Health Care and Behavioral Service Use by Medicaid-Enrolled Adults After Release From Incarceration

We used longitudinal data to describe Medicaid enrollment and utilization among returning citizens who enrolled within 120 days of release from state prison and to characterize their health care utilization within the 120 days after receiving coverage.

We combined data from IDOC and the Family and Social Services Administration, which oversees the Office of Medicaid Policy and Planning and the Division of Mental Health and Addiction (DMHA). First, we used IDOC data to identify data from adults who were incarcerated and released from state prisons between February 2015 and December 2018. Data were excluded for those who died while incarcerated, were age >64 years when released, absconded while in custody, or were transferred from IDOC jurisdiction. Information about incarceration, release, demographic characteristics, mental health needs, and substance use history was also obtained from the IDOC data. Mental health need and substance use history were identified through an assessment by IDOC medical professionals with a determination on the likelihood of the need for mental health services. The assessment used an ordinal scale, based on an assessment conducted during intake, ranging from A, indicating no known history of mental health need or substance use disorder and no excess likelihood of service needs, to F, indicating need for a high level of service use. We aggregated this variable into a binary indicator of A versus all other levels to specify a subpopulation of individuals at elevated risk for needing services upon release.

Release and demographic data from IDOC were linked to Medicaid enrollment and claims data by Indiana’s Management Performance Hub, as previously described (6). Medicaid data included effective coverage dates and all paid claims for covered health services. The combination of IDOC release and Medicaid enrollment data was used to identify adults released from Indiana state prisons during the study period who also enrolled in Medicaid within 120 days of release.

Additional DMHA data provided information about care at the six state psychiatric hospitals and 23 community mental health centers across Indiana. These data also included information regarding adults who participated in the Recovery Works program, a safety-net program designed to provide behavioral health services immediately upon release for returning citizens who lack health insurance coverage.

Our main measures of health care utilization were the timing and occurrence of four types of health care utilization: general emergency department (ED) encounters and hospitalizations, general outpatient care, ED encounters or hospitalizations related to behavioral health needs, and outpatient or residential treatment for behavioral health needs (hereafter referred to as subacute care). We calculated the frequencies of the first service type utilized and any service type use within the first 120 days after Medicaid coverage. All individuals in the sample were observed for 120 continuous days after enrolling in Medicaid.

Encounters were classified on the basis of Current Procedural Terminology codes and ICD-9 and ICD-10 diagnosis codes. We classified behavioral health services to include services for mental health needs and substance use disorders and diagnoses by using the Clinical Classification Software tool developed by the Health Care Cost and Utilization Project. (Additional details are available in an online supplement to this report.)

A binary variable of early versus late Medicaid coverage was created on the basis of the number of days from release date to the effective Medicaid coverage date. Enrollment between 0 and 30 days after release was considered early, and enrollment between 31 and 120 days was considered late.

Individual demographic characteristics and information about the incarceration episode included age group, race-ethnicity, gender, length of incarceration, release type, number of times the individual had been incarcerated, and utilization of the Recovery Works program. The length of incarceration was calculated as the release date minus the intake date for the incarceration and was categorized into five groups (0–6 months, >6–12 months, >12 months to 2 years, >2–5 years, and >5 years). Release types were denoted as releases to parole, probation, complete discharge, or community transition program.

We presented demographic and incarceration characteristics and utilization measures by early versus late coverage. Bivariate comparisons were conducted with chi-square tests. The study was approved by the institutional review board at Indiana University.

From February 2015 to December 2018, 16,447 releases from Indiana state prison (15,929 unique adults) resulted in Medicaid enrollment within 120 days, representing 39.5% of all releases (N=41,604). Of the releases with Medicaid enrollment, 56.9% (N=9,361) resulted in enrollment within 30 days of release. Most release episodes were for individuals who were male (81.1%), White (71.4%), and incarcerated for the first time (74.9%). Within each demographic or other category, most individuals were ages 26–35 years (37.2%), incarcerated for between 12 months and 2 years (27.3%), and released on parole (50.0%). Additionally, 26.2% of individuals had a documented history of mental health needs or substance use disorder. Additional demographic and other characteristics of the individuals with early or late Medicaid enrollment are shown in Table 1.

