The Patient Equation

by Glen de Vries

  An article recently appeared in The Atlantic about Bluebird Bio, a company developing a gene therapy for thalassemia, a blood disorder that leads to low hemoglobin, affecting approximately 100,000 babies born each year.7 Bluebird's therapy will cost approximately $2 million, split into five yearly installments—but if the patient doesn't improve, only the first installment will need to be paid by the patient's insurer.8

  The problem, Peter Bach of Memorial Sloan Kettering Cancer Center tells The Atlantic, is that value‐based agreements make it easier for companies to charge exorbitantly high prices in the first place. “When Bluebird says, ‘We've got a $2 million therapy,’ they just made that up,” Bach told The Atlantic. “Markets should set prices—not financing mechanisms.”9

  Making Value‐based Care the Future

  Peter Bach's concern is legitimate—but it just means that we need better pricing standards, not that outcomes‐based reimbursement is the wrong way to go. The value—and the efficacy—of the therapy may be up for debate, but the model is a step in the right direction for value‐based care. The challenge is that we simply aren't there yet as far as the data—we don't yet agree on what we should be pegging reimbursement against and how to move an entire legacy industry toward a new financial model.

  Therefore, my advice to companies that come to me right now, in the course of developing the next great pharmaceutical, is to think about socioeconomic value in their therapeutics as quickly as possible while developing them. Put activity trackers on patients, leverage ideas like territory, record a breadth of digital data on everyone in their trials, whether or not they think it matters today. They will collect an interesting biospecimen that could enable the creation of a richer disease model to help subset patients, even if it is not part of the evidence used for demonstrating why the price point for their medication makes sense.

  They might also expose inconsistencies not just in the patients themselves, but in their environments. We don't know the formulas yet, but I am convinced there are ways that measurements like territory or other high‐resolution behavioral markers will contribute to models of disease progression. How someone is moving around the planet or interacting on social media might—and I believe should and will—contribute to our ability to detect whether or not they are responding to treatment, or whether (in oncology cases particularly) they need a new therapy. There will be other behavioral or cognitive biomarkers we've yet to even think of that will be additive to predictions. The information is out there, and we have the ability to collect it, even the obligation to collect it, and use it to help us create better disease models—because that is the road to better outcomes.

  We talked about my Alzheimer's graph way back in Chapter 2, the cone of possibilities for my future. If a doctor can change my trajectory, move me from someone who appears as if he will need treatment in the future to someone who likely won't, then in a perfect outcomes‐based world, that doctor should be compensated. There is a modern statistic in baseball called win probability added (WPA), the change in a team's win expectancy for a particular game based on the actions of each player. If it's the bottom of the ninth inning with two outs and your team is losing by a run, the chance of you winning that game is just 4%.10 But if you hit a home run, the odds of victory jump to 56%. Your action is responsible for that rise from 4% to 56%, and in a world of outcomes‐based reimbursement in baseball, you would be paid whatever that half a win might be worth. We can't be quite so precise in medicine (at least not yet), but prescribing you a statin may increase your odds of surviving another decade by a few percentage points. The doctor who makes that decision should get paid more than the one who doesn't.

  That's at the individual level. At the population level, in most countries, regulatory approval of a drug just means that it is safe enough from a medical perspective. But we can also consider economic approval. Is the drug safe and effective enough from the perspective of economic returns? If we can use our new streams of knowledge to get patients off drugs that aren't working and onto new ones faster than before, there's economic value if the patients are quicker to contribute to society again. We don't want to give a drug to eight billion people. We want to give it to the ones for whom it will make a difference, where it will meaningfully affect their odds in a positive way. The endgame is being able to demonstrate this, and that's why the new tools are so critical. Giving the wrong treatment costs money. Using the wrong app, one that doesn't produce the actionable advice it claims to, also costs money. We want to spend our limited pool of money on effective therapies, not ineffective ones. We need our breakthroughs to reach as many people as possible—and wasting money on cases where they won't be of benefit doesn't help anyone. This is a case of money as a limited resource—but it is really an argument about opportunity cost. We could have used those resources on efforts that drive outcomes. Why are these aligned incentives and value‐based systems so important? Because they reduce waste.

  I spoke to a class of Columbia Business School students getting a certificate in digital health, and many of the questions were about value‐based contracts. To those students, I explained that you need to worry about three things when considering a value‐based approach that we just don't pay enough attention to today.

  First, consider who should take your drug. “Take it if you have this mutation, but not that other one. Not if you have a previous history of heart disease. Not if you're under 18.” These are natural ways life scientists think about medication, inclusion and exclusion criteria in clinical trials, and drug labels that are approved when we make medications available to the public. We can get even more specific as we refine our patient equations. Take it if you're at least this active in a typical day. Take it if you can also avoid these three foods. Take it if you combine it with these exercises, these other medications, and fall in a particular range based on this wearable sensor. Maybe it's a binary yes/no decision that we end up producing, or maybe it's a weighted score of how well we expect you to respond. But in any case, there is a patient equation that can predict who will benefit, and we should continue to refine it to figure out who the target patient population is, as precisely as possible.

