تاثیرات روانی اجتماعی در خودکشی زندانی: مطالعه مورد-شاهدی از خودآسیبی نزدیک به مرگ در زنان زندانی

Abstract We examined the psychosocial influences on female prisoner suicide by carrying out a study of near-lethal self-harm. We interviewed 60 women prisoners who had recently engaged in near-lethal self-harm (cases) and 60 others who had never carried out near-lethal acts in prison (controls) from all closed female prison establishments in England and Wales, using mixed quantitative and qualitative methods. We gathered information on socio-demographic and criminological variables, life events and childhood trauma, exposure to suicidal behaviour, contributory and precipitating factors for near-lethal self-harm, social support and psychological characteristics. While socio-demographic factors were only modestly associated with near-lethal self-harm, being on remand, in single cell accommodation, and reporting negative experiences of imprisonment were strong correlates. Recent life events and past trauma, including different forms of childhood abuse, were also significantly associated with near-lethal self-harm, as were a family history of suicide and high scores on measures of depression, aggression, impulsivity and hostility, and low levels of self-esteem and social support. Our findings underline the importance of both individual and prison-related factors for suicide in custody, and hence the need for a comprehensive approach to suicide prevention in women’s prisons. Given the multiple needs of female prisoners at-risk of self-harm and suicide, complex psychosocial interventions are likely to be required, including interventions for abused and bereaved women, and initiatives to improve staff–prisoner relationships and reduce bullying. The findings of this research may provide insights into factors leading to suicidal behaviour in other forensic and institutional settings, such as detention centres and psychiatric hospitals, and may assist in developing suicide prevention policies for prisoners and other at-risk populations.

Introduction Internationally, suicide rates in prisoners are considerably higher than in the general population. In a recent study of 12 countries, rates of prison suicide were above 100 per 100,000 prisoners in the majority of countries, compared to an average general population suicide rate of 21 per 100,000 (Fazel, Grann, Kling, & Hawton, 2010). Although this problem has been traditionally associated with male prisoners, who worldwide represent on average 95% of the prison population (Walmsley, 2009), there is evidence that rates of suicide may be as high, or even higher amongst female prisoners (Charles et al., 2003, Mackenzie et al., 2003 and Ministry of Justice, 2010), despite women’s lower risk of suicide in the community (WHO, 2002). In England and Wales, between 1978 and 2004, female prisoners were twenty times more likely to die by suicide than women of the same age in the general population, a proportional excess greater than for male prisoners (Fazel & Benning, 2009). Theoretical models of prison suicide, and suicidality more generally, suggest that suicidal behaviour is rarely the consequence of a single cause or stressor, but rather depends on several state and trait-dependent factors (Hawton and Van Heeringen, 2009, Jenkins et al., 2005 and Liebling and Krarup, 1993). Understanding these factors, the ways in which they interact, and their role in the disproportionately high rates of suicide in prison might assist in developing models of prevention policy in custody and in institutions in general, including detention centres and psychiatric hospitals. However, much of the research in this area has focused on a relatively narrow range of variables (Fazel, Cartwright, Norman-Nott, & Hawton, 2008). Previous research has tended to study these factors within two relatively isolated bodies of literature. On the one hand, psychiatric and psychological studies focussing on the “imported vulnerability” of at-risk prisoners; on the other, sociological analyses of the role of imprisonment itself in precipitating self-harm. More recently, developments in the field have led to the view that “prisons expose already vulnerable populations to additional risk” (Liebling, Durie, Stiles, & Tait, 2005, p. 210), and thus that prison suicide is best understood as a complex phenomenon resulting from the dynamic interactions between individuals and their surroundings. Related attempts to bridge situational and dispositional models of prisoner suicide are consistent with wider theoretical models of suicide as a process