Incidence and predictors of non-fatal drug overdose after release from prison among people who inject drugs in Queensland, Australia

Drug and Alcohol Dependence 153 (2015) 43–49

Contents lists available at ScienceDirect

Drug and Alcohol Dependence journal homepage: www.elsevier.com/locate/drugalcdep

Incidence and predictors of non-fatal drug overdose after release from prison among people who inject drugs in Queensland, Australia R.J. Winter a,b,∗ , M. Stoové a,b , L. Degenhardt c,d , M.E. Hellard a,b , T. Spelman a,b,e , R. Jenkinson a,f , D.R. McCarthy a , S.A. Kinner b,d,g a

Centre for Population Health, Burnet Institute, Australia School of Public Health and Preventive Medicine, Monash University, Australia c National Drug and Alcohol Research Centre, University of New South Wales, Australia d Melbourne School of Population and Global Health, University of Melbourne, Australia e Victorian Infectious Diseases Service, Doherty Institute, Australia f Australian Institute of Family Studies, Australia g School of Medicine, University of Queensland, Australia b

a r t i c l e

i n f o

Article history: Received 12 March 2015 Received in revised form 23 May 2015 Accepted 4 June 2015 Available online 16 June 2015 Keywords: Drug overdose Prisoners Drug users Injecting drug use

a b s t r a c t Introduction: Release from prison is a period of elevated risk for drug-related harms, particularly among people who inject drugs (PWID). Non-fatal overdose can cause serious morbidity and predicts future fatal overdose, however neither the incidence nor the risk factors for non-fatal overdose following release from prison are well understood. Methods: Structured health-related interviews were conducted with 1051 adult prisoners in Queensland, Australia prior to release and approximately 1, 3 and 6 months post-release. Incidence of self-reported overdose in the community was calculated for PWID and all prisoners for three discrete time periods. Negative binomial regression with robust error variance was used to identify pre-release predictors of overdose among PWID. Results: The incidence of reported overdose was highest between 1 and 3 months post-release (37.8 per 100 person-years (PY) among PWID; 24.5/100 PY among all ex-prisoners). In adjusted analyses, the risk of post-release non-fatal overdose was higher for PWID who reported: being unemployed for >6 months before prison, having been removed from family as a child, at least weekly use of benzodiazepines and/or pharmaceutical opiates in the 3 months prior to prison, and ever receiving opioid substitution therapy (OST). Pre-release psychological distress and a lifetime history of mental disorder also predicted overdose, whereas risky alcohol use in the year before prison was protective. Conclusions: PWID have a high risk of overdose following release from prison. Imprisonment is an opportunity to initiate targeted preventive interventions such as OST, overdose prevention training and peer-delivered naloxone for those with a high risk profile. © 2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Release from prison is a period of elevated mortality risk, particularly resulting from suicide and drug overdose (Merrall et al., 2010; Kinner et al., 2013a; Chang et al., 2015). The risk of drugrelated death is highest in the first few weeks post-release (Merrall et al., 2010) and, although it decreases over time, excess mortality and morbidity continue among ex-prisoners (Hobbs et al., 2006).

∗ Corresponding author at: Centre for Population Health, Burnet Institute, GPO Box 2284, Melbourne 3001, Australia. Tel.: +61 3 8506 2328. E-mail address: [email protected] (R.J. Winter). http://dx.doi.org/10.1016/j.drugalcdep.2015.06.011 0376-8716/© 2015 Elsevier Ireland Ltd. All rights reserved.

People who have a history of injecting drug use (PWID) are at particular risk of drug-related death, and are over-represented in the criminal justice system (Fazel et al., 2006); around 70% of people entering Australian prisons report illicit drug use in the past 12 months and around half report a history of injecting (Australian Institute of Health and Welfare, 2013). Return to drug use is common after release from prison (Kinner, 2006; Binswanger et al., 2012) and ex-prisoners may be at increased risk of non-fatal drug overdose (Ochoa et al., 2005; Kerr et al., 2007; Kinner et al., 2012). Between 40% and 70% of PWID report having experienced a (mostly opioid) non-fatal overdose (Ochoa et al., 2005; Pollini et al., 2006; Darke et al., 2007; Kerr et al., 2007; Milloy et al., 2008). These events carry a risk of direct and

