With his previous professional experience ranging from CEO of Massachusetts-based hospital system Partners HealthCare to CEO of global non-governmental organization Partners in Health, to his current roles as a Professor of Psychiatry at Harvard Medical School and Executive Partner at Flare Capital Partners, Gary Gottlieb, M.D., brings a wide-angle perspective on the use of real-world evidence (RWE) in the challenge of “whole patient care.”
Dr. Gottlieb spoke with Carolyn Magill, CEO of Aetion, at Aetion’s first annual summit, "Evidence at the Apex," in a thought-provoking conversation about data, patients, and clinicians.
Here are highlights from their exchange on the promise of RWE to enable clinicians to tailor treatment strategies to individual patients, the impact of real-world data (RWD) accessibility, and the relationship between whole patient care and payment reform.
These responses have been edited for clarity and length.
Q: What does “whole patient care” mean to you?
A: I think it means transforming our system from illness care to health care. Transactions form the basis of illness care, and much of the data we have access to today are those most germane to those transactions—to the condition in front of us—without understanding how the patient fits in among a cohort of similar individuals, or within his or her environment. When we speak today about the social determinants of health, we are just grazing the surface.
In illness care, decisions are made under conditions of uncertainty. Whole patient care reduces that uncertainty by taking into consideration a broader set of data than what one symptom or symptom complex presents. It requires an understanding of all of a person’s challenges and what their or their family’s objectives are, rather than focusing solely on a specific disease state.
I am hopeful that changes in policy, payment structures, and how we envision health care will transform today’s system into one where we manage populations as well as individual patients, and move toward the management of health rather than the transactions of illness care.
Q: How can we capture data that are relevant to whole patient care? And how can we make these data accessible to the clinician?
A: Collecting and curating real-world data thoughtfully and responsibly is extremely important.
First, we must take into account housing, food, exposure to violence, education, and other critical elements, and to do that, we need to have a broader set of data than those that relate to medical care directly. The data have to help us understand the context in which an individual lives.
For example, if we use antiretroviral therapy (ART) alone to treat a person living with human immunodeficiency virus (HIV) who doesn’t have secure housing or enough food, we’re not really treating the whole person, or helping to keep them healthy. Even with the best possible therapies, a person’s life situation may still make him or her vulnerable to other infectious diseases.
Second, I think we need truly legible data that reliably apply to specific patients. If we can embed RWE that relates to data that are consistent with that patient’s presentation, and array it within the availability of an existing electronic health record, RWE can enhance the data we have in narrative form and in other pure, unbiased data. We have to consider the data relative to that person’s circumstance in order to determine which therapies work best for that individual.
Q: Let’s imagine a world in which physicians have access to real-time RWD to make clinical decisions. What if we’re able to look even more holistically at defining an intervention, so that it’s not just a medication? It could be a combination of medications, or a care management program. What impact do we then see on a given patient population?
A: I think it’s inspiring. Today, we perform randomized controlled trials (RCTs) once or a couple of times on relatively narrow populations. Physicians generalize their conclusions, which may not work for everyone.
Let’s say I work with a patient who has multiple comorbidities and who is on numerous medications that have toxic or enhancing effects on the organ systems that I’m targeting. RCT data may not be effective in that circumstance.
Access to RWD will allow clinicians to see the potential for effectiveness more broadly, in a specific patient population under real-life conditions. Those kinds of data will better help determine the outcome of an intervention. But we have to make those data available in a way that is thoughtful, and that fits into the workflow without being overwhelming.
For example, one of the major challenges most medicine faces is phenotypic specificity—which traits cause a greater differentiation for active treatment versus placebo. As a psychiatrist, I have to ask myself, what am I really looking at that I’m describing as “depression”? By definition, it is a set of criteria that have some probabilistic algorithms associated with them, that can be reduced down to create some specificity and to create cohorts of like patients.
But cohorts of real people are not that easily defined. They are still a mass of heterogeneity, even in the populations living with what I would describe as “major depressive disorder”—severely depressed patients who we would differentiate from those with moderate symptoms. If I am going to prescribe a treatment, I want to understand what that treatment will look like in my patient’s symptom complex, but also through a variety of other phenotypic specifiers, such as a digital analysis of their imaging data, a variety of laboratory data, medical comorbidities they might be living with, and tissue data that could tell me something about their genotypic specificity.
The more that I can use RWD to build RWE around the person I am treating, the more that phenotypic specificity starts to diminish as a confounder in my treatment choices.
Q: What is your perspective on innovative payments, especially as we think about novel and high-cost therapies?
A: Innovative payments are critical. I think that RWD and RWE will help us understand how to measure the value of a specific intervention beyond a singular outcome. The challenge is that, because we are focused on singular outcomes rather than all of the ways that a therapy might improve someone’s life, we create barriers when faced with costs and finite resources.
Much of the financial risk needs to be with providers, as long as there is enough of an actuarial understanding of the population so that patient populations can be risk-stratified in ways that the payments and their tie to outcomes make sense.
From the perspective of specialty care, innovative payments could move to bundled payments, wherein a specialist is providing a kind of warranty. That way the specialist is not just responsible for a procedural transaction, and then abandons the care to the rest of the system or to a primary care doctor.
Q: One of the ideas we keep hearing about is that patients should own their data. What do you think about that? How do you feel about the data that academic medical centers collect—and the accessibility that pharmaceutical companies, for example, might have to that data when patients do not?
A: I endorse the perspective that patients should own what is indeed their data. I think we have to figure out how they permit us to use it, and in what form. For our part, we need to give them the best possible understanding of how those data can be used to improve decision-making, to improve their own condition, to improve the science overall, and to improve the way that we think about how we deliver health care.
But it has to be an iterative and continuous process, particularly as data are considerably more disseminatable than they have been in the past. People fear that their data may be used in ways that will cause challenges. For instance, they may not be able to get insurance or won’t have access to other choices in their lives.
Our investments in security are critical in order to gain the public’s trust—to ensure that data ownership resides with the individual, that they control the “toll booths,” and that they understand how their data can be used in a variety of settings. That is going to be expensive and complex. But if we don’t do it, we will lose their trust. And we won’t be able to use better and clearer data as it accumulates.
In academic medical centers, we are slowly but surely getting out of our silos and the paranoia about the use of data. Within departments and sections and labs, there’s a temptation to covet and protect data. Or, within a given system, there may be resistance to comparing outcomes across various areas.
Decision-making in academic medical centers is slow—just as it is in other big, complex organizations—but I’m optimistic about the progress that has been and will be made.