Prisoners are at increased risk of suicide. Investigation of both individual and environmental risk factors may assist in developing suicide prevention policies for prisoners and other high-risk populations. We conducted a matched case-control interview study with 60 male prisoners who had made near-lethal suicide attempts in prison (cases) and 60 male prisoners who had not (controls). We compared levels of depression, hopelessness, self-esteem, impulsivity, aggression, hostility, childhood abuse, life events (including events occurring in prison), social support, and social networks in univariate and multivariate models. A range of psychosocial factors was associated with near-lethal self-harm in prisoners. Compared with controls, cases reported higher levels of depression, hopelessness, impulsivity, and aggression, and lower levels of self-esteem and social support (all p values <0.001). Adverse life events and criminal history factors were also associated with near-lethal self-harm, especially having a prior prison spell and having been bullied in prison, both of which remained significant in multivariate analyses. The findings support a model of suicidal behaviour in prisoners that incorporates imported vulnerability factors, clinical factors, and prison experiences, and underscores their interaction. Strategies to reduce self-harm and suicide in prisoners should include attention to such factors.
Citation: Rivlin A, Hawton K, Marzano L, Fazel S (2013) Psychosocial Characteristics and Social Networks of Suicidal Prisoners: Towards a Model of Suicidal Behaviour in Detention. PLoS ONE 8(7): e68944. doi:10.1371/journal.pone.0068944
Editor: Marianna Mazza, Catholic University of Sacred Heart of Rome, Italy
Received: February 8, 2013; Accepted: June 4, 2013; Published: July 29, 2013
Copyright: © 2013 Rivlin et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The study was funded by the NHS Forensic Mental Health R&D Programme (MRD 12/101 and MRD 12/105). KH is also supported by Oxford Health NHS Foundation Trust and is a National Institute for Health Research Senior Investigator. SF is funded by the Wellcome Trust (095806). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Rates of suicide in male prisoners in many high income countries are around three to seven times higher than in the general population . Therefore, several national suicide prevention strategies specifically target this high risk group , . Such strategies would benefit from more information on specific prisoner subgroups with different risk profiles , and a deeper understanding of psychosocial factors, particularly those that are not routinely collected by prison administrations.
Theoretical models of suicide suggest that suicidal behaviour is rarely the result of a single cause or event, but rather depends on the cumulative and interactive effects of several social, environmental, familial, personality and mental health factors , , . These are likely to involve an underlying vulnerability to suicide (mostly defined in terms of biological and psychological traits), which becomes heightened under the influence of particular stressors . In a prison setting, these may include general aspects of the regime, such as adjustment to the prison situation, loss of freedom, and removal from a familiar environment , as well as more specific aspects of prison life, including a lack of purposeful activity (i.e. access to activities to keep occupied such as work or education) , withdrawal from drugs or alcohol , receiving bad news , being in a single cell or in segregation , violence and victimization , and boredom .
A prisoner’s vulnerability to these factors may in turn be influenced by personal characteristics and predispositions that are ‘imported’ into the prison. Among them are current and lifetime psychopathology , physical illness , adverse life events, such as a history of childhood trauma, and personality characteristics likely to influence an individual’s opinion of themselves, perceptions of and adaptations to the environment, and the likelihood of acting on suicidal feelings , .
Consistent with this life-course model of the aetiology of suicide, there is growing awareness of the need to investigate a wide range of both individual and environmental factors in order to better understand and reduce the incidence of suicidal behaviour in prisons. Yet much of the research in this area has focused on a relatively narrow range of variables . An important exception is a study in the UK by the Office for National Statistics (ONS) , in which demographic, social and psychiatric correlates of suicidal behaviour in prisons were explored in a large sample of male and female prisoners. However, this study did not include direct assessment of psychological states or traits, and examined as an outcome lifetime and previous suicidality (based on self-reported intent), rather than suicidal behaviour occurring exclusively during incarceration. Furthermore, the ONS study focused on the broad categories of suicidal ideation and attempts as proxies for suicide. However, there is evidence that physically dangerous and medically severe self-harm acts provide a better approximation of actual suicide than other forms of self-harming behaviour or suicide attempts , . In addition, near-lethal acts of self-harm are an important and prevalent problem in their own right in prisons, for which targeted interventions should be considered .
We studied male prisoners who had made near-lethal suicide attempts in prison and compared them with prisoners who had not engaged in near-lethal self-harm in custody. We had two main research questions. The first was how do psychological characteristics, namely depressive symptoms, hopelessness, self-esteem, impulsivity, aggression and hostility, all of which have been implicated in suicide risk in the general population , differ between prisoners who had made near-lethal suicide attempts and other prisoners? Secondly, we investigated distal (e.g. childhood trauma) and more proximal (e.g. recent life events) factors, together with a range of environmental factors that related to the extent and quality of prisoners’ social networks. A deeper understanding of psychological and environmental factors will potentially contribute to understanding suicidal behaviour in prisoners, and assist in developing effective suicide prevention initiatives in prisons, and possibly in other institutional settings such as the military  and psychiatric inpatient units . The latter settings also have elevated suicide rates, however the contribution of social and environmental factors is sometimes neglected. This more detailed understanding of suicidal behaviour in prisoners is especially important given the known difficulties in developing effective prison screening instruments .
