A randomized crossover clinical trial comparing 24-h blood pressure measurements from two different cuffless wrist ambulatory monitors with cuffed ambulatory monitoring for translation to clinical practice: Rationale and Design of Continuum BP study

Abstract

Study objective

Cuffless blood pressure (BP) measuring devices have potential to improve screening, diagnosis and management of hypertension. However, translation of BP estimates from such devices to cuffed BP readings, which currently inform hypertension and cardiovascular risk management guidelines, remains uncertain. We designed Continuum BP to understand how BP measurements from two independent CE-marked wrist worn devices can be contextualized for clinical use.

Design

Prospective observational randomized crossover study.

Setting

William Harvey Heart Centre, Queen Mary, University of London.

Participants

48 adult volunteers.

Interventions

Aktiia Bracelet (utilizing photoplethysmography) and the Healthstats BPro Evo (utilizing applanation tonometry) worn independently, each alongside cuffed OnTrak ambulatory BP monitor (ABPM), across two 24 h non-overlapping periods of monitoring by the same individuals.

Main outcome measures

Difference between average daytime, 24 h and night-time systolic and diastolic BP estimated by each cuffless device and ABPM; and acceptability of wearing each device type compared to ABPM. Data will be used to estimate a numerical calibration of readings from each of Aktiia/Healthstats BPro Evo to ambulatory BP.

Conclusion

Continuum BP received ethical approval from West of Scotland National Health Service research ethics committee (24/WS/0131) and was advertised across Queen Mary University of London campuses and affiliated hospitals. Recruitment is complete. Collected data is being analysed. Findings will enable physicians to interpret and use cuffless BP from two commonly available wrist devices in routine practice, and disseminated through study website, scientific meetings and publications. Trial registration https://clinicaltrials.gov/ (NCT06573801).

Graphical abstract

Highlights

• •Cuffless wearables can provide near continuous estimation of blood pressure (BP).
• •Hypertension management is based on evidence linking outcomes to cuffed brachial BP.
• •Continuum BP will translate BP from two cuffless wearables to cuffed ambulatory BP.
• •Recruitment is complete and collected data is being analysed.
• •Findings will help clinicians interpret and use cuffless BP in routine practice.

1. Introduction

Elevated blood pressure (BP) or hypertension is the leading modifiable risk factor for cardiovascular diseases [1]. Recently, hypertension has contributed to >10 million deaths globally [2]. However, worldwide many people remain undiagnosed, and many on treatment have their BP uncontrolled [3], [4], [5]. Among factors underlying this is lack of access to, or motivation for, easy reliable regular clinical or self-monitoring [6].

Conventionally, BP is measured using upper arm cuffed manual auscultatory or automated oscillometric devices which provide systolic (SBP) and diastolic (DBP) readings. Diagnosis and management of hypertension is based on cuffed BP measured during clinic/office visits and increasingly out-of-office e.g., home and 24 h ambulatory BP monitoring (ABPM) [7]. The use of an inflatable cuff is associated with discomfort and anxiety, and errors related to fitting and positioning of the cuff or tubing; which can all influence BP readings. Clinic readings are also subject to ‘white-coat’ [8] or ‘masked’ hypertension [9] while ABPM is expensive and not universally available [10]. Importantly, cuffed methods allow only for a limited number of readings that fail to capture BP variability (BPV), associated with physical and mental activities that occur within the 24 h period and are associated with clinical outcomes [11].

Innovation in cuffless wearables such as smart watches and bands has potential to offset these limitations [12]. Novel technology that does not rely on inflation of an arm cuff, utilizes characteristics of the arterial pulse travelling from the heart to a peripheral location such as the wrist e.g., pulse wave velocity or pulse waveform analysis to mathematically estimate SBP and DBP [13]. With wearables an increasingly affordable, less obtrusive, and acceptable option for recording BP; improved awareness, early hypertension diagnosis and treatment, adherence to and titration of treatment can be facilitated.

While availability and use of cuffless BP devices have exponentially increased [14], uncertainty remains regarding clinical utility of cuffless BP monitoring [15], [16]. For example, evidence is limited on how accurately cuffless wearables can track changes in BP over 24 h [14], [17], with experts suggesting robust examination of their performance in both static and dynamic states of activity [13], [18]. There is also ambiguity concerning interpretation of cuffless BP in terms of thresholds for diagnosis and management of hypertension, use in estimating cardiovascular risk, and how well cuffless devices reflect BP control compared to ABPM, for example by metrics such as time in target range (TTR) [19] which are all currently based on cuffed measurements.

