Abstract
Background and Aims
Opioid overdose is a leading cause of death during the immediate time after release from jail or prison. Most jails in the United States do not provide methadone and buprenorphine treatment for opioid use disorder (MOUD), and research in estimating its impact in jail settings is limited. We aimed to test the hypothesis that in-jail MOUD is associated with lower overdose mortality risk post-release.
Design, Setting and Participants
Retrospective, observational cohort study of 15 797 adults with opioid use disorder who were released from New York City jails to the community in 2011–17. They experienced 31 382 incarcerations and were followed up to 1 year.
Measurements
The primary outcomes were death caused by accidental drug poisoning and all-cause death. The exposure was receipt of MOUD (17 119 events) versus out-of-treatment (14 263 events) during the last 3 days before community reentry. Covariates included demographic, clinical, behavioral, housing, healthcare utilization, and legal characteristics variables. We performed multivariable, mixed-effect Cox regression analysis to test association between in-jail MOUD and deaths.
Findings
A majority were male (82%) and their average age was 42 years. Receiving MOUD was associated with misdemeanor charges, being female, injection drug use, and homelessness. During 1 year post-release, 111 overdose deaths occurred, and crude death rates were 0.49 and 0.83 per 100 person-years for in-jail MOUD and out-of-treatment groups, respectively. Accounting for confounding and random effects, in-jail MOUD was associated with lower overdose mortality risk (adjusted hazard ratio = 0.20, 95% CI = 0.08–0.46), and all-cause mortality risk (adjusted hazard ratio = 0.22, 95% CI = 0.11–0.42) for the first month post-release.
Conclusions
Methadone and buprenorphine treatment for opioid use disorder during incarceration was associated with an 80% reduction in overdose mortality risk for the first month post-release.
Keywords: Jail, drug-related mortality, all-cause mortality, opioid use disorder, medication for opioid use disorder, urban population
INTRODUCTION
Opioid overdose is a leading cause of death during the immediate time after release from jail or prison [1–5]. This indicates an acute risk of overdose mortality post-release among individuals with opioid use disorder (OUD) who disproportionately experience incarceration in the United States (about 15–20%) [6,7]. Methadone and buprenorphine, two medications for opioid use disorder (MOUD), are proven interventions that reduce overdose risk among people with OUD. Studies from Australia, Taiwan, Scotland, and England show that MOUD provided during incarceration is associated with lower mortality after release from prison [8–11]. For example, Marsden and colleagues examined mortality risk by specific post-release time periods among people with OUD incarcerated in England and found that in-prison MOUD substantially reduced risk of all-cause and overdose mortality within one-month post-release [11].
Notwithstanding this evidence, MOUD remains largely unavailable in the United States (US) correctional settings due to regulatory and logistical reasons. Stigma is another major barrier to implementation of MOUD. A widely held misunderstanding is that MOUD replaces one addiction with another [12, 13], and drug addiction is moral failing rather than an illness [14]. This is true not just in prisons, but also in jails, where individuals are incarcerated for shorter periods of time for pre-trial detention and misdemeanor convictions [15,16]. Because MOUD programming is scarce, research on MOUD’s impact on mortality and other outcomes in US jail populations is limited, which in turn may hinder wider implementation. To address this gap, we assessed the impact of the in-jail MOUD program in New York City (NYC) jails, the oldest and largest correctional-based licensed opioid treatment program in the US. Using matched health and administrative data, we tested the hypothesis that in-jail MOUD for people with OUD was associated with lower risk of overdose mortality after release from NYC jails to the community during May 2011-December 2017.
METHODS
Setting
This retrospective, observational cohort study is part of a larger study involving researchers, clinicians, and public health practitioners from academic institutions, NYC Department of Health and Mental Hygiene (DOHMH), and NYC Health + Hospitals/Correctional Health Services (CHS) to evaluate impact of jail-based MOUD on mortality and other health outcomes post jail release [17].
NYC’s jail system received over 85 000 admissions in 2011 and nearly 56 000 in 2017 [18]. As the provider of healthcare, social work services, and discharge planning for NYC jails, CHS medical staff perform a comprehensive medical intake evaluation, including substance use screening to identify current use or misuse of substances (e.g., frequency, type), risks of withdrawal from opioid, alcohol, and other drugs based on assessment of self-reported behaviors, self-reported (and where possible confirmatory) enrollment in community MOUD programming, urine toxicology tests, or other OUD or MOUD records. Additional screening by providers may take place at clinical encounters after intake. Individuals with positive screening results for opioid use based on the totality of these factors receive an OUD diagnosis, which is documented in the record by International Classification of Diseases, Ninth and Tenth revisions, Clinical Modification (ICD-9-CM, ICD-10-CM) codes which serve as a marker of current OUD in the electronic medical records. Individuals with a recorded OUD diagnosis may then be eligible for jail-based MOUD programming.
