1. INTRODUCTION

The epidemic of opioid overdose is a growing crisis in the US. In 2016, 42,249 Americans died of an opioid-related overdose, more than any prior year in recorded history (Seth et al., 2018). This evolving crisis has disproportionately affected persons with recent experiences of incarceration (Binswanger et al., 2013; Brehm Christensen et al., 2006; Bukten et al., 2017; Huang et al., 2011; Merrall et al., 2010; Winter et al., 2015). In the US, the two-week period following release from prison or jail is associated with a 129-fold increase in the risk of death due to overdose relative to the general population (Binswanger et al., 2007). While medications for opioid use disorder (OUD)—including methadone, buprenorphine, and depot naltrexone—are the most effective treatment, they are not routinely available in most US prison and jails. There have been recent calls to ensure access to treatment within prisons and jails from health advocates, scientists, policy makers, and legal experts. The National Academies of Sciences recently stated that withholding these medications is unethical (Leshner and Mancher, 2019), and the Law Enforcement Action Partnership, a collection of current and former elected sheriffs, prosecutors, and other law enforcement professionals, has publicly called for the provision of medications for OUD within correctional facilities (Law Enforcement Action Partnership (LEAP), 2019). While there is increased support for provision of medications for OUD within correctional settings, available data in the US are currently limited and preliminary (Green et al., 2018).

In England, investigators documented a 31% reduction in drug-related poisoning deaths post-release among those who received medications for OUD while incarcerated compared to those who were not (Marsden et al., 2017). In New South Wales, Australia, investigators documented a 74% reduction in accidental drug-induced deaths when comparing those who had received medications for OUD while incarcerated and retained in treatment post-release to those who had not (Degenhardt et al., 2014). In the present study, we sought to estimate the expected reduction in opioid-related overdose deaths if wide scale uptake of screening and medications for OUD had occurred in all US prisons and jails in 2016; we make the assumption that similar program effectiveness would be achieved and that reductions in opioid-related overdose mortality in the US would be similar to those observed in England and Australia. For this analysis, we developed a model using data from several national and international sources. Using data from England, we conservatively estimated of the number of lives that might be saved by provision of medications for OUD in prisons and jails alone, and using data from Australia, we calculate a more ambitious estimate of the number of lives that might be saved by availability of medications for OUD in prisons and jails and post-release retention in treatment.

2. MATERIAL AND METHODS

We estimate the reduction in opioid-related overdose mortality in each state if wide scale uptake of screening and medications for OUD had occurred in all US prisons and jails in 2016. We produce estimates for two counterfactual scenarios: (1) all persons who are clinically indicated receive medications for OUD while incarcerated, and (2) all persons who are clinically indicated receive medications for OUD while incarcerated and are retained in treatment post-release. The steps below were used to generate estimates for each counterfactual scenario.

First, we used data from the National Center for Health Statistics database to determine the total number of opioid overdose deaths for each state in 2016. Opioid-related overdose deaths were defined as having the following International Classification of Diseases, 10^th Revision (ICD-10) codes: opium (T40.0), heroin (T40.1), natural and semisynthetic opioids (T40.2), methadone (T40.3), synthetic opioids other than methadone (T40.4), or other and unspecified narcotics (T40.6) as a contributing cause. We then multiplied the total number of opioid-related overdose deaths in each state by the proportion expected to occur during the one year post-release period (fORM_postrelease). This proportion was derived from several empirical sources, including estimates from North Carolina, Pennsylvania, Washington, and Rhode Island, and ranged from 3% to 25% (Binswanger et al., 2013; Binswanger et al., 2007; Green et al., 2018; Pizzicato et al., 2018; Ranapurwala et al., 2018). To reflect the uncertainty in this model parameter, we performed a Monte Carlo simulation which drew values from a uniform distribution ranging from 3 to 25%, as these values reflect the upper and lower bounds in published data. This simulation process was repeated 10,000 times for each state. The following equation describes this process:

where N_iORM is the reported number of opioid overdose deaths within the state, fORM_postrelease is the proportion of deaths expected to occur during the post-release period, and N_iORM_postrelease is the resulting estimate for the number of deaths occurring during the post-release period within the state.

