Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Oct 1.
Published in final edited form as: J Subst Abuse Treat. 2022 Jul 2;141:108835. doi: 10.1016/j.jsat.2022.108835

Cost-effectiveness of extended-release injectable naltrexone among incarcerated persons with opioid use disorder before release from prison versus after release

Ali Jalali 1, Philip J Jeng 1, Daniel Polsky 2, Sabrina Poole 3, Yi-Chien Ku 3,4, George E Woody 3, Sean M Murphy 1
PMCID: PMC9508988  NIHMSID: NIHMS1833574  PMID: 35933942

Abstract

Introduction:

Opioid use disorder (OUD) is highly prevalent among incarcerated populations, and the risk of fatal overdose following release from prison is substantial. Despite efficacy, few correctional facilities provide evidence-based addiction treatment. Extended-release injectable naltrexone (XR-NTX) administered prior to release from incarceration may improve health and economic outcomes.

Methods:

We conducted an economic evaluation alongside a randomized controlled trial testing the effectiveness of XR-NTX before release from prison (n=38) vs. XR-NTX referral after release (n=48) of incarcerated participants with OUD, both groups continuing treatment at a community addiction treatment center. The incremental cost-effectiveness ratio (ICER) assessed the cost-effectiveness of XR-NTX before release compared to referral after release for three stakeholder perspectives at 12- and 24-week periods: state policymaker, health care sector, and societal. Effectiveness measures included quality-adjusted life-years (QALYs) and abstinent years from opioids. In addition, we categorized resources as OUD-related and non-OUD-related medical care, state transfers payments, and other societal costs (productivity, criminal justice resources, etc.).

Results:

Results showed an association between XR-NTX and greater OUD-related costs and total costs from the state policymaker perspective. QALYs gained were positive but statistically insignificant between arms; however, results showed XR-NTX had an estimated 15.5 more days of opioid abstinence over 24 weeks and statistically significant at a 95% confidence level based on the distribution of bootstrapped samples. We found that estimated ICERs to be > $500,000 per QALY for all stakeholder perspectives. For the abstinent-year effectiveness measure, we found XR-NTX before release to be cost-effective at a 95% confidence level for willingness-to-pay values >$49,000 per abstinent-year, across all perspectives.

Conclusions:

XR-NTX administered to persons who are incarcerated with OUD before release may provide value for stakeholders and bridge a well-known treatment gap for this vulnerable population. Lower than expected participant engagement and missing data limit our results, and study outcomes may be sensitive to methods that address missing data if replicated.

Keywords: Cost-effectiveness, Opioid use disorder, Criminal justice, Health econometrics, Naltrexone, Medications for opioid use disorder

1. Introduction

Access to evidence-based treatment for opioid use disorder (OUD) remains limited among persons involved in the criminal-legal or justice system, particularly for those incarcerated, despite the high prevalence of OUD among this population (Fazel et al., 2017; Zaller et al., 2019), and their elevated risk of overdose mortality shortly after release from incarceration (Binswanger et al., 2013; Binswanger et al., 2007; Bukten et al., 2017; Mattson et al., 2018; Pizzicato et al., 2018; Ranapurwala et al., 2018). Pharmacotherapy (i.e., buprenorphine, methadone, and naltrexone) is the recommended first line treatment for OUD (Leshner & Dzau, 2019; National Academies of Sciences). Research has shown all three medications to be effective and potentially cost-effective in numerous settings (Murphy & Polsky, 2016; Onuoha et al., 2021), including opioid agonist treatment among recently incarcerated persons (Gisev et al., 2015; Zarkin et al., 2020); however, methadone and buprenorphine are not readily available to those with OUD who are involved in the justice system (Krawczyk et al., 2017). Concerns regarding the diversion of narcotics, stigma/preference for non-narcotic treatment, and cost have been barriers to adoption of these medications among justice agencies (Nunn et al., 2009; Wakeman & Rich, 2015). Naltrexone, a non-narcotic, full opioid antagonist often administered as an extended-release injection (XR-NTX), could help to fill the OUD treatment gap for persons leaving incarceration, while providing approximately 30 days of protection from opioid relapse and overdose (Farabee et al., 2020; Friedmann et al., 2018; Gordon et al., 2021; Lincoln et al., 2018).

Unlike opioid-agonist (methadone) or partial-agonist (buprenorphine-naloxone) pharmacotherapies, providers recommend that patients be opioid-free for 7–10 days before initiating XR-NTX to reduce the risk of precipitated withdrawal (Sigmon et al., 2012), which can be severe. Prior work has shown a significant detoxification hurdle to XR-NTX relative to buprenorphine-naloxone, among persons seeking treatment for OUD in U.S. community-based inpatient detoxification or residential treatment programs; however, the medications were equally effective among participants who successfully initiated their medication (Lee et al., 2018). Another barrier to widespread adoption of XR-NTX is its relatively high price. According to the U.S. Department of Veterans Affairs Federal Supply Schedule (“US Department of Veterans Affairs Federal Supply Schedule”, 2020) (VA FSS), the 2020 price for an XR-NTX injection was $859.66, versus $124.23 for a 30-day supply of 16mg generic sublingual buprenorphine-naloxone, or $65.98 for a 30-day supply of 60mg (stabilization dose) generic methadone (“SAMHSA/CSAT Treatment Improvement Protocols,” 2018).

Theoretically, correctional settings should face relatively low barriers to XR-NTX induction among incarcerated individuals with OUD interested in treatment, assuming: a) the individuals have had little-to-no access to opioids; and b) the opportunity cost of their time is comparatively low. Upon successful induction, XR-NTX could result in downstream cost reductions that help to offset the price of the program, including the cost of the medication, by reducing opioid-relapse rates after release (Jarvis et al., 2018; Krupitsky et al., 2011; Lee et al., 2016; Lee et al., 2015).

The estimated economic burden of OUD to U.S. society was $787 billion in 2018, including excess health care expenditures, lost productivity, criminal justice resources, and premature mortality (Murphy et al., 2020). For the U.S. health care sector and taxpayers, Murphy et al. estimated annual OUD costs to be $89 and $93 billion, respectively (Murphy et al., 2020). Evaluating the cost of treatment for a particular stakeholder, in relation to the units of effectiveness gained, provides critical information for decision-makers about the best use of their scarce resources. Murphy and colleagues conducted the only two cost-effectiveness analyses alongside clinical trials to test the effectiveness of XR-NTX as a treatment for OUD (Murphy et al., 2019; Murphy et al., 2017), one of which focused on a community-dwelling population with prior justice involvement (not necessarily incarceration; Murphy et al., 2017). To date, no such information exists for the provision of XR-NTX to the high-risk population of individuals with OUD who are currently incarcerated.