44

R.J. Winter et al. / Drug and Alcohol Dependence 153 (2015) 43–49

indirect, acute and chronic morbidities such as physical injury, seizure, pulmonary conditions, peripheral neuropathy, temporary paralysis, chest infection, neurological damage and cognitive impairment (Warner-Smith et al., 2002). In a study of 136 Sydneybased heroin users who reported overdose, 79% reported one or more of these sequelae (Warner-Smith et al., 2002). A substantial body of research from Australia, North America and Europe has identified correlates of heroin overdose, including patterns and types of drugs used (e.g., injecting route of administration (Brugal et al., 2002; Ochoa et al., 2005)), poly-drug use (Seal et al., 2001; Darke and Hall, 2003; Dietze et al., 2005; Kerr et al., 2007) and demographic (e.g., male gender (Darke and Hall, 2003; Stoové et al., 2009), older age (Kerr et al., 2007; Stoové et al., 2009)) and psychosocial factors (e.g., poor mental health (Bartoli et al., 2014), unstable housing (Kerr et al., 2007)). While there is evidence of a strong, dose-dependent protective effect of opioid substitution therapy (OST) against opioid overdose mortality (van Ameijden et al., 1999; Brugal et al., 2005), transitioning on or off drug treatment has been associated with increased risk (Strang et al., 2003; Davoli et al., 2007). Most of the available research has focussed on heroin-related overdose, but there is increasing research into overdose related to other nervous system depressants such as prescription opioids and benzodiazepines, in line with the burgeoning use of these drugs in many countries (Centers for Disease Control and Prevention, 2013; Cerdá et al., 2013). There has been comparatively little investigation of non-opioid illicit drug overdose, such as from psychostimulants (Kaye and Darke, 2004; Darke et al., 2008). Psychostimulants are the primary cause of only a minority of fatal overdoses in Australia (Darke et al., 2008), however their contribution to nonfatal overdose is less clear, obfuscated by the high prevalence of poly-drug use among users (Darke and Hall, 2003; Kaye and Darke, 2004; Kerr et al., 2007; Darke et al., 2008). The lower risk of death and ambiguity in definition (Kaye and Darke, 2004; Darke et al., 2008) mean that non-fatal psychostimulant overdoses may also be documented less frequently than opioid overdoses. Nevertheless, 13% of 200 regular cocaine users from Sydney reported cocaine overdose (Kaye and Darke, 2004), and some Australian jurisdictions have seen recent increases in psychostimulant-related harms (Heilbronn et al., 2013). Psychostimulant-related overdose morbidity may confer a range of short and long term cardiovascular, respiratory, autonomic and neuromuscular harms (Kaye and Darke, 2004; Darke et al., 2008; Cruickshank and Dyer, 2009). Non-fatal overdose is a strong predictor of future fatal overdose (Stoové et al., 2009; Darke et al., 2011), suggesting the potential for non-fatal overdose to serve as a flag for early intervention to prevent death. Release from prison may present an opportunity to intervene to reduce the risk of non-fatal drug overdose after release. To date, most studies of non-fatal overdose have relied on crosssectional data; few studies have examined overdose longitudinally to enable identification of risk and protective factors. Based on data from a cohort of adults interviewed in prison and at multiple time points after release from custody in Australia, the aims of this study were to: (1) estimate the incidence of self-reported non-fatal overdose at three discrete time periods following release from prison, among all released prisoners and among PWID, and (2) identify the pre-release predictors of non-fatal overdose among PWID.

2. Methods 2.1. Study design and setting The Passports study was a multi-site, single-blinded, randomised controlled trial of a case-management re-entry intervention for sentenced adult prisoners in the state of Queensland, Australia.

The study methods are described in detail elsewhere (Kinner et al., 2013b). Baseline interviews were conducted within 6 weeks of expected release from prison and before randomisation, in the seven prisons from which the majority of sentenced prisoners were released. Participants were randomised to receive either usual care or a transitional intervention that included individualised casemanagement in the first 4 weeks following release (Kinner et al., 2013b). Follow-up interviews occurred approximately 1 (FU1), 3 (FU2) and 6 months (to a maximum of 12 months) (FU3) after release from prison. 2.2. Sample selection and recruitment Prisoners due to be discharged from selected prisons from August, 2008 to July, 2010 were identified through correctional records and screened for eligibility. Eligibility criteria included (1) expected release within 6 weeks, (2) sentenced (i.e., not a pre-trial detainee), (3) imprisoned for at least 4 weeks, and (4) able to give informed consent. Participation was limited to prisoners who met these criteria so that baseline data were a true reflection of the status of participants immediately prior to release and measures which refer to a specific in-prison time period could be accurately collected. Researchers explained the study and supplied a plainlanguage information sheet; participants provided signed informed consent to participate. Of 1665 prisoners eligible and approached to participate, 1325 (80%) consented and completed a pre-release interview (Kinner et al., 2013b). By key demographic indicators, participants were broadly representative of all persons released from prison in Queensland during the recruitment period, with the exception that women were oversampled to allow adequate numbers for sex-stratified analyses (Kinner et al., 2013b). Pre-release data were collected via a face-to-face, researcheradministered structured questionnaire, typically taking 60–90 min to complete. Follow-up interviews were conducted by telephone in the community, and for participants who had been reimprisoned, in prison either by telephone or face-to-face. Participants who were released more than 8 weeks after their baseline interview or who did not complete at least one follow-up interview were excluded from the analyses presented here. 2.3. Measures 2.3.1. Post-release non-fatal overdose. At each follow-up interview, participants were asked whether they had overdosed or become unconscious as a result of taking drugs since release or last followup. The definition of non-fatal overdose was intentionally broad to accommodate poly-drug use in the sample (Kerr et al., 2007). Participants were asked how many times they had overdosed, what drugs they had been using and – for participants who had been reimprisoned during follow-up – whether the overdose/s occurred in the community or in custody. In this paper only community overdoses are examined. Reported non-fatal overdoses after using only cannabis and/or alcohol were also excluded; all other illicit drug overdoses, including those related to non-prescribed use of pharmaceuticals, were included. 2.3.2. Potential predictors of non-fatal overdose. Potential predictors of non-fatal overdose among PWID were identified based on a review of the literature. Demographic indicators included age at baseline (<30/30+ years), sex (male/female), indigeneity (yes/no), years of schooling (<10/10+ years), >6 months unemployed prior to imprisonment (yes/no), unstable housing in the month prior to imprisonment (yes/no), having been removed from family (e.g., fostered) as a child (yes/no), history of previous adult incarceration (yes/no) and duration of current incarceration. Drug-related measures included: years since first injection, lifetime overdose