Data were collected between 2007 and 2009 in 19 male prisons in England. These were selected in consultation with the Ministry of Justice because of high rates of suicide attempts and completed suicides (and their being within 100 miles of Oxford). Participating establishments included three Young Offenders’ Institutes (prisoners aged 18–21), three Category ‘A’ (maximum security) prisons, 12 Category B prisons (“establishments for those who do not require maximum security but for whom escape must be made difficult”) and one Category C prison (“for prisoners who cannot be housed in open conditions but who are unlikely to try to escape”).
Near-lethal suicide attempts were defined as acts of self-harm which a) could have been lethal had it not been for intervention or chance, and/or b) involved methods which are associated with a reasonably high chance of death . These criteria have been outlined elsewhere in more detail . Prison officers used these criteria to identify prisoners to refer to the study. Cases were interviewed within four weeks of the suicide attempts.
We included 60 male prisoners in the study, and excluded 42. The 60 participating prisoners were significantly more likely than those excluded to be white (52/60 (87%) vs. 25/42 (60%); χ2 = 9.8, p = 0.01) and to be on a life sentence (13/39 (33%) vs. 2/23 (9%); χ2 = 4.6, p = 0.03). There were no other significant differences between the included and excluded prisoners with regard to tested socio-demographic or criminological characteristics.
Prisoners who had never made a near-lethal suicide attempt whilst in prison were randomly selected by the Ministry of Justice from the Prison Service’s daily list of prisoners. As confounding of risk of suicide in prisons has been shown for age, gender, and facility type , controls were matched with cases in terms of age (five years older or younger), gender and type/category of prison, although each control was from a different prison to the prisoner to which they were matched. All participants were over the age of 18 years.
The interviews lasted between 90 and 120 minutes and took place in private in the prison. They included the following measures, some of which were abbreviated to allow for their use within the constraints of custody (Table 1):
Table 1. Measures used in study of near-lethal suicide attempts in male prisoners.doi:10.1371/journal.pone.0068944.t001
Sociodemographic and criminological variables.
Sociodemographic information was gathered using an adapted version of a structured questionnaire which has been reliably used in many previous studies of self-harm . A questionnaire to collect criminological data was adapted from one used in a major study of psychiatric morbidity amongst prisoners in England and Wales .
The Beck Depression Inventory I-A (BDI) was used to assess levels of depression , . This is a 21-item self-report measure where each item can score between 0 and 3. It has high levels of internal consistency, stability and validity .
We used Item 2 (Hopelessness) of the BDI to assess levels of hopelessness. Beck and colleagues  reported that this item was almost as predictive of eventual suicide in 211 suicide ideators as the 20-item Hopelessness scale.
We measured self-esteem using a modified version of Robson’s Self Concept Scale . We modified the scoring from the original 7-point scale to a 4-point Likert scale (completely agree, agree, disagree, completely disagree). Each of the positive items scores between 1 and 4, with negative items reversed. Total scores range from 12 to 48.
We measured impulsivity using the Plutchik Impulsivity Scale (PIS) , , which includes 15 4-point Likert scale items. Each item scores between 1 and 4, providing a total score range of 15 to 60. This scale is reported to have good validity and internal consistency .
This was measured with the Brown-Goodwin Assessment for Lifetime History of Aggression questionnaire (BGLA) . We excluded two of the original nine items as they relate specifically to military issues, but retained the original scoring.
This was assessed using a modified version of the Buss-Durkee Hostility Inventory (BDHI) , , . As administration of the original 75-item questionnaire is time consuming, we used only two of its seven subscales: ‘assault’ (physical violence towards others) and ‘irritability’ (readiness to explode with negative affect at the slightest provocation). These two combined subscales have amongst the highest individual internal consistency coefficients of all the subscales , and have previously been used in studies investigating the relationship between suicide attempts and hostility .
The Childhood Trauma Questionnaire (CTQ) is a 28-item self-report inventory which assesses histories of emotional, physical and sexual abuse, and of emotional and physical neglect . It has good test-retest reliability and internal consistency, and its criterion validity has been found to be acceptable . However, to simplify the questionnaire, we abridged the original 5-point Likert scale to three points: often, sometimes and never, allowing a total score of between 25 and 75.
Life events and prison experiences.
This checklist was adapted from a psychiatric morbidity survey amongst English and Welsh prisoners . Each item is rated ‘yes’ or ‘no’. In addition, we included questions on whether the prisoner had ever been in local authority care, had a family history of suicide or self-harm, and whether they knew people in prison who had self-harmed or died by suicide.
The Social Support Scale (SSS) is a 7-item instrument measuring self-perceived social support. It has been adapted for use both in the general community  and in prison , . It has three response categories: not true, partly true and certainly true. Each item is scored between 1 and 3, and overall scores range from 7 to 21. Scores under 17 indicate that participants perceive a severe lack of social support, between 18 and 20 a moderate lack of social support, and 21 indicates no lack of social support.
We asked participants about their social networks in relation to extent (number of external contacts via letters, telephone calls and visits since being in prison) and quality (number of close friends and relatives that they feel close to outside and inside prison). The questionnaire was based on that used in the ONS prison study (1998).
We have elsewhere reported on associations between near-lethal self-harm in male prisoners and diagnosed psychiatric disorders , and on the psychosocial and psychiatric influences on female prisoner suicide , .
The analyses were conducted using the Statistical Package for the Social Sciences  and STATA . We adopted a 95% (p<0.05) significance level. For continuous data, paired sample t-tests were used. Odds ratios, 95% confidence intervals and associated p-values for analyses of categorical factors were calculated using McNemar’s Test to account for matching of cases and controls.
