Offending, custody and opioid substitution therapy treatment utilisation among opioid-dependent people in contact with the criminal justice system: comparison of Indigenous and non-Indigenous Australians

Background Although Indigenous Australians are over-represented among heroin users, there has been no study examining offending, time in custody, and opioid substitution therapy (OST) treatment utilisation among Indigenous opioid-dependent (including heroin) people at the population level, nor comparing these to non-Indigenous opioid-dependent people. The aims of this study were to compare the nature and types of charges, time in custody and OST treatment utilisation between opioid-dependent Indigenous and non-Indigenous Australians in contact with the criminal justice system. Methods This was a population-based, retrospective data linkage study using records of OST entrants in New South Wales, Australia (1985–2010), court appearances (1993–2011) and custody episodes (2000–2012). Charge rates per 100 person-years were compared between Indigenous and non-Indigenous Australians by sex, age and calendar year. Statistical comparisons were made for variables describing the cumulative time and percentage of follow-up time spent in custody, as well as characteristics of OST initiation and overall OST treatment utilisation. Results Of the 34,962 people in the cohort, 6,830 (19.5%) were Indigenous and 28,132 (80.5%) non-Indigenous. Among the 6,830 Indigenous people, 4,615 (67.6%) were male and 2,215 (32.4%) female. The median number of charges per person against Indigenous people (25, IQR 31) was significantly greater than non-Indigenous people (9, IQR 16) (p < 0.001). Overall, Indigenous people were charged with 33.2% of the total number of charges against the cohort and 44.0% of all violent offences. The median percentage of follow-up time that Indigenous males and females spent in custody was twice that of non-Indigenous males (21.7% vs. 10.1%, p < 0.001) and females (6.0% vs. 2.9%, p < 0.001). The percentage of Indigenous people who first commenced OST in prison (30.2%) was three times that of non-Indigenous people (11.2%) (p < 0.001). Indigenous males spent less time in OST compared to non-Indigenous males (median percentage of follow-up time in treatment: 40.5% vs. 43.1%, p < 0.001). Conclusions Compared to non-Indigenous opioid-dependent people, Indigenous opioid-dependent people in contact with the criminal justice system are charged with a greater number of offences, spend longer in custody and commonly initiate OST in prison. Hence, contact with the criminal justice system provides an important opportunity to engage Indigenous people in OST. Electronic supplementary material The online version of this article (doi:10.1186/1471-2458-14-920) contains supplementary material, which is available to authorized users.

