Role of wearable rhythm recordings in clinical decision making—The wEHRAbles project

Abstract Background Multiple wearable devices for rhythm analysis have been developed using either photoplethysmography (PPG) or handheld ECG. Hypothesis The aim of this survey was to assess impact of these technologies on physicians' clinical decision‐making regarding initiation of diagnostic steps, drug therapy, and invasive strategies. Methods The online survey included 10 questions on types of devices, advantages, and disadvantages of wearable devices as well as case scenarios for patients with supraventricular arrhythmias and atrial fibrillation (AF). Results A total of 417 physicians (median age 37 [IQR 32‐43] years) from 42 countries world‐wide completed the survey. When presented a tracing of a regular tachycardia by a symptomatic patient, most participants would trigger further diagnostic steps (90% for single‐lead ECG vs 83% for PPG, P < .001), while a single‐lead ECG would be sufficient to perform an invasive EP study in approximately half of participants (51% vs 22% for PPG, P < .001). When presented with a single‐lead ECG tracing suggesting AF, most participants (90%) would trigger further diagnostic steps. A symptomatic AF patient would trigger anticoagulation treatment to a higher extent as an asymptomatic patient (59% vs 21%, P < .001). PPG tracings would only rarely lead to therapeutic steps regardless of symptoms. Most participants would like scientific society recommendations on the use of wearable devices (62%). Conclusions Tracings from wearable rhythm devices suggestive of arrhythmias are most likely to trigger further diagnostic steps, and in the case of PPG recordings rarely therapeutic interventions. A majority of participants expect these devices to facilitate diagnostics and arrhythmia screening but fear data overload and expect scientific society recommendations on the use of wearables.


| INTRODUCTION
Wearable devices have rapidly evolved over the last decade allowing a consumer-driven rhythm analysis on large segments of symptomatic or healthy individuals.
Devices utilize two methods for heart rhythm analysisphotoplethysmography (PPG) or ECG (Table 1). 1 PPG is based on light emitting and light sensing diodes that estimate heart rate from changes in blood volume caused by peripheral pulsations. Commercially available smartphones provide the technology to record PPG tracings, which is utilized by third-party applications. Some devices involve automated algorithms that can detect pulse irregularity and notify the consumer regarding a possible arrhythmia. Wearable single-lead or multiple lead ECG use electrodes that can be hand-held or implemented in a wristband or a smartwatch. Health and fitness technology is a rapidly evolving market showing a doubling in revenue within the past 5 years. 2 The aim of this survey was to assess impact of these technologies on physicians' clinical decision-making regarding initiation of diagnostic steps, drug therapy, and invasive strategies.

| METHODS
An online questionnaire was prepared using the EHRA Young EP infrastructure and distributed to EHRA Young EP members, members of national electrophysiology (EP) working groups and via social media platforms (Twitter, Facebook). The questionnaire included baseline questions on demographics as well as 10 questions on types of devices, advantages and disadvantages of wearable devices, as well as gaps in evidence. Three case scenarios for (a) a young patient with palpitations, (b) symptomatic atrial fibrillation (AF), and (c) asymptomatic AF were presented to ask for clinical decision-making. The full questionnaire is available in Supporting Information.
Questions were classified as nonmandatory. Responses were excluded if no answer was given at any clinical case scenario or two responses were submitted by one person. In case of missing data, pairwise deletion was performed.
Continuous variables are presented as mean ± SD or median (interquartile range-IQR). Categorical variables are presented as percentages and counts. Questions of clinical decision-making were compared using Wilcoxon test for dependent and Mann-Whitney-U test for independent variables. Consensus between respondents was measured using the consensus measure "C," ranging from 0 (no consensus) to 1 (complete consensus). 3 A two-sided P-value of <.05 was considered significant. Statistical analyses were performed using the SPSS 20.0 (IBM, Armonk, New York) and R 3.6.1 (The R Foundation, Vienna, Austria). Most of the participants were in a position to take clinical decisions either independently (75%) or under supervision (21%).

| Patient-scenarios for clinical decision-making
In a case example of a young patient with palpitations with on/offphenomenon presenting a 30 seconds tracing from a wearable device indicating narrow-complex tachycardia (in a single-lead ECG) or regular tachycardia (in a PPG tracing), most participants would trigger further diagnostic steps (90% in single-lead ECG, C = 0.70, 83% in PPG, Figure 1A). For half of the participants (51%), a single-lead ECG would be sufficient to perform an invasive electrophysiological study, while only 22% would indicate an invasive electrophysiological study based on a PPG tracing. Participants would be more reluctant to start antiarrhythmic drug therapy in this patient (28% for single-lead ECG, C = 0.55, 9% for PPG, C = 0.67).
In a case example of a patient with palpitations presenting a 30 seconds tracing indicating AF, most participants would trigger further diagnostic steps (90% for single-lead ECG, C = 0.72, 86% for PPG, Figure 1B). The majority of participants would start anticoagulation (if indicated by CHA 2 DS 2 -VASc-Score) based on a singlelead ECG (59%, C = 0.58), but not on a PPG tracing (21%, C = 0.59).
In a case example of an asymptomatic patient presenting a 30 seconds tracing indicating AF, most participants would also trigger further diagnostic steps (87% for single-lead ECG, C = 0.71, 79% for PPG, C = 0.68, Figure 1C). Participants would start anticoagulation (if indicated by CHA 2 DS 2 -VASc-Score) in 44% (C = 0.57) based on a

| Advantages and disadvantages of wearable devices
Participants saw the potential of wearable devices mainly in allowing faster diagnosis while the main concern was data overload (Figure 3).
In addition to a consensus document from the scientific society,

