Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Risk of death in prisoners after release from jail

Risk of death in prisoners after release from jail Abstract Objective: T compare the r isk of death in o a cohor t of Wester n Australian released prisoners with the risk experienced by the general population of Western Australia. L. M. Stewart School of Population Health, University of Wester n Australia C. J. Henderson Formerly Director of Prison Health Services, Department of Justice of Western Austr alia Methods: A cohort study of prisoners in Wester n Australia whose last date of release ranged from 1 January 1994 to 1 Januar y 1999. Overall mortality and cause of death were determined by data linkage to the Registrar General’s record of deaths. Results: Aboriginal prisoners had a significantly lower sur vival rate after release than non-Aboriginal prisoners (p<0.0001). When compared with their peers in the Wester n Australian community, both Aboriginal and non-Aboriginal prisoners were found to have an increased relative risk of death. Female non-Aboriginal released prisoners aged between 20 and 40 years were 17.8 (95% CI 8.1-27.5) times more likely to die than other female nonAboriginals in Western Australia in the same age range. Male non-Aboriginal prisoners aged 20-40 years were 6.3 (95% CI 5.2-7.4) times more likely to die than their counterpar ts in the WA community. Female Aboriginal released prisoners were 3.4 (95% CI 1.2-5.6) times more likely to die than their peers, while male Aboriginal released pr isoners were 2.9 (95% CI 2.23.5) times more likely to die. In their first six months after release, female non-Aboriginal prisoners aged 20 to 40 years were 69.1 (95% CI 17.9-120.3) times more likely to die than their counterpar ts in the WA community. The main causes of excess death were related to drug and alcohol abuse. Conclusion: All prisoners w ere at greater than expected relative r isk of death after release from prison, with female nonAboriginal prisoners at par ticularly high relative risk. ( Aust N Z J Public Health 2004; 28: 32-6) M. S. T. Hobbs School of Population Health, University of Wester n Australia S. C. Ridout School of Population Health, University of Wester n Australia M. W. Knuiman School of Population Health, University of Wester n Australia risoners are not always isolated from the general community. They begin as members of the community, move into the prison system and then back into the community again. They may cross the prison dividing wall a number of times in their lives. Their chronic health problems and infectious diseases do not remain behind when they leave prison. Prisoners’ health problems are the community’s health problems. The community must be concer ned with the risk of the spread of infectious disease and the cost of the treating chronic ill health. All members of the community (including e x-prisoners) are entitled to equitable health care. Released prisoners are a vulnerable group who are likely to e xperience higher mortality rates than other members of the community. This has been conf irmed in a number of studies. Harding-Pink and Fryc1 found an increased risk of sudden death in released prisoners in Geneva. Joukamaa2 found an approximately four times greater than expected mortality rate in a sample of male released prisoners in Finland. Putkonen et al.3 found a 17 times greater than expected risk of death in a cohort of 132 female homicide offenders, although it is not clear whether all of these prisoners had been released from jail. Lattimore et al.4 examined mortality in two cohorts of young Californian males paroled during the 1980s and found a particularly high rate of death due to homicide. The purpose of the present study is to examine whether there is excess mortality in released prisoners and, if so, whether there are specific causes of death that contrib ute to excess mortality. The cohort under study comprises male, female, Aboriginal and nonAboriginal West Australian released prisoners. The risk of death in these groups was compared with the risk experienced by the general population of Western Australia. Methods The cohor t under study includes all prisoners whose last date of release from prison in Western Australia ranged from 1 Januar y 1994 to 1 Januar y 1999. Information collated from Ministr y of Justice records included the prisoner’ name, s Submitted: Februar y 2003 Revision requested: June 2003 Accepted: October 2003 Correspondence to: Associate Professor Michael Hobbs, School of Population Health, Unit of Clinical Epidemiology, University of Western Austr alia, Crawley, Western Australia 6009. Fax: (08) 9380 1188; e-mail: mikeh@dph.uwa.edu.au AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH VOL. 28 NO. 1 Vulnerable Groups Risk of death in prisoners after release from jail Indigenous status, sex, date of bir th and date of release from prison. These data were linked with the Registrar-General’s record of deaths to determine how many of the released prisoners had died and their date and cause of death. Deaths were followed up from 1 January 1994 to 1 December 1999. Once data linkage was complete, prisoners’ names were deleted from the dataset. Sur vival of Aboriginal and non-Aboriginal released prisoners was compared and log rank statistics were calculated to determine whether the Aboriginal and non-Aboriginal curves were significantl y different from one another. Sur vival curves were produced for male Aboriginals, female Aboriginals, male nonAboriginals and female non-Aboriginals. Where the released prisoner had died, the cause of death was coded according to ICD-9 or ICD-10, depending on which coding system was current at the time of death. Deaths registered after 1 January 1999 were coded according to ICD-10, while deaths registered prior to this date were coded according to ICD-9. ICD-9 and ICD-10 codes provide logical groupings of deaths. Deaths were g rouped into a small number of categories according to their ICD-9 and ICD-10 codes. The risk of death in the prisoner cohort was compared with the risk of death in the WA population using indirect