Abstract
Wave reflection at central arteries consists of a major component of left ventricular afterload. Central augmentation index (AIx) is the most widely used surrogate of wave reflection. Recent technological developments now provide the ability to obtain, non‐invasively, aortic, or carotid pressure waves and measure AIx based on various algorithms of pulse wave analysis. The aim of this study was to compare AIx measurements performed by the Arteriograph, Complior, and Mobil‐O‐Graph apparatuses. Recordings by each device in randomized order were performed with 5‐minute interval at 211 individuals (age 55.1 ± 14.1 years, 67.8% males) who underwent diagnostic cardiovascular assessment. All measurements were obtained at the supine position, and AIx was calculated using the formula AIx = 100 × (Augmentation pressure)/(Pulse Pressure). Bland‐Altman analysis was performed. Mean difference (bias) ± one standard deviation of difference (with limits of agreement) of AIx between different devices was as follows: (a) Mobil‐O‐Graph vs Complior: −2.1 ± 14.8% (−31.1% to 26.9%), (b) Arteriograph vs Complior: 12.9 ± 14.6% (−15.7% to 41.5%), and (c) Mobil‐O‐Graph vs Arteriograph: −10.8 ± 16.9% (−43.9% to 22.3%). The three examined devices exerted significant differences in central AIx estimation which makes the three devices non‐interchangeable for wave reflection assessment. However, the Mobil‐O‐Graph device showed the highest agreement (lowest bias) with the Complior system as regards to the AIx measurement.
Keywords: aorta, applanation tonometry, oscillometry, pulse wave analysis, wave reflection
1. INTRODUCTION
Wave reflection occurring during the cardiac cycle, especially at the level of ascending aorta, is an important determinant of left ventricular load and coronary blood flow. Especially, the reflected waves arriving at the ascending aorta at early systole augment systolic blood pressure (SBP) and thereby increase ventricular load.1 In contrast, occurrence of the reflected waves during late systole or even diastole is highly desirable because they increase mean diastolic blood pressure (DBP), promoting coronary perfusion.
Wave reflections depend upon three major factors: (a) the distance to the reflecting site(s), (b) the speed of wave transmission which is predominantly determined by arterial stiffness, and (c) the intensity of the reflected waves.1, 2 Wave reflections are generated in several regions of impedance mismatch (ie, due to arterial diameter or elasticity changes) in the arterial tree. Reflected waves are combined and form a relatively discrete global reflected wave which arrives at the proximal aorta either prematurely, during early systole, leading to a secondary systolic pressure peak and increased central pulse pressure, or delayed during late systole after the systolic pressure peak.1, 3
Augmentation index (AIx), defined as the ratio of the augmentation of SBP to pulse pressure (PP), is the most widely applied surrogate of wave reflection. For its calculation, the merging point of the incident (forward) and the reflected (backward) wave (inflection point) needs to be identified. According to previous observations,4, 5 this inflection point corresponds to the peak of flow velocity and several algorithms have been developed for its identification.3, 4
Nowadays, several commercial devices exist for the non‐invasive measurement of central blood pressure and AIx,6 either directly at the carotid artery (without the use of transfer functions or mathematical model) or indirectly at the aortic level by use of mathematical transformation of peripheral arterial pulses. These apparatuses also use different techniques for arterial pulse acquisition such as applanation tonometry or automated oscillometry. Finally, different computational algorithms are used for aortic wave derivation (when necessary) and pulse wave analysis for the determination of AIx. All these methodological differences are evident at the Complior, Arteriograpgh, and Mobil‐O‐Graph devices which are commonly used for central blood pressure and wave reflection estimation. Nonetheless, the consistency and agreement of central AIx measurement by Mobil‐O‐Graph compared with other older existing technologies (namely Arteriograph and Complior) are unknown, and therefore, we aimed to explore them in the present study.