Results: Of the 34,962 people in the cohort, 6,830 (19.5%) were Indigenous and 28,132 (80.5%) non-Indigenous. Among the 6,830 Indigenous people, 4,615 (67.6%) were male and 2,215 (32.4%) female. The median number of charges per person against Indigenous people (25,IQR 31) was significantly greater than non-Indigenous people (9, IQR 16) (p < 0.001). Overall, Indigenous people were charged with 33.2% of the total number of charges against the cohort and 44.0% of all violent offences. The median percentage of follow-up time that Indigenous males and females spent in custody was twice that of non-Indigenous males (21.7% vs. 10.1%, p < 0.001) and females (6.0% vs. 2.9%, p < 0.001). The percentage of Indigenous people who first commenced OST in prison (30.2%) was three times that of non-Indigenous people (11.2%) (p < 0.001). Indigenous males spent less time in OST compared to non-Indigenous males (median percentage of follow-up time in treatment: 40.5% vs. 43.1%, p < 0.001).
(Continued on next page) (Continued from previous page) Conclusions: Compared to non-Indigenous opioid-dependent people, Indigenous opioid-dependent people in contact with the criminal justice system are charged with a greater number of offences, spend longer in custody and commonly initiate OST in prison. Hence, contact with the criminal justice system provides an important opportunity to engage Indigenous people in OST.
Keywords: Indigenous population, Aboriginal and Torres Strait Islander people, Opioid-related disorders, Opioid substitution treatment, Crime, Prisons, Data linkage Background Substance use is a major social and health challenge for Indigenous peoples worldwide, and contributes to 20% of the disease burden (in disability adjusted life years) experienced by Aboriginal and Torres Strait Islander (Indigenous) Australians [1]. Illicit drug use accounts for 3% of the total disease burden, and is responsible for 4% of the disparity in health between Indigenous and non-Indigenous Australians [1]. Indigenous people are over-represented among samples of injecting drug users in Australia -despite comprising about 3% of the population [2], between 11% and 12% of respondents in a national survey of needle and syringe program attendees identified as Indigenous [3]. Similarly, studies in Australia and Canada have shown that Indigenous injecting drug users are over-represented in experiencing the harms associated with injecting drug use, including higher incidences of HIV and hepatitis C [4][5][6].
Indigenous people worldwide are known to have contact with the criminal justice system at disproportionately high levels [7,8]. About one-quarter of the prisoner population in Australia are Indigenous [9]. The most common charges for Australian Indigenous prisoners are acts intended to cause injury (34%), unlawful entry with intent (16%), and offences against justice procedures (11%) [9]. In addition to being more likely to be arrested, charged and imprisoned, the frequency of criminal justice contact is higher among Indigenous people [10], and occurs at an earlier age than non-Indigenous people [11]. About three-quarters (77%) of Indigenous people in Australian prisons have previously been imprisoned, compared with 51% of non-Indigenous people [9].
The factors contributing to over-representation of Indigenous people in the criminal justice system are complex and often inter-related [10]. Indigenous people experienced major disruption to every aspect of health, social, economic, cultural and spiritual systems following colonisation, and inter-generational trauma is ongoing. The Royal Commission into Aboriginal Deaths in Custody in the early 1990s proposed that the over-representation of Indigenous people in prison was due to the combined effect of bias in the criminal justice system and Indigenous economic and social disadvantage [12]. The Royal Commission recommended that Indigenous economic, social and cultural disadvantage, as well as substance abuse, be addressed [12]. Currently, there is little evidence to support the idea that discrimination and racial bias in policing and court decisions explain the over-representation of Indigenous people in prison [10,13]. After adjusting for sentence-related factors such as current and past offending, Indigenous people are no more likely to be issued longer sentences than non-Indigenous people in New South Wales (NSW) courts [14].
Alcohol abuse and illicit drug use are strongly associated with the likelihood and number of arrests Indigenous people have experienced in the last five years [15]. Opioid substitution therapy (OSTmethadone or buprenorphine), the preferred treatment for opioid dependence, has shown to be beneficial in reducing heroin use [16,17], other drug use, crime, HIV infection, exposure to viral hepatitis, and mortality [18]. However, maintaining retention in treatment is necessary in order to maximise OST treatment outcomes [19]. Higher legal severity (corresponding with more complex and severe criminal justice histories) have been associated with shorter retention in methadone treatment [20]. However, there are limited data comparing outcomes related to OST use between Indigenous and non-Indigenous people. Ethnic minority groups have been found to experience significant delays in admission to methadone programs and lower retention once in treatment [21,22]. Similar findings have been observed among Aboriginal people in studies of people who inject drugs in Canada [23]. In contrast, an evaluation of OST provision in an urban Australian Aboriginal Health Service found that rates of treatment retention and heroin use reduction were equivalent to those observed in mainstream treatment programs [24]. There is therefore a clear need for further research to resolve these differences in study findings, given the potential benefits of treatment in this population.
To date, there has been no study examining offending, time in custody, and OST treatment utilisation among Indigenous opioid-dependent people at the population level, nor comparing these to non-Indigenous opioiddependent people. Using a population-level cohort of opioid-dependent people in contact with the criminal justice system in NSW, Australia, the aims of this study (relating to three main themes), were to:

Study cohort
The population-level cohort was defined on the basis that individuals had a recorded history of opioid dependence (as evidenced by the receipt of OST), at least one criminal charge during the study period, and valid information regarding their Indigenous identity (n = 34,962).

Data sources
Three administrative datasets were used to define the cohort in this study and to compare patterns of offending, time in custody and opioid-substitution therapy (OST) treatment utilisation between Indigenous and non-Indigenous Australians.