1.1. Rationale and aim of Continuum BP study

We describe the Continuum BP study (Fig. 1 and Graphical Abstract), designed to understand how BP estimates from two CE-marked wrist-worn cuffless devices, employing two independent commonly used methodologies (photoplethysmography (PPG) and applanation tonometry (AT)) for measuring BP from radial pulse characteristics, can be contextualized for clinical use.

Fig. 1.

Rationale and aim of Continuum BP.

The Aktiia Bracelet G11 (Aktiia, Neuchatel, Switzerland [20]) a wrist band utilizing PPG (Fig. 2A), and the Healthstats BPro Evo [21] wrist worn tonometric monitor using AT (Fig. 2B) differ in how they apply pulse wave analysis for estimating BP. PPG sends optical signals from a light emitting diode through pulsating peripheral vasculature and converts optical energy transmitted/reflected in accordance with blood volumetric changes to BP values using machine learning algorithms with reference to a brachial (calibration) BP [22]. AT uses a semi-occlusive mechanism to apply gentle pressure at the wrist, and a tonometric sensor to capture outward intravascular pressure waveforms which are then scaled to BP values again with reference to a brachial artery (calibration) BP [23].

Fig. 2.

Cuffless wrist devices (A) Aktiia and (B) Healthstats BPro Evo examined in Continuum BP.

The Aktiia bracelet has been reported to estimate similar daytime but not night-time BP when compared to ABPM [17], [24], and to estimate BP in most common body positions compared to auscultation with varying precision [25]. Similarly, BP measurements under ambulatory conditions using Healthstats BPro Evo have been noted to have moderate agreement with cuffed upper arm ambulatory readings [26].

In the Continuum BP study, we aim to examine how BP estimated by Aktiia and Healthstats BPro Evo worn for two 24 h non-overlapping periods (2–14 days apart) by the same volunteers, compares with BP measured by 24 h cuff-based ABPM. While ABPM is the current gold-standard for BP monitoring, it has been shown to have poor short-term intra-individual reproducibility [27]. We will therefore compare BP estimates from each cuffless device with cuffed ambulatory BP from the same period of monitoring as well as with cuffed ambulatory BP pooled across the two monitoring periods. In estimating how real-time BP can be calibrated to cuffed ABPM readings, our findings will inform how cuffless BP can be interpreted and potentially used in current clinical practice for diagnosis and management of hypertension and cardiovascular risk.

2. Materials and methods

2.1. Study registration and ethical approval

Continuum BP was registered at ClinicalTrials.gov (NCT06573801, https://clinicaltrials.gov/study/NCT06573801) and received a favorable opinion from the West of Scotland National Health Service research ethics committee (24/WS/0131). The study forms part of the research portfolio of the National Institute of Health and Research Barts Biomedical Research Centre at Queen Mary University of London (QMUL) and was sponsored by QMUL. Minor non-notifiable amendments were approved by the sponsor. The trial sponsor and funder had no role in design, conduct, analysis, and reporting of the study.

2.2. Study design and setting

Continuum BP, a prospective observational single-centre randomized cross-over study, sequentially assigned Aktiia and Healthstats BPro to participants for two non-overlapping periods of 24 h of BP monitoring alongside the OnTrak ABPM device. The study was conducted at the William Harvey Heart Centre, QMUL.

