This transcript has been edited for clarity.
Stephen M. Strakowski, MD: Hello, and thank you so much for listening to us today in this conversation. We really hope that you’ll find it interesting and useful. I’m Stephen Strakowski. I’m a professor of psychiatry at Indiana University. I’m just thrilled to be joined today by a couple of true international experts who happen to be friends of mine, so that’s nice.
Dr Andrew Nierenberg is a professor of psychiatry at Harvard and the director of the Dauten Family Center for Bipolar Treatment Innovation at Massachusetts General Hospital. Dr Peter Margolis is a professor of pediatrics at the University of Cincinnati and Cincinnati Children’s Hospital, one of the highest rated in the world.
Welcome, gentlemen. I look forward to our conversation today. We’re going to talk about learning health systems, which is something that I don’t believe has ever been discussed within the context of psychiatry on this platform. I’m looking forward to it. Welcome.
Andrew A. Nierenberg, MD: Great. Thanks so much, Steve. I’m very happy to be here with my good friend and colleague, Peter.
Peter Margolis, MD: Thank you to you both.
Strakowski: Recently, there was a publication that you all were responsible for that came out in Bipolar Disorders called “A Bipolar Learning Health Network: An Innovation Whose Time Has Come.” For transparency, I’m on the paper too, but what led the two of you to put this paper together? What drove it? What was happening in the field?
Nierenberg: We had an opportunity to think carefully about how to get better outcomes, and we had several things that came together that allowed it to happen, including a generous donation from the Dauten family to seed this particular endeavor. In our initial conversations, Peter and I were saying that there’s a tremendous opportunity to try to get better outcomes for bipolar disorder, specifically, and maybe for psychiatric disorders in general.
Strakowski: Are outcomes a problem in bipolar disorder? Tell me about that.
Nierenberg: I think one of the things that’s most remarkable to me in the journey that we’ve had is that most systems don’t even know what their outcomes are. If you look deeply in the literature and try to see if there are trends over time in the actual outcomes, those trends would be things like quality measures about the use of lithium, about the frequency of hospitalization, emergency room visits, and so forth. Then you try to see what the trends are in terms of the proportion of people who are doing well or not doing well.
It doesn’t seem to be much of a change, which is extraordinary. Again, in my journey here, getting into the messy world of quality, of the definitions of quality, and of quality metrics, there’s nothing that’s been done in the field for at least 15 years. We saw this as a great opportunity to improve the outcomes, not only for adults but also for children and adolescents.
Strakowski: Peter, why a learning health system approach? Why would we do that instead of something else?
Margolis: One of the problems that we face is, as Andy said, an inexorable and incredibly lengthy time to get ideas and evidence into practice. What Andy mentioned in psychiatry is widespread across healthcare.
It is said that it takes 17 years to go from trials and new evidence emerging in the literature to their application and practice, and unfortunately, that’s what we’re seeing in mental health care. It’s also what we’re seeing in many other fields.
That gap is coming at a time of tremendous progress where we have new availability of data to allow us to theoretically assess what our current performance is, reflect on that, and then more and more tools to help physicians that change their practices, even in the context of extremely busy practices and problems that we’re facing every day.
These methods of system change, and data are becoming increasingly available, and there’s an opportunity now to take advantage of them in what’s called a learning health system. The idea is that data and observations from one-on-one interactions with patients should fuel our understanding of how well we’re doing as well as the ability to create new evidence and then put the evidence into practice, so this virtuous cycle, if you will, of doing better than we are today. That’s come to be known as a learning health system.
Even the learning health system concept, as first articulated by the National Academy of Medicine, that took place in 2008. We’re now 20 years past the time that even that idea has been promoted, that we’re now seeing it used more and more in practice.
Strakowski: My memory, if it’s correct, is that one of the major points of all of us being mandated to use electronic medical records by the government was specifically to improve outcomes, but then it did not happen, right? What do you think the barriers have been?
Margolis: There are many barriers. Probably the biggest barrier is the time, including concerns about how much time and how hard it is to actually use these tools in new ways. There are also barriers related to knowledge about how to change systems. There are barriers even within the electronic health record. Although there’s all kinds of information in the electronic health record, often it’s not configured in a way that we can actually get the measures we need.
It takes work, time, and money to actually bring that about. Despite the promise, we haven’t been able to be very nimble at taking advantage of the tools that now exist.
Nierenberg: If I could add, I think there’s also a big cultural gap in terms of how we are trained, what we do day to day as clinicians, and especially in psychiatry, I think we have remnants of the psychoanalytic, one-on-one privacy, nobody else should know what’s going on in the room mentality. This really breaks apart that model and almost brings a peer-to-peer learning, almost coaching, and constantly using the data, but in the context of other people helping you so you can figure out what you could do better.
