Enrollment and Service Use Patterns Among Persons With Severe Mental Illness Receiving Expedited Medicaid on Release From State Prisons, County Jails, and Psychiatric Hospitals

At any given time, as many as 50,000 persons with severe mental illness are in our nation’s state hospitals, more than 100,000 persons with severe mental illness are in jails, and more than 250,000 persons with severe mental illness are in prisons (1,2). Most of these individuals will return to their communities, which has significant implications for mental health and criminal justice practice and policy given evidence that persons with mental illness who are released from these institutional settings, particularly criminal justice settings, have difficulty accessing services, housing, and social supports and have high recidivism rates (3–10).

For persons with mental illness, a lack of health insurance is viewed as a significant barrier to accessing community behavioral health care in a timely manner on release from institutional settings (11–13). A study of released prisoners in Texas and Ohio, for example, found that only 8% of men and 21% of women were enrolled in Medicaid eight to ten months after release, whereas 68% and 58%, respectively, were without any health insurance (14). These findings are of particular concern given prior research in Washington and Florida that suggests that having Medicaid on release from jail enables persons with severe mental illness to access mental health services more quickly and more frequently, compared with jail inmates with mental illness who do not have Medicaid on release (15,16).

Programs and policies to quickly facilitate Medicaid enrollment for persons with severe mental illness who are leaving institutional settings have become increasingly common in recent years (17,18). A recent study of a prison prerelease pilot program to expedite Medicaid enrollment in Oklahoma reported increases in Medicaid enrollment and mental health service use after release (19,20). Further interest in these policies and programs has been stimulated by the Affordable Care Act (ACA) and its potential to increase access to general medical and behavioral health services for persons in need, especially for those in criminal justice settings (21,22). This is particularly important given the high mortality rates experienced by prison inmates immediately after release to the community (23).

More information is needed about whether policies to expedite enrollment in Medicaid are associated with greater Medicaid enrollment and increased service utilization for persons with severe mental illness and whether these patterns vary across institutional settings. In this study, we addressed these evidence gaps by comparing patterns of coverage and outpatient mental health service use of persons with severe mental illness who were referred for expedited Medicaid enrollment from state prisons, county jails, and psychiatric hospitals in Washington State during 2006, the first year the expedited practice was authorized. Specifically, we compared the characteristics of persons who were approved or denied for Medicaid enrollment within and across settings. We also compared patterns of postrelease Medicaid enrollment and outpatient mental health service use for these two groups. Because the study did not otherwise control for selection bias into the three settings or into referral for expedited Medicaid, results should not be interpreted as causal estimates.

Methods

Study Design and Context

A retrospective cohort study design with linked administrative data was used to examine approval versus denial of expedited Medicaid for persons with severe mental illness who were referred for expedited Medicaid from state prisons, county jails, and psychiatric hospitals in Washington State during 2006. Also examined were Medicaid enrollment status and outpatient mental health service utilization at 30, 60, and 90 days after release. The study was occasioned by the enactment of state legislation in January 2006 (House Bill 1290) that authorized expedited Medicaid enrollment for persons with severe mental illness on release from institutional settings.

To prioritize the processing of expedited Medicaid referrals, the legislature appropriated funds for 14 full-time-equivalent staff positions that were assigned to local community service offices around the state. The expedited process involved two steps. First, prior to release, referrals were initiated by facility staff who identified eligible individuals and invited them to volunteer for expedited Medicaid enrollment, assisted those who agreed in completing an application for Medicaid benefits, and then referred the application to community service offices prior to the participant’s release date. Second, after release, approval for Medicaid enrollment was activated at a local community service office, where applicants had to appear for an application review and a means test (income limits).

Sample

We used administrative data obtained from Washington State agencies (see below) to identify individuals who were referred for expedited Medicaid and who had a diagnosis of schizophrenia or other psychotic disorder, bipolar disorder, or major depression. Individuals who were likely to be eligible for referral and processing for expedited Medicaid typically met one of two criteria: enrolled in Medicaid or state-funded general assistance coverage in the five years prior to an institutional stay, or evidence of need for mental health services in the prior five years as demonstrated by a mental illness diagnosis in medical claims, receipt of mental health services, or receipt of psychiatric medications.

