Within a 12-month period, a TCN program yielded an annual return of $2.55 for every dollar invested, when considering Medicaid and criminal justice system costs. This return on investment was primarily driven by differences in monthly criminal justice system costs per participant, not Medicaid costs, between the two groups, as TCN participants had reduced probation costs. This result confirms our previous work showing that participation in TCN lowered the odds of reincarceration due to a parole or probation technical violation and confirms there are cost benefits to such an effect. Interestingly, we did not find significant differences in Medicaid costs between the group, despite our previous findings that TCN participation shortens hospital stays and lowers preventable hospitalizations for individuals hospitalized [20]. This could be a function of our sample size.
While our analysis does not establish the mechanism by which TCN lowers probation costs, there are plausible mechanisms to explain the association found. First, TCN patients may better be able to address their health and social needs, allowing them to meet probation requirements, which is consistent with other intervention work focused on this population [35]. Another possible reason is that TCN program providers and community health workers communicate regularly (when given permission by their patients) with criminal justice entities, especially parole and probation officers. These relationships may be important in identifying alternatives to re-incarceration for technical violations, including enrolling patients in substance use treatment when individuals have relapsed to drug use. Such a potential mechanism emphasizes the importance of the TCN intervention being embedded within the healthcare system to have the expertise and resources to fully understand a person’s health context.
To our knowledge, this is the first economic analysis of a US health system–based community health worker intervention for adults leaving the correctional system. Our analysis provides a more thorough estimate of the return on investment compared with estimates derived from a single source (either Medicaid or the criminal justice system) or pre-post trial designs, and further advances the evidence on how to best meet the needs of individuals belonging to a highly marginalized population. To be sure, other costs to the state were unmeasured, including housing, employment, and food access programs, and reflect future work that can better delineate how health system investments may affect state costs for this population. These data add to the limited evidence examining costs of healthcare interventions for individuals released from incarceration on criminal justice expenditures [25,26,27]. In contrast to our findings, a study of 1325 recently released individuals in Australia found that individuals randomized to a low-intensity case management program (the ‘Passports Study’) had higher health, criminal justice, and intervention costs, with an average increase of $1790 AUD per participant over 2 years compared to a control group [36]. The key difference in our study and the ‘Passports Study’ is that the TCN intervention has been found to have no impact on preventable ED visits in Connecticut, while the ‘Passports Study’ led to more healthcare utilization, including ED visits, which are costly [36, 37].
Policy makers and criminal justice and health care organizations interested in making similar investments utilizing Medicaid dollars to impact criminal justice expenditures should interpret this study in the context of four key points. First, the financial value of a TCN program depends on the baseline costs among the targeted patient pool. Given that individuals leaving incarceration generally have high health needs, there is a limited need to target the interventions to contain a high-risk patient pool. The program is offered to those in need of primary care with a chronic health condition and those over 50 years of age, which is a broad and medically complex group. Even when participation includes a broader population, and not focused on a specific disease category or a predetermined “high cost group,” the community health worker intervention returned $2.55 for every dollar invested.
Second, return on investment relies critically on who is making the investment and who is receiving the return. We have presented an economic analysis from the perspective of a state, especially as it is responsible for both Medicaid and the correctional system. It is especially important to note that our findings may not generalize to other states given the unified prison-jail system and Medicaid fee-for-service model in Connecticut, nor does it account for the costs of Medicaid to the federal government. We found that the TCN program was cost neutral for Medicaid (i.e., was not significantly different than care as usual), but we would predict that such findings may be different in systems with Medicaid managed care which may incentivize prevention and value-based care.
Third, this study suggests that TCN programs are beneficial, even from a narrow financial perspective. That said, the financial return on investment underestimates the true societal return because the benefits of the program related to improvements in health, remaining in the community, employment and labor market outcomes, or even spillover effects to families and communities are not assessed. For example, some interventions such as recommended cancer screening or identifying patients with chronic conditions like hypertension through community outreach may not affect healthcare costs over a 12-month time horizon, but may lead to valuable improvements in health (and lower costs, over time) [38].
Last, returns that accrue to the state from investing in TCN programs are important, especially as states have long desired to control criminal justice system and health care costs and have worked to reduce incarcerated populations before and during the COVID-19 pandemic. Before COVID-19, the Pew Center on the States surveyed 41 states and estimated that if these states reduced their recidivism rates by 10%, 635 million dollars would be saved in 1 year [39]. In the context of COVID-19, as states across the US release people from correctional facilities in order to mitigate virus transmission, there is a dire need to ensure those individuals access high quality, effective primary care and do not return to the correctional system, which would lower both healthcare and criminal justice system costs [12]. Criminal justice reforms and decarceration efforts that address the unique health needs of this population may both address the health-harming effects of incarceration, but also minimize additional costs to states.