We conducted conditional logistic regressions to determine which factors were independently predictive of near-lethal suicide attempts. Specifically, within the criminological and life events domains, we identified which variables were associated univariately with near-lethal self-harm at a 95% significance level (p<0.05), and present in at least ten pairs of participants (to avoid model instability). In addition, we dropped collinear variables to reduce the risk of over-adjustment. We analysed the sensitivity and specificity of factors that were potentially important in three separate models. These models included combinations of risk factors to examine their predictive accuracy for near-lethal self-harm. The first model included those factors that remained significant in multifactorial analyses. The second and third models added one factor each that was of borderline significance in these multifactorial analyses.
Finally, we tested associations among psychological variables, and within scores on the childhood trauma scale and subscales, using Pearson’s r (for normally distributed data) and Spearman’s rho correlations (for non-normal distributions).
Informed Consent, Confidentiality and Ethical Approval
Prisoners who met the inclusion criteria received a participant information sheet which explained details of the study, including its purpose, what the interview would entail, and the ability of the prisoner not to participate without any adverse consequences to themselves, any medical treatment they may have been receiving, or their sentence plan, parole or any other aspect of their life in prison.
Before the interview took place, participants were reminded of the purpose of the study and what the interview would entail. Participants were asked whether they had any questions. They were assured that they could withdraw their consent for the interview at any stage without any adverse consequences. Confidentiality was assured to the prisoner, except in cases where a serious threat was posed to their, or someone else’s life. Following from these discussions, the prisoner was asked if he was willing to participate and, if so, was asked to sign a consent form.
Participants had access to support both before and after the interview from a Suicide Prevention Coordinator, Chaplain, Samaritan, Listener (trained peer support) or Psychologist.
The study had ethical approval from the UK’s Central Office for Research Ethics Committee (Ethics number 06/MRE12/83) and the Prison Service for England and Wales (Reference PG 2006 063).
Our data contains potentially identifiable information. Because of the strict confidentiality agreements in place with participants at the time of the interviews we do not intend to make our data publically available.
Two-thirds (n = 40, 67%) of the near-lethal suicide attempts were by hanging or ligaturing, 12 (20%) by severe cutting, three (5%) by self-asphyxiation, three (5%) by overdose of analgesics, one (2%) by ingestion of a foreign object and one (2%) by self-immolation.
Sociodemographic and Criminological Variables
Sociodemographic and criminological characteristics of the cases and controls are presented in Table 2. Cases were significantly more likely than controls to be white and to have no educational qualifications. There were no differences between cases and controls in their marital or employment status prior to prison, nor in whether they had children.
Table 2. Sociodemographic and criminological characteristics of male prisoners who made near-lethal suicide attempts (cases) and those who had not (controls).doi:10.1371/journal.pone.0068944.t002
Cases were significantly more likely than controls to have had a prior prison sentence, and to have had two or more (compared to none or only one) prior prison sentences. Cases were significantly younger when they received their first conviction (mean age in years = 15.8, SD = 5.4, range = 10–28 v. 17.9, SD = 6, range = 10–28; p = 0.034). Whilst there was no statistical difference between cases and controls in terms of their prison status (remand or sentenced), cases were significantly more likely to have been in prison for less than 30 days since first reception and for less than 30 days in the current prison.
There was no significant difference between cases and controls with respect to their category of index offence. In particular, cases were no more likely than controls to have been arrested or convicted of a violent act. Whilst it was not possible, due to lack of statistical power, to determine whether there was a difference in sentence type (life or determinate) between cases and controls, the data do not suggest any strong differences: 13 (33%) of 39 cases who had been sentenced were serving a life sentence compared to 16 (33%) of 48 controls who were also on life sentences. Similarly, for those prisoners serving a determinate sentence, there were no differences in the length of sentences between cases and controls (Table 2).
Whilst in prison for the current offence, cases were significantly more likely than controls to have been held in a ‘safer cell’ (cells with reduced ligature points and often with a clear Perspex door to facilitate observation). Cases were also significantly less likely to have been employed in prison. There were no significant differences between cases and controls in whether they had been held in solitary confinement (for disciplinary reasons), given ‘added days’ for disciplinary offences, or taken part in drug or alcohol misuse or education programmes on their current sentence.
No multifactorial analysis was conducted as none of the possible factors met the inclusion criteria of univariate significance and statistical stability except having had a prior prison sentence.
Compared to controls, cases had significantly more depressive symptoms, higher hopelessness, impulsivity, aggression and hostility scores, and significantly lower self-esteem scores (Table 3). All psychological variables were significantly intercorrelated (Table 4).
Table 3. Psychological characteristics and reported childhood trauma of male prisoners who made near-lethal suicide attempts (cases) and those who had not (controls).doi:10.1371/journal.pone.0068944.t003
Table 4. Correlation matrix of impulsivity, hostility, self-esteem, aggression and depression scores in all prisoners participating in the study (N = 120).doi:10.1371/journal.pone.0068944.t004
Cases scored significantly higher than controls on the Childhood Trauma Questionnaire, and on three of its subscales: emotional abuse, emotional neglect, and physical neglect (Table 3). Scores on all subscales were significantly intercorrelated at p<0.001 (Table 5).
