Abstract
Purpose
Rates of self-harm are elevated in prison, and there is limited evidence to support the efficacy of brief risk screening at reception to predict and prevent self-harm. This study aims to examine the predictive validity of the self-harm/suicide screening items embedded in a prison mental health screening tool from two key domains strongly associated with risk: previous suicidal/self-harm behaviour, and recent ideation.
Design/methodology/approach
A sample of men and women were screened on entry to prison, with eight screening items covering the two key domains of risk. Follow-up data on self-harm incidents were collected for 12 months post-screening. The predictive validity of individual screening items, item combinations and cumulative screening score was examined for the overall sample and for men and women separately.
Findings
Individual screening items across the two domains were all strongly associated with self-harm in the follow-up period, with odds ratios varying from 2.34 to 9.24. The predictive validity of both individual items, item scores and item combinations demonstrated high specificity but low to moderate sensitivity, and modest area under the curves (AUCs). Predictive validity was generally better for men than women; however, differences were not statistically significant.
Practical implications
Identifying those at risk of self-harm in prisons remains challenging and brief universal screening at prison entry should be only one component of a broader prison risk assessment and management strategy.
Originality/value
To the best of the authors’ knowledge, this study is one of very few to prospectively examine self-harm behaviour following risk screening. Predictive validity was examined in a representative sample of individuals in custody, and for men and women separately.
Keywords
Citation
Browne, C., Chemjong, P., Korobanova, D., Jang, S., Yee, N., Marr, C., Rae, N., Ma, T., Spencer, S.-J. and Dean, K. (2023), "Self-harm risk screening on prison entry: assessing the predictive validity of self-harm history and recent ideation in men and women", International Journal of Prisoner Health, Vol. 19 No. 3, pp. 414-426. https://doi.org/10.1108/IJPH-12-2021-0115
Publisher
:Emerald Publishing Limited
Copyright © 2022, Emerald Publishing Limited
Internationally, rates of self-harm and suicide in prison are significantly higher than in the general population (Fazel et al., 2016; Hawton et al., 2014). Those in custody present an elevated risk of self-harm and suicide even at the time of prison entry, reporting high rates of suicidal behaviour, suicide attempts and self-harm before incarceration (Larney et al., 2012; Fleming et al., 2012). Individuals in prison also have elevated rates of factors known to be associated with self-harm/suicide risk, including mental illness, substance use and trauma (Gottfried and Christopher, 2017).
Australian prison studies have reported findings in keeping with those from other jurisdictions. A 2015 health survey of 1,132 individuals in custody in New South Wales (NSW) found that 30.5% of participants reported experiencing suicidal ideation in their lifetime, 17.8% had attempted suicide and 11.8% had engaged in self-harm (Justice Health and Forensic Mental Health Network, JHFMHN, 2017). The proportion of women who reported a history of suicide attempts and self-harm was higher than for men, a consistent finding in prison samples (Daigle and Côté, 2006; Larney et al., 2012), with the relative rates of suicide in women in custody compared to those in the general population also being much higher than for men (Fazel and Benning, 2009).
Recent meta-analyses of factors associated with self-harm and suicide in custody (Favril et al., 2020b; Zhong et al., 2021) have identified the strongest associations to be for antecedent behaviours/cognitions, including recent suicidal ideation and a history of self-harm or suicide attempt. For the few studies where men and women are considered separately, both commonalities and differences have been identified with regard to risk factors for suicide and self-harm (Miranda-Mendizabal et al., 2019), including amongst those in prison (Horton et al., 2018).
Although much is known about the risk factors for self-harm and suicide in custody, the accurate identification of those who are at risk remains challenging. Risk factors known to be strongly associated with self-harm, such as suicidal ideation and psychiatric diagnosis, are highly prevalent in the general prison population, and pathways from risk to suicidal behaviour are complex (Favril et al., 2020a). Even outside of custodial environments, the performance and clinical utility of suicide risk scales has been questioned (Franklin et al., 2017; Large et al., 2016). In this context, it is perhaps unsurprising that there is limited evidence to support risk screening in prison to predict and prevent self-harm. Systematic reviews by Perry et al. (2010) and Gould et al. (2018) which examine the validity of screening tools in identifying those in this population at risk of suicide or self-harm have found that there are no “gold standard” assessments for this purpose and a small number of studies overall. Of studies examined, they found few prospective studies, evidence of limited predictive validity, small samples and a range of other methodological issues. Despite this, there is some indication that the implementation of self-harm/suicide screening for prison entrants may be followed by a lowering of rates of subsequent self-harm (Gallagher and Dobrin, 2005).
