Background
The United States (US) leads the world in both the number of incarcerated persons and rate of incarceration. Figures from 2016 estimate the total number of incarcerated persons in the US at 1.5 million and a rate of incarceration in jails or prisons of 670 per 100,000 [
1]. Compared to the general population, incarcerated persons have a disproportionally higher rate of opioid use disorders (OUDs) [
2‐
4]; 13.1% and 9.2% of people incarcerated in state and federal prisoners respectively reported using heroin or other opiates regularly in the community before their incarceration [
5]. Given that the vast majority of incarcerated persons will be released from prison at some point and that 626,024 individuals were released from state and federal prisons in 2016 [
6], there is a considerable need to deliver effective treatment to this population in order to reduce relapse to drug use upon release.
A return to illicit psychoactive substance use upon community re-entry poses significant risks to health and public safety [
7]. Individuals recently released from jail or prison are at increased risk for overdose death within their first month in the community [
8‐
17] and a return to opioid use is associated with criminal activity [
7,
18,
19] and re-incarceration [
18,
20,
21]. Relapse to drug use is also a significant public health concern as it heightens the risk for HIV and hepatitis B and C infections [
2,
4,
7].
The Centers for Disease Control and Prevention (CDC) estimated that in the US there are approximately 1.1 million people who were HIV positive in 2015 [
22]. Of that group, one in seven individuals were unaware of their HIV status [
22]. In 2015, HIV was the 9th leading cause of death for those 25 to 44 years of age [
22]. As with OUDs, HIV infection is overrepresented in the criminal justice population with estimated rates ranging between three and five times greater than that of the general population [
23‐
25]. The higher incidence of HIV infection amongst criminal justice populations is at least partially attributable to their increased rates of substance use disorders. Substance use exacerbates the risk of HIV through the practice of injection drug use, the sharing of needles and injection paraphernalia (cookers, cotton, and rinse water), and its association with risky sex, sex with multiple partners, and transactional sex/sex work [
26‐
30]. For example, a longitudinal study by MacGowan and colleagues found that among men recently released from prison, the only factor independently associated with risky sex was the use of alcohol or illicit substances before sex [
31]. Substantial research evidence indicates that men and women with OUDs differ in terms of their health and substance use treatment needs and their risk of HIV infection. Among individuals with OUDs, women are more likely than men to suffer from serious medical conditions [
32‐
34], mental health problems [
33,
35], unemployment [
35,
36], the stress of having responsibility for child care [
37], and the burden of having a spouse or partner with addiction problems [
33,
34]. Moreover, an analysis by Binswanger and colleagues [
38] of HIV risk behaviors for both men and women post-release found that a higher proportion of women than men engaged in several risk behaviors including unprotected sex and sex with multiple partners. In addition, they found that women were more prone than men to exchange sex for drugs and/or money.
While the practice of mandatory or opt-out HIV testing has become more commonplace in state and federal prisons, [
25] there is still a need for HIV risk-reduction interventions for criminal justice involved individuals [
26]. One promising avenue for improving public health outcomes, is the use of opioid agonist therapy (OAT) for the treatment of OUD. Reviews of studies involving OAT and HIV-risk behaviors have shown that both methadone and buprenorphine may help to reduce injection drug use, needle sharing, and risky sexual behaviors [
26,
39]. However, these studies often focus on community samples and there is limited research on the effects of prison-initiated OAT on HIV-risk behaviors [
39,
40] presenting an opportunity for study.
For the treatment of OUDs, the three pharmacotherapies approved by the U.S. Food and Drug Administration (methadone, buprenorphine, and naltrexone) represent the highest standard of care, but are rarely implemented within the criminal justice system [
41‐
43]. Given their demonstrated efficacy in community settings [
44,
45], implementation of pharmacotherapies prior to release may help to prevent illicit opioid use in prison and relapse upon release for individuals who have maintained opioid abstinence while incarcerated. In turn, a reduction in the rates of relapse to opioid use upon release can help to reduce the incidence of HIV infections through a reduction in injection drug use and risky sex behaviors. A review of studies of pre-release opioid agonist therapies (OAT) found that pre-release OAT in prison is associated with significantly increased treatment uptake after release and treatment retention, with differences observed as far as 12 months post-release [
46]. Given the importance of substance use treatment retention for positive treatment outcomes [
47,
48], this finding supports the use of pre-release OAT for reducing relapse to opioid use and its associated harms, including the risk of HIV infection.
Our research group has previously reported on findings from a randomized clinical trial that compared post-release outcomes of incarcerated persons with pre-incarceration histories of opioid dependence (defined by DSM-IV) who were randomly assigned to begin sublingual buprenorphine/naloxone prior to versus post-release from prison [
49,
50]. The study found that participants who were randomly assigned to initiate buprenorphine in prison were significantly more likely to enter and to remain in buprenorphine treatment in the community compared to participants who were assigned to begin buprenorphine treatment after release [
50]. However, despite greater community treatment exposure, there were no significant differences between treatment conditions in heroin and cocaine use at the 12-month follow-up [
50].