We conducted a prospective economic evaluation alongside a randomized controlled clinical trial that tested the effectiveness of XR-NTX administered to individuals with OUD before release from incarceration, compared to referral to XR-NTX after release from incarceration, both continuing XR-NTX treatment at NET Steps, a publicly funded community substance use disorder treatment program in Philadelphia, PA. We estimated costs according to the health care sector, state policymaker, and societal perspectives. Effectiveness measures included quality adjusted life-years (QALYs) and abstinent-years gained.

2. Methods

2.1. Clinical trial

The trial was a two setting, 2-group, open-label randomized controlled effectiveness trial, stratified by sentencing status (sentenced/non-sentenced) and gender (male/female). Participants had a history of OUD, fully detoxified from opioids, incarcerated in the Philadelphia Department of Prisons at the time of enrollment, interested in receiving XR-NTX treatment, lived in Philadelphia, and not planning to leave Philadelphia for 6 months after release from incarceration. The parent study randomized one hundred and forty-six participants to receive (1) XR-NTX at baseline before release from incarceration, continuing treatment at NET Steps where they could receive 3 additional injections of XR-NTX (n=74), or (2) referral to a NET Steps for their baseline XR-NTX injection 3–7 days following reentry (n=72), and 3 additional study injections of XR-NTX (n=72). The study collected data at baseline, with payments provided to participants for weekly urine drug screenings combined with monthly Time Line Follow Back (TLFB) reports of substance use (Linda C. Sobell, 2007) The parent study also collected detailed health care utilization and incarceration data monthly, and health-related quality of life (HRQoL), wage and labor productivity data at 12- and 24-week assessment points, with follow-up phone interviews by study staff to help mitigate data loss from missed appointments.

Sixty of the 146 randomized participants became ineligible for treatment after randomization due to administrative withdrawal, correctional facility transfers, being mandated to an alternative treatment, or being released from incarceration prior to receiving an XR-NTX injection and not providing data beyond baseline necessary for the economic evaluation. Thus, our analysis includes a total of 86 participants (n=38 XR-NTX before release, n=48 XR-NTX referral after release) in the analytic sample for both the primary and economic evaluations of the trial (Woody et al., 2021). Analysis by Woody et al. (2021) demonstrated that the lower-than-expected sample size did not bias primary group comparisons but lowered statistical power of the study (Woody et al., 2021). By design, all participants in the before release group received their first injection compared to only 33% in the comparison group, with lower and statistically insignificant known relapse rate, and a statistically significant treatment duration (Woody et al., 2021). Additional details on the parent trial, data collection, and clinical trial design are available elsewhere (Woody et al., 2021).

2.2. Design

The analyses in this study followed well-established guidelines for conducting an economic evaluation alongside a clinical trial (Glick et al., 2014; Neumann et al., 2014; Neumann et al., 2016) and reported outcomes in a manner directly comparable to previous economic evaluations of substance use disorder interventions (Jalali et al., 2020; Murphy et al., 2016). A detailed description of the economic study design, measures, and statistical analyses have been previously published (Murphy et al., 2020), which we briefly review here.

2.3. Measures

2.3.1. Effectiveness

QALYs and opioid abstinent years (a measure of time abstinent from opioids) were the measures of effectiveness. The QALY is a widely generalizable, longitudinal measure of a treatment’s effectiveness, combining the health-related quality-of-life (HRQoL) associated with an individual’s health state, and the time spent in that health state. We calculated U.S.-based HRQoL values using the EuroQoL 5D-3L instrument, which can generate an index value ranging from −.596 to 1, where 1 indicates perfect health, 0 indicates death, and health states perceived to be worse than death are below 0 (EuroQol). The abstinent year is a measure of time abstinent from opioids, and, indirectly, pharmacotherapy retention. We calculated opioid abstinence as a combined measure of urinalysis and self-reported opioid use, with missed or refused urinalysis tests counted as positive, except for instances where XR-NTX had recently been administered (within 30 days), following prior studies (Lee et al., 2018; Murphy et al., 2016).

2.3.2. Cost

We valued resources according to the resource-costing method, which entails weighting each resource unit utilized/consumed by participants with an appropriate unit cost. We chose unit costs according to stakeholder perspective, with the intent of representing their real-world costs. As in prior studies by the Murphy team (Murphy et al., 2016; Murphy et al., 2019; Murphy et al., 2017; Neumann et al., 2014; Polsky et al., 2010), the Non-study Medical and Other Services (NMOS) form elicit self-reported information on health care resource utilization, and out-of-pocket health care, pharmaceutical, and travel costs; we collected self-reported nonstudy medication use using a concomitant medication questionnaire; participants also self-reported state transfers (e.g., unemployment benefits, welfare), criminal activities, hours worked, and wages received based on the Addiction Severity Index (ASI) assessment (McLellan et al., 1992). Supplemental Appendix A contains the Impact Inventory (Peter J. Neumann, 2016), which lists the resources relevant to each stakeholder, and Table B1 in Supplemental Appendix B lists all unit costs, according to the stakeholder’s perspective.

Implementation costs attributable to XR-NTX before release from incarceration were previously estimated by our team using a micro-costing analysis and included the costs of the intervention per participant for all stakeholder perspectives (Jeng et al., 2021). The micro-costing analysis solicited estimates of resources required to operate the XR-NTX program, including medication, supplies, and labor, from relevant study and site personnel via semi-structured interviews, guided by the Drug Abuse Treatment Cost Analysis Program (DATCAP) instrument (French, 2003; French et al., 1997).

2.3.3. Cost-effectiveness

The incremental cost-effectiveness ratio (ICER) was the primary measure of cost-effectiveness. We calculated ICERs for each stakeholder perspective at the 12- and 24-week periods by dividing the difference in the multivariable-adjusted average cost between the XR-NTX before release and XR-NTX referral after release treatment arms, by the difference between the arms’ multivariable-adjusted average effectiveness; thus, the ICER measures the additional cost incurred on behalf of the average participant by a particular stakeholder to gain 1 additional unit of effectiveness. A health intervention is considered cost-effective if the value of the ICER falls below the stakeholder’s willingness-to-pay value for each unit increase in effectiveness. The health economic literature generally considers ICERs between $100,000 to $200,000 per QALY as cost-effective (Glick et al., 2014; Neumann et al., 2014; Neumann et al., 2016). No widely accepted value threshold has been established for the abstinent year measure; however, time abstinent from opioids is an important clinical measure, and has been, and continues to be, used as an effectiveness measure in many economic evaluations of OUD interventions (Jalali et al., 2020).

2.4. Statistical analyses

Clinical and demographic covariates of interest included baseline measures of the participant’s gender, age, race (non-Hispanic black or white and other), ethnicity (Hispanic), insurance status (Medicaid, uninsured, other), marriage status, baseline history of injection drug use (opioids and other drugs), medical care costs (past 30 days), and 7 composite scores of potential problem areas (medical, employment, alcohol, drug, legal, family/social, and psychiatric) from the ASI. ASI scores range from 0 and 1, with higher scores indicating greater severity (McLellan et al., 1992).

We modeled all cost and effectiveness measures using generalized linear models (GLMs) with clustered standard errors to account for longitudinal correlation at the participant level, and multiple imputation to address uncertainty due to missing data. The GLM is a family of flexible statistical models, which allows the mean and variance function to be selected according to the model’s fit to the observed data, with the assistance of statistical tests such as the modified Park test, Pregibon link test, Pearson correlation test, and modified Hosmer-Lemeshow test (Henry A. Glick, 2014). Health care and non–health care resources were grouped into societal (criminal activity and labor productivity), OUD treatment costs, non-OUD medical and other costs dependent on the stakeholder perspective (out-of-pocket pharmaceutical costs for the health care sector, state welfare transfers for the state policymaker) categories and analyzed. We utilized two-part models when necessary to account for excess zeros (applied to OUD-related and other medical costs GLM regressions).

Prior to employing multiple imputation, we applied inverse probability weights to all regressions within the GLM framework to account for missingness while determining the appropriate family and link functions with observed data. Imputation methods provide valid inferences when applying bootstrapped resampling in the presence of missing data as long as the imputation model is valid (Schomaker & Heumann, 2018). Additionally, imputing missing data then applying bootstrapped resampling to estimate uncertainty of cost-effectiveness statistics are computationally efficient and can perform well in trial-based cost-effectiveness analyses (Brand et al., 2019). To assess the best imputed dataset used in the non-parametric bootstrap estimation procedure, we used the two-sample Kolmogorov-Smirnov test of equality of distribution to assess the goodness-of-fit of the imputed data. In addition, imputation models addressing missing data in epidemiological and clinical trials helped to guide all analysis (Sterne et al., 2009), including how to combine resampling and missing data methods within the GLM framework (Jalali et al., 2022). Supplemental Appendix B provides further detail on patterns of missingness and the bootstrapping procedure.

2.5. Study outcomes

We utilized the GLMs to calculate multivariable-adjusted, predicted means for each category of cost and effectiveness at the relevant study time points using the statistical method of recycled predictions (Henry A. Glick, 2014). Then, we aggregated these values to report outcomes and cost-effectiveness statistics at the 12- and 24-week periods. Moreover, we calculated QALYs for each arm using the predicted HRQoL utility weights from the EQ-5D and the area-under-the-curve methodology. Furthermore, we used estimated QALYs and abstinent years to report two separate sets of ICERs for each stakeholder perspective at 12 and 24 weeks. All effectiveness outcomes are reported in annualized values (recalculated assuming constant effect over a 1-year period) as reported by prior analyses in the literature (Murphy et al., 2019).

To account for sampling uncertainty, we estimated standard errors and percentile confidence intervals of predicted costs, effectiveness, and ICERs by performing the GLM regression procedure within a non-parametric bootstrap (Barber & Thompson, 2000; Henry A. Glick, 2014). We calculated ICERs regardless of the statistical significance of the cost and effectiveness estimates, because the statistical power to detect a joint difference in cost and effects is greater than that of detecting a difference in cost and effect individually (Glick et al., 2014).

2.6. Sensitivity analysis

The non-parametric bootstrap produces a distribution of cost and effect pairs by treatment arm, which we used to construct cost-effectiveness acceptability curves (CEAC) for different stakeholder perspectives for both the 12- and 24-week aggregation periods. The CEACs display the probability that the ICER would fall below a given willingness-to-pay threshold for stakeholders, i.e., the likelihood that the treatment is considered cost-effective, across a range of values per QALY or abstinent years gained. XR-NTX injection was $859.66 based on the VA FSS, which is based on the “Big Four” listed price (VA, Department of Defense, Public Health Service, and Coast Guard), but also listed at the higher price of $1,226.91 for other governmental agencies. We repeat the estimate of OUD-related costs to examine sensitivity of the overall results with the higher XR-NTX injection price.

The GLM model for the primary results utilized a battery of statistical tests recommended to determine the most appropriate family and link functions when modeling costs. In addition, we applied the two-part regression models to account for excess zeros in longitudinal costs. We tested the sensitivity of the cost-effectiveness results using two-part and single ordinary least squares (OLS) regressions, as well as two-part GLMs with the canonical link functions (e.g., natural logarithm for the GLM-gamma). Supplemental Appendix B reports the variation in the multivariable-adjusted predicted means of the OUD and other medical cost categories based on GLM and OLS specifications. The research team conducted all analyses using STATA 16.1 (StataCorp LLC, College Station, TX).

3. Results

Figure 1 displays the study XR-NTX injections in the analytic sample by study arm. We reported patient characteristics by treatment arm in Table 1. Pooled two-sample t-tests (Binswanger et al., 2013), used to assess statistical difference between the XR-NTX before release and XR-NTX after release from incarceration arms, showed that patients’ characteristics differed across baseline insurance status, age, and history of intravenous opioid drug use. Unadjusted costs and effectiveness outcomes are biased in the presence of treatment imbalance. Therefore, regression-adjusted outcomes using the method of recycled predictions and resampling that account for baseline characteristics and sampling uncertainty are more likely than unadjusted values to provide unbiased estimates (Glick et al., 2014).

Figure 1.

Figure 1.

Open in a new tab

Study Injection by Treatment Arm

Notes. XR-NTX denotes extended-release injectable naltrexone.

Table 1.

Study Participant Baseline Characteristics

XR-NTX Before Release (n=38) XR-NTX After Release (n=48)
Variables Mean Std. Dev. Mean Std. Dev. p-value
Female 29% 0.46 25% 0.44 0.6572
Uninsured 34% 0.48 13% 0.33 0.0157
Medicaid 37% 0.49 60% 0.49 0.0300
Married 18% 0.39 8% 0.28 0.1681
Age 40.53 8.78 35.98 8.09 0.0145
Black 29% 0.46 13% 0.33 0.0582
Hispanic 21% 0.41 17% 0.38 0.6087
High School Education 79% 0.41 79% 0.41 0.9805
Intravenous Drug Use 45% 0.50 77% 0.42 0.0018
Addiction Severity Indices
Medical 0.12 0.28 0.16 0.29 0.4858
Employment 0.82 0.19 0.79 0.22 0.4713
Alcohol 0.14 0.24 0.06 0.16 0.0766
Drug 0.41 0.10 0.42 0.12 0.7611
Legal 0.45 0.26 0.45 0.25 0.9410
Family-Social 0.29 0.26 0.23 0.29 0.3188
Psych 0.32 0.25 0.36 0.21 0.4821
Open in a new tab

Notes: XR-NTX denotes extended-release injectable naltrexone. p-values reported in the table test for the statistical difference in the mean calculated using a two-tailed t-test.

3.1. Costs

Table 2 reports the predicted mean costs and incremental costs for the health care sector, state policymaker, and societal perspectives by 12- and 24-week time periods. We report all incremental costs as XR-NTX before release compared to referral after release. The average intervention cost was $976 per participant in the treatment group based on detailed micro-costing analysis (Jeng et al., 2021). Incremental OUD treatment costs were $955 (95% CI: $223, $1,687) and $1,626 (95% CI: $541, $2,710) for the 12- and 24-week periods. From the health care sector and societal perspectives, incremental other-medical and out-of-pocket costs associated with treatment were $415 (95% CI: −$2,154, $2,984) and $1,154 (95% CI: −$2,568, $4,875) for the 12- and 24-week periods. For the state policymaker perspective, the incremental other-medical and state-transfer costs were $546 (95% CI: −$1,088, $2,179) and $1,312 (95% CI: −$1,362, $3,986) for the 12- and 24-week periods. Incremental non–health care societal costs associated with treatment were −$699 (95% CI: −$3,709 $2,311) and $627 (95% CI: −$3,736, $4,990) for the 12- and 24-week periods.

Table 2.

Predicted Mean Costs and Outcomes from GLM Regressions

12 Weeks 24 Weeks
Cost and Effectiveness Measures XR-NTX before release XR-NTX After Release Difference 95% Confidence Interval XR-NTX before release XR-NTX after release Difference 95% Confidence Interval
Outcomes
Annualized QALYs 0.793 0.790 0.003 (−0.063, 0.069) 0.749 0.743 0.006 (−0.062, 0.073)
Annualized Abstinent Years 0.587 0.266 0.322 (.174, 0.469) 0.386 0.201 0.185 (0.046, 0.324)
Cost, 2019 US $
Intervention 976 0 976 976 0 976
OUD-related 3,250 2,295 955 (223, 1,687) 5,412 3,786 1,626 (541,2,710)
Other Medical and Out-of-Pocket 3,522 3,107 415 (−2,154, 2,984) 5,817 4,663 1,154 (−2,568, 4,875)
Other Medical and State Transfers 3,194 2,648 546 (−1,088, 2,179) 5,463 4,151 1,312 (−1,362, 3,986)
Societal 276 975 −699 (−3,709, 2,311) 1,113 485 627 (−3,736, 4,990)
Total Cost, 2019 US $
State Policymaker Perspective 7,419 4,943 2,477 (675, 4,278) 11,851 7,937 3,914 (1,091,6,737)
Healthcare Sector Perspective 7,748 5,402 2,346 (−341,5,033) 12,205 8,449 3,755 (−87, 7,598)
Societal Perspective 8,024 6,377 1,647 (−1,363, 4,657) 13,317 8,935 4,382 (19, 8,745)
Open in a new tab

Notes: QALY denotes quality-adjusted life-years. XR-NTX denotes extended-release injectable naltrexone. OUD denotes opioid use disorder. The sample size for XR-NTX before release and after release consisted of 38 and 48 participants, respectively.

The total cost, including the cost of the intervention, associated with XR-NTX before release was highest for the societal perspective ($8,024 at 12 weeks and $13,317 at 24 weeks), followed by total costs for the health care sector ($7,748 at 12 weeks and $12,205 at 24 weeks), and state policymaker ($7,419 at 12 weeks and $11,851 at 24 weeks). We found that for the state policymaker perspective, incremental costs of XR-NTX before vs. after release was positive and statistically different from zero for both time periods of aggregation. On average, the expected additional costs faced by a state policymaker was $2,477 (95% CI: $675, $4,278) and $3,914 (95% CI: $1.091, $6,737) for the 12- and 24-week periods, respectively.

3.2. Effectiveness

We report QALYs and abstinent years by treatment and time period at the top of Table 2. The predicted mean QALYs associated with XR-NTX before release were slightly higher than the control arm, with incremental annualized QALYs of 0.003 (95% CI: −0.063, 0.069) at 12 weeks postintervention, and 0.006 (95% CI: −0.062, 0.073) at 24 weeks; however, the QALYs gained were not statistically significant at the 95% level. Annualized abstinent years were greater in the XR-NTX before release arm, compared to referral after release for both the 12-week (0.322, 95% CI: 0.174, 0.469) and 24-week (0.185, 95% CI: 0.046, 0.324) periods. Abstinent years were statistically significant at the 95% level using bootstrapped resampling.

3.3. Cost-effectiveness

Table 3 reports the ICERs based on the predicted means of cost and effectiveness for the 12- and 24-week periods. We found estimated ICERs to be > $500,000 per QALY for all stakeholder perspectives in all periods studied. Cost-per-abstinent-year ICERs ranged from $5,122 for the societal perspective, to $7,701 for the state policymaker perspective in the first 12 weeks. Cost-per-abstinent-year ICERs for the 24-week periods were $20,293 for the health care sector, $21,149 for the state policymaker, and $23,681 for the societal perspectives. No standard willingness-to-pay threshold exists for an abstinent year, but if we proxy willingness-to-pay thresholds based on historically minimum values reported as cost-effective for both QALYs and life-years-saved effectiveness measures (Grosse, 2008), we may consider XR-NTX to be cost-effective.

Table 3.

Incremental Cost-effectiveness Ratio XR-NTX Before Release

Stakeholder Perspective ICER per QALY ICER per Abstinent Year
12 Weeks
State Policymaker Perspective 880,600 7,701
Healthcare Sector Perspective 834,126 7,295
Societal Perspective 585,649 5,122
24 Weeks
State Policymaker Perspective 703,675 21,149
Healthcare Sector Perspective 675,175 20,293
Societal Perspective 787,918 23,681
Open in a new tab

Notes: QALY denotes quality-adjusted life-years. ICER denotes the incremental cost-effectiveness ratio and calculated as the incremental difference in costs and effects of XR-NTX before release compared to after release. XR-NTX denotes extended-release injectable naltrexone. OUD denotes opioid use disorder.

Figures 2 and 3 display the cost-effectiveness acceptability curves for all stakeholder perspectives for the 12- and 24-week periods in panels. Results from the cost-effectiveness acceptability curve analysis indicated that XR-NTX is cost-effective at a 95% confidence level for willingness-to-pay values > $49,000 per abstinent year across all perspectives.

Figure 2.

Figure 2.

Open in a new tab

Cost-effectiveness Acceptability Curve of XR-NTX Before Release (QALYs, 24 weeks)

Notes. XR-NTX denotes extended-release injectable naltrexone. QALY denotes quality-adjusted life-years. The figure displays the probability that XR-NTX before release is cost-effective given willingness-to-pay per QALY gained ranging from 0 to $500,000.