  Second, how do we prove the drug is working? This is a newer idea in life sciences, particularly in the commercial part of a drug's life cycle—but even in the research and development phase, from the perspective of not just the entire population, but individual patients and what really matters to them. What aspects of physiology, cognition, and behavior can we measure that track with effectiveness and safety? How do we measure them, and what can we do to enable that measurement—or find proxies for it—at scale? This is the argument for tracking behavior in all clinical trials. If we're not keeping these measurements in mind while evaluating our drug physiologically, we've missed a huge opportunity to relate physiology to behavior in our modeling of disease as it responds to a new therapy. Figuring out how you engage with a patient in that process goes right along with the measurements. Are there things you need to do (think back to our disease management platforms in Chapter 12) to get the best outcome with your drug in each individual?

  Third, make sure you know when you should stop the patient from using your therapy. This third consideration in value‐based environments is antithetical to how most life sciences companies have traditionally operated. Yet it is just as important as the first two—if not the most important in eliminating waste in a value‐based system. What will be the signals that come out of the data you are collecting that indicate a patient isn't benefiting from the drug? How will you feed that information back into the criteria for who should take it? What will you do to help everyone across the value chain—including the patient—realize as quickly as possible that the therapy is failing, and how can you facilitate the transition to an alternative (even competitive—see, I told you it would be antithetical!) therapy? Every time a patient takes a pill after the point where we could have known it's not working is wasting the patient's time, decreasing
the area under their QOL value curve, and degrading your margin in a value‐based world.

  Managing all three of these concepts is a difficult task for the life sciences industry. But if it was easy, everyone would do it. The companies that embrace this thinking, understand how patient equations are the key to getting there, and manage to execute will be the winners—both financially and in terms of bringing the most value to patients.

  Value‐based care absolutely requires something different from fee‐for‐service. It's a huge change to the traditional business model, and requires new ways of thinking. We need to be able to predict outcomes and trajectories. And we also need to effectively measure the outputs that we decide matter. But it's not science fiction. We are on a path to value‐based care.

  But this isn't enough. We can change the way payers are reimbursed and how drugs are valued, but that doesn't get us to systemic change in every aspect of the health care system. We haven't spent much time in the book talking about two huge stakeholders in the future of medicine: doctors and patients. Doctors need to come on board the data revolution in order to propagate these technologies to the world, and patients need to understand how all of this is set to make radical improvements to their health.

  In the final chapter, I talk to internist and cardiologist—and my own physician—Dr. Dan Yadegar, and patient advocate Robin Farmanfarmaian about where doctors and patients fit into a patient equation–powered world, and how we can make sure incentives are aligned for the betterment of individuals and of society.

  Notes

  1. John Kuelper, “Community Providers Will Help Drive the Future of Precision Medicine,” STAT, February 23, 2018, https://www.statnews.com/2018/02/23/precision-medicine-community-providers/.

  2. Ibid.

  3. Manasi Vaidya, “Outcome‐Based Contracts Viable for Kymriah, but US Payers Still Unsure,” Pharmaceutical Technology, July 30, 2018, https://www.pharmaceutical-technology.com/comment/outcome-based-contracts-kymriah/.

  4. Ibid.

  5. “Noom—The Anthology of Bright Spots,” The diaTribe Foundation, 2016, https://anthology.diatribe.org/programs/noom/.

  6. Steven Loeb, “How Does Omada Health Make Money?,” VatorNews, February 3, 2017, https://vator.tv/news/2017-02-03-how-does-omada-health-make-money.

  7. Suzanne Falck, “Thalassemia: Types, Symptoms, and Treatment,” Medical News Today, January 10, 2018, https://www.medicalnewstoday.com/articles/263489.php.

  8. Sarah Elizabeth Richards, “Pharma Should Pay for Drugs That Don't Work,” The Atlantic, April 22, 2019, https://www.theatlantic.com/ideas/archive/2019/04/pharma-should-pay-drugs-dont-work/587104/.

  9. Ibid.

  10. Greg Stoll, “Win Expectancy Finder,” Gregstoll.com, 2018, https://gregstoll.com/∼gregstoll/baseball/stats.html#H.0.9.2.1.

  15

  Aligning Incentives

  To get the most out of a patient equation–powered world, lots of pieces we've discussed already need to come together. Data needs to be integrated across health care in much more complete ways than it is now. Pharmaceutical companies need to embrace the power of sensors and adaptive trial designs, and bring clinical trials into the twenty‐first century. Best practices for disease management apps need to be developed and tested. Companies and organizations across life sciences and health care need to think about how they can use new streams of data to provide better and more actionable insights to patients who use their products and services. Reimbursement models need to move in the direction of value‐based care. And, finally, doctors and patients need to come fully on board.

