Comparison of Two Noninvasive Devices for Measurement of Central Systolic Blood Pressure With Invasive Measurement During Cardiac Catheterization

Abstract

J Clin Hypertens (Greenwich). 2012; 14:575–579. © 2012 Wiley Periodicals, Inc.

Recently, a new device for noninvasive assessment of central systolic blood pressure (cSBP) (BPro device with A‐Pulse) was approved by the US Food and Drug Administration, but available data are limited. In 52 patients undergoing invasive elective cardiac evaluation, central hemodynamics were measured invasively. Immediately thereafter, radial artery waveforms were sampled by two noninvasive techniques, the BPro and, as a comparator, the SphygmoCor System. Then, central hemodynamics were measured invasively for a second time. The invasively recorded cSBP (137±27 mm Hg) did not differ with both noninvasively assessed cSBP by BPro (136±21 mm Hg, P=.627 vs invasive cSBP) and by SphygmoCor (136 ± 23 mm Hg, P=.694 vs invasive cSBP) and correlated highly between invasively recorded and both noninvasively assessed cSBP. However, using Bland‐Altman plots, spreading of compared data of both devices can be found (BPro: 0.87±13 mm Hg vs invasive cSBP; SphygmoCor: 0.77±14 mm Hg vs invasive cSBP). There was an excellent correlation of both noninvasive devices for the calculation of cSBP (r=0.961, P<.001). cSBP differed by only 0.1±6 mm Hg (P=.913) between the two noninvasive devices. Therefore, both noninvasive devices showed an accurate agreement in cSBP compared with invasively measured cSBP.

Although the importance of hypertension as a predictor of cardiovascular (CV) risk has been known for a long time, the relative role of the individual components of peripheral blood pressure (BP) is still a matter of debate. ¹ , ² , ³ , ⁴ Currently, however, the importance of central hemodynamics to more precisely assess the pressure load on the CV system becomes more obvious. This is driven by two important points. First is the opportunity by noninvasive devices (eg, SphygmoCor; Atcor Medical, Sydney, New South Wales, Australia) to instantaneously generate from the peripheral arterial pressure waveforms the corresponding central (aortic) pressure waveforms ⁵ , ⁶ , ⁷ , ⁸ and to quantify central pulse pressure (PP) and augmentation index (AIx). Second are the findings of previous studies that central hemodynamics are independent predictors of clinical outcomes. ⁹ Moreover, in the Strong Heart Study it was shown that central PP relates more strongly to vascular disease and outcome than (peripheral) brachial PP. ¹⁰ This was confirmed by Huang and colleagues, ¹¹ who found that central BP predicts more all‐cause and CV mortalities better than peripheral BP. Furthermore, recent studies have also shown substantial differences in the ability of antihypertensive drugs to affect central in contrast to peripheral hemodynamics and cardiovascular risk. ⁹ , ¹²

Most recently, another device for noninvasive assessment of central systolic BP (cSBP) (BPro device with A‐Pulse; HealthSTATS, Singapore) was approved by the US Food and Drug Administration (K072593). This device potentially offers the opportunity to measure cSBP under ambulatory conditions over 24 hours.

Data on the validity of obtained information on central hemodynamics with this novel device were lacking, however, and no comparison between assessed cSBP with this device and invasively measured central hemodynamics, which is the gold standard of central hemodynamics measurement, has been published at the time of our study.

Methods

Study Population

The study population was composed of 52 adult patients from the Department of Cardiology (Hospital of Nuremberg) who were undergoing cardiac catheterization for diagnostic and/or therapeutic purposes between July and August 2009. Patients were consecutively enrolled. Exclusion criteria were any arrhythmias that would disturb the required sinus rhythm during pulse wave recordings, eg, atrial fibrillation.

