There is an increasing number of wrist blood pressure measurement devices that successfully passed the validation procedures of the British Hypertension Society (BHS) and the European Society of Hypertension (ESH). It remains unknown, however, whether pulse pressure as a marker of arterial stiffness and vascular ageing affects the accuracy of these devices. An ESH protocol validated wrist device was compared with the upper arm mercury sphygmomanometry in a study population (33 patients, 99 measurements) including a relevant number of subjects with pulse pressure >50 mm Hg (84.8%) and isolated systolic hypertension (27.3%). Mean systolic bias was 10.2 mm Hg with 95% limits of agreement of −13.1 and 33.6 mm Hg, mean diastolic bias was 4.8 mm Hg with limits of agreement of −11.0 and 20.7 mm Hg. The impact of body mass index, age, systolic blood pressure and pulse pressure on the absolute value of blood pressure bias was tested by stepwise multiple regression analysis. The systolic bias significantly depended on pulse pressure, whereas there was no significant effect of the independent variables on the diastolic bias. Separate correlation analysis showed a significant correlation between pulse pressure and both absolute systolic bias (Pearson r=0.48, P<0.001) and relative systolic bias (systolic bias divided by systolic blood pressure, Pearson r=0.29, P=0.003). Even well-validated wrist blood pressure devices can show a clinically relevant bias in patients with elevated pulse pressure. Increased arterial stiffness may impair the accuracy of oscillometric blood pressure measurement at the wrist.
Home blood pressure measurement has a stronger predictive power for cardiovascular mortality than office blood pressure screening.1 Oscillometric automated blood pressure measurement devices have markedly simplified blood pressure self-monitoring, and a lot of devices have proven high accuracy in validation procedures according to the British (British Hypertension Society, BHS), European (European Society of Hypertension, ESH) or US (Association for the Advancement of Medical Instrumentation, AAMI) protocols.2, 3, 4 Among these home blood pressure measurement devices the wrist monitors have gained increasing popularity. These devices are small, easy to handle and can be used without the need to undress. Therefore, especially elderly people tend to prefer wrist blood pressure measurement monitors to the upper-limb devices. Initially, there was a lack of wrist devices that have been validated by international protocols. In recent years, however, there is an increasing number of wrist devices that fulfilled the required accuracy criteria of BHS, ESH and AAMI. Validation with these protocols examines the degree of agreement between clinical readings from the automated monitor and those on a mercury sphygmomanometer, still widely accepted as the gold standard for non-invasive blood pressure measurements. The accuracy of wrist blood pressure monitors strictly depends on the correct use of the device, including positioning of the cuff on heart level and avoidance of palmar flexion and extension. We have observed, however, that—even though properly used—well-validated wrist devices show a marked bias in elderly patients with advanced atherosclerosis and increased pulse pressure. Pulse pressure depends on vascular stiffness and may be regarded as a footprint of vascular ageing.
In this study we examined the impact of pulse pressure on the accuracy of an ESH-validated wrist blood pressure measurement device in a study population including a relevant number of subjects with pulse pressure >50 mm Hg and isolated systolic hypertension.
As proposed by the ESH criteria, 33 subjects were included in the study.3 There were 20 female and 13 male probands (60.6 vs 39.4%). Nineteen participants (57.6%) were normotensive, 14 (42.4%) were hypertensive and 9 (27.3%) had isolated systolic hypertension. A total of 28 subjects (84.8%) had a pulse pressure >50 mm Hg. Mean age was 59.2±14.9 years, mean body mass index (BMI) was 27.9±4.6 kg m−2. Systolic blood pressure was 135.8±23.2 mm Hg, diastolic blood pressure was 77.0±12.5 mm Hg, resulting in a mean pulse pressure of 58.8±17.1 mm Hg. The characteristics of the study population are summarized in Table 1. Informed consent was obtained from all the probands.
Overall, 99 measurements were performed in 33 subjects with a wrist blood pressure monitor and with an upper-arm reference device. The wrist device was an automated oscillometric blood pressure monitor with standard cuff size range of wrists from 13.5 to 19.5 cm (Omron RX-Classic, Omronmedizintechnik, Mannheim, Germany). The device was validated according to the ESH protocol.3 Exclusion criteria were defined to be cardiac arrhythmia (atrial fibrillation, frequent extrasystoles), age <18 years, inability to declare informed consent for participation, a history of upper limb trauma or upper limb arterial stenosis and wrist circumference <13.5 or >19.5 cm. All the measurements were performed by the same two trained observers. The team had experience in validation procedures according to both BHS and ESH protocols.5 Before the study's measurements, agreement of readings of observers was shown to assure the validity of measurement results as published previously.5 A calibrated mercury sphygmomanometer (Erkameter 3000, Erka Kallmeyer Medizintechnik, Bad Tölz, Germany) was used as reference device. Arm circumferences were measured and recorded to allow correct choice of cuff size. The measurements took place in a quiet room with an ambient temperature of 20–22 °C. Subjects had to rest seated for at least 5 min before the measurement procedure was initiated. The two observers were blinded to each other. Mercury readings were taken by one mercury column and a two-person stethoscope with a Y-connector. To avoid venous congestion and to minimize variability in blood pressure, the time between measurements was determined to be 30–60 s. Wrist blood pressure measurement was performed according to the generally accepted recommendations: The cuff was held midway between jugular notch and xiphoid process.6 Before beginning the measurements, the observers excluded palmar flexion or extension of the wrist. All the measurements were performed at the same arm of the subject. Readings were noted and differences between device and observers were determined.
