• Open Access

Imprisonment of opioid-dependent people in New South Wales, Australia, 2000–2012: a retrospective linkage study

Authors


  • The authors have stated they have no conflict of interest.

Abstract

Objective: There are few data about the incarceration of opioid-dependent people involving large representative cohorts. We aimed to determine the prevalence and duration of incarceration in a large cohort of opioid-dependent people in Australia using data linkage methods, and estimate the costs associated with their incarceration.

Method: Retrospective linkage study of all entrants to opioid substitution therapy (OST) for the treatment of opioid dependence in NSW, 1985–2010, with data on incarceration, 2000–2012. The number and duration of incarcerations were calculated. The average daily cost of incarceration was applied to days of incarceration in the cohort.

Results: Among 47,196 opioid-dependent people, 37% (43% of men and 24% of women) had at least one episode of incarceration lasting one or more days. Men had a median of three (ranging between 1–47) incarcerations, and women, two (1–35). Indigenous men spent 23% of follow-up time incarcerated, compared with 8% for non-Indigenous men; similarly, Indigenous women spent a substantially greater proportion of time incarcerated than non-Indigenous women (8% vs. 2%). Costs of incarceration of this cohort between 2000 and 2012 totalled nearly AUD$3 billion.

Conclusions: This is the first study to examine incarceration of opioid-dependent people across an entire population of such users. Our findings suggest that a substantial minority of opioid-dependent people experience incarceration, usually on multiple occasions and at significant cost. Treatment for opioid dependence, inside and outside prisons, may help reduce incarceration of this cohort.

Studies of illicit opioid-dependent people have identified high rates of criminal involvement, contact with police and incarceration.1–3 Important reasons for understanding patterns of incarceration in this population include the association between exposure to prison and unemployment, financial hardship and an increased risk of stress-related illnesses.4,5 Among people who inject drugs, having been in prison is a risk factor for hepatitis C seroconversion.6,7 Furthermore, the four weeks immediately post-release from prison are a period of particularly heightened risk of drug overdose death.8 Describing incarceration experiences at a population level is a step towards understanding how incarceration may contribute to the overall burden of disease and mortality in opioid-dependent people.

Most studies presenting data on the involvement of this group in the criminal justice system are limited by the use of convenience or respondent-driven sampling. First, such findings may not be representative of the broader opioid-dependent population. Second, self-report data may be subject to biases that under- or over-estimate incarceration. Third, incarceration data collected via self-report are typically limited to prevalence within a specified – often short – time period; accurate data on the number of episodes of incarceration over time and duration of each incarceration are difficult to ascertain from self-report. Such details are important in order to examine issues such as costs of incarcerating people with opioid dependence (and therefore the burden on the criminal justice system and potential opportunity costs with respect to other interventions).

In this paper, we examined incarceration in a cohort of all opioid-dependent people who had registered for opioid substitution therapy (OST) for opioid dependence in New South Wales (NSW). We obtained data for this cohort on the number and duration of incarceration episodes (available since 2000 in NSW) over a twelve-year period. We aimed to determine:

  • The proportion of opioid-dependent people incarcerated on a full-time basis between 2000 and 2012;
  • The median number and length of full-time incarceration episodes;
  • Cumulative time and proportion of follow-up spent in prison, 2000–2012; and
  • The cumulative costs associated with incarceration of this cohort.

Methods

Data sources

The study cohort was identified from the Pharmaceutical Drugs of Addiction System (PHDAS), a database of the NSW Ministry of Health that includes details of all people entering OST since 1985. Identifiers in this database are considered reliable, as individuals must provide identification to the prescribing doctor. Data were extracted for all persons registering for OST in the calendar years 1985 to 2010 inclusive.