2. METHODS
The study population consisted of 211 individuals (age 55 ± 14 years, 67.8% males, 36% smokers) who underwent diagnostic cardiovascular assessment in the 2nd Department of Cardiology of our Institute. Out of them, 22.7% had diabetes, 50.2% had hypertension, 29.9% had hyperlipidemia, 29.9% had coronary artery disease, and 26.1% had inflammatory disease.
All measurements were performed by a single experienced operator (medical doctor) as previously described.7 Each subject rested in a supine position for 10 minutes in a quiet room at 23°C before the recording of the baseline hemodynamic measurements. The study protocol was in accordance with the ethical standards of Helsinki Declaration, and it was approved by Attikon hospital's scientific committee. All individuals provided informed consent prior to study enrollment.
Brachial BP and heart rate (HR) were measured in the right arm with an automated digital oscillometric sphygmomanometer (TensioMed, Ltd). Brachial blood pressure was measured before using each device (Complior, Arteriograph, or Mobil‐O‐Graph). Measurements of AIx by the three devices were performed in a randomized order as illustrated in Figure 1, with at least 5 minutes of interval between measurements. For the cuff‐based devices (Arteriograph and Mobil‐O‐Graph), arm circumferences were first measured for the proper cuff size selection (two sizes were available: 24‐34 and 32‐42 cm).
Figure 1.
Each individual was assigned to a sequelae of augmentation index measurements in a randomized fashion using orange: Arteriograph, blue: Complior, and green: Mobil‐O‐Graph device
2.1. Augmentation index measurement
Augmentation index was calculated by using the formula AIx = 100 × (Augmentation pressure)/(Pulse Pressure), as previously described.8, 9 For this purpose, the central (carotid or aortic) pulse waveforms were analyzed.
2.1.1. Complior device
The automated system Complior (Alam Medical) was used to record carotid pulse waves and calculate AIx. The apparatus uses pressure sensitive transducers that record distension waveforms over superficial arteries (ie, carotid, femoral, etc). This system measures directly central/carotid pressure waveforms without the use of mathematical models or transfer function. The recorded carotid signals were calibrated using the mean and diastolic pressure measured at the brachial level.7
2.1.2. Arteriograph device
The Arteriograph (TensioMed Kft.) device uses an upper arm cuff equipped with a sensitive a sensor. The cuff was applied to the subject's upper arm, and after a first BP measurement, the cuff was over‐inflated with a pressure 35‐40 mm Hg above the recorded systolic pressure. During systole, the blood volume ejected into the aorta generates pulse wave (early systolic peak). This pulse wave propagates toward the periphery, and it is reflected at the bifurcation of the aorta, creating a second wave (late systolic peak). Both early and late systolic peaks were obtained and recorded on the computer as pulse waves. The software of Arteriograph decomposes the early, late systolic and diastolic waves and also determines the onset and peaks of the waves as previously described.10
2.1.3. Mobil‐O‐Graph device
The Mobil‐O‐Graph apparatus (IEM GmbH) is based on the recording of brachial pressure wave using a conventional arm cuff and the oscillometry principle. At first, the device measures the brachial systolic and diastolic blood pressure, and after that, the cuff (equipped with a high fidelity pressure sensor) re‐inflates at the level of the diastolic blood pressure for approximately 10 seconds. The sensor records brachial pressure waveforms. An aortic pulse wave is then derived by using a generalized transfer function. Several pressure and cardiovascular parameters including AIx are computed by using the ARCSolver algorithm as previously described.11
2.2. Statistics
The continuous variables were expressed as mean ± one standard deviation. Distribution normality of the three AIx measurements was evaluated by Kolmogorov‐Smirnov test. Difference in mean AIx values measured by the three devices was compared between different devices by ANOVA for repeated measurements using Bonferroni correction for multiple comparisons. Linear correlations between different AIx readings were assessed by Pearson or Spearman correlation coefficient as appropriate. Agreement between AIx values estimated by the tested apparatuses was further evaluated by the intraclass correlation coefficient (ICC) as previously described.12, 13 Mean differences and standard deviation of differences (SDD) between different AIx measurements (Mobil‐O‐Graph vs Complior, Arteriograph vs Complior and Mobil‐O‐Graph vs Arteriograph) were also calculated. The limits of agreement between different measurements of AIx were defined as follows: lower limit = mean difference − 1.96 × SDD and upper limit = mean difference + 1.96 × SDD. Bland‐Altman analysis was performed as previously described.14 P‐values of .05 were considered to represent statistical significance, and all tests were two‐sided. Statistical analysis was performed by IBM SPSS Statistics for Windows, version 23.0. (IBM Corp).
