The Patient Equation

by Glen de Vries

  And then, of course, these innovations must be brought to patients at scale. One of the biggest things standing in the way of that progress is the need for new reimbursement models that truly incentivize the right behaviors and investment in the right initiatives across the landscape of precision medicine. The way we get there is by keeping our eye on the big picture—and tying our compensation to those big‐picture results. It's about reimbursing life sciences companies and health care organizations based not on what we do but on the outcomes of what we do.

  There has been a trend toward value‐based reimbursement, but we are still in the relatively early days, and only a small percentage of what we do is successfully compensated under these new models. As an industry, we need to encourage the move toward value‐based reimbursement models, and as a society we need to understand where we have the technology and data to measure value. To align incentives, we need to move closer to a system of outcome‐based rewards. The next chapter will explore this issue more fully as we drive toward a better understanding of how to move the world toward a patient equation–driven future.

  Notes

  1. Courtesy of William Carson.

  2. Wikipedia Contributors, “Jawbone (Company),” Wikipedia, October 1, 2019, https://en.wikipedia.org/wiki/Jawbone_ (company).

  3. Jonathan Shieber, “Apple Partners with Aetna to Launch Health App Leveraging Apple Watch Data,” TechCrunch, January 29, 2019, https://techcrunch.com/2019/01/29/apple-partners-with-aetna-to-launch-health-app-leveraging-apple-watch-data/.

  4. “Apple and Amazon's Moves in Health Signal a Coming Transformation,” The Economist, February 3, 2018, https://www.economist.com/business/2018/02/03/apple-and-amazons-moves-in-health-signal-a-coming-transformation.

  5. “Project Baseline,” Verily Life Sciences, 2017, https://www.projectbaseline.com.

  6. “Apple and Amazon's Moves in Health Signal a Coming Transformation.”

  14

  Value‐based Reimbursement

  For too long, our reimbursement systems have been driven by what we do, what we make, and what we sell—and not whether a treatment works, or how effective it turns out to be. The truth is—and you probably realize this by this point in the book—our ability to measure value in health care has been limited until recently. We could track gross measures like survival rates across populations, but there's a big difference between surviving in a hospital room, with limited quality of life, and surviving at the baseline level established before an illness hit. We could also track individual patient outcomes in detail, but that data remained trapped within each individual's chart, not connected to or compatible with an integrated view of outcomes for patients like them.

  The future of medical reimbursement is all about value—from a broader perspective than just survival—hinging on the integration of the details of individuals. Incentives for pharmaceutical companies and device manufacturers are changing as we realize we can measure more. We can measure whether something is working for each patient, and how much it is working—and we can move patients from treatments that don't work to treatments that do.

  Moreover, in a value‐based world, the calculations can go far beyond each individual. Data can give us the ability to make convincing arguments for regulators to let our drugs and devices enter new markets, give us new paths to approval, and new ways to value the work we do that isn't just based on how much cost we incur developing our treatments—but based instead on how much improvement these treatments bring to our patients and to the world.

  These new therapies may be more expensive than the broad‐audience drugs of the present, but their costs will be justified based on realized value—not just theoretical improvements to quality of life and longevity. The alignment of incentives across health care delivery and the research and development of better tools for health care will change the business equations. This will be thanks to focusing on patient equations, and on investing and delivering a better future for patients as individuals, and as populations—regardless of whether the population size is in the billions or perhaps just dozens with a particular rare disease.

  Beyond Survival

  Improving survival rates is terrific, there's no doubt. But—looking back at the discussion of the territory algorithm in Chapter 2—is survival the only measure we should value? We have the capability to show the economic impact that survival has on society and on each individual in a quantitative way. We can measure—or at least start to measure—how much quality of life is included in an extra 18 months of survival for a patient, as well as how much increase in GDP those extra 18 months might lead to for a government or society. We could never do this before, but with cognitive, behavioral, and nontraditional physiological inputs, we now can. We can present to regulators not just those traditional survival numbers, but more—to make the case that our drugs and devices are of real value across new dimensions.

  Regulators are conservative. They should be. Their job is not to enthusiastically trust, but to skeptically question the evidence presented to them. Let's turn back to the idea of territory as a proxy for overall health. Maybe it will prove to be a surrogate measure for how much someone is able to work. Maybe we can use it as a quantitative input that helps estimate how much someone is contributing to GDP—a proxy endpoint for socioeconomic engagement and output. If that's true, it can help justify our entering a market and asking for a certain price on our product.

  In a truly value‐based world, if territory turns out to be a quantitatively valuable predictor, it could be used to justify a life sciences company being paid a certain percentage of the increased GDP their product leads to. If my drug causes 10 million people who would have otherwise died to each contribute $10,000 a year to the world, that's $100 billion in value per year. Is that worth $1 billion, $10 billion, or more to the pharma company that generated it? The answer to that question lies in economics and the negotiations of payers and consumers. But the ability to make that case—responsibly, reliably, and objectively—lies with the life sciences industry. And our regulators are the best starting point to make sure the evidence for such a case is sound.

