There is a lot of excitement about the potential of wearable technology to transform healthcare delivery and improve patient care. The Consumer Technology Association (CTA) predicts that the U.S. could reach a “critical mass” of physicians using patient-generated data from devices such as wearables by 2020. At Boston-based Brigham and Women's Hospital, clinical leaders are using wearable technology as part of a new “hospital at home” program with the aim of reducing healthcare costs.
David Levine, M.D., is a practicing general internist and research fellow in the Division of General Internal Medicine and Primary Care at Brigham and Women's Hospital and Harvard Medical School. His research focuses on digital health, care redesign, and the quality of delivered healthcare. Among other projects, he leads Brigham's Home Hospital program to bring acute care to the home for patients who would normally be admitted to an inpatient facility. Healthcare Informatics Associate Editor Heather Landi recently spoke with Dr. Levine about Brigham and Women’s Home Hospital program and what he sees as both the opportunities, and the limitations, of using wearable technology as part of clinical care. Below are excerpts from that interview, edited for length.
What is the Home Hospital program and how does the program use wearable technology?
The program is pretty innovative; we essentially create a hospital that goes to your home, so that when you show up at the emergency department, right before you are wheeled upstairs to be admitted, our team is allowed to come in and see if you are interested in getting your acute care at home. Essentially, we bring all the parts of the hospital that you would need to your home—the doctor goes to your home, the nurse goes to your home, meaning IV medicines, imaging, lab tests and, particularly, wearables as well. We use a patch wearable that is a sticker that goes on the patient’s chest, and we are able to monitor continuous heart rate, continuous respiratory rate, single lead telemetry, steps, sleep, falls, and that data gets streamed to our team. And, it’s actually better than if the patient were in the hospital. Most people are not using this technology in the hospital right now. So, we don’t have to wake up our patients in the middle of the night for vitals. We know when our patients are walking, when they are sleeping, and when their basic vital signs are stable.
We had our pilot late last year and it ran for about two months. It was very successful, but it was small, only about 20 patients. We just relaunched in early June for a much larger, randomized control trial. It’s a study, but at the same time, it’s also an active clinical operation. It’s part of my passion to make research very pragmatic. So, while it is a randomized control trial, it is also very much embedded in the hospital’s operations. We take patients of any age, but our median age is older, in the 70s to 80s, and that really represents the fact that patients who get admitted [to the hospital] are mostly older.
David Levine, M.D.
Can you tell me about some results or care outcomes that you have seen from this program?
Our result paper is under review, and will hopefully be published in the next month. I can’t share any of the details, but what I can tell you is that, in general, we saw the same safety and the same quality of care, but we saw large improvements in patients’ physical activity compared to the control groups [receiving care in the hospital]. Patients who went home compared to patients who stayed in the hospital slept more, they walked more, and they were upright more often compared to their compatriots who were in the randomized control arm in the traditional hospital. And then we saw reductions in cost as well, which is our primary outcome—wanting to show that we could reduce the cost of acute care, but maintain quality and safety, while also improving the patient experience. We did see patient experience ratings go up when they were able to stay at home for their acute care.
How are you using the wearables data and incorporating that into patient care?
We use the data in different ways; it’s very much a work in progress. The standard pieces are heart rate, respiratory rate and telemetry. We’ve built alarms that essentially tell us when a variable is outside the range that we’ve selected. Those alarms are then sent directly to the physicians’ and/or nurses’ phones, and they are immediately told that there is an issue. Some hospitals still have what we call telemetry nurses; we don’t have that role, we instead have machine algorithms that are watching all this data. And then, after an alarm is sent, we have the ability to call the patients or to start a video visit immediately to see what’s going on. And, our team also is able to activate and be at the patient’s home within 20 minutes.
Some of the more advanced, research-based work that we’re doing is creating smart alarms, and that’s essentially combining different values at once to make an alarm even smarter. For example, if a patient walks up the stairs, and their heart rate goes up, I’m OK with that. But, if patient is lying in bed and, all the sudden, has a very fast heart rate, that’s something that might be more concerning to me. So, combining something like exertion with heart rate can be a very powerful factor. And with the wearable, we’re able to grab, not just heart rate, but also three-dimensional (3D) accelerometry. Those are more detailed analyses we’re starting to do.