3. RESULTS
Augmentation index measurements performed by the three devices for the total population are reported in Table 1. Paired comparisons of AIx values between different devices showed that the Arteriograph apparatus overestimated AIx, significantly, compared to Complior (P < .001) and Mobil‐O‐Graph (P < .001). In contrast, Mobil‐O‐Graph and Complior provided statistically comparable AIx values (P = .121).
Table 1.
Augmentation index measurements by Complior, Mobil‐O‐Graph, and Arteriograph devices
| Augmentation Index (%) | Mean ± SD | Minimum | Maximum | 95% CI |
|---|---|---|---|---|
| Complior | 15.7 ± 11.2 | 0 | 49.4 | 14.1‐17.2 |
| Mobil‐O‐Graph | 17.8 ± 11.3 | −6 | 47 | 16.2‐19.3 |
| Arteriograph | 28.5 ± 12.7 | −8 | 49.8 | 26.8‐30.5 |
Abbreviations: CI, confidence interval; SD, standard deviation.
Augmentation index measured by Arteriograph was significantly correlated with AIx by Complior (r = .254, P < .001), but not with AIx measured by Mobil‐O‐Graph (r = .007, P = .924). AIx by Complior was marginally correlated with AIx by Mobil‐O‐Graph (r = .135, P = .05). The ICC between AIx estimates by the examined devices was below 0.4 indicating poor agreement according to previous classification and interpretation of ICC values.13
We additionally performed correlation analysis (using Spearman correlation coefficient and ICC) as well as Bland‐Altman analysis (mean and SD of differences) for the AIx measurements by the three different apparatuses, in a subgroup of 37 healthy subjects (mean age 45 ± 12 years, 51.4% females). Measurements of AIx by Arteriograph and Complior were significantly correlated (r = .677, P < .001 and ICC = .557, P = .015). No significant correlation was observed between AIx readings by Complior vs Mobil‐O‐Graph (r = −.004, P = .982 and ICC = .03, P = .467) and by Arteriograph vs Mobil‐O‐Graph (r = −.09, P = .63 and ICC=−.098, P = .6). Mean ± SD of differences in AIx readings by the tested devices was 15 ± 27.6% between Mobil‐O‐Graph vs Complior, −8.2 ± 17.5% between Mobil‐O‐Graph vs Arteriograph, and 23.2 ± 21.8% between Arteriograph vs Complior.
The mean difference ± one standard deviation of difference (with limits of agreement) of AIx between different devices was as follows: (a) Mobil‐O‐Graph vs Complior: −2.1 ± 14.8% (−31.1% to 26.9%), (b) Arteriograph vs Complior: 12.9 ± 14.6% (−15.7% to 41.5%), and (c) Mobil‐O‐Graph vs Arteriograph: −10.8 ± 16.9% (−43.9% to 22.3%). Bland‐Altman plots of AIx differences between different devices are illustrated in Figure 2. It was observed that the mean difference did not vary remarkably according to the mean AIx level for all comparisons. However, a wide range of limits of agreement (high SDD) was note for all three devices when compared between each other.
Figure 2.
Bland‐Altman plots of augmentation index (AIx) differences between different devices (black thick line indicates mean difference, whereas the dashed lines represent the upper and lower limit of agreement)
4. DISCUSSION
The present study, to the best of our knowledge, compared for the first time the measurement of central AIx using a novel automated, oscillometric, cuff‐based sphygmomanometer (Mobil‐O‐Graph) against two other existing technologies (Arteriograph and Complior). Notably, all these technologies combine different principles for pulse wave acquisition (oscillometry and tonometry), and different approaches for non‐invasive estimation of central pulse waves (direct recording of carotid pressure waveforms and aortic wave derivation by use of transfer functions and mathematicals models). Therefore, it is currently unknown whether AIx measurement by this new apparatus is interchangeable with AIx values derived by other available technologies. The findings of the present study provide novel information supporting that they are not.