Study Protocol

Patients undergoing invasive elective cardiac evaluation were tagged prior to cardiac catheterization with a standard oscillometric BP device (Dinamap Pro 100V2; Criticon, Norderstedt, Germany) and the BPro device on the same arm. Arterial access was established in all patients via the A. femoralis dextra (6F pigtail catheter), ie, there was no interference with brachial measurements. The first invasive measurement (fluid‐filled manometer system) of central hemodynamics was done. Immediately thereafter, brachial systolic BP and diastolic BP (for calibration of both devices) by oscillometry BP device were measured. Then, radial artery waveforms were sampled by noninvasive techniques using BPro with A‐Pulse and, thereafter, in the same arm, by noninvasive technique with the SphygmoCor System. Subsequently, central hemodynamics were measured invasively and values were averaged. All measurements were performed prior to the cardiac evaluation and done before any drugs were infused intravenously. At the beginning and the end of the study, calibration of cardiac catheter was verified with a mercury sphygmomanometer.

BPro with A‐Pulse

In brief, the A‐Pulse central aortic systolic pressure (CASP) pulse wave application software (version C2.6.0 ML) was used with the BPro pulse wave acquisition device attached to the patient’s wrist (like a wristwatch) to capture and display the radial pulse wave in real time. Radial artery waveform was averaged from single waveforms recorded consecutively for 10 seconds per block of waveforms. From the radial pulse wave, the software estimates the CASP using an n‐point moving average method (NPMA), a mathematical low pass filter. It was shown that an NPMA with a denominator of one quarter of the tonometer sampling frequency (here: 60 Hz) accurately defines CASP when applied to noninvasivley acquired calibrated radial artery pressure waveforms (RAPWFs). The NPMA is designed to accurately derive CASP and does not generate an aortic waveform. Thus, it cannot be used to derive other indices such as central augmentation index.¹³

SphygmoCor System

In brief, radial artery waveforms were recorded from the radial artery at the wrist, using high‐fidelity applanation tonometer (Millar Instruments, Houston, TX) directly into the SphygmoCor System. Radial artery waveform was averaged from single waveforms recorded consecutively for 8 seconds. Corresponding central (aortic) waveforms were then automatically generated from the radial artery waveform by a validated transfer function. ⁵ , ⁶ , ¹⁴ From the derived central waveforms, data are given for central hemodynamics. All recordings included in the analysis had high‐quality, defined as in‐device quality index >80% (as derived from an algorithm that includes average pulse height; pulse height variations; diastolic variations; and the maximum rate of rise of the peripheral waveform).

Statistical Analyses

Normal distribution of data was confirmed by Kolmogorov‐Smirnov tests before further analyses. None of the parameters studied differed from normal distribution. Hence, data were compared by paired and unpaired Student t tests and are expressed as mean±standard deviation (SD). Two‐tailed values of P<.05 were considered statistically significant. Univariate correlation analyses were performed using Pearson’s correlation coefficient. Bland‐Altman plots were used to assess agreement between methods. All analyses were performed using SPSS 16.0 (SPSS Inc, Chicago, IL).

Results

Clinical characteristics of the entire study population are given in the Table. Medical treatment was not withheld prior to the measurements. There was no significant difference between first and second invasive measurement of cSBP (138±28 vs 136±28 mm Hg, P=.139). There was no difference in the results whether BPro was compared with the first invasive measured and SphygmoCor was compared with the second invasive measured central hemodynamics and vice versa (data not shown). Therefore, invasively measured cSBP was averaged for comparisons with the noninvasive measurements.

Table TABLE.

 Clinical Characteristics

Age, y	63.7±11
Sex, male/female	30/22
Weight, kg	81.3±19
Height, cm	169±9
Body mass index, kg/m2	28.1±5.0
Casual systolic BP, mm Hg	147±24
Casual diastolic BP, mm Hg	79±11
Smoker, %	34
Hypertension, %	90
Diabetes mellitus, %	25
Hypercholesterolemia, %	62
CHD, %
One vessel	29
Two vessels	21
Three vessels	19

Abbreviations: BP, blood pressure; CHD, coronary heart disease. Values are expressed as mean±standard deviation unless otherwise indicated.

Comparison Between Invasive and Noninvasive Central Pressure Measurements

There was no difference between the invasively recorded cSBP (137±27 mm Hg) and both noninvasively assessed cSBP by BPro (136±21 mm Hg, P=.627 vs invasive cSBP) and by SphygmoCor (136±23 mm Hg, P=.694 vs invasive cSBP). Moreover, there was a high correlation of cSBP between invasively recorded and both noninvasive calculated devices (1, 2).

