• Open Access

Increased mortality among Indigenous persons during and after release from prison in New South Wales


Correspondence to: Azar Kariminia, The Kirby Institute for Infection and Immunity in Society, Faculty of Medicine, University of New South Wales, Cliffbrook House, Building CC4, 45 Beach St, Coogee, NSW 2034; e-mail: akariminia@kirby.unsw.edu.au


Objective: To estimate the overall and cause specific mortality of Aboriginal offenders in New South Wales (NSW), Australia.

Methods: The study cohort consisted of all Aboriginal men and women aged 18 years and older who had experienced full-time imprisonment in NSW between 1 January 1988 and 31 December 2002. Their data were linked probabilistically to the Australian National Death Index to obtain information on death. Standardised mortality ratios were calculated for all causes of death and adjusted for age, sex, and calendar year.

Results: The cohort comprised 7,980 men and 1,373 women with 75,801 person years of observation. During a median follow-up period of 8.3 years, 485 men and 73 women died, giving an overall mortality rate of 733 and 755 deaths per 100,000 person-years. The risk of death in men was 4.8 (95% CI: 4.4–5.3) times and among women 12.6 (95% CI: 10.0–15.8) times that of the NSW residents, with a markedly elevated risk for almost all conditions. The leading cause of death was cardiovascular disease in men (112 deaths, 23%) and mental and behavioural disorders (17 deaths, 23%) in women. The risk of death was greatest following release from prison.

Conclusions and Implications: High mortality rates for cardiovascular disease, a preventable and treatable condition, were seen among Aboriginal offenders. Prison has an important role to play in screening marginalised populations for a range of health conditions. This is particularly true for Indigenous offenders.

The health of Aboriginal Australians is acknowledged to be poor compared with non-Aboriginal Australians, and worse than their Indigenous counterparts in New Zealand, Canada and the United States.1–3 This has been attributed to socio-somatic illness arising from factors such as dispossession, disadvantage, forced separation of Aboriginal families and legal discrimination.4

The overall death rate for Indigenous people is around three times higher than the rate for the rest of the population. At the national level for 2005–07, Aboriginal people were estimated to have a life expectancy at birth of 72.9 years for females and 67.2 for males; 9.7 years less for females and 11.5 years less for males compared to non-Indigenous Australians.3 A recent Commonwealth Government statement has set a target of Closing the Gap [in life expectancy] by addressing a range of health determinants (housing, safer communities, schooling, health, childhood factors, economic participation and governance).4

During the mid-1980s the number of Aboriginal people dying in custody created widespread community concern and resulted in the establishment of a Royal Commission in 1987. It determined that while Aboriginal people do not die in custody at a greater rate than non-Aboriginal prisoners, they come into custody at a rate which is “overwhelmingly higher” than that of the general community.5 The Royal Commission recommended ongoing monitoring of Indigenous and non-Indigenous deaths in prison.

Overall, the prisoner population across Australia has risen, particularly since 1988, and Aboriginal people have not been spared this increase. Currently, there are 9,253 men and 692 women in full-time custody in New South Wales (NSW).6 More than 30% of women and 22% of men are of Indigenous heritage, the highest proportion since the Royal Commission's report in 1991. Only around 2% of the general NSW population are Indigenous.7

Alongside the increasing number of Aboriginal prisoners, a high percentage enter the judicial system with a multitude of health problems, including chronic illnesses, substance abuse and mental illness. Despite the over-representation of Aboriginal people in prisons, and the fiduciary responsibility for officials to use evidence when planning health services for offenders, there is still a lack of comprehensive and high-quality statistics on the pattern of mortality among this population. While the focus of the Royal Commission was on deaths in custody, studies demonstrate the need for attention to go beyond the period of custody and examine the post-release period and the difficulties faced by released prisoners attempting to reintegrate into society. Two Western Australia studies8,9 have shown that Aboriginal people released from prisons are at least four times more likely to die than people in the general population. These studies focused on all-cause mortality and did not report the cause-specific standardised mortality ratio (SMR).

In a previous study of a cohort of offenders in NSW, we observed higher risk of all-cause mortality for Aboriginal offenders than non-Aboriginal groups when compared with the NSW general population.10 We extend this work to describe major causes of death among Aboriginal offenders in our cohort. Describing patterns of death for Aboriginal offenders will enable health service planners to focus their efforts on particular areas to reduce the number of such deaths among the Aboriginal offender population both in prison and the community.