within the individual and in interaction with their environment, involving an underlying vulnerability (mostly defined in terms of biological and psychological trait characteristics) which becomes heightened under the influence of specific stressors (Van Heeringen, 2001). Research findings support a life-course model of the aetiology of suicidal behaviour in which risk depends on cumulative exposure to social, environmental, social, personality and mental health factors (Mann, Waternaux, Haas, & Malone, 1999). Personality characteristics, especially hopelessness and low self-esteem (Van Heeringen, Hawton, & Williams, 2000), as well as environmental/contextual factors, and social circumstances, for example recent adverse events, social isolation and institutional bullying, have been highlighted (Blaauw, Winkel et al., 2001, Leese et al., 2006, Rojas and Stenberg, 2010 and Shiner et al., 2009). Despite repeated calls for a comprehensive approach to understanding and addressing the problem (see e.g. McHugh & Snow, 2002), few studies have examined both individual and environmental factors related to prisoner suicide. With much of the previous research on prisoner suicide being in male-only and predominantly male samples (Fazel, Cartwright et al., 2008), risk factors and indicators of vulnerability for suicide in female prisoners are little understood. An important exception is a study by the Office for National Statistics (ONS) (Jenkins et al., 2005, Meltzer et al., 1999 and Singleton et al., 1998), 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, which may provide important insight into the pathways leading to suicidal behaviour. In addition, no qualitative data about triggers and motivations for suicidal behaviour were collected. These may complement and triangulate quantitative analyses of the relationship between particular risk factors and suicidal behaviour, and thus provide a potentially richer and more complex understanding of the meaning and significance of different associations. A further limitation of the ONS study is its focus on lifetime and previous suicidality (based on self-reported intent) in prisoners, rather than suicide in prison. To further theoretical understanding of this problem and develop appropriate intervention strategies, it is potentially more useful to investigate the social and environmental influences on suicidal behaviour occurring (exclusively) during incarceration. Also, although self-harm and attempted suicide have previously been used as proxies for suicide (Marzano, Rivlin, Fazel, & Hawton, 2009), 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 more generally. Medically serious suicide attempters are epidemiologically very similar to individuals who die by suicide (Douglas et al., 2004 and Moscicki, 1995), and twice as likely as other suicide attempters to kill themselves (Rosen, 1976). As means to self-harm are restricted in prison and therefore fatal and near-fatal injuries are more likely to result from behaviour that had not been motivated by suicidal intentions, individuals who have been involved in a ‘near-lethal’ act (based on severity of method and/or injuries) provide a useful focus for analysis (Marzano, Rivlin et al., 2009). In women’s prisons, where self-harming and suicidal behaviours are widespread (Daigle and Côté, 2006 and Völlm and Dolan, 2009), near-lethal self-harm is an important problem in its own right, as well as a valid proxy for suicide. We report findings of a case-control study of women who had recently engaged in a near-lethal act in prison, utilising experiences gained from an earlier pilot study (Borrill, Snow, Medlicott, Teers, & Paton, 2005). We aimed to identify socio-demographic, criminological and psychological variables associated with near-lethal self-harm in order to provide further understanding of this behaviour and inform preventive initiatives. We have elsewhere reported on psychiatric disorders (Marzano, Fazel, Rivlin, & Hawton, 2010).

