Discussion: (0 comments)
There are no comments available.
View related content: Health Care
The year in digital health has whiplashed between a broader appreciation of the promise of digital health and an increased recognition of just how difficult it is, meaningfully and measurably, to improve health and drive revenue. My selections for this year’s digital health awards capture important aspects of the evolving landscape, and also highlight where future opportunities for innovation might lie.
Digital Health Company of the Year: Epic Systems
Epic is the EMR company that everyone loves to hate, and that has clearly emerged as the go-to EMR platform for large hospital systems – a great segment to dominate, given the progressive consolidation of physician practices, as discussed recently in the NYT. (Tory Wolff and I discussed Epic at length this year, see Part 1 and Part 2.)
To technologists in Silicon Valley, Epic epitomizes everything that’s wrong with modern medicine – it’s inelegant and clunky, it delivers a user experience many physicians view as awful, and worst of all, it’s a closed, highly proprietary system. I imagine struggling EMR companies must regard them the way SNL’s Jon Lovitz’s Dukakis looked at Dana Carvey’s Bush and said “I can’t believe I’m losing to this guy!”
Yet Epic seems to be growing ever-more dominant, while the rest of the field remains highly fragmented, and it’s important to understand why.
Quite simply, Epic figured out that what their customers needed – in this case, a way to connect the various parts of their system, so that established physician groups and hospital functions can reliably communicate using a common platform, and share patient data.
The appeal of Epic isn’t that its solution is especially elegant, but simply that it works. Epic has a reputation for “flawless implementation” (a phrase I’ve heard multiple times), and while that’s surely an exaggeration, Epic clearly places significant emphasis on the nuts-and-bolts of the installation, working intensively with each individual client, in a highly customized fashion, holding the client’s hand throughout the process.
The key lesson, in my mind, is the importance of understanding Dave McClure’s brilliant maxim, “your solution is not my problem.” Hospitals systems want a solution that reliably solves their most immediate, most pressing problem: basic connectivity – and this is what Epic effectively delivers.
There are a slew of remaining problems that Epic doesn’t effectively address (or as I like to call them, opportunities), functionalities that sit around the EMR, from improved input of data to more sophisticated analysis of the data that’s already been acquired, to the improved ability of care providers to have the relevant information at their fingertips, clearly and accurately presented, whether the provider is at the bedside, the clinic, or at home in the barcalounger.
Going forward, a critically important question is the degree to which such innovation can be developed outside of Epic, which obviously would like to offer these functionalities itself. I am inspired by Leigh Drogen’s “Law of Unbundling” (h/t to Bijan Salehizadeh for putting me onto this), which basically asserts that at least on the internet, companies that try to win by bundling everything (examples cited include AOL, Craigslist, Facebook) get disrupted as customers realize they have other options, and can get better results from a series of other sources.
The fascinating question is whether Epic will be able to exclude all competitors while improving its own adjacent offerings or whether competitor products will prove so compelling to customers that Epic will be forced to at a minimum accept them, perhaps embrace them (e.g. facilitate development of ecosystem), or if necessary, acquire them.
Digital Health Person Of The Year: Vinod Khosla
Starting with his notorious “Do we need doctors or algorithms” post in TechCrunch in January, continuing with his much-discussed interview this summer at the Rock Health Digital Health Summit, and culminating in his just-posted white paper, “‘20% doctor included’: speculations & musings of a technology optimist,” Vinod Khosla has brought his boldly disruptive spirit – and real investment capital – to digital health. (See here and here for my discussion of the controversy.)
His argument, in essence, is that data and computers should enable significantly better health decisions; he writes,
“Computers are much better than people at organizing and recalling information. They have larger and less corruptible memories, remembering more complex information much more quickly and completely, and make far fewer mistakes than a hot shot MD from Harvard. Contrary to popular opinion, they’re also better at integrating and balancing consideration of patient symptoms, history, demeanor, environmental factors, and population management guidelines than the average physician (emphasis added).”
In Khosla’s future, computers will eventually do almost all the work, with providers relegated perhaps to facilitating the information exchange between patient and computer – sort of like the way an attendant at PepBoys will hook up a computer to your car to figure out why the “check engine” light is on.
I continue to believe Khosla fundamentally underestimates and undervalues the degree to which the practice of medicine is (or at its best, can be) fundamentally more humanistic, relationship-oriented than he recognizes, and also far less transactional, the data and distinctions far less discrete. Consequently, I am intrigued by opportunities that use technology to enhance, not effectively replace, the provider/patient relationship (approaches that most easily lend themselves to either an ACO/capitated environment or concierge care, and work less well in context of traditional fee-for-service).
At the same time, Khosla is absolutely right about the extent to which existing data are profoundly underutilized – and about what an opportunity fixing this would present, from both a health and economic perspective. This is arguably where there’s the single greatest opportunity for value capture in the digital health space.