The Pharmaceutical Drugs of Addiction System (PHDAS) dataset
The PHDAS is a comprehensive record of all people in NSW to whom pharmaceutical drugs of addiction were dispensed by authorised clinicians through the NSW Opioid Treatment Program since 1985. The PHDAS records each client's full name, date of birth, sex, OST program entry and exit dates, the OST medicine received (buprenorphine or methadone), the approved prescriber, the treatment setting (community or prison) and the reason for program exit [25]. A new treatment program is recorded in the PHDAS when an application to prescribe OST is approved by the NSW Ministry of Health, or if a client changes their prescriber, OST medicine, or point of administration. A person can therefore have multiple treatment programs during a period of continuous dosing. Consistent with earlier studies, a continuous OST treatment program was defined where there was less than seven days between program exit dates and subsequent program entry dates [26,27].

Definitions
Receipt of OST was used as a marker of opioid dependence, which is a clinical criterion for receiving treatment with OST in NSW. We assumed ongoing opioid dependence from first entry to OST to the end of follow-up, as it is known that opioid dependence is a chronic relapsing disorder with low remission rates [29,30]. The term Indigenous in this paper refers to individuals who identify as Australian Aboriginal and/or Torres Strait Islander people. An individual was considered Indigenous if they were ever-identified as an Aboriginal and/or Torres Strait Islander person in either the ROD or PHDAS datasets (determined by self-report). Therefore, individuals with conflicting records (who were identified as Indigenous in one dataset and non-Indigenous in the other) were also considered as Indigenous. Records for people with an unknown/missing Indigenous identity were excluded from the cohort (n = 13,107).
Offending was evaluated using data relating to charges (recorded crime) and were coded according to the Australian and New Zealand Standard Offence Classification (ANZSOC) system which has 16 major categories of offences [31]. This study used definitions consistent with the BOCSAR's standard crime and statistic reporting [32]. Violent offences included murder, assault, robbery, sexual assault and indecent assault/act of indecency/other sexual offences. Property offences included break and enter, motor vehicle theft, theft and fraud.

Statistical analyses
Through the linkage of the datasets described, this study provided a unique opportunity to conduct longitudinal population-level analyses. All analyses were conducted using SAS Enterprise Guide 5.1 (SAS Institute Inc., Cary, NC, USA). Descriptive statistics were used to summarise charges, time in custody and OST treatment utilisation. Comparisons between Indigenous and non-Indigenous Australians in the cohort were made using chi-square tests of association for categorical variables and Wilcoxon-Mann-Whitney tests for continuous variables. Due to the different date ranges available for the each of the datasets, we defined distinct follow-up periods to determine the total observation period for each analysis, taking into account the date of death (if recorded).

Offending
The follow-up time for the analyses of charges commenced on 1 December 1993, or whenever the individual turned 10 years of age (whichever was later); 10 years is the age of criminal responsibility in NSW. Follow-up ceased on 31 December 2011, or when death occurred (whichever was earlier). In addition to comparisons of offence frequencies, rates of charges per 100 person years (PY) were also calculated for Indigenous and non-Indigenous males and females, by age group, and calendar year to determine if rates varied across age groups and over time.

Time in custody
The follow-up time for the analyses of custody episodes commenced on 1 January 2000, or whenever the participant turned 10 years of age (whichever was later). The follow-up time ceased on 31 March 2012, or when death occurred (whichever was earlier). Custody episodes where the individual was received and released on the same day were excluded. In calculating the duration of custody episodes, both complete and incomplete custody episodes were included. Complete custody episodes were those completely contained within the follow-up period (i.e. reception date of or after 1 January 2000 and release date of or before 31 March 2012). Incomplete custody episodes were those which started prior to January 1 2000 and/or had not ended by 31 March 2012. In these cases, only days within the follow-up period were counted. The percentage of total follow-up time each individual spent in custody was compared using the cumulative time spent in custody (including incomplete episodes) and the total length of follow-up (start of custody to 31 March 2012 or death).

OST treatment utilisation
The follow-up time for analyses relating to OST treatment utilisation commenced on the day of OST initiation (with records from 1 January 1985) and ended at death or the date of data extraction (18 May 2012), whichever was earlier. Individuals who had commenced treatment and had no treatment end date listed, were deemed to still be receiving treatment at the date of extraction. Treatment retention was evaluated at three, six, nine and 12 months after first commencing OST and compared between Indigenous and non-Indigenous offenders. The percentage of total follow-up time each individual spent in treatment was compared using the cumulative time spent in treatment and the total length of follow-up (start of treatment to 18 May 2012 or death).