Table 5. Correlation matrix of scores on the childhood trauma scale and subscales in all prisoners participating in the study (N = 120).doi:10.1371/journal.pone.0068944.t005
Approximately 40% of cases reported having been emotionally (n = 25, 42% vs. n = 16, 27% in controls; OR = 2.0, 95% CI 0.8–5.1, p = 0.08) or physically (n = 25, 42% vs. n = 12, 20%; OR = 3.6, 95% CI 1.3–12.4, p = 0.01) abused as children. There was no statistically significant difference between the numbers of cases and controls who reported having been sexually abused as children (n = 7, 12% vs. n = 5, 8%; OR = 1.5, 95% CI 0.4–7.2, p = 0.53).
Most life events were reported more frequently by cases than controls (Table 6). The association between having experienced adverse life events and a near-lethal act was significant for bullying, having been homeless, and having experienced the death of a parent or sibling. Cases were also significantly more likely than controls to have been in local authority care under the age of 16 years. Cases had also experienced significantly more types of life events than controls (n = 6.5, SD = 3.0, range = 1–13 vs. n = 5.1, SD = 3.0, range = 0–12; p = 0.01). Finally, significantly more cases than controls had experienced their most recent life event in the last year and there was a trend towards more having experienced it in the last 6 months.
Table 6. Life events, experiences of victimization in prison, and exposure to suicide and self-harm of male prisoners who made near-lethal suicide attempts (cases) and those who had not (controls).doi:10.1371/journal.pone.0068944.t006
When having been bullied, homeless, and having a history of being in Local Authority care were entered into a multifactorial regression model, only bullying (OR = 2.5, 95% CI 1.0–6.3, p = 0.04) remained significant.
More cases than controls had experienced non-fatal deliberate self-harm or death by suicide in biological family members and when these phenomena were combined the difference was significant (Table 6). There was no difference between cases and controls in terms of having friends who had deliberately self-harmed or died by suicide. Cases were no more likely than controls to have been exposed to self-harm or suicidal behaviours whilst in prison.
Experiences in Prison
Almost all experiences of victimization in prison (apart from sexual abuse where no cases or controls reported being a victim) were reported more frequently by cases than controls, although these differences did not reach statistical significance (Table 6). Compared to controls however, cases had experienced significantly more types of victimization in prison (n = 1.1, SD = 1.3, range = 0–4 vs. n = 0.7, SD = 1.0, range = 0–4; p = 0.01).
Prisoners making a near-lethal suicide attempt reported lower levels of self-perceived social support than controls on the Social Support Scale (mean = 16.6, SD = 3.7, range = 8–21 vs. mean = 18.9, SD = 2.4, range = 11–21; p<0.001).
Cases were significantly more likely than controls to report none or few close or good friends outside prison, and controls to report having no close or good friends living or working inside prison (Table 7).
Table 7. Social networks and external contacts of male prisoners who made near-lethal suicide attempts (cases) and those who had not (controls).doi:10.1371/journal.pone.0068944.t007
Sensitivity and Specificity of Possible Models
We analysed the sensitivity and specificity of factors that were potentially important in three different models. The first included those factors that remained significant in multifactorial analyses (i.e. having had a prior prison sentence and having been bullied). Both factors were present in 34 cases and 13 controls (23 cases and 34 controls had only one of these factors; 3 cases and 13 controls had none). The model’s sensitivity was 0.57 and specificity was 0.78. In other words, using these factors to predict near-lethal self-harm in prisoners means that one would correctly identify 57% of cases, and also correctly identify 78% who were not at risk.
The second model included being white, having had a prior prison sentence, and having been bullied. All three factors were present in 30 cases and 9 controls. The model’s sensitivity was 0.50, its specificity was 0.85.
The third model included being white, having a family history of self-harm or suicide, a prior prison sentence, and having been bullied. All four factors were present in 14 cases and 4 controls. The model’s sensitivity was 0.23 and specificity was 0.90. Thus, as expected, the sensitivity of the model was reduced by adding to the numbers of factors in any predictive model, but the specificity increased.
There was no improvement in these models when physical abuse was added or when it substituted another factor.
We interviewed 60 men who made near-lethal suicide attempts in prison and 60 prisoners who had never made a near-lethal suicide attempt whilst incarcerated using a semi-structured interview covering a wide range of psychological and environmental factors. We found that a number of psychological characteristics, and factors measuring childhood trauma, life events, social support and social networks, significantly differentiated prisoners who made near-lethal suicide attempts from those who did not. Below we consider these findings in more detail, before bringing them together with our previously reported findings for psychiatric disorder and history of previous self-harm  into an explanatory model for near-lethal suicidal behaviour in prisoners, which we argue is applicable to completed suicide .
Our findings have different implications depending on the comparison group. Compared with the first comparison group, prisoners who did not make near-lethal suicide attempts, our data provide novel information on psychological and environmental risk factors for suicidal behaviour in prison. This has implications for suicide prevention strategies in prison and developing a model of near-lethal self-harm in prisoners. An alternative comparison is with the general population, and examining the findings in this way provides some insight into suicidal behaviour in prison.
Compared with controls, cases had significantly higher levels of depressive symptoms, hopelessness, impulsivity, aggression, hostility, and lower levels of self-esteem. These results are in keeping with previous research on the role of personality characteristics in suicidal behaviour, both in prison , , ,  and in the community , .