Data and Measures

We obtained demographic, clinical, and service utilization data from the Research and Data Analysis Division of the Washington State Department of Social and Health Services (DSHS), which linked utilization information from Medicaid, psychiatric hospitals, and outpatient mental health services for the period 2001–2010. This file was linked to Department of Corrections data for persons released from state prisons during the same period. Demographic and diagnostic variables included sex (0, male; 1, female); race (0, white; 1, black or other); age in years; a hierarchical diagnostic category of schizophrenia, bipolar disorder, or depression; an indicator of co-occurring substance abuse treatment; and previous DSHS eligibility status (that is, aged, blind, or disabled [ABD]; state general assistance–unemployable [GA-U]; or other). Diagnoses were mutually exclusive and were obtained from Department of Corrections records or community mental health claims records. With respect to psychiatric hospitals, most referrals for expedited Medicaid came from Washington State’s two large state psychiatric hospitals (84%), with the remainder from two local psychiatric hospitals and several residential settings (16%).

We created dichotomous indicators of administrative approval for expedited Medicaid, of Medicaid enrollment, and of outpatient mental health service use at 30, 60, and 90 days following the date of release. Outpatient mental health services included intake assessment, medication management, case management, individual and group counseling, and other services.

Data Analysis

Bivariate and multivariate analyses were conducted to model differences between cohorts in the demographic and clinical characteristics of persons approved for expedited Medicaid enrollment and those denied, as well as differences in their Medicaid enrollment and outpatient mental health service use in the 90 days after the index release, with analyses controlling for demographic characteristics. All analyses were carried out in Stata, version 13 (24). Chi square tests and t tests were used to examine bivariate relationships between approval status (1, yes; 0, no) and demographic, clinical, and service use variables. Logistic regression was used to examine factors related to approval for expedited Medicaid, Medicaid enrollment, and use of mental health services. We also examined the interaction between approval status and the facility type to capture nonlinearities in the enrollment and service use models. Average marginal effects are reported from these models overall; differences by approval status within each facility type (jail, prison, and hospital) were calculated by using the method of recycled predictions (25) and are reported in separate tables.

Results

Demographic and clinical characteristics of the three cohorts of individuals referred from state prisons, county jails, and psychiatric hospitals are shown in Table 1. Among 1,248 referrals across the three settings, 80% (N=993) were approved for expedited Medicaid and 20% (N=255) were denied. The approval rate was highest for psychiatric hospitals (91%), followed by state prisons (83%) and county jails (66%).

Bivariate results suggest that across the three institutional settings, individuals who were approved for expedited Medicaid were more likely than those who were not approved to be white (p<.05) and female (p<.01), to have a diagnosis of schizophrenia (p<.001) or bipolar disorder (p<.05), and to be more likely to have had prior Medicaid benefits (that is, ABD) in Washington State (p<.001); however, individuals who were denied expedited Medicaid were more likely to have a diagnosis of depression (p<.001) and were more likely to have a co-occurring substance use disorder (p<.001) (Table 1). Results from each institutional setting were generally consistent with these findings, although a few exceptions and some differences in characteristics, such as substantially higher rates of substance use disorder diagnoses among those released from jails, were noted.

Table 2, which is discussed in more detail below, shows Medicaid approval rates and Medicaid enrollment and mental health service use at 30, 60, and 90 days postrelease for the total sample and for each institutional setting.

As shown in Table 3, multivariate regression results suggest that the probability of being approved for expedited Medicaid was 13.6 percentage points lower for individuals released from jails compared with those released from psychiatric hospitals (p<.001). No difference in the probability of approval was noted between those released from prisons and from hospitals. The association of expedited Medicaid approval with race, gender, and age remained significant in the multivariate analysis. The probability of approval for females was 7.0 percentage points higher than for males (p<.001); for persons of color, the probability of approval was 5.5 percentage points lower than for whites (p<.01); and younger individuals had a lower probability of approval than older individuals (p<.01). Also, the probability of approval for individuals with schizophrenia was 18.4 percentage points higher than for those with depression (p<.001), and the probability of approval for those with a diagnosis of schizophrenia was 9.8 percentage points higher than for those with a diagnosis of bipolar disorder (p<.001).