standardisation. Age-specific death rates for the WA population (by gender and Indigenous status) were calculated using data on deaths in WA over the time period of the study (1 January 1994 to 1 December 1999) collated by the W Registrar General’ Office, and the esA s timated resident population of Western Australia aged 15 years and older as at 30 June 1996, as determined by the 1996 Census.5 Person-years of exposure from release to death or the end of follow-up (1 December 1999) were calculated for the prisoner cohor t. Expected numbers of deaths for each group (female Aboriginal, female non-Aboriginal, male Aboriginal and male non-Aboriginal) and each five-year age band were calculated by multiplying the age-specif ic death rates for this group in the WA population by the person-time of exposure. The comparison of observed to expected deaths yielded the standardised mortality ratio (SMR). Figure 1: Survival after release in male and f emale Aboriginal and non-Aboriginal prisoners. Within each group (female Aboriginal, female non-Aboriginal, male Aboriginal and male non-Aboriginal), the variation in age specif ic observed to expected deaths was tested using Poisson regression to deter mine if there was signif icant variation from age g roup to age group, since if there was then the overall SMR would not accurately summarise the stratum specific rates. Agespecif ic observed/e xpected ratios did not vary signif icantly between the ages of 20-40 years. SMRs presented in this report refer only to this age range. 95% conf idence intervals for SMRs were calculated using this formula: 95% CI=SMR±1.96√(SMR/exp) where exp = Σ age specif ic expected deaths.6 Ethics approval for this study was obtained from the Human Research Ethics Committee of the University of Wester n Australia. Results The cohort The cohort under study included all prisoners whose last date of release from jail fell between 1 January 1994 and 1 January 1999. Deaths were followed up from 1 January 1994 to 1 December 1999. There were 9,381 prisoners in total; 326 died after release. Indigenous status, sex and age Five hundred and fifty-four (6% of total) prisoners were female Aboriginals, 628 (7%) were female non-Aboriginals, 1,927 (20%) were male Aboriginals and 6,272 (67%) were male nonAboriginals. Age at release (all prisoners) ranged from 16 years to 82 years, with mean age 31 years, median age 29 years and mode 21 years. Follow-up time The mean length of follow-up was 1,228 days, the median 1,223 days and the range 0 to 2,160 days. Three prisoners died on the day they were released. Deaths in custody were excluded from this analysis. Survival analysis Survival curves for Aboriginal and non-Aboriginal released prisoners were significantly different (p<0.0001). The survi val outcome for Aboriginals was poorer than that for non-Aboriginals. Figure 1 shows survival cur ves for female Aboriginal, female non-Aboriginal, male Aboriginal and male non-Aboriginal released prisoners. Female non-Aboriginal Male non-Abor iginal Female Aboriginal Causes of death Male Aboriginal The main causes of death in prisoners after release from jail are listed (see Table 1). Suicide was the main cause of death, followed by deaths due to alcohol and drug dependence (particularly opioid-type drugs) and accidental poisoning due to dr ugs (especially heroin). Deaths due to diseases of the circulatory system were significant in males. Deaths due to motor vehicle accidents were common in male 2004 VOL. 28 NO . 1 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Stewart et al. Article Table 1: Causes of death (number of deaths). Cause of death Suicide Alcohol and Accidental drug poisoning dependence due to drugs related 303-304 F10-F19 1 6 9 41 57 E850-E858 X40-X44 1 5 3 29 38 Diseases of the circulatory system 390-459 I00-199 3 0 15 20 38 Motor vehicle (transpor t) accidents E810-E819 V01-V99 3 2 19 15 39 Othera Total ICD9/ ICD10 codes Female Aboriginal Female non-Aboriginal Male Aboriginal Male non-Aboriginal Total E950-E959 X60-X84 2 2 20 40 64 Note: (a) Other causes of death include, in order of frequency: gastro-intestinal, accidental death, cancer, assault, unkno wn (including death subject to coronial inquiry), diabetes , and respiratory illness. Aboriginals. Alcohol and dr ug-related deaths were more common in non-Aboriginals than Aboriginals (39% of total deaths in non-Aboriginals compared with 12% in Aboriginals). In terms of actual numbers, as presented above, there were ver y few female deaths compared with male deaths. Females make up only a small proportion of the cohor t (around 13%). The absolute death rate was highest in male Aboriginals at 17 per 1,000 person-years. In female Aboriginals the rate was 9 per 1,000 person-years, in male non-Aboriginals it was 9 per 1,000 person-years, and in female non-Aboriginals it was 8 per 1,000 person-years. Standardised mortality ratios Standardised mortality ratios were calculated in order to compare the observed rate of death in released prisoners aged 20-40 years with that which would have been expected had the prisoners experienced the same mortality rate as the population of WA aged 20-40 years. The age range 20-40 years was chosen because age specific obser ved/expected values do not vary signif icantly within this age range and, therefore, the SMR provides an accurate summary of the age-specif ic values. Two sets of comparisons were made for all-cause mortality. In the first set (SMR 1), female Aboriginal and female non-Aboriginal prisoners were compared with the whole female population of WA, and male Aboriginal and male non-Aboriginal prisoners were compared with the total male population of WA. In the second set of comparisons (SMR 2), female Aboriginal prisoners were compared with the female Aboriginal population of WA, female non-Aboriginal prisoners were compared with the female non-Aboriginal population of W male Aboriginal prisoners were A, compared with the male Aboriginal population of WA and male non-Aboriginal prisoners were compared with the male non-Aboriginal population of WA (see Table 2). Table 2 shows that all prisoners