Influence of Sex on the Accuracy of Oscillometric-Derived Blood Pressures

Authors


George A. Stouffer, MD, Division of Cardiology, University of North Carolina, Chapel Hill, NC 27599-7075
E-mail: rstouff@med.unc.edu

Abstract

The effect of sex on the correlation between oscillometric and central aortic blood pressures (BP) is largely unknown. BP was simultaneously measured in the brachial artery using an oscillometric device and in the aorta using a fluid-filled catheter in 98 patients undergoing coronary angiography. Mean age (±standard deviation) was 58±12 years, with 55% of the patients being male. Mean BP (MBP) measured by the oscillometric device and in the aorta were similar when the group was examined as a whole, but the oscillometric device overestimated MBP in men and underestimated MBP in women. Oscillometric pressures accurately estimated diastolic BP in women but overestimated diastolic BP in men and underestimated systolic BP in both sexes. The oscillometric device underestimated aortic pulse pressure in both sexes but less in men than in women. The accuracy of oscillometric MBP and diastolic BP varied as a function of aortic MBP, but sex-related differences in the accuracy of oscillometric pressures remained significant after adjusting for MBP, body mass index, height, age, race, heart rate, diabetes, smoking status, and BP-lowering therapies using a multivariate logistic regression model. Sex is an important determinant of the accuracy of oscillometric BP. J Clin Hypertens (Greenwich). 2011;13:112–119. © 2010 Wiley Periodicals, Inc.

Hypertension, the most common primary diagnosis in the United States, is a major risk factor for cardiovascular (CV) disease. Recently it has been recognized that central aortic pulse and/or systolic pressures correlate better with major adverse CV outcomes than does pressure measured in the brachial artery. For example, in the Strong Heart Study, noninvasively measured central pulse pressure in a group of American Indians was more strongly related to vascular hypertrophy, extent of atherosclerosis, and CV events than was brachial blood pressure (BP) measured manually. Similarly, in the Aortic Blood Pressure and Survival study, central pulsatility (defined as aortic pulse pressure divided by mean aortic pressure) was a powerful predictor of CV events.1–3

There is also evidence that there are negative consequences when treatment decisions are made using peripheral rather than central BP. In the Conduit Artery Function Evaluation (CAFÉ) study of the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT), brachial BP was reduced to a similar degree in both the atenolol ± thiazide and in the amlodipine ± perinopril arms. However, central systolic and pulse pressures were reduced more in the amlodipine ± perinopril group and there were fewer CV events in that group.4

Oscillometric devices measure mean BP (MBP) and calculate systolic BP (SBP) and diastolic BP (DBP) based on a proprietary algorithm that is the same for all patients irrespective of age, height, sex, vascular disease, or other clinical variables. While a recent study showed that oscillometric MBP correlates well with central aortic MBP,5 the assumption that a common algorithm can be used to determine SBP and DBP for all patients may not be correct, as recent studies have shown that the accuracy of oscillometric pressure is dependent on height6 and age.5 The effect of sex on the accuracy of oscillometric-derived BP has not been determined despite studies showing sex-related differences in pulse pressure, arterial properties, and arterial wave reflection that are independent of age, mean arterial pressure, and body height.7,8

Methods

Patients

The study population consisted of 98 nonselected, noncontinuous patients who underwent cardiac catheterization at the University of North Carolina Memorial Hospital. Recruitment into the study was performed only when a research coordinator was available, which limited enrollment. Exclusion criteria included acute coronary syndromes, contraindication to the placement of BP cuff on either arm, arrhythmia, or upper extremity arterial disease. This study was approved by the institutional review board of the University of North Carolina at Chapel Hill.

Study Protocol

All recordings were obtained after the patient was in the supine position for at least 15 minutes while under sedation and prior to the injection of any contrast media. Three noninvasive BP measurements were obtained from each arm with simultaneous assessment of aortic pressure and, thus, for each patient, 6 aortic pressure measurements were recorded. Each inflation-deflation cycle required <1 minute for completion, and measurement recordings were separated by a 5-second pause (slightly more time was required when the BP cuff was switched to the alternate arm). Central aortic BP was measured via a 6 French fluid-filled pigtail catheter placed in the ascending aorta, attached to a strain gauge that was amplified by a commercially available catheterization laboratory system (Witt Series IV Physiomonitoring System, Philips Medical Systems, Boeblingen, Germany). The physiomonitoring system was balanced and calibrated before each patient was studied, in accordance with the manufacturer’s instructions. The zero reference was taken as the midaxillary line. Aortic MBP was obtained by the integrative method.