Figure 3.

Figure 3.

Open in a new tab

Cost-effectiveness Acceptability Curve of XR-NTX Before Release (Abstinent Years, 24 weeks)

Notes. XR-NTX denotes extended-release injectable naltrexone. The figure displays the probability that XR-NTX before release is cost-effective given willingness-to-pay per abstinent year gained ranging from 0 to $100,000.

4. Discussion

This study was the first comprehensive economic evaluation of a pre-release XR-NTX intervention among the high-risk population of individuals with OUD released from incarceration, including the first to provide detailed estimates of health care and non–health care costs for this population during the particularly risky period immediately following release. We found an association between initiation of XR-NTX treatment before release with higher OUD-related costs over both the 12- and 24-week periods, which positively correlated with the unit cost of XR-NTX (Table B5 in Supplemental Appendix B). This association was consistent with results from the parent trial, which also demonstrated an association between XR-NX before release with a statistically significant increase in weeks in OUD treatment. Resources accounted for in this category included outpatient OUD treatment days, residential detoxification days, hospital detoxification days, XR-NTX injections, and nonstudy buprenorphine, and methadone treatment. The costs associated with non-OUD health care resources, such as inpatient and outpatient hospital days, emergency room encounters, and physician office visits, were also slightly higher for the XR-NTX before release group but were not statistically significant at the 95% confidence level.

The additional health care resource utilization/costs associated with the XR-NTX before-release intervention translated to statistically significantly increased time abstinent from opioids, but not QALYs. Non–statistically significant incremental QALYs from XR-NTX compared to treatment-as-usual have been previously reported among justice-involved populations (Murphy et al., 2017), and future studies may benefit from a longer study period to examine long-term QALY differences. Stakeholders reviewing results from this study should consider the study limitations when interpreting these results. Primarily, the less than expected number of successfully randomized patients in the analytic sample lowered the power to detect meaningful differences in outcomes measures, and, therefore, we do not preclude the presence of Type II errors. Conversely, the difference in predicted abstinent years between study arms was positive and found to be statistically significant but decreased over time. While the utility of using quality of life measures in economic evaluations rests in its application in comparative resource allocation decisions across treatments strategies for varying disorders and illnesses, clinical measures of effectiveness such as abstinence or treatment retention provide an alternative, OUD treatment-centered approach for decision-making.

We calculated that XR-NTX before release had an estimated 15.5 more days of opioid abstinence over the 24-week period using the point estimates reported in Table 2 compared to after release. The decrease in estimated time abstinence with the observation length is not surprising since the pre-release XR-NTX injections failed to promote sustained treatment adherence (monthly injection rates) among the intervention group (Figure 1). While opioid overdose mortality rates are highest immediately following release from incarceration, mortality risk remains elevated compared to non-justice-involved populations even a year after incarceration (Ranapurwala et al., 2018). The results of this study, therefore, support implementation of further interventions that address treatment discontinuations in combination with pre-release XR-NTX e.g., contingency management or integration of medical and social services—though we know little about the most effective methods (Chan et al., 2021). A comprehensive review and meta-analysis of naltrexone treatment for OUD in the justice-involved population found that we still do not know strategies that improve retention in community care (Bahji et al., 2020). Digital therapeutics may be an alternative, cost-effective option (Wang et al., 2021), but these therapies are relatively new and need further evidence to determine efficacy.

Mean study and nonstudy costs of XR-NTX before release for both treatment groups, along with the total costs for the 24-week period, are within the range of values reported by recent economic evaluations of OUD pharmacotherapy across different populations (Murphy et al., 2019; Murphy & Polsky, 2016; Murphy et al., 2017). Two recent economic evaluations of OUD treatment conducted alongside a clinical trial are relevant comparators to this study and merit further discussion. Murphy et al. (2017) evaluated XR-NTX compared to brief counseling and referral to treatment among community-dwelling, justice-involved persons with OUD (Murphy et al., 2017). Murphy et al. (2017) evaluated the impact of XR-NTX on costs, QALYs, and abstinence, and employed similar statistical methods as those implemented in the current analysis (Murphy et al., 2017). The point estimates of ICERS were much higher in this study than those reported by Murphy et al. (2017), and cost-effectiveness acceptability curves did not indicate a high likelihood of economic value for stakeholders in the $100,000 to $200,000 willingness-to-pay range (Murphy et al., 2017). Similar to this study, however, Murphy et al. (2017) did not find evidence of cost-offsets associated with XR-NTX treatment (Murphy et al., 2017). The driver of value in both studies is based on improvements in effectiveness. The predicted incremental abstinent years between treatments were comparable to Murphy et al. (2017), but incremental QALYs as well as predicted mean levels were lower in this study. The lower incremental QALYs may be due to the difference in the risk profile of the justice-involved populations. Moreover, the comparator treatment group in Murphy et al. (2017) was brief counseling and referral and not necessarily participants screened to be interested in XR-NTX treatment when referred to community care after release (Murphy et al., 2017).

Murphy et al. (2019) reported results from an economic evaluation comparing buprenorphine-naloxone vs. XR-NTX alongside a clinical trial of adults with OUD at inpatient detoxification and residential treatment programs (Murphy et al., 2019). Both the incremental QALYs and predicted mean levels reported in this study are comparable to Murphy et al. (2019). Murphy et al. (2019) reported similar mean cost estimates and did not find statistically significant cost-offsets between treatments (Murphy et al., 2019). In contrast to the pattern of effectiveness over time observed in this study, Murphy et al. (2019) reported an increase in both QALYs and abstinent years for the XR-NTX treatment arm (Murphy et al., 2019). The population examined in their analysis were not primarily justice-involved, and the length of study in Murphy et al. (2019) was greater than the current analysis (36 weeks compared to 24 weeks). Whether the justice-involved population receiving XR-NTX before release experience potential effectiveness gains after the first 24 weeks is unknown. Uncertainty of treatment effectiveness beyond the study length analyzed in this economic evaluation may be addressed in decision-analytic models that simulate hypothetical cohorts over extended time periods. Studies using noneconometric methods may provide further insight.

As is the case in all studies, the results reported in this analysis were subject to limitations. The most important limitation is the missingness and less-than-planned number of participants successfully randomized. While we applied inverse probability weights and subsequent multiple imputation methods to calculate multivariable-adjusted cost- effectiveness statistics, the application of these methods in small sample studies with missing data may reduce the external validity of the results. However, bootstrapped resampling allowed us to assess uncertainty by re-estimating adjusted mean outcomes of treatment by varying the underlying participant populations. Further, the short follow-up period did not provide enough data on patterns of costs and effectiveness outcomes over time given that OUD is a chronic condition and patients are encouraged to maintain pharmacotherapy. Although the follow-up period is shorter in prospective clinical trials for practical reasons, the use of patient-level health, treatment, and health care utilization data allowed for strong internal validity of the results relative to modeling approaches that simulate patient data based on consolidation of multiple studies (e.g., Markov or simulation models). Further, cohort analysis with simulated data does not directly observe the relationship between effectiveness and cost outcomes over time, which limits the value of such studies absent real-world data.