are at signif icantly increased risk of death after release. It also highlights the high mortality rate experienced by all Aboriginal West Australians aged 20-40 years. Aboriginal released prisoners appear at much greater risk of death when compared with the whole WA population, which is predominantly non-Aboriginal, than when compared with only the Aboriginal population. Female Aboriginal released prisoners in this age range are at 10.8 times greater risk of death than other female Western Australians (Aboriginals and non-Aboriginals together), but they are at 3.4 times g reater risk of death than other female Aboriginals. Similarly, male Aboriginal released prisoners aged 20-40 years are 9.4 times more likely to die than other male West Australians in the same age g roup, but are 2.9 times more likely to die than other male Aboriginals in Western Australia. Table 2 also highlights the vulnerability of female nonAboriginal released prisoners, who are 17.8 times more likely to die than other female non-Aboriginal Western Australians in this age range. Aboriginal male and female released prisoners are Table 2: Standardised mortality ratios (SMR, 95% confidence interval) for all-cause mortality in per sons aged 20-40 years (reference populations: 1. the total population of WA and 2. the Aboriginal and non-Aboriginal populations of WA separately). Female Aboriginal SMR 1a (95% CI) SMR 2b (95% CI) 10.8 (3.7-17.8) 3.4 (1.2-5.6) Female non-Aboriginal 14.2 (6.5-22.0) 17.8 (8.1-27.5) Male Aboriginal 9.4 (7.1-11.6) 2.9 (2.2-3.5) Male non-Aboriginal 5.0 (4.2-5.9) 6.3 (5.2-7.4) Notes: (a) SMR 1 calculated by comparing male and female Aboriginal and non-Aboriginal released prisoners with the total male or female WA population. (b) SMR 2 calculated by comparing Aboriginal released prisoners with Aboriginal population of WA and non-Aboriginal released prisoners with non-Aboriginal population of WA, separately f or males and females. AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH VOL. 28 NO. 1 Vulnerable Groups Risk of death in prisoners after release from jail Table 3: Standardised mor tality ratios (SMR, 95% confidence interval) for specific causes of death (reference populations: the Aboriginal and non-Aboriginal populations of WA). Cause of death Female Abor iginal Female non-Abor iginal Male Aboriginal Male non-Aboriginal Suicide 15.0 (0-35.7) 17.4 (0-41.5) 4.3 (2.3-6.2) 4.9 (3.0-6.7) Drug and alcohol related 3.3 (0-9.7) 115.9 (40.2-191.6) 2.9 (0.6-5.2) 20.1 (15.1-25.2) Motor vehicle (transport) accidents 2.2 (0-6.6) 9.7 (0-28.8) 4.2 (1.9-6.5) 3.1 (1.3-4.9) Diseases of the cir culatory system 2.1 (0-6.3) 0 1.2 (0.1-2.2) 4.6 (0.6-8.7) approximately three times more likely to die than their Aboriginal counterparts, while male non-Aboriginal released prisoners face a six-fold increased risk of death. Risk of death soon after release from jail The relative risk of death within six months after release from jail for prisoners aged 20-40 years w as determined by restricting follow-up to six months and then calculating SMRs. The results of these calculations are presented in Table 4. Female non-Aboriginal released prisoners were at especially high relative risk of death in their first six months after release from jail. During this time, they were 69.1 times more likely to die than their peers in the WA community. Specific causes of death Standardised mortality ratios were calculated for specif ic causes of death for released prisoners aged 20-40 years (see Table 3). Alcohol and drug dependence-related deaths and deaths due to accidental poisoning were combined into one category, labelled ‘drug and alcohol related’. Aboriginal prisoners were compared with the Aboriginal population of WA and non-Aboriginal prisoners were compared with the non-Aboriginal population of WA. The most striking f inding presented in Table 3 is the ver y high relative risk of dr ug and alcohol-related death in female non-Aboriginals. Released female non-Aboriginal prisoners aged 20-40 years were 115.9 times more likely to die due to dr ug and alcohol-related causes than other non-Aboriginal female West Australians in the same age range. While the relative risks are high, the confidence intervals are also wide due to the small sample size. Male non-Aboriginal released prisoners are also at increased risk of death due to drugs and alcohol. They are 20.1 times more likely to die of drug and alcohol-related causes than other male non-Aboriginal West Australians. Female Aboriginal prisoners aged 20-40 years were at 3.4 times greater risk of death due to any cause than their counter parts in the Aboriginal community (see Table 2). However, because the cohort was small and causes of death were varied, it was not possible to accurately determine their cause-specif ic relative risk of death. Confidence inter vals for specific causes of death were wide and included 1. Male Aboriginal released prisoners in the 20-40 year age group had significantly elevated risks of excess deaths due to suicide and motor vehicle accidents. They were 4.3 times more likely to die from suicide and 4.2 times more likely to die as a result of a motor vehicle accident than their peers in the community. Discussion In interpreting any epidemiolo gical study, it is important to consider the possibility that bias may explain par t or all of the observed effects. There are a number of possible causes of bias in the present study. 1. Losses to follow-up. Losses to follow-up would have resulted if the released prisoners moved interstate or overseas, since only deaths occurring in WA are recorded. Losses to follow-up may also have occur red if the prisoner had a number of aliases, so that the name appearing on the prisoner dataset was different to the name appearing on the death record. Any such losses would mean that the number of observed deaths and consequently the SMR was underestimated. 