Results Near-lethal self-harm episodes The near-lethal incidents included in the study involved hanging (28, 47%), ligaturing (15, 25%), severe cuts and lacerations (9, 15%), overdosing (7, 12%), and one self-induced diabetic coma (2%). Most acts were carried out with suicidal intent (only three prisoners stated not being suicidal at the time of their self-harm). The mean suicide intent score was relatively high (18.9, SD = 5.5) compared with an average score of 9.2 (SD = 6.2) in females presenting to a general hospital in England following a self-harm act (Harriss, Hawton, & Zahl, 2005). Socio-demographic factors The majority of prisoners who had engaged in near-lethal self-harm were white, single and under 30 years of age (38, 63%). These and other socio-demographic features did not differ significantly between cases and controls, except that more cases had no educational qualifications (Table 1) (see Appendix 1 for comparisons of cases and female self-inflicted deaths, and controls and the general female prison population). Table 1. Socio-demographic characteristics of female prisoners who engaged in near-lethal self-harm (cases) and those who had not (controls). Cases N = 60 Controls N = 60 Odds ratio (95% CI) p n (%) n (%) Age, median (years) 25.5 26.0 0.581 White ethnicity v. non-white 52 (87) 50 (83) 1.30 (0.48–3.56) 0.609 Singlea 40 (67) 32 (53) 1.75 (0.84–3.66) 0.136 Parent/guardian of children 31 (52) 27 (45) 1.31 (0.64–2.68) 0.465 Educational qualificationsb 30 (50) 41 (68) 0.46 (0.22–0.97) 0.041 Unemployedc 36 (60) 32 (53) 1.30 (0.64–2.71) 0.461 a Including divorced, widowed and separated. b Any vs. none. c Including sick/disabled and housewives. Table options Criminological factors Prisoners who had engaged in near-lethal self-harm were more likely than controls to be on remand, to have had a previous sentence, to have been in their current prison less than 30 days, and to be in single and ‘safe’ cell accommodation (Table 2). Near-lethal self-harm was also significantly associated with not being on normal wing location (vs. segregation, healthcare or intensive residential unit) (13, 22% vs. 1, 2%; OR = 16.3, 95% CI 2.1–129.3, p = 0.001). In contrast, cases and controls did not differ significantly in terms of age at first conviction (case median = 16, control median = 17, z = −1.56, p = 0.12), having two or more previous sentences (n = 26, 43% vs. n = 21, 35%; OR = 1.42, 95% CI 0.68–2.97, p = 0.350), offences, sentence type and length of sentence. However, the control group differed from the general female prison population in having a greater proportion of prisoners remanded or convicted for violence (see Appendix 2 for comparisons on criminological variables of cases and female self-inflicted deaths, and controls and the general female prison population). Table 2. Criminological characteristics of cases and controls. Cases N = 60 Controls N = 60 Odds ratio (95% CI) p n (%) n (%) Prior prison spell 34 (57) 22 (37) 2.26 (1.09–4.70) 0.028 Index offence Violence 21 (35) 24 (40) Sexual 1 (2) 0 (0) Robbery 12 (20) 9 (15) Burglary 5 (8) 1 (2) Other theft 5 (8) 7 (12) Fraud and forgery 0 0 4 (7) Criminal damage 6 (10) 1 (2) Drug offences 4 (7) 11 (18) Other 6 (10) 3 (5) Violent vs. non-violent offence b 34 (57) 33 (55) 1.07 (0.52–2.20) 0.854 Status Remand vs. sentenced 21 (35) 4 (7) 7.54 (2.40–23.68) <0.0001 Sentence type (sentenced prisoners only) Indeterminate vs. determinate sentence 12/39 (31) 10/56 (18) 0.49 (0.19–1.28) 0.142 Sentence length Less than or equal to 6 months 4/39 (10) 2/56 (4) Greater than 6 months to less than a year 1/39 (3) 8/56 (14) 12 months to less than 4 years 9/39 (23) 21/56 (38) 4+ years (including indeterminate sentences) 25/39 (64) 25/56 (45) 18+ months 33/39 (85) 39/56 (70) 2.40 (0.85–6.78) 0.094 Latency Less than 1 month since first reception 11 (18) 4 (7) 3. 14 (0.94–10.5) 0.053 Less than 1 month in current prison 13 (22) 5 (8) 3.04 (1.01–9.16) 0.041 Single cell accommodationc 57 (95) 40 (67) 9.50 (2.64–34.14) <0.0001 ‘Safe cell’ accommodation and 12 (22) 5 (8) 3.07 (1.01–9.38) 0.042 During current prison term: Imprisonment difficult or very difficulta 38/59 (64) 11/58 (19) 7.73 (3.34–17.85) <0.0001 Intimidated to hand over belongings 20 (33) 2 (3) 14.5 (3.2–65.5) <0.0001 Threatened with violence 23 (38) 12 (20) 2.47 (1.10–5.64) 0.027 Victim of actual abuse 12 (20) 6 (10) 2.25 (0.78–6.46) 0.125 Belongings stolen 26 (43) 17 (28) 1.93 (0.91–4.13) 0.087 Received unwanted sexual attention 11 (18) 10 (17) 1.12 (0.44–2.88) 0.810 Victim of forced sexual attentions 5 (8) 5 (8) 1.0 (0.27–3.65) 1.0 a Denominators vary because of missing information. b Including violence, sexual offences and robbery. c At the time of the near-lethal act in cases, and of the interview in controls. d ‘Safe cells’ have reduced ligature points. Two of the cases in safe cell accommodation, together with the five controls in this category, were from HMP Peterborough, where all cells have reduced ligature points. Table options In a multifactorial analysis, having had a prior prison spell, serving a sentence longer than 18 months and being in single cell accommodation were entered simultaneously in a logistic regression