Study population

For this retrospective cohort study, the NSW Department of Corrective Services (DCS) provided a list of all adult men and women, 85,203 in total, who had spent some time in full-time custody in NSW prisons between 1 Jan 1988 and the time of data extraction on the 31 May 2003. The list was extracted from the DCS Offender Integrated Management System (OIMS). The OIMS is a large, computerised information system used to collect, store and retrieve information on NSW offenders while in custody and on conditional release in the community. A unique identification number called the Master Index Number (MIN) is assigned to each person when they first enter prison. This unique number is used in all subsequent imprisonments. For each individual we received information on demographic factors (name, date of birth, sex, country of birth, Indigenous status), and imprisonment history (the start and end date of each incarceration).

The offender cohort was linked to the Australian National Death Index (NDI), a register held by the Australian Institute of Health and Welfare (AIHW) by the Health Registers and Cancer Monitoring Unit of the AIHW using the following common personal identifiers: surname, first name, middle name and date of birth, plus date of last contact with the prison system. These identifiers are not unique and are also subject to recording error. Furthermore, the use of multiple names is common in prisoners, which causes important challenges for data linkage. For these reasons, linkage was undertaken by a probabilistic method using the software called Integrity (formerly known as Automatch). Only highest-level matches (that is, matches on sex, family name, first name, middle name and date of birth accurate within one year) were accepted. Classification codes ICD-9 (deaths before 1997) and ICD-10 (deaths since 1997) for underlying cause of death were extracted from NDI for all the accepted matches. We also established a separate category ‘drug-related death’ by combining those cases where the underlying cause of death was inferred to be directly related to drug use. These included deaths due to mental and behavioural disorders (ICD-9 304, 305.2–305.9; ICD-10 F11-F16, F19, F55). We have previously found that our linkage with the NDI has a sensitivity of 88.4% and specificity of 99.7% for identifying deaths.11 At the time we conducted the linkage, the NDI database contained information on all deaths reported up to 31 December 2002.

Using information on country of birth and Indigenous status, 9,353 of the cohort were identified to be of Aboriginal or Torres Strait Islander descent. Indigenous status is self-reported in this system and is believed to have been recorded correctly for over 90% of prisoners after 1990 (personal communication, Mr Luke Grant, Acting Commissioner, NSW DCS). These 9,353 indigenous people constitute the cohort for the current study.

Statistical analysis

Overall mortality rates (the number of deaths divided by the person-years at risk) and 95% confidence intervals were calculated. The time at risk was defined as the period between the first entry into prison after 1 January 1988 until the date of death or the end date of the study (31 December 2002). Those who were imprisoned during the study period, and who first entered prison prior to 1 January 1988, were considered to be at risk from 1 January 1988. To obtain the overall rates of death during imprisonment and after release, the person-years at risk in prison and after release were also calculated separately. The time at liberty ceased at the earlier of date of death and date of next re-incarceration to control for re-entry into prison. This means that a person with more than one episode of imprisonment contributed person-years, several times, to the prison and after-prison period. This gives us a reliable estimate of mortality, as for each person, days between imprisonment have been aggregated with total days out of prison, and each person is counted only once in the calculation.

For each cause of death, SMRs were estimated as ratios of observed and expected number of deaths. Age- and sex-specific annual mortality rates were calculated for the NSW general population. The expected number of deaths was derived by multiplying the accumulated person-years by the NSW rates across sex, age and calendar year strata. Ninety-five per cent confidence intervals based on the Poisson distribution of observed number of deaths were obtained. SMRs were not calculated if the number of expected deaths was zero. Stata software was used to perform the statistical analyses.12


Cohort characteristics

The study sample consisted of 7,980 (85.3%) men and 1,373 (14.7%) women with a total follow-up time of 66,127 and 9,673 person-years, respectively. Eighteen per cent of the total person-years of follow-up in men and 11% in women were in prison (Table 1). The median age at the study entry was 23.2 years (range 18–79) for men and 24.8 years (18–64) for women. Approximately two-thirds of the study cohort were imprisoned more than once during the study period. The mean (SD) length of incarceration was 1.5 (1.33) years in men and 0.8 (0.19) years in women.