2. Misclassification bias. Misclassif ication of Indigenous status may occur in any of the three sources of data on Aboriginality used in the present study. In the population Census used to determine the total Aboriginal population of WA, individuals may or may not have chosen to reveal their Aboriginal heritage. When death cer tif icates were issued, the person completing the death certif icate may not have known the deceased’s Indigenous status and may have been reluctant to ask. According to the Australian Bureau of Statistics, 7 while most Aboriginal deaths are registered, it is likely that a signif icant proportion are not registered as Aboriginal. On entr y to the prison system, prisoners may or may not have wanted their Aboriginal heritage known. Further, individuals’ reasons for identifying as Aboriginal in the prison system Table 4: SMR values (95% confidence intervals) for released prisoners aged 20 to 40 years, where follow-up was limited to the first six months after release. Female Aboriginal SMR (95% CI) 7.8 (0-16.6) Female non-Aboriginal 69.1 (17.9-120.3) Male Aboriginal 5.9 (3.4-8.4) Male non-Aboriginal 18.2 (13.2- 23.2) 2004 VOL. 28 NO . 1 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Stewart et al. Article may be different to their reasons for identifying as Aboriginal in the Census. The Census, the prison records and records of deaths may well be quite inconsistent with one another where Indigenous status is concerned. If this misclassification is random, the SMR would be biased towards the null; if it is not, the SMR could either over- or under-estimate the tr ue relative risk. In an attempt to evaluate the consistency between death and prisoner records, we compared Indigenous status as recorded on prisoner records and WA death records for the prisoners in our cohort who died. The two sets of records showed fairly good agreement. For 83% of deaths, the same Indigenous status was recorded on both datasets. Only 3% differed. For the remaining 14% of deaths, Indigenous status was not recorded on the death records. 3. Choice of reference population. In calculating the SMR it was assumed that the comparison being made was between the risk of death in the exposed (released prisoners) and the risk of death in the unexposed (the remainder of the community). However, what was actually being compared was deaths in released prisoners with all community deaths (those who had never been to prison, members of the cohor t and all other ex-prisoners). This calculation should approximate a calculation of exposed versus unexposed when ex-prisoners make up a very small proportion of the total population. In the case of non-Aboriginals this is likely to be true; whether it is also true in the case of Aboriginals is another matter. Imprisonment rates in Aboriginals are much higher than in non-Aboriginals. For example, in Western Australia in 1996 the Indigenous imprisonment rate was 2,483 per 100,000 adult Indigenous population, while the imprisonment rate for all West Australians was 170 per 100,000 adult population.8 The Aboriginal community may well include a signif icant proportion of individuals who have been to prison. The SMR calculated for Aboriginals may have underestimated the true relative risk. Despite these limitations, it is clear from the results of this study that all prisoners – male, female, Aboriginal and non-Aboriginal – were at g reater than expected risk of death after release from prison. The relative risk of death was as high as 17.8-fold for allcause death in non-Aboriginal females (69.1-fold for all-cause death in non-Aboriginal females in the first six months after release). The risk of cause-specific death was even more pronounced, with female non-Aboriginals at 115.9 times g reater risk of drug and alcohol-related death and male non-Aboriginals 20.1 times more likely to die of drug and alcohol-related causes. The results of this study are in broad agreement with those already published for released prisoners in Geneva1 and Finland.2 In Geneva, Harding-Pink and Fryc1 found that during their f irst year after release from jail, prisoners experienced a rate of sudden death that was four times greater than the age-adjusted rate in the general population. In the Finnish study, Joukamaa2 found that the study sample of released male prisoners experienced a mortality rate appro ximately four times that of a population-based control g roup. Putkonen et al. 3 found a 17 times greater than expected risk of death in a cohort of 132 female homicide offenders, and a 425 times greater risk of suicide. In contrast to the results of Lattimore et al.,4 who found that homicide was the main cause of death in y oung male offenders after release from jail in California, we found that homicide was a relatively insignif icant cause of death. Only 3% of deaths in this cohort were classif ied as homicide. Tragicall y for those members of this cohort who died, many of these deaths were potentially preventable. Suicides, complications of drug and alcohol dependence, drug overdoses and motor vehicle accidents were responsible for 198/326 (61%) of the deaths. Although potentially preventable, these deaths would have been a consequence of very complex behavioural problems that do not have easy solutions. To properly address them would require co-ordinated programs both pre- and post-release. In the present study, we found that the group of prisoners at particularly high relati ve risk of death after release was female non-Aboriginals. Further, these prisoners were at g reat risk of drug and alcohol-related death. In Victoria, Davies and Cook9 have also examined mortality in women after release from prison. They reported on the deaths of 62 women, of which 45 were due to drug-related causes. Clearly, this problem is not new and it is not restricted to Western Australia. Released female prisoners are an extremely vulnerable group. Non-Aboriginal women make up a relatively small par t of the prison population in Western Australia (only 7% of this cohort), however they are an e xtremely vulnerable group. The results of this study suggest that with particular reference to women, attention should be focused on mental health issues, psychosocial support and dr ug rehabilitation programs. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

Loading next page...