2.3. Study devices

• 1.Aktiia: each participant was assigned a unique Aktiia account to which a specific device was paired using the Aktiia Research App. The app was also used for initialization i.e., calibrating signals from the wrist to cuffed BP using the oscillometric Aktiia Cuff (removed once initialization was successful) and to sync data from the bracelet, after the 24 h period of monitoring was complete, to Aktiia's cloud platform. Here signals were converted to SBP and DBP values, and available as time stamped output for download through the Aktiia DataHub. The Aktiia bracelet is a silent device programmed to make several random attempts per hour to capture signals, requiring the user to be still for this to succeed.
• 2.Healthstats BPro Evo: each participant was fitted with an assigned BPro device, calibrated using average unobserved seated BP measured using Omron HEM-907. The BPro attempts to capture a radial pulse waveform every 15 min for the subsequent 24 h, when monitoring automatically stops. The BPro is sensitive to hand and body movements but can indicate start of signal capture using an audible beep, which was set to work throughout the day and silenced between midnight and 6 am. Data captured from a completed monitoring session was then transferred via the PC-based B-Pro Connect program to the secure Healthstats portal, and matched to the relevant participant ID, from where a time-stamped report of BP measurements was available to download.
• 3.Spacelabs Healthcare OnTrak 90227 is a CE-marked oscillometric ABPM. An upper arm cuff is attached via a tubing to a battery-operated unit worn at the waist on the opposite side. For the study, the OnTrak was programmed using the associated Sentinel software, to record BP every 20 min during 7 am to 11 pm and every 30 min during 11 pm to 7 am. Each monitoring session (2 per participant) was labelled using the combination of participant assigned identification number and visit number. For each monitoring period, the unit was fitted with fully charged batteries. Data collected at end of monitoring was downloaded using linked participant details onto Sentinel, from where a time stamped report was available to export in various formats.

2.4. Study population and recruitment strategy

The study was open to adult volunteers between 21 and 85 years of age (the range for which Aktiia was CE-marked), with or without a diagnosis of hypertension, fulfilling eligibility criteria aligned with intended use of both cuffless wrist devices (Table 1).

Table 1.

Continuum BP eligibility criteria.

Inclusion criteria 1. Age ≥ 21 years and < 85 years at the time of screening 2. Able to speak and understand English. 3. Able and willing to give informed consent. 4. Independent and mobile. 5. Willing and able to wear devices/no upper limb restrictions.

Exclusion criteria 1. Individuals who are or have been involved in interventional research within a period of 3 months. 2. Vulnerable individuals including those with mental ill-health or who are care dependant. 3. Individuals with serious comorbidities or end organ damage (e.g., previously diagnosed heart failure with NYHA class III/IV, verbal or documented evidence of renal failure or history of dialysis, stroke or myocardial infarction within the last 1-year, previous stroke with residual deficit); and doctor diagnosed hyperthyroidism. 4. Individuals with arterio-venous fistula; or reduced peripheral perfusion identified by a history of Raynaud's syndrome or a pulse pressure of <30 mmHg or a capillary refill time of >3 s in the dominant arm. 5. Individuals with heart rhythm disorders e.g., persistent Atrial Fibrillation and those who have tachycardia >120/min after 5 min of rest 6. 7. Individuals with skin irritation injury or damage. Self-reported pregnancy. Participation will be withdrawn in the event of pregnancy following enrolment. 8. 9. Life threatening or terminal illness with limited lifespan of <12 months. Obesity (BMI 35 kg/m2 or higher) or arm/wrist circumferences outside the range highlighted in each device's specification. 10. Those with very high office BP (greater than Stage 3 hypertension). 11. Interarm difference > 15 mmHg in SBP and > 10 mmHg in DBP. 12. Where the PI thinks that it would be technically difficult to obtain or interpret blood pressures measurements.

We used posters, leaflets and social media platforms to advertise the study at QMUL campuses and in clinical and non-clinical areas of the affiliated Barts Health NHS Trust hospitals. Interested individuals approached the study team in person or using telephone/email contact details provided. All such individuals were then provided with an electronic or paper participant information sheet and recontacted after a week to answer any queries and/or confirm interest. More time was provided for individuals requiring further consideration.

Those confirming interest in participating were recontacted and, with permission, asked a few screening questions before being invited (to avoid a wasted journey) for an in-person visit (Visit 1) when the inclusion/exclusion criteria were re-assessed and clinical measurements were collected. A second visit (Visit 2) was booked no less than 2 days and up to 14 days after Visit 1. Thus, each participant confirmed to be eligible made two separate visits to the William Harvey Heart Centre.

2.5. Study procedures

For a summary of study procedures and data collection please see Table 2.

Table 2.

Continuum BP schedule of procedures for data collection.

*Includes demographic, clinical and medication history.

