Prison suicide in 12 countries: an ecological study of 861 suicides during 2003–2007
- First Online:
- Cite this article as:
- Fazel, S., Grann, M., Kling, B. et al. Soc Psychiatry Psychiatr Epidemiol (2011) 46: 191. doi:10.1007/s00127-010-0184-4
- 1.1k Views
Although suicide rates among prisoners are high and vary between countries, it is uncertain whether this reflects the importation of risk from the general population or is associated with incarceration rates.
We collected data on suicides and undetermined deaths in 12 countries (Australia, Belgium, Canada, Denmark, England and Wales, Finland, Ireland, Netherlands, New Zealand, Norway, Scotland, and Sweden) directly from their prison administrations for 2003–2007. These were compared with rates of suicides in the general population separately by gender using Pearson’s correlations. In addition, they were compared with rates of incarceration. Linear regression was used to examine any association after adjustment for rates of incarceration.
Data were collected on 861 suicides in prison, of which 810 were in men. In the men, crude relative rates of suicide were at least three times higher than the general population. Western European countries had similar rates of prisoner suicide which were mostly higher than those in Australia, Canada, and New Zealand. There was no association between rates of suicide in prisoners and general population rates or rates of incarceration. In the women, inmate suicide rates varied widely and were mostly raised compared with rates in the general population. In addition, these rates did not appear to be associated with general population rates of suicide.
Rates of prison suicide do not reflect general population suicide rates, suggesting that variations in prison suicide rates reflect differences in criminal justice systems including, possibly, the provision of psychiatric care in prison.
In the few countries where it has been studied, suicide rates in prisoners have been found to be substantially higher than the general population. In England and Wales over a quarter of a century, suicide rates in prisoners were reported to be approximately five times higher in men than age-standardised general population rates , and in women, the difference was 20-fold . In one US state, rates of suicide in jails were eight times higher than general population rates  and in Austria, Australia, and New Zealand the rate has been calculated to be 5–10 times higher than in the general population [4–6]. Beyond these countries, little is known about inmate suicide rates compared with population rates.
When explaining elevated rates of prison suicide, different theories have been proposed. Criminologists, using predominantly qualitative data, have argued for the importance of social and environmental factors. One of these is the importation of risk from the general population [7, 8]. Simply put, the hypothesis is that prisoners ‘import’ their risk from outside custody, and hence prison suicide rates will reflect to some extent those in the general population. Another factor that might be relevant is the rate of incarceration as overcrowding has been shown to be associated with suicide rates in a recent ecological study .
Another theoretical approach to prison suicide that is drawn from the fields of psychology and psychiatry has examined the characteristics of those that have died in custody by psychological autopsy and found high rates of psychiatric morbidity, previous suicidality, and recent stressful events [10–12]. A recent systematic review of the case–control comparisons has shown that both criminological and psychiatric factors are important. The strongest risk factors were environmental (being in a single cell), psychiatric (a history of attempted suicide, recent suicidal ideation, and a current psychiatric diagnosis), and criminal history (being on remand, having received a life sentence, and having a violent index offence) . However, this review concluded that the research on environmental and ecological associations with prison suicide was very limited . With the numbers of prisoners increasing in over 70% of the countries in the world , suicide in prisoners will remain an important challenge to public health and policy.
In this study, we aimed to examine rates of prison suicide in different countries and compare them with general population rates of suicide and rates of incarceration in each country. We collected data from prison administrations in 12 countries.
The study was conducted based on data collected from 12 countries: Australia, Belgium, Canada, Denmark, England and Wales (where prison data is reported together), Finland, Ireland, Netherlands, New Zealand, Norway, Scotland, and Sweden.
Prison suicide deaths
Information on all self-inflicted deaths was collected from the 12 countries directly from their prison services’ research departments during 2003–2007 inclusive. In the case of Australia, this information was publicly available in a recent government report. From this, we added self-inflicted and accidental deaths . For the other countries, we requested data on deaths from suicide and undetermined causes separately for men and women. We then calculated rates per 100,000 prisoners for each year by requesting data on mid-year numbers of prisoners for each year from 2003 to 2007, and calculating annualised averages. In Australia, Finland and the Netherlands, numbers of female suicides were unavailable, and we assumed for the analysis that all suicides in custody occurred in men.
General population suicide rates
Information on rates of suicides and undetermined deaths (ICD-10 codes X60-X64 and Y10-Y34) per 100,000 in the population was collected from government statistical offices. In Canada, data for 2003–2005 were used. In Denmark, New Zealand, Norway and Belgium, data for 2003–2006 were used. In Belgium, we relied on statistics for the Flemish part of the country as they are unavailable for the whole nation. In New Zealand, information was not available on rates of deaths due to undetermined causes.
Rate of incarceration
Information on rates of incarceration was drawn from the World Prison Population List, seventh edition, which reported on prisoner numbers for 2005 and 2006 . These rates are based on communication with the national prison administration in these countries or related government sources.