R.J. Winter et al. / Drug and Alcohol Dependence 153 (2015) 43–49

history (yes/no), frequency (
45

conducted excluding participants who were re-imprisoned during follow-up. 2.4.2. Predictors of non-fatal overdose. The primary outcome variable was a numeric count of non-fatal overdoses reported across the follow-up period. Analysis of the predictors of overdose was restricted to PWID because the vast majority of overdoses in this cohort were reported by PWID. Negative binomial regression with robust error variance, adjusted for age, sex, incarceration history and previous overdose, was used to calculate risk ratios for baseline predictors of the number of non-fatal overdoses. This approach was used since most participants reported no overdoses, resulting in over-dispersion of the number of overdoses (i.e., the standard deviation was considerably greater than the mean). Adjustment was limited to four covariates due to the small number of participants reporting overdose and the small number of events. We adjusted for age, sex and adult incarceration history as these three factors have consistently emerged as predictors of fatal overdose after release from prison (Hobbs et al., 2006; Binswanger et al., 2007; Farrell and Marsden, 2008; Binswanger et al., 2013; Kinner et al., 2013a; Chang et al., 2015; Pierce et al., 2015). Lifetime history of overdose was selected as it is a strong predictor of future fatal overdose among PWID (Stoové et al., 2009; Darke et al., 2011). Data were analysed using StataTM version 13.1 (Stata Corporation, 2013). 3. Results

2.4.1. Calculating non-fatal overdose incidence rates. For all participants and for PWID, incidence of non-fatal overdose per 100 person years (PY) was calculated across all follow-up and for three discrete time periods: release to FU1, FU1 to FU2, and FU2 to FU3. 95% confidence intervals around incidence estimates were calculated assuming a Poisson distribution. Time at risk for overdose was calculated as the time between the index prison release and followup interview, or between follow-up interviews. Any time spent in prison during the follow-up period was subtracted from time at risk. Since the actual date of overdose event/s was not collected, overdoses reported during a follow-up period were estimated to have occurred at a date equidistant between release and interview, or between interviews. This approach has been used to estimate the date of events such as hepatitis C infection, where the actual date is unknown (Aitken et al., 2008). Since not all participants completed all three follow-up interviews, the risk of over or under-estimating overdose incidence between time points arises; that is, overdoses reported by participants who missed their first or second follow-up but participated in a second or third may have occurred at an earlier time point. To correct for this, missing time at risk data were imputed using the sample median and reported non-fatal overdoses were redistributed across person time accordingly. The median time to follow-up for the sample was 33 and 63 days between release and FU1, and FU1 and FU2, respectively. For example, if participant A completed follow-up interviews at 28 days (FU1) and at 182 days (FU3) but missed FU2, and reported two overdoses occurring between FU1 and FU3, their time-at-risk for non-fatal overdose for each discrete time period therefore became 28 days (release to FU1), 63 days (FU1 to FU2, imputed sample median) and 91 days (FU2 to FU3) – totalling 182 days. Participant A’s two overdoses were then allocated to have occurred equidistantly at 51 days and 102 days after release. Imputation of time at risk was only undertaken for participants who missed interviews between follow-up time points, not where attrition was permanent. Missing time at risk data were imputed for 260/1051 participants (25%) and reported overdoses for four participants were re-distributed. A sensitivity analysis was

Of the 1325 participants who completed a baseline interview, 81 (6%) were not released within 8 weeks of baseline and 193 (15%) did not complete a valid follow-up interview and were excluded. Among the remaining 1051 cohort members, 850 (81%) completed FU1, 844 (80%) completed FU2 and 891 (85%) completed FU3. One in four participants (24%) completed at least one follow-up while reincarcerated. Participants lost to follow-up were more likely than those completing at least one follow-up to be Indigenous (39% vs. 22%; 2 = 24.85, p < 0.01), unemployed for longer than 6 months prior to imprisonment (56% vs. 48%; 2 = 4.51, p = 0.03), have a history of prior adult incarceration (74% vs. 64%; 2 = 7.45, p < 0.01), and less likely to have a history of suicide attempt (15% vs. 22%; 2 = 4.40, p = 0.04). 3.1. Sample characteristics at baseline Table 1 presents the sample characteristics at baseline. Twenty three percent reported a history of non-fatal overdose and 56% reported a history of injecting drug use. PWID were more likely to report a history of non-fatal overdose and, on most indicators, exhibited greater disadvantage than their non-injecting counterparts. 3.2. Incidence of post-release non-fatal overdose Most participants did not report any overdose during followup. Among those reporting at least one overdose (n = 38; 4%), the median number of overdoses was 1 (IQR = 1–2), and this did not differ by sex, age or Indigenous status (all p > 0.05). Of those reporting an overdose after release from prison, 53% reported no history of overdose. Heroin was the most commonly reported drug involved in overdose (49%), followed by other opiates (31%), alcohol (26%) and benzodiazepines (24%). Methamphetamine was reported in 16% of overdose events. For the whole cohort (n = 1051) over the entire period of observation, 85 non-fatal overdoses were reported across 526.5 PY of follow-up. Among PWID (n = 591), 79 overdoses were reported

46

R.J. Winter et al. / Drug and Alcohol Dependence 153 (2015) 43–49

Table 1 Socio-demographic characteristics stratified by lifetime injecting history.