Prisoners who undertook near-lethal suicide attempts were also more likely than controls to have experienced childhood trauma, especially emotional or physical abuse, or emotional neglect, as has previously been reported in the community . Similarly, suicidal ideation and attempts were significantly associated with childhood trauma in studies of male prisoners in Italy , England and Wales , and in female prisoners in the US . However, in this report, the difference between cases and controls in self-reported rates of child sexual abuse did not reach statistical significance, in contrast to our parallel study of female prisoners  and other previous research . This apparent difference might be because previous studies have tended to use either mixed-sex or women-only samples of inmates, which have a higher prevalence of child sexual abuse . It is possible that child sexual abuse amongst men is not as important a risk factor for suicide in prison as it is for women. Alternatively, it might be that the method of data collection through face-to-face interview dissuaded male prisoners from admitting to sexual abuse in childhood.
Adverse life events were common amongst the prisoners in this study. However, because of the high rates of these life events in the control group, a finding consistent with previous research , , the only significant differences between cases and controls were for bullying, homelessness, death of a parent or sibling, and having been in Local Authority care. Of these, bullying remained significant in multivariate analysis. It is possible that more modest associations with near-lethal suicide attempts may have become apparent with a larger sample because the prevalence of specific life events was generally higher in cases than in controls. Also, there may have been differences in their degree or impact, rather than simply their presence or absence.
Cases had experienced significantly more types of life events than controls. Similarly, the ONS prison study  found that nearly three times as many male prisoners in England who had tried to kill themselves in the previous year had experienced seven or more events than those who had never attempted suicide. Thus there may be a cumulative effect of life events on the likelihood of attempting suicide, i.e. it is not only the presence or absence of any life event but the number and impact of life events (possibly within a given amount of time) experienced by an individual. It may also be that the relationship between number of life events and risk of suicide is not linear, but there may exist a ‘tipping point’ at which an individual’s ability to cope with adverse events is breached.
More cases than controls had experienced a family member either self-harming (with or without suicidal intent) or dying by suicide. This is consistent with both genetic  and social learning models of suicidal behaviour . However, exposure to self-harm or suicide in friends or fellow prisoners did not differ between cases and controls, the latter finding probably reflecting the high incidence of self-harm and attempted suicide in prisons .
Poor social support was associated with making a near-lethal suicide attempt in prison. Specifically, we found the following factors were associated with such attempts at least at borderline significance levels: having fewer close or good friends both outside and inside prison, feeling less close to relatives, and having fewer external contacts in the form of letters, phone calls and visits. These findings are consistent with previous prison and community studies ,  and suggest that a person’s social environment can contribute to suicide risk. It may be that the more an individual feels connected to social surroundings, the more likely they are to be socialised into the norms of the group or society and consequently less likely to pursue self-harming behaviours and suicide. In prison, this feeling of connectedness may be even more important than in the community since incarceration has already removed the individual from their primary support group in most prisoners. Previous research ,  suggests that other institutional and environmental factors may have an impact on prisoners’ feelings of connectedness, and potentially on risk of near-lethal self-harm. However, as we matched participants by type of establishment, we were not able to fully test these hypotheses.
The risk factors identified above are similar to findings from general population studies, where individuals who die from suicide or engage in suicidal behaviour are also more likely to have suffered childhood trauma or other adverse life events, , have a family member engaged in suicidal behaviour , , have poor social support networks , , and also have higher levels of depressive symptoms, hopelessness, impulsivity, aggression, hostility, and lower levels of self-esteem , . Whilst the risk factors presented in this report appear similar compared to the general population, the background prevalence of these risk factors is much higher in prisoners than in the general population and therefore any interventions focusing on them might have the potential for a greater impact.
Towards a Comprehensive Model of Near-lethal Self-harm in Prisoners
The findings of this study, together with previously published data about the role of psychiatric disorders in prisoners’ near-lethal self-harm , provide support for a theory of prison suicide which incorporates both historical (or lifetime) factors that may make a person vulnerable to suicide as well as prison-related ones, and clinical factors. There are likely to be complex interactions between these factors (Figure 1).
Figure 1. A model of possible pathways for suicide in prisoners.doi:10.1371/journal.pone.0068944.g001
The multifactorial analyses we conducted suggest that prior prison spells and having being bullied are independent factors for near-lethal self-harm in prisoners, possibly alongside white ethnicity and having a family history of suicide and self-harm. Our predictive models were, however, limited by low sensitivity (i.e. high false negative rate), which is likely to be partly a consequence of the models not including any psychiatric diagnostic variables and, possibly, ecological factors such as overcrowding . Adding more factors to these models decreased sensitivity further, highlighting the problem of predicting rare events using risk factors that are common in the population of interest. Nevertheless, these models may be improved, and could be useful in settings where identification of high risk is problematic. Further research is currently underway to determine whether adding psychiatric variables to this model could improve its predictive power.
Strengths and Limitations of the Study
Studying survivors of near-lethal self-harm appears to provide a good proxy for completed suicide in prison , and allows assessment of a wide range of potentially contributing factors . However, due to time limitations, we measured some of these variables using shortened or simplified versions of existing questionnaires. Whilst most of the altered versions of the questionnaires were psychometrically validated, some (e.g. the Robson Self-Concept Scale) were not. Also, all of the measures used relied on prisoners’ self-report.