2.5.1. Visit 1: screening and enrolment

Each potential participant was assigned a unique study identifier (ID) and written informed consent was obtained by trained members of the study team (cardiac research nurse and research associate) prior to collecting any personal information or physical measures. This was followed by screening for eligibility using (a) questionnaire-based sociodemographic, medical and medication history, (b) physical measurements for weight, height, arm and wrist circumference, (c) seated observed automated office BP (average of 3 measurements starting with a 5-min rest and an interval of 1-min between readings) on each arm using the OMRON HEM-907 and appropriately sized cuff, and (d) pulse rate, capillary refill time and oxygen saturation on the dominant arm. Participants also completed an Equality, Diversity and Inclusion questionnaire. This information was recorded on paper and electronic case report forms on the Castor Electronic Data Capture platform [28], pre-programmed to identify screen failure. If eligibility criteria were met, enrolment was confirmed to participants. Enrolled participants additionally provided information regarding perceived stress (4-item Perceived Stress Scale (PSS4, Supp File S1) [29]) and recent alcohol intake.

At the Chief Investigator's (CI) judgement, screening could be repeated once. Participants were required to re-consent, and a new study ID was issued.

2.5.2. Visit 1: randomization

Continuum BP was a crossover study with each enrolled participant eventually having BP measured using both wearable devices. Assignment to a cuffless wearable took place through simple randomization with variable block sizes via Castor [28]. Research staff and participants were only aware of (and hence unblinded to) device sequence once randomized.

2.5.3. Visit 1: device application

Following randomization, 3 unobserved BP measurements were recorded using OMRON HEM-907 from the dominant arm, starting with a 5-min rest and an interval of 1-min. The study team left the room, and readings were concealed using the machine's hide function until all 3 had been recorded and the study team returned. The average of these 3 measurements was noted as unobserved office BP.

A cuffless wrist device, with an appropriately sized strap, was applied on the participant's dominant wrist in the order dictated by randomization and calibrated as per instruction manuals/training provided by device companies. Information about how the device worked and how to care for it was provided.

Participants were then fitted with the OnTrak 90227 using an appropriately sized cuff on the non-dominant arm and explained that the device was programmed to operate by inflating the upper-arm cuff at set frequencies during day and night. The first reading was manually triggered, and participants were observed for the following 20–30 min to ensure an automatic inflation occurred as scheduled.

Visit 1 lasted a maximum of 2 h. At the end, participants were provided with (1) a paper Activity and Medications Log to record physical activity, medications, waking and sleeping hours, and any unusual symptoms (Supp. File S2) and (2) a (non-validated, study-specific) Likert scale based Device Wearability Questionnaire rating separately the experience of wearing the wrist device and the ABPM, to complete for the 24 h of BP monitoring (Supp. File S3).

The study team also provided each participant with the time (spanning a little more than 24 h from the point when the allocated wrist device was calibrated) after which monitoring could end. Participants were shown how to switch off and remove devices safely, reminded to continue any prescribed medications and routine activities except heavy physical exercise, showers, swimming, and protect devices from getting wet during monitoring.

2.5.4. Visit 2

The study team confirmed participant's willingness to continue and checked for any change in health/medication status that would preclude participation. Each participant had their unobserved office BP measured, the second wrist device fitted on the dominant wrist and the OnTrak monitor on the non-dominant arm as before. Activity and Medications Log, Device Wearability Questionnaire and return instructions were provided for the second period of monitoring. Visit 2 lasted a maximum of 1 h.

2.5.5. Return of devices and paperwork

On both visits participants were supplied with return packages and pre-paid postage labels, and instructed on how to pack devices and paperwork for posting. Where convenient, participants returned all materials in-person.

2.5.6. End of participation or withdrawal

Participation ended when devices and paperwork from Visit 2 were received by the study team. Participants were able to withdraw from the study voluntarily at any time. The CI could also withdraw a participant if it was in the participant's best interest or in compliance with study procedures/protocol. If a participant withdrew/was withdrawn, there was no further data collection, but data already collected was retained for analysis.

2.5.7. Data collection and entry

All self-reported and clinically measured data collected on paper case report forms, was transferred by the study team, linked only by the study ID to the study database on Castor. This was designed and tested jointly by the study team and study coordinators at the William Harvey Barts Cardiovascular Clinical Trials Unit (CVCTU).

Data from each pair of returned devices was downloaded as per manufacturer instructions (through Sentinel, BPro Connect and the Aktiia DataHub) to a designated study laptop into a source folder. From here, the different file formats were converted to a standardized csv format using a custom written Python script, labelled using study ID, date and device type, and uploaded on study-specific Castor database. At least 20% of converted device data files were checked for accuracy against source files (Supp File S4 Data Management Plan).