The present study
The aim of the present study is to examine the impact of initiating buprenorphine prior to versus post-release on relative incidences of four key HIV risk behaviors: (1) sex without a condom (2) injection drug use (3) using unsterilized needles, and (4) sharing injection paraphernalia. Here we present findings from a secondary analysis of data from the above-mentioned clinical trial. We hypothesized that because of the potential advantages of initiating buprenorphine treatment in prison (higher rates of treatment entry and treatment retention) that there would be greater levels of improvement (greater decreases) in the number of self-reported incidences of each of the four HIV risk behaviors over time when controlling for gender and community treatment entry. The parent study found that the study condition assigned to initiate buprenorphine in prison compared to after release was associated with significantly higher rates of community treatment entry and community treatment exposure. While greater participation in treatment post-release did not produce significant differences in heroin or cocaine use, it may have impacted other aspects of substance use that are related to increased HIV risk (i.e. using unsterilized needles, exchanging sex for drugs) especially when treatment is based on the principle of harm reduction.
Discussion
Overall, the present study did not find support for the initiation of buprenorphine in prison (as opposed to the community) as a means to reduce incidences of HIV-risk behaviors. When examining injection drug use, the condition assigned to begin buprenorphine in the community reported fewer instances of injecting over time when compared to the group assigned to initiate buprenorphine in prison. This result is in contrast to the parent study [
49] that did not find any differences in self-reported heroin use, cocaine use, or treatment retention at 12 months. However, these results should be interpreted cautiously as only 40.3% of participants at baseline reported injecting drugs with a needle at least once in the last 30 days they were in the community suggesting they may prefer alternative routes of administration [
60]. Therefore, changes in heroin or cocaine use over time would not necessarily mirror changes in injection drug use.
There were no significant predictors of decreasing incidences of sex without a condom, using unsterilized needles, or sharing injection paraphernalia in the present study. However, it should be noted that in the case of using unsterilized needles, self-reported frequencies of this behavior were very low at baseline compared to the other HIV-risk behaviors (see Table
2) as some individuals use opioids exclusively intranasally.
The present findings contribute to the existing efforts to reduce HIV-risk behavior among prisoners treated with OAT. We are aware of only one other randomized trial of pre-release buprenorphine treatment in the US, which was conducted among short-sentenced inmates in New York City [
61]. That study did not report HIV risk behavior [
61]. In a review of the research on prison-based OAT and its effects on HIV-risk behaviors, Larney found some support for the use OAT in prison to reduce post-release injection drug use and needle sharing [
39]. However, the author noted a paucity of research in this area: only one of the five studies evaluated was a randomized controlled trial, and none of them took place in the US. Subsequent to the report by Larney, our group reported on post-release HIV-risk behavior from a randomized trial comparing initiating methadone treatment during vs. post-release from prison compared to a counseling in prison and referral condition [
40]. That study did not find not find significant differences between treatment conditions in the rate of change of HIV-sex or -drug risk behavior. However there were significant effects of treatment condition on drug-risk behaviors such that participants assigned to initiate methadone pre-release reported fewer incidences of drug-risk behaviors irrespective of time and there was a significant effect of time such that participants reported fewer incidences of drug-risk behavior as the study progressed.
While there is limited research on the effects of prison-initiated OAT on HIV-risk behaviors, there is evidence supporting the use of community-based OAT to reduce HIV-behaviors for both criminal-justice-involved persons [
26] and the general population [
62]. A review of HIV-risk reduction strategies for criminal-justice-involved adults found that OAT significantly reduced injection drug use, but was less effective at reducing risky sexual behaviors [
26]. These findings are echoed in reviews of studies involving general community samples [
62]. Treatment adherence appears to be a significant component for the success of OAT in reducing HIV-risk behaviors [
62], consistent with the broader literature which reports that adherence to substance abuse treatment is key in producing positive treatment outcomes [
63]. While the present study did not find support for the use of pre-release buprenorphine to reduce HIV-risk behaviors, the parent study found that participants who initiated buprenorphine in prison had a higher mean number of days receiving buprenorphine treatment in the community, suggesting that prison-initiated buprenorphine may improve treatment adherence in the community.
The present study has a few important limitations. First, response rates to certain items were quite low at the most distal follow-up periods (6 and 12 months) compared to response rates at baseline and follow-ups closer to release and the overall rate of missing responses was high (54%). While multilevel modelling is equipped to handle missing data, if data are not missing completely at random [e.g., there is some systematic factor(s) accounting for missingness that is not included in the model], then information is lost in the analysis, potentially biasing parameter estimates. Second, an assumption of hierarchical linear models is a consistent effect of predictors over time, which may not be the case in reality when examining treatment effects that might be most potent in the time immediately after release. There is also an assumption of linear change over time when using such models. While model comparisons showed that quadratic and cubic functions did not appear to fit the data better than a linear one, changes in instances of a particular behavior may be asymptotic in reality as they have floors (e.g. counts of behaviors cannot be negative) and ceilings that cannot be exceeded due to real world constraints. Third, self-reported data on HIV risk is subject to potential bias although such bias would likely be present across both conditions. Finally, recall of these behaviors, particularly for the 30 days prior to index incarceration, is subject to potential inaccuracy, which again would likely be equally present across both conditions.
More research is needed on the intersection of prison-initiated pharmacotherapy and HIV treatment. At least one study [
64] has found a positive association between 24-week retention in buprenorphine treatment and maximal viral suppression (which is associated with improved HIV treatment outcomes). However, this study was quasi-experimental, as participants in a RCT of directly administered antiretroviral therapy for prisoners who met the DSM-IV criteria for opioid dependence were offered buprenorphine pharmacotherapy while incarcerated and assessed post-release. Further research is needed to explore the relationship between prison-based OAT and HIV treatment outcomes such as viral suppression and medication adherence.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.