Interviews were conducted up to four weeks after the act. However, as some of the risk factors that were investigated were trait measures, these should largely have been unaffected by the time delay. Also, whilst the study was adequately powered to test for associations with the main variables, it was underpowered to test for more modest associations. It should be emphasized, however, that conducting a study like this is complex and very laborious and therefore restricted sample size is an inevitable consequence. As we tested multiple associations, it is possible that some of the significant findings at the 5% level were chance findings. Therefore caution is warranted in some of the less strong findings, and these will need replication. Nevertheless, our findings highlight the co-occurrence of criminological and psychosocial problems associated with the risk of suicide, and provide a useful basis for future research.
There are high absolute and relative rates of suicide and self-harming behaviours in prisons in many countries, and prisoners have been identified as a high risk group in national suicide prevention strategies. The findings of this study have implications for suicide prevention both in prisons in the UK and elsewhere, and also perhaps for other institutionalised populations. They support a model of suicidal behaviour in prisons that incorporates both imported risk factors (i.e. characteristics and experiences that individuals already have at the time of entry to prison) and environmental risk factors, including influences in the prison setting. This suggests assessment of suicide risk at the time of reception into prison could include any history of suicidal behaviour in the family, childhood trauma, previous self-harm, adequacy of social networks, mental health, and levels of self-esteem, aggression, and impulsivity. Those who experience adverse life events while in prison, especially if their social support appears to be restricted, are at heightened risk, and careful monitoring of such individuals who have concurrent risk factors should be considered.
We are grateful to Tunde Adeniji, Debra Baldwin, Pat Baskerville, and Jenny Rees of the Ministry of Justice Safer Custody and Offender Policy Group for their support and advice, Mary Piper of the Department of Health, Jo Borrill and Jo Paton for their assistance with the study, Adam Spriggs of the Ministry of Justice for providing control data, and all Area and Local Suicide Prevention Coordinators who helped with recruitment for the study. We also thank Karen Smith of the Centre for Statistics in Medicine for statistical assistance. Lastly, we thank all the prisoners who participated in this study.
Conceived and designed the experiments: AR KH LM SF. Performed the experiments: AR LM. Analyzed the data: AR KH LM SF. Contributed reagents/materials/analysis tools: AR KH LM SF. Wrote the paper: AR KH LM SF.
- 1. Fazel S, Grann M, Kling B, Hawton K (2011) Prison suicide in 12 countries: an ecological study of 861 suicides during 2003–2007. Soc Psychiatry Psychiatr Epidemiol 46: 191–195. doi: 10.1007/s00127-010-0184-4
- 2. Department of Health (2012) Preventing suicide in England: a cross-government outcomes strategy to save lives. London: Department of Health.
- 3. US Department of Health (2012) 2012 National strategy for suicide prevention: goals and objectives for action. Rockville: US Surgeon General and National Alliance for Suicide Prevention.
- 4. Humber N, Piper M, Appleby L, Shaw J (2011) Characteristics of and trends in subgroups of prisoner suicides in England and Wales. Psychol Med 41: 2275–2285. doi: 10.1017/s0033291711000705
- 5. Fergusson DM, Woodward LJ, Horwood LJ (2000) Risk factors and life processes associated with the onset of suicidal behaviour during adolescence and early adulthood. Psychol Med 30: 23–39. doi: 10.1017/s003329179900135x
- 6. Mann J, Waternaux C, Haas G, Malone K (1999) Toward a clinical model of suicidal behviour in psychiatric patients. Am J Psychiatry 156: 181–189.
- 7. Hawton K, Van Heeringen K (2009) Suicide. Lancet 373: 1372–1381. doi: 10.1016/s0140-6736(09)60372-x
- 8. Van Heeringen K (2001) The suicidal process and related concepts. In: Van Heeringen K, editor. Understanding suicidal behaviour: The suicidal process approach to research, treatment and prevention. Chichester: Wiley. 3–14.
- 9. Liebling A (1992) Suicides in prison. London: Routledge.
- 10. Leese M, Thomas S, Snow L (2006) An ecological study of factors associated with rates of self-inflicted death in prisons in England and Wales. Int J Law Psychiatry 29: 355–360. doi: 10.1016/j.ijlp.2005.10.004
- 11. Paton J, Jenkins R (2005) Suicide and suicide attempts in prisons. In: Hawton K, editor. Prevention and treatment of suicidal behaviour: from science to practice. Oxford: Oxford University Press.
- 12. Fazel S, Cartwright J, Norman-Nott A, Hawton K (2008) Suicide in prisoners: a systematic review of risk factors. J Clin Psychiatry 69: 1721–1731. doi: 10.4088/jcp.v69n1107
- 13. Blaauw E, Winkel F, Kerkhof A (2001) Bullying and suicidal behavior in jails. Crim Justice Behav 28: 279–299. doi: 10.1177/0093854801028003002
- 14. HM Chief Inspector of Prisons (2004) Expectations: Criteria for Assessing the Conditions in Prisons and the Treatment of Prisoners. London: HM Inspectorate of Prisons.