The study had 8 units of each of the 3 device types. These were used in pairs of cuffless and cuffed ABPM devices. Each pair of returned devices underwent functionality checks once BP data from a monitoring period was transferred as described above, before being assigned to a new participant.

2.5.8. Monitoring and oversight

QMUL as the Sponsor retained the right to audit the study, or study site. In addition, any part of the study could be audited by the funders where applicable.

A trial management group consisting of the CI, the study team, and Barts CVCTU statistician and study coordinating team met before the study opened and monthly while the study was ongoing to review progress and conduct. Considering this was a low-risk, single-centre, small sample study, a monitoring plan was agreed between the study team and Barts CVCTU, who also coordinated source data verification (Supp File S5 Monitoring Plan).

2.5.9. End of study and data storage

Recruitment ended when the estimated sample size was reached. Data collection ended with return of all study devices. The CI and study team ensured all participant identities were protected, and all information regarding participants was confidentially managed in accordance with the Data Protection Act (2018), the UK Policy Framework for Health and Social Care and Research Ethics Committee approval. Device data was linked to participants only through assigned study IDs.

Study data will be available to the funder, sponsor, Barts CVCTU staff, and study site following the criteria laid out in relevant contracts. Following final analysis and publication of main results a cleaned version of the study data along with codebook and analysis code will be stored in secure locations at QMUL.

An anonymized version of the database excluding all personal identifiable information will also be stored securely. This dataset will include a new ID number for each participant which is not associated with recruitment order.

2.6. Study outcomes

The primary endpoint of Continuum BP was difference between average SBP and DBP measured using each cuffless device and (1) its comparative ABPM on the same visit, and (2) from pooled ABPM across both visits, over daytime (a defined period of 7 am to 10.59 pm). Secondary endpoints included similarly estimated differences between cuffless devices and ABPM in average SBP and DBP over 24 h and night-time (11 pm to 6.59 am), and differences in BPV over selected monitoring periods. These will inform development of correlation factors for separate relationships between Aktiia and Healthstats BPro Evo, and ABPM readings to allow translation of cuffless BP for clinical usage.

Participant feedback on acceptability of wearing each device will be evaluated for each cuffless device and its comparator ABPM.

As an exploratory endpoint, we will compare TTR estimated by each cuffless device and its comparative ABPM on the same visit.

2.7. Risks/benefits and safety reporting

As all devices were CE-marked. Risks to participants were minimal except for some discomfort, especially during sleeping hours. Research staff checked with participants that the device was comfortably positioned – snug but not tight. Where participants indicated this was tight after calibration, the device was repositioned and recalibrated. Participants were able to self-report unusual symptoms on the Activity Log and had a free text option on the Device Wearability Questionnaire to report adverse experiences, which were reviewed and appropriately reported. Adverse events (AEs) between visits were not recorded as they were not the aim/focus of this study.

It was made clear to participants that enrolling in the study would not mean getting diagnosis or care for hypertension. However, all participants were provided with printed copies of their OnTrak 24 h ABPM reports once reviewed by the study clinician. In cases where raised BP was noted, this was highlighted to the participant and if granted permission, their primary care physician was additionally notified.

2.8. Statistical considerations

2.8.1. Sample size

Continuum BP was a within subject study, comparing BP values obtained by cuffed and cuffless devices for the same individual. The sample size for comparing paired BP readings with a power of 80% and (two-sided) alpha significance of 5% that would have allowed detection of an expected difference in mean BP between cuffless and cuffed devices of 5 mmHg (SD 10 mmHg)2 was 34 (number of pairs, in this case individuals). To ensure representation of main characteristics affecting BP, we aimed for a sample size of 48, with 16 participants in each of the following age categories: 21–45 years, >45–60 years, and > 60- <85 years of age. We also aimed for balanced representation by sex with no more than 9 participants of one sex in each age group.

Based on evidence examining ambulatory BPs [30] and completeness of initially returned device pairs, and to account for variable signal capture by wrist devices, a sufficient dataset was considered as 14 daytime readings (7 am-10:59 pm) and 6 nighttime readings (11 pm-6:59 am) for each device. The planned sample size of up to 48 accounted for 25% of individuals not achieving a sufficient dataset for all devices and for 5% withdrawal rate while retaining 80% power in paired analysis. Characteristics of withdrawing participants/those with insufficient data were monitored to ensure that equal representation described above was maintained.