Oscillometric Pressure Measurement

Prior to catheterization, each patient was fitted with an appropriately sized BP cuff.9,10 Noninvasive BP recordings were taken from the brachial artery with the physiomonitoring system, which uses an oscillometric device employing stepwise pressure deflation. The device measures MBP (which is not reported), and then SBP and DBP are estimated using proprietary software. For the purposes of this study, MBP was calculated using two different formulas. The average oscillometric (±standard error of the mean) MBP was 93.5±1.5 mm Hg using MBP=DBP+1/3 (SBP−DBP) and 93.7±1.5 mm Hg using a formula that corrects for varying amount of time in diastole (MBP=DBP+[0.33+ (heart rate×0.0012)]×[SBP−DBP]).11 Since the formulas resulted in similar MBP (P=.83 for a difference in MBP as determined by the two formulas), only MBP as calculated by the second formula was used for the rest of the analysis. The oscillometric device used in this study met Association for the Advancement of Medical Instrumentation accuracy requirements, most recently updated in 2002,12 and was calibrated with a mercury standard according to manufacturer specifications prior to the start of the study (available at http://www.casmed.com/NE%20Specs%20022205.pdf).

Data Analysis

Mean SBP and mean DBP for each patient were calculated as the average of the 6 independent measurements. Paired t test analysis was performed on the differences in means to assess for statistical significance. Correlations between oscillometric and central aortic pressures were calculated using Pearson’s coefficient of correlation, and scatter plots described the pressure relationship between sites with a linear regression line. To test whether age and sex predicted differences between brachial pressure and central pressure, multivariate linear regression modeling was performed, adjusting for body mass index (BMI), height, race, heart rate, diabetes status, smoking status, and use of angiotensin-converting enzyme (ACE) inhibitors and β-blockers. All analyses were conducted with Stata Statistical Software release 9.0 (College Station, TX), with a 2-tailed P<.05 selected for the level of statistical significance.

Results

Patient Demographics and BP

The mean age (±standard deviation) of the 98 patients in the study was 58±12 years (range, 32–84 years). Fifty-five percent of patients were men, 90% were receiving BP-lowering therapy, and 29% had a diagnosis of diabetes mellitus (Table I). β-Blockers and ACE inhibitors were the two most commonly used BP-lowering therapies. Two thirds of patients had significant coronary artery disease as defined by a history of prior revascularization (percutaneous coronary intervention or coronary artery bypass surgery) or having an untreated lesion >70% stenosis in an epicardial artery or major branch.

Table I.   Demographics and Pharmacologic Treatment
CharacteristicTotal, %Men, %Women, %
  1. Abbreviations: ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; BMI, body mass index.

No.985444
Caucasian74 (76%)43 (80%)31 (70%)
African American24 (24%)11 (20%)13 (30%)
Weight, kg86±1992±1878±18
Height, m1.71±0.11.79±0.11.63±0.1
BMI, kg/m229.2±5.828.8±5.829.7±6.0
Heart rate, beats per min70±1069±1071±10
Hemoglobin, g/dL13.6±1.814.3±1.812.7±1.5
Coronary artery disease65 (66%)36 (66%)29 (66%)
Hypertension77 (79%)43 (80%)34 (77%)
Congestive heart failure20 (20%)10 (19%)10 (23%)
Diabetes mellitus28 (29%)13 (24%)15 (34%)
Hyperlipidemia65 (66%)35 (65%)30 (68%)
Current smoker23 (24%)16 (30%)7 (16%)
End-stage renal disease3 (3%)2 (4%)1 (2%)
Blood pressure–lowering therapy87 (89%)47 (87%)40 (91%)
 ACE inhibitor53 (55%)30 (56%)23 (52%)
 ARB13 (13%)7 (13%)6 (14%)
 β-Blocker59 (60%)34 (63%)25 (57%)
 Diuretic44 (45%)21 (39%)23 (52%)
 Calcium channel blocker15 (15%)7 (13%)8 (18%)
 α-Blocker4 (4%)3 (6%)1 (2%)
 Nitroglycerin33 (34%)18 (33%)15 (34%)
 Clonidine3 (3%)1 (2%)2 (5%)
 Minoxidil1 (1%)1 (2%)0