- 15. Rivlin A, Hawton K, Marzano L, Fazel S (2010) Psychiatric disorders in male prisoners who made near-lethal suicide attempts: case-control study. Br J Psychiatry 197: 313–319. doi: 10.1192/bjp.bp.110.077883
- 16. Meyer JP, Qiu J, Chen NE, Larkin GL, Altice FL (2012) Emergency Department Use by Released Prisoners with HIV: An Observational Longitudinal Study. PLoS ONE 7: e42416. doi: 10.1371/journal.pone.0042416
- 17. Brezo J, Paris J, Tremblay R, Vitaro F, Zoccolillo M, et al. (2006) Personality traits as correlates of suicide attempts and suicidal ideation in young adults. Psychol Med 36: 191–202. doi: 10.1017/s0033291705006719
- 18. Mann J (2003) Neurobiology of suicidal behaviour. Nature Reviews Neuroscience 4: 819–828. doi: 10.1038/nrn1220
- 19. Meltzer H, Jenkins R, Singleton S, Charlton J, Yar M (1999) Non-fatal suicidal behaviour among prisoners. London: Office for National Statistics.
- 20. Marzano L, Rivlin A, Fazel S, Hawton K (2009) Interviewing survivors of near-lethal self-harm: a novel approach for investigating suicide amongst prisoners. J Forensic Leg Med 16: 152–155. doi: 10.1016/j.jflm.2008.08.011
- 21. Rivlin A, Fazel S, Marzano L, Hawton K (2012) Studying survivors of near-lethal suicide attempts as a proxy for completed suicides in prisons. Forensic Sci Int 220: 19–26. doi: 10.1016/j.forsciint.2012.01.022
- 22. Williams M, Pollock L (2000) The psychology of suicidal behaviour. In: Hawton K, Van Heeringen K, editors. The international handbook of suicide and attempted suicide. Chichester: John Wiley & Sons. 79–93.
- 23. Pinder R, Iversen A, Kapur N, Wessely S, Fear N (2012) Self-harm and attempted suicide among UK Armed Forces personnel: results of a cross-sectional survey. Int J Soc Psychiatry 58: 433.439. doi: 10.1177/0020764011408534
- 24. Powell J, Geddes J, Hawton K, Deeks J, Goldacre M (2000) Suicide in psychiatric hospital in-patients. Br J Psychiatry 176: 266–272.
- 25. Robinson L, Spencer MD, Thomson LDG, Stanfield AC, Owens DGC, et al. (2012) Evaluation of a screening instrument for autism spectrum disorders in prisoners. PLoS ONE 7: e36078. doi: 10.1371/journal.pone.0036078
- 26. Kresnow M, Ikeda RM, Mercy JA, Powell KE, Potter LB, et al. (2001) An unmatched case-control study of nearly lethal suicide attempts in Houston, Texas: research methods and measurements. Suicide Life Threat Behav 32: 7–19. doi: 10.1521/suli.220.127.116.11.24210
- 27. Hawton K, Harriss L, Hall S, Simkin S, Bale E, et al. (2003) Deliberate self-harm in Oxford, 1990–2000: a time of change in patient characteristics. Psychol Med 33: 987–996. doi: 10.1017/s0033291703007943
- 28. Singleton N, Meltzer H, Gatward R (1998) Psychiatric morbidity among prisoners in England and Wales. London: Office of National Statistics.
- 29. Beck A, Ward C, Mendelson M, Mock J, Erbaugh J (1961) An inventory for measuring depression. Arch Gen Psychiatry 4: 561–571. doi: 10.1001/archpsyc.1961.01710120031004
- 30. Beck A, Steer R (1987) Manual for the Beck Depression Inventory. San Antonio: The Psychological Corporation.
- 31. Ambrosini PJ, Metz C, Bianchi MD, Rabinovich H, Undie A (1991) Concurrent validity and psychometric properties of the Beck Depression Inventory in outpatient adolescents. J Am Acad Child Adolesc Psychiatry 30: 51–57. doi: 10.1097/00004583-199101000-00008
- 32. Beck A, Steer R, Kovacs M, Garrison B (1985) Hopelessness and eventual suicide: a 10-year prospective study of patients hospitalized with suicidal ideation. Am J Psychiatry 142: 559–563.
- 33. Robson P (1989) Development of a new self-report questionnaire to measure self esteem. Psychol Med 19: 513–518. doi: 10.1017/s003329170001254x
- 34. Plutchik R, van Praag H (1986) The measurement of suicidality, aggressivity and impulsivity. Clin Neuropharmacol 9: 380–382. doi: 10.1016/0278-5846(89)90107-3
- 35. Plutchik R, van Praag H, Picard S, Conte H, Korn M (1989) Is there a relation between the seriousness of suicidal intent and the lethality of the suicide attempt? Psychiatry Res 27: 71–79. doi: 10.1016/0165-1781(89)90011-5
- 36. Brown L, Goodwin F, Ballenger JC, Goyer PF, Major LF (1979) Aggression in humans correlates with cerebrospinal fluid amine metabolites. Psychiatry Res 1: 131–139. doi: 10.1016/0165-1781(79)90053-2
- 37. Gunn J, Gristwood J (1975) Use of the Buss-Durkee Hostility Inventory among British prisoners. J Consult Clin Psychol 43: 590. doi: 10.1037/h0076882
- 38. Buss AH, Durkee A (1957) An inventory for assessing different kinds of hostility. J Consult Psychol 21: 343–349. doi: 10.1037/h0046900
- 39. Renson GJ, Adams JE, Tinklenberg J (1978) Buss-Durkee assessment and validation with violent versus nonviolent chronic alcohol abusers. J Consult Clin Psychol 46: 360–361. doi: 10.1037/0022-006x.46.2.360
- 40. Vassar M, Hale W (2009) Reliability reporting across studies using the Buss Durkee Hostility Inventory. J Interpers Violence 24: 20–37. doi: 10.1177/0886260508314931
- 41. Koller G, Preuss UW, Bottlender M, Wenzel K, Soyka M (2002) Impulsivity and aggression as predictors of suicide attempts in alcoholics. Eur Arch Psychiatry Clin Neurosci 252: 155–160. doi: 10.1007/s00406-002-0362-9
- 42. Bernstein DP, Fink L, Handelsman L, Foote J, Lovejoy M, et al. (1994) Initial reliability and validity of a new retrospective measure of child abuse and neglect. Am J Psychiatry 151: 1132–1136.
