In September 2017, the U.S. Food and Drug Administration (FDA) made a striking announcement. Transforming its current regulatory practice for approving and certifying medical devices—the FDA announced a bold new plan, the Digital Health Software Precertification (Precert) Program, to offer an entirely new regulatory model to assess smartphone apps, wearables, sensors, and software. This transformation and medicalization of the consumer health market present both opportunities and obstacles, by opening up large markets for health monitoring and diagnosis using inexpensive mass-market, off-the-shelf devices. It also raises challenges, both related to privacy and effective uses of the devices to promote health. The Fitbit and Apple Watch are examples.
Initially, a glorified pedometer and electronic watch, Fitbit and Apple Watch, respectively, have expanded their focus (and sales pitches) increasingly toward healthcare, and many competing devices have been introduced with similar capabilities. Wearable monitors can now measure pulse rate, the electrocardiogram (ECG), blood oxygen levels via pulse oximetry, and other physiological parameters. These devices are being accompanied by smartphone apps that are intended to detect health conditions such as cardiac arrhythmias, or to monitor workers’ activity for occupational health improvement.
Apple Series 4 watch, released in September 2018, is a case in point. In addition to cell phone connectivity and GPS, and acceleration sensors, the watch has a pulse rate sensor (using photoplethysmography) and a set of ECG electrodes. On Dec. 6, 2018, Apple released an app that will analyze the ECG and detect atrial fibrillation (AF) in the wearer. “Capturing meaningful data about someone’s heart in real time is changing the way we practice medicine,” said Ivor Benjamin, president of the American Heart Association at the product’s introduction. The watch also tells time, by the way.
However, what are the health benefits of all this technology and what are its possible harms for the user? Many studies have been conducted on healthcare applications of wearables and answers are starting to emerge. The most reliable answers come from systematic reviews, which summarize results from numerous individual assessments.
Steps to better health
By far, the most research studies on health benefits of wearables have been done on activity monitors, chiefly the Fitbit but including other devices as well. As of January 2019, Clinicaltrials.gov lists 366 clinical trials involving the Fitbit. Most evaluate the Fitbit for monitoring or promoting physical activity in patients with various health conditions, but other applications have been examined as well, for example, to evaluate sleep. A search of Clarivate Web of Science found 186 articles on the search terms “Fitbit and health”—mostly published within the past 3 years—including a number of randomized trials and systematic reviews. Some assessed the accuracy of fitness monitors for assessing physical activity, others assessed health benefits from the use of the monitors.
The results are encouraging but also point to limitations of consumer-grade wearables for healthcare. In their April 2018 systematic review of 25 studies on weight loss programs that incorporate the use of wearable activity monitors, Cheatham et al. (California State University at Dominguez Hills) [1] found that, for middle-aged and older patients, the use of activity monitors increased the effectiveness of weight loss programs. In a 2018 review, Strath and Rowley [2] (University of Wisconsin) found that exercise programs using consumer-grade fitness monitors led to health benefits in patients being treated for cancer, pulmonary disease, and other conditions. However, the authors noted that most of the studies were short-term, and the exercise programs generally suffered from high dropout rates (in some studies, only one-third of the patients beginning a program were still participating after 6 months).
How well do Fitbit and other consumer activity monitors stack up against research-grade activity monitors? According to an extensive 2018 systematic review by Feehan et al. (University of British Columbia), the situation is mixed [3]. Although “Fitbit devices may provide similar measures for time in bed and time sleeping (as professional grade equipment),” they are prone to “markedly overestimating time spent in higher-intensity activities and underestimating distance during faster-paced ambulation.” They also tend to be unreliable when used with individuals with gait abnormalities. In addition, consumer-grade devices use proprietary algorithms that the manufacturer can change without warning, which makes it difficult to compare study results over time. However, the price is right, and for undemanding applications, they clearly work quite well.
But, for many applications, they may be inadequate without further testing. For fatigue monitoring of workers, for example, “devices that are used for making decisions about work, such as work–rest periods, breaks, and task reassignments, should be evaluated for accuracy against a gold standard across a range of the workplace settings where they would be expected to be used,” David Rempel, MD (professor of Occupational and Environmental and Medicine at the University of California at San Francisco, San Francisco, CA, USA) told one of the present authors recently. “If medical decisions will be made from this information, the devices will require FDA certification.”