Some participants’ characteristics in the study differed on average between treatment groups due to the exclusion of 60 randomized participants who became ineligible prior to treatment initiation. This exclusion introduced potential random bias in the data leading to unreliable average treatment effects in unadjusted mean outcomes. However, the econometric methods used in this study adjust for the imbalanced patient characteristics at baseline and patient attrition over time. The robust resampling methods, using the non-parametric bootstrap in combination with multiple imputation, addressed many of the pitfalls associated with loss-after-randomization.

Nevertheless, the reported estimates of incremental costs exhibited high variance, and the study failed to report statistically significant differences in some cost categories. To the best judgment of the authors, results reported in this study are statistically sound and provide important information for stakeholders given available data from the trial.

This study provided needed data to address an emerging public health emergency among the high-risk population of persons with OUD recently released from incarceration. We observed an association between the comprehensive economic evaluation of XR-NTX before release with higher OUD-related costs, but found no evidence for potential cost-offsets. An association did exist between earlier initiation of XR-NTX with additional QALYs and abstinent years. QALY-based ICERs had lower likelihood of the intervention being cost-effective for all three stakeholder perspectives at willingness-to-pay values greater than $100,000 compared to more favorable results for abstinent years in a comparable range. Although studies with longer follow-up data and improved participant engagement are needed, this study is the first to provide evidence that XR-NTX administered before release from incarceration may provide value for stakeholders and bridge a well-known treatment gap for this vulnerable and high-risk population.

Supplementary Material

1
2

Highlights.

  • Extended-release naltrexone (XR-NTX) before release from prison provides economic value for stakeholders.

  • XR-NTX before release increased opioid use disorder related costs and decreased opioid use

  • XR-NTX before release is cost-effective compared to referral after release from a state policymaker perspective

Funding:

This work was supported by the National Institute on Drug Abuse [grant numbers R01DA046721, P30DA040500].

Abbreviations:

OUD

Opioid Use Disorder

XR-NTX

extended-release injectable naltrexone

QALYs

quality-adjusted life-years

HRQoL

health-related quality of life

TLFB

Time Line Follow Back

DATCAP

Drug Abuse Treatment Cost Analysis Program

ICER

incremental cost-effectiveness ratio

CEAC

cost-effectiveness acceptability curve

Footnotes

Declaration of interest: None

CRediT Author Statement

Ali Jalali: conceptualization, methodology, software, validation, formal analysis, writing – original draft, visualization; Philip J. Jeng: investigation, writing – review & editing, project administration; Daniel Polsky: conceptualization, writing – reviewing and editing, supervisions; Sabrina Poole: resources, data curation, writing – review & editing, project administration; Yi-Chien Ku: investigation, data curation, writing – review & editing, supervisions; George E. Woody: conceptualization, methodology, investigation, resources, data curation, writing – review & editing, supervisions; Sean M. Murphy: conceptualization, methodology, investigation, writing – review & editing, validation, supervision, funding acquisition.