| Device types
The wearable ECG devices that participants of our survey labeled as most known to them, most available, used and recommended to patients or colleagues are Apple Watch and two AliveCor devices

| Interaction between user and healthcare team
Over the past decade, new wearable devices have allowed consumers instead of providers to take charge of collecting their heart rhythm data, thereby significantly expanding the amount of information collected on large segments of symptomatic or healthy individuals. In addition to data storage, deep-learning algorithms are now capable of distinguishing between sinus rhythm and arrhythmia (primarily AF) with reasonable accuracy. 6,7 The user-driven unselected recording of different parameters challenges the interaction between the potential patient and the healthcare team as there is a fear of data-overload from otherwise healthy people. However, this fear may be overestimated as screening for AF with a median time of 117 days with PPG-based tachogram in the Apple Heart Study only resulted in 0.5% notification of irregular rhythm. 4 The Huawei Study revealed that the rate of false positive alarms decreased with age, suggesting screening for AF could benefit at-risk populations. 8 When asked how the data from wearable devices should be shared with the healthcare team, only 9% of the present survey stated that they wanted a third party like the industry to provide the data.
F I G U R E 3 Advantages and disadvantages of wearable rhythm devices (multiple answers possible) However, as a group they were not settled who should receive the data, illustrating the challenges ahead with regards to current positions of the various sectors of the healthcare system in heart rhythm management.
Recently, the Heart Rhythm Society and the Consumer Technology Association has launched a joined guidance paper to the users of wearable technologies, that is, potential arrhythmia patients. 9 This is an important step to ensure appropriate use of wearable technology, but scientific society guidelines or position papers on wearable heart rhythm monitoring devices are warranted to help the healthcare teams tailor their interaction with the wearable device users but also addressing the established healthcare system's desired requirements to the wearable device industry.

| Clinical decision-making in patients with palpitations
Evaluation of intermittent palpitations and asymptomatic arrhythmias has long presented an unmet need in cardiology.
12-lead-ECG remains a cornerstone element for the diagnosis.
However, the sporadic and infrequent nature of supraventricular tachycardias (SVTs) makes it difficult to capture an episode on ambulatory monitoring. The 2019 ESC guidelines indicate that mobile recording devices may be required for the diagnosis of SVT. 10 They specifically highlight that wrist-worn, optically based heart rate monitors are user-friendly, but at the same time require an appropriate validation. Most participants of the present survey would refer their patients for an invasive EP study based on a single lead ECG tracing.
In a study involving healthy adults, Wang et al reported variable accuracy among different wrist-worn monitors, all being less accurate than a chest strap-based electrode containing monitor. 11 Guidelines for the management of AF state that the diagnosis of AF requires an ECG recording and that episodes of 30s are diagnostic. 12 Since AF is associated with increased risk of stroke, other morbidity and mortality, opportunistic screening is recommended in

| Advantages and disadvantages of wearable devices
The possibility of unlimited monitoring resulting in a faster diagnosis is one important advantage of wearable devices. It provides the opportunity to screen large populations. The traditional Holter and event recorders only monitor the heart rhythm for a limited period of time, are uncomfortable to wear and require healthcare staff to properly apply it to the patient. Implantable loop recorders can monitor for years but require an invasive procedure. Therefore, it is not inconceivable that wearables will lead to a reduced use of the current systems and therefore may be cost-effective.
Data overload is by far the most important disadvantage. Dealing with vast amounts of information coming from technology that might have limited accuracy is a big concern. Especially false positive results will lead to overdiagnosis, over treatment, and more workload for the physician.
The other important disadvantage of wearable devices is that this technology is patient-driven rather than clinician-driven. As shown in the Apple Hearth Study and the Huawei study a minority of the participants were over 64 years of age, and in the case of arrhythmia detection would rarely indicate a change of therapy. It is a concern that these technologies mainly reaches the younger, healthier population, leading to an overconsumption of healthcare in otherwise healthy subjects. 4

| Perspectives
It is anticipated that the use of wearable devices will increase from 325 million connected devices in 2016 to 1.1 billion devices by 2021. 18 As no proper guidance exists for interpretation of wearable device recordings, it is expected that this surge of wearable technology will most likely disrupt the traditional delivery of healthcare.
There is a lack of scientific data making it difficult to choose the right device type for the right patient. Trials comparing the diagnostic properties of the different wearable technologies and the effect on outcome are needed.
There will be a demand for reimbursement, not only of the specific technology, but also of the medical support to process all the generated data.

| Limitations
The present survey has limitations attributed to target respondents and questionnaire design. The survey was mainly spread through the scientific network of EHRA Young EP and participation was completely voluntary, therefore being prone to selection bias. Since the survey was distributed by the EHRA Young EP network and had relatively young participants more prone to use wearable technologies, this may reflect a selection bias and needs to be taken into account when interpreting the data. Furthermore, in the presented case scenarios, the preformulated answers may not represent all possible choices that could be made in individual clinical settings.
The current survey focusses on patient-triggered recordings presented to physicians and does not differentiate between advantages, disadvantages and availability of different types of wearable devices.

| CONCLUSION
Physicians from more than 40 countries are well aware of current wearable rhythm devices and already use and recommend these novel technologies. Tracings from wearable rhythm devices suggestive of arrhythmias are most likely to trigger further diagnostic steps, and in the case of PPG recordings rarely therapeutic interventions. A majority of participants expect these devices to facilitate diagnostics and arrhythmia screening but fear data overload and expect scientific society recommendations on the use of wearables.