|Year : 2019 | Volume
| Issue : 3 | Page : 109-118
Evaluating the accuracy of the VitalWellness device
Nicole Polanco1, Sharon Odametey2, Seyyedeh Neda Derakhshani1, Mark Khachaturian3, Connor Devoe1, Kamal Jethwani2, Sujay Kakarmath2
1 Connected Health Innovation, Partners HealthCare; Department of Dermatology, Massachusetts General Hospital, USA
2 Connected Health Innovation, Partners HealthCare; Department of Dermatology, Massachusetts General Hospital; Harvard Medical School, USA
3 Vital USA, USA
|Date of Web Publication||30-Dec-2019|
Partners Healthcare, 25 New Chardon St., 3rd Floor, Suite 300, Boston, 02114, Massachusetts
Source of Support: None, Conflict of Interest: None
Background and Objective: Portable and readily accessible wellness devices can aid vital sign measuring for those interested in tracking their health. In this diagnostic accuracy study, we evaluated the performance of the VitalWellness device (VW), a wireless, compact, noninvasive device that measures four vital signs (VS) – blood pressure (BP), heart rate (HR), respiratory rate (RR), and body temperature (BT) – using the index finger and forehead. Methods: Adult volunteers with VS that fell both within and outside of the normal physiological range were enrolled to provide BP, HR, RR, and BT measurements using both the VW and Food and Drug Administration-approved reference devices. A subgroup of participants underwent an additional test to analyze the VW's performance on HR and RR outside of normal physiological ranges. Statistical measurements were plotted on scatter and Bland–Altman plots. Sensitivity analyses to evaluate the VW's performance by gender, skin color, finger size, and auxiliary activities were performed. Results: A total of 263 participants completed the study. On an average, systolic BP measured using the VW was 10 mmHg lower than that of the reference device (correlation coefficient r = 0.7), whereas diastolic BP was 3 mmHg lower (r = 0.6), and RR was 2 bpm lower (r = 0.7). VW HR and BT measurements were, on average, 1 bpm and 0.3°F higher than the corresponding reference measurements (r = 0.9 and r = 0.3), respectively. Conclusion: The VW device is well-suited for home-based, nonmedical monitoring of HR, RR, and BP. Further improvement in measurement accuracy is required to enable applications for medical use.
Keywords: Cuff-less blood pressure monitor, monitoring, vital signs, Wellness device
|How to cite this article:|
Polanco N, Odametey S, Derakhshani SN, Khachaturian M, Devoe C, Jethwani K, Kakarmath S. Evaluating the accuracy of the VitalWellness device. Digit Med 2019;5:109-18
|How to cite this URL:|
Polanco N, Odametey S, Derakhshani SN, Khachaturian M, Devoe C, Jethwani K, Kakarmath S. Evaluating the accuracy of the VitalWellness device. Digit Med [serial online] 2019 [cited 2020 Mar 28];5:109-18. Available from: http://www.digitmedicine.com/text.asp?2019/5/3/109/274384
| Introduction|| |
Wellness devices for remote health tracking and monitoring have been reported to help up to 56% of the United States population living with chronic conditions to actively manage their disease(s). In the US, patients living with chronic conditions account for about 75% of all healthcare expenditures per year: over $1 trillion dollars annually. Even among healthy individuals this trend is growing, driven largely by the availability of a variety of health tracking tools and mobile devices such as smartphones and smartwatches. National data provided by the Pew Research Center indicates that “69% of U.S. adults keep track of at least one health indicator.” Access to tracking and monitoring tools may help individuals become more aware of their own health status and may facilitate change toward healthy behaviors.,,
When monitoring patient health, health-care providers (HCP) depend on vital signs (VSs) – including height, weight, blood pressure (BP), heart rate (HR), respiratory rate (RR), and body temperature (BT). Regular VS monitoring may allow HCP to detect issues early so they can provide the care needed to mitigate negative health outcomes and contain health-care costs., VS measurement can be a time- and people-intensive process, requiring several sessions per day. Moreover, hospital monitoring devices require staff to transfer data to a patient's electronic medical record, introducing opportunities for human error.
Advances in technology have provided multiple tools to monitor VS remotely, thereby allowing HCPs to closely monitor patients outside of the hospital. However, most commercially available VS remote monitoring devices target a single VS measurement, and users may need up to five different devices. Using multiple devices to transmit data to caregivers is cumbersome and may cause delays in reporting. Having a single device that can perform multiple measurements, within a short span of time may offer a better experience for users.