Table 1.  Characteristics of 9,353 Indigenous people imprisoned in NSW, Australia, 1988–2002.
CharacteristicsMen (n=7,980)Women (n=1,373)
Age at entry into prison – n (%)
18–19 yr1,678 (21)187 (13.6)
20–24 yr2,815 (35.3)478 (34.8)
25–34 yr2,319 (21.9)485 (35.3)
35–44 yr889 (11.1)185 (13.5)
≥45279 (3.5)38 (2.8)
Number of imprisonments
12,384 (29.9)548 (39.9)
21,439 (18.0)251 (18.3)
≥34,157 (52.1)574 (41.8)
Total time in prison (years)12,1611,032
Follow-up time in years, median (range)8.6 (1 day to 15 years)7.1 (1 day to 15 years)
Most serious offence
Property2,400 (30.1)520 (37.9)
Drug related212 (2.7)59 (4.3)
Violent2,912 (36.5)419 (30.5)
Sexual376 (4.7)2 (0.1)
Other offence2,080 (26.1)373 (27.2)

Comparative mortality

Over a median follow-up period of 8.3 years (range=1 day to 15 years), 485 men and 73 women in the study cohort died. Of these, 36 men and one woman died in custody. The median age at death was 34.3 (range 18–80) years in men and 31.5 (range 18–69) years in women. Rate of death from all causes combined was 733 deaths per 100,000 person-years in men and was 755 per 100,000 person years in women (Table 2). The mortality rate among Aboriginal men was 4.8 times (95% CI: 4.4–5.3) and among Aboriginal women 12.6 times (95% CI: 10.0–15.8) that among NSW residents of the same age and sex.

Table 2.  Age-specific rates (per 100,000 person-years) and SMRs for all-cause mortality in 9,353 Aboriginal people imprisoned in NSW, Australia, 1988–2002.
Age group (years)Aboriginal MenAboriginal Women
 DeathsRateSMR (95% CI)DeathsRateSMR (95% CI)
18–199409 (213–786)4.0 (2.1–7.7)2848 (212–339)24.7 (6.2–98.7)
20–2456378 (291–492)3.4 (2.6–4.4)11588 (326–106)16.5 (9.1–29.8)
25–2994550 (449–673)4.5 (3.7–5.6)16594 (364–969)16.7 (10.2–27.3)
30–3495704 (576–861)5.3 (4.3–6.5)19893 (569–1,399)18.2 (11.6–28.5)
35–3969793 (626–100)5.4 (4.2–6.8)7521 (248–1,093)7.7 (3.7–16.1)
40–4480155 (125–193)8.3 (6.6–10.3)81,006 (503–2,011)10.0 (5.0–20.1)
45–4941156 (115–212)5.9 (3.4–7.5)61,588 (713–3,525)10.2 (4.6–22.8)
50–5425201 (136–298)5.1 (2.5–3.2)1654 (92–4,639)2.6 (0.4–18.3)
55–598163 (815–326)2.4 (1.2–4.9)12,004 (282–14,229)5.1 (0.7–36.3)
60–645235 (977–564)2.0 (0.8–4.8)00
65–691109 (153–773)0.6 (0.1–4.1)227,337 (6,837–109,303)27.3 (6.8–109.0)
70–741290 (408–206)1.0 (0.1–6.8)
≥751522 (735–370)10.7 (1.5–76.2)
Total485733 (671–802)4.8 (4.4–5.3)73755 (600–949)12.59 (10.0–15.8)

Cause-specific mortality

The leading cause of death in men was cardiovascular disease (112 deaths), representing nearly one quarter of all deaths (Table 3). Women had a lower incidence of death from cardiovascular disease than men (103 vs 169 deaths per 100,000 person years), but had a higher SMR (14.8, 95% CI: 8.0–27.5 vs 8.5, 95% CI: 7.1–10.2). In women, the most common cause of death was from mental and behavioural disorders (17 deaths, 23%). Death from this category had the highest SMR (57.1) in women and the second highest SMR (10.4) in men. All deaths in women and 68 (81%) deaths in men within the mental and behavioural causes of death category were coded as resulting from drug-dependence. Overall, drug overdose, was responsible for 97 (29%) excess deaths among men and 26 (39%) excess deaths among women.