Male Aged 30 years or more Indigenous Fewer than 10 years schooling Unemployed longer than 6 months prior to prison Unstable accommodation 1 month prior to prison Removed from family as a child History of previous adult incarceration History of non-fatal drug overdose

PWID (n = 591) n (%)

Non-PWID (n = 460) n (%)

Total (N = 1051) n (%)

454 (76.8) 296 (50.1)** 125 (21.2) 297 (50.3)*** 219 (37.1)*** 117 (19.8)*** 129 (21.9)** 484 (82.2)*** 207 (35.0)***

375 (81.5) 270 (58.7) 105 (22.8) 144 (31.4) 122 (26.5) 52 (11.3) 70 (15.2) 189 (41.1) 31 (6.8)

829 (78.9) 566 (53.8) 230 (21.9) 441 (42.1) 341 (32.4) 169 (16.1) 199 (19.0) 673 (64.2) 238 (22.7)

PWID = people who inject drugs. * p < 0.05. ** p < 0.01. *** p < 0.001. Table 2 Non-fatal overdose incidence density (per 100 PY) at three discrete time periods, and overall, following release from prison. All prisoners a

Overall (median 202 days post-release) Release–FU1 (median 33 days post-release) FU1–FU2 (median 33–96 days post-release) FU2–FU3 (median 97–194 days post-release) a

PWID

Non-PWID

NFOD

n

PY

Incidence (95% CI)

NFOD

n

PY

Incidence (95% CI)

NFOD

n

PY

Incidence (95% CI)

85

1051

526.55

16.14 (12.89–19.96)

79

591

292.12

27.04 (21.41–33.71)

6

460

234.44

2.56 (0.93–5.57)

12

1051

104.16

11.52 (5.95–20.12)

10

591

58.70

17.04 (8.17–31.33)

2

460

45.46

4.40 (0.53–15.89)

40

997

177.49

22.54 (16.10–30.69)

37

560

97.88

37.80 (26.62–52.11)

3

438

79.61

3.77 (0.78–11.01)

33

891

244.90

13.47 (9.27–18.92)

32

494

135.53

23.61 (16.15–33.33)

1

397

109.37

0.91 (0.02–5.09)

Total number of reported non-fatal overdoses (NFOD); FU = follow-up; PY = person-years; PWID = people who inject drugs; 95% CI = 95% confidence interval.

across 292.1 PY. Table 2 shows the incidence of non-fatal overdose for the entire period of observation and at each discrete time period since release, for the full sample and stratified by PWID status. Fig. 1 displays the trend over the three post-release time periods. A sensitivity analysis was conducted excluding participants who were re-imprisoned during follow-up; incidence point estimates did not change substantively, varying by a maximum of 10%.

3.3. Pre-release predictors of post-release non-fatal overdose Given that most overdoses (n = 79/85) occurred among PWID, analysis of the predictors of overdose was restricted to this group (n = 591). Table 3 shows the unadjusted relative risk (RR) of postrelease non-fatal overdose for key demographic, drug use, mental health and risk behaviour variables. For each indicator variable, the RR is then adjusted for sex, age, previous history of adult incarceration and history of non-fatal overdose. There was no bivariate association between reporting overdose and receiving the case-management intervention (Kinner et al., 2014) so to preserve statistical power this variable was not adjusted for in multivariable analyses. Adjusted for sex, age, prior incarceration and non-fatal overdose history, the risk of non-fatal overdose was greater for those who reported: being unemployed for longer than 6 months prior to imprisonment, being removed from their family as a child, ever being diagnosed with a mental illness, experiencing very high psychological distress at baseline, using pharmaceutical opioids or benzodiazepines other than prescribed at least weekly in the 3 months prior to imprisonment, and ever being on OST. Preimprisonment high-risk alcohol consumption was protective for post-release overdose (Table 3).

4. Discussion Fig. 1. Non-fatal overdose incidence density (per 100 PY) at three discrete time periods following release from prison. (a) Median days between release and followup 1 = 33 days. (b) Median days between follow-ups 1 and 2 = 63 days (median 33–96 days post-release). (c) Median days between follow-ups 2 and 3 = 97 days (median 97–194 days post-release). NB: PWID = people who inject drugs.

To our knowledge, this is the first study to estimate the incidence and pre-release predictors of non-fatal overdose among adults leaving prison. We observed a high incidence of non-fatal overdose among PWID throughout the period of observation. The

R.J. Winter et al. / Drug and Alcohol Dependence 153 (2015) 43–49

47

Table 3 Unadjusted and adjusted risk ratios for pre-release predictors of non-fatal overdose among PWID following release from prison (n = 591).

Demographics and social circumstances Male Aged 30 years or more History of previous adult incarceration History of non-fatal drug overdose Indigenous Fewer than 10 years schooling Unemployed longer than 6 months prior to prison Unstable accommodation Removed from family as a child Months served in prison Randomised to intervention Mental health Very high psychological distressb Diagnosed mental disorder Ever attempted suicide Ever self-harmed Pre-imprisonment alcohol and other drug use High risk alcohol consumptionc At least weekly cannabis use At least weekly heroin use At least weekly non-prescribed opioid used At least weekly non-prescribed benzodiazepine use At least weekly amphetamine use Injecting behaviour and drug treatment Years since first injection Ever used drugs in prison Injected drugs during current prison sentence Ever on opioid substitution treatment a b c d * ** ***

Unadjusted relative risk (95% CI)

Adjusted relative riska (95% CI)

1.02 (0.31–3.36) 2.03 (0.75–5.48) 4.92 (1.44–16.82)* 2.22 (0.83–5.94) 1.63 (0.49–5.38) 1.18 (0.44–3.16) 5.36 (2.23–12.90)*** 2.75 (1.02–7.43)* 5.82 (2.44–13.87)*** 0.95 (0.89–1.01) 1.15 (0.43–3.05)

– – – – 1.93 (0.64–5.75) 0.69 (0.30–1.59) 4.43 (1.89–10.37)*** 1.66 (0.59–4.70) 4.37 (1.75–10.88)** 0.95 (0.89–1.02) 1.58 (0.62–4.04)

