Every Step You Take: Tracking Patients Using Wearable Technology
If your wrist has recently told you that it is time to get up and walk around, you are not alone. Wearable technology, exemplified by household brand names like Fitbit, Garmin and Apple Watch, has become a multibillion-dollar industry that is only set to expand in coming years. They are everywhere, it seems.
Direct-to-consumer ads show people with these devices walking, running, swimming, climbing mountains, becoming healthier and leaving their tired old selves behind. However, the reality may be somewhat different, not only for healthy individuals, many of whom stop wearing the watch after a short period of time, but for those with diseases like diabetes or knee osteoarthritis, who are increasingly being led to believe that such wearable technology is the answer to their very complicated disease state.
Although there are a number of potential benefits to a wearable device that constantly tracks activity and other health data, whether current devices are gathering the right data — and whether clinicians and researchers are making optimal use of those data — remains to be seen. In the rheumatology field, although most experts acknowledge that anything that can help patients get up and moving is a good thing, questions persist.
Many clinicians, including William G. Dixon, PhD, director of the Arthritis Research UK Centre for Epidemiology, and chief investigator of the Remote Monitoring of Rheumatoid Arthritis (REMORA) Study, see significant opportunities for the type of data that can come from apps and wearable devices. “We as clinicians see our patients for about 20 minutes every 6 months, or much less than 1% of their lives,” he said. “The rest of the time, we need to support them as they self-manage their disease, doing things like physical activity. If we can see — or track — what they are doing in the months between their visits, we can use this technology to deliver interventions suited to their specific needs and activity levels.”
To this point, Jeffrey R. Curtis, MD, MS, MPH, of the University of Alabama at Birmingham, and a member of the Healio Rheumatology peer perspective board, stressed the importance of the human element in the equation. “This isn’t a technology problem, this is a social and behavioral problem,” he said. “We need to generate evidence, which is what these devices are doing, but we also need to use this evidence to create personalized care.”
In a perfect scenario, clinicians could use streamed data from a device to show patients where they are lacking and how they can improve. However, Linda Li, PT, PhD, a senior research scientist of clinical epidemiology at Arthritis Research Canada, warned of overreliance on hard numbers. “Many patients have said it’s great to have some idea about their physical activity level,” she said. “But they also reported that sometimes they are experiencing a flare-up and are simply not able to do as much as they would like. Then they are showed evidence of this, and just seeing that alone can be discouraging.”
In an effort to capitalize on the possibilities and deal with the issues surrounding emerging technologies, the FDA recently launched the Digital Health Unit, a broad-spectrum initiative in health IT, mobile health, personalized medicine, telehealth and telemedicine and wearable devices. The FDA hopes that the use of smart phones, social networks and apps can help patients focus more directly on health and wellness outcomes, and help clinicians tailor interventions to the individual patient. More importantly, tech developers have been involved in the initiative, and are encouraged to collaborate with patients, health care experts, researchers and traditional medical device companies to ensure that subsequent generations of products improve on current technology.
It is difficult to determine what all of this means for patients with rheumatic and autoimmune diseases. One concern is that, at the moment, there are few data looking at how these devices have impacted this patient population. In addition to concerns such as cost and access, privacy and integration with other health data, there remains considerable debate surrounding the clinical importance of various thresholds of the most basic data point these devices deliver: step count.
In their study published in Evidence-Based Mental Health, Chum and colleagues examined how the Fitbit could impact a nonrheumatologic parameter — depression — in a cohort of 36 patients who completed a 28-week behavioral activation therapy program to improve physical activity and mental health. According to study results, 23 of the patients found the Fitbit a useful tool for improving physical activity. While they reported self-awareness, peer motivation and goal-setting opportunities as positive outcomes of the device, the patients also reported inconvenience, inaccuracies and disinterest. The researchers concluded that although the device was a good complement to the behavioral intervention, there are both positive and negative associations with it.
“Digital health is not a computer science, it’s a social science,” Brennan Spiegel, MD, MSHS, director of Health Services Research for the Cedars-Sinai Health System and director of the Cedars-Sinai Center for Outcomes Research and Education, said in an interview. “Although we have electronic tools and wearable biosensors, we haven’t solved the age-old problem of how to have people change their behavior in measurable ways.”