The PHDAS data were probabilistically linked to prison data from the Re-offending Database (ROD) of the NSW Bureau of Crime Statistics and Research (BOCSAR). These data are supplied to BOCSAR by Corrective Services NSW and include records of all episodes of full-time incarceration. Data were available from 1 January 2000 to 31 March 2012. Linkage between the PHDAS and ROD was undertaken by BOCSAR staff. Vital status of cohort members was determined through probabilistic linkage of the PHDAS to the National Death Index; this linkage was undertaken by staff of the Australian Institute of Health Welfare. Data on costs of incarceration were obtained from the Australian Government Productivity Commission (http://www.pc.gov.au/gsp/rogs).

Analyses

Cohort follow-up time started at January 2000, or whenever the participant turned 10 years of age (whichever was later). Follow-up ceased in March 2012, or when death occurred (whichever was earlier). For analysis purposes, incarceration episodes comprising reception and release on the same day were excluded. In calculating the duration of the episode(s), only those incarceration episodes completely contained within the follow-up period were included (i.e. reception date on or after 1 January 2000 and release date on or before 31 March 2012). Median numbers of incarceration episodes and duration of completed episodes were calculated, as well as cumulative days in prison. Total time in prison during the follow-up period was calculated by including the relevant portions of any episodes that started before 1 January 2000 or ended after 31 March 2012.

Indigenous people are highly over-represented in prison populations in Australia.9,10 Participants were defined as Indigenous if flagged as Aboriginal and/or Torres Strait Islander in any of their prison or OST records. Participants were defined as non-Indigenous if flagged as such in either database. Indigenous status was coded as missing for remaining participants.

Costs of incarcerating opioid-dependent people were calculated by applying the average daily cost – system-wide – of incarcerating an individual, to the number of days in prison (from the annual publications of the Australian Government's Productivity Commission (see http://www.pc.gov.au/gsp/rogs). Notably, these costs do not include the costs of health interventions delivered to prisoners, as these are borne by the NSW Ministry of Health; they are estimates of the costs of incarceration only. Costs were indexed to 2012 Australian dollars.

Ethical approvals

Support for this study was obtained from all data custodians (Pharmaceutical Services Unit, NSW Health, BOCSAR and AIHW). Ethical approval was obtained from the University of New South Wales, NSW Health's Population & Health Services Research Ethics Committee, AIHW, the Alfred Hospital, Corrective Services NSW, NSW Justice Health, and the NSW Aboriginal Health and Medical Research Council (AHMRC).

Results

The cohort

There were 47,196 people in the cohort at the beginning of follow-up. The cohort was 33.0% female, with 15.0% identified as Indigenous, 70.0% non-Indigenous, and 15.0% with missing or unknown Indigenous status as flagged in the databases. In total, 43.3% of men and 23.9% of women in the cohort had at least one prison episode. Overall this amounted to 36.9% of the cohort overall (n=17,402; Table 1).

Table 1. Completed incarceration episodes among opioid-dependent people in NSW, January 2000 – March 2012.
 MalesFemalesPersons
 N = 31,623N = 15,573N = 47,196
N (%) with any completed prison episode13,679
(43.3%)
3,723 (23.9%)17,402
(36.9%)
Median (LQ-UQ) (min-max) number of completed prison episodes per person
Sentenced
Not sentenced
Juvenile
Adult
3 (1–6) (1–47)
2 (1–3) (0–19)
1 (1–3) (0–29)
0 (0–0) (0–35)
3 (1–6) (0–38)
2 (1–5) (1–35)
1 (0–2) (0–20)
1 (1–3) (0–28)
0 (0–0) (0–28)
2 (1–5) (0–26)
3 (1–6) (1–47)
1 (1–3) (0–20)
1 (1–3) (0–29)
0 (0–0) (0–35)
3 (1–6) (0–38)
Mean (SD) number of completed prison episodes per person
Sentenced
Not sentenced
Juvenile
Adult
4.6 (4.3)
2.4 (2.4)
2.2 (2.7)
0.3 (1.7)
4.3 (3.8)
3.9 (4.0)
1.6 (2.0)
2.3 (2.7)
0.3 (1.6)
3.6 (3.6)
4.5 (4.2)
2.2 (2.3)
2.2 (2.7)
0.3 (1.7)
4.1 (3.8)

There were a total of 101,490 prison episodes for people in the cohort. Of these, almost one in five (18,889, 18.6%) involved the person entering and leaving custody within the same day, and were excluded from the analyses. For the purpose of examining the duration of incarceration episodes, we also excluded 5,114 episodes that started before 1 January 2000 and/or finished after 31 March 2012, leaving 77,485 complete episodes. Four in ten men (43.3%) and one in four women (23.9%) in the cohort had at least one incarceration episode lasting longer than a day. There was a median of three episodes for adult men and two episodes for adult women across the entire cohort (n=47,196; Table 1).