Temporal relationship between age of first offence and first commencing OST treatment
In order to examine the temporal relationship between age of first charge and first commencing OST treatment, we identified a sub-set of individuals from the total study population who were below the age of criminal responsibility in NSW (10 years) at the beginning of the charges dataset (1 December 1993) (n = 2,815). We therefore had their complete criminal charge histories and were able to analyse these in relation to their first OST commencement.

Ethical approval
Ethical approval to conduct this study was obtained from the ethics committees of the NSW Aboriginal

Offending
A total of 210,593 charges were laid against Indigenous offenders, and 424,203 charges were laid against non-Indigenous offenders ( Table 1). The median number of charges for Indigenous offenders (25, IQR 31) was almost three times that of non-Indigenous offenders (9, IQR 16) (p < 0.001), indicating that repeat offending was common. The percentage of charges which were proven (i.e. those with a guilty verdict) was similar between Indigenous and non-Indigenous males (83.5% vs. 85.6%) and females (87.3 vs. 88.4%) (Additional file 1). Charges for theft and related offences were the most prevalent, representing 23.9% of all charges by Indigenous offenders and 24.9% of all charges by non-Indigenous offenders. Property offences represented 25.0% of all charges for both Indigenous and non-Indigenous offenders. However, the percentage of charges for a violent offence was greater among Indigenous offenders (14.9%) than non-Indigenous offenders (9.4%). Between 68.7% (males) and 73.5% (females) of charges for violent offences by non-Indigenous people were proven. Similarly, between 69.6% (males) and 76.3% (females) of charges for violent offences by Indigenous people were proven.
Although Indigenous offenders comprised 19.5% of the cohort, they contributed to 33.2% of the total charges laid against the cohort; ranging between 24.0% (fraud and illicit drug offences) and 45.8% (robbery and related offences) of specific charge types ( Figure 1). Considering all violent offences, Indigenous offenders contributed to almost half (44.0%) of all violent offence charges laid against the cohort.
The distribution and outcomes of charges laid against Indigenous and non-Indigenous males and females are shown in Table 2. Among male offenders, the largest difference in the percentage of charges laid against Indigenous and non-Indigenous people were for offences relating to property damage and environmental pollution such as noise, air or water pollution (61.1% vs. 29.0%, p < 0.001), followed by offences against justice procedures (80.3% vs. 49.9%, p < 0.001). Among female offenders, the largest difference in the percentage of charges laid against Indigenous and non-Indigenous people were for offences relating to acts intended to cause injury (62.4% vs. 29.3%, p < 0.001), followed by offences against justice procedures (69.8% vs. 37.8%, p < 0.001). Figures 2 and 3 show that across age categories, charge rates (number of charges/100 person-years) were approximately two times greater among female and male Indigenous offenders than non-Indigenous female and male offenders, and were higher among younger The charge rates (number of charges/100 person-years) between non-Indigenous and Indigenous males and females across the 1993-2011 calendar years are shown in Additional files 2 and 3. For each group, charge rates were relatively consistent across all years, with the exception of a clear peak in overall charge rates in 2001, corresponding with the time that a heroin shortage was observed across Australia [33,34].

Time in custody
A total of 17,967 individuals spent at least one full day in custody between January 2000 and March 2012, with Indigenous people representing 5,303 (29.5%) of all those who were incarcerated among the cohort (Table 3). Overall, Indigenous people spent twice the median percentage of follow-up time in custody compared with non-Indigenous people (21.7% vs. 10.1%, p < 0.001 for males, and 6.0% vs. 2.9%, p < 0.001 for females). This was both the result of Indigenous people having more custody episodes (median of 5 episodes for males and 4 for females), and episodes were of a longer duration (median 75 days for males and 30 days for females). In comparison, the median number of custody episodes for non-Indigenous people was 3 episodes for males and 2 episodes for females; the median duration of custody episodes was 68 days for males and 22 days for females. Table 4 compares OST treatment utilisation during the first treatment episode and in total over the follow-up period, between Indigenous and non-Indigenous males and females. More Indigenous offenders first commenced OST while in custody than non-Indigenous offenders (30.2% vs. 11.2%, p < 0.001), which was consistent among both males (37.7% vs. 13.8%, p < 0.001) and females (14.8% vs. 4.6%, p < 0.001).