Clinical Trial Registration: NCT02617628

References

  1. Bahji A, Carlone D, & Altomare J (2020). Acceptability and efficacy of naltrexone for criminal justice-involved individuals with opioid use disorder: a systematic review and meta-analysis. Addiction, 115(8), 1413–1425. 10.1111/add.14946 [DOI] [PubMed] [Google Scholar]
  2. Barber JA, & Thompson SG (2000). Analysis of cost data in randomized trials: an application of the non-parametric bootstrap. Stat Med, 19(23), 3219–3236. [DOI] [PubMed] [Google Scholar]
  3. Binswanger IA, Blatchford PJ, Mueller SR, & Stern MF (2013). Mortality after prison release: opioid overdose and other causes of death, risk factors, and time trends from 1999 to 2009. Ann Intern Med, 159(9), 592–600. 10.7326/0003-4819-159-9-201311050-00005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Binswanger IA, Stern MF, Deyo RA, Heagerty PJ, Cheadle A, Elmore JG, & Koepsell TD (2007). Release from prison--a high risk of death for former inmates. N Engl J Med, 356(2), 157–165. 10.1056/NEJMsa064115 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Brand J, van Buuren S, le Cessie S, & van den Hout W (2019). Combining multiple imputation and bootstrap in the analysis of cost-effectiveness trial data. Stat Med, 38(2), 210–220. 10.1002/sim.7956 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bukten A, Stavseth MR, Skurtveit S, Tverdal A, Strang J, & Clausen T (2017). High risk of overdose death following release from prison: variations in mortality during a 15-year observation period. Addiction, 112(8), 1432–1439. 10.1111/add.13803 [DOI] [PubMed] [Google Scholar]
  7. Chan B, Gean E, Arkhipova-Jenkins I, Gilbert J, Hilgart J, Fiordalisi C, Hubbard K, Brandt I, Stoeger E, Paynter R, Korthuis PT, & Guise JM (2021). Retention Strategies for Medications for Opioid Use Disorder in Adults: A Rapid Evidence Review. J Addict Med, 15(1), 74–84. 10.1097/ADM.0000000000000739 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. EuroQol. Retrieved 20 February from https://euroqol.org
  9. Farabee D, Condon T, Hallgren KA, & McCrady B (2020). A randomized comparison of extended-release naltrexone with or without patient navigation vs enhanced treatment-as-usual for incarcerated adults with opioid use disorder. Journal of Substance Abuse Treatment, 117, 108076. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Fazel S, Yoon IA, & Hayes AJ (2017). Substance use disorders in prisoners: an updated systematic review and meta-regression analysis in recently incarcerated men and women. Addiction, 112(10), 1725–1739. 10.1111/add.13877 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. French MT (2003). Drug Abuse Treatment Cost Analysis Program (DATCAP): User’s Manual Eighth Edition, University of Miami, Florida. [Google Scholar]
  12. French MT, Dunlap LJ, Zarkin GA, McGeary KA, & McLellan AT (1997). A structured instrument for estimating the economic cost of drug abuse treatment. The Drug Abuse Treatment Cost Analysis Program (DATCAP). J Subst Abuse Treat, 14(5), 445–455. 10.1016/s0740-5472(97)00132-3 [DOI] [PubMed] [Google Scholar]
  13. Friedmann PD, Wilson D, Hoskinson R, Poshkus M, & Clarke J (2018). Initiation of extended release naltrexone (XR-NTX) for opioid use disorder prior to release from prison. Journal of Substance Abuse Treatment, 85, 45–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gisev N, Shanahan M, Weatherburn DJ, Mattick RP, Larney S, Burns L, & Degenhardt L (2015). A cost-effectiveness analysis of opioid substitution therapy upon prison release in reducing mortality among people with a history of opioid dependence. Addiction, 110(12), 1975–1984. [DOI] [PubMed] [Google Scholar]
  15. Glick HA, Doshi JA, Sonnad SS, & P. D (2014). Economic evaluation in clinical trials (2nd Edition). OUP Oxford. 10.1093/med/9780199685028.001.0001 [DOI] [Google Scholar]
  16. Gordon MS, Mitchell SG, Blue TR, Vocci FJ, Fishman MJ, Murphy SM, Couvillion K, Maher K, Ryan DA, & Wenzel K (2021). A clinical protocol of a comparative effectiveness trial of extended-release naltrexone versus extended-release buprenorphine with individuals leaving jail. Journal of Substance Abuse Treatment, 128, 108241. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Grosse SD (2008). Assessing cost-effectiveness in healthcare: history of the $50,000 per QALY threshold. Expert Rev Pharmacoecon Outcomes Res, 8(2), 165–178. 10.1586/14737167.8.2.165 [DOI] [PubMed] [Google Scholar]
  18. Glick Henry A. J. A. D., Sonnad Seema S., Daniel Polsky. (2014). Economic evaluation in clinical trials (2nd Edition). OUP Oxford. 10.1093/med/9780199685028.001.0001 [DOI] [Google Scholar]
  19. Jalali A, Ryan DA, McCollister KE, Marsch LA, Schackman BR, & Murphy SM (2020). Economic evaluation in the National Drug Abuse Treatment Clinical Trials Network: Past, present, and future. J Subst Abuse Treat, 112S, 18–27. 10.1016/j.jsat.2020.02.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jalali A, Tamimi RM, McPherson SM, & Murphy SM (2022). Econometric issues in prospective economic evaluations alongside clinical trials: Combining the nonparametric bootstrap with methods that address missing data. Epidemiological Reviews, (In Press). [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Jarvis BP, Holtyn AF, Subramaniam S, Tompkins DA, Oga EA, Bigelow GE, & Silverman K (2018). Extended-release injectable naltrexone for opioid use disorder: a systematic review. Addiction, 113(7), 1188–1209. 10.1111/add.14180 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jeng PJ, Jalali A, Ryan DA, Poole SA, Vocci FJ, Gordon MS, Woody GE, Polsky D, & Murphy SM (2021, October 15). The cost of providing extended-release naltrexone treatment to pre-release prisoners with opioid use disorder Addiction Health Services Research, Virtual. [Google Scholar]
  23. Krawczyk N, Picher CE, Feder KA, & Saloner B (2017). Only One In Twenty Justice-Referred Adults In Specialty Treatment For Opioid Use Receive Methadone Or Buprenorphine. Health Aff (Millwood), 36(12), 2046–2053. 10.1377/hlthaff.2017.0890 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Krupitsky E, Nunes EV, Ling W, Illeperuma A, Gastfriend DR, & Silverman BL (2011). Injectable extended-release naltrexone for opioid dependence: a double-blind, placebo-controlled, multicentre randomised trial. Lancet, 377(9776), 1506–1513. 10.1016/S0140-6736(11)60358-9 [DOI] [PubMed] [Google Scholar]
  25. Lee JD, Friedmann PD, Kinlock TW, Nunes EV, Boney TY, Hoskinson RA Jr., Wilson D, McDonald R, Rotrosen J, Gourevitch MN, Gordon M, Fishman M, Chen DT, Bonnie RJ, Cornish JW, Murphy SM, & O’Brien CP (2016). Extended-Release Naltrexone to Prevent Opioid Relapse in Criminal Justice Offenders. N Engl J Med, 374(13), 1232–1242. 10.1056/NEJMoa1505409 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lee JD, McDonald R, Grossman E, McNeely J, Laska E, Rotrosen J, & Gourevitch MN (2015). Opioid treatment at release from jail using extended-release naltrexone: a pilot proof-of-concept randomized effectiveness trial. Addiction, 110(6), 1008–1014. 10.1111/add.12894 [DOI] [PubMed] [Google Scholar]
  27. Lee JD, Nunes EV Jr., Novo P, Bachrach K, Bailey GL, Bhatt S, Farkas S, Fishman M, Gauthier P, Hodgkins CC, King J, Lindblad R, Liu D, Matthews AG, May J, Peavy KM, Ross S, Salazar D, Schkolnik P, . . . Rotrosen J. (2018). Comparative effectiveness of extended-release naltrexone versus buprenorphine-naloxone for opioid relapse prevention (X:BOT): a multicentre, open-label, randomised controlled trial. Lancet, 391(10118), 309–318. 10.1016/S0140-6736(17)32812-X [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Leshner AI, & Dzau VJ (2019). Medication-Based Treatment to Address Opioid Use Disorder. JAMA, 321(21), 2071–2072. 10.1001/jama.2019.5523 [DOI] [PubMed] [Google Scholar]