model. Prior prison spell (adjusted OR = 3.33, 95% CI 1.30–8.56, p = 0.012) and single cell status (adjusted OR = 10.5, 95% CI 2.1–52.2, p = 0.04) remained significant. Adverse experiences during current prison term Despite some prisoners with near-lethal self-harm reporting that they felt safer and more supported in prison than outside (7/59, 12%), the majority described being in prison as difficult or very difficult (Table 2). Forty cases (67%) reported having experienced at least one stressful event during the current prison term, compared with 28 (47%) controls (OR = 2.29, 95% CI = 1.09–4.78, p = 0.027). The most common of these was having had belongings stolen. However, this was also the most prevalent prison experience reported by controls and did not significantly distinguish the two groups. In contrast, being threatened with violence and intimidated to hand over belongings were significantly associated with near-lethal self-harm, and eight (13%) near-lethal cases reported being bullied because of their self-harm. In both cases and controls, a relatively small proportion of prisoners had suffered physical or sexual abuse while in prison. Social networks in prison Although there were no significant differences in the size of social networks inside prison between cases and controls (case median = 2, control median = 2, z = −0.28, p = 0.778), or in the number of cases and controls who had no close staff member and/or prisoner (15, 25% vs. 11, 18%; OR = 1.49, 95% CI 0.62–3.57, p = 0.375), those with near-lethal self-harm were more likely to describe their relationship with other prisoners as difficult or very difficult (15/58, 26% vs. 3/58, 5%; OR = 6.40, 95% CI 1.85–21.88, p = 0.002), and to speak of all or most staff in negative terms (19/59, 32% vs. 9/58, 16%; OR = 2.59, 95% CI 1.07–6.23, p = 0.034). Even those who were positive about staff were often critical of their reactions to self-harm and their inability to address or understand their needs (13/40, 33%), because of being untrained, overstretched or simply “uncaring”. Qualitative analysis of prison-related influences on near-lethal self-harm Primary factors Problems with staff were mentioned by several prisoners when asked about their reasons for self-harming (9 cases, 15%): I just didn’t want to be around. I had enough of these [staff] pushing me and everything. I did. (Case 52) Arguments with other prisoners (8 cases, 13%) were also mentioned by as a reason – or the primary reason – for having self-harmed: Because I’m on valium-based medication – what everybody wants – I’m on methadone. I kept giving a girl like my methadone all the time. She was bullying me into it. (Case 18) Other primary factors included medication and detoxification issues (4), feeling let down or discriminated against by the system (3), and spending too much time in their cells (3). Eight prisoners (13%) attributed their acts to anxieties about sentencing, prison transfers or being released: I’d just been sentenced on the Thursday…and I was due to get shipped out two days after…I hadn’t got my head around the fact that my sentence was a lot more than what I thought it would be. (Case 32) However, prison-related factors were seldom described as the only influences on prisoners’ near-lethal self-harm. Indeed, 27 prisoners reported having carried out the acts for reasons unrelated to their being in prison, and most had previously self-harmed (39/59, 66%) and attempted suicide (49/58, 85%) outside prison. Triggering factors Factors prisoners said had triggered (rather than caused) near-lethal self-harm were mostly linked to their being in prison. Once again, difficulties with prison staff (12, 20%) and fellow prisoners (12, 20%) appeared to be especially prevalent: There was, there was a bit of an argument I had with somebody [another prisoner] which upset me. (Case 12) Further precipitating factors for near-lethal self-harm included concerns surrounding trials, sentencing and parole boards (4 cases), as well as prison transfers (4), being denied medication (2) or a visit (1), lacking distractions whilst in cell (1) and having no tobacco (1). Additional factors Other factors prisoners said had contributed to, but not necessarily caused or triggered the near-lethal acts, were symptoms of drugs or medication withdrawal (6), missing family and friends outside prison (6), feeling upset after a visit (1), and being disappointed at having to interrupt counselling due to being transferred to another prison (1). Others spoke of finding imprisonment difficult (14), not least because being bullied and