The predictive validity of two key domains of risk most strongly associated with risk of self-harm and suicide in prison – previous behaviour and recent ideation – has been examined in a number of previous studies. In non-prison samples, no significant difference has been found in the relative strength of association of these two factors with regard to risk of suicide (Large et al., 2021), but in prison samples there is some indication that suicidal ideation is a stronger predictor than past behaviour (Favril et al., 2020a, 2020b; Zhong et al., 2021). It is not known, however, whether having a combination of both previous behaviour and recent ideation is more predictive of self-harm behaviours in prison than either factor alone. There is some evidence to suggest that the presence of mental health problems, more broadly, on prison entry might be missed if only one approach to screening is taken (i.e. if only a past history or recent symptoms are considered; Korobanova et al., 2021), and it is possible this principle extends to self-harm/suicide risk.
In NSW, as in many jurisdictions, all prison entrants undergo a standardised reception screening process. The health component of this process is completed by a primary health nurse who is responsible for completing the screening tool and generating referrals to appropriate services in custody (e.g. custodial mental health, general practitioner, drug and alcohol services). The NSW Prison Mental Health Screening study was undertaken to develop and test a new tool for identifying those entering custody who should be prioritised for further mental health assessment and/or management of self-harm/suicide risk, and to elicit important clinical information to inform their treatment and management in prison. The current study describes the incidence of self-harm events occurring over a 12-month period in prison amongst those recruited for the NSW Prison Mental Health Screening study and examines the predictive validity of the self-harm/suicide screening items embedded in the Prison Mental Health Screening tool, in a sample of both men and women, with the self-harm/suicide screening items covering both previous behaviour and recent ideation.
Method
Setting and sample
Data collection was conducted at the main reception centres for women and men entering custody in NSW: Silverwater Women’s Correctional Centre and the Metropolitan Reception and Remand Centre. Between 2016 and 2018, 341 participants were recruited to the study; 112 women and 229 men. Individuals entering custody were identified through the electronic Patient Administration System (PAS) and were approached for participation in random order within 48 h of reception. During the recruitment period, 7,685 individuals (6,619 men and 1,066 women) entered the two centres. Of these, 647 (377 men and 270 women) were approached, of whom 60 (39 men and 21 women) declined to participate and 245 (108 men and 137 women) were unable to be interviewed for a range of reasons (e.g. too mentally unwell, aggressive, unavailable because of attending court or other appointments or insufficient comprehension or use of English). In total, 342 participants consented to, and completed, the baseline data collection interview with one male participant subsequently excluded from the analyses because of missing data.
Measures
Baseline interviews were completed by a research project officer, a registered mental health nurse (PC). The interview comprised a questionnaire developed for the study to record sociodemographic, clinical and criminal justice information, followed by the Prison Mental Health Screening (PMHS) tool covering psychiatric diagnoses, history of mental health treatment and psychiatric symptoms across the diagnostic spectrum experienced currently (within the last month) and historically, as well as questions around suicide and self-harm. The PMHS was designed as a screening tool to identify prison entrants that may require further assessment of their risk, and as a method of collecting important clinical information about the individual to assist with treatment and management decisions.
For this study, eight suicide and self-harm screening items from the PMHS were examined. Four were historical: lifetime history of self-harm; self-harm in past 12 months; lifetime history of suicide attempt; and suicide attempt in past 12 months. The other items measured recent suicidal ideation, with two reflecting passive ideation: “In the past week, have you been thinking that you might be better off dead?” and “In the past week, have you had any thoughts that life is not worth living?”; and two reflecting active ideation: “In the past week, have you had thoughts about hurting or killing yourself?” and “In the past week, have you had thoughts about how you might end your life?” Each of the items were scored as present (1) or not present (0). They were also added to create a “total screening score” out of 8. Additional combination variables were created to identify those with: any history of self-harm/suicide attempts, any self-harm/suicide attempt in the past 12 months, any recent suicidal ideation and any recent history/recent active ideation (any self-harm/suicide attempt in the past 12 months plus any recent thoughts of hurting/killing self or having a suicide plan).
Outcomes
All forms of self-harm were examined as a single outcome type, defined as deliberately causing harm or injury to oneself. A distinction were not made between suicidal and non-suicidal self-injury because of difficulties determining intent based on official records of incidents. Outcome data were collected from electronic custodial and health records for each participant for up to 12 months following initial interview (based on amount of time spent in prison during this time). A binary outcome variable was created with “1” being recorded if the outcome occurred at least once during the follow-up period. A variable was created reflecting the number of days each participant spent in custody during the follow-up period, and all those interviewed at baseline were included in the follow-up analysis, irrespective of time spent in custody during this period.