3.84 (1.15–12.78)* 4.22 (1.74–10.24)*** 3.44 (1.38–8.63)** 1.53 (0.42–5.59)

4.31 (1.41–13.18)** 2.97 (1.25–7.03)* 2.40 (0.92–6.28) 1.43 (0.42–4.85)

0.26 (0.10–0.70)** 0.41 (0.17–0.96)* 1.77 (0.67–4.68) 3.55 (1.28–9.83)* 2.30 (1.00–5.34)* 0.91 (0.35–2.38)

0.33 (0.13–0.89)* 0.46 (0.19–1.09) 1.65 (0.60–4.58) 4.03 (1.53–10.65)** 2.64 (1.09–6.35)* 0.94 (0.40–2.20)

1.01 (0.95–1.08) 1.02 (0.38–2.70) 1.45 (0.52–4.03) 2.32 (0.86–6.21)

0.97 (0.90–1.05) 0.84 (0.31–2.30) 1.92 (0.68–5.43) 3.02 (1.20–7.61)*

Adjusted for age, sex, previous incarceration history and lifetime non-fatal overdose history. Compared with low, medium or high distress as measured by the Kessler 10. Compared with low/no risk, measured using the Alcohol Use Disorder Identification Tool (AUDIT). Non-prescribed use of opioids other than heroin (includes methadone, buprenorphine). p ≤ 0.05. p ≤ 0.01. p ≤ 0.001.

period of greatest risk in this cohort was between 1 and 3 months following release, where the incidence was approximately twice as high as in the immediate post-release period. Although there are no directly comparable, published studies, the overall incidence of non-fatal overdose in this study (27.0/100 PY) is similar to that observed in a cohort of recently incarcerated illicit drug users in Vancouver, Canada (26.3/100 PY) (Kinner et al., 2012). In that study, recent incarceration was associated with a doubling of the odds of non-fatal overdose, however the period of observation was over 15 years. Most community studies of PWID have focussed on lifetime overdose and those that have measured overdose over a specific time frame have produced inconsistent results, with studies in the United Kingdom and Australia reporting 12-month prevalence rates ranging from 9% to 41% (Darke and Ross, 1996; Gossop et al., 1996; Warner-Smith et al., 2002). Studies of overdose mortality among ex-prisoners have shown a marked increase in risk of death in the first 2 weeks post-release compared with the following 10 weeks (Merrall et al., 2010). There is a high concentration of overdose deaths immediately following release from prison and a downward trend is usually observed thereafter (Bird and Hutchinson, 2003; Singleton et al., 2003; Farrell and Marsden, 2008). By contrast, in this cohort the incidence of non-fatal overdose was lower in the first month post-release than in the subsequent 2 months. The higher incidence of non-fatal overdose between 1 and 3 months post-release could be explained by participants’ patterns of return to drug use. A significant proportion of participants may have initially tried to abstain from drug use but failed after a few months, a common experience among ex-prisoners (Binswanger et al., 2012). Nevertheless, release from prison may precipitate a period of acutely elevated risk; for 43% of

participants reporting post-release overdose, it was their first ever overdose. After adjusting for age, sex, previous incarceration and lifetime overdose history, the pre-release predictors of post-release overdose included being unemployed for longer than 6 months prior to imprisonment, having been removed from family as a child, reporting current very high psychological distress, a diagnosed mental disorder, at least weekly use of benzodiazepines and/or pharmaceutical opioids in the 3 months prior to imprisonment, and ever having been on OST. High-risk alcohol consumption was protective against overdose. The findings did not differ greatly from those of previous community studies of non-fatal overdose among PWID. Poly-drug use (Brugal et al., 2002; Dietze et al., 2005; Kerr et al., 2007), poor mental health (Bartoli et al., 2014), social isolation – which may be compounded by unemployment, family disconnectedness (Rossow and Lauritzen, 1999) and childhood trauma (Lake et al., 2015) may all increase the chances of engaging in high-risk drug use and increase the likelihood of accidental overdose (Farrell et al., 1996). Although lifetime OST history predicted overdose in this cohort, this is possibly a marker of severity of dependence; OST at the time of overdose was not assessed in this study. Alternatively, it could be a marker of OST interruption: OST is unavailable for most prisoners in Queensland. The results highlight that not all prisoners with a history of injecting drug use are at equal risk for non-fatal overdose, and that identifying those with a confluence of risk markers has the potential to improve their post-release outcomes through informing targeted preventive interventions both in prison and post-release. The risk markers identified here are easily identifiable through routine reception and pre-release screening; those most at risk could be flagged for support during the transition into the community.

48

R.J. Winter et al. / Drug and Alcohol Dependence 153 (2015) 43–49

While it was not the aim of this study to construct a pre-release screening tool, these data provide a useful first step in doing so. Transitional programmes should, at a minimum, include a component of drug education incorporating overdose risk reduction strategies and overdose response training. The provision of naloxone post-release, which when administered reverses the effects of overdose, is also currently being trialled (Strang et al., 2013) and has had success in community studies of PWID (Strang et al., 2008). The provision of OST in prison and on release into the community also has demonstrated effectiveness in preventing drug-related mortality (Degenhardt et al., 2014), and is a clear gap in the suite of evidence-based responses to opiate dependence in the jurisdiction where this study was conducted. Drug education should also cover a wide variety of drug types; while the majority of overdoses reported in this study were opioid-related, many also involved other drugs and some jurisdictions have recently experienced an increase in methamphetamine-related harms related to an increase in purity (Heilbronn et al., 2013; Scott et al., 2015). The fact that the study intervention, which provided individual support and linkage to general health and welfare services, had no effect on post-release overdose suggests that specific, targeted and tailored education and intervention are necessary for PWID transitioning from prison to community.

