So, my name is Sandra DeYoung and I've had the pleasure of co-chairing this session with Dr. Anita Everett. And she's going to tell you a lot more about the roots of this presentation. But let me just start with what we're going to do today. So, we have a variety of presentations. Sadly, our first speaker, Dr. Glenda Wren, isn't able to be here today. But I'm going to share her presence through a video recording in just a moment. Then Dr. Everett will tell us about the bed crisis and the Presidential Task Force on the beds crisis, which she chaired. And then we'll hear about financial incentives and who pays for what by Dr. Steve Sharfstein. Then some variables, population, community and system factors by Dr. Trestman. I'll talk a little bit about kids and what makes the world of youth different. And then we'll have Dr. Kristin Hasmiller-Lick, who is our expert modeler. And she'll tell us about the simulated model that's been developed. And then we'll have a Q&A. So, let me start by sharing this. Well, we're figuring this out. I'll just say that one of the things that came out of the work of this Presidential Task Force was a great deal of media attention. And so a number of us have spent quite a bit of time talking to media about this. And CBS was very interested. And they wanted to hear about a family's experience with the mental health system for their child. And Glenda Wren is a wonderful psychiatrist and very active in the APA. And she herself had to deal with the mental health system for her child during COVID. And so she spoke to CBS about that experience. And that's what this video is. I can't see it. I can't see it, but hopefully you can. It was hard on all the kids. He didn't really complain a lot about it. But I know it was much harder for him to focus in online school. It's a really common story. There's not too much unique. And I've heard my own story from the mouths of others so many times. I was a mom raising my kid. Was aware of some challenges. Got help in the private sector to address those challenges. And my son did okay, but started to show some more severe signs of really decompensation. And then had a crisis occur. And when the crisis happens, I mean, you have to address the crisis. It's not really an option. The thing that was different about my experience, though, was that within the first couple of moments of being aware of the fact that we were in a crisis, I was thinking through, like, five steps of what needs to happen next. We moved very quickly through the ER because I knew exactly what to say and how to say it to get us where we needed to go. Were you frustrated, even though you knew what you were doing? I was extremely frustrated. Frustrated for myself for how hard it was. But then, I mean, always in the back of my mind, I was thinking about these other families and, like, what hope do they have and what it must be like for someone who... I mean, I still held all the feelings of shock and horror and concern and fear about the life of my child. And I have all these skills to, like, put it to the side, to call these phone numbers. How do other people do it? So you cannot lose sight of the root cause of that issue, which is the lack of supports upstream. I do think that the great majority of crises could be averted if the right interventions took place at a different time. Earlier. I do believe that. I definitely believe that. If we don't address this? We will continue to bury our children. That's a bleak prognosis. That's just, it's true. More kids will kill themselves. Yeah. Fortunately, they had... It's done? Yeah. Okay. Oh. That's it. It's done. Okay, and I'm done. Okay, great. So with that, setting the stage for us all, Dr. Anita Everett. Thank you. Yeah, and her story really is a story that's not uncommon, of course. Accessing psychiatric services, particularly if you're interested in outpatient services, can be a real challenge, as all of us know, for children, but also for adults as well. So this is a... We're here to tell you a little bit about the project, which is a work product of an American Psychiatric Association task force. That is one of the options that as presidents have is to do a body of work over the course of their year. Dr. Jeffrey Geller was then president of APA, and at the very beginning of his term, he set aside the notion that he wanted to study and come up with a report for the ideal number of psychiatric beds that would be needed to meet the needs of a community. And for those of you who are familiar with Dr. Geller's career, he's been very focused and is really a master of advocacy for quality of care in institutions. So that's sort of very much a part of what he's concerned about. I had the privilege of being asked by Jeff to chair the committee, which, again, went over the course of a year and has had quite a tale beyond the year in many different aspects. I do need to, at this point, disclose that I am here in my role, and I accepted this in my role as a past president of the American Psychiatric Association. I'm employed by SAMHSA, a federal agency, but I'm not here in this particular context representing SAMHSA. This is not a SAMHSA-funded program in any way and doesn't overlap at the current time with any work that SAMHSA's doing directly. Okay, so this is a group that started out, and the notion for me as a person who's fairly skilled in administrative processes, you've got to have a group that you can work with that's a reasonable size. And it started out with about a dozen individuals, but quickly grew. We picked up a lot as we went along and wound up with about 30 or so individuals that were part of the meetings when we met and beyond that when the subcommittees started happening. So it became a very sort of large project to manage. We started out with the concept that it was a sort of a binary question, inpatient, outpatient. But very early in the process we realized that we needed more sophistication than just binary. We don't live in a binary world now, and the notion of in versus out, particularly for those of us who've worked primarily in community-based services, we know there's a lot of variability in what constitutes outpatient care or non-hospital-based care. And so we recognized from the beginning that we need inpatient care, but also there are multiple other layers of care that can be very helpful. That right away got to the point of we have to do something that's a little bit more sophisticated to make it a worthwhile, meaningful project, because what we don't want is the old-style sort of way of calculating number of beds per 100,000 people, which is sort of the previous metric of what's the right number of beds per 100,000 people. And so that's sort of how we started. And I'm going to hand it back to Dr. Leck, who's going to talk with us in more detail, but this very sort of elegant system, this process that Kristen is expert at called Dynamic Systems Simulation. And so that's sort of our end product, our end sort of system that we landed on, our way of thinking about this project. We divided into six and then added a seventh subcommittee. To be honest, the child one was the one that we added later. At first we thought we could just infuse the child across the system, but the children advocates and our child psychiatrists kept telling us that children are not, in fact, little adults, and there are certain things that are very distinct about the children's services. The committees that we had, and I do have a copy of the report, which I'll show you in a minute, the seven committees were the historic and contemporary use of psychiatric beds, which we felt was very important to understand the context, the current context. What is a psychiatric bed? So in modern community mental health services, we don't use staffed hospitals per se, but a lot of things that sort of approximate the use of beds. So we had to sort of come up with a system or a way to identify what beds were. The financing, of course, is a major problem. It would have been easy to jump right into, well, if only we were paid more, we could have more care, but it's not as simple as that. How should things be paid for and how inconsistently are things paid for across the United States? We then had a subcommittee on population variables, and that looks at, you know, when you look at national information, hospitals and other services are used differently by different populations, by different customs, and also by different cultures and demographics. And so this is a group that sort of looked at what are some of the variables that impede access to care and or promote access to care, or promote access to care, I guess you'd say, facilitate access to care. Then we had a group of community-based psychiatry types who looked at what are sort of definitions of what are contemporary community services, what is a range of services, what makes sense. So there we looked at things, for example, as sort of community treatment teams, which are not inpatient-based or not place-based care, but mitigate the need for inpatient care when they're working as they should be. And then the last committee, which was a little bit later but very welcomed, and I'm really glad in the end that we did have a separate focus just on the children, was the Child and Adolescent Psychiatric Beds Committee. We did wind up in the end with two different constructs for how we were going to do that, so we'll talk primarily about the adult side today, but we'll also look at the child side. I think those are my main remarks. It was a privilege for me to work with all those amazing individuals that were part of it. We did create a report, which is available on the APA website. I brought my signed copy so I don't have copies for others to have, but it's available in a PDF form on the website, and for anyone that's interested, it is retrievable on the APA website. It ends at a point in time which was toward the end of 2021, which is when the body of work that was in the report, it's the work as of then, but some work has been ongoing after then, and so you'll have to understand that if you're interested in looking at the report, that it ended and the work is still going on. Okay, so I think with that, I'll turn it back over to Sandra. Thanks, Anita. And thanks to Anita for doing such a terrific job in chairing this huge task force that we had. It was a privilege to participate in this task force and to work with Anita, who is superb in terms of her leadership skills. What's interesting about the report is it says something about policy, mental health policy at a global level, but then it immediately goes to a very practical