The cohort included people who had first entered OST in the 1980s, and who may have ceased illicit opioid use and offending by the time period for which prison data were available. In order to assess if incarceration prevalence was being underestimated through the inclusion of these individuals, a sensitivity analysis was undertaken using only cohort members who had an episode of OST between 2000–2012. There were negligible differences in the proportion of clients with an incarceration episode and the median and mean number of episodes, as almost all people who had first entered OST prior to 2000 also re-entered OST after 2000.

As shown in Table 2, 28.3% of completed incarceration episodes were 1–7 days in length; an additional 16.0% ranged between one week and a month (Table 2). Almost half of all episodes (46.2%) ranged from 32 to 365 days, with 9.4% lasting more than one year. The median length of adult incarceration episodes was 59 days overall, being longer for men (69 days) than women (28 days). There was considerable variation in the length of episodes, ranging from one day to 4,231 days (Table 2).

Table 2. Details of completed incarceration episodes among opioid-dependent people in NSW, January 2000 – March 2012.
 MalesFemalesPersons
Total number (%) of episodes62,95614,52977,485
Sentenced
Not sentenced
32,865 (52.2)
30,091 (47.8)
5,967 (41.1)
8,562 (58.9)
38,832 (50.1)
38,653 (49.9)
1–7 days
8–31 days
32–365 days
366+ days
16,691 (26.5)
9,732 (15.5)
29,939 (47.6)
6,594 (10.5)
5,270 (36.3)
2,660 (18.3)
5,873 (40.4)
726 (5.0)
21,961 (28.3)
12,392 (16.0)
35,812 (46.2)
7,320 (9.4)
Juvenile
Adult
4,572 (7.3)
58,384 (92.7)
1,136 (7.8)
13,393 (92.2)
5,708 (7.4)
71,777 (92.6)
Median episode length (LQ–UQ) (min–max)59 (7–183) (1–4,231)24 (3–114) (1–3,070)50 (6–181) (1–4,231)
Sentenced (n=38,832)
Not sentenced (n=38,653)
181 (90–341) (1–4,231)
7 (1–28) (1–3,007)
122 (60–240) (1–2,463)
5 (1–20) (1–3,070)
180 (89–309) (1–4,231)
6 (1–26) (1–3,070)
Juvenile (n=5,708)
Adult (n=71,777)
10 (2–38) (1–1,186)
69 (8–198) (1–4,231)
6 (1–21) (1–337)
28 (3–121) (1–3,070)
9 (1–34) (1–1,186)
59 (7–182) (1–4,231)
Mean (SD) length of episodes161.1 (286.8)91.9 (174.2)148.1 (270.7)
Sentenced
Not sentenced
281.8 (347.8)
29.3 (82.1)
193.7 (224.0)
21.0 (65.3)
268.2 (333.3)
27.5 (78.8)
Juvenile
Adult
40.1 (82.0)
170.6 (294.9)
22.1 (41.3)
97.8 (179.8)
36.5 (76.0)
157.0 (278.5)
Days in prison (for completed episodes)10,143,0881,335,37711,478,465
Sentenced
Not sentenced
9,260,031
883,057
1,155,717
179,660
10,415,748
1,062,717
Juvenile
Adult
183,222
9,959,866
25,053
1,310,324
208,275
1,1270,190

We examined the proportion of the cohort who spent time in incarceration each year (Table 3). Males had higher levels than females. In 2000, for example, 19.5% of males spent time in incarceration compared to 7.5% of females. There was some decline in the proportion of the cohort spending time in incarceration in the latter years of the study, with 14.0% of males and 5.2% of females spending time in custody in 2011.