This study demonstrated that, overall, there is a low agreement and consistency in central AIx measurement among the three examined techniques. This observation is documented by the significant bias in AIx readings between different technologies which is accompanied by a quite wide range of the reported limits of agreement, and by the weak correlations observed between the different AIx measurements. The closest agreement of AIx readings was observed between Mobil‐O‐Graph and Complior as shown by the lowest observed bias (−2.1%). In contrast, the lowest agreement of AIx measurements was observed between Arteriograph and Complior, as indicated by the highest observed bias (12.9%). Similar observations were made for AIx measurements performed only in healthy individuals.
Similar comparisons of AIx estimations by different devices have been previously reported,15, 16, 17, 18, 19 evaluating various methodologies based on: (a) different pulse acquisition technique (ie, tonometry, oscillometry, photoplethysmography), (b) different signal processing techniques (ie, with or without transfer function, algorithms for inflection points identification), and (c) different arterial sites for wave reflection estimation (ie, finger, aorta, radial, brachial, carotid arteries) (Table 2). It should be acknowledged that all these differences render the interpretation and explanation of the observed discrepancies in AIx measurements among various devices quite difficult.
Table 2.
Key points from available literature data discussed in the present study, concerning augmentation index (AIx) measurements by different devices and techniques
| Year | Author(s) | Tested devices | Sample | Wave reflection index(es) | Key outcome |
|---|---|---|---|---|---|
| 2009 | Jatoi et al20 | Arteriograph, Complior, SphygmoCor | 254 untreated hypertensive patients | Aortic AIx | “Although Arteriograph values of AIx are in close agreement with corresponding parameters obtained by Complior and SphygmoCor, the techniques are not Interchangeable” |
| 2011 | Dhindsa et al16 | SphygmoCor, Omron, Itamar | 40 healthy subjects | Aortic, radial, and finger AIx | “There were high and significant correlations between AIx values even though the absolute values derived by each technique were different” |
| 2011 | Kips et al17 | SphygmoCor, Omron | 143 rural black South Africans | Radial and central AIx | “Although the average AIx was very similar in Omron and SphygmoCor‐waves, large individual differences in AIx were observed” |
| 2012 | Zhang et al19 | SphygmoCor, PulsePen, and A‐Pulse | 66 patients | Carotid and radial AIx | “Tonometry‐based devices were not consistent in measurements of central BP and wave reflections in clinical practice, with considerable and significant differences among them” |
| 2014 | Agnoletti et al15 | SphygmoCor, PulsePen | 38 patients | Central AIx | “No difference in cAIx between the two devices” |
| 2016 | Peng et al18 | SphygmoCor (cuff), SphygmoCor (tonometer) | 82 subjects with treated hypertension | Central AIx | “The agreement was good with a small mean difference and no systematic bias” |
| 2019 | Papaioannou et al (present study) | Mobil‐O‐Graph, Arteriograph, Complior | 211 subjects (37 healthy) | Central AIx | “The three examined devices exerted significant differences in central AIx estimation which makes the three methods non‐interchangeable” |
Dhindsa et al16 compared AIx at peripheral and central arteries derived from SphygmoCor, Itamar, and Omron devices, in 40 apparently healthy subjects. The Omron and Itamar apparatuses measured peripheral AIx at the radial artery and finger, respectively, whereas SphygmoCor provided both radial and aortic measures of AIx. Nevertheless, that study examined only the correlations between AIx measures.16 Another group explored differences in peripheral (radial) AIx measured by SphygmoCor and Omron HEM‐9000AI apparatuses, concluding that although the average AIx was very similar between the examined devices, large individual differences in AIx readings were observed.17