Vital USA has developed a wireless device – the VitalWellness (VW) – to measure four VS using the index finger and forehead. The device can transmit information electronically and is part of the VW platform, which also includes the Vital app. Using a single, small wireless device that transmits data in real time through Bluetooth to a smartphone application could enable individuals to be more involved in managing their health. The device is intended to monitor VS outside of the medical environment; can be used remotely, without trained clinical staff; and reduces personnel and equipment needed to measure VS.
| Methods|| |
The primary aim of this study was to assess the accuracy of the VW in measuring four VS (BP, HR, RR and BT) in comparison with the Food and Drug Administration (FDA)-approved reference devices typically used in a hospital environment. The study protocol was approved by the Partners Healthcare Institutional Review Board and is registered on the NIH clinical trials website (NCT03589716).
The VitalWellness device
The VW is a compact VS measurement device that uses proprietary sensor technology to non-invasively measure four VS [Figure 1]. BP, HR, and RR are measured from a person's left index finger and BT at the forehead. Results are transmitted through Bluetooth to a smartphone which displays the results via a mobile application. The app collects and tracks data to show longitudinal trends, and it also guides users on how to capture their VS measurements. The VW is connected to a smartphone for storage and has a USB interface for charging, thereby requiring no electrical connection during use. Measurements are taken with the device, independent of the phone.
BP is calculated with an inflatable finger cuff and dynamic light scattering (DLS) sensor subsystem which measures blood flow after occlusion to calculate systolic BP (SBP) and diastolic BP (DBP). HR is calculated by photoplethysmography (PPG)-based subsystem from the ratio of 660 (nm) and 940 (nm) light absorption in the finger. RR is evaluated through a combination of DLS and PPG subsystems. BT is calculated through an FDA-510(k)-approved infrared sensor and proprietary BT algorithm.
The FDA-approved Welch Allyn 767 Mobile Aneroid Sphygmomanometer and the 3M Littman Master Classic II Teaching 40-inch Stethoscope, used together, measured BP. The Welch Allyn Connex Spot VS Monitor measured HR and BT. The Capnostream™ Portable Bedside Capnograph obtained EtCO2 waveforms for RR assessment, which were then read by a physician.
Sample size was determined based on consensus guidelines recommended by the International Standardization Organization (ISO) for each VS [Table 1]., A total of up to 80 participants with VS within the normal physiological range and up to 190 participants with VS outside of the normal physiological range were required to participate. Assuming that up to 5% of the study participants who attended the in-person enrollment visit might withdraw their consent during the study; we estimated the final sample size at 284.
Study volunteers were recruited from the public and clinical practices within the Partners HealthCare System (PHS). The study was advertised on three online portals: The PHS clinical trials website, Craigslist.com and the NIH clinical trials website. Study investigators also sought patients from the Wellesley Family Care Associates clinic, the Massachusetts General Hospital Revere Health Center, and the Brigham and Women's Hospital Division of Renal Medicine. In addition, study staff sent out letters and made recruitment phone calls to study volunteers from previous studies at PHS who had consented to being contacted for future research.
Volunteers age ≥18 years with enough understanding of the English language to understand and consent to study procedures were eligible to participate. The study participants with physical disabilities that precluded safe and adequate testing, and those with implanted or body-worn electronic medical devices were excluded. For measurements using exercise or aerobic step testing, the following study volunteers were excluded: VS outside of normal physiological range during initial measurements; age ≥65 years; self-report of food, alcohol or caffeine consumption within 3 h of testing; pregnant; self-reported history of fall or fracture in the past year; those with arthritis, dizziness, neurological illness, chest pain, myocardial infarction, heart disease, stroke, transient ischemic attack; self-report of being under active treatment for anemia, electrolyte imbalance or hyperthyroidism; and mental impairment with limited ability to cooperate. Participants with nail polish or artificial nail on the left index finger were excluded from HR measurements. The following exclusions applied for the paced breathing test: VS outside of normal physiological range; history of seizures or epilepsy; panic attacks; history of respiratory or heart disease; hyperthyroidism; and/or pregnancy.
The study participants were instructed not to consume food, alcohol, or caffeine in the 30-min period before the study visit.
