Individual placement and support (IPS) is an evidence-based practice that improves competitive employment outcomes for people with serious mental illness (
1). Originating in New Hampshire in the 1990s (
2), IPS is now found in most states throughout the United States, although availability remains severely limited compared with the need (
3). A 2016 national survey of state mental health and vocational rehabilitation (VR) agencies identified 523 IPS programs in 38 (75%) of the 50 states and the District of Columbia, with states varying widely in the number of IPS programs and clients served (
4). The states also varied greatly in support for the implementation of new IPS programs and maintenance of existing programs through policy decisions, funding, technical assistance, and fidelity monitoring. The survey found more extensive support for IPS expansion in states that had joined a quality improvement collaborative devoted to IPS, called the IPS Learning Community (
5).
The 2016 survey noted several trends suggesting continued, if not accelerated, growth in IPS services nationwide, including continued growth of the IPS Learning Community, legal settlements requiring states to provide IPS services based on the 1999
Olmstead decision (
6), and federal initiatives led by the Substance Abuse and Mental Health Services Administration (SAMHSA) and the Office of Disability Employment Policy. These developments suggested that a follow-up survey would be useful.
The goal of our study was to determine the changes in available IPS services in the United States since 2016. We hypothesized that IPS services would continue to expand in the United States and that states in the IPS Learning Community would have a greater increase, compared with states not in the learning community. We also hypothesized that state infrastructure support would be more common in learning-community states. Using methods similar to those of the 2016 survey, our study updates that survey, comparing changes over time and addressing these hypotheses.
Methods
Using similar measures and data collection procedures as in the 2016 survey (
4), we conducted a national telephone survey of representatives from state mental health and VR agencies in all 50 states and the District of Columbia. The interviews focused exclusively on IPS services, unlike the more general 2016 survey. The Westat Institutional Review Board, which followed the principles outlined in the Declaration of Helsinki, approved the study. Respondents gave verbal consent for participation.
Sample
The sampling frame consisted of the 50 states, the District of Columbia, and two counties: Alameda County, California, and Broward County, Florida. Both Alameda County and Broward County are members of the IPS Learning Community and are treated as separate jurisdictions apart from the rest of the state. When calculating the total number of states with IPS services, we included California in the counts for both the learning-community states (i.e., Alameda County) and non–learning-community states (i.e., California, excluding Alameda County), but we counted California only once in the national total to avoid duplicate counting. Florida did not pose a duplicate count issue because in 2019, Florida had no IPS programs outside Broward County.
Our sampling plan was to interview at least one key informant from each state—specifically, an official employed by the state mental health or VR agency and identified as the administrator responsible for monitoring employment services in the state for people with serious mental illness. In three large states where the state agency’s functions are decentralized (Florida, California, and Arizona), we modified the plan, interviewing county officials, who were more knowledgeable than state officials about employment services in their counties.
We identified respondents by using several sources: the 2016 survey contact list, current state leaders in the IPS Learning Community, participant lists from prior IPS Employment Center training sessions, state behavioral health and VR agency websites, a contact list of state commissioners available from the National Association of State Mental Health Program Directors, and recommendations from interviewed participants. We sent introductory e-mails describing the study and requesting participation in a telephone interview.
Respondent Sample
We interviewed 70 representatives from 50 states and the District of Columbia, including 46 state and county mental health agencies and 12 state VR agencies. We conducted all interviews with a single representative, except for six interviews with two representatives. Interviews with county representatives included one in Florida, six in California, and one in Arizona.
Interview Procedures
Two authors (J.A.P. and S.M.L.) interviewed 44 and 26 respondents, respectively. The first author randomly assigned states to each interviewer. The interview protocol was structured into a series of 26 open-ended and closed-ended questions. The interviewers received training in the interview protocol before their first interviews. All interviews were conducted by phone between February 2019 and September 2019 and averaged 45 minutes to 1 hour. The interviewers introduced themselves as researchers and trainers at Westat and the IPS Employment Center. During the interviews, they took detailed notes and entered each question's response into a spreadsheet. After interview completion, the interviewer reviewed the data in the spreadsheet for accuracy and added any additional explanatory notes to the spreadsheet.
Program Measures
Learning-community membership.
We identified states that belonged to the IPS Learning Community in 2019. We considered the five states that joined between 2016 and 2019 as learning-community members when comparing 2016 and 2019 survey data.
State IPS services.
This dichotomous measure indicated whether a state had any IPS programs. To ensure consistency of terminology, the interviewer first determined whether the respondent was familiar with IPS, then briefly described the IPS model and explained that several other terms sometimes were used to refer to IPS, including the “SAMHSA Toolkit on supported employment,” “evidence-based supported employment,” and the “Dartmouth” model.
Number of IPS programs.