Another study compared wave reflection indices obtained only from tonometry‐based devices (SphygmoCor, PulsePen, and A‐Pulse) in 63 patients.19 In line with our findings, it was observed that the examined tonometry‐based devices were not consistent in measurements of AIx, exhibiting considerable and significant differences among them.19 However, the limits of agreement in carotid AIx measurement by SphygmoCor and PulsePen (around −15% to 26%) were somewhat narrower than the limits of agreement reported in our study. In another comparison study, Agnoletti et al15 found that PulsePen and SphygmoCor devices are not equivalent and provide different wave shapes (despite similar harmonics content) with a greater discrepancy observed on radial than on carotid derived parameters. Jatoi et al20 also assessed AIx measurements by three commercial devices (Arteriograph, Complior, and SphygmoCor) in 254 untreated hypertensive patients. The authors concluded that the three devices are not interchangeable for the AIx measurement.20
Finally, Millasseau et al21 used a single tonometry‐based device (SphygmoCor) and examined the agreement in AIx measurements obtained: (a) directly at the carotid artery and (b) indirectly at the aorta after transformation of the recorded radial pressure waveforms. It was found that the agreement between aortic AIx estimated from the transformed radial and transformed carotid waveforms was relatively poor with a mean difference and mean difference −6.2 ± 9.2% for the total examined population.21 Similar results were obtained when aortic AIx obtained from the transformed radial pressure waves was compared with aortic AIx estimated directly from the carotid artery (without using a transfer function).21
A few limitations of our study should be acknowledged. Unfortunately, our study design does not allow us to derive any conclusion concerning the accuracy or/and precision of these devices, neither to point out any technology or technique as the gold standard for AIx measurement. Furthermore, our methodology did not involve any intervention that might change AIx, and thus, it was not possible to assess and compare the ability of these technologies to detect changes in AIx values at the same individual.
In conclusion, the three examined devices exert significant differences in central AIx estimation which means that the three methods are not interchangeable. The Mobil‐O‐Graph device showed the highest agreement (lowest bias) with the Complior system as regards to central AIx measurement. Still, among the non‐invasive available technologies for central AIx estimation, there is no established gold standard method. A consensus on this matter would further assist toward the standardization of validation protocols concerning wave reflection estimation, the enhancement of the accuracy of AIx measurements and possibly the advancement of its diagnostic and prognostic accuracy.
CONFLICT OF INTEREST
None.
AUTHOR CONTRIBUTIONS
TGP, JT, DB, HT, GK, GP, FK, KK, DV, VL, EP, JP, DT, and II substantially contributed to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work. TGP, JT, DB, HT, GK, GP, FK, KK, DV, JP, DT, and II drafted the work or revising it critically for important intellectual content. TGP and II involved in final approval of the version to be published. TGP and II agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Papaioannou TG, Thymis J, Benas D, et al. Measurement of central augmentation index by three different methods and techniques: Agreement among Arteriograph, Complior, and Mobil‐O‐Graph devices. J Clin Hypertens. 2019;21:1386–1392. 10.1111/jch.13654
REFERENCES
- 1. Vlachopoulos C, O'Rourke M. Genesis of the normal and abnormal arterial pulse. Curr Probl Cardiol. 2000;25:303‐367. [DOI] [PubMed] [Google Scholar]