2.8.2. Planned statistical analysis

Participant characteristics will be summarized using mean ± SD for normally distributed continuous variables, median ± IQR for non-normally distributed variables, and number and percentage to represent categorical variables for the whole cohort and by randomized group. Trial adherence will be reported as proportions of participants with sufficient data for all four devices (both cuffless devices and their comparator ABPMs), and for each cuffless device and ABPM pair.

Analyses of primary and secondary endpoints will be run on a per-protocol basis using definitions of sufficient data described above, on two analysis populations. Population 1 (to ensure comparable estimation of endpoint and interpretation) comprising only those with sufficient data for each device pair across both visits. Population 2 (to maximize use of collected data) comprising those with sufficient data for the cuffless device and ABPM for each time frame.

For the various study endpoints, mean (SD) of each participant's set of SBP and DBP measured by all devices will be calculated for three timeframes (24 h, daytime and nighttime) for readings from each (cuffless-ABPM) device pair at each visit, and as pooled mean (SD) for readings from the ABPM across both visits. For BPV endpoints, SBP and DBP variability will be estimated for similar periods using SD, Coefficient of Variation and Variability Independent of Mean. Methods from published work based on linear interpolation [31], [32], area under curve for time spent within target BP [33] and proportion of the distribution of BP reading will be considered for calculating TTR for SBP.

Effect size for primary endpoints will be reported as difference in mean daytime SBP (or DBP) between each cuffless device and ABPM with 95% confidence intervals and examined using paired t-tests. Achieving statistical significance will demonstrate whether a substantial difference in measurements exists. If a statistically significant difference in BP is not found, then extent of agreement will be reported using Bland-Altman plots and Intraclass Correlation Coefficient. All continuous secondary and exploratory endpoints will be examined similarly.

For correlating cuffless BP estimates to cuffed BP, the relationship between each wrist device and ABPM will be explored using linear or non-linear regression models depending on linearity checks, keeping cuffless BP as independent and BP measured by ABPM as dependent variables. The regression models will adjust mean readings obtained from cuffless devices to translate them to cuffed readings. Factors that may influence wrist device signal capture, for example age, wrist size, ethnicity, pulse pressure, physical activity will be assessed and adjusted in the model. The idea is to show proof of the concept that daytime-average cuffless BP estimates can be approximated to corresponding daytime-average cuffed readings and made usable for current clinical practice as opposed to one time office BP. Relationships with pooled ABPM readings as dependent variable will be examined in separate models. Model adjustments will include demographic variables (e.g., age and BMI) and covariates e.g., stress and alcohol intake – which may vary between visits.

Effect size for measure(s) of acceptability of wearing different devices will be reported as difference in proportions (with 95% confidence intervals) of responses across Likert categories, between each cuffless device and ABPM, for each of 5 statements and examined using chi-squared tests.

As sensitivity analysis, examination of key endpoints will be repeated using participant waking and sleeping hours, for which the condition of data sufficiency will be relaxed.

If numbers allow, we will consider reporting comparisons between devices while factoring in enrolment BP/antihypertensive treatment status.

Considering the crossover design, we will test period and crossover effects. As only participants meeting described definitions of data sufficiency will be included for evaluation; the study will conduct complete case analysis. Missing data will be reported for participant characteristics and for data collected from devices and generally assumed missing-at-random.

Self-reported AEs will be summarized as counts and percentages. The number of AEs and number of participants experiencing at least one AE will be presented for the whole cohort.

Statistical analysis will be completed using Stata and/or R. Statistical reporting will follow the updated CONSORT statement (see Fig. 3 for participant flow). Statistical significance will be taken as 5% and corrections for multiple comparisons will apply.

Fig. 3.

Consort diagram for Continuum BP participant flow.

Adapted from: Hopewell S, Chan AW, Collins GS, Hróbjartsson A, Moher D, Schulz KF, et al. CONSORT 2025 Statement: updated guideline for reporting randomised trials. BMJ. 2025; 388:e081123. https://dx.doi.org/10.1136/bmj-2024-081123 © 2025 Hopewell et al. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

No interim analysis was planned. The full statistical analysis plan and the complete research protocol CONTINUUM BP Protocol V1.3_08.05.2025 can be provided upon reasonable request.

2.9. Patient and public involvement

The Barts Biomedical Research Centre QMUL hosts a patient and a public involvement panel whose members have lived experience of cardiovascular disease. The panel was involved with study design, suggested areas and groups to consider for advertising the study, received updates and provided regular input on progress of the study.