approach to how any community can decide whether they have too many beds, too few beds, and what to do about it. So I think it's got both an application for people who work at SAMHSA, and people who work in Baltimore. So I think that's a real strength of this report, and I have the somewhat guarded optimism that it's going to be influential for a long period of time. But my talk is going to be very practical and concrete. It's all going to be about money. I'm going to teach a mental health policy tutorial for fourth-year residents at the University of Maryland, Sheppard Pratt. My first sentence when we meet is, policy is all about money. So we're going to talk about this this afternoon. The slides that you see, I have to give credit to my 22-year-old grandson, who I needed very desperately. And if you look at the actual label for the slides, it says, Grandpa's Slides. So you'll see that. Okay, so here we go. It's not advancing. Oh, I was pressing up and not down. Okay, so there are two adages about money. One is Watergate, which was, follow the money. And the other is Willie Sutton, which is, why do you rob banks? That's because that's where the money is. So I'm going to give a bit of a case study having to do with financial incentives, and that's something that I'm intimately familiar with, the 30-year period where I was at Sheppard Pratt from 1986 to 2016. In 1986, when I arrived, it was 320 beds, an esteemed psychoanalytically oriented inpatient facility with 1,000 admissions and an average length of stay of 80 days for adults and 120 days for children. It had $40 million in revenue, which was 90% inpatient, 10% outpatient, which, by the way, was a community mental health center. It was the only private psychiatric hospital in the country that actually sponsored its own community mental health center, which was important in terms of the reinvention that took place. The payer mix was 80% private insurance, 10% Medicare, and 10% other. It did not take Medicaid. 2016, rocketing forward, 360 beds in two locations. At one point, after managed care diminished the length of stay from 80 down to 20, down to 10, we were down to 90 beds, but now we're up to 360 beds in two locations, 10,000 admissions, an average length of stay of 10 days for adults and children, $320 million in revenue, which is 40% is inpatient, and 60% is the expanded outpatient continuum of care, the hospital without walls, the special services including schools, residential care, and psychosocial rehab. People used to ask me, what happened? Shepherd Pratt had long stays. Well, we do have long stays. There are nearly 1,000 patients who live with us in various levels of supported housing. The payer mix was very different, with 30% is private insurance. Now 33% is Medicaid, the strategic payer for public services, 20% Medicare, and 17% other. So in 30 years, an esteemed long-term private inpatient hospital was transformed into a large public-private partnership, comprehensive community mental health center and health system. This is what the campus looks like. I don't know how many of you have been to the campus. Shepherd Pratt is arguably the most beautiful asylum in the country. For those of you who haven't been there, it's something to see. Still there? Ah, that's me when I arrived in 1986. It says, relax, honey, change is good. This was my existential experience. This is one of our schools. School buses are there. There are 13 different schools all across the state. We're the largest provider of special education in the state. These are kind of one of our group homes that we have built across the state and have acquired psychosocial rehab programs. So how did this happen? So here, survival and opportunity. Necessity is the mother of invention. My mantra was, we're not for profit, but we're also not for loss. Monetary incentives, but not just monetary. Our values are really important, and our values being public health and community care. At one point, a group of the old faculty at Shepherd came to me and said, we should downsize to 40 beds and only take self-paid patients. And I said, you've got the wrong CEO for that. Shepherd Pratt is going to become a big public health entity with the largest not-for-profit behavioral health system in the country. And my successors at Shepherd have expanded it even more. So getting to the money, how did this happen in terms of the finances? Well, first we have to ask the question, who pays? There are three basic buckets. One is government, and that's supported by taxes, and it's either the federal government or the state government or some combination of the two. Private insurance, and that's paid for by premiums in contrast to taxes. And then, of course, there's self-pay, which is an important piece, and an important piece of a part of Shepherd called the retreat at Shepherd Pratt, which is a self-pay residential program. Now, the financial incentives in each of these buckets are different, and there's always the effort on the part of the payer to cost shift. Let's see if we can get someone else to pay. We don't want to pay. So in the government, the financial incentive is to cut taxes, and so you get the governor or the legislature, and they say, this is too much, we don't want to pay, where can we cap the costs, how can we shift the costs, what can we do? And people want to get re-elected by cutting taxes. Nurse, this Johnny is impossible to figure out, doesn't cover enough, and is downright embarrassing. This is just like the rest of our health care system. Private insurance, financial incentives there, a little bit different. Private insurance, the biggest incentive is they want to cover the well and not the sick. That's called the loss ratio. Those of you who are familiar with private insurance, the name of the game is how can we enroll as many healthy people and not cover as many sick people as we can, and then we can make a profit. Whether it's for-profit insurance or not-for-profit, that's the incentive. So to make a long story short, the insurance company tells us in the midst of it all that it will pay for only half of the liposuction. And then the hospital, what about the financial incentives for me as the CEO in the hospital and all this cost shifting that's going on? The main thing is survival, self-preservation. How can we sort of manage in this kind of situation? Where can we go to find funding? Well, one of the funders we found out that there was money for special education. You know that in this country there's a right to an education. There's not a right to health care, but there's a right to education. So we were able to get special education funds and expand that dramatically. This says you don't know how lucky you are, a quarter of an inch either way, and it would have been outside the area of reimbursable coverage. And then there's the self, getting the self, and, of course, the financial incentives for people who are paying out of pocket is to not go bankrupt. Half the bankruptcies in this country are medical bankruptcies. It hurts when I go like this. And then for what? Well, for what kind of care? What's the benefit package? What are the limits, the inside limits in managed care? Managed care has been all about utilization review and deciding on what to pay for and what not to pay for, sometimes quite arbitrarily, but that's their wish. And what has evolved are a number of quality measures, which this talk is not all about, but it is something that's important. And as I will indicate later, paying for value is becoming much more of an issue. How? And how is very much caught up with the financial incentives, whether it's retrospective reimbursement, basically fee for service, which is the American healthcare system primarily, or prospective payment or bundle payments ahead of time. So I'm going to do a little digression to the Maryland system. I don't know how many of you are familiar with the People's Republic of Maryland when it comes to paying for hospital care. Any of you? A little bit, a little bit. We lead the nation. We have a special CMS waiver to do this particular all-payer global hospital budget system. It's an approach to paying for hospital care that sets an expected revenue in advance. Hospitals receive a budget revenue irrespective of patient volume. They're told exactly what they're going to get. The budget is adjusted year over year based on a variety of factors, including population changes, inflation, market shifts, and quality programs. Why all-payer global hospital budgets? Well, fee for service, hospital reimbursement, is expensive and poorly aligned with improved health outcomes. As volumes fall, fee for service threatens the viability of hospitals in many areas. By contrast, global budgets provide stability to the hospitals. You know what you're going to be paid, allowing them to shift services based on community needs and to achieve their not-for-loss on a different basis. It creates a way for hospitals to make money through basically prevention. So this story is actually a 50-year story that started in the 70s with a rate-setting process, and then a charge-per-case rate-setting from the 80s to 2014. And then in 2014, the big change to global revenue, all-payer model, and then another change in 2018 to global revenues or total cost-of-care model. So all Maryland hospitals have been paid by an all-payer global budget since 2014. The shift has been associated with substantial overall cost savings and improvements in care quality. Now, just a side note, these are all the general hospitals in Maryland, including the psychiatric units in the general hospitals. The psych hospitals, like Shepard Pratt, are not in the global system. Why? Because there was a real concern that if the incentives were to just lower length of stay, that a lot of very complex patients would be left out in the cold. And so Shepard Pratt is not in the global system. They are in a system where their rates are set, their volumes are set, their cost of care is set, but they are not in a basically global revenue situation. The evaluation took place of the all-payer model, and it was very positive that hospitals were able to operate within their global budgets. And I'm not going to go into great detail except the last bullet here, that with the savings that took place, one of the advantages for behavioral health was additional coordination with community providers, such as Shepard Pratt with behavioral health. So monies that were saved were actually put into practice in terms of better coordinated care. Here are some of the cost results. In order for it to continue to get the waiver, it has to show