A, Scatterplot of the IBP app and standard BP measurements with line of identity in black. The gray lines represent the cutoff points for hypertension. Spearman rank-order correlation coefficient is a nonparametric test for strength of association for non–normally distributed continuous variables with ρ of −1 or +1 indicating perfect correlation and 0.4 to 0.6 representing moderate correlation. IBP measurements were not normally distributed by the Shapiro-Wilk test. B, Bland-Altman plot for IBP and the standard BP measurement. The short dashed line represents the means (SDs) of the absolute value of the difference (IBP minus standard), which were −1.2 (16.2) and 7.1 (10.8) mm Hg. The long dashed lines represent ±2 SDs for this range (32.4 and 21.6, respectively).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Plante TB, Urrea B, MacFarlane ZT, et al. Validation of the Instant Blood Pressure Smartphone App. JAMA Intern Med. 2016;176(5):700–702. doi:https://doi.org/10.1001/jamainternmed.2016.0157
Mobile health (mHealth) technologies include unregulated consumer smartphone apps.1 The Instant Blood Pressure app (IBP; AuraLife) estimates blood pressure (BP) using a technique in which the top edge of the smartphone is placed on the left side of the chest while the individual places his or her right index finger over the smartphone’s camera. Between its release on June 5, 2014, and removal on July 30, 2015 (421 days), the IBP app spent 156 days as one of the top 50 best-selling iPhone apps; at least 950 copies of this $4.99 app were sold on each of those days.2 Validation of this popular app or any of the similar iPhone apps still available (eg, Blood Pressure Pocket, Quick Blood Pressure Measure and Monitor), have not been performed. Using a protocol based on national guidelines,3 we investigated the accuracy and precision of IBP.
A Johns Hopkins University School of Medicine institutional review board approved this study. In August and September 2015, we enrolled patients and staff who were at least 18 years old from 5 ambulatory Johns Hopkins sites (1 clinic each in general internal medicine, nephrology, and the ProHealth clinical research unit, and 2 in cardiology clinics). Clinicians referred patients with or without hypertension or a baseline hypertensive reading for enrollment. After prescreening 105 individuals, written informed consent was obtained from 101 participants. Participants were given $5 gift cards for their time. Per prespecified dropping rules, data from 16 individuals were discarded because of unavailable cuff sizes (n = 1), standard device errors (n = 2), and excessive variation in sequential standard BP measurements, greater than 12 mm Hg for systolic BP or greater than 8 mm Hg for diastolic BP (n = 13).3
For IBP measurements, research staff were trained to measure BP according to manufacturer guidelines using the IBP app, version 1.2.3, on a smartphone (iPhone 5s and 6; iOS version 8.3; Apple Inc); IBP required entry of date of birth, sex, height, and weight for each measurement. For standard BP measurements, research staff were trained to follow a standard protocol using calibrated, validated automated sphygmomanometers (Omron 907 and 907 XL).4 Measurement order (IBP and standard) was random. Following 5 minutes of quiet sitting, 2 sequential BPs were taken by each device, separated by 60 seconds. The standard BP measurement was an average of the 2 BP measurements by sphygmomanometry. Sensitivity and specificity for detection of hypertensive BP were calculated using systolic BP of at least 140 mm Hg or diastolic BP of at least 90 mm Hg.
Calculation of the mean of the absolute value of the difference and data visualization were examined using Stata statistical software (version 13.1; StataCorp).
Of 85 participants, 44 (52%) were women. Their mean (SD) age was 56.6 (16.3) years, and their body mass index, calculated as weight in kilograms divided by height in meters squared, was 27.8 (5.8); 45 (53%) self-reported hypertension, and 91% of these (41) reported taking antihypertensive medications.
The mean (SD) of the absolute values of the difference between IBP and standard were 12.4 (10.5) mm Hg for systolic BP and 10.1 (8.1) mm Hg for diastolic BP. IBP underestimated higher BPs and overestimated lower BPs (Figure). IBP systolic BP readings were within 5, 10, and 15 mm Hg of the standard BP measurement 24%, 44%, and 59% of the time, respectively, whereas respective proportions for diastolic BP were 26%, 48%, and 70%, correlating with the lowest possible accuracy grade in all categories by British Hypertensive Society scoring.5 Spearman ρ was 0.44 (P < .001) for systolic BP and 0.41 (P < .001) for diastolic BP. Sensitivity and specificity of IBP for hypertensive BPs were 0.22 and 0.92, respectively.
The BP measurements from an mHealth app with more than 148 000 units sold were highly inaccurate. The low sensitivity for hypertensive measurements means that approximately four-fifths (77.5%) of individuals with hypertensive BP levels will be falsely reassured that their BP is in the nonhypertensive range.
Our study has both clinical and public health relevance. While IBP recently became unavailable for unclear reasons,6 it is installed on a vast number of iPhones; furthermore, several “me-too” apps are still available. Hence, we remain concerned that individuals may use these apps to assess their BP and titrate therapy. From a public health perspective, our study supports partnership of app developers, distributors, and regulatory bodies to set and follow standards for safe, validated mHealth technologies.
Corresponding Author: Timothy B. Plante, MD, Division of General Internal Medicine, Johns Hopkins University School of Medicine, 2024 E Monument St, Ste 2-300B, Baltimore, MD 21287 (email@example.com).
Published Online: March 2, 2016. doi:10.1001/jamainternmed.2016.0157.
Author Contributions: Dr Plante had full access to all of the data and takes responsibility for the integrity of the data and accuracy of the data analysis.
Study concept and design: Plante, Miller, Appel, Martin.
Acquisition, analysis, or interpretation of data: Plante, Urrea, MacFarlane, Blumenthal, Appel, Martin.
Drafting of the manuscript: Plante, Martin.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Plante, Martin.
Obtained funding: Martin.
Administrative, technical, or material support: Plante, Urrea, Blumenthal, Appel, Martin.
Study supervision: Blumenthal, Miller, Appel, Martin.
Conflict of Interest Disclosures: Dr Martin reports grants from the PJ Schafer Cardiovascular Research Fund, during the conduct of the study; and grants from American Heart Association and Aetna Foundation and personal fees from American College of Cardiology outside the submitted work. No other disclosures are reported.
Funding/Support: This study was supported by a PJ Schafer Cardiovascular Research Grant. Dr Plante is funded by the Institutional National Research Service Award National Institutes of Health training grant No.T32HP10025B0.
Previous Presentation: This study was presented at the American Heart Association’s Epidemiology and Prevention/Lifestyle 2016 Scientific Sessions; March 2, 2016; Phoenix, Arizona.
Additional Contribution: We thank Jeanne Charleston, PhD, BSN, Department of Epidemiology, the Johns Hopkins Bloomberg School of Public Health, for her assistance in obtaining of the standard devices, standard device QA management, and research staff training; the clinical staff at Johns Hopkins General Internal Medicine, Cardiology, Nephrology, and ProHealth sites for assistance in enrollment; and Morgan Grams, MD, PHD, MHS, Satish Misra, MD, and Haitham Ahmed, MD, for their assistance in trial design. These individuals were not compensated for their contributions.
Create a personal account or sign in to: