Health economic analyses of the Justice Community Opioid Innovation Network (JCOIN)

Abstract

The Justice Community Opioid Innovation Network (JCOIN) will generate real-world evidence to address the unique needs of people with opioid use disorder (OUD) in justice settings. Evidence regarding the economic value of OUD interventions in justice populations is limited. Moreover, the variation in economic study designs is a barrier to defining specific interventions as broadly cost-effective. The JCOIN Health Economics Analytic Team (HEAT) has worked closely with the Measures Committee to incorporate common economic measures and instruments across JCOIN studies, which will: a) ensure rigorous economic evaluations within each trial; b) enhance comparability of findings across studies; and c) allow for cross-study analyses of trials with similar designs/settings (e.g., pre-reentry MOUD), to assess questions beyond the scope of a single study, while controlling for and evaluating the effect of intervention-, organizational-, and population-level characteristics. We describe shared trial characteristics relevant to the economic evaluations, and discuss potential cross-study economic analyses.

1. Background

Prior work has shown that evidence-based treatment for opioid use disorder (OUD), most specifically, pharmacotherapy (methadone, buprenorphine, naltrexone), can help to improve health-related quality of life, while reducing the negative externalities that various stakeholders incur. Medications for OUD (MOUD) have been associated with reductions in opioid overdose mortality (Sordo et al., 2017), high-cost health care utilization (e.g., emergency department [ED] and inpatient visits), and criminal activity, as well as increased school and workplace productivity (Murphy & Polsky, 2016). However, variation in research designs, perspectives, and economic effectiveness measures is a barrier to defining specific interventions as broadly cost-effective (Murphy & Polsky, 2016). Existing knowledge is severely limited with regard not only to the comparative effectiveness of MOUD and linkage to MOUD strategies in justice settings, including peer-support strategies, but also their relative economic value, making the aforementioned issues even more prominent (Blanco & Volkow, 2019; Eddie et al., 2019; National Academies of Sciences & Medicine, 2019).

Despite high rates of OUD among justice populations, few receive MOUD (Krawczyk, Picher, Feder, & Saloner, 2017). The Justice Community Opioid Innovation Network (JCOIN), a key component of the National Institutes of Health’s Helping to End Addiction Long-term (HEAL) Initiative, intends to develop new insights into how best to manage OUD among justice populations (National Institutes of Health, 2020). JCOIN consists of 11 hubs, each leading a clinical trial that tests the effectiveness of initiating MOUD in justice settings or of improving linkage to MOUD and other services, and 2 cross-hub resource centers. The JCOIN funding announcement highlighted the importance of integrating rigorous economic evaluations, specifically cost-effectiveness analyses (CEAs) (National institutes of Health, 2018). Of the 11 clinical trials, 10 are planning economic evaluations. The JCOIN Health Economics Analytic Team (HEAT) comprises economic investigators from each relevant hub, and the JCOIN Methodology and Advanced Analytics Resource Center. The HEAT has worked closely with the JCOIN Measures Committee to incorporate common economic measures and instruments across studies, which will: a) ensure rigorous economic evaluations within each trial; b) enhance comparability of findings across studies; and c) allow for cross-study analyses of trials with similar designs/settings (e.g., pre-reentry MOUD), to assess questions beyond the scope of a single study, while controlling for and evaluating the effect of varying interventional, organizational, and population-level characteristics across studies, such as toxicity of the drug supply, access to MOUD, and policies pertaining to the coverage of MOUD. Herein, we describe shared trial characteristics relevant to the economic evaluations, and discuss potential cross-study economic analyses.

2. Methods & design

2.1. Studies with planned economic analyses

Within the hubs planning an economic evaluation there are shared design characteristics that will facilitate cross-study analyses (Table 1). Four studies are testing models of linkage to MOUD and other services upon reentry. The Chestnut Health Systems hub is testing two versions of the Recovery Management Checkup (RMC) intervention to support linkage to MOUD and improve retention, relative to standard reentry processes (i.e., passive referral to treatment). The intervention conditions are RMC with quarterly checkups and RMC-Adaptive with checkups based on participants’ progress (Scott et al., 2021, this issue). The Yale University hub is testing the effectiveness of the Transitions Clinic Network: Post-incarceration Addiction Treatment, Healthcare, and Social Support intervention, which provides wrap-around social and OUD treatment services, including MOUD, plus a primary care team, compared to primary care alone (Howell et al., 2021, this issue). The University of Chicago hub is evaluating the effectiveness of a hub-and-spoke case-management model, with peer-recovery coaches and overdose education and naloxone distribution (OEND), for linking persons to MOUD and other services upon reentry, compared to OEND alone (Pho et al, 2021, this issue). The University of Kentucky hub is testing pretreatment telehealth, with and without peer navigation, compared to services as usual, while increasing the utilization of MOUD among incarcerated women (Staton et al., 2021, this issue).

Table 1.

JCOIN Hub Descriptive Characteristics

Hub	Study Title	Setting	Intervention	Control	Economic Evaluation Type	Stakeholder Cost Perspectives	Intervention Period	Key Data Collection Points
Baystate Medical Center	Massachusetts JCOIN	Jails	MOUDa in jail + Community-Based MOUD	Treatment as Usual	CEAb	Healthcare Sector; State Policymaker	12 months pre- and post-reentry	NA - Administrative Data
Brown University	Using Implementation Interventions and Peer Recovery Support to Improve Opioid Treatment Outcomes in Community Supervision	Community Supervision	Local Change Teamc + Peer Support Specialist	Local Change Team	Cost-Offset Analysis	Healthcare Sector; State Policymaker; Societal	12 months	BLd, 3, 6, &12 months
Chestnut Health Systems	Improving Retention across the OUD Service Cascade upon Re-Entry from Jail using Recovery Management Checkups (RMC)	Jails	RMCe: quarterly follow up	RMC - Adaptive: frequency and content based on progress	CEA	Healthcare Sector; State Policymaker; Societal	24 months	BL, 3, 6, 9, 12, 15, 18, 21, & 24 months
Friends Research Institute	A Comparative Effectiveness Trial of Extended Release Naltrexone versus Extended-Release Buprenorphine with Individuals Leaving Jail	Jails	In-Jail XR-BUPf (CAM2038/Brixadi) + Community-Based MOUD	In-Jail XR-NTXg + Community-Based MOUD	CEA	Healthcare Sector; State Policymaker; Societal	6 months	BL, 1–6, 7, 12, & 18 months
New York State Psychiatric Institute / Columbia University	Facilitating Opioid Care Connections: System Level Strategies to Improve Use of MAT and Movement Through the Opioid Care Cascade for Defendants in a new Opioid Court System	Drug Courts	Opioid Court Modelh	Usual Drug Court Model	CEA	State Policymaker	12 months implementation; 12 months sustainment	NA - Administrative Data
New York University School of Medicine	Long-Acting Buprenorphine vs. Naltrexone Opioid Treatments in CJS-Involved Adults (EXIT-CJS)	Jails	In-Jail XR-BUP (Sublocade) + Community-Based MOUD	In-Jail XR-NTX + Community-Based MOUD	CEA	Healthcare Sector; State Policymaker; Societal	6 months	BL, 1–6, 7, & 12 months
Texas Christian University	JCOIN: TCU	Community Supervision	Vertical Integration of Opioid Treatment Linkage Modeli	Horizontal Integration of Opioid Treatment Linkage Model	CEA	Healthcare Sector; State Policymaker; Societal	12 months	BL, 6, & 12 months
University of Chicago	Reducing Opioid Mortality in Illinois	Jails/Prisons	Case Management & Peer Recovery Support for Linkage to OUD Services+OENDj	Services as Usual + OEND	CEA	Healthcare Sector; State Policymaker; Societal	12 months	BL, 3, 6, &12 months
University of Kentucky	Kentucky Women’s JCOIN	Jails	MOUD & Pre-Treatment Telehealth + Peer Navigator	MOUD & PreTreatment Telehealth	CEA	Healthcare Sector; State Policymaker; Societal	12 months	BL, 3, 6, & 12 months
Yale University	Transitions Clinic Network: Post Incarceration Addiction Treatment, Healthcare, and Social Support	Jails	Transitions Clinic Network Interventionk	Standard Primary Care	CEA	Healthcare Sector; State Policymaker; Societal	12 months	1, 3, 6, 9, &12 months

Notes:

^aMedications for opioid use disorder.