Statistical analysis
Statistical analyses were carried out using SPSS Statistics 27. Descriptive sociodemographic, clinical and criminal justice statistics were obtained for the overall sample and chi-square analyses conducted for a comparison by gender. The predictive validity of each of the individual screening items, cumulative screening score and variable combinations were examined through calculation of sensitivity (proportion of those endorsing that item who go on to engage in the outcome, or true positive rate), specificity (proportion of those not endorsing that item who do not go on to engage in the outcome, or true negative rate), positive predictive value (the probability of the outcome occurring with endorsement of that item) and negative predictive value (the probability of the outcome not occurring if item is not endorsed). Logistic regression analyses were conducted for each exposure, adjusting for days in custody in the 12-month follow-up period. Receiver Operating Characteristic curves were generated and Area Under the Curve (AUC) values obtained. The AUC reflects the probability that the item will correctly predict the outcome of interest: an AUC of 0.90 reflects a 90% chance that endorsement of that item will predict the outcome.
Ethics approval
The Prison Mental Health Screening study was approved by the Justice Health and Forensic Mental Health Network Human Research Ethics Committee (Ref: G185/14), the NSW Aboriginal Health and Medical Research Council Ethics Committee (Ref: 1137/15) and the Corrective Services NSW Ethics Committee (Ref: D16/139081).
Results
Sample description
Table 1 summarises the sociodemographic, clinical and criminal justice characteristics of the overall sample and separately by gender. Participants’ ages ranged from 18 to 75 years (Median = 32, IQR = 25–40). Just over one-quarter (26.4%) identified as Aboriginal and/or Torres Strait Islander with a significantly larger proportion of women reporting Indigenous status (37.5% vs 21% of males; χ 2 (1) = 10.59, p = 0.001). Nearly three quarters (73.8%) reported educational attainment below Year 12 level and a similar proportion (73.9%) were unemployed at the time of offence. The majority of the sample (65.4%) were on remand at baseline interview and around a quarter (27.0%) of participants were in custody for the first time.
Two-thirds (66.9%) of the overall sample reported previous treatment for a mental health problem and this was significantly higher for women (82.1% vs 59.4% of males; χ 2 (1) = 17.58, p < 0.001). A substantial majority (85%) of the sample reported experiencing at least one psychiatric symptom within the last month (82.1% of men and 91.1% of women; χ 2 (1) = 4.76, p = 0.029).
Suicide/self-harm screening characteristics of the sample
At baseline interview, almost one-quarter (23.8%) of the total sample reported a history of self-harm (excluding suicide attempts), with nearly one in ten (9.1%) of the total sample reporting that they had engaged in self-harm in the previous 12 months (Table 2). One-third of the sample (32.8%) reported a previous suicide attempt, and 8.8% of the total sample reported an attempt in the previous 12 months. Women were significantly more likely to report a history of self-harm (31.3% vs 20.1% of males; χ 2 (1) = 5.18, p = 0.023), with no gender differences observed for self-harm within the past 12 months or history of suicide attempts. Almost a quarter (22.3%) of the overall sample endorsed at least one of the questions relating to suicidal ideation in the past week and this was higher for women (32.1% vs 17.5% of men; χ 2 (1) = 9.35, p = 0.002).
Around half of the total sample (53.4%) recorded a screening score of 0, meaning that they reported no history of self-harm or suicide attempt or recent suicidal ideation. Women were significantly more likely to endorse at least one screening item (60.7% vs 39.7% of men; χ 2 (1) = 13.30, p < . 001), and two or more items (42% vs 25.8% of men, χ2 (1) = 9.21, p = 0.002). In terms of previous self-harm or suicide attempts, of the total sample, 38.7% reported any history; 15% reported such behaviour in the past 12 months. No significant differences were observed between men and women. Nearly a third of women in the sample (32.1%) reported recent ideation, which was a significantly higher proportion than in men (17.5%; χ2 (1) = 9.35, p = 0.002).
Incidence and prediction of self-harm events in the 12-month follow-up period
Only 72 (21.1%) of participants remained in custody for the entire 12-month follow-up period, although 46.8% of those released were reincarcerated during the follow-up period. The median number of days spent in custody over the 12 months was 190 (IQR = 75–330), or approximately six months. Of the total sample interviewed at baseline, 15 (4.4%) engaged in self-harm in prison during the follow-up period, with no significant gender difference: 4.8% of men vs 3.6% of women (χ2 (1) =. 0.27, p = 0.602). No deaths by suicide occurred.