3. Discussion

Continuum BP is a crossover study comparing cuffless BP estimates from Aktiia Bracelet and Healthstats BPro Evo with cuffed ambulatory BP from the OnTrak monitor, across two non-overlapping periods of monitoring in the same group of adult volunteers, with the aim of translating cuffless BP readings for clinical use.

The Study opened in Jan 2025. Over 100 individuals contacted the study team. Fifty in-person screening sessions were conducted during which there were 2 screen failures. At time of writing, recruitment of 48 adult volunteers was completed with data undergoing checks and initial analysis.

Despite their promise as near continuous means of monitoring BP, current guidelines do not recommend cuffless wearables for diagnoses or management of hypertension [15], [16]. This relates to heterogeneity in cuffless technology and populations studied. PPG signals for example require adequate contrast between sensor and skin, and blood volumes sensed are impacted by applied pressure. Signals are affected by peripheral circulation, gravity, and autonomic innervation, with significant inter-user variability in morphology of waveforms [13]. While clinical validation frameworks for cuffless BP monitors are being developed, such devices are marketed following regulatory approvals based on a favorable benefit-to-risk ratio e.g., some trade-off between ease of use leading to better compliance and monitoring versus acceptable limits of accuracy [12]. Examining clinical utility of cuffless devices thus remains a top hypertension research priority [34].

Hypertension diagnosis and BP control thresholds have been well defined for separate office and home/ambulatory settings for cuffed devices. Cuffless wearables requiring user calibration may be sufficiently indicative of BP trends. However, there is little evidence on how cuffless BP relates to clinical outcomes. Furthermore, interpreting absolute values as they relate to cuffed BP, which has been linked to clinical outcomes through robust evidence and upon which clinical management is based, remains unreliable. In this context, Continuum BP, while not a validation study, will provide evidence on where cuffless BP measurements from two commonly available wrist devices stand with regards to current gold-standard of cuffed ambulatory BP as an ‘out of office’ measurement. Short of artificially invoking BP changes in a lab setting, Continuum BP will be making a pragmatic clinical field comparison of performance at different levels of activity and positions of everyday life [18].

The study was successful in achieving equitable enrollment across age and sex categories. A good number of participants self-reported non-White ethnicity, which will enable assessment of device performance across skin tones. The study was open to adults whether or not they had hypertension. While this helped recruitment, it meant we enrolled a limited number of known hypertensives, arguably the group for whom BP self-monitoring is most crucial. Importantly, the study was not designed to test ability of the selected wrist devices to track BP changes following titration of antihypertensive therapy [18]. However, a large proportion of the global adult population remains unaware of their BP, for whom easy monitoring that can be translated to current cuffed BP thresholds may facilitate timely diagnosis and appropriate intervention. Larger studies may still be needed to examine how commonly cuffless and cuffed ambulatory devices similarly classify adults without a diagnosis of hypertension as having elevated BP.

Differences in the way BP measurements are triggered by the cuffless devices (BPro which is pre-programmed at 15-min intervals, Aktiia which makes attempts based on user activity) and OnTrak (as described above), mean that each device pair is not comparable for ‘simultaneous in time’ measurements, and hence some difference in mean BP and BPV is expected. The study also applied each cuffless device on the dominant arm, likely making it more susceptible to motion artefact. However, current recommendations suggest making comparisons between time averaged rather than absolute BP values [18], which has been widely employed in published performance studies and in Continuum BP. Further, while we fitted devices on opposite arms, the eligibility criteria ensured that interarm differences in BP were within acceptable limits [18].

The study compares Aktiia bracelet and Healthstats BPro Evo wrist monitor not with each other but with the gold-standard. Findings will inform how real-time intermittent BP from each device may be used. A quantification of any systematic bias [12] allowing calibration of real-time cuffless BP to equivalent cuffed ground truth, will guide clinical utility and additionally make cuffless BP device data usable in models that predict clinical cardiovascular outcomes and are developed using cuff measured BP values. This could drive a paradigm shift, enabling cuffless technologies to play a transformative role in alleviating the health burden related to hypertension.