^bCost-effectiveness analysis.

^cLocal Change Team = justice and community service providers to develop and implement interorganizational linkage strategies to improve uptake of MOUD.

^dBL=baseline.

^eRecovery Management Checkup = provides regular check-ups to support linkage to a designated MOUD provider upon release from jail, and improve retention.

^fExtended-release buprenorphine.

^gExtended-release naltrexone.

^hOpioid court model entails intervening the day of the individual’s arrest to link them to MOUD and other treatment services, while under daily judicial supervision.

ⁱOpioid Treatment Linkage Model = designed to link individuals under community supervision with community-based MOUD and other services.

^jOEND Opioid education and naloxone distribution

^kTransitions Clinic Network = a primary care team and support for social needs, including housing, food access, and social support.

Three studies are examining the effectiveness of MOUD provided pre-reentry, and continued in the community. The Friends Research Institute (Gordon et al., 2021, this issue) and the New York University (Waddell et al, 2021, this issue) hubs are conducting head-to-head comparative-effectiveness trials of two different MOUD; extended-release naltrexone (XR-NTX) versus extended-release buprenorphine (XR-BUP, CAM2038/Brixadi), and XR-NTX versus XR-BUP (Sublocade), respectively. The Baystate Medical Center hub is conducting an evaluation of the “Chapter 208” Massachusetts legislation that funded a 4-year pilot program offering all three MOUD in select jails, plus continued community-based treatment (Evans et al., 2021, this issue).

Two studies are testing models of linkage to MOUD and other services, within community supervision settings (e.g., probation or parole). The Brown University hub is testing the independent and combined effectiveness of interorganizational and person-level linkage strategies in community supervision settings (Martin et al., 2021, this issue). The organization-level component uses local teams of service providers in both justice and community settings to develop linkage strategies, while the person-level component will test a peer-support specialist intervention, compared to a non-peer-support arm. The Texas Christian University hub is comparing clinical- and agency-level interventions to support linkage to care for adults in the justice system, as well as agencies’ abilities to implement strategic and adaptive processes (Knight et al., 2021, this issue).

Finally, the New York State Psychiatric Institute/Columbia University hub is evaluating different implementation strategies of New York State’s opioid court model (Elkington et al., 2021, this issue). The opioid court is an alternative to typical drug court models, and entails intervening the day of the individual’s arrest to link them to MOUD and other treatment services, while under daily judicial supervision.

2.2. Measures

2.2.1. Stakeholder perspectives

Stakeholder perspectives define the parameters of an economic evaluation with regard to resources that should be valued, and the value that is placed on them. The perspectives proposed (Table 1), and the resources to be included vary across studies due to different data generating mechanisms (e.g., administrative vs. self-report). As per current recommendations (Neumann, Sanders, Russell, Siegel, & Ganiats, 2017), where possible, studies include health care–sector and societal perspectives. Most studies also include a state policy-maker perspective. The health care–sector perspective includes all direct health care (OUD- and non-OUD-related) costs incurred on behalf of the participant, including the day-to-day cost of managing the intervention, and participant out-of-pocket costs. The societal perspective accounts for all streams of costs incurred on behalf of participants, regardless of who incurs them, and should include all health care-sector costs, direct costs to the justice system, and broader societal costs, such as those associated with reduced school and workplace productivity, time required by participants and caregivers to obtain treatment, and those incurred by victims of crime (e.g., medical expenditures, property damage, pain and suffering). The state policy-maker perspective includes health care costs from a public-payer perspective, costs associated with social safety net programs, and direct costs to the justice system.

2.2.2. Intervention costs

Sources for all measures, by study, are listed in Table 2. Start-up (pre-implementation, implementation) and ongoing management cost analyses will be guided by the Costs of Implementing New Strategies (COINS) tool (Saldana, Chamberlain, Bradford, Campbell, & Landsverk, 2014), and the Drug Abuse Treatment Cost Analysis Program (DATCAP), a costing survey (French, Dunlap, Zarkin, McGeary, & McLellan, 1997).

Table 2.

Economic measures, by hub.

Measures	Data Source	Collection Method	HUB
Intervention Costs (start-up and ongoing management)	DATCAPa	Administrative & Semi-structured interviews	Baystate Brown University Chestnut Columbia Friends New York University Texas Christian University University of Chicago University of Kentucky Yale
	COINSb		University of Chicago
Healthcare Resource Utilization	NMOSc	Self-report	Brown University Chestnut Friends New York University Texas Christian University University of Chicago University of Kentucky Yale University
	PHDd	Administrative	Baystate
Criminal Activity	CLAFe	Self-report	Brown University Chestnut Friends New York University Texas Christian University University of Chicago University of Kentucky Yale University
	PHD	Administrative	Baystate
	NY State Unified Case Management System	Administrative	Columbia
Other social costs (e.g., workplace and school productivity, participant time, social safety net, caregiver time)	NMOS	Self-report	Brown University Chestnut Friends New York University Texas Christian University University of Chicago University of Kentucky Yale University
	PHD	Administrative	Baystate
Economic effectiveness measures
QALYs	PROPrf	Self-report	Brown University Chestnut Friends New York University Texas Christian University University of Chicago University of Kentucky Yale
Abstinent Years	Study records	Urine toxicology + Self-report	Brown University Chestnut Columbia Friends New York University Texas Christian University University of Chicago University of Kentucky Yale University
Opioid overdoses (fatal and non-fatal)	PHD	Administrative	Baystate

Notes:

^aDATCAP= Drug Abuse Treatment Cost Analysis Program.

^bCOINS= Costs of Implementing New Strategies.

^cNMOS= Non-Medical and Other Services.

^dPHD= Public Health Data Warehouse.

^eCLAF=Criminal and Legal Activities Form

^fPROPr=Patient Reported Outcomes Measurement Information System (PROMIS)-Preference (PROPr).

2.2.3. Health care resource utilization

The studies will capture health care resource utilization using administrative records, or self-report via the Non-study Medical and Other Services (NMOS) form (Murphy et al., 2020). The NMOS utilizes timeline follow-back (Sobell & Sobell, 1992) methodology to assess participant use of resources not directly related to the intervention.

2.2.4. Criminal activity and justice resources

The studies will measure criminal activity and justice resources via administrative records (e.g., official criminal records) or self-report. The studies will capture self-report using the Criminal and Legal Activities Form (CLAF) (Murphy et al., 2020), which includes questions on criminal activity for which there was no subsequent legal interaction (e.g., an arrest); and questions on arrests, charges, convictions, incarcerations, length of incarceration, and days on community supervision. Questions on criminal activity without justice-system interaction are important from the societal perspective, as society incurs certain costs for crimes committed, regardless.

2.2.5. Social safety net resources

The studies will capture social safety net resources (e.g., housing services, financial assistance, foster care) via the NMOS.

2.2.6. Effectiveness measures

For studies conducting a cost-effectiveness analysis, effectiveness measures include quality-adjusted life-years (QALYs), a measure of time abstinent from opioids (abstinent year), and opioid overdoses. The QALY is a combined measure of the health-related quality of life associated with a participant’s health state at a given point in time, and the duration of time spent in that state. The studies will measure health-related quality of life using the validated Patient-Reported Outcomes Measurement Information System (PROMIS)-Preference (PROPr) instrument (Hanmer & Dewitt, 2020). PROPr measures a participant’s capabilities (no problems to extreme problems) across eight health domains. The participants’ scores are used to define their health state, which is attached to a health-utility value representative of the general U.S. population’s preference for that state. The health-utility value typically ranges from 0 (death) to 1 (perfect health), although PROPr can also account for states perceived to be worse than death (i.e., <0). The manner in which opioid use by participants is assessed will vary according to hub, but will typically consist of a combination of urine toxicology and self-report. The hub will then calculate outcome “abstinent years” as the proportion of the year that the participant was abstinent from opioids. The Baystate Medical Center hub will use opioid overdoses as the measure of effectiveness, and they will obtained these data from the Massachusetts Public Health Data Warehouse (Massachusetts Department of Public Health, 2020). Although other hubs will collect overdose information, they are unlikely to have sufficient power to include it as an economic effectiveness measure.