Until late 2017, MOUD eligibility was based on legal charges. Though all individuals with OUD were offered short course methadone or buprenorphine for the treatment of acute opioid withdrawal during incarceration, they were ineligible for MOUD maintenance if they had felony charges, parole violations, or other active warrants which could result in discharge to state prison or other custody, partly due to lack of access to MOUD in New York State (NYS) prisons. Ineligible individuals who had been enrolled in community MOUD would be slowly tapered off MOUD (“withdrawal-only”). Eligible individuals, by contrast, were offered voluntary enrollment in the MOUD program, which involved induction or continuation on methadone or buprenorphine for maintenance along with linkage to community (“maintenance”). Change in legal status during incarceration could change eligibility. In late 2017, this legal eligibility impediment was removed by CHS.
Data sources
From their electronic medical records, CHS extracted identifiers, incarceration, and health records for adults aged ≥18 years who were admitted and released from NYC jails during May 1, 2011 through December 31, 2017 and had an OUD diagnosis. These data were transferred to DOHMH, and probabilistically matched with the NYC death certificate data using QualityStage Software (IBM). Additionally, the identifiers were deterministically matched with NYS Medicaid claims and Statewide Planning and Research Cooperative System (SPARCS) hospitalization and emergency department (ED) records [19]. An independent human review of 300 match cases (0.7%) and 300 unmatched counterfactuals (0.2%), which were randomly selected using stratified sampling with stratification by matching algorithm and data sources, determined that matching quality was acceptable (sensitivity=97%; specificity=96%).
The Institutional Review Boards of DOHMH (18–106), two academic institutions (i18–00445 and 1811019740), and Biomedical Research Alliance of New York (19‐PRS‐156‐419(HHC)) reviewed and approved this study. It also received certification from the Office for Human Research Protections of the US Department of Health and Human Services. The analysis was not pre-registered and the results should be considered exploratory.
Study population
The study sample comprised of 79 115 incarceration events from 29 566 adults. Of these, we excluded incarceration events with non-community discharge or death in custody (Fig 1). We also excluded incarceration events where no MOUD was provided during the entire incarceration largely because OUD was deemed not current. We further excluded events where only withdrawal was provided for the last three days before discharge because such discharges would be on indeterminate doses of MOUD that would range between maintenance and out-of-treatment, depending on the timing of discharge. Lastly, we excluded records of individuals who had subsequent incarceration events recorded after death, indicating potential recording or matching errors. The final sample included 31 382 incarceration events from 15 797 adults. The incarceration events were further categorized into MOUD (n=17 119) and out-of-treatment (n=14 263) groups using the definitions described in the exposure section below.
Figure 1:
Flow Chart of the Study Sample Selection
Outcomes
The main study outcome was death caused by accidental drug poisoning, identified from NYC death registry data using ICD-10-CM codes (X40-X49; X60-X69; X85-X90; Y10-Y19; Y35.2; U01{.6–.7}, F11–16, and F18–19), text searches for overdose, poisoning, acute and chronic substance, acute complications, acute intoxication (by drug name), and intoxication (by drug name) in the multiple causes of death fields, and manner of death certified by the NYC Office of the Chief Medical Examiner as accidental. We then categorized overdose death cases into three non-mutually exclusive groups, including heroin-related, fentanyl-related, and cocaine-related deaths. Additionally, we used all-cause deaths as a secondary study outcome.
Exposure
The exposure variable was receipt of in-jail MOUD. Throughout incarceration, treatment options for OUD (withdrawal-only, maintenance, or out-of-treatment/no MOUD offered) could be modified based on an individual’s decision or less often, involuntarily (e.g., non-compliance with conditions of MOUD program). These changes were represented by 11 unique trajectories (Supporting information, Table S1). We combined in-jail MOUD trajectories into two groups (MOUD and out-of-treatment) based on MOUD received during the last three days before discharge. Note that receipt of MOUD versus out-of-treatment was only measured during incarceration.