According to Spiegel, the question is how to apply the decades-old best practices of health to new technologies. “These devices should be used as tools to provide feedback and monitor patients in a cycle of improvement, rather than just as a one-way information stream that isn’t meaningful, engaging or tailored to the patient’s life,” he said. “That is why people stop using Fitbits so quickly — they don’t feel like they are getting any additional information, guidance or insight that they couldn’t already obtain on their own.”
Many recent data sets bear that out. Published in Digital Medicine, Noah and colleagues conducted a meta-analysis of thousands of studies regarding remote patient monitoring and clinical outcomes that appeared in PubMed between January 2000 and October 2016. Results from 16 high-quality studies showed that remote patient monitoring failed to impact six reported outcomes in a statistically significant way, including BMI (0.73; 95% CI, 1.84 to 0.38), weight (1.29; 95% CI, 3.06 to 0.48), waist circumference (2.41; 95% CI, 5.16 to 0.34), body fat percentage (0.11; 95% CI, 1.56 to 1.34), systolic blood pressure (2.62; 95% CI, 5.31 to 0.06), and diastolic blood pressure (0.99; 95% CI, 2.73 to 0.74).
“Interventions based on health behavior models and personalized coaching were most successful,” the researchers concluded. “We found substantial gaps in the evidence base that should be considered before implementation of remote patient monitoring in the clinical setting.”
Li acknowledged the necessity of coaching and clinical intervention in the use of wearables, but she nonetheless stressed the positives that can come from these devices. “It can serve as a reminder, just wearing something on your wrist or your belt that notifies you to get moving,” she said. “It becomes part of your personal environment.”
The interplay between patient, health care professional and technology is critical, according to Curtis. “We can work with patients using a guided interview to set goals,” he said. “Then we can talk about personalized reminders, via the wearable device or text, that promote health behaviors that the patient has selected. In this way, it is not the doctor or the Fitbit nagging you, it’s the patients nagging themselves. Patients can set their own messaging and triggers, but it’s shared goal setting with their clinician.”
Clinicians, then, should be capable of interpreting the data that come from the patient and the device, according to Spiegel. “We have to constantly be thinking about how we gain information, knowledge and wisdom from the ones and zeroes that come from these sensors,” he said. “We need to convert data into information, information into knowledge, knowledge into wisdom and then convert all of this into improved care.”
The Myth of 10,000 Steps
An increasing number of Americans can cite 10,000 steps per day as a significant benchmark for physical activity. However, a cursory investigation traces this number back to the 1960s when the number was used in a slogan in Japanese walking clubs and a marketing campaign by a consumer pedometer. The number was also used during the Tokyo Olympics to encourage citizens to be active, but according to Li, “this number wasn’t based on any research study that says everyone needs to reach that threshold to achieve good health.”
In fact, evidence is still lacking for that particular number, according to Li. “It can make patients frustrated,” she said. “They ask: Am I actually doing enough? Should I do 10,000 steps even if my knee hurts? More? Less?”
Li acknowledged that some sort of baseline or benchmark is important, both for patients to track their progress and for physicians to observe how various activity levels are impacting disease outcomes. Yet many experts, including Daniel Kenta White, PT, ScD, MSc, assistant professor in the department of physical therapy at the University of Delaware STAR Health Sciences Complex, believe that closer to 6,000 steps may be a more optimal target, particularly for patients with knee OA. “The target of 6,000 steps per day is a good benchmark to determine whether a patient with knee OA is at risk of developing functional limitations or not,” he said.
White also said that the number itself is not as important as having a goal in mind. “The key feature of this is having a step goal in mind, and working toward it each day,” he said. “It doesn’t necessarily matter whether it’s 10,000 or 6,000 or some other number that you and your patient have worked out.”
Perhaps more importantly, White outlined two critical components of the equation. “A step is a step, which makes it a useful parameter as a research tool,” he noted. “And walking is the most common physical activity older adults employ for exercise. This is incredibly useful information to work with.”
For Spiegel, the issue comes back to integrating technology with the patient. “You already have 100 million sensors in your body,” he said. “You don’t need a Fitbit to know if you are walking a lot or not. The question is, are we going to develop the type of sensors we can listen to in a different way?”