Participants were enrolled in a quiet, temperature-controlled room. To enable the assessment of variability in measurements due to differences in finger size, a Mudder Ring Gauge was used to measure participants' left index finger size. Skin color was assessed using the NIS Skin Color Scale, which required trained study staff to make a subjective judgment of skin color and assign a score based on the scale. VS were measuredfirst with the VW, then with the respective reference device. Different staff members obtained measurements using the VW and the reference devices, respectively, and each was blinded to measurements taken by the other.
Study staff recorded any VW device failure as a “failed measurement” while collecting VS measurements. This included connection loss between device and cellphone, device freezing or turning off during testing, and errors in participants' finger position.
Blood pressure measurement
Participants placed their left index finger inside of the VW finger cuff, held the device in their palm, and placed it on their chest at heart level for 3 min. Two consecutive measurements were taken. Thereafter, two reference BP measurements were obtained by two staff simultaneously. The ISO 2009 guidelines were followed using the aforementioned sphygmomanometer and stethoscope.
Per VW user guidelines, if two consecutive VW measurements for a given participant differed by ≥12 mmHg for SBP or by ≥8 mmHg for DBP, the complete set of measurements (including the reference device measurements) was discarded. Otherwise, thefirst measurement was recorded as the VW value for SBP and DBP, and the second one was discarded. Similarly, if the two reference measurements differed by ≥4 mmHg, the set of measurements was discarded, and a fresh set was obtained after a 10-min wait.
Heart rate measurement
Using the VW and the Welch Allyn Connex Spot Monitor one after the other, HR was administered three times on the participant's index finger for 1 min each.
To obtain HR values >130 beats/min, eligible participants underwent either a cycle ergometer test or an aerobic step test.
- Exercise physiologists used the Borg Scale to assess the participant's perceived exertion on a Scifit 100R Cycle Ergometer every 3 min. The resistance level of the ergometer was increased by 10 watts every 3 min until the participant reached a perceived exertion level of “somewhat hard”
- The Harvard step test protocol was used for aerobic step testing. The reference device was placed on the participant's finger to enable real-time monitoring of HR. Participants were asked to step up and down in synchrony with a metronome set at 96 beats/min. The procedure lasted for 5 min, or until the participant's HR reached 140 beats/min.
Respiratory rate measurement
RR was measured using the Capnostream™ Monitor and the VW simultaneously. The participant breathed into a nasal prong capnography sensor that captured exhaled air. The EtCO2 waveform was then printed and read by a physician to confirm the RR value.
To obtain RR values >30 breaths/min, eligible study participants underwent a paced breathing test. They matched their breathing to a metronome set variably between 60 and 80 beats/min (that is, 30–40 breaths/min) for up to 100 s.
To measure temperature, study staff placed the VW 0.5–1” away from a participant's forehead and held the power button for 1 s. Staff then placed the temperature probe in the posterior pocket under the participant's tongue to obtain BT measurements.
To test the variability of VS measurements, a sub-sample of participants (n = 15) were asked to cough (n = 5), drink water (n = 5) or talk (n = 5) while VS were measured in accordance with the procedures outlined above.
We calculated correlation coefficients, mean, standard deviation, mean difference, standard deviation of mean difference, and standard error of the mean difference with 95% confidence intervals for measurements obtained by the VW and respective reference devices. The measurements were plotted on a scatter plot and a Bland–Altman plot.
We conducted exploratory data analysis and identified data points that might bias the estimated correlation between the VW and reference devices. For BT measurements, implausible data points (<96°F or >104°F) were identified and excluded. For BP measurements, a complete set of measurements was excluded if the reference device measurements were ≥12 mmHg for SBP or ≥8 mmHg for DBP. Thereafter, data points having the greatest influence on the correlation coefficient were identified using Cook's distance and excluded from the final set of measurements, using a cutoff value of 0.05.
For sensitivity analyses, mean difference with 95% confidence intervals was calculated for subgroups based on gender, finger size, skin color, and the presence of auxiliary activities. We compared the mean differences between subgroups using an independent sample t-test.
| Results|| |
The study enrolled 265 participants, with two withdrawing before the study completion [Figure 2]. The total sample (n = 263) were aged 18–97, with a median of 38 years old. Nearly two-thirds of participants were female, most were Caucasian, and two-thirds self-reported their educational status as “college or post college.” More than half of the participants self-reported their occupation as “employed,” followed by “student” (14%) and “unemployed” (14%). About three-fourths of the study participants were subjectively assigned an NIS skin color score of 1–3 by study staff, 20% of them were given skin color scores of 4–6, and <4% were determined to have a skin color score >6 [Table 2].