Second, we calculated the number of overdose deaths expected under a counterfactual scenario where all persons who are clinically indicated receive medications for OUD while incarcerated in 2016. We created a normal distribution representing the potential percent reduction in opioid overdose mortality attributable to provision of medications for OUD in correctional settings (fMOUD_effect). This distribution was derived using data from England reporting a 31% (standard deviation [SD]: 17.3) reduction in drug-related poisoning deaths post-release when comparing those who received medications for OUD while incarcerated to those who had not. We implemented stochastic processes to select the percent reduction from the normal distribution using Monte Carlo simulations. This process was repeated 10,000 times for each state to yield the expected number of opioid-related overdose deaths among individuals within a year post-release, had wide scale uptake of screening and medications for OUD occurred in prisons and jails in 2016.

We used the difference between this estimate and the previously calculated estimate of the number of opioid-related overdose deaths post-release in which no medications for OUD programs were implemented (broadly reflecting real-world conditions in 2016) to estimate the number of lives saved attributable to wide scale uptake of screening and medications for OUD in prisons and jails, or averted mortality (N_iORM_averted):

A national estimate was derived by summing state-specific estimates.

To estimate the number of lives saved per 10,000 persons incarcerated, we divided the simulated number of lives saved in each state (N_iORM_averted) by the state-specific combined prison and jail population point prevalence values (N_iIncarcerated), and multiplied these values by 10,000 to yield the number of lives saved per 10,000 persons incarcerated (N_iORM_{averted per 10K}):

The steps above were repeated using a second normal distribution representing the potential percent reduction in opioid overdose mortality attributable to provision of medications for OUD in correctional settings and retention on treatment post-release (fMOUD_effect). This distribution was derived using data from Australia reporting a 74% (SD: 6.5) reduction in accidental drug-induced deaths when comparing those who had received medications for OUD while incarcerated and were retained in treatment post-release to those who were not.

R Studio was used to conduct the Monte Carlo simulations and produce all maps. All mean estimates are reported using simulation intervals (SI) which accounts for stochastic processes by reporting the 95% upper and lower limits of the simulated output. As a simulation study involving aggregate level data, this analysis did not require oversight from an institutional review board.

3. RESULTS

The estimated number of lives saved and lives saved per 10,000 persons incarcerated among persons with recent incarceration if wide scale uptake of screening and medications for OUD had occurred in US prisons and jails, by state in 2016, and stratified by estimates with receiving treatment alone and receiving treatment with post-release retention are reported in Table 1. Under the first scenario, the model predicts that if all persons who were clinically indicated had received medications for OUD while incarcerated in 2016, approximately 1,840 (95% SI: −2,757 – 4,959) lives would have been saved nationally. We also estimated that 668 (95% SI: −1,008 – 1,812) lives would be saved per 10,000 persons incarcerated.

𝑁𝑖⁢𝑂𝑅𝑀*𝑓𝑂𝑅𝑀𝑝𝑜𝑠𝑡𝑟𝑒𝑙𝑒𝑎𝑠𝑒=𝑁𝑖⁢𝑂𝑅⁢𝑀𝑝𝑜𝑠𝑡𝑟𝑒𝑙𝑒𝑎𝑠𝑒𝑁𝑖⁢𝑂𝑅⁢𝑀𝑝𝑜𝑠𝑡𝑟𝑒𝑙𝑒𝑎𝑠𝑒*𝑓𝑀𝑂𝑈𝐷𝑒𝑓𝑓𝑒𝑐𝑡=𝑁𝑖⁢𝑂𝑅⁢𝑀𝑝𝑜𝑠𝑡𝑟𝑒𝑙𝑒𝑎𝑠𝑒𝑤𝑖𝑡ℎ𝑀𝑂𝑈𝐷𝑁𝑖⁢𝑂𝑅⁢𝑀𝑝𝑜𝑠𝑡𝑟𝑒𝑙𝑒𝑎𝑠𝑒−𝑁𝑖⁢𝑂𝑅⁢𝑀𝑝𝑜𝑠𝑡𝑟𝑒𝑙𝑒𝑎𝑠𝑒𝑤𝑖𝑡ℎ𝑀𝑂𝑈𝐷=𝑁𝑖⁢𝑂𝑅⁢𝑀𝑎𝑣𝑒𝑟𝑡𝑒𝑑(𝑁𝑖⁢𝑂𝑅⁢𝑀𝑎𝑣𝑒𝑟𝑡𝑒𝑑𝑁𝑖⁢𝐼𝑛𝑐𝑎𝑟𝑐𝑒𝑟𝑎𝑡𝑒𝑑)*10,000=𝑁𝑖⁢𝑂𝑅⁢𝑀𝑎𝑣𝑒𝑟𝑡𝑒𝑑𝑝𝑒𝑟10⁢𝐾