 
/lp/wiley/risk-of-death-in-prisoners-after-release-from-jail-k8nY0oK78I

References (8)

Publisher
Wiley
Copyright
Copyright © 2004 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1467-842X.2004.tb00629.x
Publisher site
See Article on Publisher Site

Abstract

Abstract Objective: T compare the r isk of death in o a cohor t of Wester n Australian released prisoners with the risk experienced by the general population of Western Australia. L. M. Stewart School of Population Health, University of Wester n Australia C. J. Henderson Formerly Director of Prison Health Services, Department of Justice of Western Austr alia Methods: A cohort study of prisoners in Wester n Australia whose last date of release ranged from 1 January 1994 to 1 Januar y 1999. Overall mortality and cause of death were determined by data linkage to the Registrar General’s record of deaths. Results: Aboriginal prisoners had a significantly lower sur vival rate after release than non-Aboriginal prisoners (p<0.0001). When compared with their peers in the Wester n Australian community, both Aboriginal and non-Aboriginal prisoners were found to have an increased relative risk of death. Female non-Aboriginal released prisoners aged between 20 and 40 years were 17.8 (95% CI 8.1-27.5) times more likely to die than other female nonAboriginals in Western Australia in the same age range. Male non-Aboriginal prisoners aged 20-40 years were 6.3 (95% CI 5.2-7.4) times more likely to die than their counterpar ts in the WA community. Female Aboriginal released prisoners were 3.4 (95% CI 1.2-5.6) times more likely to die than their peers, while male Aboriginal released pr isoners were 2.9 (95% CI 2.23.5) times more likely to die. In their first six months after release, female non-Aboriginal prisoners aged 20 to 40 years were 69.1 (95% CI 17.9-120.3) times more likely to die than their counterpar ts in the WA community. The main causes of excess death were related to drug and alcohol abuse. Conclusion: All prisoners w ere at greater than expected relative r isk of death after release from prison, with female nonAboriginal prisoners at par ticularly high relative risk. ( Aust N Z J Public Health 2004; 28: 32-6) M. S. T. Hobbs School of Population Health, University of Wester n Australia S. C. Ridout School of Population Health, University of Wester n Australia M. W. Knuiman School of Population Health, University of Wester n Australia risoners are not always isolated from the general community. They begin as members of the community, move into the prison system and then back into the community again. They may cross the prison dividing wall a number of times in their lives. Their chronic health problems and infectious diseases do not remain behind when they leave prison. Prisoners’ health problems are the community’s health problems. The community must be concer ned with the risk of the spread of infectious disease and the cost of the treating chronic ill health. All members of the community (including e x-prisoners) are entitled to equitable health care. Released prisoners are a vulnerable group who are likely to e xperience higher mortality rates than other members of the community. This has been conf irmed in a number of studies. Harding-Pink and Fryc1 found an increased risk of sudden death in released prisoners in Geneva. Joukamaa2 found an approximately four times greater than expected mortality rate in a sample of male released prisoners in Finland. Putkonen et al.3 found a 17 times greater than expected risk of death in a cohort of 132 female homicide offenders, although it is not clear whether all of these prisoners had been released from jail. Lattimore et al.4 examined mortality in two cohorts of young Californian males paroled during the 1980s and found a particularly high rate of death due to homicide. The purpose of the present study is to examine whether there is excess mortality in released prisoners and, if so, whether there are specific causes of death that contrib ute to excess mortality. The cohort under study comprises male, female, Aboriginal and nonAboriginal West Australian released prisoners. The risk of death in these groups was compared with the risk experienced by the general population of Western Australia. Methods The cohor t under study includes all prisoners whose last date of release from prison in Western Australia ranged from 1 Januar y 1994 to 1 Januar y 1999. Information collated from Ministr y of Justice records included the prisoner’ name, s Submitted: Februar y 2003 Revision requested: June 2003 Accepted: October 2003 Correspondence to: Associate Professor Michael Hobbs, School of Population Health, Unit of Clinical Epidemiology, University of Western Austr alia, Crawley, Western Australia 6009. Fax: (08) 9380 1188; e-mail: mikeh@dph.uwa.edu.au AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH VOL. 28 NO. 1 Vulnerable Groups Risk of death in prisoners after release from jail Indigenous status, sex, date of bir th and date of release from prison. These data were linked with the Registrar-General’s record of deaths to determine how many of the released prisoners had died and their date and cause of death. Deaths were followed up from 1 January 1994 to 1 December 1999. Once data linkage was complete, prisoners’ names were deleted from the dataset. Sur vival of Aboriginal and non-Aboriginal released prisoners was compared and log rank statistics were calculated to determine whether the Aboriginal and non-Aboriginal curves were significantl y different from one another. Sur vival curves were produced for male Aboriginals, female Aboriginals, male nonAboriginals and female non-Aboriginals. Where the released prisoner had died, the cause of death was coded according to ICD-9 or ICD-10, depending on which coding system was current at the time of death. Deaths registered after 1 January 1999 were coded according to ICD-10, while deaths registered prior to this date were coded according to ICD-9. ICD-9 and ICD-10 codes provide logical groupings of deaths. Deaths were g rouped into a small number of categories according to their ICD-9 and ICD-10 codes. The risk of death in the prisoner cohort