Compared with individuals who enrolled after 30 days, those who enrolled within 30 days were more frequently male (81.8% vs. 80.2%, p=0.014), non-Hispanic White (72.9% vs. 69.5%, p<0.001), older (i.e., ages 46–64 years, 25.7% vs. 20.5%, p<0.001), had longer incarcerations (>2 years, 35.7% vs. 28.5%, p<0.001), and were released on parole (52.9% vs. 46.3%, p<0.001) or probation (32.1% vs. 28.9%, p<0.001). Overall, 47.5% of individuals who enrolled in Medicaid after release from prison had an ED encounter and 39.9% had a subacute behavioral health encounter within 120 days of initiating coverage. In the entire population studied, 80.7% had utilized at least one health care service appearing in Medicaid claims data within the first 120 days of coverage (Table 1). Those who enrolled early were less likely than those who enrolled later to have an ED encounter (46.2% vs. 49.2%, p<0.001) or an acute behavioral health encounter (18.8% vs. 20.2%, p=0.023) but were more likely to have a subacute behavioral health encounter (41.6% vs. 37.7%, p<0.001) while enrolled in Medicaid. Similarly, those who enrolled early were less likely to have either an all-cause ED encounter (36.7%, vs. 41.2%, p<0.001) or an acute behavioral health encounter (9.7% vs. 10.9%, p=0.016) as their first health care service type. Several of the differences between early and late enrollments were driven by individuals released in or before 2016, when the full combination of policy changes was not yet in effect.

We observed frequent care utilization, with 47.5% of individuals who enrolled in Medicaid after release from prison having an ED encounter and 39.9% having a subacute behavioral health encounter within 120 days of initiating coverage. Furthermore, service use was more frequent among those who enrolled late than among those who enrolled early, and the types of care received differed between the two groups.

Health care needs are high among returning citizens (8, 9, 12–15). Consequently, Indiana and other states have policies to improve access to health insurance coverage for this population (2, 10, 11, 16). These policies are hypothesized to broadly improve access to health care, which may help avert poor outcomes and unnecessary utilization. Many of the differences in utilization we observed appeared to be related to the 2015 policy changes in Indiana. It has been shown that Medicaid coverage is obtained at higher rates for those with greater need, such as those with high blood pressure, diabetes, or asthma (17). High ED utilization in our study may reflect a similar phenomenon, where enrollment reflects a need for care. Conversely, an acute health care encounter within the health care facility may have triggered application for coverage. Such patterns may result in suboptimal management of general medical or mental health conditions. Questions remain regarding how to strengthen the utilization of primary and preventive services to ensure the optimal management of health conditions and optimal provision of care. Future research may be needed to characterize the reasons for use of services and the proportion of ED encounters that could be better managed in other settings.

Those who had enrolled early in Medicaid used subacute behavioral health services more frequently than acute behavioral health services as their first encounter within the first 120 days of coverage. This subgroup also more frequently had a history of behavioral health concerns. These patterns suggest that those with behavioral health needs are gaining coverage early and are utilizing lower-acuity care. Although we cannot assess appropriateness of the setting of care, in many cases, subacute treatment may prevent costlier and more severe ED encounters for behavioral health needs. Further research is needed to better characterize health needs and services received among this population.

Our study had several limitations. First, we limited our sample to those who became enrolled in Medicaid within 120 days of release. At most, this represented 45% of returning citizens during this time (6). Second, our results may not be generalizable to other states, given Indiana’s 1115 Medicaid waiver. Additionally, we lacked geographic indicators to assess within-state heterogeneity. Third, we were unable to distinguish whether individuals enrolled on their own or with assistance from IDOC or health care staff because of a need for services. Fourth, early versus late coverage was based on effective dates, which may have included retrospective enrollment, underestimating time to coverage. Fifth, individuals may have received unobserved care between the time of release and receiving Medicaid coverage. We could observe their utilization of care in Recovery Works and community mental health centers during this time but not of other types of care. However, those seeking acute care would likely have been presumptively enrolled in Medicaid, if eligible.

Stable health insurance coverage is a key part of transitioning from prison to the community, promoting access to care for adults involved in the criminal legal system. Indiana has enacted policies and procedures to expedite Medicaid enrollment among returning citizens and to provide such coverage; however, we observed that ED visits and other acute service utilization were frequent in this population. Additional research is warranted to determine opportunities for such care to be provided in the primary care setting. The extent to which more proactive enrollment in health coverage can avert ED visits remains an important question.