Years ago, when I was a medical student, one of my mentors regularly told us to order a test on a patient only if we knew how the result would change the plan of care. This was his way of teaching us the difference between data and actionable information. I was reminded of this foundational wisdom as I read reports and was interviewed about the new Apple Watch Series 4 (AW4) with fall-detection and electrical heart-rate sensor capabilities. It’s clear AW4 will generate new data. It’s less clear how that data can be harnessed to improve health outcomes in the real world.
Health IT has been “Data-Rich, Information Poor” (DRIP) for decades. It’s essential that data from devices like AW4 be turned into useful, actionable information. So, hoping to gain additional insights on some possible practical applications of AW4, I called up my old friend and colleague Deepak Talreja, M.D., F.A.C.C., a practicing interventional cardiologist and experienced medical informaticist.
There is exciting potential in devices like AW4. They represent a significant step on the journey towards the Internet of Things (IoT) for healthcare. As Talreja noted, “Similar innovations have already reached the marketplace—with step-counting and biometrics of ‘Fitbit’-type wearable devices, CardioMEMS™ for heart failure, and the Alivecor Kardia portable phone peripheral that also offers an optional e-cardiology ‘over-read’ for an additional monthly fee.”
Talreja and I agree that the widespread deployment of a diverse array of devices that connect, communicate and collaborate to promote better health is inevitable. While this vision is full of promise for the long run, it will face many challenges and limitations in the near term. Ultimately the “Iron Triad” of people, process, and technology will determine the success of these efforts so, it’s worth thinking about AW4 from this perspective.
“The competition between consumer-driven wearables and medical-grade devices will test the limits of HIPAA privacy laws, medical-provider and subspecialist access, the willingness of providers to evaluate nontraditional data streams both during scheduled visits and after hours, and medicolegal laws and precedents,” Talreja said. For devices like AW4, he concluded, “One key will be to ferret out what measurables are both useful and interpretable by consumers.”
They will also test the reimbursement system since someone will have to cover the cost of all of this. The devices aren’t free. Providers are rightly reluctant to give their services away. It is unclear at this time who will pay for the device or services. Sure, some patients will pay out of pocket, but that’s not a scalable or sustainable approach.
The Limits of Technology
It’s important to be precise about capabilities when it comes to functions like electrocardiogram (ECG) production and analysis. AW4 is FDA-cleared to identify normal heart rhythms or signs of Atrial Fibrillation (AFib). Essentially, AW4 can tell if you are “fine” or should "get this checked out.” Also, it does not address other important, common and lethal cardiac arrhythmias like Ventricular Fibrillation.
AFib itself can be benign or lethal. Analysis based on the available FDA data suggests AW4 will have a positive predictive value of around 45 percent. This means more than half the time the ECG app flags a problem, it will be wrong. Throw in the expectation that the population of AW4 owners is likely to skew towards younger and healthier and the number of false positives will be even higher.
This raises the real possibility of needlessly scaring a lot of people and wastefully taxing an already overburdened healthcare system. Based on his experience with remote monitoring and consumer devices, Talreja points to the “important implications of increasing utilization of primary care and subspecialty medical services, additional medical testing, and emergency room and urgent care visits by anxious patients who in some cases will be responding to monitoring artifacts.”
Interoperability with the rest of the health IT ecosystem is another big challenge. AW4 “interoperability” appears to be limited to production of a PDF. This will be highly limiting and potentially counter-productive. How will that PDF flow into the system of care in a timely and reliable manner? EHRs (electronic health records) are already overstuffed with PDFs which can be hard to find and are unstructured. Will this PDF be one more “needle” in that “haystack”? And who, exactly, is supposed to read and act upon these reports?
People and Process: Who is Watching the Watch?
It does no good to have devices like AW4 generating data if we don’t know what to do with it or don’t have processes in place to reliably respond. Designing and deploying devices like AW4 is the easy part. Building the systems of care that can act upon the information they supply will be much harder and essential if they are to have a significant impact on health.
And, this is a consumer-facing device so an effective system of care is highly likely to include the patient, their family, other caregivers in addition to traditional healthcare providers. It will not be easy to sort this all out into a reliable, scalable system. AW4’s fall-detection function provides an excellent window into these kinds of people and process challenges.
The AW4 app can detect when the wearer has fallen and send an alert. I’m a member of the sandwich generation. The idea that I could support my elderly mom in her efforts to live independently and safely by having her watch tell me if she has “fallen and can’t get up” is appealing.
But I have concerns about process and making the info actionable. What if I am out of range, busy, asleep or incapacitated in some way? Who is “watching the watch” and able to act if I can’t? Those concerns are magnified if there are issues with sensitivity, specificity or if the device is prone to user error. There are also very basic “people” questions like, “Will Mom consistently wear the watch?”
The Cure for DRIP: A Population Health – AI Mash-up
Talreja believes it’s feasible to build “a reliable, scalable and useful IoT for healthcare out of devices like AW4, but this will ultimately require a combination of population health and advanced analytics.” He’s right. As the number and variety of devices grows, the tidal wave of data they generate will have to be filtered for human consumption and action. If we don’t, we will just make DRIP worse.
For patients, there is the promise of AI-assisted interpretation and decision support that empowers them to better sort out what’s going on and what they should do. For providers, it is the promise of automated monitoring and notification—air traffic control as it were—to help them find and focus on the individual patients with serious problems. The addition of automated protocols for routine care will further reduce the burden on human providers while advanced decision support will assist them in dealing with complex situations even more effectively.
The Best Day or the Worst?
As several others have noted, upon hearing about the AW4, Ethan Weiss, M.D., a University of California, San Francisco cardiologist, tweeted, “I can’t figure out whether today is the best day in the history of Cardiology or the worst.” Talreja and I agree the answer to Dr. Weiss’s question is “Yes.” It will be some of both. Eventually, a well-designed IoT for healthcare comprised of many such devices will be a boon to better heath at lower cost. However, the combination of limited functionality, weak interoperability, low positive predictive value and lack of well-defined care process points to a rocky start and limited benefits at first.
Achieving meaningful success will be a challenge with many ups and downs and frequent reminders that turning data into actionable information requires serious attention to people, process and technology. Otherwise, it’s just more DRIP, DRIP, DRIP.
Dave Levin, M.D., has been a physician executive and entrepreneur for more than 30 years. He is a former Chief Medical Information Officer for the Cleveland Clinic and serves in a variety of leadership and advisory roles for health IT companies, health systems and investors. You can follow him @DaveLevinMD or email DaveLevinMD@gmail.com.
Deepak R. Talreja, M.D., F.A.C.C. is a practicing interventional cardiologist and medical informaticist. You can follow email him at firstname.lastname@example.org.