was compared with the risk of death in the WA population using indirect standardisation. Age-specific death rates for the WA population (by gender and Indigenous status) were calculated using data on deaths in WA over the time period of the study (1 January 1994 to 1 December 1999) collated by the W Registrar General’ Office, and the esA s timated resident population of Western Australia aged 15 years and older as at 30 June 1996, as determined by the 1996 Census.5 Person-years of exposure from release to death or the end of follow-up (1 December 1999) were calculated for the prisoner cohor t. Expected numbers of deaths for each group (female Aboriginal, female non-Aboriginal, male Aboriginal and male non-Aboriginal) and each five-year age band were calculated by multiplying the age-specif ic death rates for this group in the WA population by the person-time of exposure. The comparison of observed to expected deaths yielded the standardised mortality ratio (SMR). Figure 1: Survival after release in male and f emale Aboriginal and non-Aboriginal prisoners. Within each group (female Aboriginal, female non-Aboriginal, male Aboriginal and male non-Aboriginal), the variation in age specif ic observed to expected deaths was tested using Poisson regression to deter mine if there was signif icant variation from age g roup to age group, since if there was then the overall SMR would not accurately summarise the stratum specific rates. Agespecif ic observed/e xpected ratios did not vary signif icantly between the ages of 20-40 years. SMRs presented in this report refer only to this age range. 95% conf idence intervals for SMRs were calculated using this formula: 95% CI=SMR±1.96√(SMR/exp) where exp = Σ age specif ic expected deaths.6 Ethics approval for this study was obtained from the Human Research Ethics Committee of the University of Wester n Australia. Results The cohort The cohort under study included all prisoners whose last date of release from jail fell between 1 January 1994 and 1 January 1999. Deaths were followed up from 1 January 1994 to 1 December 1999. There were 9,381 prisoners in total; 326 died after release. Indigenous status, sex and age Five hundred and fifty-four (6% of total) prisoners were female Aboriginals, 628 (7%) were female non-Aboriginals, 1,927 (20%) were male Aboriginals and 6,272 (67%) were male nonAboriginals. Age at release (all prisoners) ranged from 16 years to 82 years, with mean age 31 years, median age 29 years and mode 21 years. Follow-up time The mean length of follow-up was 1,228 days, the median 1,223 days and the range 0 to 2,160 days. Three prisoners died on the day they were released. Deaths in custody were excluded from this analysis. Survival analysis Survival curves for Aboriginal and non-Aboriginal released prisoners were significantly different (p<0.0001). The survi val outcome for Aboriginals was poorer than that for non-Aboriginals. Figure 1 shows survival cur ves for female Aboriginal, female non-Aboriginal, male Aboriginal and male non-Aboriginal released prisoners. Female non-Aboriginal Male non-Abor iginal Female Aboriginal Causes of death Male Aboriginal The main causes of death in prisoners after release from jail are listed (see Table 1). Suicide was the main cause of death, followed by deaths due to alcohol and drug dependence (particularly opioid-type drugs) and accidental poisoning due to dr ugs (especially heroin). Deaths due to diseases of the circulatory system were significant in males. Deaths due to motor vehicle accidents were common in male 2004 VOL. 28 NO . 1 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Stewart et al. Article Table 1: Causes of death (number of deaths). Cause of death Suicide Alcohol and Accidental drug poisoning dependence due to drugs related 303-304 F10-F19 1 6 9 41 57 E850-E858 X40-X44 1 5 3 29 38 Diseases of the circulatory system 390-459 I00-199 3 0 15 20 38 Motor vehicle (transpor t) accidents E810-E819 V01-V99 3 2 19 15 39 Othera Total ICD9/ ICD10 codes Female Aboriginal Female non-Aboriginal Male Aboriginal Male non-Aboriginal Total E950-E959 X60-X84 2 2 20 40 64 Note: (a) Other causes of death include, in order of frequency: gastro-intestinal, accidental death, cancer, assault, unkno wn (including death subject to coronial inquiry), diabetes , and respiratory illness. Aboriginals. Alcohol and dr ug-related deaths were more common in non-Aboriginals than Aboriginals (39% of total deaths in non-Aboriginals compared with 12% in Aboriginals). In terms of actual numbers, as presented above, there were ver y few female deaths compared with male deaths. Females make up only a small proportion of the cohor t (around 13%). The absolute death rate was highest in male Aboriginals at 17 per 1,000 person-years. In female Aboriginals the rate was 9 per 1,000 person-years, in male non-Aboriginals it was 9 per 1,000 person-years, and in female non-Aboriginals it was 8 per 1,000 person-years. Standardised mortality ratios Standardised mortality ratios were calculated in order to compare the observed rate of death in released prisoners aged 20-40 years with that which would have been expected had the prisoners experienced the same mortality rate as the population of WA aged 20-40 years. The age range 20-40 years was chosen because age specific obser ved/expected values do not vary signif icantly within this age range and, therefore, the SMR provides an accurate summary of the age-specif ic values. Two sets of comparisons were made for all-cause mortality. In the first set (SMR 1), female Aboriginal and female non-Aboriginal prisoners were compared with the whole female population of WA, and male Aboriginal and male non-Aboriginal prisoners were compared with the total male population of WA. In the second set of comparisons (SMR 2), female Aboriginal prisoners were compared with the female Aboriginal population of WA, female non-Aboriginal prisoners were compared with the female non-Aboriginal population of W male Aboriginal prisoners were A, compared with the male Aboriginal population of WA and male non-Aboriginal prisoners were compared with the male non-Aboriginal population of WA (see Table 2). Table 