We calculated Pearson correlations between rates of prison suicide and those of the general population in men, women, and both genders combined. We also calculated Pearson correlations between rates of prison suicide and incarceration rates in men. Determining 95% CIs for the rates of prison suicide in women was not possible, as base rates were mostly under five. In men, linear regression was used to examine the association between suicide rates in prisoners and the general population when adjusting for rates of incarceration. SPSS 16 (SPSS Inc, 2008) was used for all analyses. As data were anonymised, ethics approval was not sought and the guidelines governing research from the Declaration of Helsinki were followed.
Data were collected on 861 suicides in 12 countries: Australia, Belgium, Canada, Denmark, England and Wales, Finland, Ireland, Netherlands, New Zealand, Norway, Scotland, and Sweden.
Prison suicides: males
Numbers and rates of suicide in male prisoners compared with the male general population and incarceration rates
No. of prison suicides
Suicide rate per 100,000 prisoners (95% CI)
Suicide rate per 100,000 general population
Incarceration rate per 100,000 prisoners
Crude relative ratea
England and Wales
Prison suicides: females
Numbers and rates of suicide in female prisoners compared with the female general population
No. of prison suicides
Suicide rate per 100,000 prisoners (95% CI)a
Suicide rate per 100,000 general population
Crude relative rateb
England and Wales
Comparisons with the general population
Comparisons with incarceration rates
There was no significant correlation between male prison suicide rates and overall rates of incarceration (r = 0.36, p = 0.25). When adjusted for incarceration rates, rates of prison suicide were not associated with general population suicide rates (r = 0.39, β = 0.16, p = 0.64).
We have reported a study of 861 prison suicides in 12 countries. There were three main findings. First, although rates of suicide in male prisoners varied between 58 and 150 per 100,000 prisoners, they were at least three times higher than estimates of male general population rates of suicide. Rates of suicide in female prisoners also varied widely and a similar pattern emerged of higher rates than the general suicide rates. Second, we found no correlations between suicide rates in prisoners and those in the general population, and prison suicide rates and rates of incarceration. Third, there was little difference in suicide rates between the Western European countries studied. However, Australia, Canada, and New Zealand appeared to have lower rates than the Western European countries.
The results underline the observation that prisoners are a high risk group for suicide  and should therefore be considered as part of national suicide prevention strategies. Rates of suicide in men that we have reported in the present study were all at least three times higher than general population rates of suicides. In the women, they were also higher except in Ireland where there were no suicides in women prisoners during 2003–2007.
A second implication is that our study finds little ecological evidence to support the hypothesis that prison suicide rates partly reflect the underlying rate of suicide in the general population. Furthermore, suicide rates do not appear to be related to incarceration rates.
There will be different caseness criteria used to categorise suicide across the 12 prison systems, and differences in suicide rates between countries may partly reflect this. However, we have tried to minimise this by requesting data on the number of suicides that included undetermined deaths. A study has shown that this approach more accurately reflects suicide numbers . Also we have mostly used data from an established network on prison services that have been working towards a common set of guidelines. Furthermore, the analyses in the paper are necessarily ecological. A more accurate comparison to the prison sample than the general population would be individuals from similar socio-demographic backgrounds . Age-standardised comparisons will provide more accurate risk estimates [18, 19]. Other ecological associations, which may be important that we have not been able to test include overcrowding, amounts of purposeful activity, and rates of severe psychiatric morbidity in the general population and in the homeless [20, 21].
Finally, in view of the finding that rates of prison suicide are not associated with general population suicide rates, this suggests that variations in prison suicide rates reflect differences in criminal justice systems. Future work could examine how Australia, New Zealand, and Canada manage prisoners and prisons to see if there are systemic differences compared with other countries. One possibility is that their prison health services are configured differently. Previous work has shown that changes to psychiatric services in New Zealand prisons in the 1980s led to a striking increase in suicide , implying that changes to the provision and delivery of psychiatric care may be important.
We are grateful to Lars Nylén, Director General of the Swedish Prison and Probation Service, for writing to colleagues in various countries requesting data, and to the following individuals in prison administrations for supplying prison suicide data: Tunde Adeniji for England and Wales, Bill Staubi for Canada, Julie Miller for New Zealand, Ragnar Kristoffersen for Norway, Yvonne Walsh for Ireland, Peter Løvgreen for Denmark, Willy Van den Bergh for Belgium, and Ed Wozniak for Scotland. We also appreciate the assistance of Jan Hoogenboezem, Kees van Heeringen, Sunny Collins, Steve Platt, Lars Mehlum, Grethe Westby, Udo Reulbach, Merete Nordentoft, and Ella Arensman in gathering data on national suicide rates. Sophie Anwar helped collect Australian data. There was no specific funding for this project.
Conflict of interest statement
The authors report no conflict of interests.