The measurement procedure was performed as follows:
BPa Wrist device (to check the automated device, not included in the analysis)
BP1 Mercury (observers 1 and 2)
BP2 Wrist device (test device)
BP3 Mercury (observers 1 and 2)
BP4 Wrist device (test device)
BP5 Mercury (observers 1 and 2)
BP6 Wrist device (test device)
BP7 Mercury (observers 1 and 2)
Results are presented as mean±s.d. Comparison of systolic and diastolic blood pressure of test and reference device was performed by Tukey mean-difference plots (Bland–Altmann plots). Bias and the limits of agreement (bias±2 s.d.) are reported. The impact of age, BMI, systolic/diastolic blood pressure and pulse pressure on the absolute value of the systolic and diastolic measurement bias was analysed by stepwise multiple linear regression analysis. Furthermore, the association of pulse pressure and absolute/relative measurement bias was analysed by correlation analysis. Pearson r and P-value (two-tailed) are reported. P<0.05 was regarded significant.
Measurement procedures with the automated wrist device were successful in all the 33 probands. Figure 1 provides the Tukey mean-difference plots (Bland–Altmann plots) of systolic (A) and diastolic (B) blood pressure difference of device and observers (mean of observers A and B, n=99 measurements). Mean systolic bias was 10.2 mm Hg with 95% limits of agreement (±2 s.d.) of −13.1 and 33.6 mm Hg, mean diastolic bias was 4.8 mm Hg with limits of agreement of −11.0 and 20.7 mm Hg. To analyse the impact of age, BMI, systolic blood pressure and pulse pressure on the absolute value of the systolic and diastolic measurement bias, we performed a stepwise multiple linear regression analysis. The absolute value of the blood pressure bias was defined as the dependent variable. The results of the linear regression model for the systolic blood pressure bias are presented in Table 2. The variable ‘pulse pressure’ was included in the model with a standardized β-coefficient of 0.485 and a T-value of 5.43. Analysis of variance testing for the acceptability of the model showed that the variation accounted for by the model is described by an R-value of 0.485, R2 of 0.235 and an F-value of 29.52 (P<0.001). The null-hypothesis that there was no impact of pulse pressure was rejected with P<0.001. Systolic blood pressure (P=0.184), age (P=0.244) and BMI (P=0.489) were excluded. Stepwise multiple linear regression analysis with the absolute diastolic bias as dependent variable and age, BMI, diastolic blood pressure and pulse pressure as independent variables showed no significant impact of the independent variables on the dependent variable (P>0.05 each). As there is the possibility of multicollinearity between systolic/diastolic blood pressure and pulse pressure, a univariate correlation analysis of pulse pressure and both absolute and relative measurement bias was added. The results are shown in Figure 2. As indicated by the regression lines there was a significant correlation between pulse pressure and systolic bias (Pearson r=0.48, P<0.001). The correlation remains significant for the association of pulse pressure and the ‘relative systolic bias’ (systolic bias divided by systolic blood pressure) with a Pearson r=0.29, P=0.003. As evident in the figure, there is no correlation between pulse pressure and diastolic absolute/relative blood pressure. Pearson r is −0.10 (P=0.33) for the absolute diastolic bias and −0.15 (P=0.13) for the relative diastolic bias.
Ageing is associated with increasing vascular stiffening and reduced large artery Windkessel function.7 Thus, systolic blood pressure gradually increases in the course of life, whereas diastolic blood pressure starts falling again in the sixth life decade, leading to a rise of pulse pressure and isolated systolic hypertension.8 The extent of pulse pressure may be regarded as a footprint of vascular ageing. Isolated systolic hypertension is the most common form of hypertension in the elderly.9 The present study population included a relevant number of subjects with increased pulse pressure and isolated systolic hypertension. Although the device has proven sufficient accuracy in a validation procedure according to the ESH protocol, it showed a clinically relevant bias in the present study population. The multiple regression analysis shows that the systolic bias depends on pulse pressure. With regard to a potential intercollinearity between systolic pressure and pulse pressure, we additionally performed a correlation analysis of the association of pulse pressure and both absolute and relative systolic measurement. The correlation is highly significant for the association of systolic measurement bias and pulse pressure, and remains significant for the ‘relative systolic bias’ (systolic bias divided by systolic blood pressure). Thus, the effect of pulse pressure on the measurement bias cannot be attributed merely to the extent of systolic blood pressure.