2.2.7. Unit costs

Table 3 lists the unit cost sources. The costs associated with the start-up and ongoing management of each intervention will be calculated using the COINS and DATCAP tools. Health care-sector and societal perspective costs should reflect only the value of the resources, as opposed to accounting for factors such as profit and risk (Neumann et al., 2017). Therefore, health care resource utilization costs for these perspectives will be estimated using Medicare fee-for-service expenditures, given the intent/design of these payments to only reimburse providers for the resources that would be used to treat a typical patient with a given condition (Brady & Robinson, 2001). Pharmaceutical costs for the health care–sector and societal perspectives will be obtained from the Federal Supply Schedule (Neumann et al., 2017). Under the state policy-maker perspective, health care costs, including pharmaceuticals, will be valued according to nationally representative Medicaid expenditures, which will be calculated using Transformed Medicaid Statistical Information System data (Centers for Medicare & Medicaid Services, 2020).

Table 3.

Unit cost sources, by perspective.

Measure	Healthcare Sector Perspective	State Policymaker Perspective	Societal Perspective
Substance use disorder services
Pharmacotherapy (buprenorphine, methadone, naltrexone)	FSS	T-MSIS, CMS	FSS
Behavioral therapy	Medicare FFS	T-MSIS, CMS	Medicare FFS
Inpatient Detoxification	Medicare FFS	T-MSIS, CMS	Medicare FFS
Residential treatment	Medicare FFS	T-MSIS, CMS	Medicare FFS
Other healthcare services
Hospital stays	Medicare FFS	T-MSIS, CMS	Medicare FFS
Outpatient visits	Medicare FFS	T-MSIS, CMS	Medicare FFS
Emergency department visits	Medicare FFS	T-MSIS, CMS	Medicare FFS
Mental health visits	Medicare FFS	T-MSIS, CMS	Medicare FFS
Other prescriptions	FSS	T-MSIS, CMS	FSS
Criminal justice activities
Specific crimes	NA	McCollister et al. (2010): direct costs	McCollister et al. (2010): societal costs
Probation visits	NA	BLS	BLS
Other resources
Social safety net	NA	NMOS: Self-reported amount received	NA
Workplace productivity	NA	NA	Self-reported wages/benefits & hours worked
Educational productivity	NA	NA	Card (1999) & Max et al. (2004)
Participant time	NA	NA	Self-reported hourly compensation; Value of time in school; Federal minimum wage
Participant travel	NA	NA	Federal mileage reimbursement rate & self-reported mileage; Public transportation fee

Notes: FSS = US Department of Veterans Affairs Federal Supply Schedule; Medicare FFS = Medicare Fee-for-Service; T-MSIS = Transformed Medicaid Statistical Information System; CMS = Centers for Medicare and Medicaid Services; BLS = U.S. Bureau of Labor Statistics; NMOS = Non-study Medical and Other Services form

Criminal activity will be valued according to estimates that McCollister et al. (2010) develiped for unique offenses. These estimates include direct costs to the justice system, as well as tangible and intangible costs to victims. The direct costs to the justice system will be used to estimate justice-system costs from the state policy-maker perspective, while the societal price weights are a combination of the direct costs to the system and the costs incurred by victims.

Self-reported employment compensation will be used to value time spent working (i.e., workplace productivity). School productivity will be valued by applying the estimated return for a year of education in the U.S. (Card, 1999), according to the participant’s age (Max, Rice, Sung, & Michel, 2004), to time spent in school. Participant time costs (e.g., time spent traveling to and receiving treatment) will be calculated according to their hourly workplace/school productivity, or the federal minimum wage if unemployed. Direct travel costs will be estimated by applying the cost associated with the mode of travel (e.g., federal milage reimbursement rate, cost of public transportation) to the distance traveled. The value of social safety net resources utilized will be self-reported via the NMOS.

2.3. Within-study analyses

The studies will evaluate intervention costs using financial/administrative records from sites, where possible, and semi-structured interviews with site leaders and staff regarding the resources required to both implement and manage the intervention on a day-to-day basis. The sites will report start-up and ongoing management costs separately, and only the latter will be included in the analyses, given that one-time start-up costs become negligible over time.

