Physiological Measurement
COMMENT
Comment on ‘New photoplethysmogram indicators for improving cuffless and continuous blood pressure estimation accuracy’
Recent citations - Reply to Comment on ‘New photoplethysmogram indicators for improving cuffless and continuous blood pressure estimation accuracy’ Wan-Hua Lin et al
To cite this article: Noud van Helmond and Jeffrey I Joseph 2018 Physiol. Meas. 39 098001
View the article online for updates and enhancements.
This content was downloaded from IP address 147.140.233.30 on 16/10/2018 at 17:14
Physiol. Meas. 39 (2018) 098001(3pp)
https://doi.org/10.1088/1361-6579/aadf11
COMMENT
RECEIVED
21 May 2018
Comment on ‘New photoplethysmogram indicators for improving cuffless and continuous blood pressure estimation accuracy’
RE VISED
30 July 2018 ACCEP TED FOR PUBLICATION
5 September 2018 PUBLISHED
27 September 2018
Noud van Helmond1,2 and Jeffrey I Joseph1 1
Department of Anesthesiology, Thomas Jefferson University Hospital, Philadelphia, PA, United States of America Author to whom any correspondence should be addressed. 1020 Locust Street, Philadelphia, PA 19107, United States of America.
2
E-mail:
[email protected] Keywords: blood pressure measurement, pulse transit time, photoplethysmogram
Abstract Objective: A recent study investigated the potential utility of new photoplethysmogram (PPG) indicators to improve cuffless continuous blood pressure (BP) measurement. Approach: In this Comment we provide additional discussion of the physiology underlying PPG- and pulse arrival time (PAT)-based BP measurement. We also discuss validation of these measurements. Main results: Changes in PPG features and PAT can occur independently of BP. Any study validating BP measurement based on PPG features or PAT should include a variety of calibration independent physiological challenges affecting BP. Significance: PPG/PAT-based BP measurement has been investigated extensively as an alternative to cuff-based BP measurement. We feel the inherent physiological confounding in PPG/PAT-based measurement makes it unlikely that it will be of clinical value. We read with interest the recent study on new photoplethysmogram (PPG) indicators to improve cuffless continuous blood pressure (BP) estimation by Lin et al (2018). We agree with the authors that the development of accurate cuffless continuous BP monitors may improve rates of out-of-office BP measurement for adults with hypertension and that the value of such devices depends on the devices’ accuracy. We feel further discussion of the physiology underlying the employed methods and their validation is critical. The BP estimation algorithms that were used in the report were based on PPG features and pulse arrival time (PAT). The PPG feature that performed best as a single estimator of BP was the ascending slope of the 2nd derivative of the PPG and a related feature, the ascending slope of the 1st derivative of the PPG improved the performance of PAT-based estimation the most. Overall, the PAT model combined with the ascending slope of the 1st derivative of the PPG was clearly the most promising in terms of absolute precision and accuracy, although it would be informative to have bias measurements versus a true reference (intra-arterial measurement) (Kim et al 2014). As pointed out by the authors, PAT-based BP measurements have two major shortcomings; the first shortcoming is that PAT includes the pre-ejection period (PEP) in addition to pulse transit time (PTT). The PEP component depends on the ventricular electromechanical delay and isovolumetric contraction phase, which vary based on contractility and afterload. PEP can be shorter or longer at the same BP in the same individual, e.g. PEP shortens during dynamic exercise (Wong et al 2011) and elongates during peripheral vasoconstriction (Zhang et al 2011). The PTT component of PAT is dependent on the Moens–Korteweg equation, according to which PTT decreases as the arteries stiffen. Since arterial stiffness increases with BP via the mechanical properties of the arterial wall, PTT often shows an inverse relationship with BP. The shortcoming in the present study is that the PTT component represents the transit time through both arteries and arterioles, if it is measured at a distal location. Smooth muscle tone, mainly in arterioles, can be different at the same BP in the same individual, and thus induce PTT differences. For example, arm PTT increases despite no change in diastolic BP during vasodilation (Gao et al 2017). Importantly, the total bias induced in PAT by combined bias in PEP and PTT may be cumulative elongation (e.g. decreased contractility/peripheral vasodilation, vasovagal syncope), cumulative shortening (e.g. increased contractility/peripheral vasoconstriction, hypovolemia), or off-setting (e.g. increased contractility/peripheral vasodilation, exercise), depending on the physiological condition.