While predictive analytics may change the way we think about risk and disease in the future, I suspect there’s a tremendous amount of value to be captured from basic descriptive analytics – from “simply” (as if) putting together a coherent, accurate, comprehensive timeline of a patient’s health experiences and expenses over the years, and comparing it to some basic standards.
Steward’s de la Torre seems to share this perspective as well, telling Fortune,
“From an IT perspective, health care is at least 15 to 20 years behind the rest of the world. It’s amazing to me that Wal-Mart can tell you where one T-shirt is in one of its factories or warehouses at any point in time, but we don’t even know if a patient had an MRI or a CAT scan, a test that costs $1,000-plus, in the past few months. That’s how far behind our IT systems are.”
While I may not share de la Torre’s apparent enthusiasm for the strict institution of so-called “best practices” (see this useful, cautionary discussion by Groopman and Hartzband ) I certainly believe in avoiding “worst practices,” and making sure providers don’t do things that are manifestly wrong (whether by accident or by ignorance). As Ashish Jha has argued, there’s a lot of good we can do – and money we can save – simply by reducing medical errors.
Furthermore, while it’s tremendously exciting to contemplate the use of sensors, activity monitors, and other approaches to provide rich new phenotypic information (and I love the science here), it’s critically important to
develop a clearer sense of the information we already have, and to better understand what care actually looks like today.
If I were a harried hospital executive facing the (false) choice between (Option 1) the ability to layer into my system new, elaborate cutting-edge sensor data or (Option 2) improved visibility into basic data that exist (or should exist) today, I’d pick Option Two in a heartbeat – it’s hard to move forward and improve processes intelligently without a clear understanding of where you’re starting from.
As an inquisitive physician passionate about advancing standards rather than simply maintaining them, however, I’d look eagerly to the dense phenotyping possibilities enabled by Option 1. (This is also exactly what we’re pursuing in the MGH/MIT CATCH digital health initiative [Disclosure: I am a SF-based co-founder]).
Digital Health Book Of The Year: “Why Nobody Believes The Numbers,” by Al Lewis
Disease management and wellness programs (two related but distinct category of offerings) typically describe both health goals and financial goals. The inconvenient truth, as Al Lewis, a leading figure of the disease management movement, explains, is that far too often, we use faulty techniques to assess the performance of these programs, resulting in essentially cooked data that, well, nobody believes. As Fielding Mellish might say with only slight exaggeration, it’s a travesty of a mockery of a sham. (See here for my recent discussion of corporate wellness programs.)
Lewis isn’t suggesting that disease management can’t reduce costs – in fact, he argues explicitly that it can – only the reduction takes much longer, is far less dramatic, and is appreciably more difficult to achieve (on those occasions when cost savings are actually achieved) than most vendors typically acknowledge.
This is an important book (albeit oddly overpriced, and also poorly rendered on the Kindle, so avoid the digital version) because it explicitly addresses one of the most significant hurdles in healthcare in general, and digital health in particular: credibly demonstrating that a particular intervention delivers return on investment.
I am reminded of a question Stanford professor Arnold Milstein asked me last year while we were walking around campus. “Can you think of a healthcare technology that has reduced net costs to the system?” he asked me. He acknowledged vaccines would qualify, but beyond that, it’s pretty tough going – and it strikes me this is a challenge faced by all new offerings in healthcare, including disease management and wellness programs.
One response to this challenge is to point out that what really matters is the value over time, something especially relevant to the consideration of new technologies. As Gottlieb and Makower recently discussed, and as Clay Christensen has famously argued, disruptive technologies often start off as extremely expensive, but over time, can become dramatically less so, thus evaluating a new technology by original cost can be both unfair and misleading.
Nevertheless, it’s today’s cost that matters most to potential purchasers, which leads to the phenomenon Lewis describes of using faulty metrics to artificially demonstrate benefit. For example, comparing the performance of program participants to a group who didn’t op-in can lead to “success” that seems attributable to the program but more likely reflects the differential motivation of the starting populations.
While the fragility of ROI claims for disease management and wellness programs likely comes as no great shock to potential purchasers of programs (as Lewis’s title suggest), it’s important that entrepreneurs in this space appreciate the magnitude of the challenge as well, and understand what they’re getting into. There are far too many South Park underpants gnome business plans around (Step 1: Offer behavior app, invoking BJ Fogg; Step 2; Step 3: Profit), and comparatively few robust demonstrations of value delivery here.
Fact is, it’s inherently quite difficult to come up with an innovation that improves health in a significant, measurable, and durable fashion, and it’s extraordinarily difficult to improve health to the degree that you actually reduce costs.
Without question, robustly modifying behavior remains one of the greatest opportunities for digital health; but achieving this to the extent that you actually reduce healthcare costs represents a far more difficult task than most entrepreneurs initially recognize
There are no comments available.
1150 17th Street, N.W. Washington, D.C. 20036
© 2016 American Enterprise Institute for Public Policy Research