Central aortic SBP ranged from 79 mm Hg to 253 mm Hg and DBP ranged from 45 mm Hg to 104 mm Hg. There was minimal variation in the 6 measurements taken for each patient for central aortic SBP (1.5 mm Hg; P=.04) and no variation in central aortic DBP (0.3 mm Hg; P=.48). Oscillometric measurements of SBP and DBP in the brachial artery ranged from 87 mm Hg to 210 mm Hg and 42 mm Hg to 106 mm Hg, respectively. There was no difference between BPs measured in the left arm and in the right arm (P=.90 for difference in SBP and P=.72 for difference in DBP).

Correlation Between Oscillometric MBP and Aortic MBP and the Influence of Sex

Oscillometric and central aortic MBP were highly correlated (R=0.8738, P<.001) (Figure 1A) and there was no difference between the average oscillometric MBP and the average central aortic MBP (93.7±1.5 mm Hg vs 92.7±1.7 mm Hg; P=.58) (Figure 1B) when the group was examined as a whole. Sex influenced the accuracy of oscillometric estimates of MBP as oscillometric MBP significantly overestimated aortic MBP in men (93.8±2.0 mm Hg vs 89.1±1.8 mm Hg; P<.01) (Figure 1B) and significantly underestimated aortic MBP in women (92.7±2.6 mm Hg vs 97.1±2.8 mm Hg; P<.01).

Figure 1.

 Comparison between aortic and oscillometric mean blood pressure (MBP). (A) Correlation between aortic and oscillometric blood pressures in individual patients. (B) Oscillometric MBP and aortic MBP in all patients and stratified by sex.

There was a trend toward higher MBP in women compared with men (P=.052) in this study so we examined the relationship between accuracy of oscillometric MBP and level of aortic MBP. The amount of difference between oscillometric MBP and central aortic MBP varied as a function of aortic MBP (R=0.461; P<.0001) (Figure 2A), and when patients were divided into quintiles based on aortic MBP, there were small statistically significant differences between oscillometric MBP and central aortic MBP in the quintile of patients with the lowest central aortic MBP (7.3±2.2 mm Hg; P=.002) and in the quintile of patients with the highest central aortic MBP (−3.2±1.5 mm Hg; P=.03) (Figure 2B).

Figure 2.

 Effect of mean blood pressure (MBP) on accuracy of oscillometric blood pressure. (A) Relationships between aortic and oscillometric-derived MBP are shown as a Bland Altman plot and (B) when patients were divided into quintiles based on aortic MBP. Ranges of MBP in each quintile were 60.3 mm Hg to 78.2 mm Hg, 80.2 mm Hg to 88.1 mm Hg, 88.2 mm Hg to 94.2 mm Hg, 95.0 mm Hg to 106.6 mm Hg, and 107.0 mm Hg to 152.1 mm Hg, respectively. *P<.05 for comparison of oscillometric blood pressure vs central aortic BP.

In a multivariate logistic regression model, the sex-related difference in accuracy of oscillometric MBP remained significant after adjusting for value of MBP, BMI, height, age, race, heart rate, diabetes, smoking status, β-blocker usage, and BP-lowering therapies.

Effect of Sex on the Accuracy of Oscillometric Estimates of Central DBP and SBP

Oscillometric DBP was significantly greater than aortic DBP (5.2±0.8 mm Hg; P<.0001), with the slope of the linear regression line being <1.0. The accuracy of oscillometric DBP decreased with increasing central aortic MBP (R=0.26; P=.01) (Figure 3A) or aortic DBP (R=0.315; P<.002) (Figure 3B).

Figure 3.

 Comparison between aortic and oscillometric diastolic blood pressure (DBP) and systolic blood pressure (SBP). Differences between oscillometric and aortic (A) DBP or (C) SBP are plotted as a function of mean blood pressure (MBP). Bland Altman plots for (B) DBP and (D) SBP are also shown. All measurements are in mm Hg.