that it's saving Medicare money, and it does save Medicare money, but there are also quality results and a reduction of complications. Better than the national Medicare readmission rate, for example, that's another quality measure. The total cost of care, which is since 2018, continues the global budgets. It adds a primary care medical home model to coordinate the hospital model, and it permits hospitals to make investments in related services, for example, including behavioral health. Here are some of the findings. I'm not going to go into great detail, but it really is very positive. Now, let me ask a question. There are no for-profit hospitals or really for-profit medicine in Maryland. Why? Why? Can't make a profit. Okay, so that's why I call it the People's Republic of Maryland. How much is important when you go to prospective payment and capitation? There's a difference between capitation and decapitation. This came home to me when we tried to go in the business of providing HMO-type capitated care. We became the provider of behavioral health services for Kaiser in Maryland, 80,000 lives, and the per member per month that we received was a decapitated amount, so we couldn't deal with it. So, you know, how much is still really important? There's another case study here that I will not go into, but anybody should be the subject of another symposium on for-profit hospitals and the inpatient scandals of the 1980s and 90s, financial incentives, the adolescents and the substance users. There was no managed care. There was no utilization review, and these for-profit systems made a lot of money, and a number of their leaders went to jail. So that's just another story in terms of financial incentives. And the future is paying for value. I think more and more people are going to be looking toward very concrete evidence of better outcomes, and I think that's true for us as well in all of medicine. And then the final thing is a pitch for the second edition of the textbook for hospital psychiatry, which is edited by my successor and myself. It's got all this stuff in about how Shepard Pratt survived and thrived. And that's the end of my talk. Thank you. Thank you and good morning, everyone. And just as difficult as it can be to follow the money is following Dr. Sharfstein. So thank you so much, Steve. Because it's, you know, the context of what we're trying to do is extraordinarily complicated with so many different factors and variables. The, you know, what I'd like to cover in my few minutes with you are really some of the issues of this is not a dichotomous model we're trying to understand inpatient versus everything else. And so what are the variables we really need to look at? What matters when we're trying to talk about policy? How many psychiatric beds of what kind are really needed in what context? What are the community variables? And certainly what are the system and systemic variables? So focusing initially on some of the population variables, you know, we first really had to think about just the definition of what's needed. And when we're talking about inpatient psychiatric care, we're usually talking about threshold admission requirements. Someone is acutely suicidal, acutely at risk of harming someone else due to a psychiatric illness, or in some states, unable globally to take care of themselves. Some combination of those criteria contribute to the threshold for admission. And then you start drilling down into, okay, for which psychiatric illnesses and in what context? So there are so many levels that we really began to think about and it kept getting more and more complex. So the generally accepted criteria for inpatient admission, as I mentioned, are more or less agreed upon in concept. It's the implementation and the adjudication that gets tricky on the individual level. The population variables, which of those actually correlate to these risk factors? And, you know, how do the variables change? Because as most everybody in this audience knows, so many of our hospital beds are voluntary admissions, but in the public sector, many are involuntary admissions or forensic admissions. So all of those carry different expectations, demands, requirements for the kind of care, the context of care, the sequence of how they got there and where they're going after that level of care. So which are the populations that are in the context that we are thinking about in this decision process? And which are the populations that we aren't currently thinking about or traditionally aren't counted in terms of bed need or access? We had to think about, well, are we including veterans and those that are in the VA beds? What about those in jails and prisons and forensic psychiatric hospitals? Do we include those? Because they still are requiring for many, 300,000 a day in the jails and prisons are requiring active psychiatric care. So what we did in the context of all of that, thinking through, you know, what are the key issues that we need to think about as we're looking at what databases do we try to access? These were not yet complete list of some of the population variables. I'm going to interrupt you, Bob, because your slides aren't showing again. So we're having the same problem. Sorry. Nobody out there said, hey, stupid, look at the slides. Come on, wake up. That was my job. Yeah. Oh, well. So moving right along, this is a list of the potential population variables. And even here, not everything is included. What do we look at? As most of us know, someone's zip code can all too often determine their health status better even than a GWAS study can of their genetics. And you go from there, all of these different variables, which ones actually make a difference in how we think about the modeling? Because in any kind of a model, you have to choose what to include and what not, and what loading. How important is it? How much does this relate? So when we started looking more deeply into this, as opposed to how many beds do we need per 100,000 people, it gets really challenging. But it becomes a much more potentially useful tool. Because keep in mind, just to restate what Dr. Everett had framed this as, we're hoping to build a tool that different jurisdictions, different states, different regions can use in policy decision making where they're going to allocate money to build, to develop, to support programmatic activity. That's the goal in all of this. Because it's not just the inpatient units. It's also the emergency rooms. If anyone manages a system, you know that the more beds available, the shorter the dwell time in the ED. But then you have to discharge people from the inpatient units. Where are you discharging them to? Discharging them to a shelter means within a few days, they're going to be back in the ED. So what other resources are necessary as you're trying to come up with a comprehensive model? Does this make sense? So that's really what we were trying to think through. In advance. Excellent. So, in context, we're also looking at levels of impairment and the clinical severity of the illness. Obviously, a critical component is suicidal ideation. But even that is a nontrivial variable. Because is it with a plan? Is it, you know, what level of intent? Is there a history? All of these different components that we know as clinicians are critical in making the decision whether to admit someone or not. What's their legal status, as I mentioned? Voluntary or involuntary? And the potential for involuntary outpatient commitment. And different names, different states. And some states have the legal authority to do so, but never use it because they don't have the capacity to make use of it in their region. What are the socioeconomic variables that actually correlate and are predictive of inpatient use? The education index, residential stability, housing instability is an enormous factor that every one of us who lives in a health system and deals with inpatient discharge or emergency room discharge knows is critical. Employment. You know, keeping in mind, and this goes way beyond this, but, you know, if someone has a stable job, a purpose in their life, a goal, that creates a different dynamic as well in their process of recovery and flourishing in their life. Their overall physical health and the costs of care, as we know. And their criminal justice involvement. In a bizarre way, as some of us have seen, criminal justice involvement can be a ticket of entry into care that otherwise people may not be able to receive. So these are the dynamics that we needed to think through. So after we muddled through all of this, spent a solid year debating a lot of these issues, challenges, opportunities, what we found in reviews of literature were some of the higher priority variables included straightforward an area deprivation index. That was one of the greatest predictors of bed needs. And this boils down to what are the resources generally available in your community? How many supported residential housing beds are available? What's the capacity of the ACT team? Is there transitional housing available? Is there job training available? All of those that are part of this more formal ADI index. Next in this priority list were the SES characteristics. Was someone employed? What was their overall health? Did they have a primary care physician who was participating in supporting their overall care? Were they justice involved over the last year? Were they insured or uninsured? Did they have access, theoretically, to ambulatory care? As well as their educational status. These variables did contribute significantly. For example, things that were relatively low priority variables were actually the number of deaths from suicide or the number of suicide attempts per 100,000. That was not a strong predictor of increased bed needs. And people eligible for coverage or need for antipsychotics, per se, was not predictive of bed needs. So when we looked at the overall system, we started, again, building a map. And it really was one of these really overly complicated, you know, if anybody remembers the biochem charts, Krebs cycle, how that links with everything else, that's what it starts looking like when you start thinking about all of the interconnections of a genuine continuum of care. So a comprehensive community system would really provide an appropriate array of services to meet each of the levels of intensity and complexity at the right time. Hospitalizations within such a community system would be appropriately utilized and readily available for individuals who need that level of intensive supervision and intervention. And, again, what's likely to