BParm low)/2) for the right (a) and left might have induced an increased difference in the (b) arm. The closed symbols represent the systolic BP and the BP readings between the two sequences of positions. open symbols the diastolic BP. Dotted lines represent ± 2 s.d. More importantly however, the differences in BP between the two positions were in both position of normo- and hypertensive out-patients. The right sequences significant and in the same direction. atrial level was approximated by exact determi- To avoid observer bias and digit preference, BP nation of the mid-sternum. Both the systolic and was measured with the random-zero sphygmoman- diastolic BPs were significantly lower when the arm ometer. Recent studies16–19 showed potential prob- was placed at right atrial level than on the support lems with the random zero sphygmomanometer of the chair. The differences were of the same magni- such as an underestimation of the BP and the possi- tude with both the random-zero sphygmoman- bility of adding the digit preference for zero-value ometer and the semi-automated oscillometric as a new source of error when computing the real device. A slight sequence effect was also found BPs. Nevertheless, the random zero sphygmoman- regardless of the instrument used for the BP ometer is still preferable when blinding is important measurement. to avoid observer bias in recording repeated BP read- BP measurement is a widely used procedure in ings. To support the reliability of the measurements, clinical practice on which the diagnosis and treat- simultaneous BP determinations with an automatic ment of hypertension are based. To obtain an accur- oscillometric device were included in the protocol. ate reading and to be able to compare readings taken The individual differences between the two pos- on separate occasions, it is important to standardize itions with respect to both the systolic and the dias- the BP measurement. A factor that is well known tolic BPs obtained with the two different measure- but often neglected is the position of the patient’s ment techniques were significantly correlated. arm during the measurement. Other interfering factors such as stress and noise In previous studies important differences between were reduced in the present study as all the patients BP readings taken with the arm raised at the so were familiar with the procedure and the room was called ‘heart level’ and with the arm vertical, paral- quiet with a pleasant temperature. Talking was Blood pressure measurement and arm position RT Netea 108 avoided during the relaxation time and the measure- 7 Merendino J, Finnerty FA. Importance of the position ments. of the arm on the level of arterial blood pressure. JAMA Our findings are supported by our earlier results 1961; 175: 51–53. with respect to the influence of arm position on 8 Mitchell PL, Parlin RW, Blackburn H. Effect of vertical intra-arterial BP readings20 and by the results displacement of the arm on indirect blood pressure obtained in a previous epidemiological study11 in a measurement. N Engl J Med 1964; 271: 72–74. 9 Webster J, Newnham D, Petrie JK, Lovell HG. Influence large group of healthy volunteers. The hydrostatic of arm position on measurement of blood pressure. Br factors were considered the most important for the Med J 1984; 288: 1574–1575. differences observed between the BP readings at dif- 10 Parr GD, Poole PH. Effects of sphygmomanometer type ferent arm levels. In this case, to correct a BP value and position of the arm on blood pressure measure- measured at a certain level below the heart, one ment. J Hum Hypertens 1988; 2: 153–156. should substract the hydrostatic pressure of the 11 Terent A, Breig-Asberg E. Epidemiological perspective blood column between the heart and the level of of body position and arm level in blood pressure measurement from the measured BP value. measurement. Blood Pressure 1994; 3: 156–163. The importance of the present findings for clinical 12 Bailey H, Bauer JH. Review of common errors in the practice can be better appreciated if one considers indirect measurement of blood pressure. Arch Intern that initiation and modification of antihypertensive Med 1993; 153: 2741–2748. therapy are based on BP levels in categories of 5 to 13 Roche V, O’Malley K, O’Brien E. How ‘scientific’ is 10 mm Hg.2,4 Changes in BP as low as 10 mm Hg can blood pressure measurement in leading medical jour- nals? J Hypertens 1990; 8: 1167–1168. consistently modify the risk of cerebral and cardio- 21,22 14 Sloan PJ et al. Standardised methods for comparison vascular complications of hypertension. As of sphygmomanometers. J Hypertens 1984; 2: 547–551. shown by the present study, such erroneous elev- 15 Van Loo JM, Peer PG, Thien Th. Twenty-five minutes ations in BP might be obtained when the patient’s