OST treatment utilisation
Examining treatment retention during individuals' first OST treatment episode, the percentage of people in treatment at three, six, nine and 12 months was lower among Indigenous offenders (decreasing from 53.9% at three months to 24.7% at 12 months) than non-Indigenous offenders (decreasing from 58.6% at three months to 31.3% at 12 months). Furthermore, the lower treatment retention observed among Indigenous offenders was consistent for both males (decreasing from 53.4% at three months to 24.0% at 12 months) and females (decreasing from 54.9% at three months to 26.4% at 12 months).     Although there was no difference in the median number of OST treatment episodes between Indigenous (2, IQR 3) and non-Indigenous (2, IQR 3) males (p = 0.348), the median number of treatment episodes was greater for Indigenous females (3, IQR 4) than non-Indigenous females (2, IQR 3) (p < 0.001). Indigenous males spent less time in treatment over the follow-up period compared to non-Indigenous males (40.5% vs. 43.1%, p = 0.002). There was no difference in time in treatment among Indigenous females compared to non-Indigenous females (57.9% vs. 55.8%, p = 0.885).
The temporal relationship between age of first offence and first commencing OST treatment among the sub-cohort of 2,815 people with full offending data available is examined in Additional file 4. Among males, the median age of first OST entry among Indigenous offenders was 6.6 years after their first charge, compared to 3.9 years among non-Indigenous offenders, a difference of 2.7 years. Similarly, among females, the median age of first OST entry among Indigenous offenders was 4.2 years after their first offence, compared to 1.5 years among non-Indigenous offenders.

Discussion
The results from this study provide the first populationlevel comparison of offending, time in custody, and OST treatment utilisation among opioid-dependent Indigenous and non-Indigenous Australians in contact with the criminal justice system.

Offending
Despite comprising less than one fifth of the cohort, Indigenous Australians accounted for one third of all charges, indicating that disproportionately more charges were laid against Indigenous people in the cohort, and that repeat offending was common; a finding consistent with previous studies [10,35,36]. Theft, traffic offences and illicit drug offences were the three most common offence types among non-Indigenous males and females. Conversely, among Indigenous males and females, theft, acts intended to cause injury, and offences against justice procedures (e.g. failing to appear before court) were most common. Based on 2012 estimates, acts intended to cause injury, unlawful entry with intent, and offences against justice procedures accounted for almost 60% of offences recorded among Indigenous Australians [37]. Therefore, compared to the broader population of Indigenous offenders in Australia, theft was more common among our cohort. Acquisitive crime is known to be higher among people who use drugs as it has the potential to generate income to support their drug use [38], a key factor likely to be responsible for the higher rate of theft and related offences observed in our cohort. In addition, a history of criminal arrest may be a barrier to gaining employment, resulting in further criminal activity in order to generate income. It has previously been estimated that about 15% of the difference in employment-population rates between Indigenous and non-Indigenous Australians is due to the difference in arrest rates [39]. Indigenous Australians historically experienced higher unemployment rates, lower income and inter-generational poverty than other Australians, and fare worse across other related social determinants of health [40]. Also relevant to our cohort, was the peak in the rate of charges that occurred in 2001; a direct consequence of the heroin shortage that occurred in Australia (most notably in NSW) during that time [34]. The shortage led to an increase in the price of heroin, which subsequently resulted in an increase in the rate of acquisitive crime [33].
Indigenous people contributed to almost half (44%) of all violent-related offences. The higher rate of violent offending among Indigenous people is well documented [41]. In addition, compared to non-Indigenous violent   offenders, Indigenous violent offenders are more likely to be re-incarcerated for a violent offence, and within shorter periods of time [41]. Multiple inter-related factors embedded in the historical experiences of Indigenous people are likely to contribute to their high rates of violent offending, including for example, loss of land and culture, trans-generational trauma, grief and loss, and social exclusion [42]. Although alcohol use is often a major factor implicated in violent crimes [41,43], the relationship between illicit drug use and violent crime is less clear. In national surveys of police detainees and prisoners, Indigenous offenders were less likely than non-Indigenous offenders to self-report heroin use in the 30 days prior to being detained by police, and also in the six months prior to imprisonment [11]. However, in a survey of drug use and crime behaviour among male offenders incarcerated in prison, drug markets were identified as being associated with high levels of violence [44], with the highest being among those whose preferred drug of choice was heroin -29% of respondents reported using force or threats of violence and 17% reported using weapons to obtain heroin. Hence, despite the various factors influencing Indigenous people to commit violent crimes, the role of illicit drug use cannot be excluded.
Previous studies have identified that Indigenous male offenders tend to have earlier and more serious contact with the criminal justice system [11]. In our cohort, offending commenced about three years earlier among Indigenous people. Indigenous females in particular, had charge rates that peaked earlier and for longer (15)(16)(17)(18)(19) years to 20-24 years) compared to non-Indigenous females (20-24 years). Conversely, charge rates peaked between 15-19 years for both Indigenous and non-Indigenous males. To date, offending patterns among Indigenous men have been more widely studied than that of Indigenous women [45,46]. Current evidence suggests that discrimination among Indigenous women is compounded by being both female and Indigenous, and that few criminal justice system services and interventions are targeted to their needs [46]. Most services are either designed for Indigenous men, or for women in general, and are not culturally specific [41,46]. Consequently, our results highlight that there is a great need to further understand and address Indigenous females' over-representation in the criminal justice system.
There was also a clear reduction in the rate of offending with increasing age, particularly among Indigenous people. The low offending rates observed beyond 65 years for females and 69 years for males is likely to be related to poorer health, and multiple co-morbidities experienced by older Indigenous people, and life expectancy being only 72.9 for females and 67.2 for males -on average 13 years less than non-Indigenous Australians [47].