Sites will calculate all other costs using the resource-costing method, which entails identifying “real-world” unit costs associated with each resource category, for each stakeholder, and multiplying the unadjusted number of resource units consumed by the relevant unit cost. Each study arm will estimate the predicted mean cost for each resource category within a given perspective and time-point using a multivariable generalized linear mixed model (GLMM). The GLMM is a flexible framework that allows for the most appropriate family and link functions to be chosen based on the observed data, as well as for the inclusion of random effects. The sites will then calculate final predicted mean values using the method of recycled predictions.

Differences in costs among arms will be tested for individual resources (e.g., ED visits, inpatient stays, criminal activity), according to perspective. Additionally, each arm will calculate average total costs, and each arm will test these to determine whether the intervention resulted in downstream cost-offsets sufficient to cover its cost, on average.

The outcome for the cost-effectiveness analyses will be the incremental cost-effectiveness ratio (ICER), which reports the marginal cost to achieve a unit of the effectiveness measure via the intervention, relative to the control. The ICER numerator is the difference in the average total costs described in the prior paragraph; the denominator is the difference in the predicted mean effectiveness measure, calculated using the same methods described above. Predicted mean health-utility values will be generated for each arm and time-point, then used in the calculation of average QALYs gained via the area-under-the-curve method. For “abstinent years”, predicted mean days abstinent from opioids will be calculated, then converted to represent the proportion of the year that the participant was abstinent. Regardless of whether the difference in the numerator or denominator is statistically significant, it is important to calculate the ICER and evaluate its statistical uncertainty (Glick, Doshi, Sonnad, & Polsky, 2014). The degree of uncertainty associated with the ICER will be illustrated via cost-effectiveness acceptability curves, which display the probability that the intervention would be considered cost-effective across a range of value thresholds (i.e., willingness-to-pay for an additional unit of effectiveness) (Glick et al., 2014).

Finally, each economic evaluation will include sensitivity analyses. Unit costs for which we are uncertain, and could be large cost drivers, will be altered according to our findings in the white and grey literature. Additionally, models will be estimated using various regression techniques (GLMM, ordinary least squares), and compared to unadjusted mean values, to evaluate the robustness of the parameter estimates.

2.4. Cross-study analyses

As mentioned, the similarities in study designs, settings, and measures will facilitate cross-study comparability and analyses. All hubs planning an economic evaluation will estimate the resources/costs required to start and sustain their intervention, both in aggregate and per participant. Additionally, by design, the hubs will be able to aggregate both cost and effectiveness measures to common timeframes to allow for rigorous, multivariable cross-study analyses within a given design/setting. A consistent measure of opioid use (e.g., urinalysis) will be used to calculate the effectiveness measure of “abstinent years” for the cross-study analyses. For example, among the three studies focused on pre-reentry MOUD, each hub will be able to test for significant cost-offsets associated with each medication (XR-NTX, methadone, both types of XR-BUP) from the health care-sector and state policy-maker perspectives, while controlling for potentially confounding factors, such as sociodemographic characteristics and site heterogeneity. Two of the three studies are collecting self-report information via the NMOS, CLAF, and PROPr, allowing for the evaluation of the cost-effectiveness of XR-NTX, XR-BUP (Sublocade), and XR-BUP (Brixadi), relative to one another, all from previously discussed perspectives.

3. Discussion

Each JCOIN study represents an important contribution to the field, and collectively the studies will cover critical touchpoints throughout the justice system for persons with OUD. Advancing our understanding of the relative economic impact of each intervention is central to JCOIN, and crucial for not only determining which strategies should be pursued in the interest of improving the well-being of the study populations, as well as the public, but also what it will take to start and sustain these interventions.

