Hypertension is a leading risk factor for heart disease (1). Current guidelines recommend to measure blood pressure (BP) using a validated auscultatory or (semi)automatic upper-arm sphygmomanometer (2, 3). While in the past BP measurement was mainly performed in the doctor’s office, nowadays it is known that the addition of out-of-office BP measurement provides better prognostic information than office BP measurement alone (4, 5). New evolutions in development of BP monitors aim to increase convenience while preserving accuracy (6).

Novel methods replace the oscillometric measurements of BP by cuffless measurements often using photoplethysmography (PPG), ballistocardiography (BCG) and/or electrocardiography (ECG) to estimate BP. This is performed by either analysing the pulse wave of the PPG signal (pulse wave analysis, PWA) (7), as is the case in the smartwatch that was used in this study, or by analysing a combination of PPG and ECG or BCG to calculate the pulse transit time (PTT) (8, 9). Other methods that are being investigated include the use of bio-impedance sensors for BP measurement (10). Advantages in using these methods are increased convenience as well as the possibility to easily increase the number of measurements per day, up to even continuous measurement. Other advantages are an increased role for the patient in managing their own health, possibly resulting in increased adherence to antihypertensive treatment (5).

While guidelines recommend only upper-arm inflatable cuff measurements of BP, commercial devices such as smartwatches capable of measuring BP using cuffless methods are already coming to market. In recent years, smartwatch devices first fulfilled the requirements for a medical device in the International Organization for Standardization (ISO) 81,060–2:2018 Non-invasive sphygmomanometers (5). However, it was recently emphasized that using these standards for new devices is inappropriate, and that new standards are needed (11). More recently, some devices were cleared by the United States Food and Drug Administration (FDA) and the European Union (Conformity Marketing, CE-label) (12). The Samsung Galaxy Watch Active 2 that was used in this study recently received CE-clearance, but algorithms and validation studies are not openly disclosed (8). In those studies about cuffless measurement that are openly available, devices mostly use data collected solely from normal healthy individuals (8). Because of the current scarce validation of the technology and low reliability of wrist BP measurement in general, the Korean Society of Hypertension – currently the only society that has published a position paper on smartwatch-based BP measurements - acknowledges the potential of smartwatch-based BP measurements for increasing awareness of BP and possibly to detect hypertension early in the general population, but is hesitant to recommend its use to monitor treatment response in hypertensive patients and patients with cardiovascular comorbidities (5).

In clinical practice immediate problems are now to be faced. One problem with unvalidated devices is that they cannot be actively used and prescribed before proper validation. A second and more pressing problem is that, once on the market, patients will use these devices for monitoring their health, and proper advice and interpretation by clinicians will be necessary. In a recent review, Elgendi et al. therefore emphasize that to increase reliability and validity of medical use of smartwatch-based BP measurements, more research is required in both normotensive and hypertensive subjects (8).

This study aims to investigate a recently CE-cleared smartwatch using cuffless BP measurement in a population with normotensive and hypertensive individuals scheduled for 24-h BP measurement. In doing so, this study aims to contribute to the evidence base for validating new technologies, already available on the consumer market, for medical use.

To our knowledge, this is the first study comparing BP measured by a smartwatch to ABPM in a mixed normotensive and hypertensive population.

The main findings of this study are a proportional bias in the Taffé method, an association in the results needing correction in the Bland-Altman method and a BP variability that is systematically lower in the smartwatch as compared to the ABPM.

It should first be noted that of all 50 patients that were eligible for participation (i.e., all patients that did not meet an exclusion criterion, that were willing to participate and that were able to work with the technology) and that the device was tested on, 20% (10 patients) were not able to participate due to limitations of the technology (either not fitting, too high BP at calibration or unspecified error message). The fact that a high BP at calibration impedes the calibration process already limits the usage of the device in a hypertensive population. Also, 7 patients of the total screened population indicating not being able to work with the technology limits the usage in populations with low digital literacy.

It is known that validation standards for cuffless BP estimating devices are lacking, and the World Health Organization (WHO) and independent research groups have emphasized the urgent need for developing such validation standards (23, 24). With standards lacking, multiple analyses have been performed to provide as much insight as possible in the comparison of the two methods.

The results of the Taffé method indicate that, based on the bias plots, BP measurements are reliable in normal BP ranges. The farther BP deviates from a normal range, the larger the bias on the measurements becomes. Precision of both measurements is comparable in normal ranges. As already indicated by the developers of this method, Taffé et al., current validation methods suffer from the same limitations as the Bland-Altman analysis when comparing a new measurement method with a gold standard reference method and new international validation protocols using this new method are urgently needed (15). Existing validation protocols and criteria, such as the ISO 81060-2:2018 for non-invasive sphygmomanometers, are not applicable to this study as the smartwatch is a cuffless device in which a calibration step is required (11). Due to the fact that a calibration is necessary, providing three consecutive measurements as prescribed in the ISO protocol would not suffice to reliably validate the smartwatch. Still, it could be useful to compare the results of this study with the cut-offs that are proposed. While no 85 participants were included as proposed in the ISO standards, the amount of total paired measurements in this study (1,063) greatly exceeds the required 255 valid paired measurements. Criterion 1 of the ISO requirements states that for systolic and diastolic BP, mean difference is lower than or equal to ± 5 mmHg with a SD no greater than 8 mmHg (25). In our analysis the criterion for mean BP is met in systolic BP (−2.05) and is not met in diastolic BP (−5.58), and the criterion for SD is met in neither systolic (15.5) nor diastolic BP (22.5).