Table 3. Percentage of cohort with any custody in each calendar year.a
Calendar yearMales
(n= 31,623)
Females
(n= 15,573)
Persons
(n= 47,196)
 %%%
  1. a Includes all people alive at 1/1/2000: percentages do not take account of deaths during observation period

200019.57.015.4
200120.57.916.3
200219.77.215.6
200319.67.115.5
200419.37.015.2
200519.27.015.2
200618.76.814.8
200718.36.814.5
200817.46.313.7
200916.65.813.0
201015.35.712.1
201114.05.211.1

We further considered the cumulative length of time cohort members spent incarcerated according to their age at the start of follow-up (1 January 2000). Table 4 shows that there was a steep decline in the proportion of follow-up time spent in prison with increasing age: males younger than 20 years at the beginning of follow-up spent 16.6% of their follow-up time incarcerated, compared to 3.8% of females younger than 20 years at the beginning of follow-up. These levels declined such that 2.6% and 0.2% of follow-up time was spent incarcerated among males and females respectively who were aged 55–59 years in January 2000.

Table 4. Time spent in custody by the cohort during follow-up according to age as at the beginning of follow-up (1st January 2000), 2000 – 2012.
 PeopleFollow-up time% of observation time in custody
Age at January 1st 2000NTotal days in custodyTotal days of observationMalesFemalesPeople
< 20 years6,7903,591,50529,991,04816.6%3.8%12.0%
20–24 years7,8753,525,54734,451,96813.7%3.6%10.2%
25–29 years8,3543,375,93036,224,94712.5%2.6%9.3%
30–34 years7,6262,265,31032,938,8889.5%1.8%6.9%
35–39 years7,5661,280,98232,452,2045.5%1.1%3.9%
40–44 years5,710616,55124,068,4693.4%0.6%2.6%
45–49 years2,535200,24310,536,2012.5%0.3%1.9%
50–54 years58549,9122,372,8662.5%0.9%2.1%
55–59 years1068,272414,8642.6%0.2%2.0%
60 + years4918182,9150.0%0.0%0.0%
 47,19614,914,270203,634,3709.9%2.3%7.3%

Table 5 presents the cumulative time, including incomplete episodes (those that started before or continued after the end of the study period), spent by the cohort in prison during the follow-up period. Out of a total of 474,596 person years (PY) of follow-up, the cohort spent 38,030 PY in prison. Men spent proportionally more time in prison (9.1% of total follow-up) than women (2.2%). Indigenous male offenders spent one-quarter of their follow-up time in prison (22.8%), compared to 7.9% of follow-up time for non-Indigenous male offenders. A similar finding was observed for women: Indigenous female offenders spent 7.5% of their follow-up time in prison, compared to 1.6% of non-Indigenous offenders.

Table 5. Cumulative time in prison (including incomplete episodes)a among opioid-dependent people in NSW, January 2000 – March 2012.
 Males
(N = 31,623)
Females
(N = 15,573)
Persons
(N = 47,196)
Total days of follow-up
(Person years of follow-up)
118,704,948
(324,996 PY)
54,641,096
(149,599 PY)
173,346,044
(474,596 PY)
Total days in prison (including incomplete episodes)a
(Person years)
 Juvenile
 Adult
12,415,268
(33,991 PY)
191,457
12,223,811
1,475,184
(4,039 PY)
27,013
144,8171
13,890,452
(38,030 PY)
218,470
13,671,982
% of follow-up time spent in prison9.1%2.2%6.8%
Clients for whom Indigenous status was knownMales
(N = 27,514)
Females
(N = 12,528)
Persons
(N = 40,042)
  1. a Includes days in the follow-up period spent in custody for episodes that had started prior to 2000, and/or which had not ended by March 2012.