decrease functional bed utilization? Having a full continuum of crisis services available. So elements of a crisis continuum typically are designed to meet the needs of a person in crisis and would include such things as someone to talk to, think 988. Someone to respond, not just to talk to, and a place to go. All of those are critical components of a meaningful continuum of crisis services. So additional components for a full continuum, the call centers, mobile crisis teams as opposed to police response, crisis hubs or centers, medical triage and screening, psychiatric services in emergency care, crisis residential services, intensive community-based crisis intervention, and critical time interventions, all of these, none of this is new science, none of this requires more evidence. We know these elements work. There's no question. It's the issue of both follow the money and figure out in each community, each jurisdiction what's the right percentage of each that we need to deliver the care in the context of environments that are richer or poorer, more dense, more rural. Additionally, ongoing chronic care for psychiatric illnesses can substantially reduce the need for inpatient beds. Again, this is not neuroscience. This is really straightforward. Having better access to ambulatory psych services, partial hospital as well as intensive outpatient programs, team-based care. It's not only an ACT team. Really, as we're struggling now to expand access when we know we don't have enough psychiatric expertise, we need to think more and more about ambulatory teams providing appropriate care, leveraging all of our interdisciplinary team members. We've known for 50 years that ACT teams and FACT teams work well. And I think all of us can say, yeah, in our community we have an ACT team. And it needs to be five times bigger than it currently is. As well as coordinated specialty care. The vast majority of our patients, certainly those with severe and persistent mental illness, have medical comorbidities. And they're going to die ten years sooner than they would otherwise. And so we need to think about integrated medical care in this as well. Because we're dealing with acute episodes within chronic illness. Just checking. And so we're also talking about care coordination. Care management services. Which CMS finally is beginning to fund with codes. Homeless outreach services. Peer support services. You know, I'm delighted in my own health system, we're now paying for peer recovery specialists. And community health workers. Because we know they work. They make a difference. They engage people. And they can keep them in care. Psychosocial rehab. Clubhouses. For those with severe mental illness. Are critical resources. That are not typically funded in the normal stream of psychiatric care. Except in very well-funded, robust, publicly-funded systems. And community-based recovery support services. So additionally, you know, going through this continuum. Facilitation of transitions to the community. From the hospital. From correctional facilities. Tens of thousands of people a day. With severe mental illness are being discharged. From facilities. Without adequate support or coordination of care. Additionally, people are being discharged from emergency rooms. Without adequate care and follow-up. We need specialized services for specific sub-populations. We know these. Whether it's those with eating disorders. Specialized needs for substance use disorders. Traumatic brain injury. Those that are complex med psych patients. Transitional placements. And certainly, for the limited number of state hospital facilities. You know, remember the time when our state psychiatric hospitals were acute level of care chronically? All too often, certainly during the pandemic, they were flooded in many of our states with acute care needs short-term. And that's not what they were for. That's not what they're appropriate for. Forensic beds. The majority of our psychiatric hospitals have greater than 50% forensic beds when that wasn't their original intent or purpose. Decreasing this, again, going through all of these additional levels. So, with that context, you can see the extraordinary complexity of the system we work in, day in and day out. And each of us may see from only a narrow perspective. But we started integrating all of this. This is what it started fueling. The need for a highly complex, sophisticated model that could help us balance out what would happen if. And the paper that Dr. Everett said is available on the APA website for free downloading, gives you a more detailed explication of a lot of this. And so, all of these, you know, emergency evaluations, emergency holds. So, a better functioning system is expected to reduce our inpatient psychiatric bed need overall. But, it's also going to result in improved identification of people who are currently suffering in silence or in isolation, who currently are not getting the services they really need, such as ACT or assisted outpatient therapies for whom it would be appropriate. So, for some domains, it would actually increase service utilization. But until we pull this together and look at it comprehensively, we're just guessing. So, I'm excited to have had the opportunity to participate in this process, and it's an ongoing process. And, you know, we're going to hear next about the model itself and the opportunities that the phenomenal state of Michigan has helped us in, thanks Greg and company, to be able to start making it concrete. So, we can see in the real world how a model would function. Thank you. I'm actually going to keep you on tantrum hooks for a very brief digression into the world of youth before Kristen presents the model. Fingers crossed here. Aha. So, let me just say that when we started, there were a few child psychiatrists on the task force, and we were all assigned to separate subcommittees. And we felt really good about the fact that Dr. Everett and Dr. Geller had decided to include children as part of the population that this work was going to be directed for. And we loved that idea because it doesn't often happen and because the risk is if you don't include the kids, they get marginalized and forgotten. But what our experience was that increasingly as we sat through this committee meetings, we realized that actually things felt different enough in the child world that we decided we probably needed a separate subgroup. So, I'm just going to try to talk a little bit today about why we felt that way and why the needs of kids seemed different enough to have a different chapter in the report. I have no disclosures relevant to this presentation, and I just want to acknowledge those child folks who were involved in the work. So, for those of you perhaps not from the United States, I use a child and adolescent psychiatry in this country. Oh, it's not advancing. Thank you, Bob. I thought somebody was having a coughing fit over there. I don't know why not. Do you, should we get our dear friend again? So, I'm going to keep talking in the interest of time just with some basic definitions. So, a child psychiatrist in this country is somebody who has attended medical school and who has typically completed at least three years of general psychiatry residency and two years of child and adolescent psychiatry fellowship. So, there's a minimum five years after medical school. We use the term youth here for simplicity so that I don't have to keep using the phrase children and adolescents, but by that we mean anybody who is 18 years and younger. There is some debate, for example, my state of Massachusetts has formally increased the youth definition to 21 by the Department of Mental Health, so there's some fluctuation about that. And family is broadly defined as the functional caretaking group. I think it's, the other one's still open. Okay. Thank you. I don't know what he does, but anyway. I'm just going to move on. So, I have ten ways in which I think youth and patient admissions are different from those of adult. And I'll just run through them and then and cover them each in more depth. So, the first is that children are, of course, a vulnerable and dependent population. We have a higher bar for safety for them and family is part of their treatment. In general, we prefer to treat youth in the community and in the least restrictive setting and regard acute inpatient as actually a somewhat traumatic experience for kids and so we prefer not to do it, potentially traumatic. There's also a greater role, we would say, of developmental and psychosocial factors in admission and oftentimes in children, a full mental health diagnosis has not yet fully declared itself. And for us, we really rely on a continuum of care, which I'll talk about later, but even more than the adult world, that continuum is not continuous too often. Over the last few decades, the supply of child psychiatry beds has decreased while we have seen an increase in demand and that was happening long before COVID, but one of the silver linings of COVID has been that it has really drawn attention to the insufficiencies of the youth mental health system and certainly its insufficiency in meeting the pandemic and youth mental health that we're seeing now. There's also, even more than in general psychiatry, a shortage nationally of child psychiatrists and other child mental health professionals and there are interesting sort of clinical and systemic factors that affect whether children need admission. For example, it turns out that what season you're in actually matters and I'll talk about that. Next, we also use inpatient admissions not only to deal with crises. There are circumstances in which, if we are gonna make certain treatment interventions, do certain medication trials and have a chance to really observe, or for example, we would like to observe a child outside of their family system because we have concerns about that system, we often need to admit them. We've just heard about financing from Dr. Sharfstein and I would say it's even more complicated in child psychiatry that we have more different sources of funding. And finally, the population in need is actually not clearly defined, so the factors that Dr. Trestman was addressing are again, we would argue, even more complex. So let's talk about each of those. So youth, of course, are a dependent and vulnerable population. We're legally mandated to protect them and if we feel we don't have a safe place to discharge a child from the emergency room to, we may need to keep them. But we also know that there have been real stresses historically on that family system and for one of the