  29. Lincoln T, Johnson BD, McCarthy P, & Alexander E (2018). Extended-release naltrexone for opioid use disorder started during or following incarceration. Journal of Substance Abuse Treatment, 85, 97–100. [DOI] [PubMed] [Google Scholar]
  30. Sobell Linda C. M. B. S. (2007). Timeline Follow-Back: A technique for assessing self-reported alcohol consumption. (Measuring Alcohol Consumption; ) [Google Scholar]
  31. Mattson CL, O’Donnell J, Kariisa M, Seth P, Scholl L, & Gladden RM (2018). Opportunities to Prevent Overdose Deaths Involving Prescription and Illicit Opioids, 11 States, July 2016-June 2017. MMWR Morb Mortal Wkly Rep, 67(34), 945–951. 10.15585/mmwr.mm6734a2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. McLellan AT, Kushner H, Metzger D, Peters R, Smith I, Grissom G, Pettinati H, & Argeriou M (1992). The Fifth Edition of the Addiction Severity Index. J Subst Abuse Treat, 9(3), 199–213. 10.1016/0740-5472(92)90062-s [DOI] [PubMed] [Google Scholar]
  33. Murphy SM, Campbell AN, Ghitza UE, Kyle TL, Bailey GL, Nunes EV, & Polsky D (2016). Cost-effectiveness of an internet-delivered treatment for substance abuse: Data from a multisite randomized controlled trial. Drug Alcohol Depend, 161, 119–126. 10.1016/j.drugalcdep.2016.01.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Murphy SM, Jeng PJ, Poole SA, Jalali A, Vocci FJ, Gordon MS, Woody GE, & Polsky D (2020). Health and economic outcomes of treatment with extended-release naltrexone among pre-release prisoners with opioid use disorder (HOPPER): protocol for an evaluation of two randomized effectiveness trials. Addict Sci Clin Pract, 15(1), 15. 10.1186/s13722-020-00188-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Murphy SM, McCollister KE, Leff JA, Yang X, Jeng PJ, Lee JD, Nunes EV, Novo P, Rotrosen J, & Schackman BR (2019). Cost-Effectiveness of Buprenorphine-Naloxone Versus Extended-Release Naltrexone to Prevent Opioid Relapse. Ann Intern Med, 170(2), 90–98. 10.7326/M18-0227 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Murphy SM, & Polsky D (2016). Economic Evaluations of Opioid Use Disorder Interventions. Pharmacoeconomics, 34(9), 863–887. 10.1007/s40273-016-0400-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Murphy SM, Polsky D, Lee JD, Friedmann PD, Kinlock TW, Nunes EV, Bonnie RJ, Gordon M, Chen DT, Boney TY, & O’Brien CP (2017). Cost-effectiveness of extended release naltrexone to prevent relapse among criminal justice-involved individuals with a history of opioid use disorder. Addiction, 112(8), 1440–1450. 10.1111/add.13807 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. National Academies of Sciences, E., and Medicine. Medications for opioid use disorder save lives. National Academies Press. [PubMed] [Google Scholar]
  39. Neumann PJ, Cohen JT, & Weinstein MC (2014). Updating cost-effectiveness--the curious resilience of the $50,000-per-QALY threshold. N Engl J Med, 371(9), 796–797. 10.1056/NEJMp1405158 [DOI] [PubMed] [Google Scholar]
  40. Neumann PJ, Ganiats TG, Russell LB, Sanders GD, & S JA. (2016). Cost-effectiveness in health and medicine. Oxford University Press. 10.1093/acprof:oso/9780190492939.001.0001 [DOI] [Google Scholar]
  41. Nunn A, Zaller N, Dickman S, Trimbur C, Nijhawan A, & Rich JD (2009). Methadone and buprenorphine prescribing and referral practices in US prison systems: results from a nationwide survey. Drug Alcohol Depend, 105(1–2), 83–88. 10.1016/j.drugalcdep.2009.06.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Onuoha EN, Leff JA, Schackman BR, McCollister KE, Polsky D, & Murphy SM (2021). Economic Evaluations of Pharmacologic Treatment for Opioid Use Disorder: A Systematic Literature Review. Value Health, 24(7), 1068–1083. 10.1016/j.jval.2020.12.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Neumann Peter J., G. TG, Russell Louise B., Sanders Gillian D., and Siegel Joanna E.. (2016). Cost-effectiveness in health and medicine. Oxford University Press. 10.1093/acprof:oso/9780190492939.001.0001 [DOI] [Google Scholar]
  44. Pizzicato LN, Drake R, Domer-Shank R, Johnson CC, & Viner KM (2018). Beyond the walls: Risk factors for overdose mortality following release from the Philadelphia Department of Prisons. Drug Alcohol Depend, 189, 108–115. 10.1016/j.drugalcdep.2018.04.034 [DOI] [PubMed] [Google Scholar]
  45. Polsky D, Glick HA, Yang J, Subramaniam GA, Poole SA, & Woody GE (2010). Cost-effectiveness of extended buprenorphine-naloxone treatment for opioid-dependent youth: data from a randomized trial. Addiction, 105(9), 1616–1624. 10.1111/j.1360-0443.2010.03001.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Ranapurwala SI, Shanahan ME, Alexandridis AA, Proescholdbell SK, Naumann RB, Edwards D Jr., & Marshall SW (2018). Opioid Overdose Mortality Among Former North Carolina Inmates: 2000–2015. Am J Public Health, 108(9), 1207–1213. 10.2105/AJPH.2018.304514 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. SAMHSA/CSAT Treatment Improvement Protocols. (2018). In Medications for Opioid Use Disorder: For Healthcare and Addiction Professionals, Policymakers, Patients, and Families. Substance Abuse and Mental Health Services Administration (US). [PubMed] [Google Scholar]
  48. Schomaker M, & Heumann C (2018). Bootstrap inference when using multiple imputation. Stat Med, 37(14), 2252–2266. 10.1002/sim.7654 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Sigmon SC, Bisaga A, Nunes EV, O’Connor PG, Kosten T, & Woody G (2012). Opioid detoxification and naltrexone induction strategies: recommendations for clinical practice. Am J Drug Alcohol Abuse, 38(3), 187–199. 10.3109/00952990.2011.653426 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Sterne JA, White IR, Carlin JB, Spratt M, Royston P, Kenward MG, Wood AM, & Carpenter JR (2009). Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ, 338, b2393. 10.1136/bmj.b2393 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. US Department of Veterans Affairs Federal Supply Schedule (2020). https://www.va.gov/opal/nac/fss/pharmPrices.asp
  52. Wakeman SE, & Rich JD (2015). Addiction Treatment Within U.S. Correctional Facilities: Bridging the Gap Between Current Practice and Evidence-Based Care. J Addict Dis, 34(2–3), 220–225. 10.1080/10550887.2015.1059217 [DOI] [PubMed] [Google Scholar]
  53. Wang W, Gellings Lowe N, Jalali A, & Murphy SM (2021). Economic modeling of reSET-O, a prescription digital therapeutic for patients with opioid use disorder. J Med Econ, 24(1), 61–68. 10.1080/13696998.2020.1858581 [DOI] [PubMed] [Google Scholar]
  54. Woody GE, Poole S, Yu E, Carroll J, & Lynch KG (2021). Extended release injectable naltrexone before vs. after release: A randomized trial of opioid addicted persons who are in prison. J Subst Abuse Treat, 127, 108355. 10.1016/j.jsat.2021.108355 [DOI] [PubMed] [Google Scholar]
  55. Zaller N, Donadeo K, Coffey J, Zielinski M, & Brinkley-Rubinstein L (2019). Screening for Opioid Use Disorder in the Largest Jail in Arkansas: A Brief Report. J Correct Health Care, 25(3), 214–218. 10.1177/1078345819852133 [DOI] [PubMed] [Google Scholar]
  56. Zarkin GA, Orme S, Dunlap LJ, Kelly SM, Mitchell SG, O’Grady KE, & Schwartz RP (2020). Cost and cost-effectiveness of interim methadone treatment and patient navigation initiated in jail. Drug and alcohol dependence, 217, 108292. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1833574-supplement-1.pdf (45.3KB, pdf)
2
NIHMS1833574-supplement-2.docx (40.2KB, docx)

RESOURCES