With artificial intelligence rapidly being employed for personalized treatments across health domains, experts foresee a time when hypertension care would be able to leverage real-time BP data in algorithms similar to those employed in continuous glucose monitoring devices [35] for reliably predicting absolute BP and clinically useful trends, that would optimize self-management, enable prompt lifestyle or pharmacotherapeutic interventions, and move care from clinics to community. By assessing utility of BP recordings using wearable devices, we anticipate that evidence generated by Continuum BP will be contributing to this broader vision.

CONTINUUM BP Study Team

• -Chief Investigator: Dr. Ajay Gupta
• -Co – Investigators: Prof Anthony Mathur and Prof Amrita Ahluwalia
• -Other Investigators: Dr. Ayesha Ahmed, Dr. Krishnaraj Rathod
• -Trial Statistician: Dr. Annastazia Learoyd
• -Trial Delivery Team: Ruth Bowles, Jayasree Sivakumar, Dr. Richard Burns, Cindy Sharma
• -CVCTU, William Harvey Research Centre, Queen Mary University of London: Emily Kirkpatrick, Nikeeta Gurung, Georgia Mannion-Krase
• -CVD Hub, William Harvey Research Centre, Queen Mary University of London: Jack Biddle, Alex Aboud, Kris Lavery, Alice Reid

Trial Management Group

Ajay Gupta (Chair), Emily Kirkpatrick, Ruth Bowles, Ayesha Ahmed, Yousaf Bhatti, Nikeeta Gurung, Annastazia Learoyd, Jack Biddle, Alex Aboud, Krishnaraj Rathod, Richard Burns, Jayasree Sivakumar.

Dissemination

Data arising from the study is owned by the sponsor. Funders will be acknowledged in the publications but do not have review and publication rights of the data from the study.

The full study report will be accessible via ClinicalTrials.gov within one year of the end of the study notification. A plain language summary of the study findings will be posted on the CVCTU website (https://www.qmul.ac.uk/whri/clinical-activities/cvctu/trials-portfolio/continuum-bp-/), along with links to all peer reviewed scientific publications arising from the study.

Collaborators

We acknowledge Aktiia (now Hilo, Rue de Bassin 8A, 2000 Neuchatel, Switzerland) and Healthstats (82 Playfair Road, #07-02, D'Lithium, Singapore 368001) for providing study devices on rent, and related training and ongoing support.

CRediT authorship contribution statement

Ajay K. Gupta: Writing – review & editing, Supervision, Methodology, Conceptualization. Ayesha Ahmed: Writing – review & editing, Writing – original draft, Methodology, Investigation, Data curation, Conceptualization. Ruth Bowles: Writing – review & editing, Project administration, Methodology, Investigation, Conceptualization. Annastazia Learoyd: Writing – review & editing, Methodology, Formal analysis, Data curation. Emily Kirkpatrick: Writing – review & editing, Resources, Project administration. Nikeeta Gurung: Writing – review & editing, Resources, Project administration. Krishna Rathod: Writing – review & editing, Methodology, Investigation. Amrita Ahluwalia: Writing – review & editing, Supervision, Funding acquisition, Conceptualization. Anthony Mathur: Writing – review & editing, Supervision, Funding acquisition, Conceptualization.

Ethical statement

Continuum BP received a favorable opinion from the West of Scotland National Health Service research ethics committee (24/WS/0131).

All participants provided written informed consent and were made aware of the right to withdraw from the study at any time.

The Chief Investigator and study team ensured all participant identities were protected, and all information regarding participants was confidentially managed in accordance with the Data Protection Act (2018), the UK Policy Framework for Health and Social Care and Research Ethics Committee approval.

Funding

This work forms part of the research portfolio of the National Institute for Health Research Barts Biomedical Research Centre (NIHR203330). The funding body has no role in the design, conduct, analysis and interpretation of the study, or writing and decision to submit the article for publication.

Declaration of competing interest

AM and AmA are co-applicants on NIHR Barts Biomedical Research Centre Grant (NIHR203330). AM, AG and AyA are inventors on Queen Mary University of London's UK Patent Application No.2508620: Method for managing blood pressure. AM has received honoraria and travel bursary from Abbott. AM is a Board Member - IoNa Therapeutics: Company number 15704565.

Appendix A. Supplementary data

Supplementary material 1

Supplementary material 2

Supplementary material 3

Supplementary material 4

Supplementary material 5

Supplementary material 6

Data availability

The anonymized dataset will be available for data sharing upon request. The TMG will be given the opportunity to approve or deny any request for data sharing.