White suggested that Fitbits and personal activity tracking software could represent a significant frontier for the rheumatology community. “Physical activity such as this is an incredible area of research that is perhaps even more important than medications, and maybe even surgery, particularly in patients with knee OA,” he said, and noted that attempts to cure this disease have mostly been unsuccessful. “Medications have side effects. Exercise is incredibly effective. If we can find ways to harness physical activity and promote it in our patients, we may move the field further than we ever have before with pharmacotherapies.”
Evaluating Other Parameters
In a recent study in Gait & Posture, Richards and colleagues studied the utility of a 6-week toe-in gait training program in a cohort of 16 patients with knee OA to determine whether patients could learn to change their foot progression angle. Clinicians evaluated patients walking naturally, walking with feedback, walking without feedback, and walking with a dual-task at the start and end of the training program. Natural walking yielded a foot progression angle difference from median 6.9 to median 3.6 degrees — for example, by 3.3 degrees in week 6 (P < .001). However, adding feedback reduced the difference in foot progression angle nearly to zero, according to the findings.
“Traditionally, there are three Vs of big data,” Curtis said. “Those are velocity, variety and volume. But I would add a fourth: Value. Patients want value from their data: Why are you asking me to give this data? Why are you asking me to share it? We, as clinicians, need to be able to show patients what we are going to do with it. More importantly, if we can show them that we have improved their condition, they feel like they are getting value for it.”
In their study published in Telemedicine Journal and E-health, Liu and colleagues compared the Fitbit Alta with a consensus sleep diary among healthy young adults to determine parameters of recording sleep. Results showed that the Fitbit recorded 437.15 minutes of total sleep time per night, compared with 442.61 minutes for the diary, a difference of 5.46minutes. The researchers also noted that the Fitbit recorded 2.15 more awakenings per night than the diary, comprising 13.09 minutes more wake time after sleep onset. In short, the device was more accurate than the patient in reporting sleep habits.
Dixon underscored the importance that this type of technological accuracy can bring to both the research arena and to patient care. “In a study we were doing recently, we had patients self-report symptoms on a daily basis for three months with >90% completeness,” Dixon said. “We successfully integrated these patient-reported outcome data into electronic health records. This changed the nature of the consultation positively. Instead of the patient having to remember how they have been doing over the last 6 months, with their long-term memory biased by recent symptoms, we have real-time data illustrating how they have been day-to-day.”
In an effort to monitor physical activity and total sleep time among patients with RA, McKenna and colleagues evaluated the use of the SenseWear Pro3 Armband in their study in Rheumatology International. The researchers reported a significant positive relationship between total sleep time and physical activity (P = .018), with a low total sleep time for patients with RA.
“People with RA who are more physically active have longer total sleep time,” the researchers wrote. “These findings provide an objective profile of total sleep time and physical activity duration in people with RA and suggest a relationship between increased physical activity duration and longer total sleep time.”
“If I can monitor your sleep and mobility, I can take better care of you,” Curtis noted. He also stressed that the devices can help doctors illustrate to patients the association between symptoms — or, in this case the relationship between physical activity and sleep. “As rheumatologists, we can show our patients that if they can, say, increase their step count by 1,500 steps, their fatigue will go down in a significant and measurable way,” he said. “It’s all part of the quantified self. When you do X, then Y can happen.”
A big question to consider is how data from wearable technologies will fit into the paradigm of traditional epidemiology and electronic health records. “What became very apparent in the research that the Patient-Centered Outcomes Research Institute [PCORI] has funded is that patients often don’t care about the same metrics that rheumatologists care about in clinical trials,” Curtis said. However, PCORI has endeavored to make use of EHR, claims and registry data to determine, more accurately, how to better serve patient needs. “What was missing from that equation was a way to see what patients were doing 365 days a year. Now we are starting to get that real-world data.”
At the moment, much of this data integration is still in the speculation stage, according to Dixon. “If we look to the future, it’s easy to see where we will be able to develop algorithms where physical activity, as measured by wearable technology, can serve as a surrogate marker for disease state,” he said. At the moment, much of this data integration is still in the speculation stage, according to Dixon. “If we look to the future, it’s easy to see where we will be able to develop algorithms where physical activity, as measured by wearable technology, can serve as a surrogate marker for disease state,” he said. “We will be able to see different patterns for ankylosing spondylitis, RA, or osteoarthritis, and we will be able to see patterns for when flares occur. This opens opportunities for timely intervention or ‘anticipatory medicine’.”
