With 988—the nationwide three-digit number for mental health, substance use, and suicide crisis—going “live” on July 16, Kana Enomoto says a massive cultural shift is happening in communities across the nation. That’s why she and her McKinsey Health Institute colleagues collaborated with RI International to develop a user-friendly, interactive Crisis Resource Need Calculator. “We quickly realized that many of our state and local leaders would need strong, useful tools to help them make a case for developing a full crisis care continuum and plan their own work,” she says. Enomoto is the global director of Brain Health for McKinsey Health Institute and a national leader in mental health and substance use policy, data, programs, and practice improvement.
The previous iteration of the Crisis Now calculator, developed by RI International in 2017, has helped people estimate the optimal allocation of crisis system resources and associated healthcare costs. Nikhil Seshan, a McKinsey & Company consultant who supported the calculator redesign, says the model stays true to the original by helping users estimate a community’s capacity needs and behavioral health system costs. “The new version answers the same two questions for the user: what’s the potential healthcare cost associated with crisis and inpatient care, assuming the state or locality serves all people who need in-person care?” he says. “And what’s the potential capacity needed for their different sites of care if they’re going to serve all people who need in-person crisis care?”
However, the third element of the new iteration of the calculator is novel. Users can now examine the shift in costs and benefits of incorporating specific Crisis Now elements, separately or in conjunction. They can alter the sites of care they’d like to implement by adding crisis receiving facilities, short-term crisis beds, or mobile crisis teams. (If the user adds all three, text pops up that reads, “This is equivalent to the Crisis Now model. Please refer to the Crisis Now model already on display.”) “For example, a county can look at the potential changes in annual crisis system costs if they only added mobile crisis teams or crisis receiving chairs or both,” Enomoto says.
Some calculator findings might be surprising for communities. For example, Paul Galdys, deputy CEO at RI International and former assistant director at Arizona Medicaid, notes that when a county, region, or state without crisis receiving facilities or short-term crisis beds toggles on mobile crisis teams, they’ll quickly find that potential annual behavioral health acute inpatient and crisis care system costs could also rise. “In this scenario, mobile crisis teams responding to a person who needs in-person care have no place to go but the hospital.” That said, more people would likely receive more appropriate levels of care, and shifting to a mobile crisis response instead of a law enforcement one helps communities divert people in crisis from justice system involvement, benefits that aren’t included in the calculator.
As a user changes the assumptions on the left-hand side of the calculator, the right recalibrates. For example, the below bar graph of Orleans Parish shows an emergency department-inpatient only scenario (left bar) compared to a modified scenario where the user has added crisis receiving facilities (middle bar) and a fully implemented Crisis Now model (right bar).
The Crisis Now model aligns with SAMHSA’s National Guidelines for Behavioral Health Crisis Care, which provides a framework for the elements a community’s crisis system must have to be effective and comprehensive. That includes regional or statewide crisis call centers that coordinate in real-time, centrally deployed 24/7 mobile crisis, and 23-hour crisis receiving and stabilization programs. The guidelines also integrate crisis care principles and practices such as not requiring medical clearance from an emergency medical facility before admission and “no wrong door”—an approach where facilities accept everyone who comes.
The objective of the collaboration was to refine the calculator’s functionality and pressure test assumptions where possible. The team also hopes that the tool helps ensure 988 doesn’t become “a bridge to nowhere,” especially for people who need in-person crisis care. The first step was for the team to determine the monthly prevalence of nationwide behavioral health crises. They started with Vibrant Emotional Health’s estimate that the potential serviceable population for the 988 hotline could be 39 million people. (Vibrant Emotional Health administers the SAMHSA-funded 988-Lifeline.) This figure was then multiplied by published estimates for the number of calls per person (ranging between 1.2-1.75) and the proportion of those calls that would need in-person care. Estimates for how many people in a community need in-person care vary but typically hover between 10 and 20%.