The extent to which within-study economic analyses will be directly comparable across hubs, varies. First, all hubs will be estimating the resources/costs necessary for intervention start-up and ongoing management. Although these figures are directly comparable within a given design/setting (e.g., linkage within a probation/parole setting), a comparison of program costs alone is not sufficient to inform resource allocation decisions. Second, all hubs will be collecting resource utilization data to estimate the health/policy-maker/social costs associated with each strategy, and evaluate the relative cost-offsets within each design/setting. The existence of significant cost-offsets and where they occur (e.g., fewer ED visits) is typically of great interest to stakeholders, but, again, this information alone is not sufficient to inform resource allocation decisions. Furthermore, only studies utilizing the NMOS and CLAF instruments will be capable of calculating costs from all three perspectives of interest, most specifically societal costs; however, the comparability of cost estimates from the health care-sector and state policy-maker perspectives could be problematic among studies relying on administrative data, depending on the extent those data are comprehensive.

Ensuring that resources are captured comprehensively is a key concern for most economic evaluations. Due to the fragmented nature of the U.S. health care system, administrative health records do not often follow participants over time, nor are they capable of being linked with justice or social-services records. Fortunately, self-reported service utilization has shown good validity over recall periods similar to those in the JCOIN studies (Brown & Adams, 1992; Harlow & Linet, 1989; Roberts, Bergstralh, Schmidt, & Jacobsen, 1996; Wallihan, Stump, & Callahan, 1999). Similarly, self-reported criminal activity has acceptable validity and reliability, in general, and among populations similar to those in JCOIN (Jansson, Hesse, & Fridell, 2008; Nieves, Draine, & Solomon, 2000; Thornberry & Krohn, 2003).

The final, but most informative, component of an economic evaluation is the incorporation of an effectiveness measure. The balance between costs and effectiveness is critical to inform efficient resource allocation decisions for any stakeholder. The effectiveness measure drives the usefulness of the cost-effectiveness analysis, but the importance of any particular measure can vary across stakeholders. The QALY is generalizable across interventions/programs, as well as diseases/disorders, and is associated with widely accepted cost/QALY thresholds that indicate whether the intervention is a “good value”, hence its recommendation as the primary outcome for CEAs (Neumann et al., 2017). Although the “abstinent year” measure does not yet have generally accepted value thresholds, the measure is clinically relevant, straightforward to interpret, and the cost/“abstinent year” outcome allows for direct comparison to economic evaluations in the OUD literature that have utilized time-abstinent measures, including studies that have it as their sole effectiveness measure (Jalali et al., 2020; Murphy & Polsky, 2016). Studies can apply objective values to fatal, but not nonfatal, opioid overdoses averted. Furthermore, most studies lack the statistical power to formally evaluate fatal overdose as an effectiveness measure, and in addition to nonfatal overdose being a difficult measure to capture, it is not generalizable beyond the OUD literature. Because the importance of, and thus willingness to pay for, a particular effectiveness measure can vary by stakeholder, cost-effectiveness acceptability curves will be calculated for each ICER, to display the likelihood that the intervention would be considered cost-effective across a wide range of value thresholds.

Given that each trial is prospectively designed, we expect few limitations with regard to within-study analyses; however, because each site designed its study independently, sites will encounter limitations when conducting cross-site analyses. Fortunately, the HEAT worked closely with the JCOIN Measures Committee to incorporate common economic instruments across studies, where possible. To provide further clarity with regard to limitations of within- or cross-study analyses, each economic study will include an Impact Inventory (Neumann et al., 2017) and a Consolidated Health Economic Evaluation Reporting Standards (CHEERS) Checklist (Husereau et al., 2013).

4. Conclusion

JCOIN promises to advance understanding of the impacts of the opioid crisis among justice populations, and inform decisions on which interventions/strategies represent the most efficient use of resources in a given setting. The studies will address questions of feasibility, economic efficiency, and sustainability to highlight interventions of “good value.” For example, an intervention that is slightly more effective than the next-best alternative, but far more expensive, will ultimately result in fewer people being treated, or a reduction in other stakeholder-provided resources. The HEAT and JCOIN economic studies will generate evidence on relative return on investment to inform best and sustainable practices for addressing OUD in justice populations.

Highlights.

• Justice populations are disproportionately affected by opioid use disorder (OUD).

• Few in justice population receive evidence-based OUD treatment.

• Evidence of economic value of OUD interventions in justice populations is sparce.

• Existing evidence severely limited by variation in economic study designs.

• JCOIN economic studies will inform best and sustainable practices.