For the assessment of predictive validity in relation to self-harm incidents for the total sample (Table 3), the best performing individual screening item was a history of suicide attempt, with a balance of both moderate sensitivity (66.7%) and specificity (68.7%) and an AUC of 0.68 [95% confidence interval (CI) = 0.54, 0.82]. Those who reported a history of suicide attempt were over four and a half times more likely to self-harm in prison in the 12 months after entering custody than those without a history, adjusting for time in prison during that period [adjusted odds ratio (aOR) = 4.65, 95% CI = 1.54, 14.05]. Specificity of individual items tended to be higher than sensitivity, with the most specific items being those measuring active suicidal ideation (i.e. 96.9% specificity for “had thoughts about how to end life”; 95.7% specificity for “thoughts of hurting or killing self”).
Endorsing any of the eight screening items yielded a moderately high sensitivity of 73.3%, however, the associated specificity was low (54.6%). A screening score of 2 or more produced more balanced sensitivity and specificity (66.7% and 70.6%, respectively) with an AUC of 0.69 (95% CI = 0.55, 0.83).
Although there were no statistically significant differences in AUCs for men and women (Table 4), AUCs were higher for men, with the exception of prediction based on history of suicide attempt (AUC = 0.69 for women vs 0.67 for men). For men, endorsing four or more screening items yielded a moderate AUC of 0.73 (95% CI = 0.54, 0.91) and any self-harm or suicide attempt in the past 12 months, or any suicidal ideation in the past week, were also both moderately predictive (AUC = 0.71, 95% CI = 0.53, 0.90; and AUC = 0.70, 95% CI = 0.52, 0.88, respectively). Previous suicide attempt was the most predictive item for women (AUC = 0.69, 95% CI = 0.44, 0.85), followed by endorsing two or more screening items (AUC = 0.67, 95% CI = 0.42, 0.93) and history of self-harm/suicide attempt in the previous 12 months (AUC =0.67, 95% CI = 0.37, 0.98).
Discussion
In a sample of men and women recently received into prison, rates of previous self-harm behaviour and recent ideation were high. Over 12 months following prison entry, just under 1 in 20 engaged in self-harm in custody. While the eight individual screening items from across the two key domains of risk, previous self-harm/suicidal behaviour and recent ideation, were all strongly associated with self-harm during the follow-up period, predictive validity findings were mixed. There appeared to be some advantage to items from the previous history domain over recent ideation which may reflect the duration of observation and the dynamic nature of ideation. There was some evidence to support the use of variable combinations over individual items and, while findings were generally stronger for men than women, differences were not significant. While individuals entering prison are at high risk, brief screening on entry to predict in-custody self-harm is challenging. Future research should move beyond prediction to focus on evaluating the reduction in self-harm achieved through screening.
Characteristics of prison-entry sample and differences by gender
The current sample reflected the typical sociodemographic profile of individuals entering custody in NSW (Justice Health and Forensic Mental Health Network, JHFMHN, 2017), including indicators of pre-prison disadvantage such as educational attainment below year 12 and high rates of unemployment and mental illness. Approximately a quarter of the overall sample identified as Aboriginal and/or Torres Strait Islander, a similar proportion to that reported in the 2018 NSW inmate census (Corben and Tang, 2019), with the proportion of women of Indigenous background being even higher at over one-third. These figures highlight the enduring overrepresentation of Indigenous people in custody in NSW, noting that Aboriginal people made up only 2.8% of the general population in the last national census in 2016 (Australian Bureau of Statistics, 2021).
At prison entry, almost one-quarter of the total sample reported a history of self-harm, one-third reported a history of suicide attempt and almost a quarter reported experiencing suicidal ideation in the past week. The rates of past self-harm and suicide attempts are higher than previously found in the 2015 Network Patient Health Survey (Justice Health and Forensic Mental Health Network, JHFMHN, 2017), likely because of the remand population of the current study in which rates of mental illness and self-harm risk are known to be higher (Tyler et al., 2019; Willis et al., 2016). The rates found in the current study were similar to those reported in prison surveys conducted in other Australian jurisdictions such as Western Australia (Fleming et al., 2012) and the Australian Capital Territory (Young et al., 2017). Although not all comparisons were statistically significant, the current study found a higher proportion of women than men reported a history of self-harm and suicide attempts, consistent with previous research (Larney et al., 2012).
While a minority of the sample remained in custody for the full 12-month follow-up period, 4.4% of the total sample were reported to have engaged in self-harm in custody in the 12 months post reception, a rate comparable to a prospective study in the UK in which 3.1% of participants engaged in self-harm in the 6 months after reception screening (Ryland et al., 2020).