  2. b Includes OST clients for whom at least one record indicated they were not Aboriginal or Torres Strait Islander, and for whom there were no records indicating that they were Aboriginal or Torres Strait Islander. Those with unknown Indigenous status are excluded.

Indigenous clients with a charge history   
 Total days of follow-up
Days in prison
% of follow-up time spent in prison
20,493,131
4,667,898
22.8%
10,275,415
767,304
7.5%
30,768,546
5,435,202
17.7%
Non-Indigenous clients with a charge historyb   
Total days of follow-up
Days in prison
% of follow-up time spent in prison
98,211,817
7,742,851
7.9%
44,365,681
707,878
1.6%
142,577,498
8,450,729
5.9%

The costs per prisoner per day were applied to the number of days spent in prison by the cohort each year (Table 6). In total over the period 2000–2012, it was estimated that incarceration of opioid-dependent people in this cohort cost $2.98 billion in constant AUD. There was variation across years in the annual costs, a function of changes in daily incarceration costs and the number of days of incarceration (Table 6).

Table 6. Estimated costs (2012 AUD) of incarceration of opioid-dependent people in NSW, January 2000 – March 2012.
 Cost per prisoner
per day (2012 AUD)
No. days in custody by cohort membersEstimated cost of custody
(2012 AUD)
  1. a Used 2010–2011 figure as this was the most recent available.

    
Jan–June 2000$209.52500,452$104,854,703
2000–2001$217.651,066,354$232,091,948
2001–2002$236.321,112,149$262,823,052
2002–2003$213.831,106,608$236,625,989
2003–2004$215.871,134,151$244,829,176
2004–2005$212.101,176,853$249,610,521
2005–2006$221.721,181,589$261,981,913
2006–2007$224.151,206,573$270,453,338
2007–2008$230.721,169,069$269,727,600
2008–2009$222.571,243,475$276,760,231
2009–2010$207.501,103,424$228,960,480
2010–2011$201.871,005,530$202,986,341
2011–Mar 2012$201.87a665,755$134,395,962
Total 2000–2012 13,671,982$2,976,101,254

Discussion

Among a large cohort of opioid-dependent people who had ever registered for OST, over a 12-year period, we found that more than one-third were incarcerated at least once, typically more often, and that the costs associated with this are considerable. In any given year, around one in seven was incarcerated, with some variation across calendar years in such levels. The cumulative incidence of incarceration in the cohort is lower than has previously been reported in studies using smaller or convenience samples of opioid users or people who inject drugs,2,11 but sensitivity analyses suggested that our results were not biased downwards by the inclusion of older opioid users in our analysis. Our findings clearly suggest that care should be taken in extrapolating incarceration prevalence from selected samples of opioid users, given the lower levels in this cohort compared to convenience samples.

We found that some groups spent much longer in prison than others: males spent much longer than females in prison, as did Indigenous clients compared to those who were not Indigenous. There was also marked age-related changes in the proportion of follow-up time spent in prison, with the youngest spending the longest times in prison.

There is clearly a much elevated rate of custody among this group compared to the general population. We observed that between one-in-six to one-in-nine cohort participants were imprisoned each year (11–16%), whereas across NSW, there are around 180 people imprisoned per 100,000 population.12 This finding is consistent with the highly elevated rates of court appearances we observed in this cohort compared to general population levels.13

Incarceration of opioid users has important public health implications. Injecting drug use occurs within prisons,14,15 with around half of opioid-dependent injectors reporting that they continue to inject while incarcerated.9,10,15,16 Drug injection in prison typically involves shared use of needles and syringes, and incarceration is a risk factor for hepatitis C infection among people who inject drugs.7,17 Furthermore, the weeks immediately following release from prison are a time of elevated risk of fatal opioid overdose.8 Most of the cohort had multiple releases from prison and thus were exposed to this period of elevated mortality risk multiple times.