Table 3.  Observed and expected number of deaths and SMRs for the Aboriginal people imprisoned in NSW, Australia, 1988–2002.
Cause of death (ICD-10; ICD-9)Aboriginal Men (n=7,980)Aboriginal Women (n=1,373)
 ObservedExpectedSMR (95% CI)ObservedExpectedSMR (95% CI)
  1. * Includes other disease-related (ICD–9: 210–239, 280–289, 580–677, 680–759, 780–799; ICD–10: D00–D48, D50–D89, L00–N99, R00–R99); other external causes of death (ICD–9: 970–999; ICD–10: Y10–Y34) and 8 deaths reported as unknown.

Mental, behavioural (F00-F99; 290–319)848.110.4 (8.4–12.9)170.357.1 (35.5–91.8)
Cardiovascular (I00-I99; 390–459)11213.28.5 (7.1–10.2)100.714.8 (8.0–27.5)
Cancer (C00-C97; 140–208)1715.61.1 (0.7–1.8)42.02.0 (0.7–5.3)
Digestive system (K00-K93; 520–579)232.68.9 (5.9–13.3)70.241.7 (19.9–87.4)
Respiratory system (J00-J99; 460–519)152.17.2 (4.4–12.0)10.24.6 (0.6–32.6)
Endocrine system (E00-E89; 240–279)73.81.8 (0.9–3.9)20.212.2 (3.0–48.6)
Infectious diseases (A00-B99; 001–139)23.00.7 (0.2–2.6)0
Nervous system (G00-G99; 320–389)92.73.3 (1.7–6.3)0
Accidental (V01-X59; E800-E949)10824.84.4 (3.6–5.3)150.917.7 (10.7–29.4)
Suicide (X60-X84; E950-E959)5719.42.9 (2.3–3.8)40.66.5 (2.4–17.3)
Homicide (X85-Y09; E960-E969)282.312.0 (8.3–17.4)80.253.0 (26.5–106)
Other causes*233.17.5 (5.0–11.3)50.413.6 (5.6–32.6)
Total485100.64.8 (4.4–5.3)735.812.6 (10.0–15.8)
Drug-related11113.78.1 (6.7–9.8)270.643.6 (29.9–63.6)

Homicide had the highest SMR in Aboriginal men (12.0, 95% CI: 8.3–17.4) and the second highest SMR in women (53.0, 95% CI: 26.5–106.0) reinforcing the high levels of violence to which some Aboriginal people are exposed. Excess mortality from diseases of the digestive system was evident for which there were around nine times more deaths in men and 42 times more deaths in women. All seven deaths from digestive system diseases in women, and 17 of these deaths in men, were due to chronic liver conditions and cirrhosis. Deaths from overdoses, suicides, and homicide were more common among those younger than 45 years, whereas deaths for cardiovascular disease and digestive system disease were more common in the older (over 45 years) group (Table 4).

Table 4.  Deaths among the Aboriginal people imprisoned in NSW, Australia, 1988–2002, according to age group at risk.
 Total deathsPerson-yearsCause of death
Age at risk  Cardiovascular diseasesDigestive system diseasesSuicideHomicideDrug overdose
18–19 years
20–24 years
25–34 years
35–44 years
≥45 years

Of the 485 deaths in men, 36 (7%) took place while in prison for a mortality rate of 296 deaths per 100 000 person-years (SMR=2.1, 95% CI: 1.5–3.0). During incarceration, deaths from drug overdose occurred at a rate of 41 (SMR=2.1, 95% CI: 0.9–5.0), suicide at a rate of 82 (SMR=2.8, 95% CI: 1.5–5.3), and cardiovascular disease at a rate of 82 (SMR=5.9, 95% CI: 3.2–11.0) deaths per 100,000 person years. Apart from suicide (post release rate 87 per 100,000 person-years, SMR=3.0), these rates are all considerably less than the mortality rates following release. In women, one death occurred in prison, which was related to cardiovascular disease.

Between 1988 and 1999 the overall SMRs increased from 2.5 to 7.5 among Aboriginal offenders (Figure 1). It then dropped for the next two years before increasing again in 2002. During this period, SMRs for cardiovascular conditions increased steadily. In 1989 the increased risk of death from these conditions was 5.3; the risk increased to 9.2 in 2002 (Figure 2). For drug-related deaths, SMRs had large fluctuations during the entire period.

Figure 1.