4.1. Limitations Given evidence of a concentration of fatal overdose very soon after release from prison (Merrall et al., 2010), our approach of using the mid-point between follow-ups to calculate non-fatal overdose incidence is likely to have produced a conservative estimate, if overdoses were more likely to have occurred closer to the prison release date (Merrall et al., 2010). Participants lost to follow-up in this study were more likely to have an incarceration history and to have been unemployed prior to imprisonment, both of which were also strong predictors of post-release overdose. This may have resulted in underestimation of the true rate of overdose. Our study relied on ex-prisoners self-reporting overdose events. While this may result in either over- or under-reporting, self-report is arguably the best way to ascertain these events since many overdoses never come to the attention of treatment services (Davidson et al., 2002) and thus administrative data sources do not provide the whole picture. There is some evidence that PWID’s self-reported data are sufficiently reliable and valid to provide descriptions of drug use, drug-related problems and the natural history of drug use (Darke, 1998). The small number of participants reporting overdose, and the small number of overdose events, precluded the construction of a fully adjusted multivariable predictor model. Further work is required to investigate independent predictors for internal and external validity. Analysis of the predictors of postrelease overdose utilised a combination of current (pre-release) and past (pre-imprisonment) measures which may not represent post-release circumstances surrounding risky drug use. Nevertheless, the ability to identify risk characteristics prior to release is an important step in identifying those most at risk of drug-related harms. A useful next step for research in this area would be to examine how post-release circumstances and experiences shape drug using behaviours and associated risk of overdose. Finally, drug use behaviours and characteristics may differ between users of different types of drugs. In this study we included both opioid and non-opioid overdoses. Consistent with previous research (Darke and Hall, 2003; Kerr et al., 2007) poly-drug use was a striking feature of participants who reported overdose, such that use of a broad overdose category is unlikely to have affected our estimates.

4.2. Conclusion The findings of this study highlight the significant morbidity experienced by ex-prisoners, particularly PWID, and the need for better preventive measures. Imprisonment provides a unique opportunity to identify those most at risk of post-release drugrelated harms such as overdose. Effective interventions to prevent both fatal and non-fatal overdose, such as the provision of OST during imprisonment and continued after release into the community, are needed along with further research on pre-release education and overdose response training, and the provision of naloxone. Worldwide, OST is currently unavailable for many prisoners. Delivering low-cost OST to high-risk groups such as PWID in prison, and during transition to the community, is likely to reduce both fatal and non-fatal overdose. Given the complexity of the health and social needs of this group, OST should be provided in the context of broader transitional support, particularly for those most likely to return to high risk drug use following release. Role of funding source The Passports project was funded by National Health and Medical Research Council (NHMRC) Strategic Award #409966. RW is supported by NHMRC Postgraduate Scholarship #603756 and the NHMRC Centre for Research Excellence on Injecting Drug Use (CREIDU, #1001144). MS is supported by NHMRC Career Development Fellowship #1090445. LD is supported by NHMRC Principal Research Fellowship #1041742. MH is supported by NHMRC Senior Research Fellowship #1062877. Rebecca Jenkinson is supported by NHMRC Early Career Fellowship #1054674. Damien McCarthy is partially supported by CREIDU. SK is supported by NHMRC Career Development Award #1004765. The funding sources had no role in the study design, collection or analysis of data, or in the writing or submission of the manuscript. The National Drug and Alcohol Research Centre at the University of NSW is supported by funding from the Australian Government under the Substance Misuse Prevention and Service Improvements Grants Fund. The authors acknowledge the contribution to this work of the Victorian Operational Infrastructure Support Programme. Contributors SK designed the Passports study; RW, SK, LD, MH and RJ devised the sub-study aims and approach. TS advised on statistical approach. RW conducted data analysis and wrote the manuscript. DM contributed to data analysis and manuscript editing. All authors contributed to multiple drafts and approved the final manuscript. Conflict of interest None declared. Acknowledgements The authors wish to thank Queensland Corrective Services for assistance with data collection, and Passports study participants for sharing their stories. The views expressed herein are solely those of the authors, and in no way reflect the views or policies of Queensland Corrective Services. References Aitken, C.K., Lewis, J., Tracy, S., Spelman, T., Bowden, D.S., Bharadwaj, M., Drummer, H., Hellard, M.E., 2008. High incidence of hepatitis C virus reinfection in a cohort of injecting drug users. Hepatology 48, 1746–1752.