The team also incorporated a behavioral health claims analysis at the national level, looking at acute inpatient, crisis receiving facilities, and mobile crisis services claims. Seshan points out they also had to consider people experiencing a behavioral health emergency who don’t enter the health system but, instead, come into contact with law enforcement or other first responders. The analysis was informed by an estimate from the Treatment Advocacy Center’s Road Runners report that 34% of all emergency law enforcement pickups don’t get transported to in-person healthcare facilities. (This is on page 32 of the report.)
The layered analysis resulted in a startling figure: the total number of crisis episodes across the nation is roughly 230 per 100,000 people per month. Enomoto points out that the estimate reflects the total addressable market and isn’t meant to predict 988 call volume nor how the launch of 988, and likely corresponding increase in Lifeline calls, texts, and chats, will affect crisis care referrals. “The estimated prevalence rate is based on population need,” she says, “rather than the potential influx from 988-Lifeline.”
That said, Seshan shares that 988, and associated data collection, could influence future iterations of the calculator, especially state-specific information. Among the state-specific data are the average cost of an acute psychiatric inpatient bed and the average length of inpatient stay. These estimates come from commercial and Medicare claims analysis. However, he says user input could help further refine the calculator. “The team could enhance the calculator to capture user-input numbers,” says Seshan, “so that over time, trends can be identified like whether users are consistently changing the average length of stay for a particular state.”
The newly launched calculator is pre-populated with national data and those tailored to a community’s population, such as per diem costs, the average length of inpatient stay, and estimated travel time for mobile crisis teams. “This makes it intuitive and easier to use than the previous iteration,” says David Covington, CEO at RI International. In addition, users can choose to look at a single county, multiple counties, or an entire state. (The calculator’s Frequently Asked Questions page provides transparency on each pre-populated assumption and its sources.)
While the calculator is pre-populated, users can change any data assumptions, whether monthly episodes, length of stay, or even utilization assumptions like initial “triage to” or “referrals from” mobile crisis services, crisis receiving facilities, short-term crisis beds, and the emergency department. “The team used estimates that were backed by data wherever possible, but also wanted to enable people to make changes as they see fit,” says Seshan. The right-hand bar graph will reflect the edited assumptions as the user makes modifications.
Seshan points out that while communities often focus on behavioral health costs, interviews with state leaders highlighted they also need to make a case for the potential human cost of an insufficient system and the benefits of a robust one. “That’s much harder to do in a purely numbers-based calculator,” he says, “but working with RI, the team was able to create a module to help people visualize potential differences in utilization between user-input ‘current state’ and the Crisis Now model. This could enable a state or county leader to say, ‘Hey, in the current scenario, we might not be serving as many people who need to be served and, by adding crisis stabilization and mobile crisis services, more people could get the most appropriate level of care.’”
This visualization can also help leaders determine whether a state or county is underserving or overserving its population at a particular site of care. “If mobile crisis teams are underserving people, that could represent a need and suggest the community might benefit from investing in more teams to serve more people,” says Enomoto. Similarly, higher emergency department use could be a red flag, indicating a system issue that could lead to problems like psychiatric boarding—a high cost to people in crisis and the community.
Users of the calculator can visualize the number of in-person behavioral health episodes a state, region, or county serves annually compared to when the Crisis Now model is implemented in a community with the same population size. By inputting their current site usage—mobile crisis, crisis receiving, short-term stabilization, emergency department, and acute inpatient hospitalization on the right-hand side, they can see the two side by side on the left.
The team recognizes that SAMHSA’s national guidelines and the Crisis Now model are the “gold standard” of how states, counties, and regions should respond to crises. However, Enomoto says they also realize that not every community can immediately produce a comprehensive crisis continuum. “The calculator gives people the flexibility to experiment with different model components that might best fit their community and its readiness,” she says. “We collaborated with RI to develop the calculator as an ongoing tool communities use throughout their 988 redesign as they expand the continuum of their crisis response system.”