Under the second scenario, the model predicts that if all persons who were clinically indicated had received medications for OUD while incarcerated and were retained in treatment post-release in 2016, 4,400 (95% SI: 2,675 – 5,557) lives would have been saved nationally, and 1,609 (95% SI: 972 – 2,037) lives would be saved per 10,000 persons incarcerated. The estimated number of lives saved and per 10,000 persons incarcerated if wide scale uptake of screening and medications for OUD had occurred in US prisons and jails, by state in 2016, and stratified by estimates with receiving treatment alone (panels A and B) and receiving treatment with post-release retention (panels C and D) are presented graphically in Figure 1.

4. DISCUSSION

Our model projected that if wide scale uptake of screening and medications for OUD had occurred in US prisons and jails in 2016, an estimated 1,840 lives would have been saved by provision of medications for OUD in prisons and jails alone, and 4,400 lives would have been saved by provision of medications for OUD in prisons and jails and programs to ensure post-release retention in treatment. For perspective, in a single year, the number of deaths that might be averted by provision of medications for OUD in prisons and jails with post-release retention in treatment is estimated to be greater than the number of lives lost in the War in Afghanistan (2,216 lives), in Pearl Harbor (2,403 lives), and in the terrorist attacks on 9/11 (2,996 lives). Our estimates also represent approximately 4-10% of all opioid overdose deaths in 2016 (Scholl et al., 2019). By comparison, prescription drug monitoring programs were expected to prevent 600 overdose deaths in 2016—or 1.4% of all opioid overdose deaths in the same year (Patrick et al., 2016).

The current model examined the impact of screening and medications for OUD programs on mortality alone. Additional benefits to expanding access to medications for OUD in correctional settings have been documented, including post-release reductions in non-medical opioid use, nonfatal overdose risk, and criminal behavior (Brinkley-Rubinstein et al., 2018; Lee et al., 2016; Rich et al., 2015; Sharma et al., 2016), as well as improved perceptions of prison and jail safety and facility environment (Brinkley-Rubinstein et al., 2019). Undertreating OUD in correctional settings also has complex socioeconomic implications, including higher post-release healthcare costs, criminal justice costs, and overall community burden (Kinner and Wang, 2014). In addition to the number of human lives that might be preserved, jurisdictions might consider these supplementary benefits in weighing their decision to expand access to medications for OUD.

Current findings underscore the importance of post-release retention in treatment among persons who receive medications for OUD while incarcerated, as well as the need for concurrent interventions that mitigate overdose risk for this population. Naloxone distribution to individuals at release from incarceration (Bird et al., 2016) and to facility visitors (Huxley-Reicher et al., 2018) have demonstrated promise in reducing overdose mortality. Critically, pre-arrest diversion to community-based medications for OUD programs produces favorable outcomes for persons with OUD—in the absence of the myriad of harms that are associated with incarceration (Freudenberg and Heller, 2016).

Several limitations should be noted. First, our estimated rate of reductions in opioid-related mortality are derived from studies in England and Australia; these settings may differ from the US in critical ways (e.g., treatment capacity, healthcare access, medication treatments available) that may lead to increased or decreased program effects in different settings. Second, our estimates rely on opioid mortality data reported by the National Center for Health Statistics. These values are known to underestimate the true number of opioid-related deaths, as toxicological laboratory tests and the circumstances under which these tests are performed can vary by jurisdiction (Rudd, 2016); therefore, the estimates presented may be conservative. Third, limited data are available to estimate the proportion of opioid-related overdose deaths among individuals with prior-year incarceration. The data used to inform our input parameter were derived from studies that examined mortality rates among persons released from US prison systems; however, the risk of opioid overdose death post-release may differ for individuals who are released from jails. Fourth, medications for OUD are currently available in a limited number of prisons and jails nationwide. Our model assumes that the 2016 “standard of care” in prisons and jails (i.e., no access to medications for OUD) is applied nationwide; therefore, our model may overestimate the number of lives that might be saved in those jurisdictions that did provide medications for OUD to individuals who are incarcerated during this year. However, this limitation is somewhat minor, as very few prisons and jails nationwide provided access to medications for OUD to persons who were incarcerated during that time.