R.J. Winter et al. / Drug and Alcohol Dependence 153 (2015) 43–49 Andrews, G., Slade, T., 2001. Interpreting scores on the Kessler Psychological Distress Scale (K10). Aust. N. Z. J. Public Health 25, 494–497. Australian Institute of Health and Welfare, Cat. no. PHE 170, 2013. The Health of Australia’s Prisoners 2012. Australian Institute of Health and Welfare, Canberra. Babor, T.F., Higgins-Biddle, J.C., Saunders, J.B., Monteiro, M.G., 2001. AUDIT: The Alcohol Use Disorders Identification Test. Guidelines for Use in Primary Care, second ed. World Health Organisation, Geneva. Bartoli, F., Carrà, G., Brambilla, G., Carretta, D., Crocamo, C., Neufeind, J., Baldacchino, A., Humphris, G., Clerici, M., 2014. Association between depression and non-fatal overdoses among drug users: a systematic review and meta-analysis. Drug Alcohol Depend. 134, 12–21. Binswanger, I.A., Stern, M.F., Deyo, R.A., Heagerty, P.J., Cheadle, A., Elmore, J.G., Koepsell, T.D., 2007. Release from prison – a high risk of death for former inmates. N. Engl. J. Med. 356, 157–165. Binswanger, I.A., Nowels, C., Corsi, K.F., Glanz, J., Long, J., Booth, R.E., Steiner, J.F., 2012. Return to drug use and overdose after release from prison: a qualitative study of risk and protective factors. Addict. Sci. Clin. Pract. 7, 3, http://dx.doi. org/10.1186/1940-0640-7-3 Binswanger, I.A., Blatchford, P.J., Mueller, S.R., Stern, M.F., 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, 592–600. Bird, S.M., Hutchinson, S.J., 2003. Male drugs-related deaths in the fortnight after release from prison: Scotland, 1996–99. Addiction 98, 185–190. Brugal, M.T., Barrio, G., Fuente, L.D.L., Regidor, E., Royuela, L., Suelves, J.M., 2002. Factors associated with non-fatal heroin overdose: assessing the effect of frequency and route of heroin administration. Addiction 97, 319–327. Brugal, M.T., Domingo-Salvany, A., Puig, R., Barrio, G., Garcia de Olalla, P., de la Fuente, L., 2005. Evaluating the impact of methadone maintenance programmes on mortality due to overdose and aids in a cohort of heroin users in Spain. Addiction 100, 981–989. Centers for Disease Control and Prevention, 2013. Vital signs: overdoses of prescription opioid pain relievers and other drugs among women – United States, 1999–2010. MMWR 62, 537–542. Cerdá, M., Ransome, Y., Keyes, K.M., Koenen, K.C., Tracy, M., Tardiff, K.J., Vlahov, D., Galea, S., 2013. Prescription opioid mortality trends in New York City, 1990–2006: examining the emergence of an epidemic. Drug Alcohol Depend. 132, 53–62. Chang, Z., Lichtenstein, P., Larsson, H., Fazel, S., 2015. Substance use disorders, psychiatric disorders, and mortality after release from prison: a nationwide longitudinal cohort study. Lancet Psychiatry 2, 422–430. Cruickshank, C.C., Dyer, K.R., 2009. A review of the clinical pharmacology of methamphetamine. Addiction 104, 1085–1099. Darke, S., Ross, J., 1996. Overdose among heroin users in Sydney, Australia: prevalence and correlates of non-fatal overdose. Addiction 91, 405–412. Darke, S., 1998. Self-report among injecting drug users: a review. Drug Alcohol Depend. 51, 253–263. Darke, S., Hall, W., 2003. Heroin overdose: research and evidence-based intervention. J. Urban Health 80, 189–200. Darke, S., Williamson, A., Ross, J., Mills, K.L., Havard, A., Teesson, M., 2007. Patterns of nonfatal heroin overdose over a 3-year period: findings from the Australian treatment outcome study. J. Urban Health 84, 283–291. Darke, S., Kaye, S., McKetin, R., Duflou, J., 2008. Major physical and psychological harms of methamphetamine use. Drug Alcohol Rev. 27, 253–262. Darke, S., Mills, K.L., Ross, J., Teesson, M., 2011. Rates and correlates of mortality amongst heroin users: findings from the Australian Treatment Outcome Study (ATOS), 2001–2009. Drug Alcohol Depend. 115, 190–195. Davidson, P.J., Ochoa, K.C., Hahn, J.A., Evans, J.L., Moss, A.R., 2002. Witnessing heroin-related overdoses: the experiences of young injectors in San Francisco. Addiction 97, 1511–1516. Davoli, M., Bargagli, A.M., Perucci, C.A., Schifano, P., Belleudi, V., Hickman, M., Salamina, G., Diecidue, R., Vigna-Taglianti, F., Faggiano, F., 2007. Risk of fatal overdose during and after specialist drug treatment: the VEdeTTE study, a national multi-site prospective cohort study. Addiction 102, 1954–1959. Degenhardt, L., Larney, S., Kimber, J., Gisev, N., Farrell, M., Dobbins, T., Weatherburn, D.J., Gibson, A., Mattick, R., Butler, T., 2014. The impact of opioid substitution therapy on mortality post-release from prison: retrospective data linkage study. Addiction 109, 1306–1317. Dietze, P.M., Jolley, D., Fry, C., Bammer, G., 2005. Transient changes in behaviour lead to heroin overdose: results from a case-crossover study of non-fatal overdose. Addiction 100, 636–642. Farrell, M., Neeleman, J., Griffiths, P., Strang, J., 1996. Suicide and overdose among opiate addicts. Addiction 91, 321–323. Farrell, M., Marsden, J., 2008. Acute risk of drug-related death among newly released prisoners in England and Wales. Addiction 103, 251–255. Fazel, S., Bains, P., Doll, H., 2006. Substance abuse and dependence in prisoners: a systematic review. Addiction 101, 181–191.

