University of Michigan First to Test APA Model for Assessing Psychiatric Bed Needs

In a fragmented and underfunded mental health system, where should a community invest scarce resources to best meet the demand for psychiatric services?

Expand the number of inpatient beds (the most expensive option, demanding hospital infrastructure)? Establish a mobile crisis unit? Create a jail diversion service? Fund more assertive community outreach? Build more residential and step-down psychiatric services?

In 2015, Gregory Dalack, M.D., chair of the Department of Psychiatry at the University of Michigan Health System, organized an informal “behavioral health consortium” of local mental health service providers in Michigan to answer just that question. Partners included two hospitals in the Trinity Health System and Washtenaw Community Mental Health (CMH). The latter also operates a jail diversion service in coordination with the Washtenaw County Sheriff’s Office.

“We have been struggling with how to meet the needs we see in our institution, knowing we are part of a larger community,” Dalack told Psychiatric News. “It’s hard to coordinate services among the various providers because we each have our own populations that we serve, though sometimes they overlap. Our ER is an example—people who get services through CMH or the other institutions will sometimes come through the Emergency Department at UM. Coordinating services where the resources are scarce is a constant challenge.”

Dalack is a former treasurer of APA and former chair of the APA Registry Work Group, which helped to develop PsychPRO, APA’s mental health registry.

Five years after Dalack began working with his consortium, the APA Presidential Task Force on the Assessment of Psychiatric Bed Needs in the U.S. was charged by then-APA President Jeffrey Geller, M.D., M.P.H., with developing a model for determining how many psychiatric beds a community might need to meet its demand for psychiatric care. Chaired by past APA President Anita Everett, M.D., the task force comprised more than 30 APA leaders, mental health professionals who are experts in health services and population health, and members of the APA administration. Everett is the director of the Center for Mental Health Services at the Substance Abuse and Mental Health Services Administration.

As part of the charge, the model had to account for the wide variety of other services that might exist in the community—mobile trauma teams, assertive community treatment, residential services, jail diversion programs, and state forensic services, among others—and how those services might impact the need for inpatient beds and the overall capacity to care for people with mental illness.

The basic components of the model, which will be continually refined and built out, are described in a new report issued at APA’s 2022 Annual Meeting, “The Psychiatric Bed Crisis in the U.S.: Understanding the Problem and Moving Toward Solutions.”

At the Annual Meeting, Everett provided an overview of the work of the task force and the complexity of developing a tool that takes into account multiple levels of mental health care in a fragmented system. She was joined by Kristen Hassmiller Lich, Ph.D., an associate professor of health policy and management at the University of North Carolina School of Global Public Health; she chaired a subgroup of the task force with expertise in statistics and systems modeling.

Hassmiller Lich provided a visual display, via computer simulation, that demonstrated the power of the model to predict how changes in any aspect of mental health care in a system affects all other aspects of the system and ultimately affects the community’s ability to care for its population of people with mental illness (See “Task Force on Psychiatric Beds Produces Model for Determining Need in Any Community”).

Today, Dalack and his partners in Michigan are the first to put the task force model to the test in their community. They are now gathering data that can be fed into the APA model—a daunting task in itself, since each of the partners uses different databases and reports data in different ways.

But Dalack believes the model could offer his community, and others around the nation, the kind of information that has never been available before—the kind of information that could help his community answer the questions he began asking in 2015.

“We became aware of the tool last year and began to have a dialogue with the task force,” he said. “We told them we wanted to work with the model to animate it, plug in some numbers, and determine if it gives us information that has face validity.

“Right now, when you make a decision [about how to fund mental health services], you make that decision on your own largely parochial view of things, without knowing how that decision impacts other players in the system,” he said. “The model will give us more information about where strategic investments can be made and what the impact will be.”

Although the model is still being refined, the task force is seeking other community leaders who may be interested in testing it.

Everett said the task force model draws on artificial intelligence and is similar to a “machine learning” algorithm: The machine gets better and “learns” as it gathers more data. The participation of the University of Michigan and its partners will significantly advance the ability of the model to accurately forecast for other communities where they should focus resources, she said.

“This kind of model development gets smarter and better in terms of ability to estimate service needs with more participation from a variety of communities contributing their data,” she told Psychiatric News. “We are enormously grateful to the University of Michigan, Trinity Health, and Washtenaw County for their participation.”

Dalack emphasized that all of the points of service in the continuum of care—from community outreach and jail diversion programs to inpatient hospitalization—require an investment of money and material resources. But for most communities most of the time, building a jail diversion program is likely going to be easier than building new hospital beds.

That’s where the task force model can be so useful. “We all know we are underbedded, not just in Washtenaw County but nationally,” he told Psychiatric News. “But it’s not just about the beds. If you have more jail diversion services and a partial hospital program, what would that do to the additional bed need? It probably wouldn’t make it zero, but it might make it a smaller number than if you only were focusing on who is in your emergency department and needs admission and trying to build the requisite number of beds.

“If we learn some things, the task force will be empowered to take this to other communities that might use the model and then go to their state legislatures and say, ‘Here is where we need help, and this is what the impact would be.’ ” ■