Crossing the line
As wearable makers push more deeply into healthcare, their devices are crossing the line to become medical devices that are subject to FDA regulation. Although they may fit the definition of medical devices, FDA has expressed little interest in regulating low-risk fitness monitors that are promoted for general “wellness.” In practice, this means that companies can make hyperbolic claims for the effectiveness of their devices for promoting wellness while doctors are puzzled about how to effectively use the sometimes-unreliable data the devices provide.
Medical devices are subject to a host of regulations that consumer electronics companies are likely to be ill-prepared to deal with. To smooth the process, the FDA has introduced its Digital Health PreCert Program and so far has enrolled nine manufacturers of wearable devices including Apple and Samsung [4]. Apple worked directly with the FDA in this program to gain expedited clearance for the ECG analysis and heart rate sensing software for its new Apple Watch—formally crossing the line between consumer wearables for wellness to medical devices.
The FDA clears novel low-risk devices through its “de novo” process for new technologies such as wearables for which no “predicate” (already cleared and functionally equivalent) device is available for comparison. Thus, for low-risk (Class 1 or 2) devices, the path to FDA clearance is relatively short, as little as a couple months. Potentially, high-risk (Class 3) devices still require premarket approval, which requires clinical data to directly show that the device is safe and effective.
Some devices cleared under the de novo process can be medically quite significant. For example, in October 2018, the FDA cleared a device for continuous monitoring of blood glucose (Dexcom G6, Dexcom, San Diego, CA, USA). The device is a patch the size of a quarter that is mounted on the skin, with a small probe that penetrates the skin and provides blood glucose readings every 5 minutes. Readings are transmitted to the user’s smartphone (or, conceivably, smartwatch), joining a “wearable” to a minimally invasive sensor with potentially great significance to many diabetics.
As wearable technology progresses, medical applications are becoming more venturesome. In November 2017, AliveCor, another maker of wearable ECG devices, introduced its KardiaBand, a band for the Apple Watch that incorporates a pair of ECG electrodes and allows the wearer to record a onelead ECG. An app will identify AF or determine if the wearer is in normal sinus rhythm. (Unlike, apparently, the Apple Watch, some level of physician involvement in the process is required with Alive- Cor devices.) Clinicaltrials.gov lists 24 clinical studies involving AliveCor, mostly related to detecting AF, and evidence is strong that the technology can reliably detect the condition.
Going further, in September 2018 AliveCor won a coveted “breakthrough” designation from the FDA for an ECG-based method to detect hyperkalemia (high serum potassium), a potentially serious blood condition most commonly associated with kidney failure and other serious health problems by analysis of the ECG. According to press reports, the company is at least a year from receiving FDA approval, pending submission of clinical data. Cardiogram (https://cardiogr.am/) is developing apps for the Apple Watch to analyze heart rate data using machine learning methods to detect sleep apnea, hypertension, and diabetes. As the sensing capabilities of smartwatches, fitness monitors, and other consumer-grade wearable devices expand, we can expect a steady flow of new medical apps for smartwatches—including many of uncertain quality that claim to detect disease.
Should we use wearables to screen for health conditions?
The answer to this depends on medical judgment as well as on the technical capabilities of the devices themselves. Screening asymptomatic individuals for health problems with inexpensive off-the-shelf wearables has an undeniable appeal—one can already find testimonials on the Internet from patients who had been alerted to serious heart problems by their smartwatches. (Spoiler alert: expect many more such testimonials in the future; there are likely to be few testimonials from people who had unnecessary treatment due to false positive results.)
However, when many healthy people test themselves for conditions that they are unlikely to have, the risks of false positives are not trivial. It may lead to the detection of rare bouts of AF and diagnosis and treatment for a condition that may be present, but have questionable health relevance to the individual, a phenomenon called overdiagnosis.
Because of this, the medical community is undecided about the value of screening for AF in asymptomatic individuals. In August 2018, the U.S. Preventive Services Task Force, which reviews medical screening and preventive services, stated that “harms of diagnostic follow-up and treatment prompted by abnormal ECG results are well established” and concluded that “it is not possible to determine the net benefit of screening (of asymptomatic individuals) with ECG.” AF increases the risk of stroke, and there is undoubtedly some benefit in detecting rare sporadic bouts of AF in nonsymptomatic individuals, but the benefits are difficult to measure. By contrast, false positives and anticoagulant therapy for AF have known risks. Thus, despite the hype that accompanied the introduction of the latest Apple Watch, its net potential health benefits to consumers are unclear at present.