Person-time
Person-time was time that individuals spent in the community after release from jail. For each discharge from jail, we counted days in the community to death or up to 365 days following release. If a re-incarceration event occurred before 365 days (n=17 675) or if the end of the 365-day follow-up period occurred after the end of the study period (n=2989), the follow-up period was right-censored.
Covariates
We included self-reported or records-based demographic information (age at release from jail, sex, race, Hispanic ethnicity, marital status, education level); self-reported or confirmed HIV diagnosis; self-reported behavioral factors (lifetime injection drug use, current smoker); self-reported homelessness or transient shelter stay at the time of incarceration (homelessness); criminal legal characteristics (incarceration length, highest charge severity, parole violation on a prior felony conviction); and count of previous NYC jail admissions events since 2011. We also included psychiatric variables, recorded by CHS mental health clinicians at the current incarceration: diagnosis of serious mental illness based on criteria from the NYS Office of Mental Health (until mid-2017) and Diagnostic and Statistical Manual of Mental Disorders (DSM) 5 edition (after mid-2017), depressive disorders, schizophrenia spectrum and other psychotic disorders, bipolar disorders, post-traumatic stress disorder based on DSM 5, and personality disorder based on DSM IV or 5. We further included alcohol abuse/dependence/use disorder and cocaine abuse/dependence/use disorder presumptively based on DSM IV or 5. Lastly, we included total numbers of ED visits within three months prior to discharge date from SPARCS.
Statistical analysis
We first compared demographic, clinical, behavioral, legal, homelessness, and healthcare utilization characteristics of MOUD and out-of-treatment groups using standardized differences, which were more informative than p-values from Chi-squared tests or t-tests due to large sample size. We then calculated crude all-cause and overdose mortality rates, and Kaplan-Meier curves by exposure. Lastly, we performed Cox proportional hazards regression analysis to assess whether in-jail MOUD was associated with greater mortality risk within one year of release from jail to the community. To address potential confounding, we included both time-invariant (sex, race and ethnicity) and time-varying covariates that were updated per incarceration discharge (all variables listed above, except time-invariant variables) in the Cox models. We tested the proportionality assumption using the weighted Schoenfeld residuals [20]. Violation of this assumption for the exposure variable led to modifying the models such that data were divided using three follow-up time intervals (first 28 days, 29–56 days, 57–365 days) based on the weighted Schoenfeld residual plots/tests. We estimated a time-specific hazard ratio (HR) for MOUD for each interval via a step function [21]. Violations of the proportionality assumption for other covariates (race and ethnicity, injection drug use, alcohol use disorder) were addressed via stratification analysis. To address correlation between multiple discharge events for the same person, we included individual-level random effects in the regression model. To test whether association between in-jail MOUD and mortality differed by race and ethnicity, we stratified data by three majority groups (non-Hispanic White, non-Hispanic Black, Hispanic) and repeated the regression analysis. Lastly, to address potential bias due to informative censoring, we conducted a sensitivity analysis where overdose death, non-overdose death, and re-incarceration events were considered competing risks, and HRs for MOUD versus out-of-treatment were calculated via the Fine-Gray competing risk model [22].
All statistical analyses were conducted using SAS version 9.4/Enterprise Guide version 8.3 and R version 3.5.2. Statistical significance was determined using two-sided p-value < 0.05.
RESULTS
Table 1 shows the characteristics of all eligible discharge events from adults with OUD who were incarcerated in NYC jails (n=31 382). Fifty-two percent were for misdemeanor-related offenses, followed by 42% for felonies; 7% of incarcerations were associated with parole violations. Most incarcerations were three or more days in length (96%). Average age at jail admission was 42 years. Most individuals were male (82%), non-Hispanic Black or Hispanic (74%), and single (79%). Thirty-five percent and 5% were determined to have a mental health diagnosis and serious mental illness, respectively. Those with specific mental health diagnoses ranged from 5% (schizophrenia or psychotic disorder) to 13% (personality disorder or depressive symptoms). In addition to OUD, 25% had alcohol use disorder and 34% had cocaine use disorder. Lifetime injection drug use (46%) and current smoking (82%) were prevalent. Lastly, 17% reported experiencing homelessness.