One of the things I learned from Peter is that, as clinicians, we all want to do the best that we can, and we all want to improve rather than being static and not improving over time. I think it’s unlocking our own motivation to help our patients as much as possible, but at the same time, we don’t generally think of ourselves being in a system.
Part of the challenge is how we can help people function as best as possible, think system wide, help each other, and then have the healthcare systems actually help each other. It’s something that I’ve come to call radical collaboration, not only across the healthcare systems but with patients, with families, with clinicians, with researchers, and with data analysts — all with the single purpose of getting better outcomes.
Strakowski: That’s something I’ve said before. Although there was a national mandate to implement this, I don’t think there was a national strategy to actually optimize it, right? It was really designed for billing purposes, not for outcomes, even though outcomes were the stated goal.
Would it have helped to have some national strategy so that we weren’t all doing one-off designs on our EMRs?
Margolis: I think having leadership to change outcomes is always important. I think too often in the field of quality, the way it’s emerged has been almost like checkbox medicine: Here’s a list of things that you’re supposed to do that aren’t especially relevant necessarily to the patients that you’re seeing.
This approach that we’re working on really starts with what’s most important to patients and clinicians, the outcomes that truly matter, and then backs into the kinds of changes and the kinds of activities that we collectively think are going to make a difference. That may not be things like, did you check for whether somebody’s a smoker as the most important quality metric, but might be other things that are much more relevant to the outcomes of people living with bipolar disorder.
Strakowski: Particularly if they don’t smoke, right?
Why bipolar disorder? Why did you decide to start with that one?
Nierenberg: I think it was probably an opportunistic type of thing where we could do it. The other thing is that, out of all of the disorders, bipolar disorder is sort of an orphan, which is odd, in terms of the research that’s been going towards it and its importance. If you look at it relative to schizophrenia or major depressive disorder, it gets much less funding and much less attention.
The other thing is that, out of all the disorders, it’s the one that’s most associated with comorbid disorders and the complexity of the treatment of it is astounding. If you look at the literature and you look at the guidelines, it seems straightforward. You give somebody a mood stabilizer, or you treat them for acute episodes of mania, hypomania, or depression, and then they do fine over time.
They don’t. Even the psychopharmacology of giving somewhere between two and nine medicines is another level of complexity. Then even another level of complexity is evidence-based psychotherapy, and who’s getting it and who’s not. There are other things that can help people that simply aren’t done.
Strakowski: I know that one of the consequences of this editorial, which was also kind of behind it, is the creation of a bipolar learning health network or the Bipolar Action Network. Tell our readers and listeners a little bit about that. What is it and what’s it trying to accomplish?
Margolis: In our work at Cincinnati Children’s and my work with our team, we’ve taken this idea of a learning health system and recognize that it’s very difficult for a single physician or even for a single hospital to have all the knowledge and all the expertise that’s required to change a complex system.
Our concept is that this is work, as Andy said, that needs to be done together. Doctors, patients, psychotherapists, social workers — everyone has knowledge that can be brought to bear to change very complex, vexing problems.
We’ve used what’s called a network approach, and the concept of a learning health network is to bring together world-class experts in medicine and technology, and world-class experts from people living with the disorder — people whom we’ve come to call “experts by experience.”
This group amounts to a brain trust that’s tapped to develop many answers, to understand the systems and the processes across all the different things that need to happen for patients, and to use data to inform and drive decision-making.
In a learning health network, the community of people who are affected by a condition, together with data and this organizing framework, drives improved outcomes. What we’ve seen as we’ve developed these models in other areas is pretty striking changes in outcomes.
The very first learning health network was focused on kids living with Crohn’s disease and ulcerative colitis. Over the years, we’ve seen an improvement in the percentage of kids in remission from 60% when we started about 15 years ago to about 83% today, and for almost that entire time, which was in the face of no new medications.
The ability to take existing knowledge to share and help each other and put the knowledge into practice has produced more reliable care and better outcomes. We’ve now replicated that approach across multiple conditions. That’s a safety network involving 150 pediatric hospitals that’s cut serious safety events in half.
We also use this in the community here in Cincinnati to work on reducing racial and equity gaps in outcomes. Recently, we’ve seen a narrowing in the gap of admissions to the hospital for kids with asthma from the poorest sections of Cincinnati vs the most well-off sections, and we are reducing Black and White differences.
It’s a very powerful model when you can bring these things together. A focus on outcomes, data-driven decision-making, engaging the community, and system change methods to try to have impact. That’s what we’re designing for bipolar disorder.