- 43. Bernstein DP, Ahluvalia T, Pogge D, Handelsman L (1997) Validity of the Childhood Trauma Questionnaire in an adolescent psychiatric population. J Am Acad Child Adolesc Psychiatry 36: 340–348. doi: 10.1097/00004583-199703000-00012
- 44. Department of Health (2000) Health survey for England. London: Department of Health.
- 45. Harvey J (2005) Crossing the boundary: the transition of young adults into prison. In: Liebling A, Maruna A, editors. The effects of imprisonment. Devon: Willan Publishing.
- 46. Marzano L, Fazel S, Rivlin A, Hawton K (2010) Psychiatric disorders in women prisoners who have engaged in near-lethal self-harm: case-control study. Br J Psychiatry 197: 219–226. doi: 10.1192/bjp.bp.109.075424
- 47. Marzano L, Hawton K, Rivlin A, Fazel S (2011) Psychosocial influences on prisoner suicide: A case-control study of near-lethal self-harm in women prisoners. Soc Sci Med 72: 874–883. doi: 10.1016/j.socscimed.2010.12.028
- 48. SPSS Inc. (2006) SPSS for Windows Release 15. Chicago: SPSS Inc.
- 49. StataCorp (2005) Stata Statistical Software: Release 9. College Station, TX: StataCorp LP.
- 50. Morgan J, Hawton K (2004) Self-reported suicidal behaviour in juvenile offenders in custody: prevalence and associated factors. Crisis 25: 8–11. doi: 10.1027/0227-5910.25.1.8
- 51. Sarchiapone M, Jovanovic N, Roy A, Podlesek A, Carli V, et al. (2009) Relations of psychological characteristics to suicide behaviour: results from a large sample of male prisoners. Pers Indiv Differ 47: 250–255. doi: 10.1016/j.paid.2009.03.008
- 52. Palmer EJ, Connelly R (2005) Depression, hopelessness and suicide ideation among vulnerable prisoners. Crim Behav Ment Health 15: 164–170. doi: 10.1002/cbm.4
- 53. Ivanoff A, Jang SJ (1991) The role of hopelessness and social desirability in predicting suicidal behaviour: a study of prison inmates. J Consult Clin Psychol 59: 394–399. doi: 10.1037//0022-006x.59.3.394
- 54. Roy A, Janal M (2006) Gender in suicide attempt rates and childhood sexual abuse rates: is there an interaction? Suicide Life Threat Behav 36: 329–335. doi: 10.1521/suli.2006.36.3.329
- 55. Jenkins R, Bhugra D, Meltzer H, Singleton N, Bebbington P, et al. (2005) Psychiatric and social aspects of suicidal behaviour in prisons. Psychol Med 35: 257–269. doi: 10.1017/s0033291704002958
- 56. Verona E, Hicks B, Patrick C (2005) Psychopathy and suicidality in female offenders: mediating influences of personality and abuse. J Consult Clin Psychol 73: 1065–1073. doi: 10.1037/0022-006x.73.6.1065
- 57. Blaauw E, Arensman E, Kraij V, Winkel F, Bout R (2002) Traumatic life events and suicidal risk among jail inmates: the influence of types of events, time period and significant others. J Trauma Stress 15: 9–16. doi: 10.1023/a:1014323009493
- 58. Gorey KM, Leslie DR (1997) The prevalence of child sexual abuse: integrative review adjustment for potential response and measurement biases. Child Abuse Negl 21: 391–398. doi: 10.1016/s0145-2134(96)00180-9
- 59. Ireland JL, editor (2005) Bullying among prisoners: innovations in theory and research. Devan: Willan.
- 60. Turecki G (2001) Suicidal behavior: is there a genetic predisposition? Bipolar Disord 3: 335–349. doi: 10.1034/j.1399-5618.2001.30608.x
- 61. Joiner TE (2005) Why people die by suicide. Cambridge, Mass: Harvard University Press. 276 p.
- 62. Kawachi I, Colditz GA, Ascherio A, Rimm EB, Giovannucci E, et al. (1996) A prospective study of social networks in relation to total mortality and cardiovascular disease in men in the USA. J Epidemiol Community Health 50: 245–251. doi: 10.1136/jech.50.3.245
- 63. Magne-Ingvar U, Öjehagen A, Träskman-Bendz L (1992) The social network of people who attempt suicide. Acta Psychiatr Scand 86: 153–158. doi: 10.1111/j.1600-0447.1992.tb03244.x
- 64. Sheehan D, Lecrubier Y, Harnett-Sheehan K, Amorim P, Janavs J, et al. (1998) The Mini International Neuropsychiatric Interview (M.I.N.I.): The development and validation of a structured diagnostic psychiatric interview. J Clin Psychiatry 59: 22–33. doi: 10.1016/s0924-9338(97)83296-8