Table 1:
Demographic, Clinical, and Incarceration-Related Characteristics of the Incarceration Events
| Total (N=31 382) | MOUD (N=17 119) |
Out-of-treatment (N=14 263) |
Standardized difference | |
|---|---|---|---|---|
| Legal characteristics | ||||
| < 3 days in jail | 4% | 2% | 6% | 0.23 |
| Highest charge severity at/near admission (≤ 15 days) | ||||
| Felony | 42% | 26% | 61% | 0.90 |
| Misdemeanor | 52% | 71% | 30% | |
| Other | 6% | 3% | 9% | |
| Parole violation at/near admission (≤ 15 days) | 7% | 3% | 12% | 0.35 |
| Average length of incarceration (SD), in days | 60 (79) | 44 (60) | 79 (93) | 0.45 |
| Behavioral characteristics | ||||
| Self-reported injection drug use | 46% | 51% | 39% | 0.23 |
| Current smoking | 82% | 83% | 80% | 0.06 |
| Clinical characteristics | ||||
| Alcohol use/dependence/use disorder | 25% | 25% | 25% | 0.01 |
| Cocaine use/dependence/use disorder | 34% | 38% | 30% | 0.16 |
| Serious mental illness designation by CHS | 5% | 5% | 5% | 0.02 |
| Personality disorders diagnosis | 13% | 11% | 16% | 0.13 |
| PTSD, trauma, stress-related disorders diagnosis | 9% | 9% | 9% | 0.01 |
| Depression and depressive symptoms diagnosis | 13% | 13% | 13% | 0.01 |
| Bipolar and related disorders diagnosis | 6% | 6% | 7% | 0.02 |
| Schizophrenia and psychotic disorders diagnosis | 5% | 5% | 6% | 0.06 |
| HIV diagnosis | 8% | 9% | 7% | 0.06 |
| Housing characteristics | ||||
| Homelessness | 17% | 19% | 15% | 0.11 |
| Demographic characteristics | ||||
| Female | 18% | 23% | 11% | 0.33 |
| Race and ethnicity | ||||
| Hispanic | 44% | 42% | 45% | 0.11 |
| Non-Hispanic Black | 30% | 30% | 31% | |
| Non-Hispanic White | 24% | 26% | 22% | |
| Othersa | 2% | 2% | 2% | |
| Education | ||||
| < High school degree | 43% | 43% | 43% | 0.02 |
| High school degree or equivalent | 37% | 37% | 37% | |
| Some college/trade school | 17% | 17% | 18% | |
| Unknown | 3% | 3% | 3% | |
| Marital status | ||||
| Single, never married | 79% | 79% | 79% | 0.03 |
| Married/partnered | 16% | 15% | 16% | |
| Others | 5% | 6% | 5% | |
| Average age at admission, years (SD) | 42 (11) | 43 (10) | 41 (11) | 0.14 |
| Healthcare utilization | ||||
| Average number of emergency room visits for 3 months prior to discharge date (SD) | 0.42 (1.37) | 0.51 (1.57) | 0.31 (1.07) | 0.15 |
Notes:
Others include individuals who reported no or unknown Hispanic ethnicity and Asian or Pacific Islander, American Indian/Alaskan, other-unspecified race, or unknown race.
Incarceration events with MOUD versus out-of-treatment were more likely to have their highest charge severity recorded as misdemeanor and less likely to result from parole violation (Table 1). In-jail MOUD was more prevalent among individuals who were female, experienced homelessness, ever injected drugs, had cocaine use disorder, and had visited an ED three months prior to discharge date. Mental health-related characteristics, marital status, and the highest educational attainment were similar between two groups.
Mortality rates across MOUD and out-of-treatment groups
A total of 111 overdose deaths occurred within one year post jail release (Table 2). Crude overdose mortality rates for MOUD versus out-of-treatment groups were 0.49 per 100 person-years (95% CI = 0.35–0.66) and 0.83 per 100 person-years (95% CI = 0.64–1.04), respectively. After stratifying by drug types, the MOUD versus out-of-treatment groups had significantly lower crude heroin-related overdose death rates (rate ratio (RR) = 0.44, 95% CI = 0.28–0.70). Restricting the follow-up period to one month following community reentry, we observed lower overdose mortality rate among the MOUD versus out-of-treatment groups (RR = 0.23, 95% CI = 0.10–0.52). Lower crude mortality rates among the MOUD group within one month after release persisted when considering specific drug types. Similar to overdose mortality rates, significantly lower all-cause crude mortality rates were observed among the MOUD group in the first month following community reentry (RR = 0.26, 95% CI = 0.14–0.50), but rate differences for the one-year follow-up period were not statistically significant (RR = 0.81, 95% CI = 0.62–1.06). Figure 2 illustrates crude Kaplan-Meier survival curves by MOUD versus out-of-treatment groups, which were consistent with differences in crude overdose and all-cause mortality rates between two groups.