Stiff arteries can lead to an overestimation of non-invasively assessed systolic arterial pressure. This phenomenon is of particular relevance for oscillometric devices.10 Accordingly, the positivity of the bias values in the Bland–Altmann analysis of this study show a significant trend to overestimate both systolic and diastolic pressure. In contrast to the brachial artery at the upper arm, the anatomical site of the radial artery medioventral of the distal radius bone impedes homogenous circular compression by a cuff. This limitation of wrist blood pressure monitoring might be of particular importance in arteries with advanced atherosclerosis.
In the upper arm auscultatory sphygmomanometry, the brachial artery is occluded by a cuff and inflated to above systolic pressure. As it is gradually deflated, pulsatile blood flow is re-established and accompanied by Korotkoff sounds. The appearance of clear tapping sounds (phase I) is regarded to correspond to systolic pressure, the disappearance of sounds (phase V) is regarded to correspond to diastolic pressure.11 Wrist blood pressure monitors, however, work oscillometrically. It was shown that when the oscillations of pressure in a cuff are recorded during deflation, the point of maximal oscillation corresponds to the mean intra-arterial pressure. Systolic and diastolic pressures can only be estimated indirectly according to the empirically derived algorithms. The amplitude of the oscillations, however, depends on several factors other than blood pressure, most importantly the stiffness of the arteries. Thus, in older people with stiff arteries and wide pulse pressures, the mean arterial pressure may be significantly biased. Furthermore, it has to be taken into account that systolic and diastolic pressures vary substantially in different parts of the arterial tree. In general, the systolic pressure increases in more distal arteries, whereas the diastolic pressure decreases.12
Guidelines of validation procedures of BHS, ESH and AAMI recommend inclusion of subjects according to the predefined categories of systolic and diastolic blood pressure. Pulse pressure, however, is not considered in the inclusion criteria. This study is limited by the fact that only one wrist device was tested. If the findings can be confirmed with other wrist devices, our data suggest that there may be a relevant bias of wrist blood pressure measurement in patients with advanced atherosclerosis and isolated systolic hypertension. These patients might not be adequately represented by current validation protocols.
Ohkubo T, Imai Y, Tsuji I, Nagai K, Kato J, Kikuchi N et al. Home blood pressure measurement has a stronger predictive power for mortality than does screening blood pressure measurement: a population-based observation in Ohasama, Japan. J Hypertens 1998; 16: 971–975.
O'Brien E, Petrie J, Littler W, de Swiet M, Padfield PL, O'Malley K et al. The British Hypertension Society protocol for the evaluation of automated and semi-automated blood pressure measuring devices with special reference to ambulatory systems. J Hypertens 1990; 8: 607–619.
O'Brien E, Pickering T, Asmar R, Myers M, Parati G, Staessen J et al. Working Group on Blood Pressure Monitoring of the European Society of Hypertension International Protocol for validation of blood pressure measuring devices in adults. Blood Press Monit 2002; 7: 3–17.
Association for the Advancement of Medical Instrumentation. American National Standard. Manual, electronic, or automated sphygmomanometers. In: ANSI/AAMI SP-10 (1992 and 2002) AAMI: Arlington, VA.
Westhoff TH, Schmidt S, Zidek W, van der Giet M . Validation of the Stabil-O-Graph blood pressure self-measurement device. J Hum Hypertens 2008; 22: 233–235.
Netea RT, Lenders JW, Smits P, Thien T . Arm position is important for blood pressure measurement. J Hum Hypertens 1999; 13: 105–109.
O'Rourke MF, Nichols WW . Aortic diameter, aortic stiffness, and wave reflection increase with age and isolated systolic hypertension. Hypertension 2005; 45: 652–658.
Franklin SS, Gustin WT, Wong ND, Larson MG, Weber MA, Kannel WB et al. Hemodynamic patterns of age-related changes in blood pressure. The Framingham Heart Study. Circulation 1997; 96: 308–315.
Wilking SV, Belanger A, Kannel WB, D'Agostino RB, Steel K . Determinants of isolated systolic hypertension. JAMA 1988; 260: 3451–3455.
van Popele NM, Bos WJ, de Beer NA, van Der Kuip DA, Hofman A, Grobbee DE et al. Arterial stiffness as underlying mechanism of disagreement between an oscillometric blood pressure monitor and a sphygmomanometer. Hypertension 2000; 36: 484–488.
Pickering TG, Hall JE, Appel LJ, Falkner BE, Graves J, Hill MN et al. Recommendations for blood pressure measurement in humans and experimental animals: part 1: blood pressure measurement in humans: a statement for professionals from the Subcommittee of Professional and Public Education of the American Heart Association Council on High Blood Pressure Research. Circulation 2005; 111: 697–716.
O'Rourke MF . From theory into practice: arterial haemodynamics in clinical hypertension. J Hypertens 2002; 20: 1901–1915.
We thank Mrs Christl Harsch for her organization of the measurements.
Sources of financial support
Conflict of interest
The authors declare no conflict of interest.
About this article
Cite this article
Westhoff, T., Schmidt, S., Meissner, R. et al. The impact of pulse pressure on the accuracy of wrist blood pressure measurement. J Hum Hypertens 23, 391–395 (2009). https://doi.org/10.1038/jhh.2008.150
- blood pressure measurement
- pulse pressure
- arterial stiffness