Recent years have seen an explosion in consumer wearable devices, with many now Food and Drug Administration (FDA) cleared. These devices provide health information to the person wearing the device such as heart rate and rhythm, particularly irregular rhythms such as atrial fibrillation.
Devices are now being marketed directly to the public. For example, a $99 credit card-sized monitor advertised as being able to “record accurate EKGs (electrocardiograms).”
This technology, however, poses a challenge when it intersects with standard clinical medicine practice. Issues to consider include reliability and validity of algorithms to interpret data from the device, false-positive findings that lead to further testing, and patient expectations that physicians will review, interpret and formulate a management plan (often via chart messaging). This article will discuss the current state of wearable devices, particularly those that provide data on heart rate and rhythm, and the implications for clinical practice.
Overview of consumer wearable technology
Wearable technology includes a wide array of devices and monitoring abilities. Many consumer wearables are used by fitness enthusiasts and competitive athletes to track training and boost performance. Others focus on health-related metrics such as blood pressure, sleep quality, and heart rate and rhythm. In 2022, approximately 30% of adults in the U.S. used a wearable electronic device, and predictions are that use will continue on an upward trend, according to results from a study in the Journal of Medical Internet Research.
FDA oversight
Many of the devices are FDA cleared as Class I (low-risk) general wellness products. In 2018, the Apple Watch Series 4 received FDA clearance as a Class II medical device because of its ability to perform specific health-related functions. A Class II device is under regulatory oversight by the FDA and therefore must meet strict safety and efficacy standards. Note that FDA clearance is different from FDA approval; approval requires rigorous testing to demonstrate safety and efficacy. For example, the Fitbit device only needed FDA clearance but its app to track atrial fibrillation needed FDA approval.
Cost and third-party coverage
Are consumer wearable devices covered by third-party payers? Most payers follow Centers for Medicare & Medicaid Services (CMS) reimbursement schedules. These devices are generally not covered because they are considered to be for health and fitness purposes and not medically necessary, although Medicare Advantage plans may offer limited coverage, according to CMS. For some, flexible spending or health savings accounts can offer reimbursement.
How much do these devices cost? There is a wide range of models, with fancier ones costing more. The KardiaMobile Card EKG monitor lists for $99, and the original KardiaMobile personal EKG monitor lists for $79. At the other end, the Apple Watch Series 8 can cost up to $499.
How do these devices work?
Consumer wearable heart monitors provide data on heart rate and rhythm in real time using either an electrode-based EKG method or optical photoplethysmography (PPG). For the EKG method, an electrode (conductive pad) is attached to the skin and enables the recording of electrical currents. The EKG tracing is a graphical rendering of this electrical activity.
PPG sensing is based on light-emitting diodes at specific wavelengths directed toward the skin. The intensity and pulsations of the blood vessels throughout the cardiac cycle change the light intensity that is reflected back to the photodetector and reported as heart rate. Automated proprietary algorithms use PPG data to derive secondary variables, including heart rhythms. Some devices offer both PPG as well as capability for EKG tracing.
Note that these algorithms are designed to screen for atrial fibrillation; however, the confirmation and diagnosis require physician interpretation and often further testing. The clinical standard of care is a 12-lead EKG with electrodes recording electrical activity from 12 different locations to provide more complete information to diagnose abnormal heart rhythms.
How valid and reliable are the PPG-based algorithms for detecting abnormalities?
Many factors can affect the quality of data obtained with PPG. Faulty data can result in inaccuracies carried downstream when secondary variables are derived. The algorithms are proprietary to each manufacturer and vary in sensitivity, specificity and overall ability to determine heart rhythm. Manufacturer validation generally takes place in controlled research environments and not real-world settings, according to the Journal of the American College of Cardiology.
Sources of error include motion at the site where the device is worn, improper fit and skin tone, with decreased accuracy in darker skin tones because this affects the strength of the PPG signal. Devices with EKG capability require user interaction to record the tracing.
Unlike standard 12-lead EKGs, most devices with EKG capability have one lead that provides limited information from which to interpret the tracing. A newer model device, the KardiaMobile 6L, offers six leads. They are all limb leads as opposed to the standard 12-lead EKG with both limb and precordial or chest leads. However, the 6L has the key leads for determining presence and characteristics of the P wave of EKG tracing, which is important for detection of possible atrial fibrillation.
Is artificial intelligence (AI) being used? Not yet, and at this time it is unclear what, if any, role AI will have in the summary, integration and interpretation of information from these devices.
Clinical studies have investigated the reliability and validity of these devices. In the BASEL Wearable Study, investigators sought to validate five direct-to-consumer wearable smart devices in their ability to detect atrial fibrillation. Although atrial fibrillation was detected and confirmed in many patients, approximately 25% had inconclusive tracings, which diminished the sensitivity and specificity of the devices.
In a study published in the Canadian Journal of Cardiology, researchers tested Apple Watch Series 5’s ability to detect atrial fibrillation in people who had abnormal baseline EKG measurements, which is relatively common. They found that the Apple Watch produced high rates of nondiagnostic tracings, false-positive, and false-negative results.
Benefits of wearable consumer devices
The Heart Rhythm Society, an international professional organization of cardiologists who specialize in cardiac arrhythmia, published “Guidance for Wearable Health Solutions” in 2020, which highlighted some of the benefits of wearable technology. These include giving people the ability to participate in their own health care, increased feelings of safety and confidence, and access to real-time information on cardiac metrics.