Table 2:
Overdose and All-Cause Death Counts and Crude Mortality Rates by Receipt of In-Jail MOUDa
| MOUD | Out-of-treatment | Crude rate ratio (95% CI)b | p-valueb | |||
|---|---|---|---|---|---|---|
| No. of deaths | Rate (95% CI) | No. of deaths | Rate (95% CI) | |||
| 1-year follow-up | ||||||
| Overdose death | 41 | 0.49 (0.35, 0.66) | 70 | 0.83 (0.64, 1.04) | 0.59 (0.40, 0.87) | 0.01 |
| Heroin-relatedc | 26 | 0.31 (0.20, 0.45) | 59 | 0.70 (0.53, 0.90) | 0.44 (0.28, 0.70) | <0.01 |
| Cocaine-relatedc | 22 | 0.26 (0.16, 0.40) | 37 | 0.44 (0.31, 0.60) | 0.60 (0.35, 1.01) | 0.06 |
| Fentanyl-relatedc | 13 | 0.15 (0.08, 0.26) | 20 | 0.24 (0.14, 0.36) | 0.65 (0.33, 1.31) | 0.23 |
| All-cause deathd | 98 | 1.16 (0.94, 1.42) | 121 | 1.43 (1.17, 1.68) | 0.81 (0.62, 1.06) | 0.13 |
| 1-month follow-up | ||||||
| Overdose death | 7 | 0.59 (0.24, 1.22) | 27 | 2.62 (1.73, 3.81) | 0.23 (0.10, 0.52) | <0.01 |
| Heroin-relatedc | * | 0.34 (0.09, 0.86) | 23 | 2.23 (1.41, 3.35) | 0.15 (0.05, 0.44) | <0.01 |
| Cocaine-relatedc | 6 | 0.51 (0.19, 1.10) | 18 | 1.75 (1.04, 2.76) | 0.29 (0.12, 0.73) | <0.01 |
| Fentanyl-relatedc | * | 0.08 (0.00, 0.47) | 9 | 0.87 (0.40, 1.66) | 0.10 (0.01, 0.76) | 0.03 |
| All-cause death | 12 | 1.01 (0.52, 1.77) | 40 | 3.88 (2.77, 5.29) | 0.26 (0.14, 0.50) | <0.01 |
Notes:
Receipt of MOUD versus out-of-treatment was only measured during incarceration.
p-values for the exposure (MOUD) were derived from Poisson regression models with number of deaths as an outcome and person-years as an offset variable.
Specific drugs that were related with overdose deaths were not mutually exclusive.
Other major causes of death include malignant neoplasms (n=14), disease of heart (n=11), and HIV/AIDS (n=6) for MOUD group. For out-of-treatment group, these include malignant neoplasm (n=10), disease of heart (n=8), homicide (n=7).
suppressed due to small numbers (≤ 5).
Figure 2:
Crude Kaplan-Meier Survival Curves for In-Jail MOUD vs Out-Of-Treatment
Adjusted HR for overdose and all-cause mortality by MOUD
After accounting for potential confounding and including individual-level random effects, we found that in-jail MOUD was associated with lower risk of overdose mortality for the first 28 days post-release (adjusted HR = 0.20, 95% CI = 0.08–0.46) (Table 3). We also observed reduction of fatal overdose risk associated with in-jail MOUD in the remaining time intervals, although associations were not statistically significant.
Table 3:
Time-Specific Hazard Ratio from Multivariable, Mixed-Effect Cox Proportional Hazard Regression for Overdose and All-Cause Mortality by In-Jail MOUD Receipt
| Overdose death | All-cause death | |||
|---|---|---|---|---|
| Hazard Ratio | 95% CI | Hazard Ratio | 95% CI | |
| MOUD vs. out-of-treatment for 1–28 days | 0.20 | 0.08, 0.46 | 0.22 | 0.11, 0.42 |
| MOUD vs. out-of-treatment for 29–56 days | 0.52 | 0.09, 2.91 | 0.29 | 0.09, 0.89 |
| MOUD vs. out-of-treatment for 57–365 days | 0.84 | 0.51, 1.38 | 1.20 | 0.85, 1.70 |
Notes: A random effect was estimated including a Gaussian frailty term to the model.