Limitations
Our findings may not generalize to other states given the Medicaid fee-for-service model in Connecticut. The relatively short 12-month, follow-up period may underestimate the health- related costs associated with medical treatment if its effects of such treatment persist over the longer run, like hepatitis C treatment. Alternatively, the additional benefits accrued beyond the 12-month window might be offset by additional costs if patients require further medical treatment. Next, we chose an index date of TCN enrollment for the TCN group and release from incarceration for the comparison group, which may introduce bias if participants in the TCN group were likely to engage in costlier care prior to enrollment in the TCN program and following their release from incarceration. That said, we found that the mean number of days for TCN participants in this sample to enroll in the program was approximately 48 days following release from incarceration, and that the overwhelming majority of individuals in the comparison group (85%) did not have any ED visits, hospitalizations, or reincarceration experiences during the first 48 days following their release from incarceration—suggesting this concern did not likely influence our findings [21]. Thus, given the index date for TCN group was admission to the program and the comparison group was release from incarceration, it is still plausible that the difference in index date accounts for a difference in the observed costs though our sensitivity analysis suggests that this is unlikely. While we selected a comparison group from a similar urban area as New Haven, regional differences, such as differences in availability of reentry programs or judges’ behaviors, between the two cities may play a role in our findings. While case law of Connecticut applies to cities across the state uniformly, probation practices might be different in New Haven and the comparison community, leading to different rates of probation violations. We used propensity score matching to create a one-to-one comparison of individuals in the TCN and comparison groups, and while we were able to appropriately balance the groups on numerous sociodemographic and health characteristics, it should be noted that propensity score matching has limitations, including an inability to measure and account for unobservable covariates [40, 41]. Because TCN engagement was completely voluntary, individuals who participated may have a different level of health-seeking behaviors than other individuals leaving incarceration and that difference is not captured by the propensity match. A randomized trial would be best for establishing the relationship between TCN participation and reduced probation costs, particularly randomization at the individual-level, rather than the site, in order to better understand whether effects remain across counties. Our sample size was limited to 94 TCN patients, which impacted our confidence intervals and limits our ability for conclusions that this sample is generalizable to the formerly incarcerated population with chronic health conditions. We were unable to access the cost of services provided through the Department of Mental Health and Addiction Services, which could be potentially expensive, and our results should be interpreted in light of this limitation given that participants in both the TCN and comparison groups could have utilized such services. Our data is from 2013 to 2017, and costs may have shifted since then, but we adjusted our data for inflation costs to 2015.
Policy implications
Our study suggests that investment in enhanced primary care may be advantageous to state budgets as they attempt to manage the rising costs of correctional systems, while providing evidence-based care to individuals released from incarceration [17, 20, 21]. Policies that provide funding to support community health centers in implementing a TCN program through hiring formerly incarcerated community health workers into healthcare teams could be beneficial by decreasing criminal justice costs and improving health outcomes among individuals released from incarceration. Extending Medicaid funding to compensate for the community health worker’s time and promoting enhanced care management programs that target individuals recently released from incarceration could incentivize adaptation of TCN programs in community health centers. Lastly, this study has demonstrated the benefit of studying the costs of primary care-based programs beyond the health care system. Our findings were dependent on our ability to link Medicaid claims to criminal justice system costs, which is typically unfeasible. The inability to link such data may lead to biased estimates of interventions for individuals recently released from incarceration. States should invest in data linkage systems that facilitate cost analysis across even more systems (i.e. federal criminal justice system, social service agencies) to allow for quantifying benefits of intervention programs from a larger societal perspective. This would enable studying the collateral benefits (i.e. employment, food access) and changes to both individual and family well-being.
Future research
Future studies exploring the long-term cost impacts of primary care-based programs targeted for people released from corrections are critical to health system and criminal justice system reforms. More attention should be given to investigating the mechanisms by which programs impact costs, such as how TCN participation reduced probation costs in this study. Future analyses should consider benefits specific to certain health conditions and treatment options, including but not limited to substance use disorder treatment or preventative treatment such as cancer screening. These studies will provide evidence for how to mitigate the high costs of the criminal justice system and health systems.