The Patient Equation

by Glen de Vries

  Clearly if the future is going to look like this, collaboration is critical. Incentives have to be properly aligned, new payment models have to be put into place, every piece of the system has to be restructured in such a way that we can reach the full potential of the data. That is what the final section of the book will look at: how can we move from each of us creating and deploying our own patient equations to a world that is effectively powered by them, providing patients with everything they need to optimize their health and providing us with the business incentives to keep moving in the right direction, working together, and improving lives? We'll start with collaboration in the next chapter, and then move to payment models and incentive alignment in the chapters that follow before we reach the conclusion of the patient equation story.

  Notes

  1. Jignesh Padhiyar, “Best Health Apps for IPhone in 2019 You Shouldn't Miss Out,” Igeeksblog.com, January 17, 2019, https://www.igeeksblog.com/best-iphone-health-apps/.

  2. Mercatus Center, “Atul Gawande on Priorities, Big and Small (Ep. 26),” Medium (Conversations with Tyler, July 19, 2017), https://medium.com/conversations-with-tyler/atul-gawande-checklist-books-tyler-cowen-d8268b8dfe53.

  3. Beth Snyder Bulik, “Payers Say They'll Cover Pharma's beyond‐the‐Pill Offerings. They Just Want Proof First,” FiercePharma, August 24, 2016, http://www.fiercepharma.com/marketing/payers-want-more-info-pharma-and-healthcare-digital-health-technologies-according-to.

  4. Jing Zhao, Becky Freeman, and Mu Li, “Can Mobile Phone Apps Influence People's Health Behavior Change? An Evidence Review,” Journal of Medical Internet Research 18, no. 11 (November 2, 2016): e287, https://doi.org/10.2196/jmir.5692.

  5. Jeff Lagasse, “Groove Health Gets $1.6M for Analytics Platform Focused on Medication Adherence,” MobiHealthNews, August 8, 2017, https://www.mobihealthnews.com/content/groove-health-gets-16m-analytics-platform-focused-medication-adherence.

  6. “Maxwell: AI‐Powered Patient Engagement,” Groove Health, 2019, https://groovehealthrx.com/AI.

  7. Stephanie Baum, “Janssen Develops Mobile Clinical Trials Platform to Reduce Drug Development Costs, Improve Adherence,” MedCity News, October 14, 2017, https://medcitynews.com/2017/10/janssen-develops-mobile-clinical-trials-platform/.

  8. “A New Sort of Health App Can Do the Job of Drugs,” The Economist, February 2018, https://www.economist.com/business/2018/02/01/a-new-sort-of-health-app-can-do-the-job-of-drugs.

  9. “Scaling Impactful Digital Heath,” Welldoc, Inc., July 8, 2019, https://www.welldoc.com/outcomes/clinical-outcomes/.

  10. Simon Makin, “The Emerging World of Digital Therapeutics,” Nature 573, no. 7775 (September 25, 2019): S106–S109, https://doi.org/10.1038/d41586-019-02873-1.

  11. Ibid.

  12. Stan Kachnowski, interview for The Patient Equation, interview by Glen de Vries and Jeremy Blachman, February 10, 2017.

  13. Josh Baxt, “To Elevate Diagnostics, the Unloved Stepchild of Precision Medicine, Educate, Educate and Educate,” MedCity News, August 29, 2017, https://medcitynews.com/2017/08/teaching-payers-pharma-physicians-patients-investors-diagnostics/.

  SECTION 4

  Scaling Progress to the World

  13

  The Importance of Collaboration

  You can't do it alone. For us to scale progress to the world, companies in health care simply can't remain in their individual silos. Data needs to be combined to realize its full potential, which means players across the spectrum must work together. Finding, developing, testing, and marketing the right treatments for patients requires a coordinated effort. Companies must share data, talk to each other, and collaborate in order to thrive.

  Take an eye drop developer, for instance. In the past, that company needed only to be concerned about eye drops, to some extent at least. Now, that same company ought to be sharing data with ophthalmologists and endocrinologists, with chronic care diabetes drug manufacturers, with players across the spectrum. Easier said than done, of course. We talked already about good data versus bad data—one principle of having good data is that it's in a form that's able to be shared in the first place. Standardizing data is obviously of key importance for collaboration, because we can't join forces if our information can't interconnect. But data standardization isn't enough. We need the willingness to share—and not just the willingness, but the realization that it is an imperative, and that cooperation will lift us all to greater heights in our quest to help patients and to create solutions that drive our organizations forward.

  We need each other's data in order to enrich our models, to understand how our products and solutions are impacting patients in the world, and to fill in the gaps in our own views of patients as individuals, within populations for the modeling of disease, and across researchers who are working to manage and cure them.

  Cancer treatment, for example, has been compared to a game of whack‐a‐mole. As soon as you bop one enemy on the head—you cut off a pathway gone awry, making cells in a tumor proliferate without control—another pathway will break down or compensate somewhere else. Working together, across silos, is just like a cocktail of medications, meant to bop all of the moles on the head at the same time.