2 shows that all prisoners are at signif icantly increased risk of death after release. It also highlights the high mortality rate experienced by all Aboriginal West Australians aged 20-40 years. Aboriginal released prisoners appear at much greater risk of death when compared with the whole WA population, which is predominantly non-Aboriginal, than when compared with only the Aboriginal population. Female Aboriginal released prisoners in this age range are at 10.8 times greater risk of death than other female Western Australians (Aboriginals and non-Aboriginals together), but they are at 3.4 times g reater risk of death than other female Aboriginals. Similarly, male Aboriginal released prisoners aged 20-40 years are 9.4 times more likely to die than other male West Australians in the same age g roup, but are 2.9 times more likely to die than other male Aboriginals in Western Australia. Table 2 also highlights the vulnerability of female nonAboriginal released prisoners, who are 17.8 times more likely to die than other female non-Aboriginal Western Australians in this age range. Aboriginal male and female released prisoners are Table 2: Standardised mortality ratios (SMR, 95% confidence interval) for all-cause mortality in per sons aged 20-40 years (reference populations: 1. the total population of WA and 2. the Aboriginal and non-Aboriginal populations of WA separately). Female Aboriginal SMR 1a (95% CI) SMR 2b (95% CI) 10.8 (3.7-17.8) 3.4 (1.2-5.6) Female non-Aboriginal 14.2 (6.5-22.0) 17.8 (8.1-27.5) Male Aboriginal 9.4 (7.1-11.6) 2.9 (2.2-3.5) Male non-Aboriginal 5.0 (4.2-5.9) 6.3 (5.2-7.4) Notes: (a) SMR 1 calculated by comparing male and female Aboriginal and non-Aboriginal released prisoners with the total male or female WA population. (b) SMR 2 calculated by comparing Aboriginal released prisoners with Aboriginal population of WA and non-Aboriginal released prisoners with non-Aboriginal population of WA, separately f or males and females. AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH VOL. 28 NO. 1 Vulnerable Groups Risk of death in prisoners after release from jail Table 3: Standardised mor tality ratios (SMR, 95% confidence interval) for specific causes of death (reference populations: the Aboriginal and non-Aboriginal populations of WA). Cause of death Female Abor iginal Female non-Abor iginal Male Aboriginal Male non-Aboriginal Suicide 15.0 (0-35.7) 17.4 (0-41.5) 4.3 (2.3-6.2) 4.9 (3.0-6.7) Drug and alcohol related 3.3 (0-9.7) 115.9 (40.2-191.6) 2.9 (0.6-5.2) 20.1 (15.1-25.2) Motor vehicle (transport) accidents 2.2 (0-6.6) 9.7 (0-28.8) 4.2 (1.9-6.5) 3.1 (1.3-4.9) Diseases of the cir culatory system 2.1 (0-6.3) 0 1.2 (0.1-2.2) 4.6 (0.6-8.7) approximately three times more likely to die than their Aboriginal counterparts, while male non-Aboriginal released prisoners face a six-fold increased risk of death. Risk of death soon after release from jail The relative risk of death within six months after release from jail for prisoners aged 20-40 years w as determined by restricting follow-up to six months and then calculating SMRs. The results of these calculations are presented in Table 4. Female non-Aboriginal released prisoners were at especially high relative risk of death in their first six months after release from jail. During this time, they were 69.1 times more likely to die than their peers in the WA community. Specific causes of death Standardised mortality ratios were calculated for specif ic causes of death for released prisoners aged 20-40 years (see Table 3). Alcohol and drug dependence-related deaths and deaths due to accidental poisoning were combined into one category, labelled ‘drug and alcohol related’. Aboriginal prisoners were compared with the Aboriginal population of WA and non-Aboriginal prisoners were compared with the non-Aboriginal population of WA. The most striking f inding presented in Table 3 is the ver y high relative risk of dr ug and alcohol-related death in female non-Aboriginals. Released female non-Aboriginal prisoners aged 20-40 years were 115.9 times more likely to die due to dr ug and alcohol-related causes than other non-Aboriginal female West Australians in the same age range. While the relative risks are high, the confidence intervals are also wide due to the small sample size. Male non-Aboriginal released prisoners are also at increased risk of death due to drugs and alcohol. They are 20.1 times more likely to die of drug and alcohol-related causes than other male non-Aboriginal West Australians. Female Aboriginal prisoners aged 20-40 years were at 3.4 times greater risk of death due to any cause than their counter parts in the Aboriginal community (see Table 2). However, because the cohort was small and causes of death were varied, it was not possible to accurately determine their cause-specif ic relative risk of death. Confidence inter vals for specific causes of death were wide and included 1. Male Aboriginal released prisoners in the 20-40 year age group had significantly elevated risks of excess deaths due to suicide and motor vehicle accidents. They were 4.3 times more likely to die from suicide and 4.2 times more likely to die as a result of a motor vehicle accident than their peers in the community. Discussion In interpreting any epidemiolo gical study, it is important to consider the possibility that bias may explain par t or all of the observed effects. There are a number of possible causes of bias in the present study. 1. Losses to follow-up. Losses to follow-up would have resulted if the released prisoners moved interstate or overseas, since only deaths occurring in WA are recorded. Losses to follow-up may also have occur red if the prisoner had a number of aliases, so that the name appearing on the prisoner dataset was different to the name appearing on the death record. Any such losses would mean that the number of observed deaths and consequently the SMR was underestimated. 