things that we're seeing is an increased prevalence of single parent households and when you, and just more stress systems. Sadly, we've also seen, of course, significant domestic violence and intimate partner violence. Pandemics are historically have shown to have increased that rate and that's what in fact was found during COVID and that's what this map tries to exemplify. And so it's by some has been called the unseen pandemic of child abuse. So why do we prefer to treat kids in their communities? We like to keep them close to their families, right? So family is an integral part of the evaluation and treatment of children and we are often working with families to tweak things, to make things better for that child. We also don't want that child to lose a connection with their school. We want as much as possible to keep the child on a normal developmental trajectory and school is where they should be and what they should be doing. It's easier in general to keep their, you know, their extracurricular schedules and all of their sort of stuff that they do in a regular day going if they're not on an acute inpatient unit. We also know that seclusion and restraint can be traumatizing for kids, whether they're witnessing it or unfortunately being part of it and that the maladaptive behaviors that they see modeled by other peers on the unit are not always a good thing. So those are some of the downsides. We know kids have mental disorders. So one in five U.S. children have a mental, emotional, or behavioral disorder and we also know that half of all mental health disorders show the first signs before the age of 14 and that 75% of mental health disorders begin before age 24. So they definitely are there but they are more inchoate. They are less fully formed and often diagnostic clarity isn't there. In 2020, we know that one in six young people experienced a major depressive episode and suicide was the second leading cause of death for people aged 10 to 24 and sadly it has increased in certain age groups and demographics. But only half of those kids with a diagnosed mental illness get treatment. But as I was alluding to, many of these are not formal diagnoses. They are what we would call clinical problems. So things like aggression and behavioral disorganization or hyperactivity, irritability, as Gay Carlson likes to talk about, and explosiveness, self-harm without necessarily suicidal intent, trauma that is oftentimes not known by even the most intimate caretakers, and inappropriate or regressive behaviors often in reaction to such traumas or it could be failure to thrive because of neglect happening in the primary caretaking situation. So we want inpatient to be part of a continuum of care and that it should be the most restrictive. It also turns out to be the most expensive in part because it's the most highly staffed venue of care. So the idea of beds for us is really one end of the spectrum. We would like to start at the bottom of this list with collaborative care, remote consultation and screening in pediatric primary care practices, moving up through screening and clinics in schools, community supports for patients, caretakers, and families, things like NAMI, Autism Now, Big Brother, Big Sister programs, et cetera. Then we move into the outpatient, the formal treatment, starting with outpatient, including individual family and group treatment. We might move up to intensive outpatient or day treatment programs. And then start to get into things like respite beds, so place that are for crisis stabilization, specialized inpatient beds, so special places for kids with autism or eating disorders or substance use disorders. And then finally these locked inpatient units and after that potentially step down to intermediate or long-term care if that's what the child needs. So this is just a graphic of this model moving from the primary care setting on. And of course, as Bob was alluding to, Dr. Trestman was alluding to, we really need to have care management at all levels of the system. So here it is again. So the idea of a continuum by acuity of presentation and security of the unit moving from lowest acuity to highest acuity. But as we know that continuum is terribly more like Swiss cheese than it is a continuum across the country, but worse in settings where the social determinants of health would indicate and where all of those COVID inequities in care that we saw represented are there. It's exactly the same map that we would draw for child mental health. So that has made the system highly imperfect. And if you remember Dr. Wren saying she's convinced that her child's illness and other crises could be treated upstream, but the system upstream has to be in place to manage that. So why do kids get admitted when they do? Well, sometimes it's a mental illness and that may be co-occurring with medical issues like for example, insulin and the child being non-adherent to a drug. Non-adherent to their diabetes treatment. There may be a developmental, a neurodevelopmental issue with co-occurring mental illness. There may be a psychosocial issue. So for example, a child reports witnessing violence at home to their teacher or it comes out in some way and they may get sent to the emergency room and that needs to be evaluated and may end up in a hospitalization. And then of course, there's the whole world of forensics which I'm not gonna get into. So what we've seen in the last decades is a shortage of child inpatient beds and that's because supply has decreased with managed care, with utilization review and a drastic reduction in the number of beds as well as reduced lengths of stay. So you might have noticed on Dr. Sharfstein's slide, kids were staying for I think it was 82 days in his system. It might have been more. Now typically it's six or seven. At the same time, what we've seen is an increased evidence base and an increased rate of mental health diagnosis in youth. So that's the good news in a sense. The field of child psychiatry has matured and it has recognized that illnesses that occur in adults can occur in kids. So I remember very vividly being in college and being taught that adolescents couldn't get depressed because they didn't have the ego functioning to get depressed. I mean, it sounds funny today, right? As we talk about the youth mental health crisis, but I'm not that old really. And so it really was a recognition and then the recognition about bipolar disorder and so on and so forth. And there have been legislative mandates to require that children be studied and that treatments be studied in children and we found that actually they can be quite effective. So people, if their child needs care, they want care for their children. And of course we've had this boarding crisis because of this outstripping of supply by the demand. This is a map of practicing child and adolescent psychiatrists across the country. The orange are areas of severe shortage. So it's a pretty bleak picture, right? Enough said on that one, I think. So what are the factors that affect whether a kid is admitted or not? So some of these are individual clinical factors. So for example, if a child has a co-occurring medical problem like diabetes, the unit may not have the staff to be able to manage that and that may prohibit them from admitting them. Or if there's a developmental disorder, sometimes those kids, hospital units, feel like they can't manage them. Same thing with co-occurring substance use disorders. Or if a child is particularly aggressive or a youth has a history of inappropriate sexual behavior or even being a sexual predator, that often strikes them off the list. And even age. So for example, if a unit is a single unit that has 17-year-olds and you have a five-year-old who wants admission, maybe it's not a good age mix. And then there's the systemic factors. So things like type of insurance, how acute the unit is. I often hear that I'm told the unit's too acute right now for my patient. Right now, we all, I think, are experiencing terrible staffing problems, right? So that's another factor. The type of hospital matters. So is this a remote community hospital serving a huge catchment area? Or is this a tertiary care downtown hospital, et cetera? It also depends the location of the emergency service. So if the emergency service is housed right in the hospital, there's a higher rate of that child being able to be admitted to that hospital's inpatient unit. And then finally, the season of the year, as I alluded to earlier. So it turns out that the summer is a great time if you need to get a kid hospitalized, and any other time of year is not so great. It sort of rises up over the course of the year. So I use psychiatric beds for providing the most safe setting when a home environment can't manage or when the outpatient team can't manage. But as I said, we often use it for evaluation as well. The payment is this braided system. So between all of these things listed here, depending on the child's situation. So again, quite complex. And then this issue of how do we assess the population and define the population. Lots of different factors come into play. So I'm gonna stop there and just hopefully, you know, I've convinced you that kids are a little bit different. And now what you've all been waiting for is Dr. Hasmiller-Lick, who's gonna show us the model. No, no, no. Thank you. Okay. Okay, there's a magic button that none of us can find. Okay, we're going to get it to work. Let's see if it advances. So this is the thing. Uh-oh. No. Okay, it shows, but doesn't advance. We're going to get this. What does he hit? I don't know. What do you hit? What's the magic button? Thank you, Dr. Patel. This is job security. I know. I've never had such a hard time presenting. Well, let me just say, so I'm Kristen Hassmiller-Litch. I'm on faculty in the Department of Health Policy and Management in the School of Public Health at UNC. I get the pleasure of being a professor of public health, which means I get to teach, research, and practice public health, and I get to really try to address problems like what we're talking about today that are not owned within the organizational boundary of any, it's not within the boundary of any one organization, but really have to be addressed system-wide. Yes. Oh, we didn't close the other one. Okay. So I specialize in two things. One is decision support modeling. So trying to, coming from engineering, trained in engineering, but applying it in public health, trying to build models that help us understand the world better to make better decisions. The other part of what I do is system science, and what that's all about is