Wearable devices—addressing bias and inequity

Abstract

The increasing digitisation of health care has led to wearable devices becoming a frequent contributor to health-care decisions.1Al-Alusi MA Khurshid S Wang X et al.Trends in consumer wearable devices with cardiac sensors in a primary care cohort.Circ Cardiovasc Qual Outcomes. 2022; 15e008833Crossref PubMed Scopus (1) Google Scholar Wearable devices are a developing group of health-care technologies that include smartwatches and fitness trackers. These technologies have a substantial number of users, with a fifth of adults in the USA now using a wearable device.2McCarthy J One in five US adults use health apps, wearable trackers.https://news.gallup.com/poll/269096/one-five-adults-health-apps-wearable-trackers.aspxDate: 2019Date accessed: April 4, 2022Google Scholar These devices provide objective data that can enhance the doctor–patient relationship and improve the overall quality of clinical encounters. Unfortunately, survey data from six federally qualified health centres in the USA published in 2022 show that wearable devices and other digital technologies are not used as widely in low-income and minority populations, with cost and education being issues that substantially affect use.3Holko M Litwin TR Munoz F et al.Wearable fitness tracker use in federally qualified health center patients: strategies to improve the health of all of us using digital health devices.NPJ Digit Med. 2022; 5: 53Crossref PubMed Scopus (2) Google Scholar In this Comment, we attempt to bring attention to the increasing issues in generalisability for wearable devices driven by the non-inclusion of these populations in ongoing research studies. Historically, the implementation of any high-risk medical device or drug therapy requires thorough validation in a representative population. Digital technology companies have avoided traditional approval processes by originally marketing consumer products as low-risk wellness tools. The increase in the number of wearable devices approved by the US Food and Drug Administration (FDA) requires the conversation on bias and equity within the wearable device industry to be broadened, especially as these consumer devices (eg, smartwatches) are increasingly applied to medical indications, such as arrhythmia detection and at-home pulse oximetry. The FDA 510(k) clearance process, which is the way to FDA approval for most wearable devices, only requires equivalent safety and efficacy to products that are already available. With initial wearable device studies not being proactive in setting requirements for equity, the deficits in the representation of people of colour, rural communities, and populations with low digital literacy have continued. The future of chronic disease management relies on collaboration between telemedicine, sensor technology, and artificial intelligence. Extensive research on incorporating artificial intelligence into wearable devices has already produced positive results in arrhythmia detection and physical fitness. Nonetheless, the continued lack of diversity in study populations has increased the risk of poorly generalisable models of artificial intelligence. The Apple Heart Study,4Perez MV Mahaffey KW Hedlin H et al.Large-scale assessment of a smartwatch to identify atrial fibrillation.N Engl J Med. 2019; 381: 1909-1917Crossref PubMed Scopus (628) Google Scholar with 419 297 participants, is the most comprehensive study to evaluate the accuracy of an irregular pulse detection algorithm in a general population. The high positive predictive value of an initial irregular pulse notification for subsequent electrocardiogram-confirmed atrial fibrillation allowed for FDA clearance of this feature. However, the requirement to own an Apple product biased the study with a young, wealthy, and technology-proficient population. A 2022 clinical validation of atrial fibrillation detection with an Apple Watch in a population with a mean age of 76 years found a discrepancy in sensitivity compared with the initial results reported by Apple. Even when excluding inconclusive readings, the sensitivity of the algorithm was 50%, which was lower than the 96% from Apple.5Ford C Xie CX Low A et al.Comparison of 2 smart watch algorithms for detection of atrial fibrillation and the benefit of clinician interpretation: SMART WARS study.JACC Clin Electrophysiol. 