Death After Jail Release

Introduction

The United States incarcerates more people than any other country in the world, with a significant portion held in local jails (Walmsley, 2013). More than 740,000 persons were incarcerated in U.S. jails at midyear 2014 and approximately 11.3 million persons passed through jails in the 12 months prior (Minton & Zheng, 2015). The weeks following release from prison or jail have been shown to be a high-risk time for death, with drug overdose and homicide representing leading causes of postrelease death (Binswanger et al., 2007; Lim et al., 2012). Access to opioid agonist treatment has been shown to decrease urine drug screen positivity postrelease (Gordon, Kinlock, Couvillion, Schwartz, & O’Grady, 2012), decrease self-reported drug use (McKenzie et al., 2012), and increase retention in care (Gordon et al., 2012; McKenzie et al., 2012; Rich et al., 2015). Limited evidence suggests that prerelease opioid agonist treatment reduces postrelease mortality (Degenhardt et al., 2014; Dolan et al., 2005; Hedrich et al., 2012).

In the New York City (NYC) jail system, health care is delivered under the auspices of correctional health services (CHS), which was transferred from the local health department to the local public hospital system in August 2015. CHS quality improvement efforts include rigorous analyses of morbidity and mortality events inside the jails as a means to identify systems concerns that impact our patients. We sought to undertake a similar continuous quality improvement process using data from patients who died after release.

Method

We matched jail release records with NYC Bureau of Vital Statistics records to identify deaths occurring in NYC from June 1, 2011, to December 31, 2012. Records were matched by probabilistic record linkage using the RecordLinkage package for R (n.d.). We found 423 matches and excluded those for whom death occurred in jail or more than 42 days after release. Electronic health records (EHRs) for this limited cohort were reviewed for substance use history, opioid treatment program (OTP) interventions, and the number of prior incarcerations since November 2008. Cause of death was obtained from the Bureau of Vital Statistics.

The EHR contains self-reported information collected at the time of entry to jail including current use of drugs, history of detox, history of accidental overdose, and current methadone use. Point-of-care testing performed on urine samples during medical intake identified the presence of benzodiazepines, methadone, cocaine, or opiates. Additional abstracted data elements include methadone orders, referrals to OTP, and eligibility for methadone maintenance.

The in-jail all-cause mortality rate was estimated by dividing the number of deaths occurring inside the jail by the average daily population during the period under study. The postrelease all-cause mortality rate was estimated by dividing the number of deaths occurring within 6 weeks after release by the total number of discharges during that period; discharges were multiplied by 6 weeks to achieve comparable person-times for both the numerator and the denominator.

Results

During the 19 months under study, 27 deaths occurred within NYC jails while the average daily population was 12,300. The in-jail all-cause mortality rate was 1.39 deaths per 1,000 person-years. During the same period, NYC jails performed 86,771 discharges; of these, 59 deaths occurred within 42 days of release. The postrelease all-cause mortality rate was 5.89 per 1,000 person-years. Among postrelease deaths, the mean age was 40, the population was mostly male, and non-Hispanic Blacks accounted for more than 50% of the deaths (Table 1). Within the first 42 days, death occurred for 21 (35.6%) within the first 14 days, for 22 (37.3%) within days 15 to 28, and for 16 (27.1%) within days 29 to 42 after release (Table 2).

Causes of death in the first 42 days included opioid overdose (37.3%); other drug use (8.5%); chronic disease (25.4%); assaultive trauma (20.3%); and trauma from unintentional injury, suicide, or unspecified events (8.5%). The average number of days after release before death occurred was 18, 18, 22, 26, and 22 days, respectively.

Deaths from opioids or other drug use accounted for 27 (45.8%) deaths, though patient characteristics differed between the two groups (Table 3). While 77% of those dying from opioids self-reported a history of accidental overdose or opioid detox, only 20% of those dying from other drug use made a similar report. Similarly, 64% of those dying from opioids had a positive urine screen for methadone or opiates at the time of medical intake, though only 20% of those dying from other drug use deaths had positive screens. Most patients had prior incarceration. The median duration of incarceration in the jail was 43 days for those dying from opioids versus 16 days for those dying from other drugs. During incarceration, 50% of those eventually dying from opioids were referred to OTP. Among these 11, 5 were eligible for services, 3 were maintained on methadone until release (at dosages of 30, 90, and 100 mg), 1 requested a prolonged detox after several weeks of maintenance methadone, and 1 refused services. The remaining six were determined to be ineligible for OTP and were offered prolonged detox, and three accepted it.

Conclusions

These data show that the postrelease mortality rate is approximately 4 times greater than the in-jail mortality rate. Furthermore, they show that overdose death is the leading cause of postrelease death. Notably, only a small percentage of patients experiencing postrelease overdose were released on methadone. Many patients who died subsequent to release underwent a prolonged detoxification or were not offered methadone maintenance because of their likelihood of transfer to a setting without methadone available (e.g., New York State prisons). Others did not reveal substance use at the time of intake, suggesting that patient-centered history taking or more aggressive encouragement of enrollment into the OTP program by intake staff could reach more patients.

These findings also support CHS efforts to expand buprenorphine access and to distribute intranasal naloxone at the Rikers Island Visit House. The latter effort is intended to increase overdose prevention capacity in the communities to which our patients return and has resulted in the distribution of over 2,300 kits in the first 18 months of its existence (unpublished data).

Matching jail releases with vital statistics provided important insights into causes of death after jail release for our patient population and allowed a quality improvement review process similar to that employed for adverse events in custody. CHS encourages other correctional health systems to engage with local public health authorities to perform similar matching.