The ROC curves demonstrate acceptable AUCs (0.777 and 0.930 for systolic and diastolic BPs, respectively). For systolic BP maximal sensitivity and specificity were 71.4 and 76.9%, respectively at a BP of 136.1 mmHg and for diastolic BP maximal sensitivity and specificity were 84.6 and 88.9%, respectively at a BP of 89.9 mmHg. The PPV and NPV optimization curves demonstrate that high PPVs and NPVs can be achieved when using different cut-off points for the smartwatch compared to the reference standards. At first glance these results seem to be acceptable for a diagnostic test, but with validation standards lacking, one should keep questioning if these results are good enough to replace or complement gold standard BP measurements.

BP variability is systematically lower in the smartwatch compared to the ABPM measurement, with a systematic difference clearly being demonstrated in SD, SDw, ARV values, CV and in the ARV scatterplot.

Pulse pressure measurements are significantly different, but with a difference of 3.5 mmHg and no systematic error in the scatterplot, this latter result seems clinically acceptable.

Taking all these results together, namely the proportional bias in the Taffé method, the association needing correction in the Bland-Altman plots and the systematically lower BP variability, it seems that the smartwatch currently suffers from an anchoring point that is set when calibrating the device. After calibration, it tends to keep its BP measurements closer to the calibration point than the BPs are in reality, resulting in an overestimation of lower BPs, an underestimation of higher BPs, and thus a low BP variability. The results of our study thus indicate that a one-to-one translation of a conventional BP monitor to the smartwatch does not seem appropriate yet.

The Korean Society of Hypertension has stated that the future of cuffless devices might lie in detecting hypertension in the general population, but that they are currently not suited for monitoring treatment response in hypertensive patients (5). An additional limitation that is mentioned in their position paper is that accuracy of the smartwatch measurements may be further reduced in patients with aortic valve insufficiency, atrial fibrillation, peripheral vascular disease and other conditions affecting the cardiovascular system. The Korean Society does emphasize that, if accuracy is acceptable, smartwatch technology can increase hypertension awareness, especially in the younger population.

Other studies dare to broaden the scope of usage of these devices. Recent studies predict possibilities for integrating continuously measured BP measurements with other biological and environmental signals (26, 27). Data collected in this way can then be analyzed collectively through time-series analysis or machine learning techniques. This could in turn reveal more complex BP profiles in function of environmental conditions of the patient, or could even predict outcomes based on multiple parameters analyzed collectively. A recent review also broadens the scope to “anticipation medicine,” designed to identify increasing risk and predict the onset of cardiovascular events based on data collected over time (26). For these uses, a wearable device capable of measuring BP continuously including during the night would be most appropriate. Currently, the device used in this study is not capable of performing automatic measurements, but other wearables have been shown to perform accurate automatic measurements throughout the day (28).

At this moment, wearable cuffless BP monitors are not capable of replacing cuff-based BP monitors. If the technology is improved, it is possible that in the future there could be a role for complementary use next to cuff-based BP monitors, for example in different environments and settings, at home and away from home. Devices allowing automatic measurements can allow for continuous and nocturnal measurements, capabilities that conventional BP monitors do not offer.

To take the field forward several things are needed. First, there is an urgent need for validation standards for cuffless devices to allow for standardized research. Second, open disclosure of commercial validation studies could allow for better resource usage as studies will not have to be repeated unnecessarily. Third, further development of the cuffless devices will allow for new possibilities in the field of BP monitoring, especially unobtrusive, continuous and nocturnal monitoring. Finally, research should continue to focus on validation of these devices compared to conventional standards, but also on broadening its use and demonstrate new possibilities of continuous BP monitoring.

Patients will start using these devices today. It is thus up to researchers and clinicians to start understanding the vast range of possibilities that cuffless BP measuring has to offer.

This study demonstrates that the BP measurements by the Samsung Galaxy Watch Active 2 show a systematic bias toward a calibration point, overestimating low BPs and underestimating high BPs, when investigated in both normotensive and hypertensive patients. Standards for traditional non-invasive sphygmomanometers are not met, but these standards are not fully applicable to cuffless devices, emphasizing the urgent need for new standards for cuffless devices.

The smartwatch-based BP measurement is not yet ready for clinical usage. However, when used complementary to a cuff-based BP monitor, BP measurements might be of value. Future studies are needed to further validate wearable devices, and also to demonstrate new possibilities of non-invasive, high-frequency BP monitoring.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Ethische Toetsingscommissie Jessa Ziekenhuis. The patients/participants provided their written informed consent to participate in this study.

MF: conceptualization, experiment conduction, data analysis, and manuscript writing and editing. MS: conceptualization and manuscript writing and editing. KD: experiment conduction and data handling. ZP: data analysis. TK: conceptualization and writing – review and editing. LX: writing – review and editing. EC, PC, and AF: data analysis and writing – review and editing. GP: writing – review and editing and supervision. PD: conceptualization, writing – review and editing, resources, and supervision. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.