  Take gastric bypass surgery, for instance. It's not just about the success or failure of the procedure. People end up changing their diet, exercising more, and living fuller lives. Does that translate into more productivity at work? They lower their risk of sleep apnea and diabetes, adding years of life—higher‐quality, more productive life. Or at least the patients with good follow‐up care have all of these spillover returns. So if we base payment at least in part on these other measures, we not only incentivize performing a successful procedure, but we also incentivize good follow‐up care—which is what we should want our system to do.

  Another example: physical therapy always matters post‐surgery for musculoskeletal procedures, but a surgeon's compensation isn't affected by the outcomes of the PT. Doctors perform amazing surgeries, and then give someone a Xeroxed copy of a 20‐year‐old sheet of paper explaining what they should do in physical therapy. This doesn't make sense in an outcome‐based world. Whether or not I recover well from surgery depends as much if not more on my PT than on the surgery itself. We can measure mobility and overall activity before and after surgery and compare.

  Financial success ought to mean getting patients as close to their before‐condition numbers as we can—and, even better, financial success ought to mean preventing problems in the first place. If I'm a doctor and I intervene to stop you from ever needing diabetes medication, why shouldn't I make more money than the doctor who ignores you until he has to treat you for the diabetes he didn't help prevent—and why shouldn't that doctor make more money than the doctor who treats you for diabetes, but does a lousy job and your hemoglobin A1c gets worse? Doctors ought to be incentivized to actually care about the follow‐up and the long‐term outcome just as much as they care about making diagnoses and performing procedures. This is what will actually make a difference for patients.

  These e
xamples are simple to write, but hard to implement. The physicians and surgeons here suddenly have their compensation tied to things beyond their control—the patient's discipline in diet and exercise, or a physical therapist's ability to motivate. But it's even bigger than that. This isn't an exercise in changing the incentives and compensation for just one part of the value chain. The manufacturer of an implant, the surgeon, the physical therapist—the entire health care system ultimately responsible for delivering an outcome (in this case, an individual who can move around, enjoy life, and be productive at work) needs to be accountable as a team to deliver that outcome.

  At the front end, that means we incentivize only offering the procedure to the people who will most benefit from it. Perhaps there are other therapeutic choices for a particular patient that would potentially have better outcomes. If we know that a patient is unlikely to follow post‐surgery guidelines, perhaps they need to be incentivized or engaged differently as well. Limited resources—time and money—are unavoidable. So we need to figure out how to best drive outcomes along every step of the value chain, and to make sure we start every patient down what we predict will be the most productive path. This is how we figure out how to give the most effective treatments to the patients who will benefit from them the most.

  John Kuelper, writing for the publication STAT, explains this issue about as well as anyone I've seen: “Today's insurance models are designed around the ‘average’ patient.…If every patient with X type of cancer gets Y drug, we will see better outcomes on average, so we encourage all patients with X type of cancer to get Y drug.”1 But that approach doesn't work for targeted, personalized treatments. “[E]ven if 95 percent of patients would see no improvement with the drug,” Kuelper writes, “identifying the 5 percent who would have a profoundly beneficial response makes it possible to economically justify a price point that looks high on its surface.”2

  The (Mathematical) Fountain of Youth

  For Alzheimer's disease and other neurodegenerative disorders, even if we do a great job as a society of keeping people alive longer, we need to make sure we are also keeping them productive. Again, the incentive has to be something more than just years of survival.

  If we graph quality of life against years alive—see Figure 14.1—we want to maximize the area under the curve.

  Regardless of how “quality of life” is defined on the y‐axis, it is something that every stakeholder along the reimbursement value chain—not to mention the patient—should be able to align on. Note that the graph makes an important distinction between value and time. Quality of life and duration of life are perpendicular dimensions. Quality could mean socioeconomic engagement, or it could be measured by territory, or it could be anything that is valued by the patient or by society—or it could be proxies for any of those values. The integral of that value over time, from birth to death (point A) is the summation of that value over a lifetime. The shaded area is what we seek to maximize.

  Figure 14.1 Quality of life over duration of life

  So what is the most effective way to maximize it? At least in this theoretical illustration—and I think by general consensus—there is a time when quality of life declines, as we get closer to death. Unless unexpected or accidental, the course of most fatal illnesses leads to limited activity, consciousness, and—simply put—lower value for a period of time.

  This is getting to the point of the difference between keeping people alive longer and keeping them productively alive. If we extend the point of death (point B) out from an already declining point in quality of life (QOL) (point C), there is an increase in the total value under the curve. However, that increase is relatively small. If we can instead alter the trajectory of decline in QOL at an earlier point (point D)—change the arc of decline more fundamentally—there is a much greater effect on the total area, even with no or close to no extension of life.