Statistical measurements were plotted on scatter and Bland–Altman plots to show correlation between VW and reference device measurements [Figure 3] and [Figure 4]. [Table 3] and [Figure 5] show which VS measurements were included in the final analysis. [Table 4] shows the mean values and differences for VS measured by reference device and VW. The sensitivity analysis did not suggest any differences in the performance of the VW by gender, skin color, or auxiliary activities [Table 5] and [Table 6]. However, the mean difference of the RR – but not the other VS – between the reference and VW was lower for larger finger sizes.
|Figure 3: Correlation between vital sign measured using the reference device and that measured with VitalWellness. r = Pearson's correlation coefficient|
Click here to view
|Figure 4: Bland-Altman plots of the difference in vital sign measurements between the reference device and VitalWellness. Solid lines indicate mean difference and dotted lines indicate limits of agreement|
Click here to view
|Table 4: Mean values and difference for vital signs measured by reference device and VitalWellness|
Click here to view
|Table 5: Differences in vital sign measurements between the reference device and VitalWellness in the study participant subgroups|
Click here to view
|Table 6: Differences in vital sign measurements between the reference device and the VitalWellness in the study participant subgroup|
Click here to view
| Discussion|| |
We compared measurements for four VS from the VW in comparison with the FDA-approved reference devices and conclude that the measurements correlate well for HR, RR, and BP.
HR obtained from the VW are similar in accuracy to those reported for commercially-available wrist-worn devices and superior to those reported for smartphone-based applications that use optical input from the phone's camera to determine HR. The accuracy of the VW for SBP and DBP is similar to that of smartphone-based BP monitoring through the oscillometric finger-pressing method, but lower than FDA-approved home BP monitoring devices.,,
The accuracy of the VW for RR measurements is lower than that reported for continuous RR monitors designed for hospital use as well as that reported for human observers by counting for a full minute or counting for 15 s and quadrupling the value.,,, We did not find any peer-reviewed validation reports on wellness devices that measure RR.
The estimated correlation of the VW for BT was lower than that observed for other VS. However, the mean difference in temperature measurements is quite low compared to the reference device and suggests that the correlation coefficient may have been impacted by low variability within obtained temperature measurements. The reported accuracy of commercially-available wireless dermal temperature measurement devices is higher than that of the VW. However, the mean difference in temperature for the VW is substantially lower than that reported for temporal artery thermometers.
Previous clinic-based studies have found accuracy (difference of >5 mmHg) of SBP measured using home BP devices to differ systematically by age, gender, and size of cuff used. Similarly, HR with wrist-worn sensors can vary based on skin tone. The accuracy of the VW for all VS did not seem to be influenced by gender, skin tone, finger size, or auxiliary activities.
Strengths and limitations
The study has several limitations. First, 75% of readings collected for BP were not included in the final analysis to mitigate bias from observer error, within-person variability in BP, and statistical outliers, as well as to comply with VW device usage guidelines, and hence, we may have overestimated the magnitude of the correlation coefficient. Second, RR measurements outside of normal physiological ranges were obtained artificially by instructing study participants to synchronize their breathing to a metronome. This artificially reduced the variability in the data and possibly, underestimated the RR correlation coefficient. Similarly, the range of values obtained from temperature measurements was narrow and may have underestimated the correlation coefficient. Fourth, the race and ethnicity of participants do not represent the general population of the United States. However, results from the sensitivity testing by gender, skin color, and finger size suggest that the device may perform with similar accuracy in a more diverse sample of participants. Finally, the laboratory environment, in which the measurements were conducted by trained staff is unlike a person's home. However, as the companion app guides the user at home, this is unlikely to affect the accuracy of the device.
To provide a greater number of usable BP readings in future studies, a reassessment of study procedures could be undertaken, including rest time before taking BP measurements. In addition, obtaining more than three BP readings per participant may be considered. RR measurements outside of the physiological range included in the analysis were obtained through voluntary efforts by the participant in this study. Future studies should consider recruiting participants who naturally have RR outside of the physiological range.