2. Misclassification bias. Misclassif ication of Indigenous status may occur in any of the three sources of data on Aboriginality used in the present study. In the population Census used to determine the total Aboriginal population of WA, individuals may or may not have chosen to reveal their Aboriginal heritage. When death cer tif icates were issued, the person completing the death certif icate may not have known the deceased’s Indigenous status and may have been reluctant to ask. According to the Australian Bureau of Statistics, 7 while most Aboriginal deaths are registered, it is likely that a signif icant proportion are not registered as Aboriginal. On entr y to the prison system, prisoners may or may not have wanted their Aboriginal heritage known. Further, individuals’ reasons for identifying as Aboriginal in the prison system Table 4: SMR values (95% confidence intervals) for released prisoners aged 20 to 40 years, where follow-up was limited to the first six months after release. Female Aboriginal SMR (95% CI) 7.8 (0-16.6) Female non-Aboriginal 69.1 (17.9-120.3) Male Aboriginal 5.9 (3.4-8.4) Male non-Aboriginal 18.2 (13.2- 23.2) 2004 VOL. 28 NO . 1 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Stewart et al. Article may be different to their reasons for identifying as Aboriginal in the Census. The Census, the prison records and records of deaths may well be quite inconsistent with one another where Indigenous status is concerned. If this misclassification is random, the SMR would be biased towards the null; if it is not, the SMR could either over- or under-estimate the tr ue relative risk. In an attempt to evaluate the consistency between death and prisoner records, we compared Indigenous status as recorded on prisoner records and WA death records for the prisoners in our cohort who died. The two sets of records showed fairly good agreement. For 83% of deaths, the same Indigenous status was recorded on both datasets. Only 3% differed. For the remaining 14% of deaths, Indigenous status was not recorded on the death records. 3. Choice of reference population. In calculating the SMR it was assumed that the comparison being made was between the risk of death in the exposed (released prisoners) and the risk of death in the unexposed (the remainder of the community). However, what was actually being compared was deaths in released prisoners with all community deaths (those who had never been to prison, members of the cohor t and all other ex-prisoners). This calculation should approximate a calculation of exposed versus unexposed when ex-prisoners make up a very small proportion of the total population. In the case of non-Aboriginals this is likely to be true; whether it is also true in the case of Aboriginals is another matter. Imprisonment rates in Aboriginals are much higher than in non-Aboriginals. For example, in Western Australia in 1996 the Indigenous imprisonment rate was 2,483 per 100,000 adult Indigenous population, while the imprisonment rate for all West Australians was 170 per 100,000 adult population.8 The Aboriginal community may well include a signif icant proportion of individuals who have been to prison. The SMR calculated for Aboriginals may have underestimated the true relative risk. Despite these limitations, it is clear from the results of this study that all prisoners – male, female, Aboriginal and non-Aboriginal – were at g reater than expected risk of death after release from prison. The relative risk of death was as high as 17.8-fold for allcause death in non-Aboriginal females (69.1-fold for all-cause death in non-Aboriginal females in the first six months after release). The risk of cause-specific death was even more pronounced, with female non-Aboriginals at 115.9 times g reater risk of drug and alcohol-related death and male non-Aboriginals 20.1 times more likely to die of drug and alcohol-related causes. The results of this study are in broad agreement with those already published for released prisoners in Geneva1 and Finland.2 In Geneva, Harding-Pink and Fryc1 found that during their f irst year after release from jail, prisoners experienced a rate of sudden death that was four times greater than the age-adjusted rate in the general population. In the Finnish study, Joukamaa2 found that the study sample of released male prisoners experienced a mortality rate appro ximately four times that of a population-based control g roup. Putkonen et al. 3 found a 17 times greater than expected risk of death in a cohort of 132 female homicide offenders, and a 425 times greater risk of suicide. In contrast to the results of Lattimore et al.,4 who found that homicide was the main cause of death in y oung male offenders after release from jail in California, we found that homicide was a relatively insignif icant cause of death. Only 3% of deaths in this cohort were classif ied as homicide. Tragicall y for those members of this cohort who died, many of these deaths were potentially preventable. Suicides, complications of drug and alcohol dependence, drug overdoses and motor vehicle accidents were responsible for 198/326 (61%) of the deaths. Although potentially preventable, these deaths would have been a consequence of very complex behavioural problems that do not have easy solutions. To properly address them would require co-ordinated programs both pre- and post-release. In the present study, we found that the group of prisoners at particularly high relati ve risk of death after release was female non-Aboriginals. Further, these prisoners were at g reat risk of drug and alcohol-related death. In Victoria, Davies and Cook9 have also examined mortality in women after release from prison. They reported on the deaths of 62 women, of which 45 were due to drug-related causes. Clearly, this problem is not new and it is not restricted to Western Australia. Released female prisoners are an extremely vulnerable group. Non-Aboriginal women make up a relatively small par t of the prison population in Western Australia (only 7% of this cohort), however they are an e xtremely vulnerable group. The results of this study suggest that with particular reference to women, attention should be focused on mental health issues, psychosocial support and dr ug rehabilitation programs.

Journal

Australian and New Zealand Journal of Public HealthWiley

Published: Feb 1, 2004

There are no references for this article.