trying to bring structured but practical methods to bear on challenges like this that are affected by things that cross boundaries, that are fragmented. So, did I get the right button? Okay. No disclosures. When I think about the system, when I talk about being a system scientist, I say I think about the system. I want you to understand that I mean this broadly, not just something that can be drawn on an org chart, but I mean all the people in all the different roles across all the organizations and in the disciplines that affect an outcome that we care about. So mental wellness of our population. I also mean all of the contextual environmental factors that we're hearing in the earlier presentations. This is a wickedly complex challenge affected by so many things. There are methods that can help us really improve how we study this as a system. And it's important because this work is hard. It requires us to really come together across boundaries in the system, disciplinary, organizational, sectoral. We are braiding. You heard that word get used a couple times, braiding resources to try to get things to be enough. It's a very delicate balance between who's willing to fund what, right? And sometimes we're asked to fund things where we don't see the benefits. They're over there. That's hard. It's hard to build the case for this. We are a complex network needing to try to figure out what our role is in this and how we can get the gears better aligned to improve outcomes. This work is hard. Okay. So let's see if we can work some magic. Oh, no. There's also no mouse. Okay. Hang on. Is it playing? There was an arrow showing. What's over here? All right. We're just going to go with it. This should move. And you would see, what you would see is a model running, right? What we're doing with decision support modeling is trying to build up a virtual version of the world in which we're trying to improve things. So we have two versions of the world, one where you're waiting in line at Panera or wherever, and we've got four servers versus five servers. And what we want to do is really try to build up this reasonable replica of our world so that we can conduct experiments in it, ask what-if questions to learn how things change, where these models become repositories of how we understand the little pieces of things to work together. You know, what do we get with mobile crisis? What do we get with any improving capacity in any of these points in the continuum of care? We can build up a model that integrates knowledge and helps us do these kind of mental, right, otherwise we're doing mental experiments. It helps us play it out and have conversations around a virtual world. So that's the idea. And we're talking about trying to build decision support models of what we in engineering call queuing systems, systems of how people wait. And these are actually delightful systems to model because there's not as many of these kind of soft, hard-to-quantify relationships. What we really need to understand is demand for our services. How many people, how often, how regularly do people need service? We need to understand what capacity we have available to serve them. How long does it take? What's the length of stay? Do we have any funny rules about how we triage or prioritize people out of the queue? This is what we need to understand. And what is really powerful about this is that what you're hearing is really it's a complex mental health system, but it's really made up of a lot of different service entities that are all linked together, interconnected. And so modeling is a great way for us to really guide our intuition and decision making in this space. So I want to broaden a little bit from just the model that we built for the task force report just to contextualize this because I think it's important for you to understand a little bit more about how we're approaching using this model in Michigan, in other places, and what I think are the next steps to move this forward. So years ago, several years ago, we built a model of competency restoration, the competency restoration system. And after listening to a lot of different places about what they were struggling with, what are the decisions you're trying to make, it really boiled down to a pretty straightforward system. We have a certain number of people who need competency restoration. We need to know how many slots, beds, we have available and how long it takes. And we were able to use queuing system modeling. This is implemented in Excel. There's a link available. These slides, I've got a lot of slides, and I'm going to go quick through some of this. These slides are available for download. Contact me. I'm happy to share all the details. But there's a link here to get to this competency restoration model. And what's really cool about it is there's three tabs. You can experiment. You can ask what if. What if we change each of the three main parts of the system? What if we do some diversion and we reduce demand for competency restoration? How does that affect waiting times? Or you can increase bed capacity. Or you can decrease service time. And when we had a lot of conversations with different communities and states, what we heard is, yes, you need a certain amount of time for competency restoration. But there's a lot of logistical wiggle room. It takes a while. So that bed, that competency restoration bed is blocked. It takes sometimes a week or two weeks to move somebody into it. And then when they're stabilized, it can take a couple weeks to move them out. So we have a little wiggle room. So you can go to the model. You can go on each tab and you can say, all right, here's kind of where we are. Here's an example with one of the states we worked with. So say we have 32 people that need competency restoration each month. The length of stay is 60 days. And we have 65 beds. What would it take to get our wait times down from where they are to less than seven days or less than 14 days, whatever your target is? You can use the model to figure out what is required. And then take all of that information to have a conversation in your community about which one of these is easiest, which is most feasible, which is most efficient in our context. And with a lot of queuing systems, they are not linear. There are tipping points. Things are really bad until suddenly they're fine. And it is a pleasure when you get to work in a system where you're really close to that tipping point. And that's what we found with a lot of the communities wrestling with competency restoration is things look horrible, really long wait times. But we're actually really close to the tipping point where we can get things under control. So in this case, we can go from if we divert 1.75 people per month from the 1.32, the expected wait time went from 51.4 to 6.2 days. Right? That's a big difference for a small change. I've worked on plenty of other modeling projects where this is not true. You need to, like, multiply your capacity by two and a half to fix it. But sometimes, and I think in a lot of our cases in mental health, this is what we're talking about. Now, the modeling is not enough, though, because I think what we ran into was that there was a lot of questioning. If this isn't consistent with our mental models, the way we understand things, people don't want to believe this. Right? So we spent a lot of time trying to help demystify the model, help people understand what the assumptions are that are in it, help them really work out the math to convince themselves that this is right, and then they can run with this information. So I think it's – I want to put a plug in here for it's not just enough to get the model available. I think we also have to really help people understand and believe in it. There's no magic in it. Right? It's our system. And then use this, guide them in using this as a decision aid to make better decisions about how to invest in change. So, after having done this work in competency restoration and similarly doing a deep dive with trying to ask questions about state psychiatric hospital capacity in North Carolina, I was asked to be a part of the task force and came into this conversation, ongoing conversation, about the incredible complexity of this challenge. And really, really deeply, this is so much how I see the world, believe that if we want to improve our mental health system, the good thing is, the good news is about all this complexity is there's lots of different levers that we can pull to create change. Right? And so where we are in our communities, there's lots of ways that we can get to better. We just have to figure out how to decide which ones and what that right balance capacity is. But to be able to have a model like this, we need to have a much broader scope of our model. We can't just look at one piece of it. So, sometimes we do. We've done lots of deep dives like the competency restoration. But for this, we really need a model that's going to incorporate the broader continuum of supports. So, this is the report and the work group that worked on the modeling part. I'm going to go quickly through this because it's all described in detail in the report. But I want to go past this into kind of some of the insights that we have now as we apply this. So, just important to say that we defined an acute crisis as something that needed more than an outpatient call, an outpatient visit or a call. So, it needed a little bit more than that. And we talked about recognizing that the continuum, we tried to think about, how do we represent the breadth of the continuum but also boil this down? Everything that you add in makes the model more complex. I'll show you a picture in a minute about the underside of this model. And it is the Krebs cycle. I affectionately call these spaghetti diagrams. It looks like someone threw a bunch of pasta on a plate. They get big really fast. So, you have to really try to do the stance about what complexity is most important to include in the model to help us make decisions but not too much because we need to focus on the core decisions that we're trying to make. So, here are the different components of the system that we have included in the first version of the model. So, mobile crisis, intensive team-based care, like ACT team, other models as well, mental health crisis receiving centers, community-based crisis beds. Sorry. Okay. So, hopefully you can see the inpatient beds. And we did a lot of lumping. Had a lot of conversations about lumping and splitting. We did a lot of lumping because we're trying to really think about the decisions people are trying to make. So, I