Trend in SMR for all-cause mortality among Aboriginal offenders, NSW, 1988–2002. Note: Number of death for Aboriginal offenders in each year has been shown in brackets.

Figure 2.

Trends in the death rate from drug overdose and cardiovascular conditions among Aboriginal offenders, NSW, 1988–2002.


This is the first study to provide a comprehensive assessment of mortality among a cohort of Aboriginal prisoners in Australia. The risk of death in men was 4.8 times greater than the risk among the NSW general population of the same age and sex, and in women it was 12.6 times greater. Diseases of circulatory system accounted for the largest number of deaths in men, whereas mental and behavioural diseases were the most frequent cause of death in women.

The estimates for the study period showed that the risk of death was sharply higher after release than during incarceration and likely due to fewer overdoses, homicides, or accidents during incarceration. It is also possible that those with a physical illness may be at a reduced risk of re-offending and are more likely to die in the community, although we are unaware of any research to confirm this suggestion.

The risk of death from chronic conditions, in particular cardiovascular diseases, is extraordinary high among Aboriginal offenders. These deaths were mainly responsible for the increased mortality observed during the study period in this population. The risk of death from cardiovascular and respiratory conditions may be related to the high prevalence of smoking, increased alcohol consumption and poor diet among Aboriginal people.

While the finding in relation to cardiovascular conditions as the leading cause of death for Indigenous offenders reflects the community situation,13 other leading causes of death highlight the enduring differences between incarcerated and non-incarcerated populations. Some 17% of deaths in men and 23% in women were attributed to mental and behavioural deaths (a euphemism for drug overdose in this population). Similarly, causes of death due to accidents and injury, suicides and homicides underscore the high levels of violence this population group experiences.

Prison provides an important (but underutilised) public health opportunity to screen for chronic diseases, and assess the treatment needs of offenders who are likely to have limited interaction with the health system when in the community. One such programme (‘Tick on Kick on’)14 was established in 2001 by Justice Health NSW to address vascular health among Aboriginal prisoners. This program provides fortnightly screening, risk assessment, referral and, where appropriate, connections to local services for ongoing care following release. A key feature of this initiative is the recruitment and training of Aboriginal vascular health workers to provide awareness-raising education and health promotion on healthy lifestyles and support.

Deaths due to drug overdose contributed substantially to mortality among Aboriginal people in this cohort. The excess in drug-related deaths was much higher for women than men, as was the absolute rate of death. The significant rise in hepatitis C liver-related mortality, particularly among women, reflects the exposure of this population to viral hepatitis through injecting drug use.15,16

The high prevalence of mental illness and substance abuse may have contributed to the high risk of homicide.17–19 A report by the Australian Institute of Criminology20 found that alcohol was involved in about 80% of Indigenous homicides, as compared with 38% of non-Indigenous homicides. Improving alcohol and drug treatment programs during incarceration could reduce rates of violence and might also help to decrease the high mortality rates that are attributed to drug overdose and liver diseases in Aboriginal offenders. Suicide was more common among Aboriginal men, but women had a noticeably higher excess mortality. This could be explained mainly by the rarity of suicide among the general population of women relative to men.21

The SMR for all-cause mortality, found in this study, is higher than what we have previously reported for the entire cohort of 85,203 offenders (SMR = 4.8 vs 3.7 in men and 12.6 vs 7.8 in women). In terms of mortality from specific conditions, a different picture was found: SMR for most disease-related conditions such as cardiovascular or digestive system diseases was higher in Aboriginal offenders than the entire cohort; whereas the SMR for drug overdose and suicide was lower in Aboriginal offenders. This finding has important implications in developing evidence-based policies, planning future health services, and transitional and community based programs for Aboriginal prisoners. The prison health system needs to examine the role of incarceration in improved long term health outcomes for those in contact with the criminal justice system. Specific interventions that can effectively manage many of the long-term chronic conditions include: patient education; a structured delivery of health care services; standard disease management guidelines; the use of telemedicine and electronic medical records; and close links with academic medicine.22 These interventions combined with effective bridging of care between prison and the community may ultimately result in reduced mortality from chronic conditions among prisoners. As recommended by the Royal Commission into Aboriginal Deaths in Custody,5 any intervention for Aboriginal people should involve Aboriginal medical services wherever this is possible.