exposed to violent offenders evoked memories of their own abuse (2), and due to difficulties in dealing with problems in prison without drugs or alcohol (5), or other coping strategies (4): I never dealt with losing my son and daughter, outside I covered it up with drink and everything else. (Case 30) You’ve got more strategies when you are out. You can do more things, and you can go places and you can…Like when you are in here you’ve got a lot of time to think and dwell and things like that…And distraction helps a great deal; it stops you a great deal. (Case 41) Preventive factors Prison-related factors also featured prominently in prisoners’ accounts of what might have prevented their near-lethal acts. Although a considerable proportion of women (25/57, 44%) reported that their act could not have been prevented, all but two of those who described their self-harm as preventable made reference to prison-related factors. Of these, the most frequently mentioned was being able to talk to someone, be they a member of staff (3), a friend in prison (2), a ‘prison listener’ (a prisoner trained by the Samaritans to listen in confidence to fellow prisoners in distress) (1), or anyone available (3). Other factors that might have prevented their acts were: being treated better by prison officers and healthcare staff (especially in relation to the administration of medication) (5), not being in prison (2), having more distractions and time out of cell (2), more help with their mental health problems (2), reduced access to means to self-harm (1), being in a shared cell (1), and receiving counselling (1). Life events and childhood trauma All cases and controls had experienced at least one of 16 identified adverse life events. Events experienced more commonly included sexual abuse, having been in local authority care, violence at home, running away runaway from home, and death of a partner or child. Eighteen cases (30% vs. 9 controls, 15%; OR = 2.43, 95% CI 0.99–5.96, p = 0.049) reported having a serious physical illness at the time of their near-lethal act, including hepatitis c (6 cases), epilepsy (5), severe asthma (2), diabetes (2), and heart disease (1). Most prisoners had experienced at least five (50/60, 83% vs. 30/59, 51%; OR = 4.83, 95% CI 2.07–11.30, p < 0.0001), and a third of the cases reported ten or more adverse events (20/60, 33% vs. 6/59, 10%; OR = 4.42, 95% CI = 1.62–12.01, p = 0.002). Cases were significantly more likely to report recent life events, with almost half having suffered an adverse event within the previous six months. When prior sexual abuse, violence in the home, bullying, running away from home, and serious money problems were entered in a logistic regression model, only sexual abuse (adjusted OR = 4.10, 95% CI 1.64–10.21, p = 0.002) remained significant. We analysed the sensitivity and specificity of factors that remained significant in multifactorial analyses, namely, prior prison spell, single cell status and sexual abuse. All three factors were present in 29 cases and 7(/59) controls (1 case vs. 6 controls had none of these factors; 7 cases vs. 28 controls had only one; 23 cases vs. 18 controls had two). The model’s sensitivity was 0.48, specificity was 0.88. Compared to controls, cases had significantly greater levels of trauma on the Childhood Trauma Questionnaire (CTQ) and all of its subscales (Fig. 1). Scores on all subscales were significantly intercorrelated at p < 0.0001 ( Table 3). In childhood, 85% of cases reported having been emotionally abused (51 vs. 21, 35% controls; OR = 10.5, 95% CI 4.3–25.5, p < 0.0001), three quarters had been sexually abused (44, 73% vs. 19, 32%; OR = 5.93, 95% CI 2.70–13.07, p < 0.0001) and almost as many had been physically abused (43, 72% vs. 16, 27%; OR = 6.96, 95% CI 3.12–15.51, p < 0.0001) Table 3. Most prisoners in the near-lethal group reported having suffered all three forms of abuse as children (38, 63% vs. 11, 18%; OR = 7.69, 95% CI 3.33–17.80, p < 0.0001). Childhood trauma in cases and controls. Fig. 1. Childhood trauma in cases and controls. Figure options Table 3. Correlation matrix of scores on the childhood trauma scale and subscales in cases and controls (N = 119). Childhood trauma Sexual abuse Emotional abuse Physical abuse Emotional neglect Physical neglect 1.00 Sexual abuse 0.76∗ 1.00 Emotional abuse 0.90∗ 0.57∗ 1.00 Physical abuse 0.88∗ 0.55∗ 0.80∗ 1.00 Emotional neglect 0.85∗ 0.46∗ 0.78∗ 0.70∗ 1.00 Physical neglect 0.86∗ 0.55∗ 0.71∗ 0.75∗ 0.80∗ 1.00 ∗p < 0.0001. All correlation coefficients were calculated using Spearman’s rho. Table options Qualitative