  What is the result of drinking from the mythical fountain of youth? To live forever from point C on this chart? I think I'd pass. But if it results in the extension of the peak of the curve—literally, in this example, the top of the curve—as long as possible, creating a substantial change in the total area, then sign me up on the quest to find it!

  This is exactly how I think about our jobs in life sciences. The example may be hyperbolic, but the result is literally what motivates so many of us in health care.

  There is a case study in rheumatoid arthritis (RA) that illustrates how this theory can meet a practical, patient equation‐based reality. Full credit goes to GlaxoSmithKline's work on the PARADE study (Patient Rheumatoid Arthritis Data from the Real World), which we were lucky enough at Medidata to provide some of the technology to help run (along with Apple via HealthKit). In advanced RA, patients spend a lot of time managing their joints (for example, wrapping them in hot towels—or, rather, having them wrapped for them—before getting out of bed in the morning). Yet RA disease progression has often been quantified with measures in a clinic—like counting the number of swollen joints on a patient's hands.

  PARADE was an endeavor to deploy sensors and an easy‐to‐use app—a virtual trial brought to the patient, instead of bringing the patient into a clinic—to explore what really matters to patients, so that new baselines could be established for what “good” looks like in RA therapy. GSK focused on establishing new ways to look at value and exploring digital means to measure them. Instead of counting swollen joints in the clinic, perhaps we could measure how long it took patients to actually get out of bed in the morning. Perhaps instead of having to put patients physically in front of health care professionals to measure how well they could move their wrist, we could take those measurements with the accelerometers in their phones (which, yes, we were able to do, with reliable, quantitative results from the real world).

  Ultimately, the results of PARADE and studies like it are destined to be used to demonstrate value and create measurements that truly matter to patients. We can then align them to incentives for providers, payers, and pharma.

  Quality of life may look different for everyone. The point is not that the same measure will work for every patient. It certainly won't work for every condition. But we need to figure out what makes a difference so that it's not just about incentivizing survival, but quality survival. If drug A extends life by 18 months spent in a bed, and drug B extends life by just 12 months, but those months can be spent running on a beach, not every patient might make the same choice. Does a patient want to live to see his grandchild graduate from college, no matter the quality of those months or years? Is the goal one final road trip? Is it worth it to trade cheeseburgers and milkshakes for salads and kale juice? We may have the ability to extend our lives by minutes, months, or years, but with what pain and what cost (monetary and otherwise) to those around us and to society at large?

  The curve is probably different for everyone, although the mathematics of decreasing returns inevitably applies to us all. There is no right or wrong answer. The point is that information can be generated from objective data, and reliable predictive models can be created based on what matters to individuals—so that patients and their caregivers can make informed choices. Data can give us all more control over expected outcomes and let people decide based on their own preferences.

  Money‐back Guarantee

  We are seeing the beginnings of value‐based reimbursement take hold in pharma. We talked already about Novartis's Kymriah, where a rebate returns the cost to payers when patients don't respond to therapy. Or at least that's the idea in principle—actual outcomes‐based contracts are more complex, and, for now with Kymriah, limited to just its use in acute lymphoblastic leukemia (ALL) and not its more recent approval for B‐cell lymphomas.3

  Perhaps surprisingly, payers have not been fully supportive of outcomes‐based contracts (OBCs) for drug therapies, at least not yet. Part of the problem is that the contracts as written right now cover only the actual cost of the medication and not all of the ancillary costs associat
ed with treatment—care regimens to get the patient ready to receive the drug, for example—and so it is still a huge expense even if the drug itself is free. Part of the problem is that the current measurement—success at 30 days post‐treatment, in the case of Kymriah—may not actually reflect long‐term success. And part of the problem, at least in the United States, is navigating privacy laws that can get in the way of hospitals sharing treatment results with payers. According to an article in Pharmaceutical Technology, “There does not seem to be a specific push from private payers for treatment centres to enter into OBCs.”4 Again, we need to create aligned incentives across all participants in the delivery of therapeutic value, and systems that can support the objective measurement of that value.

  Nevertheless, Kymriah is not the only example of experimentation with payment models along these lines. The weight loss and diabetes prevention app Noom—an app‐based diet and life coaching tool, effectively a digital therapeutic—offers itself to employers on an outcome‐based model. If a user doesn't lose 5% of his or her body weight while using the app, there is no charge.5 Omada Health, which makes a digital diabetes prevention app, works similarly. According to Vator News, Omada gets paid based on the amount of weight the user loses. “[I]f the user has lost 2 percent of their weight, Omada charges 2x its fee; if they have lost 4 percent, then the company charges 4x, and so on.”6 (The base fee, according to the article, might be something on the order of $10 per user.)