Strakowski: Where are we with the network development, Andy? How many sites and how close to launching are we?
Nierenberg: We have around 12 or 13 sites at this point. Each one of them is going through a process where they’re getting their leadership to agree to share their outcomes with other healthcare systems. You can imagine the lawyers involved and everything with that.
We’re getting those done. Once those are done, then the sites can start to figure out what data that they need to gather, and how do we collectively put that together in a way that we can really help each other?
Part of the design work that we’ve done over the past year and a half or so is trying to understand the system itself for each of the places and look overall at what the general system of care is and what the patient’s journey is. You were essential in helping that, Steve, with the model that you had helped develop at Austin State Hospital.
We stole shamelessly from you, with your permission, to be able to do that. It was extraordinarily helpful to understand what a patient would go through from the beginning, when they started to have symptoms, to diagnosis, to when they had crises, to when they were in maintenance, and all of that.
Each system started to understand that better. Then we sat back and said, what are the problems and what are the things that we collectively thought we needed to improve? Beginning at the beginning, we want to see what proportion of people are in remission. What is that? Can we figure out how to measure that? Can we figure out how to compare it? That was the beginning of being able to define a series of important outcomes.
Strakowski: What’s the biggest opportunity here? What’s the best thing that can happen from something like this?
Nierenberg: I think the best thing that can happen is that people can reflect on what they’re actually doing, and if we build the community and the culture correctly, when people look at the standards and, for example, are the treatments that they’re giving concordant with guidelines, the purpose of giving them feedback is not to beat them up and say they are falling short.
Instead, it’s a learning opportunity to understand the relationship between guideline concordance and outcomes. If there isn’t guideline concordance, what’s the reason for that? In fact, there may be very good reasons for that — if, for example, someone’s treating someone who is treatment resistant, where there’s not a lot of guidelines. At the same time, we know that the use of lithium is not as much as it should be. It’s not optimal.
How do we look at that? How do we look at the variation in treatment and the variation in outcomes across healthcare systems and then learn from each other what we’re doing so we can all do better?
Strakowski: It sounds like we build a people-centered community to ultimately improve the lives of people struggling with bipolar disorder.
Nierenberg: I think that’s an important point. I’ll turn to Peter in a moment. One of the things that I love about what we have done is that it’s not patient-centered. It is human-centered. It’s all the humans involved. It’s the clinicians, it’s the families, it’s the psychologists, it’s the psychiatrists. It’s everyone involved.
Even the insurance companies and pharma companies are involved with this. How can we really work together with that single purpose of getting better outcomes and helping people thrive?
Margolis: The comment I was going to make is that one of the things that’s really struck me about this work in psychiatry is that there isn’t a paucity of ideas and there isn’t a paucity of evidence about what to do.
What’s remarkable is it’s very siloed. Somebody over here is working on evidence-based guidelines. Someone else, like yourself, Steve, is working on system redesign and has insights into that. Someone like Andy has tremendous knowledge about pharmacologic treatment. Someone else has expertise around measurement-based care. Others have expertise around team-based care.
What has become possible, and what we’re seeing come to life in the network, is that these ideas are being brought together in the service of changing the outcomes for patients. The work is not so much about checking boxes and doing what some administrator somewhere thought was important. It’s what the community thinks is important and has prioritized as a way of actually solving real problems that people are facing every day. It’s actually really exciting to see how it’s going.
Strakowski: I know, historically, mental health care tends to be underresourced. Is your major anxiety, ultimately, how we fund this effort, particularly in the absence of a national strategy to improve resources for mental health?
Margolis: Sustainability here is going to be a really big issue. My sense is that national policy is moving in this direction and there are efforts to improve the way reimbursement happens so that physicians can actually bill for the non–face-to-face time that’s required to really do the kind of job that we all want to do for the kind of care coordination, the right kind of care planning, and all those sorts of things.
There are opportunities for the system to be better. It’s certainly better now than it was 15 years ago when we started that. We’re not there yet. This group that’s working together is doing it with a pioneering spirit. We’re going to make the most of what we have in anticipation that the world will be continuing to evolve in our direction as we start to produce results.
Strakowski: Both of you, thank you again, so much, for your comments and insight.
This is, I think, very exciting. We’ll keep an eye on this over time and perhaps come back after we’ve gained more information to see how this network’s working. Information on the Bipolar Action Network is on the website, and we’ll post that with this discussion.
Thank you all for tuning in. I appreciate your time. As always, please give us comments to think about regarding this and other topics you might want to hear about.
Again, I’m Steve Strakowski. Thank you very much.