Curtis discussed next-level information that could be gathered by integrating wearable technology with social media and other types of information. “Patients with arthritis often develop symptoms with changes in weather and barometric pressure, so this is something we could look at,” he said. “Sensors that measure ultraviolet light could be beneficial for patients with photosensitivity associated with lupus. Additionally, there is technology to track whether a patient has actually ingested a pill or not.”
Li discussed other features included in several of today’s wearable technologies. “Some of the research-grade sensors that we use contain an inclinometer, sensors to measure different physiological variables like skin moisture and temperature to determine if you are sweating,” she said. “When we work with patients, we get more data than consumer-grade devices can provide. It is important to know what consumer wearables can and cannot do when we coach patients to monitor their physical activity with these devices.”
Geofencing — which uses GPS or RFID technology to form a virtual geographic boundary — is another possibility, according to Curtis. “The same technology that allows you to find the nearest healthy restaurant or pharmacy can be used to put barriers around McDonald’s or a pizza place in order to encourage patients to make healthier food choices,” he said. “The sky is really the limit on this.”
Spiegel underscored all of these potential uses and stressed the importance of keeping a focus on clinical utility. “We have to make sure we are using this to tailor the technology to individuals across multiple domains,” he said. “We can characterize patient needs like a fingerprint, and use that information to drive interactions. However, if we are just giving someone technology and foisting it upon them to change deep-rooted behaviors, like smoking or eating right, we already know this is going to fail.”
At the moment, though, this type of technology and data integration is still in its incipient stages. With that in mind, the next key question to consider is potential invasion of privacy that may come from patients essentially live-streaming their lives via a device on their wrist. “No one wants to feel like Big Brother is watching, so we need to be doing everything we do with full patient consent,” Curtis said. “What it comes back to is if patients understand that they are benefitting from wearing these devices, they will feel better about wearing them and sharing the information.”
Li highlighted cost as another factor in the current landscape of wearables. “If you have a commercially available device that is $100 or $200, it’s a good device, but it may not be as sophisticated as the ones we use for research,” she said. “It’s hard to guess when the cost of wearable technologies will come down to more affordable levels. But when the technologies continue to improve, the accuracy of the information we can gather will also improve.”
The future of these technologies, then, seems bright. In the meantime, patients will simply be looking at their wrists, waiting for their next move. – by Rob Volansky
- Chum J, et al. Evid Based Ment Health. 2017;doi: 10.1136/eb-2017-102763.
- Liu J, et al. Telemed J E Health. 2019;doi:10.1089/tmj.2018.0202.
- McKenna S, et al. Rheumatol Int. 2018;doi:10.1007/s00296-018-4009-1.
- Noah B, et al. Digital Medicine. 2018;doi:10.1038_s41746-017-0002-4.
- Richards R, et al. Gait Posture. 2018;doi:10.1016/j.gaitpost.2018.02.023.
- For more information:
- Jeffrey R. Curtis, MD, MS, MPH, can be reached at 2000 6th Ave. S, Birmingham, AL 35233; email: email@example.com.
- William G. Dixon, PhD, can be reached at Oxford Rd. Manchester, M13, UK, 9PL; email: firstname.lastname@example.org.
- Linda Li, PT, PhD, can be reached at Arthritis Research Canada, 5591 No. 3 Road, Richmond, BC, Canada V6X 2C7; email: email@example.com.
- Brennan Spiegel, MD, MSHS, can be reached at 11301 Wilshire Blvd., Los Angeles, CA 90073; email: firstname.lastname@example.org.
- Daniel Kenta White, PT, ScD, MSc, can be reached at 540 S. College Ave., Suite 210L, Newark, DE 19713; email: email@example.com.
Disclosures: Curtis reports research grants and consulting fees from AbbVie, Amgen, BMS, Corrona, Janssen, Lilly, Myriad, Pfizer, Roche and UCB. Dixon reports receiving consultancy fees from Google. Li, Spiegel and White report no relevant financial disclosures.