Multivariable models include demographic (age, sex, race and ethnicity, marital status, the highest level of education), clinical (serious mental illness, personality disorder diagnosis, schizophrenia diagnosis, bipolar disorder diagnosis, alcohol use disorder, cocaine use disorder, HIV), behavioral (injecting drug use, current smoking), housing (homelessness), healthcare utilization (total numbers of ED visits during 3 months prior to discharge date), and legal characteristics variables (length of incarceration, felony charges vs. other charges, and count of previous incarceration events).
Similarly, we found reduction of all-cause mortality risk among MOUD versus out-of-treatment groups for the first 28 days after release (adjusted HR = 0.22, 95% CI = 0.11–0.42). This association remained significant for 29–56 days post-release (adjusted HR = 0.29, 95% CI = 0.09–0.89), but became non-significant for 57–365 days (adjusted HR = 1.20, 95% CI = 0.85–1.70). When stratified by race and ethnicity, we observed reduced overdose mortality risk associated with in-jail MOUD for non-Hispanic White and Hispanic individuals, but not for non-Hispanic Black individuals (Supporting information, Table S2). On the other hand, the finding for all-cause mortality was similar between the main and race and ethnicity-specific analyses.
To assess potential bias due to informative censoring, we considered overdose mortality, non-overdose mortality, and re-incarceration competing risk events and re-ran the analysis using both cause-specific and Fine-Gray models. Findings from these two models were not qualitatively different from our main results (Supporting information, Table S3).
DISCUSSION
In this study of adults with OUD who were released from NYC jails, receiving MOUD during incarceration was associated with an 80% reduction in overdose mortality risk for the first month following community reentry. In-jail MOUD was also associated with a 78% and 71% reduction in all-cause mortality during the first- and second month following reentry, respectively.
Our findings closely align with studies from other countries such as Marsden et al., who found that MOUD was associated with 75% reduction of adjusted overdose mortality risk during the first month after release from prisons in England [8–11]. The reduced mortality risk associated with in-jail MOUD beyond the first month may likely be mediated by increased engagement with treatment in the community among those who receive MOUD while incarcerated. Although this potential mechanism could not be tested due to limited community treatment data, a Baltimore randomized control study reported evidence that in-prison buprenorphine treatment led to greater use of community-based MOUD upon release [23]. However, our null association between in-jail MOUD and mortality for 57–365 days post-release may indicate challenges of continuing long-term community MOUD despite receiving MOUD during incarceration. Further studies with community treatment data are necessary to test if use of the community-based MOUD mediates association between in-jail MOUD and mortality by different time points post-release.
Structural racism in US drug policy has disproportionately incarcerated individuals of racial and ethnic minorities who are also medically underserved [24]. This in turn may potentiate racial and ethnic disparities in fatal overdose and treatment access, as the fastest growing overdose death rates were observed in non-Hispanic Black individuals [25–28]. In NYC, 2017 data show that overdose death rates due to heroin and/or fentanyl in the 55–84 age category were substantially higher among non-Hispanic Black and Hispanic New Yorkers than non-Hispanic White New Yorkers [29]. Our findings suggest that jail-based MOUD program with linkage to community MOUD could alleviate excess burden of fatal overdose among middle-aged non-Hispanic Black and Hispanic individuals, mitigating some of the harm caused by structural racism in the criminal legal response to OUD [30]. The null impact among non-Hispanic Black individuals may be a product of limited power due to the small number of overdose deaths in this group but could potentially reflect gaps in access to MOUD in the community and warrants further investigation.
Our study has some limitations. Homelessness and comorbidities noted in electronic medical records may have been screened for differently over the study period and may be underreported, especially in individuals with shorter lengths of stay. Our analysis assumes that individuals stayed in NYC after release to the community; we do not have data on those moving out of NYC. If outmigration is higher among out-of-treatment over MOUD groups, it could bias the estimated HR away from the null since the true number of deaths among the out-of-treatment group might be underestimated. In addition, since residual confounding cannot be ruled out, we computed an E-value [31]. The E-value for the confidence interval was 3.8, implying that an unmeasured confounder would require at least that strong of an association, on the risk-ratio scale, with both overdose mortality and MOUD, in order to nullify the adjusted HR for the first 28 days following community reentry. Our study did not include more recent data where some major changes have occurred (e.g., increasing rates of overdose death rates in NYC, increasing enrollment of in-jail MOUD), and the estimated HR might be different if recent data were analyzed. Future studies with recent data are warranted to understand the extent to which the expanded in-jail MOUD has offset an increased risk of overdose deaths in NYC post-release among individuals with incarceration and OUD. Finally, we assumed that each discharge event was independent of all others, but it is highly plausible that events and outcomes are correlated within patients. To address this potential bias, we performed random effects modeling. Despite these limitations, the study has notable strengths. It is one of the largest studies of MOUD in one of the largest jail systems in the country, providing an opportunity to test important public health and policy questions with sufficient power. Another strength is the use of linked health and administrative data that examine a board range of exposures and outcomes and provide detailed baseline information.