  It is far too easy to put doctors and patients to the side when we talk about the patient equation‐driven future. After all, they aren't the ones developing the products, launching the clinical trials, deciding on reimbursement models, or looking at the big picture of the life sciences industry. They aren't dealing with FDA regulations, or, in most cases, spending much energy thinking about the business side of medicine beyond their own practices or experiences.

  And yet, doctors and patients will determine the success or failure of the data revolution. If doctors aren't on board with recommending and prescribing smart devices to their patients, diffusion won't be achieved. If doctors aren't at the forefront of data collection and sharing, looking for insights into the useful biomarkers of the future, we won't be able to move our knowledge forward. If patients don't understand how these technologies can help them, and how data can make their lives better, healthier, longer, and more productive, they won't engage, they won't help generate the information we need, and they ultimately won't see the benefits.

  It is critical that doctors and patients are not only engaged and well‐informed, but that their incentives are fully aligned with the rest of the industry. In this chapter, we look at the patient equation future from the perspectives of doctors, and then patients, in order to figure out how we can best effect change across the care continuum, and make the biggest impact on our businesses.

  Human Doctors, Digital Doctors

  It is not hard to see how doctors could get fully on board with the ideas in this book. Doctors unquestionably want rich sets of good data. They want information on their patients just like pilots want maintenance records on the aircraft they fly. Better, more complete data means doctors can work more effectively on more valuable interventions and preventative measures during patient encounters. The information streams that sensors provide can help establish useful baselines and let doctors move from being purely reactive to being able to work proactively and prevent problems down the line. I am not a physician, but it would be hard to convince me that doctors who are not on board with this vision are capable of practicing good medicine in the world of digitization and data sophistication that we live in. It's as simple as that.

  At the same time, some doctors may be worried that these tools are going to automate them out of existence. Who needs a doctor if an app can tell you what illness you have and how to treat it? The answer, of course, goes back to the failures of IBM's Watson in making smart recommendations for cancer treatment: computers aren't always right. Artificial intelligence can do a lot for us, but it can't replace human judgment, human experience, and complex human decision‐making. Garbage‐in/garbage‐out applies to data as well as to assumptions for what a desired outcome might be.

  We talked in the previous chapter about the difference between mere survival and quality of life. When faced with that trade‐off—do you want to live more years in a hospital room, or fewer years outside in the world?—there is not necessarily a right answer or a wrong one. Different patients might have different preferences. A computer can't help us make those decisions; a doctor can. Artificial intelligence can replace rote tasks like measuring blood glucose, injecting us with insulin at the right time and in the right dose—things we can get bored by, lose track of, or make a mistake with. But that doesn't replace the doctor. Instead, it frees the doctor to think, to strategize, and to do the higher‐order work that robots and predictive models can't.

  What we're talking about is expanding the doctor's toolkit. Just like some products—think wearables like Ava—expand the ability of patients to exert control, we have more and more ways for doctors to have increased information, or at least the possibility of it. What do they want their patients to be wearing, tracking, or reporting? What data is going to help fuel better medical decisions and more productive appointments? What should be measured more accurately than what a patient is normally able to report? In what situations is objective data going to potentially change treatment?

  The line of thinking that says technology will inevitably mean a loss of power for doctors is something I strenuously disagree with. The new technologies aren't taking away but giving more—more power to effect change and to identify and deliver better treatments to improve lives. Yes, automation eliminates rote tasks. Robots can replace humans on an assembly line. But the human body is incredibly complex, we live in even more complex changing environmenta
l conditions, and knowledge of disease and health continually evolves. New tools to manage us are constantly being invented. Someone who wants to follow procedures by rote as prescribed by what they learned back in medical school, and not be an active participant in the patient equation‐driven world, should absolutely be worried about being replaced by an algorithm or robot. But the kind of physician I want to be in the care of has nothing to worry about.

  Dr. Daniel Yadegar is a cardiologist in New York City, with an undergraduate degree from Harvard and a medical degree from Cornell. He has worked at some of the city's top institutions. He's also my own personal internist. I chose to interview him for this book because rather than being afraid of the future, Dr. Dan—as I have come to affectionately refer to him—has embraced it. He believes, as we all ought to, that the new technologies aren't setting us up to replace doctors but that they are hugely powerful tools to help doctors manage their patients' health and help them live longer, better lives.

  Dr. Dan wants to use these new tools to be able to diagnose heart disease and catch cancers before clinical symptoms ever manifest. To that end, he's changed the entire way he operates as a doctor. Unlike traditional physicians, Dr. Dan doesn't just want to see his patients once a year to gather some isolated data points—he wants to see a far richer picture than that. He has patients constantly monitoring their blood pressure, tracking sleep data, heart rate variability, stress markers, and more. And he puts all this data together to augment his understanding of the patient and make better and smarter decisions. He sees the practice of medicine on a path toward greater information, a greater set of tools, from drugs and devices alone to drugs and devices supplemented by digital systems that incorporate trackers, mathematical models, and more.