Figure 1.

 Left: Scatterplot of invasively recorded central systolic blood pressure (cSBP) vs noninvasively assessed cSBP by BPro. Right: Bland‐Altman comparisons of the same data. The line shows the mean difference of the measurements and the dotted lines show the 2 standard deviations (SDs) from the mean difference.

Figure 2.

 Left: Scatterplot of invasively recorded central systolic blood pressure (cSBP) vs noninvasively assessed by SphygmoCor. Right: Bland‐Altman comparisons of the same data. The line shows the mean difference of the measurements and the dotted lines show the 2 standard deviations (SDs) from the mean difference.

Using Bland‐Altman plots, however, which is more precise for agreement between methods, spreading of compared data of both devices can be found (BPro: 0.87±13 mm Hg vs invasive cSBP; SphygmoCor: 0.77±14 mm Hg vs invasive cSBP). Moreover, an increasing difference between invasive and both noninvasive measurements with increasing BP levels was found (1, 2).

Comparison Between the Noninvasive Measurements of BPro and SphygmoCor

Figure 3 depicts the excellent correlation of both noninvasive devices for the calculation of cSBP. Given in absolute values, cSBP differed in only 0.1±6 mm Hg (P=.913) between both noninvasive devices. This high agreement is also be evident by Bland‐Altman plot.

Figure 3.

 Left: Scatterplot of both noninvasively assessed central systolic blood pressure (cSBP) by BPro and by SphygmoCor. Right: Bland‐Altman comparisons of the same data. The line shows the mean difference of the measurements and the dotted lines show the 2 standard deviations (SDs) from the mean difference.

Discussion

The aim of the study was to test the validity of cSBP assessment with the BPro device with A‐Pulse compared with cardiac catheterization, the gold standard of measuring central hemodynamics. We showed that there was both an excellent agreement of the absolute values (137±27 mm Hg vs 136±21 mm Hg), eg, no shift of baseline values, with the BPro device with A‐Pulse and a high correlation (r=0.893, P<.001) between invasively and noninvasively assessed cSBP by BPro with A‐Pulse. Similar findings were obtained with respect to the comparisons between invasively recorded cSBP and noninvasively assessed cSBP with SphygmoCor, which is probably the most widely used and accepted device in clinical practice for the assessment of cSBP. However, we have to keep in mind that although the strengths of the obtained relationship is reassuring, the correlation coefficients do not indicate agreement and are potentially misleading when comparing two different techniques. Hence, we also used Bland‐Altman plots, which are accepted as the standard method of comparing two different measurements of biological variables. ¹⁵ These analyses (1, 2) disclose that the corresponding systolic BP values are widespread; hence, the level of accuracy by the Association for the Advancement of Medical Instrumentation (AAMI), ¹⁶ the British Society of Hypertension, ¹⁷ and the European Society of Hypertension (ESH) ¹⁸ would not be fulfilled. According to ESH recommendations, 19% of the differences have to be ≤5 mm Hg, 60%≤10 mm Hg, and 83%≤15 mm Hg. These recommendations have been made for the noninvasive validation of two devices with respect to peripheral BP measurements. No recommendations about validation of central hemodynamics, namely cSBP and cDBP, are currently available. An extensive literature search about validation studies for the noninvasive estimation of central hemodynamics yielded that other studies also exhibited a wide scatter of values. Previously, Smulyan and colleagues have shown similar results, namely a high correlation of invasive and noninvasive (SphygmoCor) estimated cSBP (r=0.89), as well as a mean difference of only 1.5 mm Hg, but with a large scatter (SD= 11.13  mm Hg). In accordance with our study, calibration of the noninvasive device has been carried out with oscillometric measurements of brachial BP. ¹⁹ In accordance, Zuo and associates found a nonsignificant difference between estimated cSBP (SphygmoCor) and invasively measured cSBP but a high SD of about 17 mm Hg. They concluded that the inaccurate measurement of peripheral (brachial) BP was the major limiting factor. ²⁰ However, we have to mention that previously, Williams and colleagues published a validation study using an n‐point moving average method for assessment of cSBP, which is the underlying concept of the BPro device with A‐Pulse, with invasive‐measured cSBP in 20 patients. They demonstrated an excellent correlation and agreement. ¹³