  William Carson, the recently retired CEO of Otsuka Pharmaceutical and chairman of the company's board, gave me a wonderful analogy for collaboration.1 Imagine that each pharmaceutical company working to cure a particular cancer is one person in a room. There is one door. If everyone tries to get out the door at the same time, not only will it take a long time, but it will be a congested, chaotic mess. If the people line up before they walk out the door—if the pharma companies in the metaphor work together to figure out which therapy should be first in line, second in line, and so forth—then everyone gets through the door faster. The patients we are all working to serve get their treatments—their cures—as effectively and quickly as possible.

  This isn't just a pharma problem, though. We are all collecting data across the continuum of care, and its fullest power is unleashed when we fit it all together.

  A Tiny Island or a Larger Ecosystem

  It doesn't take much looking back to remember a time when people signed up for email‐only web services like Geocities and Hotmail. But almost as soon as it launched, Gmail became king—because it was integrated with the entire suite of Google products, and enabled data to be shared across a whole range of applications. Similarly, Jawbone fitness trackers didn't integrate their data into the giants like Apple and Google—and after being valued at as much as $1.5 billion in 2012, by 2017 they were liquidating their assets.2 You can try to be a tiny island, with one crop, cut off from the larger mainland—but evidence shows that it is ultimately almost impossible to survive in the tech space without being part of a bigger ecosystem, and without integrating with large, platform players.

  Apple has built a tremendous ecosystem around its devices, particularly the iPhone and Apple Watch. Through its ResearchKit, developers can create powerful research apps, and get them in front of possible participants through the App Store. They have even entered into agreements with large‐scale health care players (starting with Aetna, to look at instrumenting their health insurance customers).3 A joint Aetna‐Apple app known as Attain offers health recommendations based on Aetna records, including vaccinations, medication refills, and information about lower‐cost lab test options. It is a small step in the direction of digital medicine, but shows the potential.

  The Economist has written about Apple, Amazon, Facebook, and Google's Verily diving into health care, trying to be the connective tissue that can bring everything together.4 Apple is exploring sensors that measure stress and blood oxygenation, and trying to find ways to measure blood glucose with the Apple Watch. Verily is building surgical robots, and in 2017 launched Project Baseline, an effort to collect comprehensive health data to be deployed in exactly the ways we've discussed in this book. “Change doesn't happen in a silo,” says the Project Baseline website.5 “Through Project Baseline, Ve
rily is building a connected ecosystem, engaging partners across healthcare, life sciences, and technology.” Facebook is using artificial intelligence to monitor posts and find users who might be depressed, so that it can potentially intervene.6 These big tech players will likely do whatever they can to find inroads into health care and do what they can to make themselves critical parts of the future, however it unfolds.

  And they ought to. The future that we're envisioning—at least the one I want to be a patient in—requires connective tissue that acquires and transports data, and that can support businesses and market infrastructure as well as apps and data flows. These digital infrastructure giants are the organizations that can make sure the right incentives are cleverly and seamlessly integrated into patients' lives in order to make our patient equations work. They are the ones who can develop activity tracker batteries that last for a year and don't need to be recharged.

  It's hard to get useful wearable data when there is any responsibility to manage infrastructure on the patient's end—when devices need to be taken off, or batteries recharged, for example. That's why ingestibles and implantables are so exciting. Of course, even they require connectivity, a digital fabric that brings all that data together. Between our phones, glasses, rings (and perhaps even cars), these companies are—or at the very least are among—the global players who will provide that network of connectivity.

  We talked in Chapter 11 about adaptive clinical trials—but the possibilities don't stop just with individual organizations. Institutions can learn even more by collaborating with their peers—especially as medicine gets more and more precise. The traditional clinical trial model was based in part on the idea that any medical center would have exposure to lots of different patients, enough to populate a trial for most reasonably common conditions. But in the new model, we are looking to develop not just drugs for a wide swath of the population, but for very specific subsets—highly‐specific cancers, for instance.

  There may be only a very small group of patients nationwide—let alone at one particular institution. Siloed data may not slow you down if you are developing a statin, but as the inclusion and exclusion criteria lists get longer and longer, and trials get more and more specific, it becomes likely that even the largest and most prominent academic medical centers might see only a few relevant patients per year—which is neither cost‐effective nor time‐effective from any research perspective.

  In Don Berry's I‐SPY 2 trial, companies are working together in order to increase the chance that the study can find drugs that work in a more cost‐effective, speedy way. Every individual life science company doesn't need to enroll control patients, and doesn't need to organize a giant new trial for one drug when instead they can all be combined under one organization. (It is also better from an ethical perspective, because fewer patients will be exposed to a placebo and the chance for great care will be maximized.) Using the I‐SPY 2 model, having a site with just one patient ends up being workable—especially as we move more of the trial to the digital realm—because costs are more limited. And, as we explored, the adaptive trials of today can lead to broader industry collaborations as we as an industry—not just one medical center, company, or country—climb the data value chain toward more and more valuable and actionable insights.