© 2018 Institute of Physics and Engineering in Medicine
Physiol. Meas. 39 (2018) 098001 (3pp)
N van Helmond and J I Joseph
In response to the ongoing interest in PAT-based BP estimation the Institute of Electrical and Electronics Engineers (IEEE) in 2014 released a guideline specifically for the validation of cuffless BP devices (IEEE 2014). This guidance, which has not been adopted (yet) by regulatory bodies, addresses some important differences between validation of cuffless BP monitors, that typically require calibration, versus validation of conventional cuff-type BP monitors, that do not require calibration. Because validation of a cuffless BP monitor immediately after calibration at the same BP could artificially increase its perceived accuracy, the IEEE protocol includes validation measurements after changes in BP from the calibration measurement and validation measurements after a significant time period since calibration. The accuracy requirements of the IEEE protocol reflect the established clinical guidelines for cuff BP monitors issued by the Association for the Advancement of Medical Instrumentation (ANSI/AAMI/ISO), which are currently used by the Food and Drug Administration (FDA) and the European Union for cuff-type BP monitors. The performance of the BP estimation models in the present study was evaluated during blood pressure changes induced by mental arithmetic stress and a Valsalva maneuver. It would be interesting to learn the performance of the algorithm during other interventions that have different confounding effects on total PAT through PEP and PTT, such as dynamic exercise, static exercise, pharmacological perturbations, and changes in body position (Payne et al 2006, Proenca et al 2010). The actual bias induced in the PAT only-estimate was fairly minimal; the best estimation model based on the combination of the ascending slope of the 2nd derivative of PPG and PAT achieved a relatively small decrease in estimation error of 0.84 ± 1.25 mmHg for systolic BP and 0.79 ± 0.63 mmHg for diastolic BP in comparison to the conventional PAT only-method. Moreover, a significant amount of the complete data set of each individual was used for training of the models. Assuming a normal heart rate of 72 beats per minute there were 684 beats in the whole 570 s experiment per subject in the study. Of these 684 beats, 256 were used to calibrate the models in each subject. The training/ calibration data set thus likely included data points from the mental stress tests and/or the Valsalva maneuver. As such, it is interesting that the best model in the study met conventional (ANSI/AAMI/ISO) bias and accuracy criteria (3 ± 7 mmHg difference versus reference), but it is unclear how it would perform after a truly novel blood pressure change in the same individual, independent of the training/calibration. The reproducibility of measurements after a significant time since calibration, e.g. 1 month (Watanabe et al 2017) or 6 months after calibration (Wong et al 2009), was not tested at all. The authors mention in the ‘Discussion’ section, that ‘blood pressure monitoring devices based on the proposed algorithms would be designed to be unobtrusive, miniaturized, portable and wearable, and could be used for long-term monitoring, which could provide a timely alarm when acute clinical events occur in daily life’. We agree that such continuous blood pressure measurement devices would truly be an improvement over current blood pressure measurement practices, but we feel the currently developed models have not been demonstrated to enable the implementation of such a device. Other research groups have tried to optimize PAT-based BP estimation by minimizing PEP confounding by using (PPG) PTT measurements between two distal points (Gao et al 2016) or by measuring and subtracting PEP from PAT (Martin et al 2016). To minimize the confounding effect of smooth muscle contraction and relaxation in small arteries on PTT there have been attempts to measure PTT in more central arteries, for example by using radar (Buxi et al 2017). It is important to note however that there is also smooth muscle modulation of tone in larger arteries, although to a lesser extent (Burton 1954). Moreover, these methods have not been miniaturized and cannot provide an accurate continuous measurement of BP to date. Therefore, it is difficult to minimize the confounding effects of smooth muscle contraction and relaxation in the small arteries on PTT. In conclusion, we applaud the authors’ effort to contribute to the advancement of cuffless blood pressure estimation by undertaking a challenging study. We feel, however, that the inherent physiological confounding of PAT-based BP estimation makes it difficult to envision that this type of measurement will have any clinical relevance in the near future, when compared to conventional validated BP measurement methods that do not contain inherent bias based on the conditions in which BP is measured.