Screening for hyperkalemia raises similar problems. At a conference in summer 2018, scientists with AliveCor and the Mayo Clinic (an investor in AliveCor) reported analyzing 2 million ECG samples, of which about 20,000 (about 1% of the total) were from patients with hyperkalemia [5]. They reported a specificity of 72% for the test, which means that 28% of ECG recordings from patients without the condition tested positive. Since hyperkalemia is rare in healthy individuals, this means that many more false compared to true positives will be found, potentially overwhelming the medical system with false positives. The medical rationale for the approach is questionable on other grounds as well. “Electrical changes (in the ECG) due to hyperkalemia can be life-threatening but they are a fairly late finding when potassium is elevated and you would hope that the elevated potassium is caught before it starts to affect the heart,” Christopher Labos, MD (a cardiologist and Associate with the McGill Office for Science and Society, Montreal) told one of the present authors recently. Elevated potassium can be found by a simple blood test “which you should be having anyway if you are at risk for hyperkalemia because of renal failure or certain medications. I’m not sure how useful this monitor would be since we should ideally be identifying and treating these people before the problem starts, not only when it becomes critical,” Dr. Labos added. It hardly seems desirable to encourage ordinary consumers to test themselves for such conditions using an over-the-counter device— not the typical purchasers of the Apple Watch, for example.
It would be a potentially fatal mistake for a consumer to self-diagnose a heart attack using a smartwatch and decide not to seek treatment. Anybody experiencing symptoms of a heart attack needs to go to an emergency room as soon as possible and not rely on a smartwatch.
Privacy and security
Use of wearables to monitor the health of individuals raises a host of ethical issues. One is privacy. The same data that can be useful for medical purposes are also valuable for marketing—and an entire industry of data brokers exists today profits from selling patients’ personal health information gathered from apps and wearables that provide real-time data on geolocation, activities, and behavioral patterns. In fact, the business model of a healthcare app or consumer device often centers around selling user data—and apps that are free or low cost are often so for that very reason. New innovative efforts led by IEEE, the Ethics in Action program, seek to align ethics and design and create new standards that will help people protect their data and control access to it [6].
Privacy issues of a different sort will arise when an individual joins a wellness program offered by an employer or insurance company. Some of these programs provide a Fitbit or other activity monitor and collect the data in exchange for financial rewards. But what happens when the employer discovers that an employee with a high-risk job has difficulty sleeping? Would that data justify cutting back on overtime or reassignment to a less demanding job? Look at the problem this way: we give our days to our employers—do we want to tell them how well we sleep at night as well?
Ultimately, the health benefits of wearable devices depend only partly on their technical capabilities. They also depend on how the information they provide can be used (or misused) in healthcare. In this transformative period, new devices, new regulations, and new stakeholder are all emerging. Now is a critical time for the IEEE community to engage the medical community and ethicists and work toward the best use of this exciting but in some ways problematic technology.
References
- S. W. Cheatham et al., “The efficacy of wearable activity tracking technology as part of a weight loss program: A systematic review,” J. Sports Med., vol. 58, no. 4, pp. 534-48, 2018.
- S. J. Strath and T. W. Rowley, “Wearables for promoting physical activity,” Clin. Chem., vol. 64, no. 1, pp. 53–63, 2018.
- L. M. Feehan et al.. “Accuracy of Fitbit devices: Systematic review and narrative syntheses of quantitative data,” JMIR mHealth uHealth, vol. 6, no. 8, e:1-527, 2018, doi: 10.2196/10527.
- FDA, “Digital health software precertification (pre-cert) program.” Accessed: 2018. [Online]. Available: https://www.fda.gov/medicaldevices/digitalhealth/digitalhealthprecertprogram/default.htm
- C. D. Galloway et al., “Noninvasive detection of hyperkalemia with a smartphone electrocardiogram and artificial intelligence,” J. Am. Coll. Cardiol., vol. 71, no. 11, p. A272, 2018.
- IEEE, “Ethics in action.” Accessed: 2018. [Online]. Available: https://ethicsinaction.ieee.org/