between blood pressure readings: the influence on the arm is inappropriately placed below the right atrial prevalence rates of isolated hypertension. J Hypertens level on the arm-rest of a common chair. A question 1986; 4: 631–635. that may arise is: in which position has the BP been 16 O’Brien E, Mee F, Atkins N, O’Malley K. Inaccuracy measured in the large epidemiologic and outcome of the Hawksley random zero sphygmomanometer. studies? In the Framingham study,23 key data on Lancet 1990; 336: 1465–1468. which diagnosis and risk of hypertension is based, 17 Conroy RM et al. Using Hawksley random zero sphyg- the BP was measured according to the American and momanometer as a gold standard may result in mis- British Heart Associations, which both recommend leading conclusions. Blood Pressure 1994; 3: 283–286. 18 Mackie A, Whincup P, McKinnon M. Does the that the arm should be placed at heart level. Unfor- Hawksley random zero sphygmomanometer underesti- tunately no mention is made with respect to the arm 24–30 mate blood pressure, and by how much? J Hum Hyper- position in most other published data, and this tens 1995; 9: 337–343. problem has already been pointed out by other 19 Kinirons MT et al. Hawksley random zero sphygmo- authors.31 This emphasizes even more the impor- manometer versus the standard sphygmomanometer: tance of standardizing the BP measuring procedure. an investigation of mechanisms. J Hum Hypertens In conclusion, our study establishes once more 1995; 9: 571–573. the importance of an adequate and general accepted 20 Netea RT et al. Influence of the arm position on definition of the correct arm position during BP intraarterial blood pressure measurement. J Hum Hyp- measurement in order to avoid that different investi- ertens 1998; 12: 157–160. gators consider their BP values interchangeable, 21 Collins R et al. Blood pressure, stroke, and coronary whereas differences in arm position are not taken heart disease. Part 2: short term reduction in blood into account. pressure: overview of randomized drugs in their epide- miological context. Lancet 1990; 335: 827–838. 22 MacMahon S et al. Blood pressure, stroke, and coron- References ary heart disease. Part 1: prolonged differences inblood pressure: prospective observational studies corrected 1 Guyton A. Textbook of medical physiology. WB Saun- for the regression dilution bias. Lancet 1990; 335: ders Company: Philadelphia: 1986, p 224. 765–774. 2 1993 Guidelines for the management of mild hyperten- sion: Memorandum from a WHO/ISH Meeting. Hyper- 23 Kannel WB, Sorlie P, Gordon T. Labile hypertension: tension 1993; 22: 392–403. A faulty concept? The Framingham study. Circulation 3 Recommendations for human blood pressure determi- 1980; 61: 1183–1187. nation by sphygmomanometers: Report of a special 24 MRC Working Party. Medical Research Council trial task force appointed by the steering committee, Amer- of treatment of hypertension in older adults: principal ican Heart Association. Hypertension 1988; 11: results. BMJ 1992; 304: 405–412. 210A–22A. 25 Medical Research Council Working Party. MRC trial of 4 The sixth Report of the Joint National Committee on treatment of mild hypertension: principal results. Br Detection, Evaluation and Treatment of High Blood Med J 1985; 291: 97–104. Pressure. Arch Intern Med 1997; 157: 2413–2446. 26 Staessen J et al. Relation between mortality and treated 5 Petrie JC, O’Brien ET, Littler WA, de Swiet M. Rec- blood pressure in elderly patients with hypertension: ommendations on blood pressure measurement. Br report of the European Working Party on High Blood Med J 1986; 293: 611–615. Pressure in the Elderly. BMJ 1989; 298: 1552–1556. 6 American Society of Hypertension Recommendations 27 Samuelsson OG et al. The J-shaped relationship for Routine Blood Pressure Measurement by Indirect between coronary heart disease and achieved blood Cuff Sphygmomanometry. Am J Hypertens 1992; 5: pressure level in treated hypertension: further analy- 207–209. ses of 12 years of follow-up of treated hypertensives in Blood pressure measurement and arm position RT Netea 109 the Primary Prevention Trial in Gothenburg, Sweden. 30 Collins R et al. Blood pressure, stroke, and coronary J Hypertens 1990; 8: 547–555. heart disease. Part 2: short-term reductions in blood 28 Gillman MW, Cook RN. Blood pressure measurement pressure: overview of the randomised drug trials in in childhood epidemiological studies. Circulation their epidemiological context. Lancet 1990; 335: 1995; 92: 1049–1057. 827–838. 29 Dahlof B et al. Morbidity and mortality in the Swedish 31 Roche V, O’Malley K, O’Brien E. How ‘scientific’ is Trial in Old Patients with Hypertension (STOP- blood pressure measurement in leading medical jour- Hypertension). Lancet 1991; 338: 1281–1285. nals? J Hypertens 1990; 8: 1167–1168.