Medicaid Enrollment

We examined whether approval for expedited Medicaid was associated with quicker Medicaid enrollment. As shown by the unadjusted rates in Table 2, for all institutional settings, 91% of persons approved for expedited Medicaid were enrolled in Medicaid in the first 30 days after release, compared with 67% of those denied (p<.001). In the first 30 days after release from prison, 89% of those approved were enrolled, compared with 56% of those denied (p<.001). In the first 30 days after release from jail, 83% of those approved were enrolled, compared with 62% of those denied (p<.001).

As shown in Table 3, results of the multivariate regression analysis, which controlled for all other model covariates, suggest that approval of expedited Medicaid across settings was associated with a probability of Medicaid enrollment at 90 days that was nearly 15 percentage points higher than when expedited Medicaid was denied (p<.001). When the analysis controlled for approval status, persons released from criminal justice settings were substantially less likely than those released from psychiatric hospitals to be enrolled in Medicaid by 90 days postrelease; the enrollment rate after prison release was 11.0 percentage points lower than after hospital release, and the rate after release from jails was 20.1 percentage points lower (p<.01). Compared with Medicaid enrollment rates for persons denied expedited Medicaid who were released from each institutional setting, the difference in enrollment rates for persons approved ranged from 10.4 percentage points for psychiatric hospitals, to 16.8 percentage points for prisons, to 21.6 percentage points for jails (Table 4). That is, if persons released from jail had not been offered expedited Medicaid, they would have been much less likely to enroll in Medicaid compared with those released from other institutions.

Mental Health Service Use

We examined whether approval for expedited Medicaid was associated with quicker use of outpatient services. The unadjusted outpatient mental health service utilization rates at 30, 60, and 90 days postrelease for the entire sample and by institutional setting are shown in Table 2. For all institutional settings combined, in the first 30 days after release, 71% of persons who were approved for expedited Medicaid received at least one outpatient mental health service, compared with 39% of those denied (p<.001). The 30-day service use rates among those released from prison were generally low, and there was no advantage associated with approval for expedited Medicaid. However, for persons released from jails, the 30-day service use rate was 59% for those approved, compared with 38% for those denied (p<.001). For persons released from psychiatric hospitals, the 30-day service use rate was 95%, compared with 54% for those denied (p<.001).

As shown in Table 3, results of the multivariate regression, which controlled for demographic characteristics and diagnoses, suggest that the probability of use of any mental health service at 90 days was 13.1 percentage points higher for individuals approved for expedited Medicaid than for those denied (p<.01). Controlling for approval status, we found that compared with persons released from psychiatric hospitals, those released from jails and prisons were substantially less likely to use mental health services within 90 days; the probability among those released from jails was 22.5 percentage points lower than for those released from hospitals, and the probability among those released from prisons was 24.7 percentage points lower (Table 3). No difference was found by approval status in the probability of service use at 90 days among those released from prison (Table 4). However, among those released from jails, the probability of mental health service use at 90 days was 14.6 percentage points higher for those approved than for those denied. For persons released from hospitals, the probability was 23.2 percentage points higher for those approved than for those denied (p<.01).

Discussion

This study presents new evidence about the utility of expediting Medicaid benefits for persons with severe mental illness at the point of release from various institutional settings. The findings underscore a number of policy-relevant issues and help to grow the evidence base for policy interventions that have the potential to improve the community reentry of persons with severe mental illness.

First, during the first year of Washington State’s new policy, 59% of persons with severe mental illness who were referred for expedited Medicaid and 54% of those who were approved came from the criminal justice system. This high percentage is consistent with many reports documenting the large numbers of people with severe mental illness who end up in jails and prisons throughout the United States. In the context of Medicaid policies, this finding also highlights the large numbers of offenders with severe mental illness in jails and prisons who are Medicaid eligible and willing to participate in a voluntary enrollment effort prior to release.