want to just quickly give you a sense of what's under the hood of the model. But, again, read the report for more detail or reach out and talk to me. But the way this model works is we're trying to understand stocks and flows, where people are within the system. So, the blue boxes are, you can think about these as stocks or bathtubs, where we have people accumulating. The pipes are how people flow through the system. So, the first part of the model is keeping track of, whoops, back here. So, it's keeping track. We're thinking about the number of, here, this is adult. I'm showing you the adult model. How many adults go into a crisis in a given month, at a given point in time? And, initially, they are in the community. And this is a function of a number of variables, trying to, again, reflect on the long lists. It's a function of the quality and adequacy of your community-based outpatient system. It's a function of the population size. It's a function of some of the aspects of Maslow's hierarchy. Like, how hard is it to be housing secure, food secure in your community? If that's harder, it's going to lead to a higher rate of incident crises. So, we're trying to think about what the equations look like to predict the rate of incident acute mental health crises in our community. You're going to see this model grow. So, as we add to the right, you're seeing different pathways that people can flow from the community. They can move into an emergency department or a crisis-receiving center. From there, they can move into inpatient beds or community-based residential treatment. If we keep going, we can start to think about adding in mobile crisis, ACT teams, which are pulling people out of the population and trying to help reduce the number of acute crises. So, that's both going to affect the arrow down to the inflow into the acute crisis, but also over to where people go. So, when there is a crisis, they're going to go to the ED. We also added in the step down, sometimes referred to as step up. The language is very tricky. But what we're thinking about here is, if these services are available, somebody who is otherwise hospitalized can move out of the hospital into community-based services, which frees up those hospital beds. So, it's important to really understand the capacity there because it affects how full our inpatient beds are. Down at the bottom, you see the competency restoration, which we felt was important because a lot of our communities and states, the competency restoration fires are putting pressure on trying to shift civil beds over to the forensic side. We're seeing this in a lot of places. A lot of our state psych hospitals are getting filled up with forensic. So, it felt important to really build this into the model because if we want to understand the impact that that has on the broader system, we need to include it. You'll notice some clouds in the diagram. And what that means is that's the boundary of the model. We're not tracking things before and after those clouds. We're focused on what's happening inside. It's very important when we're modeling like this to think about scope. It doesn't mean we ignore things outside, right? Just like the outpatient mental health system, there's a variable in here. And its strength affects the incident rate, incident acute crisis rate. Okay. And then in PLUM, you'll see all of the things that we're simulating, that we're tracking in this model. So, as a function of the capacity demand for services and how people flow through, we can see how long people are boarding in the ED. We can see wait times. We don't have this in the model right now, but we can track costs of all of this. Okay. Okay. I'm going to go real quick. The underside of the model is pretty wicked, right? It's complex. This is a system dynamics model. We wanted to really approach modeling in a way that isn't as computationally intensive. We wanted to build a model that could help with strategic thinking, that could really be used in real time with decision makers in a community to really understand how changes in capacity translate into changes in system outcomes. Part of what was really important as we were building this model, and this gets us at the cutting edge of modeling the technology here, is that for this model to be realistic, it was important for us, you all, we are living in systems that are right up at the edge of the tipping point. I think the way that we manage the mental health crisis system right now is that we pull safety valves to keep this reasonably functioning, right? And so we needed to be able to simulate that. And so for that, we had to bring in some individual-based discrete event modeling. So this is a hybrid model, but it's really cool because it runs in real time. And so what you get are these dashboards. Here's a real simple vision of it, where basically we can show lots of trends over time, how the system is behaving. Here we see in the upper left, we see the number of individuals in the ED being treated. On the bottom left, the individuals in the ED boarding. Top right, civil hospital bed capacity utilization. Bottom right, individuals in crisis arrested and divertible. So you can see how then on the left, when you push the levers, we just have a couple shown here, but in reality, there's a whole bunch more. We have all these components of the model, and we can adjust the capacity in all these parts of the system and see how that's going to ripple through to affect all of these other trends that we might care about, okay? And so in this, we see even more trends, right? On the left-hand column, it's all about things we care about in the ED. Middle is the community-based crisis, right, or a number of other outcomes that we care about just from more of a public health perspective, number of people that are arrested that maybe could have been divertible. And the capacity of all of our different, the utilization rate, how well are we using our capacities? Okay, so what we are trying to do, just to broaden this from being a model that you can, you know, use online, I think the right way to get this to really change the decisions that we're making is that we want to invest in trying to understand the system. And so this involves having a lot of conversations with people on the front lines that understand how the components are working. And this is where we are in Michigan, is really trying to understand how things fit together. What is your capacity, your working capacity right now? What are the constraints to that? Things that we're hearing is that workforce, workforce is more of a constraint than it has ever been. I'm hearing this not just in Michigan, but in North Carolina and Oregon and everywhere that we're really starting to vet this model. But really trying to map the local system, trying to understand the quick fixes and the desired fixes that people think are important across the system. Trying to also hear about the pain points, right? What do they want included in the dashboard for this to be a useful model for them? The next step in the middle, sorry, it's not showing the same thing up there. The next part of this is to then engage folks with the decision support model. Once we've got the model adapted, kind of representing the system in which we're trying to have conversations and support decisions, we want to engage folks with the model to really help them simulate and learn about the effect of their mental models about what's going to work. What changes do we want? Two minutes? Almost at time. Okay, okay. Sorry, y'all. We are running close here. I'm going to go really quick and just try to show you what I think are some of the most important things. So what we have learned in Michigan specifically is that the model structure, and it's not just Michigan. I've also continued in a couple of other states. The model structure works really well. In Michigan, we knew right away that this model structure felt right. There are a couple of changes that we made that we are needing to make. One, so replacing the crisis receiving center, which doesn't exist in the county that we're working in, with psychiatric emergency services. They function similarly, so that's a fairly easy swap. There's a shelter plus outpatient mental health services program that is really functioning as a step down in a very important way. And so considering that as an important component of the step down component of the system. And also really I'm hearing a lot about mobile crisis and needing to learn about how we're doing there. Are we stabilizing the right people well enough? How that fits in the system. Needing to allow some but not all people that are waiting for step down to, in the current version of the model, we have people waiting in the hospital before step down, but we're needing to allow for some of this waiting to happen if people are well enough stabilized in the community. And then also needing to think about how to build workforce in. But the overall structure works incredibly well. In a couple of, in Oregon, we have been really having conversations about, ooh, do we need to explicitly model the involuntary commitment process? And started down that path and then backed up to inform decisions this felt right. Okay, I'm going to, yep, I'm going to stop there. And welcome. Any questions? Do we have a minute or two? Are we out? Yeah. So. Thank you. Yep. So we are at time, but if anyone has questions, we can stay around and are happy to answer them. But for those of you who need to go, please. Thanks for joining us. Hey, I think, um, you know, if you want, I think there's still recording. So if you want to just talk into the microphone, that would be great. Thank you. Hi, how are you? Good. Good. Go ahead with your question. All right. Great. Um, my name is Caroline. I'm the co-founder of Charlie Health. We're a virtual IOP. So we're doing exactly what, um, all of you are talking about, which is trying to combat sending kids to the emergency room and trying to get them the treatment that they need in their homes since there's so little beds. Um, one of the things that I was wondering about is the model in which you talked about the tipping point. I'm curious around, like, was that real world? Because you said, for example, you said like, this is really counterintuitive to think that if you go from, you know, just making sure that an individual, 1.75 individuals does not go to the emergency room or try and get a bed, that it's going to bring it from 50 days of a waiting list down to six or, and, and you, I was staring at that slide and it's something that I'm curious about if you don't mind