Time in custody
Indigenous people represented 30% of all of people who were incarcerated among our cohort, whereas the point-prevalent estimate of the percentage of Indigenous people in prison in NSW in 2013 is 23% [9]. Given that Indigenous people comprise 2.9% of the general population in NSW [2], our findings suggest that much of the over-representation of Indigenous people in custody may be due to greater contact among people who are opioid dependent. Previously we have shown that there are marked differences in the cumulative time spent in custody among Indigenous and non-Indigenous opioid-dependent people [48]. In this study, Indigenous people spent twice the median percentage of follow-up time in custody, a consequence of these individuals having more custody episodes, as well as episodes being of a longer duration. This was consistent for both males and females. Over the twelve year period between 2000 and 2011, there was a 62% increase in the age-standardised imprisonment rate among Indigenous people in Australia [25]. In comparison, a marginal increase of 5% was observed among non-Indigenous people, highlighting the extent of disadvantage experienced by Indigenous people in contact with the criminal justice system [25]. Periods of imprisonment carry many health risks for people who inject drugs due to needle sharing and transmission of blood borne viruses [49,50], potentiating risks which are already greater among Indigenous people [5]. Therefore, encouraging the use of OST to reduce the health risks associated with injecting drug use is especially important among Indigenous people who are imprisoned.