According to AAMI, a cuff BP device is “substantially equivalent” to the reference device (for the validation) if the mean difference is ≤5 mm Hg and the SD ≤8 mm Hg. However, it must be taken into account that the reference device (for the calibration) is also “substantially equivalent” if the mean difference is ≤5 mm Hg and the SD ≤8 mm Hg compared with gold‐standard intra‐arterial pressure. ¹⁶ , ²¹ Hence, errors may sum up resulting in a possible larger scatter. However, we have used a well accepted standard oscillometric BP device (Dinamap Pro 100V2; Criticon, Norderstedt, Germany) and checked the devices inclusive the catheter manometer system carefully. In a previous study comparing peripheral BP values obtained with Arteriograph (TensioMed Ltd, Budapest, Hungary), another noninvasive device for the assessment of central hemodynamics, and measured brachial BP with a digital oscillometric device (Omron, Model HEM705‐CP, Kyoto, Japan), a high correlation of peripheral systolic BP (r=0.81, P<.001) was found, but the agreement with mean difference of 4.9 mm Hg and an SD of about 11 mm Hg was relatively poor. ²²

In the literature, good agreement between the measurements of cSBP, were found if the noninvasive device (SphygmoCor) was calibrated with the invasively recorded BP. ⁶ , ²³ We feel such an approach is not justified. Noninvasvive brachial measurements rather than invasive BP should be taken for calibration since such an approach reflects the clinical situation for which the device was designed.

Limitations

One limitation of our study is that the cuff pressure and aortic pulse analysis were not monitored simultaneously. However, because BP can differ between the left and right arms, we thought that it was absolutely mandatory to record cuff pressure and radial pulse waves in the same arm; thereby, it was impossible to record both values simultaneously. In addition, it was irrelevant whether noninvasively assessed cSBP was compared with the first, second, or averaged invasive‐measured cSBP, indicating no systemic bias due to the nonsimultaneous measurement of invasive and noninvasive assessed cSBP.

Furthermore, we investigated the accuracy between the BPro device with A‐Pulse compared with the SphygmoCor device, the most accepted and widely used noninvasive device for assessment of central hemodynamics. We found an excellent correlation (r=0.961, P<.001) and a high agreement (Figure 3) and the absolute values of cSBP differed only in 0.1±6 mm Hg (P=.913) between both noninvasive devices, therefore being under the thresholds for mean and SD recommended by AAMI ¹⁶ and the level of accuracy (71% of the differences were ≤5 mm Hg, 87%≤10 mm Hg and 96%≤15 mm Hg) according to the ESH, with the limitations as mentioned above (principles of validation), would be satisfied. ¹⁸

In accordance with our findings, a previously published study from Garcia‐Ortiz and colleagues ²⁴ also found an excellent correlation (r=0.937, P<.01) for CASP with BPro and SphygmoCor in healthy Caucasians without drug treatment, indicative of an adequate validity compared with the used reference standard (SphygmoCor).

Conclusions

Our data suggest that the BPro device with A‐Pulse is an accurate noninvasive device for the assessment of cSBP in clinical routine use. However, like any other noninvasive device for the assessment of central hemodynamics, this method is limited by the variability and accuracy of peripheral (brachial) BP measurement. It is noteworthy to mention that the BPro device (similar to the Mobil‐O‐Graph 24‐hour PWA Monitor; I.E.M. GmbH, Stolberg, Germany) potentially offers the opportunity to measure cSBP under ambulatory conditions over 24 hours. This point may be of crucial importance since 24‐hour ambulatory BP measurements as well as central BP are superior in terms of predicting target organ damage and outcomes over office BP measurements. Therefore, the advantages of both techniques may be combined for a better approach to BP measurements. ²⁵

Disclosures:  The authors report no specific funding in relation to this research and no conflicts of interest to disclose.