  How Data Collaboration Can Change the Game

  All projects in life—not just in life sciences—have limits on resources: money, people, and time. We inevitably need to think about timelines and budgets at the same time as we think about rigorous science. But data collaboration can help us accelerate value and mitigate risk by centralizing and concentrating certain core research needs. Companies can get drugs approved and go to market more quickly, if they don't each design their clinical trial data systems and instead use one from a cloud software vendor.

  This is an idea proven to be effective thanks to not just my company, but to the electronic trial industry that has attracted players as diverse as startups, database giants, and companies ostensibly in salesforce automation. We can confirm that there is a valuable drug response using synthetic control arms, and we can use integrated data across molecular to global scales to see which patients have the best chance of responding in order to seek them out, or the highest risks of adverse reactions in order to exclude them from trials. We can deliver the most valuable therapies to patients throughout the therapeutic life cycle, from the first time it is used in humans, to the moment the drug has been approved, to the last time it is prescribed before being replaced by a “better tool.” (And using the platforms we just discussed in Chapter 12, we can account for the drug's value in all of those patients' lives more easily and accurately.)

  The use of synthetic controls becomes a self‐perpetuating business model if all data is pooled, as the data for new drugs today becomes tomorrow's control data. As we discussed in Chapter 11, if patients are getting a drug in one arm of a study—today's experimental therapy—those patients can become tomorrow's control patients, as the new standard of care. This applies across drug companies. If company A is testing a new drug for a particular condition, then as soon as that drug is approved, the patients in that trial can become part of the synthetic control group for company B's new trial of their next‐generation drug, attempting to further the state‐of‐the‐art therapy for the same condition. This saves immense amounts of time and money, and creates huge value by continually making the advancement forward less burdensome for patient volunteers in research, and faster to arrive for patients outside of trials.

  You can also aggregate patients across trials and across drug companies into an experimental comparator group. A client can come to a company like Medidata and say it wants to launch a trial for, say, breast cancer with a particular set of markers over a particular period of time. Instead of the drug company needing to find 50 patients for this trial, Medidata can go into its storehouse of patient records and do that work. In the traditional model, of course doctors who are prominent in their fields and who actively participate in research can quickly help patients get access to trials. However, in models built around data sharing, with intermediaries and aggregators doing the work of connecting patients to research projects, identifying which patients will likely be the best responders (and matching them to the right trials) can be scaled beyond the connections and experience of individual investigators. We can build massively scaled adaptive trials, build synthetic control arms, and build up data in the process about people on experimental therapies for a particular disease—which all ends up enriching the next study, and the one after that, and the one after that.

  Every time we get an outcome, we can look at patients through a range of different lenses, help drug developers figure out which biomarkers matter most for their new therapies, which biomarkers matter for another company's therapies, and predict which drugs help different subpopulations the most. And, we can do all of this while constantly comparing outcomes to the current standard of care and to previous trials across life sciences and health care. We can offer the benefits of enrolling as many patients as possible, but at the same time also limit the sample to the patients most likely to be good responders. By targeting the drug more closely to likely responders, its value proposition grows. The evidence for a payer or regulator to look at becomes much stronger, and the preference for a particular treatment by a patient or by a physician (as well as the willingness to approve and reimburse for it) all go up.

  Demonstrating that value proposition more quickly—proving the drug works, even for a smaller subset of patients than originally hoped for as biomarkers are discovered and target populations narrow—still creates huge economic value. Selecting for better outcomes is better for drug companies, for payers, and for patients. Everyone ends up a winner. It's not the traditional model, but it's a disruptive way to take advantage of the promise of data and the hope of collaboration. Too often, I see companies running out of money before they find that value proposition—and then no patients end up bene
fiting. We can do better. Data collaboration can help us do better.

  Put another way, making every patient across massively scaled research a potential participant in a single denominator—not trying to create evidence‐based conclusions from data solely in single research projects, but across them—changes the equation (the business equation) around return on investment in research and development, and absolutely revolutionizes life sciences.

  Alas, good intentions regarding data collaboration aren't enough to make it all happen—and, frankly, good intentions around any of these new data‐driven technologies aren't enough to move forward at the pace at which effective change ought to progress.

  There is a naïve view that cooperation in efforts like data sharing can be sustained entirely through altruism, even if commercial factors don't support it. I think there is substantial evidence to prove otherwise (for example, the lack of progress along these lines by not‐for‐profit industry endeavors like TransCelerate BioPharma), and sufficient reason not to rely on good intentions to be the only fuel that can propel us to these incredible possibilities.

  The data in collaborations is no different than data in a single study—or in plotting the course of a single Mars probe. It must be cleaned, standardized, and made interoperable and fit‐for‐purpose for the analyses that create new comparators like synthetic controls and enable the discovery of new biomarkers and subtypes of diseases. All of that—just like every step in a laboratory, where a new molecule is synthesized and tested to see if it has the potential to become a new medication—takes time and money. Sustainable commercial business models need to be created that make this future state feasible. Incentives for the companies that create this research fabric to power innovation are just as necessary as they are in the world of smartphones and operating systems.