To determine the count of IPS programs, we asked each respondent to report the number of IPS programs in the state or jurisdiction. The number of IPS programs corresponds to the number of agencies providing IPS services. States in the IPS Learning Community report quarterly data for all participating IPS programs in a centralized data portal managed by the IPS Employment Center. We obtained program data from the portal, asked learning-community state leaders if they had any additional IPS programs, and reported the total. Leaders from non–learning-community states provided data from their own tracking systems or provided estimates based on their knowledge.
Number of IPS clients served in a recent quarter.
We determined the total number of clients receiving IPS statewide during a recent 3-month period. Learning-community states used data from the centralized portal, augmented by their own tracking systems, to report totals. Many non–learning-community states also collect IPS utilization and outcome data in their own tracking systems. In five states that did not collect such data, we estimated the total number of clients served quarterly within the state by assuming a typical caseload of 74 clients per program and multiplying by the number of programs within the state. The typical caseload was an estimate based on the 2019 sample, calculated as the mean of caseload sizes in the states reporting both the number of IPS clients served and the number of programs.
Per capita rate of IPS clients served and IPS programs.
We obtained the total population in the states based on the 2010 U.S. Census (
https://www.census.gov/quickfacts/). We followed the convention used by many others who estimated the need for mental health services based on an assumption of uniform need across geographic areas (
4,
7,
8). We calculated the rate of IPS clients served quarterly per 100,000 population within each state and did the same for the rate of IPS programs per 100,000 population.
State-Level Implementation Indicators
Training and technical assistance.
Survey questions regarding training and technical assistance included the state’s sponsorship of a technical assistance center providing comprehensive IPS training, one or more IPS trainers, and other types of training (e.g., benefits planning or services for people with dual diagnoses) that were relevant to all employment programs but not specific to IPS. Because relatively few states reported operating technical assistance centers, we collapsed the first and second categories to create a dichotomous measure of state-sponsored IPS training.
Fidelity monitoring.
In states with IPS programs, interviewers asked whether those programs received regular fidelity reviews from independent assessors using a standardized IPS fidelity scale, either the original scale (
9) or the current version (
10). For states without external fidelity reviews, interviewers asked respondents whether local programs conducted self-assessments of fidelity or if IPS programs had no fidelity reviews at all.
Interagency collaboration.
To gauge the level of collaboration between state mental health and VR agencies on employment services for people with serious mental illness, interviewers asked who provided leadership for IPS in the state, with options for both mental health and VR, only mental health, only VR, or neither agency. We provided examples of collaboration, such as the two agencies working together on developing specific employment programs for people with serious mental illness, state agencies blending funding for employment services, and state VR agencies streamlining the eligibility process for people already served by the mental health agency.
Funding.
Respondents indicated which funding sources the state used to pay for IPS services from the following list: state VR, Medicaid, federal grants, state or local general funds, Ticket to Work, and other. The number of funding sources was calculated based on the first four funding sources, because the final two were reported rarely.
Data Analysis
This article focuses exclusively on states with at least one IPS program, excluding 10 states with no reported IPS services. California has IPS programs in both Alameda County and other areas of the state, and for comparisons between jurisdictions within and outside the learning community, we counted Alameda County in the group of learning-community states and counties and also designated the rest of California as a non–learning-community state. We calculated descriptive statistics for the number of reported IPS programs, program enrollment, program outcomes, and four state-level implementation indicators: interagency collaboration, training and technical assistance, regular and independent fidelity monitoring, and funding.
Most statistics in this report are descriptive. We used chi-square analyses to test differences between learning-community states and non–learning-community states on the four program quality indicators. When expected cell frequencies in contingency tables were less than 5, we applied Fisher’s exact test instead of chi-square analysis. We used a t test to evaluate whether the number of funding sources differed between the two groups. We used two-tailed significance levels, with p<0.05.
Results
A total of 41 (80%) of 50 states and the District of Columbia provided IPS services in 2019, including 22 states and two counties in the IPS Learning Community and 18 non–learning-community states. As shown in
Table 1, state leaders reported a total of 857 IPS programs serving an estimated 43,209 clients during a recent 3-month period. The number of IPS programs per state varied from one to 110, with a mean±SD of 20.4±23.5 IPS programs per state. Despite a total population of 20 million fewer residents, learning-community states reported 115 more IPS programs, compared with non–learning-community states (N=486 versus N=371), serving nearly 10,000 more clients than served in non–learning-community states (N=26,522 versus N=16,687).
As shown in
Table 2, learning-community states were far more likely than non–learning-community states to provide implementation support through interagency collaboration between state mental health and VR agencies, external fidelity reviews, and training and technical assistance. Four-fifths of the learning-community states provided all three of these sources of support, compared with only one state outside the learning community.