Similarly, there was a highly significant correlation between central aortic MBP and the difference between oscillometric SBP and central aortic SBP (Figure 3C). Overall, central aortic and oscillometric SBP were highly correlated (R=0.9140) (Figure 3D), with a mean relative error of 8.0%. We found that on average, oscillometric SBP was significantly less than aortic SBP (7.4±1.3 mm Hg; P<.0001) with the slope of the linear regression line being 1.18. Although brachial SBP may not equate with aortic SBP because BP waveforms change in amplitude with progression through the arterial tree, in general, brachial SBP is greater than or equal to aortic SBP.13

An examination of the influence of sex on the relationships between oscillometric and direct central measurements of DBP and SBP found that the device overestimated aortic DBP in men but not in women (6.8±0.7 mm Hg vs −3.1±1.4 mm Hg; P=.002) (Figure 4) and underestimated aortic SBP more in women than in men (−11.9±2.1 mm Hg vs −3.7±1.4 mm Hg; P=.003).

Figure 4.

 Differences between aortic and oscillometric blood pressures stratified by sex. *P<.05 for comparison of oscillometric mean blood pressure (MBP) vs central aortic MBP; bars and P values represent comparison of male vs female. SBP indicates systolic blood pressure; DBP, diastolic blood pressure; PP, pulse pressure.

The differences between men and women remained significant after adjusting for MBP, BMI, height, age, race, heart rate, diabetes, smoking status, β-blocker usage, and BP-lowering therapies using a multivariate logistic regression model.

Oscillometric Device Underestimated Pulse Pressure in Both Sexes

The oscillometric device underestimated central aortic pulse pressure by an average of 12.6±1.2 mm Hg (52.1±1.7 mm Hg vs 64.7±2.4 mm Hg; P<.001). Oscillometric pulse pressure underestimated central aortic pulse pressure in men by an average of 10.5±1.3 mm Hg (45.6±1.7 mm Hg vs 56.1±2.7 mm Hg; P=.001) (Figure 4) and in women by an average of 15.0±1.8 mm Hg (60.2±2.8 mm Hg vs 75.2±3.7 mm Hg; P=.004).

The accuracy of oscillometric pulse pressures varied as a function of MBP (R=0.47; P<.001) (Figure 5A), with a trend toward an inverse correlation between accuracy of oscillometric estimation of central aortic pulse pressure and heart rate (R=−0.18; P=.07) (Figure 5B). These results are consistent with the observations that (1) aortic MBP was overestimated in men and underestimated in women, and (2) pulse pressure was underestimated in both sexes using the oscillometric device.

Figure 5.

 Difference in oscillometric and aortic pulse pressure as a function of mean blood pressure (MBP) and heart rate. Difference between oscillometric-derived pulse pressure and directly measured aortic pulse pressure plotted as a function of (A) aortic MBP and (B) heart rate.

Prediction of Central Aortic MBP and Pulse Pressure Using Equations that Incorporate Sex and Oscillometric MBP as Variables

Based on these data, it follows that the prediction of aortic MBP, SBP, and DBP (and thus central aortic pulse pressure) for the patients in this study could have been improved by including sex and oscillometric MBP as shown in the following equations:

image
image
image
image

where G is a variable reflecting sex (G=1 if male and 0 if female).

Discussion

Our results show that the accuracy of the estimation of central aortic mean and pulse pressure by oscillometric-derived brachial artery BP varies based on sex and MBP. Similar to most, but not all, previous studies, we found that oscillometric devices underestimate SBP and overestimate DBP.5,14 Novel results of the current study are that MBP and sex have an important influence on these differences, with the oscillometric device overestimating aortic DBP in men but not in women and underestimating aortic SBP more in women than in men. The current study provides further evidence that use of a single formula by an oscillometric device to estimate BP can lead to significant inaccuracies and suggests that optimization of the algorithm, based on MBP and clinical factors, could result in improved estimation of aortic pulse pressure.