Of particular concern was the high prevalence of incarceration among Aboriginal and Torres Strait Islander peoples. Indigenous cohort members were also incarcerated for a substantially greater proportion of follow-up time than non-Indigenous cohort members. Indigenous Australians continue to be over-represented in the criminal justice system,10 and we have also documented this among opioid-dependent individuals.13,18 Drug use, violence, crime and incarceration have all been identified to have a negative impact on individuals and communities as a whole.10 In this analysis, we were unable to determine if incarceration was related to offence seriousness, but our results may reflect that Indigenous people are more likely than non-Indigenous people to be denied bail if charged with an offence, and more likely to be incarcerated if convicted of an offence.19 In analyses of the criminal charges faced by this cohort, we found that there were no major qualitative differences in the nature of charges laid against Indigenous offenders (although there were slightly lower proportions ever charged with an illicit drug offence compared to non-Indigenous offenders, and slightly higher proportions of Indigenous offenders charged with violent offences). Rather, there was a quantitative difference, with higher level of all offences laid against Indigenous offenders in general.18

Turning to economic consequences, the costs of incarcerating opioid users are considerable. This outlay does produce a reduction in property and violent crime,20 but it is far from clear that prison is the most cost-effective response to crime, or whether this is the desired community response to dealing with those with substance use problems. Aos and colleagues21 have shown that a range of alternatives to prison are more cost-effective in reducing crime rates than imprisonment, including treatment for substance use disorders. OST, which is associated with reduced heroin use and incarceration risk,22–24 is significantly less costly than incarceration,25 and has other benefits including reduced offending,26–28 HIV transmission29 and mortality.30 OST can be provided in the prison setting,31 and retention in OST following release from prison is associated with significantly reduced risk of re-incarceration.32

Limitations

The administrative datasets used in this study were linked probabilistically; hence the quality of the linkage was dependent on the quality of the identifiers recorded in each database. Identifiers in the PHDAS are considered reliable, as proof of identity must be presented by the patient prior to prescription of methadone or buprenorphine. A previous linkage between the PHDAS and incarceration data had good sensitivity and high specificity.33

The cohort was based on treatment registrations, and opioid users who seek treatment may differ from those who do not in their risk of incarceration. For example, among those dependent opioid users who never register for OST, and among those who only ever use other pharmaceutical opioids such as morphine or oxycodone, the incidence of incarceration may be lower. There are no local data to permit us to test that possibility. We are, however, confident that the cohort is probably broadly representative of opioid-dependent people in NSW; for example, sentinel surveillance of people who inject drugs in NSW (98% of whom have a history of opioid use) shows that 80% have a lifetime history of OST.11

Conclusions

Although high rates of incarceration have previously been reported among opioid-dependent people, this is the first time incarceration of this group has been examined using the current study design, whereby all people in a given state are followed over time, with all linked prison episodes over more than a decade. Our findings suggest that a substantial minority of opioid-dependent people are incarcerated, usually multiple times, at significant financial cost. A significant proportion of follow-up time was spent by this cohort in prison, with men and Indigenous people in the cohort spending much longer in prison. Treatment for opioid dependence, both inside and outside prisons, may help reduce incarceration of opioid-dependent people, and reduce incarceration and offending costs to the government and to the community.

Acknowledgements

We wish to acknowledge all data custodians for providing access to the datasets used in this study: the NSW Ministry of Health (PHDAS dataset); the NSW Bureau of Crime Statistics and Research (ROD dataset); and the Australian Institute of Health and Welfare (NDI dataset). We are grateful for the expert advice received from Pia Salmelainen (NSW Health) regarding the PHDAS dataset and Jacqui Fitzgerald (NSW Bureau of Crime Statistics and Research) regarding the ROD dataset. Funding for the work undertaken for this manuscript was provided by the National Health and Medical Research Council (NHMRC). This project was also supported by a grant from the Australian Institute of Criminology (AIC) through the Criminology Research Grants Program. The views expressed are the responsibility of the author and are not necessarily those of the AIC. Louisa Degenhardt, Sarah Larney and Richard P. Mattick are supported by NHMRC Research Fellowships. 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. We also wish to thank the members of our Indigenous Reference Group: Michael Doyle, Anton Clifford, Megan Williams and Luke Bell.

Ancillary