- 2. Mitchell GF, Lacourciere Y, Arnold JM, Dunlap ME, Conlin PR, Izzo JL Jr. Changes in aortic stiffness and augmentation index after acute converting enzyme or vasopeptidase inhibition. Hypertension. 2005;46:1111‐1117. [DOI] [PubMed] [Google Scholar]
- 3. Nichols WW, O'Rourke M. McDonald's Blood Flow in Arteries: Theoretical, Experimental and Clinical Principles, 5th edn. London, UK: Taylor & Francis Ltd; 2005. [Google Scholar]
- 4. Kelly R, Hayward C, Avolio A, O'Rourke M. Noninvasive determination of age‐related changes in the human arterial pulse. Circulation. 1989;80:1652‐1659. [DOI] [PubMed] [Google Scholar]
- 5. Murgo JP, Westerhof N, Giolma JP, Altobelli SA. Aortic input impedance in normal man: relationship to pressure wave forms. Circulation. 1980;62:105‐116. [DOI] [PubMed] [Google Scholar]
- 6. Papaioannou TG, Protogerou AD, Stamatelopoulos KS, Vavuranakis M, Stefanadis C. Non‐invasive methods and techniques for central blood pressure estimation: procedures, validation, reproducibility and limitations. Curr Pharm Des. 2009;15:245‐253. [DOI] [PubMed] [Google Scholar]
- 7. Benas D, Kornelakis M, Triantafyllidi H, et al. Pulse wave analysis using the Mobil‐O‐Graph, Arteriograph and Complior device: a comparative study. Blood Press. 2019;28:107‐113. [DOI] [PubMed] [Google Scholar]
- 8. O'Rourke MF, Pauca A, Jiang XJ. Pulse wave analysis. Br J Clin Pharmacol. 2001;51:507‐522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Barraclough JY, Garden FL, Toelle B, et al. Sex differences in aortic augmentation index in adolescents. J Hypertens. 2017;35:2016‐2024. [DOI] [PubMed] [Google Scholar]
- 10. Echeverri D, Pizano A, Cabrales J, Moreno K. Validation of central and peripheral non‐invasive hemodynamic variables using an oscillometric method. High Blood Press Cardiovasc Prev. 2018;25:65‐77. [DOI] [PubMed] [Google Scholar]
- 11. Wassertheurer S, Kropf J, Weber T, et al. A new oscillometric method for pulse wave analysis: comparison with a common tonometric method. J Hum Hypertens. 2010;24:498‐504. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Bartko JJ. The intraclass correlation coefficient as a measure of reliability. Psychol Rep. 1966;19:3‐11. [DOI] [PubMed] [Google Scholar]
- 13. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med. 2016;15:155‐163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307‐310. [PubMed] [Google Scholar]
- 15. Agnoletti D, Millasseau SC, Topouchian J, Zhang Y, Safar ME, Blacher J. Pulse wave analysis with two tonometric devices: a comparison study. Physiol Meas. 2014;35:1837‐1848. [DOI] [PubMed] [Google Scholar]
- 16. Dhindsa M, Barnes JN, DeVan AE, Sugawara J, Tanaka H. Comparison of augmentation index derived from multiple devices. Artery Res. 2011;5:112‐114. [Google Scholar]
- 17. Kips JG, Schutte AE, Vermeersch SJ, et al. Comparison of central pressure estimates obtained from SphygmoCor, Omron HEM‐9000AI and carotid applanation tonometry. J Hypertens. 2011;29:1115‐1120. [DOI] [PubMed] [Google Scholar]
- 18. Peng X, Schultz MG, Abhayaratna WP, Stowasser M, Sharman JE. Comparison of central blood pressure estimated by a cuff‐based device with radial tonometry. Am J Hypertens. 2016;29:1173‐1178. [DOI] [PubMed] [Google Scholar]
- 19. Zhang Y, Agnoletti D, Safar ME, et al. Comparison study of central blood pressure and wave reflection obtained from tonometry‐based devices. Am J Hypertens. 2013;26:34‐41. [DOI] [PubMed] [Google Scholar]
- 20. Jatoi NA, Mahmud A, Bennett K, Feely J. Assessment of arterial stiffness in hypertension: comparison of oscillometric (Arteriograph), piezoelectronic (Complior) and tonometric (SphygmoCor) techniques. J Hypertens. 2009;27:2186‐2191. [DOI] [PubMed] [Google Scholar]
- 21. Millasseau SC, Patel SJ, Redwood SR, Ritter JM, Chowienczyk PJ. Pressure wave reflection assessed from the peripheral pulse: is a transfer function necessary? Hypertension. 2003;41:1016‐1020. [DOI] [PubMed] [Google Scholar]