analysis of life events and childhood trauma as contributing to near-lethal self-harm Primary factors The strong associations between near-lethal self-harm and adverse life events were further supported by qualitative data about prisoners’ primary reasons for self-harming. Indeed, adverse childhood and adulthood experiences were amongst the factors most frequently perceived as causal by the prisoners, particularly bereavement, sexual abuse and family-related problems. Just over a quarter of cases (16, 27%) said they had self-harmed because they were mourning the loss of a loved one (in three cases by suicide, in six cases of their own child) and wished to end their pain or be reunited with the person they had lost. In most cases (12/16, 75%), the death or deaths being mourned had occurred more than six months prior to the prisoner’s near-lethal act: I was just thinking about my [late] brother and I just wanted to be with my brother, so I took some tablets. That was it. I just didn’t want to be here. (Case 40) Almost a fifth of prisoners (11, 18%) explained their near-lethal act in relation to previous sexual abuse, and associated images, flashbacks, voices and negative feelings. In all but two cases, the abuse had taken place at least six months before the near-lethal act, including six instances of childhood sexual abuse: I had the thoughts as well running around in my head, and going through the pain as well like of when I was abused. So I felt…I just felt like I didn’t want to live anymore. I felt dirty. And worthless (Case 5) I’d convinced myself that he [my rapist] had killed himself in jail and he was coming into my room as a ghost…and it just, it was horrible. It was horrible…I was just getting lower and lower and I just… that’s when I thought, you know, I just want to die. I just seriously want to die. Because it’s not ever going to get any better. (Case 57) Further primary factors in prisoners’ near-lethal self-harm included serious illness and chronic pain (2), severe money problems (1), and concerns about families and children, particularly in relation to a partner or relative’s illness or problem (3), the break-up of a relationship (1), having difficult rapport and poor contact with family (4), and struggling to deal with a child or children being adopted (in 2 cases shortly before the act, and in a further 2 instances over six months before): I found out my husband didn’t want to be with me – and my, I’ve got my two kids that I mentioned, they are both adopted. So I felt that my husband was all I had left. And I woke up in the morning and I was in so much pain and I wanted it all to go away. And I just really really did want to die. (Case 30) Triggering factors Even when not identified as primary factors in the near-lethal acts, life events, mostly recent ones, were sometimes said to have triggered near-lethal self-harm (10, 17%). For example, seven prisoners (12%) reported that their behaviours had been precipitated by distressing news from outside prison: …my mum had said that there was some form I needed to fill in to, to a creditor. So that, that just flummoxed me and I couldn’t cope. (Case 48) I found out that my partner – now ex – wrote a statement against me, saying basically a lot of lies. And that really hurt me quite deeply. (Case 53) Social networks and social support The difficult family relationships which some prisoners identified as playing an important role in their self-harm act were reflected in cases’ social networks and social support scores. Prisoners who had engaged in near-lethal self-harm were significantly more likely than controls to report having no close friends outside prison. More also reported having no relatives to whom they felt close, although this difference was not statistically significant (Table 4). In the previous three months, cases had received fewer visits and phone calls from close friends and relatives, but were as likely as controls to have received letters from people outside prison. Table 4. Social networks and exposure to suicidal and self-harming behaviour of cases and controls. Cases N = 60 a Controls N = 60 a Odds ratio (95% CI) p n (%) n (%) Relatives to whom close None vs. any 18 (30) 10 (17) 2.14 (0.83–5.14) 0.084 Close friends outside prison None vs. any 28 (47) 6 (10) 7.88 (2.94–21.1) <0.0001 No close friends or relatives outside prison 12 (20) 0 (0) <0.0001 Contact with friends/family in past 3 months Any letters 54 (90) 58 (97) 0.31 (0.06–1.60) 0.143 Any phone calls 46 (77) 58 (97) 0.11 (0.03–0.52) 0.001 Any visits 32 (53) 48 (80) 0.29 (0.13–0.64) 0.002 