There are two key strengths to this study. First, as far as we know, this is thefirst large-scale demonstration of successful BP measurement from the finger. Previous studies have tested mobile applications and devices for BP but have enrolled a substantially smaller sample size, 35–85 subjects compared to 263 in this study. One study tested a BP app that required participants to use their fingers to press on the smartphone's screen and camera. This method unlike the VW, required participant's date of birth, sex, height, and weight. Users may not always have this information available to accurately use the app. Another study tested a small sensor attached to a smartphone to collect BP via the user's finger. Although the study compared the device's performance to an FDA approved BP measuring device, they did not follow rigorous recruitment guidelines to make sure that they were accounting for a wide range of possible BP measurements. Furthermore, this oscillometric method does not provide nighttime BP measurements, limiting the access to BP information. Another study that validated a cuffless device to collect BP showed promising results when tested against validated oscillometric BP monitors. However, the device does not provide DBP measurements, limiting the amount of information accessible to users. BP cuffless devices have mixed results, but like the VW, cuffless BP devices may provide a compact and efficient way for patients to measure their BP. Measurement of BP using the arm is procedurally complex, has low reliability and is associated with poor patient adherence. Furthermore, the performance of the device for BP and other VS is likely to improve substantially once it is commercially available and data from a larger user pool is available to improve the algorithm underlying the measurements. Finally, the study methods used for data collection and analysis, were rigorous and followed protocols and guidelines recommended by the FDA for evaluation of medical devices.
| Conclusion|| |
Current methods of measuring VS require multiple devices and clinical staff to record these measurements. In contrast, the VW allows for fast and accurate measurement of key VS with minimal app-guided training. It can serve as a convenient tool for users who monitor and track their VS measurements remotely.
The VW device is well-suited for home-based, nonmedical monitoring of HR, RR, and BP. Further studies with more data variability are needed to conclusively assess the accuracy of BT readings from the device, as well as to evaluate usability and acceptability of the device by end-users.
We would like to thank the following staff for helping recruit and enroll participants into the study: Stephen Agboola, M. D, M. P. H, Mursal Atif, Amanda J. Centi, P. h. D, Odeta Dyrmishi, Felicia Estrada, Harriett Gabbidon, Margaret Gorini, R. N, Saafia Masoom, Jessica Odotei, R. N, Ramya Palacholla, M. D, M. P. H. Emily de Redon and Nina Schussler. The following staff provided financial and material support by tracking and requesting sponsor payments: Colin Gorman. The following staff provided support during the editing phase of the manuscript draft: Julie R. Brown. The following staff provided their clinics and patients for enrollment in the study: Mallika L. Mendu, M. D, M. B. A, Jeffrey W. Phillips, M. D, Sushrut S. Waikar, M. D and Amy Wheeler, M. D. The following department and staff provided the equipment and their expertise necessary to administer the exercise test for the study: The MGH Cardiovascular Disease Prevention Center.
Financial support and sponsorship
Vital USA financially supported this research study.
Conflicts of interest
The only conflict of interest to declare within the manuscript is that the author, Mark Khachaturian, Ph.D., is employed by Vital USA, the company who financially sponsored the study.
| References|| |
Kutz D, Shankar K, Connelly K. Making sense of mobile – And web-based wellness information technology: Cross-generational study. J Med Internet Res 2013;15:e83.
Paré G, Jaana M, Sicotte C. Systematic review of home telemonitoring for chronic diseases: The evidence base. J Am Med Inform Assoc 2007;14:269-77.
Yilmaz T, Foster R, Hao Y. Detecting vital signs with wearable wireless sensors. Sensors (Basel) 2010;10:10837-62.
Kellett J, Sebat F. Make vital signs great again – A call for action. Eur J Intern Med 2017;45:13-9.
Appelboom G, Camacho E, Abraham ME, Bruce SS, Dumont EL, Zacharia BE, et al
. Smart wearable body sensors for patient self-assessment and monitoring. Arch Public Health 2014;72:28.
Weenk M, van Goor H, Frietman B, Engelen LJ, van Laarhoven CJ, Smit J, et al
. continuous monitoring of vital signs using wearable devices on the general ward: Pilot study. JMIR Mhealth Uhealth 2017;5:e91.