One benefit is the potential for early detection of atrial fibrillation. Patients with atrial fibrillation are at risk for blood clots that can cause a stroke, which is why many patients with atrial fibrillation are on anticoagulation medication. Early detection by consumer wearables can help prevent this from occurring. This was shown in results from the Fitbit Heart Study, which was designed to test a novel PPG-based algorithm for detecting undiagnosed atrial fibrillation.
The algorithm worked well (although it could only be used during periods of inactivity) and produced a very high positive predictive value in finding new atrial fibrillation.
Concerns, unknowns and impact on clinical practice
Impact of false-positive results
Many feel that these devices still lack adequate sensitivity and specificity. False-positive results can lead to ordering further testing and excessive use of health care resources. A 2020 study in the Journal of the American Medical Informatics Association looked at 164 patients who had an abnormal pulse when wearing Apple Watch Series 1 and newer versions through April 2019. The patients went on to have further testing, but only 11.4% had a clinically actionable cardiovascular diagnosis and 4.9% had atrial fibrillation. The authors concluded that false-positive screening results may lead to overuse of health care resources.
What information from consumer wearable devices is clinically relevant?
For instance, what findings would prompt prescription of anticoagulation? A 2023 randomized clinical trial published in the New England Journal of Medicine investigated whether anticoagulation was justified for implanted device-detected atrial high-rate episodes (AHREs). These are transient episodes of heart rates of at least 175 beats per minute and lasting at least 6 minutes that resemble atrial fibrillation and may be associated with paroxysmal atrial fibrillation. Among participants 65 years or older with AHRE episodes, anticoagulation with edoxaban (a direct factor Xa inhibitor) did not reduce the incidence of cardiovascular death, stroke or systemic embolism compared with placebo but led to a higher incidence of a composite death from any cause or major bleeding.
Data security, privacy, and HIPAA
Are wearers vulnerable to information leaks? It seems so. Although apps have their own privacy and data use policy, they are not covered by the Health Insurance Portability and Accountability Act (HIPAA). Data can be stored by some of these devices and used by industry. In 2021, 61 million health records from FitBit and Apple Watch users were leaked in a data breach. Data including the user’s name, gender, age and geographic location were exposed, leading to calls for increased data security for wearable devices.
Impact on clinical practice
Given the concerns about reliability and validity of information from these devices, can physicians act on the information without additional verification and testing? The consensus is no and that physicians need to review the device findings often and confirm with further testing.
But what about patients who are accustomed to using electronic health record patient portals to message their provider, often with attachments including tracings from their wearable devices? Clinics and practices may not be prepared to integrate this volume of data and ensure it is collected, properly encrypted and imported into the electronic record along with timely review and response from a provider.
Many institutions are now billing for messages such as these. Starting in November 2022, Cleveland Clinic announced it would bill for MyChart responses that required physician clinical time and expertise to answer. In July 2023, Johns Hopkins University in Baltimore, Maryland, developed a similar protocol. Both institutions also stated that it could take up to three business days to read and respond to a message and that messages were not reviewed in the evening or on weekends.
A physician’s perspective
Gregory M. Marcus, M.D., an electrophysiologist and professor of medicine at University of California, San Francisco, wrote the following in a 2020 opinion piece in Circulation: “The normal paradigm, where original research provides evidence that then informs guideline committees, who then communicate their consensus to clinicians, who then instruct their patients accordingly, has been disrupted. Now, industry sells devices directly to consumers, who then approach clinicians with information about the possible presence of atrial fibrillation.”
Furthermore, physicians embrace the paradigm that you only order a test if you have reason to do so and intend to act on the results. The U.S. Preventive Services Task Force recommends against screening with a resting EKG to prevent cardiovascular disease (CVD) events in asymptomatic adults at low risk for cardiovascular events.
They add that there is insufficient evidence to assess the balance of benefits and harms of screening with resting EKG to prevent CVD events in asymptomatic adults at intermediate or high risk for CVD events. Many and perhaps most wearers of these devices do not have a history of abnormal heart rhythms or signs and symptoms suggesting that they might.
“It’s important to distinguish the utility of these devices among those who are asymptomatic without an established arrhythmia diagnosis versus those with either symptoms of an arrhythmia or a previously established diagnosis,” Marcus writes. “I worry primarily about the use of these devices to screen asymptomatic individuals for a new arrhythmia diagnosis.”
Future directives
Development and direct-to-consumer marketing are not only here to stay but will continue to grow. So what should be done?
Professional medical organizations should follow the lead of the Heart Rhythm Society and its white paper on clinical use of digital health technology and continue to develop expert consensus guidelines explaining what to do with results from these devices. These can provide protocols for use of information from wearable devices, further testing algorithms, and emphasize that patients have formal appointments with their health care provider to allow appropriate evaluation.
Marcus adds that “specialists in cardiology and cardiac electrophysiology along with professional organizations have a role to play in educating the public about the meaning of the types of rhythm disturbances that can be detected by these devices. For example, because the main point is to detect asymptomatic arrhythmias that might have adverse consequences in the long term, an alert from a smartwatch in someone who is feeling fine should not be viewed as an emergency.”
More studies are needed to show cost-effectiveness and that patient care is improved by these devices. Companies marketing these devices can continue to improve algorithms; report sensitivity, specificity and negative predictive values while avoiding unsubstantiated claims.
The development of consumer wearable devices has both potential and pitfalls but is now indispensable for many and a boon for industry. Understanding the limitations and developing ways to best integrate with established clinical medicine will be needed to fully realize the benefits for health care.
Medical Economics | October 30, 2023