In comparing invasively measured aortic pressure with simultaneously measured oscillometric pressure from the brachial artery, we found that the oscillometric method accurately estimated central aortic MBP when the group was analyzed as a whole. This type of analysis, however, masked significant intraindividual variation (average absolute difference between brachial and aortic MBP of 6.9±6.1 mm Hg) and significant influences of MBP and sex on the accuracy of oscillometric BPs. Oscillometric BP overestimated aortic MBP in men (4.7±1.1 mm Hg; P<.01) and underestimated aortic MBP in women (−4.4±1.3 mm Hg; P<.01). These differences persisted after adjusting for confounding variables such as height, MBP, BMI, race, heart rate, diabetes, smoking status, β-blocker usage, BP-lowering therapies, and age.

Similar to our study, Lehmann and colleagues15 found significant differences between oscillometric-measured brachial BP (Dinamap model 1846SX; Critikon Inc, Tampa, FL) and aortic BP in patients undergoing cardiac catheterization. They found that the oscillometric device used in their study overestimated MBP (4.0±3.2 mm Hg), SBP (6.8±6.5 mm Hg), and DBP (4.9±3.0 mm Hg). Other studies examining the accuracy of oscillometric devices have reported varying results, but comparisons are limited by enrollment of a small number of patients,15–18 inclusion of few or no female patients,15,16,18 enrollment of patients from different clinical settings,17 variations in which artery was used to measure invasive BP (peripheral vs central aortic), and/or sponsorship by companies with a direct commercial interest in the results.18 Interpretation of all studies, including the present one, is limited by the proprietary nature of the algorithms used to derive systolic and diastolic pressure that vary among commercial oscillometric devices.19

Our results, together with results from prior studies, suggest that a single algorithm used by an oscillometric device may not be equally applicable to all patients. Studies have found that differences between oscillometric and central aortic pressures were not significantly influenced by cardiac index, systemic vascular resistance, heart rate, body surface area, or left ventricular ejection fraction,16 or weight.6 In contrast, Smulyan and colleagues5 found that age was an important determinant of the accuracy of oscillometric BP in a study of 100 patients undergoing coronary angiography. Using multiple regression analysis, they derived equations that better predicted aortic MBP and SBP by modifying cuff pressures by adjusting for age. In the current study, we found that sex and MBP were important determinants of the accuracy of oscillometric BPs. Taken together, these results suggest that it may be possible to increase the accuracy of oscillometric-derived SBP and DBP by modifying the algorithm based on sex, MBP, age, and possibly other variables. Furthermore, the clinical usefulness of oscillometric pressures might be increased even further by deriving a formula that correlates oscillometric pressures with aortic rather than brachial pressures.

Limitations

This study has several limitations. First, the sample size is relatively small although the differences reached statistical significance and the sample size is greater than that proposed by the guidelines of the European Society of Hypertension Working Group on Blood Pressure Monitoring.10 Second, patients were fitted with the appropriate-sized cuff but measurements of bicep diameter were not taken and controlled for in statistical analysis. Third, the group included only patients undergoing coronary angiography, and the application to patients without vascular disease is unknown. Fourth, the study used only one oscillometric device. And last, in the subgroup analysis, we adjusted for BMI, height, age, race, heart rate, diabetes, smoking status, use of β-blockers, and BP-lowering therapies, but there may be unknown confounders affecting the results. In addition, it should be acknowledged that although prediction of central aortic pressures may be improved by considering sex and MBP, values acquired with an oscillometric device are based on measurements in the brachial artery and may significantly differ from actual central pressures in individual patients. The increasing use of radial artery access for coronary angiography makes feasible future studies using direct comparison of invasively measured brachial artery pressure with oscillometric-derived brachial BP. These data would enable determination of the influence of sex and MBP on the accuracy of oscillometric-derived brachial BP.

Conclusions

Accurate noninvasive estimates of SBP and DBP are the basis for diagnosis and management of high BP. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7)20 defines elevated BP solely based on systolic and diastolic BPs, not on MBP, and, thus, inaccurate estimation of systolic and diastolic BP by an oscillometric device could have important clinical implications. Furthermore, despite progress in the estimation of central aortic pressure via noninvasive devices, recommendations on the treatment of hypertension are based on BPs measured in the brachial artery. Our results need to be validated in different and larger patient populations. Should that occur, this would provide a compelling rationale that inclusion of sex and MBP in the algorithms used to estimate DBP and SBP would enhance the accuracy of BP measurement by oscillometric devices.

Disclosures:  The authors have no conflicts of interest to disclose.

Ancillary