OST treatment utilisation
Recent data show that Indigenous people are overrepresented in OST in NSW [25]. We found that Indigenous people more frequently commenced OST in custody, which suggests under-treatment in the community. Although not directly evaluated in our study, there is evidence that OST treatment is associated with reduced offending [51][52][53]. Given that Indigenous people in our cohort also spent significantly more time in custody, these findings highlight the importance of making OST accessible through prisons. However, due to different OST policies and programs across countries and jurisdictions, access to OST in prisons vary widely [54]. For example, in Australia, although all eight States and Territories offer maintenance OST in prison, only five of these (NSW included) allow first-time initiation of OST in prison [54]. Despite NSW having one of the largest in-prison OST programs in Australia [55], culturally relevant and accessible programs for Indigenous people in the community are still needed.
Indigenous males were poorly retained in their first OST treatment episode. Given that Indigenous males more frequently commenced OST treatment in prison, poor treatment continuity at the time of release is likely to have resulted in shorter treatment episodes. This highlights the need for the development and implementation of evidence-based and culturally tailored interventions to support the transition of Indigenous males from prison to the community. This is important because maintaining retention in OST is necessary to achieve optimal OST treatment outcomes [19]. About a quarter of Indigenous people aged 15 years and over living in nonremote areas report difficulty accessing health services, a rate ten times that of the general Australian population [56]. Although efforts have been made to close the gap in health and disease burden between Indigenous and non-Indigenous Australians, Indigenous men still have the poorest health outcomes [47]. It is known that Indigenous people are more likely to access services in prison [57], and prisons provide an important opportunity for the provision of health care and screening to Indigenous people that would otherwise have difficulty accessing health services [58]. There remains a need to further examine issues around gender and access to OST among Indigenous people in order to maximise treatment outcomes.

Strengths and limitations
The cohort in this study was defined on the basis that individuals had a recorded history of opioid dependence, recorded Indigenous status (Indigenous/non-Indigenous), and at least one criminal charge in NSW during the study period. Through the use of state-wide administrative datasets, this study presented a unique opportunity to evaluate longitudinal population-level data. Given that NSW has the largest proportion of Indigenous residents in Australia, and also the largest proportion of clients in OST, a major strength of our study is that the cohort is representative of a large number of Indigenous and non-Indigenous opioid-dependent people in the Australian population. However, as the provision of OST and criminal justice services vary between States and Territories, the findings of this study may not be the same across jurisdictions. In addition, although the National Deaths Index is able to capture deaths occurring in any State or Territory in Australia, the other datasets used in this study are NSW specific. Therefore, we are not able to capture events that occurred outside NSW or follow those individuals who might have migrated to other States or outside Australia. Furthermore, although Australian OST services aim to ensure access to disadvantaged populations including Indigenous populations [59], rates of participation in the community are not consistent across Australia [60]. Hence, the results may not necessarily be generalisable to other Indigenous opioid-dependent populations.
Information regarding the identity of Indigenous Australians is not always reliable in health datasets; missing data is common, and Aboriginal and Torres Strait Islander people are often under-counted [61]. Furthermore, estimates of disparities in health between Indigenous and non-Indigenous people can be affected by the way in which Indigenous identity is recorded [62]. The everidentified strategy used in this study meant that Aboriginal and Torres Strait Islander people were included despite changes in administrative data recording or personal preference to identify as Indigenous or not over time [63]. Although recent studies have suggested that the application of other approaches to identify Aboriginal or Torres Strait Islander people in datasets may have improved accuracy [62], we were restricted to the approach used by the data custodians in constructing the datasets. However, we endeavoured to improve the accuracy of correctly identifying Indigenous people in the cohort by triangulating information from two data sources.
Rates of offending based on data from administrative datasets provide estimates of the number of charges which are made, based on offences which are reported to correctional authorities. Consequently, offences which are unreported are not able to be enumerated, which may lead to an under-estimation of actual rates of offending. Hence, the rates of offending reported in our study are also likely to have been under-estimated. As the main purpose of this study was to undertake a detailed comparison of offending, time in custody, and OST treatment utilisation among Indigenous and non-Indigenous people in an opioid-dependent population, we did not examine whether time in OST directly affects either offending or time in custody, but is an important area for future research. Also, given that we were unable to assess the temporal relationship between time of offence and initiation of opioid use, this presents another area for future research.

Conclusions
There are clear differences in the nature and levels of offending, as well as time spent in custody among Indigenous and non-Indigenous people with a history of opioid dependence and contact with the criminal justice system. Indigenous populations continue to experience social exclusion and ongoing inequity across determinants of health and our findings highlight that Indigenous people with opioid use disorders are further disadvantaged. Although the prison setting appears to be an important access point for OST among Indigenous people, the under-treatment of Indigenous people in the community is also apparent, and there appears to be a lack of continuity between the prison and community OST systems. The underlying reasons for these differences therefore need to be addressed, with a focus on the development, implementation and rigorous evaluation of more targeted and evidence-based culturally tailored interventions.