CONCLUSION
To the best of our knowledge, this is the first study to evaluate the association between in-jail MOUD and risk of opioid overdose mortality following reentry to the community using health and administrative data from a large urban jail setting in the US. Findings from this study corroborate evidence that jail-based MOUD substantially reduces overdose risk. In 2019, there were 10.3 million incarcerations in US jails, and roughly 15–20% of those incarcerated in jails have OUD or misused opioids [7,32,33]. This population is at excess risk for overdose mortality after release [2,5], yet most US carceral settings do not offer effective forms of MOUD [15,16]. Expansion of MOUD access in carceral settings and, jails, in particular, should be an urgent priority of policy efforts to address the opioid crisis.
Supplementary Material
Primary funding:
This research was supported by R01 grant (R01DA045042-01A1) from the National Institute on Drug Abuse
Footnotes
Declaration of competing interests: none declared
References
- 1.Binswanger IA, Blatchford PJ, Mueller SR, Stern MF. Mortality after prison release: opioid overdose and other causes of death, risk factors, and time trends from 1999 to 2009. Ann Intern Med. 2013;159(9):592–600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Binswanger IA, Stern MF, Deyo RA, et al. Release from prison- a high risk of death for former inmates. NEJM. 2007;356(2):157–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Gordon MS, Kinlock TW, Schwartz RP, O’Grady KE. A randomized clinical trial of methadone maintenance for prisoners: findings at 6 months post-release. Addiction. 2008;103:1333–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Fox AD, Maradiaga J, Weiss L, et al. Release from incarceration, relapse to opioid use and the potential for buprenorphine maintenance treatment: a qualitative study of the perceptions of former inmates with opioid use disorder. Addict Sci Clin Pract. 2015;10(1):1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lim S, Seligson AL, Parvez FM, et al. Risks of drug-related death, suicide, and homicide during the immediate post-release period among people released from New York City jails, 2001–2005. Am J Epidemiol. 2012;175(6):519–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Winkelman TN, Chang VW, Binswanger IA. Health, polysubstance use, and criminal justice involvement among adults with varying levels of opioid use. JAMA Netw Open. 2018;1(3):e180558–e. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Joudrey PJ, Khan MR, Wang EA, et al. A conceptual model for understanding post-release opioid-related overdose risk. Addict Sci Ciln Pract. 2019;14(1):1–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Huang YF, Kuo HS, Lew-Ting CY, et al. Mortality among a cohort of drug users after their release from prison: an evaluation of the effectiveness of a harm reduction program in Taiwan. Addiction. 2011;106(8):1437–45. [DOI] [PubMed] [Google Scholar]
- 9.Degenhardt L, Larney S, Kimber J, et al. The impact of opioid substitution therapy on mortality post-release from prison: retrospective data linkage study. Addiction. 2014;109(8):1306–17. [DOI] [PubMed] [Google Scholar]
- 10.Bird SM, Fischbacher CM, Graham L, Fraser A. Impact of opioid substitution therapy for Scotland’s prisoners on drug-related deaths soon after prisoner release. Addiction. 2015;110(10):1617–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Marsden J, Stillwell G, Jones H, et al. Does exposure to opioid substitution treatment in prison reduce the risk of death after release? A national prospective observational study in England. Addiction. 2017;112(8):1408–1418. [DOI] [PubMed] [Google Scholar]
- 12.Olsen Y, Sharfstein JM. Confronting the stigma of opioid use disorder--and its treatment. JAMA. 2014;311(14):1393–4. [DOI] [PubMed] [Google Scholar]
- 13.Volkow ND, Frieden TR, Hyde PS, Cha SS. Medication-assisted therapies—tackling the opioid-overdose epidemic. N Engl J Med 2014;370(22):2063–2066 [DOI] [PubMed] [Google Scholar]