Any contact (letters/calls/visits) 55 (92) 60 (100) 0.92 (0.89–0.99) 0.022 Family members: Died by suicide 14/53 (26) 4/60 (7) 5.03 (1.54–16.42) 0.004 Attempted suicide 11/50 (22) 8/59 (14) 1.80 (0.66–4.90) 0.247 Self-harmed 7/50 (14) 4/59 (7) 2.24 (0.62–8.15) 0.212 Friends: Died by suicide 12/50 (24) 8/59 (14) 2.01 (0.75–5.41) 0.161 Attempted suicide 5/51 (10) 10/59 (17) 0.53 (0.17–1.68) 0.276 Self-harmed 8/51 (16) 12/59 (20) 0.73 (0.27–1.95) 0.528 Know other prisoners who: Died by suicide 20/39 (51) 21/49 (43) 1.40 (0.60–3.27) 0.431 Attempted suicide 25/35 (71) 39/53 (74) 0.90 (0.65–2.33) 0.824 Self-harmedb 51/59 (86) 57/58 (98) a Denominators vary because of missing information. b Test not conducted due to lack of statistical power. Table options In addition, prisoners who had engaged in near-lethal self-harm had significantly lower scores on the Social Support Scale (case median = 19, control median = 20, z = −2.31, p = 0.021), indicating lower levels of perceived social support (in prison or outside). Exposure to suicidal and self-harming behaviour Prisoners who had engaged in near-lethal self-harm were significantly more likely than controls to have lost a family member to suicide (Table 4). However, cases were no more likely than controls to report a family history of attempted suicide and self-harm, or to have been exposed to friends’ or other prisoners’ suicidal and self-harming behaviours. Psychological characteristics and influences All psychological variables were significantly intercorrelated (Table 5). Near-lethal cases had lower self-esteem than controls, and scored more highly on the measures of hostility, impulsivity, aggression and depression. Scores on the Beck Depression Inventory (BDI) indicated that all prisoners who had engaged in near-lethal self-harm had symptoms of depression (vs. 34/59, 57.6% controls, p < 0.0001), with most having severe symptoms (BDI score ≥ 30: 44/60, 73.3% vs. 6/59, 10.2%; OR = 24.3, 95% CI 8.8–67.4, p < 0.0001) and a considerable proportion reporting symptoms of severe hopelessness (24/60, 40% vs. 5/60, 8%; OR = 7.33, 95% CI 2.56–20.98, p < 0.0001). Table 5. Correlation matrix of impulsivity, hostility, self-esteem, aggression and depression scores in cases and controls (N = 120). Impulsivity Hostility Self-esteem Aggression Depression Impulsivity (15–60) 1.00 Hostility (0–21) 0.73a,∗ 1.00 Self-esteem (12–48) −0.59b,∗ −0.49a,∗ 1.00 Aggression (0–28) 0.68a,∗ 0.76a,∗ −0.33a,∗ 1.00 Depression (0–63) 0.50 and ,∗ 0.45a,c,∗ −0.70a,c,∗ 0.38a,c,∗ 1.00 Median or Mean (SD) Cases 40.2∗ (6.83) 13.5∗ 27.3∗ (4.87) 18.0∗ 36.0∗ Median or Mean (SD) Controls 33.5 (6.80) 9.50 33.0 (5.48) 12.0 11.0 ∗p < 0.0001 for correlations and case-control comparisons. a Correlation coefficient calculated using Spearman’s rho. b Correlation coefficient calculated using Pearson’s r. c N = 119. Table options Many prisoners who had engaged in near-lethal self-harm described their acts as impulsive (28/56, 50%), with only a fifth (11/56, 20%) having made a detailed plan and just over a third (22/56, 39%) having contemplated their act for more than three hours. Primary factors The important role of psychological variables in prisoners’ near-lethal self-harm was further corroborated by participants’ accounts of their reasons for their acts. Over a quarter of participants (16, 27%) reported having been primarily motivated by intense feelings of depression and hopelessness: I was really depressed. And I just wanted to kill myself, so I tied a ligature and I don’t remember much else… I was just really low and really depressed. You know, I couldn’t see no future. (Case 22) Anger and frustration were also cited as common reasons for engaging in near-lethal self-harm (13, 22%), sometimes alongside depression: It was just mixed feelings, like. I was angry, upset, it was … I don’t know. I just didn’t want to feel like shit anymore and that. And I didn’t want to have the thoughts and that anymore. (Case 27) Feeling worthless (4), guilty (2), paranoid (2) and desiring to be at peace (temporarily or permanently) (6) were also perceived as causal by some prisoners, together with hearing voices (6) and experiencing flashbacks of past trauma (5). Triggers and additional factors Distressing flashbacks and internal voices were not always identified as primary reasons for near-lethal self-harm, with some prisoners describing them as triggers (12) or additional factors (5): I was being visited by my grandma and she was telling me that I was going to hell with her. Sort of getting a lot of hallucinations, and a lot of voices. (Case 1)