5. CONCLUSIONS

Our model projected that wide scale uptake of screening and treatment with medications for OUD in prison and jail settings would produce substantial reductions in opioid-related overdose deaths. These findings can be used by lawmakers, state public health officials, and prison and jail administrators when considering investments and potential impacts of interventions that can reduce overdose for persons who are incarcerated and for their overall jurisdiction, particularly in those states identified as having the highest numbers of potential deaths averted. Jurisdictions that lack community capacity for medications for OUD should build and expand these resources in tandem with prison and jail-based provision of medications for OUD to ensure that post-release retention in treatment can be attained.

Summary

A quantitative model was developed to estimate the potential reduction in opioid-related overdose deaths in the United States if widespread screening and medication-assisted treatment (MAT) for opioid use disorder (OUD) were implemented in all prisons and jails during 2016. Two scenarios were modeled: MAT during incarceration only, and MAT during incarceration with post-release treatment retention.

Methodology

The model utilized data from the National Center for Health Statistics to determine the total number of opioid overdose deaths per state in 2016. This was then combined with data from studies in England and Australia detailing the reduction in drug-related mortality following MAT implementation in correctional facilities. Monte Carlo simulations were employed to account for uncertainty in the proportion of overdose deaths occurring post-release and the effectiveness of MAT programs. Two distinct reduction rates were used, one for MAT alone during incarceration, and a second, larger rate for MAT during incarceration along with post-release treatment continuation.

Results

The model projected that if MAT during incarceration had been widely adopted in 2016, approximately 1,840 lives (95% SI: −2,757 – 4,959) could have been saved nationally. With the addition of post-release treatment retention, this number increased to an estimated 4,400 lives (95% SI: 2,675 – 5,557) saved nationally. State-level estimates were also generated, and these results are shown graphically elsewhere.

Discussion

The findings suggest a considerable potential for reducing opioid overdose deaths through the implementation of widespread MAT programs within correctional facilities. The model’s projections exceed the number of lives lost in major historical events, highlighting the scale of the potential impact. The significant positive effect of post-release treatment continuation is emphasized. This study focused on mortality; additional benefits of MAT, such as reduced post-release non-medical opioid use, reduced recidivism, and improved perceptions of correctional environments, were also noted. The model also accounts for several limitations and acknowledges uncertainties in the data sources used.

Conclusions

The model demonstrates substantial potential for decreasing opioid-related deaths through the expansion of screening and MAT programs within correctional facilities. The significant impact of post-release treatment retention underscores the importance of integrated community-based treatment services to fully realize the benefits of this strategy. These results can inform policy decisions and resource allocation regarding OUD treatment within correctional systems and related community services.

Summary

A model was developed to estimate the potential reduction in opioid-related overdose deaths in the US if widespread screening and medication-assisted treatment (MAT) for opioid use disorder (OUD) were implemented in prisons and jails. The study utilized data from England and Australia, which demonstrated significant reductions in drug-related deaths post-release among individuals who are incarcerated and that received MAT. Two scenarios were modeled: MAT during incarceration only, and MAT during incarceration with continued treatment post-release.

Methodology

The model estimated the number of opioid overdose deaths in each US state in 2016, focusing on deaths occurring within one year post-release. This was achieved using data from the National Center for Health Statistics and published estimates of post-release overdose risk. Monte Carlo simulations were employed to account for uncertainty in the proportion of post-release deaths. Two counterfactual scenarios were simulated: (1) widespread MAT access during incarceration, and (2) widespread MAT access during incarceration with post-release treatment retention. Data from England (31% reduction in drug-related deaths) and Australia (74% reduction) informed the model's estimates of mortality reduction associated with MAT.

Results

Under the first scenario (MAT during incarceration only), the model projected approximately 1,840 (95% simulation interval: −2,757 – 4,959) lives saved nationally and 668 (95% SI: −1,008 – 1,812) lives saved per 10,000 individuals that were incarcerated. The second scenario (MAT during incarceration with post-release retention) projected 4,400 (95% SI: 2,675 – 5,557) lives saved nationally and 1,609 (95% SI: 972 – 2,037) lives saved per 10,000 individuals that were incarcerated.