Ludikhuize J, Smorenburg SM, de Rooij SE, de Jonge E. Identification of deteriorating patients on general wards; measurement of vital parameters and potential effectiveness of the modified early warning score. J Crit Care 2012;27:424.e7-13.
Fanucci L, Saponara S, Bacchillone T, Donati M, Barba P, Sanchez-Tato I, et al
. Sensing devices and sensor signal processing for remote monitoring of vital signs in CHF patients. EEE Trans Instrum Meas 2013,62:553-69.
Dias D, Paulo Silva Cunha J. Wearable health devices-vital sign monitoring, systems and technologies. Sensors (Basel) 2018;18:2414.
Kaur A, Student ME. A review of remote patient monitoring system: Potentials, challenges and current issues. International Journal of Trend in Research and Development 2016;3:2394-933.352.
George J, MacDonald T. Home blood pressure monitoring. Eur Cardiol 2015;10:95-101. [Doi: 10.15420/ecr.2015.10.2.95].
International Organization for Standardization. Medical Electrical Equipment-Particular Requirements for Basic Safety and Essential Performance of Clinical Thermometers for body Temperature Measurement. (ISO 80601-2-56). International Organization for Standardization; 2017.
International Organization for Standardization. Non-Invasive Sphygmomanometers – Part 2: Clinical Investigation of Automated Measurement Type (ISO 81060-2:2013). Geneva: International Organization for Standardization; 2013.
Massey DS, Martin JA. 2003. The NIS Skin Color Scale. Switzerland: International Organization for Standardization; 2013.
Harvey S. The No Sweat Exercise Plan: Lose Weight, Get Health, and Live Longer (Harvard Medical School Guides). Blacklick: McGraw-Hill; 2006.
Chandrasekhar A, Kim CS, Naji M, Natarajan K, Hahn JO, Mukkamala R. Smartphone-based blood pressure monitoring via the oscillometric fi nger-pressing method. Sci Transl Med 2018;10:431.
Ruzicka M, Akbari A, Bruketa E, Kayibanda JF, Baril C, Hiremath S. How accurate are home blood pressure devices in use? A cross-sectional study. PLoS One 2016;11:e0155677.
Hodgkinson JA, Sheppard JP, Heneghan C, Martin U, Mant J, Roberts N, et al
. Accuracy of ambulatory blood pressure monitors: A systematic review of validation studies. J Hypertens 2013;31:239-50.
Shcherbina A, Mattsson CM, Waggott D, Salisbury H, Christle JW, Hastie T, et al
. Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort. J Pers Med 2017; 7:7-9.
Subbe CP, Kinsella S. Continuous monitoring of respiratory rate in emergency admissions: Evaluation of the RespiraSense™ sensor in acute care compared to the industry standard and gold standard. Sensors (Basel) 2018;18. pii: E2700.
Donnelly N, Hunniford T, Harper R, Flynn A, Kennedy A, Branagh D, et al
. Demonstrating the accuracy of an in-hospital ambulatory patient monitoring solution in measuring respiratory rate. Conf Proc IEEE Eng Med Biol Soc 2013;2013:6711-5.
Takayama A, Takeshima T, Nakashima Y, Yoshidomi T, Nagamine T, Kotani K. A comparison of methods to count breathing frequency. Respir Care 2019;64:555-63.
Rimbi M, Dunsmuir D, Ansermino JM, Nakitende I, Namujwiga T, Kellett J. Respiratory rates observed over 15 and 30 s compared with rates measured over 60 s: Practice-based evidence from an observational study of acutely ill adult medical patients during hospital admission. QJM 2019;112:513-7.
Goodwin L, Leech N. Understanding correlation: Factors that affect the size of r. J Exp Educ 2006;74:249-66.
Zsuzsanna Balla H, Theodorsson E, Ström J. Evaluation of commercial wireless dermal thermometers for surrogate measurements of core temperature. Scand J Clin Lab Invest 2019;79:1-6.
Geijer H, Udumyan R, Lohse G, Nilsagård Y. Temperature measurements with a temporal scanner: Systematic review and meta-analysis. BMJ Open 2016;6:e009509.
Milot JP, Birnbaum L, Larochelle P, Wistaff R, Laskine M, Van Nguyen P, et al
. Unreliability of home blood pressure measurement and the effect of a patient-oriented intervention. Can J Cardiol 2015;31:658-63.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6]