2022; 8: 782-791Crossref PubMed Scopus (4) Google Scholar Since 2020, other companies (eg, Fitbit and Samsung) have obtained similar FDA clearance for passive arrhythmia monitoring. However, each independent algorithm will require extensive clinical validation, highlighting how equity issues will persist as wearable device use continues. The lack of representative study populations has produced identifiable bias in wearable devices and sensors used in vital sign monitoring. In 2021, the FDA warned that at-home, digitally connected pulse oximeters might not recognise severe hypoxemia in people of colour.6Sjoding MW Dickson RP Iwashyna TJ Gay SE Valley TS Racial bias in pulse oximetry measurement.N Engl J Med. 2020; 383: 2477-2478Crossref PubMed Scopus (213) Google Scholar Similarly, the ability of smartwatches to track heart rates in people of colour has been questioned. Most heart rate monitors use photoplethysmography, an optical technique that measures changes in blood volume to find out heart rate. Despite studies published in 2020 and 2021 providing conflicting evidence on the effectiveness of photoplethysmography in people of colour, accessible subgroup analysis is not available because of the absence of reporting by ethnicity in validation studies.7Bent B Goldstein BA Kibbe WA Dunn JP Investigating sources of inaccuracy in wearable optical heart rate sensors.NPJ Digit Med. 2020; 3: 18Crossref PubMed Scopus (151) Google Scholar, 8Colvonen PJ Response to: investigating sources of inaccuracy in wearable optical heart rate sensors.NPJ Digit Med. 2021; 4: 38Crossref PubMed Scopus (9) Google Scholar For many wearable devices, validation studies were done in homogenous samples with inadequate external validation in representative populations. These measurable deficits in representation show that inclusive studies should prospectively evaluate wearable devices in diverse populations. These issues with accuracy in a general population are concerning, especially as research about wearable devices includes cardiovascular indications, such as ambulatory blood pressure measurement, heart failure detection, and drug monitoring. Despite FDA efforts to increase reporting of demographic subgroups in validation studies, enrolment of diverse study populations continues to be an issue. Most validation studies rely on participants using their own devices, which means they are dependent on volunteers who already own the relevant digital technology. A 2022 review of demographic characteristics of studies relying on participants using their own devices found that it leads to preferential enrolment of White participants compared with participants from minority groups.9Cho PJ Yi J Ho E et al.Demographic imbalances resulting from the bring-your-own-device study design.JMIR Mhealth Uhealth. 2022; 10e29510Crossref Scopus (4) Google Scholar The resulting unrepresentative study populations have a lack of generalisability that is exacerbated by the absence of accessible reporting of demographic information. Issues with equity require innovative solutions from all important stakeholders to increase the accessibility of wearable devices for under-represented communities. In this Comment, we offer potential ways to prevent an increase of existing structural disparities (figure). The research community and regulatory agencies should use processes other than traditional approval processes to address ongoing data transparency and equity concerns. Furthermore, digital device companies should make wearable devices available in community centres (eg, pharmacies and shops) as affordable alternatives to expensive consumer smartwatches to create representative study populations. For example, some pharmacies have partnered with decentralised clinical trial platforms to increase access to clinical research studies, including investigations focused on wearable devices. Furthermore, we hope that clinicians will work to proactively develop professional guidelines on validated limitations of digital health devices. Wearable devices have potential in cardiovascular management, with health-care systems having optimism about their future role in several clinical scenarios, such as early detection of heart failure exacerbation and ambulatory blood pressure management. As a substantial amount of clinical care becomes digital, it should be supported by objective data from wearable devices. This support will require a collaboration between patients, clinicians, technology companies, and the research community to incorporate equity concerns into the approval process for wearable devices. An increase in diverse representation will allow health-care systems to make informed health-care decisions with high-quality, generalisable data from wearable devices. AZ is a consultant for Heartbeat Health. JPS is a consultant for Abbott, Biotronik, Boston Scientific, Cardiologs, CVRx, Cardiac Rhythm Group, EBR Systems, Impulse Dynamics, Implicity, Medtronic, Medscape, Microport, New Century Health, Orchestra BioMed, and Sanofi.

Topics & Concepts