The two studies undertaken in Western Australia reported broadly similar findings to ours.8,9 Stewart et al.8 studied prisoners aged 20–40 years, released between 1994 and 1999 and found that Aboriginal men were three times and Aboriginal women 10 times more likely to die than the general community. Hobbs and his colleagues9 examined mortality of individuals aged 20–59 released during 1995–2001. The increased risk of death compared with the general community was around eight to nine times higher in Indigenous women and around five to six times higher in Indigenous men. The leading cause of death in women was associated with alcohol and drug dependence and in men it was cardiovascular disease.

In our study, Indigenous status was based on self-reporting and therefore, some misclassification may have occurred. Assuming random misclassification, the estimated SMRs would be biased towards the null; if it is not, the SMR could either overestimate or underestimate the true relative risk. The authors had no alternative source with which to verify the accuracy and completeness of the Indigenous status as the NDI does not report on Indigenous status mainly due to inadequate information. The accuracy of the prisoner data on Aboriginality was previously assessed by Stewart et al.8 in their study on released prisoners in Western Australia. When Stewart and colleagues compared the self-reported Indigenous status from the prisoner's record with the death record, there was an 83% agreement and only 3% disagreement between the two datasets. For the remaining 14% of deaths, Indigenous status was not recorded on the death records. Internal audits of Aboriginal or Torres Strait Islander status conducted within the NSW Department of Corrective Services suggest that it is recorded correctly for over 90% of prisoners (personal communication, Mr Luke Grant, Assistant Commissioner, NSW Department of Corrective Services). However, because the quality of Aboriginal identification may have improved over time, caution should be exercised when interpreting data with regard to year-to-year changes in SMR.

Mortality among prisoners is extremely high, especially following release from prison suggesting that programs aimed at managing the transition from prison to the community need to be improved to prevent excess mortality in the post-release period. This phenomenon likely reflects the difficulties faced by those attempting to reintegrate themselves into society and, for some, the return to drug use.23 Interestingly, in a Western Australian cohort8 of ex-offenders, the relative risk of death among Indigenous ex-prisoners was markedly lower than that for non-Indigenous ex-prisoners, suggesting that different factors may exist for Indigenous and non-Indigenous people returning to the community following incarceration. It is possible that the high levels of Indigenous incarceration have served to reduce the stigma of imprisonment, thereby reducing the alienation and shame felt by those reintegrating into the community. It is also possible that reduced heroin use by Aboriginal prisoners compared with non-Aboriginal prisoners is another factor. National information on the health and welfare of Indigenous prisoners is needed to monitor health services across jurisdiction as well as health indicators pertaining to prisoners.

The National Strategic Framework for Aboriginal and Torres Strait Islander Health specifically mentions decreasing mortality rates across all ages as an aim for reducing health inequalities between Indigenous and non-Indigenous populations. Strengthening service infrastructure in areas such as injury and poisoning, family violence, substance misuse, mental disorder, stress, trauma, and suicide are all clearly relevant to prisoner populations. It is important that efforts to improve health services at the community level do not overlook prisoner populations.

In 1989 the National Aboriginal Health Strategy Working Party defined Aboriginal health to be “Not just the physical well-being of the individual but the social, emotional, and cultural well-being of the whole community. This is a whole-of-life view and it also includes the cyclical concept of life-death-life”.24 The disproportionate incarceration of Indigenous people undoubtedly has a profound impact at both individual and community level and efforts to promote wellbeing need to take the prison experience into account. It is important that Indigenous people in prison have access to Aboriginal Medical Services as well as culturally appropriate services during incarceration and that these also play a role in the transition from prison to the community.

Programs looking beyond the incarceration window period are needed to address this problem to ensure that some good arises from what is essentially a negative event (imprisonment). While the focus of the 339 recommendations of the Royal Commission was on deaths occurring in custody, our findings demonstrate the need for attention to go beyond the period of custody and examine the post-release period. It is important to recognise that those in the community exposed to incarceration have special needs arising from risks to health such as substance use and violence which is reflected in causes of death due to injury, homicide, suicide, and overdose. Programs targeting this group and likely to require an integrated approach involving medical and mental health care, drug treatment programs, education about drug overdose risks, measures to modify risky lifestyle, and social supports.


This study is supported by National Health and Medical Research Council Research Grant No. 222849. We are grateful to the staff from the NSW Department of Corrective Services for assistance with prison data extraction.