49

Gossop, M., Griffiths, P., Powis, B., Williamson, S., Strang, J., 1996. Frequency of non-fatal heroin overdose: survey of heroin users recruited in non-clinical settings. BMJ 313, 402. Heilbronn, C., Gao, C.X., Lloyd, B., Smith, K., Best, D., Lubman, D.I., 2013. Trends in amphetamine-related harms in Victoria. Med. J. Aust. 199, 395. Hobbs, M., Krazlan, K., Ridout, S., Mai, Q., Knuiman, M., Chapman, R., 2006. Mortality and morbidity in prisoners after release from prison in Western Australia 1995–2003. Trends Iss. Crime Crim. Justice 320, 1–6. Kaye, S., Darke, S., 2004. Non-fatal cocaine overdose among injecting and non-injecting cocaine users in Sydney, Australia. Addiction 99, 1315–1322. Kerr, T., Fairbairn, N., Tyndall, M., Marsh, D., Li, K., Montaner, J., Wood, E., 2007. Predictors of non-fatal overdose among a cohort of polysubstance-using injection drug users. Drug Alcohol Depend. 87, 39–45. Kinner, S.A., 2006. Continuity of health impairment and substance misuse among adult prisoners in Queensland, Australia. Int. J. Prison. Health 2, 101–113. Kinner, S.A., Milloy, M.J., Wood, E., Qi, J., Zhang, R., Kerr, T., 2012. Incidence and risk factors for non-fatal overdose among a cohort of recently incarcerated illicit drug users. Addict. Behav. 37, 691–696. Kinner, S.A., Forsyth, S., Williams, G., 2013a. Systematic review of record linkage studies of mortality in ex-prisoners: why (good) methods matter. Addiction 108, 38–49. Kinner, S.A., Lennox, N., Williams, G.M., Carroll, M., Quinn, B., Boyle, F.M., Alati, R., 2013b. Randomised controlled trial of a service brokerage intervention for ex-prisoners in Australia. Contemp. Clin. Trials 36, 198–206. Kinner, S.A., van Dooren, K., Boyle, F.M., Longo, M., Lennox, N., 2014. Development of an intervention to increase health service utilisation in ex-prisoners. Health Justice 2, 1–8. Lake, S., Hayashi, K., Milloy, M.J., Wood, E., Dong, H., Montaner, J., Kerr, T., 2015. Associations between childhood trauma and non-fatal overdose among people who inject drugs. Addict. Behav. 43, 83–88. Merrall, E.L.C., Kariminia, A., Binswanger, I.A., Hobbs, M.S., Farrell, M., Marsden, J., Hutchinson, S.J., Bird, S.M., 2010. Meta-analysis of drug-related deaths soon after release from prison. Addiction 105, 1545–1554. Milloy, M.-J., Kerr, T., Mathias, R., Zhang, R., Montaner, J.S., Tyndall, M., Wood, E., 2008. Non-fatal overdose among a cohort of active injection drug users recruited from a supervised injection facility. Am. J. Drug Alcohol Abuse 34, 499–509. Ochoa, K.C., Davidson, P.J., Evans, J.L., Hahn, J.A., Page-Shafer, K., Moss, A.R., 2005. Heroin overdose among young injection drug users in San Francisco. Drug Alcohol Depend. 80, 297–302. Pierce, M., Bird, S.M., Hickman, M., Millar, T., 2015. National record linkage study of mortality for a large cohort of opioid users ascertained by drug treatment or criminal justice sources in England, 2005–2009. Drug Alcohol Depend. 146, 17–23. Pollini, R.A., McCall, L., Mehta, S.H., Vlahov, D., Strathdee, S.A., 2006. Non-fatal overdose and subsequent drug treatment among injection drug users. Drug Alcohol Depend. 83, 104–110. Rossow, I., Lauritzen, G., 1999. Balancing on the edge of death: suicide attempts and life-threatening overdoses among drug addicts. Addiction 94, 209–219. Scott, N., Caulkins, J.P., Ritter, A., Quinn, C., Dietze, P., 2015. High-frequency drug purity and price series as tools for explaining drug trends and harms in Victoria, Australia. Addiction 110, 120–128. Seal, K.H., Kral, A.H., Gee, L., Moore, L.D., Bluthenthal, R.N., Lorvick, J., Edlin, B.R., 2001. Predictors and prevention of nonfatal overdose among street-recruited injection heroin users in the San Francisco Bay area, 1998–1999. Am. J. Public Health 91, 1842–1846. Singleton, N., Pendry, E., Taylor, C., Farrell, M., Marsden, J., 2003. Drug-Related Mortality Among Newly Released Offenders. UK Home Office, London. Stoové, M.A., Dietze, P.M., Jolley, D., 2009. Overdose deaths following previous non-fatal heroin overdose: record linkage of ambulance attendance and death registry data. Drug Alcohol Rev. 28, 347–352. Strang, J., McCambridge, J., Best, D., Beswick, T., Bearn, J., Rees, S., Gossop, M., 2003. Loss of tolerance and overdose mortality after inpatient opiate detoxification: follow up study. BMJ 326, 959–960. Strang, J., Manning, V., Mayet, S., Best, D., Titherington, E., Santana, L., Offor, E., Semmler, C., 2008. Overdose training and take-home naloxone for opiate users: prospective cohort study of impact on knowledge and attitudes and subsequent management of overdoses. Addiction 103, 1648–1657. Strang, J., Bird, S.M., Parmar, M.K., 2013. Take-home emergency naloxone to prevent heroin overdose deaths after prison release: rationale and practicalities for the N-ALIVE randomized trial. J. Urban Health 90, 983–996. van Ameijden, E.J.C., Langendam, M.W., Coutinho, R.A., 1999. Dose–effect relationship between overdose mortality and prescribed methadone dosage in low-threshold maintenance programs. Addict. Behav. 24, 559–563. Warner-Smith, M., Darke, S., Day, C., 2002. Morbidity associated with non-fatal heroin overdose. Addiction 97, 963–967.