Disclosures Jeffrey I Joseph, DO is a founder, equity owner, and has received research support from RTM Vital Signs, LLC, a company developing a non-invasive and long-term implantable vital sign monitoring system with real-time diagnostic algorithms and data transferred via cell phone to a central monitoring system. The other author has nothing to disclose.
References ANSI/AAMI/ISO 81060-2:2013 Non-invasive sphygmomanometers—part 2: clinical investigation of automated measurement type Burton A C 1954 Relation of structure to function of the tissues of the wall of blood vessels Physiol. Rev. 34 619–42 Buxi D, Redout J M and Yuce M R 2017 Blood pressure estimation using pulse transit time from bioimpedance and continuous wave radar IEEE Trans. Biomed. Eng. 64 917–27
2
Physiol. Meas. 39 (2018) 098001 (3pp)
N van Helmond and J I Joseph
Gao M, Cheng H M, Sung S H, Chen C H, Olivier N B and Mukkamala R 2017 Estimation of pulse transit time as a function of blood pressure using a nonlinear arterial tube-load model IEEE Trans. Biomed. Eng. 64 1524–34 Gao M, Olivier N B and Mukkamala R 2016 Comparison of noninvasive pulse transit time estimates as markers of blood pressure using invasive pulse transit time measurements as a reference Physiol. Rep. 4 e12768 IEEE 2014 Engineering in Medicine and Biology Society—IEEE Standard for Wearable, Cuffless, Blood Pressure Measuring Devices (New York: The Institute of Electrical and Electronics Engineers, Inc.) Kim S H, Lilot M, Sidhu K S, Rinehart J, Yu Z, Canales C and Cannesson M 2014 Accuracy and precision of continuous noninvasive arterial pressure monitoring compared with invasive arterial pressure: a systematic review and meta-analysis Anesthesiology 120 1080–97 Lin W H, Wang H, Samuel O W, Liu G, Huang Z and Li G 2018 New photoplethysmogram indicators for improving cuffless and continuous blood pressure estimation accuracy Physiol. Meas. 39 025005 Martin S L, Carek A M, Kim C S, Ashouri H, Inan O T, Hahn J O and Mukkamala R 2016 Weighing scale-based pulse transit time is a superior marker of blood pressure than conventional pulse arrival time Sci. Rep. 6 39273 Payne R A, Symeonides C N, Webb D J and Maxwell S R 2006 Pulse transit time measured from the ECG: an unreliable marker of beat-tobeat blood pressure J. Appl. Physiol. 100 136–41 Proenca J, Muehlsteff J, Aubert X and Carvalho P 2010 Is pulse transit time a good indicator of blood pressure changes during short physical exercise in a young population? Conf. Proc. IEEE Engineering in Medicine and Biology Society vol 2010 pp 598–601 Watanabe N et al 2017 Development and validation of a novel cuff-less blood pressure monitoring device JACC Basic Transl. Sci. 2 631–42 Wong M Y, Pickwell-MacPherson E, Zhang Y T and Cheng J C 2011 The effects of pre-ejection period on post-exercise systolic blood pressure estimation using the pulse arrival time technique Eur. J. Appl. Physiol. 111 135–44 Wong M Y, Poon C C and Zhang Y T 2009 An evaluation of the cuffless blood pressure estimation based on pulse transit time technique: a half year study on normotensive subjects Cardiovasc. Eng. 9 32–8 Zhang G, Gao M, Xu D, Olivier N B and Mukkamala R 2011 Pulse arrival time is not an adequate surrogate for pulse transit time as a marker of blood pressure J. Appl. Physiol. 111 1681–6
3