expanding on it. Okay. Yeah. Um, I can't, I, I am just recalling, I'm just kind of smiling. I had so many conversations with folks that had that same question, like, this is too good to be true. How could it possibly be? Right. And, and on the flip side, if we, if we do, if we solve this by adding beds, but pulling them from the civil side, is that as bad as it, as we think and simulation model showing? Yes, indeed it is. Um, so what we found is, yeah, people were skeptical and then some of these small changes were made and it really is, it really, it re, the systems really do behave like this. There are these tipping points. There's no magic in these models. We're just keeping track of, you know, like I showed with the people waiting in line to get served. Yeah. It's really just people, it's, it's, you know, it's a question of how are people arriving? Is our service set up, you know, to, to really kind of move them in efficiently? Um, and, uh, and, and there's no magic. It's just, you know, when does one extra server, when does one extra bed, when does one, you know, 10, 10 people diverted? You just need to figure out how to get below those tipping points. It is real. It's not always that easy because a lot of these are kind of embedded and interconnected and sometimes, I think this is the other part of it, we have to also understand how other parts of the system are reacting to that because sometimes we're like, ooh, you know, this looks better here, so I'm going to, I'm going to put a bunch of people in, right? So that's why I think this big modeling that we're doing now is so critical because we need to see how things that might look good to me in my part of the world actually ripple through the bigger system. So I think that can make it not true if, if somebody else is coming in to use that excess capacity that you freed up in a different way. I have one more question. I'm sorry, just as a follow up to that. Um, I'm sure that you are going to get to this, but, um, have there been really clear proposals around ways in which to increase beds or throughout that Michigan, and you'll have to forgive me, I didn't read the report, but I plan to because it's fascinating. Um, but have there been concrete proposals on how to increase different, make those small changes to have that tipping point happen? Um, so I think, so I'm not always getting the pleasure of being around when, when those decisions get made, but in Michigan, we're early on, and Greg, you might want to say something here. We're early on. We, um, one thing that is hard about this is that our data systems and mental health are so fragmented. And so if you really want to model at this scale, that's a lot of information that needs to get put together. They are being brave and going early. Um, and so doing a lot of, a lot of the work we've done, we've approached this by trying to have lots of interviews to really get people to guesstimate, tell us approximate. And so we're going to, the model's just about up and running, but they're sort of washed in awe at the county, the Michigan county that we're working at, their adapted version. Um, my hope is that this could start to inform decisions right away. In the interviews we did, there's a hunger for it. Even the questions that we were asking folks to get the model estimates, we were told several times that those questions are helping them think through changes that they would like to make. So that's heartening, that this is work, you know, to inform a model like this, but just this kind of structured questions can help people think through kind of how things are working. So Greg, I don't know if you want to add anything to this. We're early stages still. Um, but I think it is getting people to question and really think through how the system is working. And it's getting, it's allowing us to really document some really cool things that are true about their system that help make things work, you know, better. And then the pain points that are making it hard, those insights can help us, you know, kind of figure out what to build up and avoid in other places. And Dr. Dalek is the chair of the Department of Psychiatry at Michigan, Washington County, which is where that system resides. Oh, I'm so sorry. I thought that that was him. No, sorry. Um, so, uh, appreciate the opportunity. This is such a complicated, it was a great set of presentations, by the way. Um, and it really points out the complexity. Uh, and as we start to work in our own county on this with disparate providers, CMH, our facility, uh, our system at U of M and the, uh, other community settings that are in our area, our payment incentives are different. Our patient incentives are different. And we all knew this, but getting into the responses to the questions that, uh, Dr. Hasmullick and her team are putting to us is getting us to think further and deeper about this. And so we are very early on, but very interested in continuing to do this deep dive to try and bring some rationality and organization to the whole structure. And are you all ready to make changes based on what we're learning? Um, well, changes are being made in the system even before we know how they're being driven. We had one of our institutions close their inpatient unit without any real examination of what the impact would be on the rest of the community for reasons that are understandable and largely financial in their system. But it just goes to show how much change is happening even before we have a clear idea of what the roadmap looks like. One more thing. Thanks, Greg. I'm sorry. I'm just going to go to Dr. Keepers because he's been waiting so patiently. Thank you very much. Um, I'm George Keepers. I'm the chair in Oregon. And thank you all for your wonderful presentations. And thank you even more for the work that is behind these presentations. This is really important work. Thanks for your efforts in Oregon particularly. We are building such a system in Oregon. And I just wanted to reflect on the system we built before this, which was related to bed care and coordination in acute care hospitals during the pandemic. As many of you don't know that we're short on beds, period, in Oregon. And finding enough ICU beds for all the people who needed them was an incredibly difficult problem for us there. And building this kind of a system to account to help us with that was what got us through the pandemic. So I really endorse this kind of project. My question and comment really is related to another aspect of this, which is that our systems, of course, exist in relationship with the judicial systems and with the legislative systems. And they can make interventions in our systems that can really screw things up. And I'll give just a couple of examples from Oregon. One of them was the Department of Justice injunction that required us to treat all aid and assist patients prior to any committed patients. So we basically don't treat committed patients in the state hospital. Dr. Sharfstein knows this. He's been to visit us. Another one was a recent law that was passed that basically criminalized the use of restraint for aggressive children in our residential care facilities. And you can imagine the effects of that. These are, in many cases, well-intended interventions from the judicial system and the legislative system that have disastrous effects on the mental health system. Thank you for those comments. I'm a little bit worried about people coming in. Yeah, so I'm going to go to Dr. Applebaum next as our last question, because we do have to leave the room. Dr. Applebaum. Thank you, Sandra, and the panel, for a fascinating set of presentations. So my question is, you started out with a quantitative question, right? How many beds do we need? And I understand the conceptual flow of the model. But when I look at the model, I don't see any numbers attached to it. And it seems to me for it to be truly useful to answer the question you started out with, there have to be numerical parameters associated with the effects of each of the arrows from one box to another. So my question is, are you developing, trying to develop those numbers? Where is that coming from? Is that what the Michigan work is focused on? And how likely is it that whatever parameters you come up with will, in fact, generalize across jurisdictions? Yeah. So that's a great question. So the numbers that are in the model, the model is not without numbers now. The models that are in there now came from really drawing on data that was available to us. Again, we were trying to build model structure that would work across communities, across states, so it was generalizable. Numbers we tried to pull from data that was national, and also from the literature, experts from the task force. And so it's credible. It's realistic, but not real right now. But what we're trying to do in Michigan is really ground it in their actual data. Yeah. We put certain assumptions in there to create numbers for those to start out with, and that's my term for it is calibrate, but that may not be technically correct. Yeah. What we're trying to do is to be calibrated in several different communities to assure there's... Right. So what we're trying to do is really see whether the model is consistent with all of the different glimpses of data that we have. So what we understand mobile crisis teams do, what we understand ACT intervention to do in terms of reducing acute crises. We're trying to see whether the model is consistent with all the data glimpses that we have since they're so fragmented. This modeling can help us integrate them. Yeah. So we're doing this in Michigan, a little bit behind that in North Carolina, hopefully in other places. So if you have communities that have data available that want to keep testing, building numbers into this, that's how this is going to get better. But the goal is to try to get a robust model structure that can be appropriate for lots of different communities and to get better and better at really understanding what are the data that we need. There's a lot we can do with a model in terms of uncertainty analysis to figure out how much does uncertainty in certain parameters matter. Maybe it doesn't, but others do, and that's going to guide our research data collection. Thank you. Thank you so much to all our presenters and to our wonderful audience.

mental health care

psychiatric beds

resource allocation

COVID-19

American Psychiatric Association

systemic challenges

community services

quality care advocacy

health care modeling

dynamic systems simulation

stakeholder engagement

policy development

systemic changes