Most states reported paying for IPS services by using one or more funding sources, including Medicaid, state VR, federal grants, and state or local funds, as shown in
Table 3. States within the learning community were significantly more likely than non–learning-community states to use VR funding and federal grants. On average, learning-community states accessed more of these four funding sources, compared with non–learning-community states (mean=2.83±0.92 versus mean=2.06±1.00; t=–2.62, df=40, p=0.01). In addition, four states in the IPS Learning Community also received modest funding through the Ticket to Work program.
Between 2016 and 2019, four states initiated IPS services and one state ended IPS services, for a net gain of three states. Most growth in IPS services was within states already offering IPS. During the 3-year period, the number of IPS programs increased dramatically nationwide, as shown in
Table 4. The growth was more rapid for learning-community states, compared with non–learning-community states, in the percentage increase in number of IPS programs.
Discussion
Four-fifths of U.S. states currently have IPS programs. Although state and local leaders both within and outside the learning community have successfully expanded IPS services, learning-community states have advanced further in developing state-level infrastructure to support IPS (collaboration between state leaders, external fidelity reviews, and training and technical assistance). Learning-community states also have developed access to more sources of funding. The provision of these implementation support strategies is not surprising; these are exactly the strategies recommended by the IPS Learning-Community (
5). These sources of state implementation support not only enhance the quality of IPS services but also bode well for sustainment and expansion of services (
11). The implementation support strategies are not specific to IPS but apply to other evidence-based practices as well; a recent case study of successful statewide implementation and sustainment of assertive community treatment described a very similar set of strategies (
12).
Over a 3-year period from 2016 to 2019, the number of IPS programs available grew by 64% nationally, increasing access to evidence-based employment services for people with serious mental illness. Learning-community states had greater expansion of IPS services during this 3-year period, even though these states had a smaller total population, compared with non–learning-community states. This study suggests that learning communities may enhance the quality and expansion of evidence-based practices, although controlled research is needed to rigorously assess this. More rapid growth of IPS services in learning-community states likely reflects selection bias; these states joined the learning community because state leaders demonstrated motivation and pledged commitment to expand IPS services.
Learning communities (often called learning collaboratives) are increasingly used to implement evidence-based health care of many kinds in the United States (
13). Many other countries (
14) and the U.S. Office of Disability Employment Policy's Advancing State Policy for Recovery and Employment project (
https://www.dol.gov/agencies/odep/initiatives/aspire) currently use learning communities to enhance the spread of IPS services. We expect the expanded use of learning communities to continue.
Nevertheless, even with the encouraging growth of services, the need for IPS far exceeds its availability. According to SAMHSA’s Uniform Reporting System, 2.8 million adults of working age in the public mental health system in 2019 were not in the labor force (
15). Multiple surveys show that 60% of unemployed adults with mental illness say that they want to work (
16). Many of those expressing a desire to work also would enroll in IPS services if available, although studies show a gap between an expression of the desire to work and actual acceptance of such help (
17,
18). Based on experience with providing IPS services, we assume that at least one-third of this target group would want help to gain employment through IPS. This estimate translates into a need for IPS service capacity for approximately 900,000 people, whereas current IPS programs in the United States serve less than 50,000.
These findings of unmet need apply to other evidence-based practices in mental health. Despite advances in the development of evidence-based practices in mental health (
19), people with serious mental illness seldom receive these effective treatments (
20). Even though many evidence-based practices are now found in most states throughout the United States, the need for services greatly exceeds their availability: less than 5% of clients in the public mental health system have access to these services (
3). Many stakeholder groups, including service recipients, practitioners, family advocates, and researchers, have urged state and federal policy makers to improve access to effective services (
21).
One important resource for both advocates and policy leaders is up-to-date and accurate information about the spread of evidence-based practices. This information is rarely available. For example, virtually nothing is known about evidence-based practices in rural America (
22). National surveys documenting the availability of most evidence-based practices are scarce. The SAMHSA's Uniform Reporting System documents state and national percentages of clients served within the public mental health system receiving each of seven evidence-based practices for adults with serious mental illness (
15), but this broad survey has several methodological limitations and does not provide precise details for any specific practice. Other than the SAMHSA surveys discussed above, few national surveys have addressed evidence-based practices.
The study had some limitations. In most states, a single state agency representative was the sole data source. Although chosen because of their role in monitoring employment services for adults with serious mental illness, survey respondents may have been unaware of IPS programs affiliated with private organizations, early psychosis programs, or services for transition-age young adults. In addition, the survey did not include any programs associated with the Department of Veterans Affairs. The information was collected via telephone interviews and not corroborated by any state agency reports. Outside the learning community, the terminology used to identify IPS is imprecise; respondents also referenced programs by using other labels (e.g., “evidence-based supported employment”). States do not all compile program data in the same fashion. For example, large states face greater challenges than small ones in compiling program-level data.
The most serious study limitation was that some measures were estimates and did not capture quality or amount. For example, the reported funding sources used to pay for IPS services were dichotomous, neither differentiating between large and small amounts of funding nor assessing the percentage of IPS programs within each state accessing a funding source.