Second, our findings suggest that there may be obstacles to identification and processing of Medicaid-eligible individuals in the institutional settings observed. The highest approval rates were for those released from psychiatric hospitals (91%) and state prisons (83%); rates for those released from county jails were lower (66%). Psychiatric hospitals (mostly state hospitals in this study) and prisons are both long-stay settings that allow more time to assemble and document applications than jails, which are short-stay facilities frequently cut off from Medicaid-required clinical assessments and where release dates are often unpredictable (20). Many more people with severe mental illness flow through county jails each year than through state prisons, which means that although jails are excellent screening and identification outposts for Medicaid enrollment, extra steps may be needed to assist detainees with severe mental illness to complete Medicaid applications in a timely manner (17,18).

Third, and most important, our findings show that expediting Medicaid benefits was associated with greater and timelier use of outpatient mental health services, which can facilitate successful transitions to the community. Here again, however, patterns of service use varied across institutional settings. Not surprisingly, the highest mental health service use rate in the critical first 30 days after release was for those who were released from psychiatric hospitals and who were approved for expedited Medicaid (95%); 30-day rates were much lower for those released from jails and approved (59%) and especially for those released from prisons and approved (36%). Psychiatric hospital patients are already in the mental health system so to speak. They are often admitted for inpatient care through their contacts with outpatient providers, and on hospital discharge, they return to outpatient care. However, for persons released from jails and prisons, there typically is a distance from and disconnect with the mental health system that has to be bridged to access services.

Of interest, persons released from jails and prisons who were denied expedited Medicaid rapidly increased their mental health service use by 60 and 90 days postrelease, which also corresponded with increases in their Medicaid enrollment (Table 2). Therefore, even though many individuals were denied expedited Medicaid before release, they were soon able to obtain Medicaid benefits in the community after release. This finding perhaps reflects the practice among many community providers to prioritize enrolling persons with severe mental illness in Medicaid quickly so that services provided are billable. Nonetheless, our findings show that expedited Medicaid approval was associated with quicker outpatient mental health service use after release.

Do the findings of this study have any implications for the ACA? The immediate answer is “not directly,” even though ACA provisions are resulting in Medicaid enrollment for many persons not previously covered, such as single males in contact with the criminal justice system (21,26). This study focused on persons with severe mental illness who were likely eligible for traditional Medicaid, whereas the ACA focuses on expanding Medicaid to individuals who do not qualify for traditional Medicaid. Nonetheless, the findings about relatively high rates of outpatient mental health service use among those approved for expedited Medicaid after release from prisons and jails (66%−68% by 90 days postrelease) offer evidence to support the ACA’s assumption that access to health insurance can lead to greater use of behavioral health care.

The study had several limitations. First and foremost, our observational cohort design lacked a true control group, and we are thus unable to make causal attributions about expedited Medicaid. The differences reported are a composite of program effects and selection artifacts, and we cannot disentangle the two. Consequently, we are unable to say that the positive effects observed are uniquely due to expedited Medicaid approval. Also, we were able to examine outcomes only for a three-month postrelease period. Although it is reasonable to expect that the advantages of expedited processing would show up in the months immediately following release, a longer time interval would help to determine whether there are lags in service use. Although we restricted our referral sample to persons with severe mental illness (as defined by diagnoses of schizophrenia, bipolar disorder, or depression), our comparison group of individuals who were referred for but denied expedited Medicaid could have been different in unobserved ways from the group of those who were approved. This is a limitation of using administrative data that do not contain independent measures of functioning, information about social supports, or other personal characteristics.

However, there are some compensatory benefits. Administrative data allowed us to follow all of the referred individuals across agency contacts and time periods with less concern for sample attrition and missing data, which are issues that often compromise prospective studies. Moreover, a study comparing referrals across three institutional settings on a statewide basis with use of consistent methods is a novel contribution to the literature.

Conclusions

Each year large numbers of individuals with severe mental illness are released from state prisons, county jails, and psychiatric hospitals. Many of these individuals can be voluntarily recruited to enroll in Medicaid on an expedited basis, and many of those who are approved go on to receive outpatient mental health services. Expediting Medicaid can be an important part of a continuum of practice and policy interventions designed to improve the community reentry outcomes of persons with severe mental illness who are released from institutional settings.