Discussion

The findings suggest a substantial potential for reducing opioid overdose deaths through widespread MAT implementation in correctional facilities. The projected number of lives saved under the second scenario exceeded the number of lives lost in significant historical events. The model considered only mortality reduction; additional benefits of MAT in correctional settings include reduced post-release opioid use, decreased overdose risk, and improved facility safety perceptions. The importance of post-release treatment retention is highlighted, along with supplementary interventions like naloxone distribution.

Limitations and Conclusions

Several limitations exist, including the use of data from different countries, potential underestimation of opioid-related deaths in the US, and variation in post-release overdose risk among individuals released from prisons versus jails. Despite these limitations, the model strongly suggests that widespread MAT implementation in prisons and jails could significantly reduce opioid-related overdose deaths. This information is valuable for policymakers and correctional administrators considering investments to reduce overdose mortality, particularly in states with the highest projected reductions. The necessity of building community-based MAT capacity alongside prison-based programs is emphasized to ensure successful post-release treatment.

Summary

The opioid overdose crisis in the US is severe, disproportionately impacting individuals who are recently incarcerated. The two weeks after release from prison show a drastically increased risk of overdose death. Medications for opioid use disorder (OUD) are highly effective, yet not widely available in US prisons and jails. This study estimates the potential reduction in overdose deaths if these medications were provided.

Estimating Potential Lives Saved

This study used data from England and Australia, which showed significant decreases in drug-related deaths post-release among individuals who are incarcerated and receiving OUD medications. The researchers created a model to estimate the potential impact in US prisons and jails in 2016, considering two scenarios: (1) providing medications to those clinically indicated while incarcerated, and (2) providing medications while incarcerated and maintaining treatment post-release.

Methodology

The model used data on opioid overdose deaths from the National Center for Health Statistics, focusing on deaths within one year post-release. It incorporated data from multiple sources to determine the proportion of post-release deaths and the reduction in overdose deaths achievable through OUD medication. The model used Monte Carlo simulations to account for uncertainty in these values, generating estimates for each state.

Results

The model projected that providing OUD medications in prisons and jails alone could have saved approximately 1,840 lives nationally in 2016. If treatment continued post-release, this number could rise to 4,400 lives saved. These numbers are comparable to the number of lives lost in major historical events.

Discussion and Limitations

The potential benefits extend beyond reduced mortality, including decreased opioid use, lower overdose risk, reduced criminal behavior, and improved prison environments. The study acknowledges limitations, including differences between the US and the countries used for comparison data, underreporting of opioid deaths, and variations in overdose risk between prison and jail releases. The study also assumes a baseline of zero OUD medication access, which may overestimate potential savings in areas with existing programs.

Summary

Many people die from opioid overdoses in the US each year. People recently released from jail or prison are at much higher risk of dying from an overdose. Medications can help people avoid overdoses, but these medications aren't always available in jails and prisons. This study looked at how many lives could be saved if these medications were available.

How the Study Worked

Researchers used information from England and Australia, where providing these medications in jails and prisons has helped reduce overdose deaths. They used this information to estimate how many lives could be saved in the US if similar programs were in place. They looked at two scenarios: (1) giving the medications to those who need them while they're in jail or prison, and (2) giving the medications in jail or prison and helping those people stay on the medication after they're released.

Results

The study found that if everyone who needed the medication got it while in jail or prison, about 1,840 lives might have been saved in 2016. If people also got help staying on the medication after release, that number could increase to 4,400 lives saved. That's more lives than were lost in some major events in history!

What This Means

This study shows that giving people medication for opioid use disorder in jails and prisons could save many lives. It's important to help people stay on their medication after they leave jail or prison too. More research is needed, but this study shows the potential to save many lives and reduce overdoses.

Important Considerations

The study has some limitations. The numbers are based on data from other countries, and the actual number of lives saved might be different in the US. Also, the number of opioid overdose deaths is probably higher than reported, meaning that the study's estimate may be lower than the actual number of lives that could be saved. Despite these limitations, the study still shows the huge potential of these programs.