And, then even further down the line, we’re thinking about how we can use machine learning to run some predictive analytics on our data, because we have such a breadth of data on these patients, and we’re probably only scratching the surface as to what we can do with it. For example, when I look at a graph of a patient’s 3D accelerometry, there’s probably some really important information in there. Let’s say the patient has not been moving as much that day; that’s an important variable that we should be capturing and reporting on. And, so we’re looking at if we can start to incorporate all of the data that the patch is giving us, as opposed to the more simplistic, traditional heart rate and respiratory rate data, to really create an even more important, and perhaps predictive, picture of our patients as they either get better, or potentially, get worse.
What are the biggest challenges to incorporating that more advanced data?
Validation is one of the big challenges; the more data the better, as we always say with these predictive algorithms. Essentially, being confident that the algorithms we create are clinically relevant. That’s one of the big burdens right now, in terms of the quantity of data that we need. And, then the question always is, are we measuring and looking at the right thing? We don’t know. So, is the accelerometry data really the holy grail of information? Maybe not. Is respiratory rate variability one of the more key variables that we should be looking at? Just a totally empiric question that can and should be answered, but we haven’t done it yet. And, as we enroll more patients, we eventually want to adjust our experimenting inside of Home Hospital. We’re probably going to be thinking about moving beyond wearables.
Can you give examples of other technologies you might incorporate into the program?
There’s ingestibles, that’s one category. There’s also essentially pieces of equipment that you put in the home that you use different sorts of frequencies of sound, so whether it’s ultra sound or even infrared, or even motion detectors, these have successfully been used to monitor patients. There’s also cameras that use a sensor that can detect changes in skin color to monitor heart rate. Those are some funs areas that we’re going to go into in the near future.
Is the data from the wearables being incorporated into your electronic health record (EHR) system?
That’s a huge challenge, mostly because our current EHR systems do not like outside data and they don’t know what to do with continuous data, which many wearables give us. So, our patch system gives us continuous data; essentially every millisecond we get a heart rate. That’s a first challenge; it’s a type of data source that our EHRs are not set up for. The second challenge is interoperability and getting a proprietary patch to talk to a proprietary EHR, where neither wants to open an HL7 interface or portal to push or pull data. So, that’s a huge challenge; we are not integrated with our EHR because of that. We use a standalone software to view the data, interact with the patch data and then send us alarms.
What are some of your recommendations to other patient care leaders interested in setting up a similar program using wearable devices?
One of the key things for a home hospital program is that your hospital needs to have a capacity issue, because you are taking patients who would normally go to the hospital. Likely, in the current fee-for-service environment, there’s going to be little incentive for a hospital to start a home hospital program. However, if your hospital has a large capacity issue, which most urban hospitals do, or if your hospital is part of an ACO [accountable care organization] or part of a shared risk agreement, then a home hospital program could be a very powerful, clinical program for your ACO patients, because it provides high-quality, safe care at a reduced cost. So, a hospital that’s actually on the line and has shared risk for a patient will be very interested in providing excellent care for that patient, but at a lower cost. You really need to assess the context in terms of capacity issues; and assess the context in terms of payers. If all you do is fee-for-service, I’m not sure that it would make a lot of sense in our current climate.
I want to add one thing. As you may be aware, there have been several negative randomized control trials on post-discharge wearable monitoring, in heart failure as well as COPD patients. So, while I think a lot of these things could make intuitive sense and I’m interested in different projects, I’m also a big skeptic, because the literature would suggest that it doesn’t work, right now, the way that we’re doing it. Home hospital is very different than a simple program where a patient is being discharged from the hospital and we’re going to slap a wearable on them and try to figure out if their getting worse and intervene before they come back, because that, so far, has not been shown to work. It’s not necessarily a question of the technology. It’s a question of whether or not monitoring somebody after they have left the hospital actually prevents them from coming back to the hospital. That is the key question, and, so far, the answer is, probably not.