- 14.McKenzie M, Nunn A, Zaller ND, et al. Overcoming obstacles to implementing methadone maintenance therapy for prisoners: implications for policy and practice. J Opioid Manag. 2009;5(4):219–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fiscella K, Moore A, Engerman J, Meldrum S. Jail management of arrestees/inmates enrolled in community methadone maintenance programs. J Urban Health. 2004;81(4):645–654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kerr T, Jϋrgens R. Methadone Maintenance Therapy in Prisons: Reviewing the Evidence. Canadian HIV/AIDS Legal Network. 2004: Accessed April 15, 2022. http://www.aidslaw.ca/home.htm. [Google Scholar]
- 17.NIH RePort Project Details. Impact of Jail-Based Methadone on Overdose, Recidivism, HIV and Health Outcomes, and Costs in New York City, 2011–2017. Accessed September 16, 2022. https://reporter.nih.gov/project-details/9596037
- 18.The City of New York Office of the Mayor. Breaking the frame? Remaking the criminal justice system in New York City. 2019. Accessed April 15, 2022. https://criminaljustice.cityofnewyork.us/wp-content/uploads/2019/11/Breaking-the-Frame___.pdf. [Google Scholar]
- 19.New York State Department of Health, Statewide Planning and Research Cooperative System (SPARCS), 2009–2017. Internal analyses by New York City Department of Health and Mental Hygiene. December 20, 2021. [Google Scholar]
- 20.Grambsch PM, Therneau T. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika. 1994;81(3):515–526. [Google Scholar]
- 21.Therneau T, Crowson C, Atkinson E. Using time dependent covariates and time dependent coefficients in the cox model. Survival Vignettes. 2017;2:3. [Google Scholar]
- 22.Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. JASA. 1999;94:496–509. [Google Scholar]
- 23.Gordon MS, Kinlock TW, Schwartz RP, Fitzgerald TT, O’Grady KE, Vocci FJ. A randomized controlled trial of prison-initiated buprenorphine: prison outcomes and community treatment entry. Drug Alcohol Depend. 2014;142:33–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kunins HV. Structural Racism and the Opioid Overdose Epidemic: The Need for Antiracist Public Health Practice. J Public Health Manag Pract. 2020;26(3):201–205 [DOI] [PubMed] [Google Scholar]
- 25.Lippold KM, Jones CM, Olsen EO, Giroir BP. Racial/Ethnic and Age Group Differences in Opioid and Synthetic Opioid–Involved Overdose Deaths Among Adults Aged ≥18 Years in Metropolitan Areas — United States, 2015–2017. MMWR Morb Mortal Wkly Rep. 2019;68:967–973. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Mason M, Soliman R, Kim HS, Post LA. Disparities by Sex and Race and Ethnicity in Death Rates Due to Opioid Overdose Among Adults 55 Years or Older, 1999 to 2019. JAMA Netw Open. 2022;5(1):e2142982. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Larochelle MR, Slavova S, Root ED, et al. Disparities in Opioid Overdose Death Trends by Race/Ethnicity, 2018–2019, From the HEALing Communities Study. Am J Public Health. 2021. Oct;111(10):1851–1854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Townsend T, Kline D, Rivera-Aguirre A, et al. Racial/Ethnic and Geographic Trends in Combined Stimulant/Opioid Overdoses, 2007–2019. Am J Epidemiol. 2022. Mar 24;191(4):599–612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Allen B, Nolan ML, Kunins HV, Paone D. Racial differences in opioid overdose deaths in New York City, 2017. JAMA Intern. Med 2019;179(4):576–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Krawczyk N, Mojtabai R, Stuart EA, et al. Opioid agonist treatment is highly protective against overdose death among a US statewide population of justice-involved adults. Am J Drug Alcohol Abuse. 2021;47(1):117–26. [Google Scholar]
- 31.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann. Intern. Med 2017;167(4):268–74. [DOI] [PubMed] [Google Scholar]
- 32.Office of Justice Statistics, Jail Inmates in 2019. Washington, DC: Department of Justice; 2021. https://bjs.ojp.gov/content/pub/pdf/ji19.pdf. Accessed April 25, 2022. [Google Scholar]
- 33.Bronson J, Stroop J, Zimmer S, Berzofsky M. Drug use, dependence, and abuse among state prisoners and jail inmates, 2007–2009. Washington, DC: Department of Justice; 2017. https://www.bjs.gov/content/pub/pdf/dudaspji0709.pdf. Accessed April 25, 2022. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.