Predictive validity findings for the total sample
All individual items, covering both domains of previous behaviour and recent ideation, were strongly associated with self-harm in custody during follow-up, even after adjustment for time spent in custody. Specificity rates were high but sensitivity rates were modest at best. This pattern, reflected also in high negative predictive values and low positive predictive values, indicates that individual items may be useful in screening out those not at risk of self-harm, but inadequate for capturing those who are at risk. It is also likely that, while self-harm occurs at a higher rate in prison than in the community, the rate in prison is still exerting a downward pressure on positive predictive (and upward pressure on negative predictive) values.
When the predictive validity of individual items, measured in terms of AUC, was considered, a history of suicide attempt performed best in predicting self-harm during the follow-up period, albeit modestly. A history of self-harm or suicide attempts was more predictive of self-harm than recent ideation, particularly when these events had occurred in the previous 12 months. This may be a function of the dynamic nature of self-harm/suicidal ideation, in that recent ideation may not accurately predict behaviour up to 12 months into the future, and the fact that distress is often highest during the reception period (Walker et al., 2014). When historical and ideation items were combined to form a “screening score” out of eight, a score of two or more demonstrated modest predictive validity; however, sensitivity limited the clinical utility of this measure. The best performing combination of variables was self-harm/suicide attempt in the prior 12 months in terms of AUC; however, sensitivity was highest for lifetime history of self-harm/suicide attempt.
Although there are few prospective studies with which to compare the current study in terms of predicting future self-harm, those that do report predictive validity statistics have similarly found moderate AUC levels and lower sensitivity levels relative to specificity (Naud and Daigle, 2010; Ryland et al., 2020). The optimal balance between sensitivity and specificity in the current context needs to consider the implications of false negative and false positive cases. The former will miss out on potentially effective risk management approaches and the latter may subject the individual to unnecessary restrictions. Brief risk screening tools used with prisoners entering custody should perhaps prioritise sensitivity over specificity to triage those from the total population who require further and more detailed assessment of their risks and needs (i.e. a two-stage approach). Despite the complexity, point-of-reception risk screening is likely to have value, with evidence from the juvenile justice system indicating that facilities screening all entrants within 24 h of arrival have significantly lower odds of serious suicide attempts within a 30-day follow-up period than facilities that do not (Gallagher and Dobrin, 2005). In the current study, the highest sensitivity was associated with a screening score of at least one, suggesting that this may be a threshold by which individuals being screened could be usefully referred for further assessment and management if a two-stage approach was used.
Comparison of predictive validity findings for men and women
The predictive validity of individual items, item scores and item combinations were generally better for men than for women, although the differences were not statistically significant. For men, a screening score of four or more was the most predictive of self-harm in the follow-up period as opposed to a score of two or more for women; for women, a history of suicide attempt was the most predictive item.
To date, relatively little attention has been paid to the implementation and evaluation of prison suicide screening tools specific to women (Gould et al., 2018), although there has been some research in forensic patient samples to suggest certain structured clinical assessment tools typically used for the assessment of violence risk are better predictors of self-harm in women than men (O’Shea and Dickens, 2015; O’Shea et al., 2014). This, along with our findings, suggests that different approaches to risk screening may be required for men and women entering custody.
Strengths and limitations
The current study involved a representative sample of both men and women entering custody in NSW to examine the predictive validity of screening items covering two key domains of risk, and is one of few studies to prospectively examine self-harm behaviour following risk screening. There were also a number of limitations. Although self-harm is more common in prison than in the community, it still occurs at relatively low rates, as observed in this study, which limits the predictive ability of screening approaches. Outcomes were collected for a period of up to 12 months post-reception to capture a sufficient number of events for analysis, but given, as previously mentioned, the dynamic nature of self-harm risk, long-term prediction is challenging. The sample size was also likely inadequate to power analyses of predictive validity by gender, given apparent differences between men and women were not statistically significant, highlighting the need for larger samples in future research.
Outcome data were collected from electronic records only, which, while avoiding potential information biases associated with self-report, does raise the possibility that incidents of self-harm occurred in the follow-up period that were not reported or discovered. Conversely, the predictor variables were all measured by self-report, which may be problematic with sensitive topics. Individuals with prison experience may also be wary of reporting a history of suicidal behaviours or ideation due to the potential consequences of suicide risk management in custody including restrictions, increased observations, specialised cell placements and further assessment or intervention around their risk. Additionally, our self-report methods excluded individuals who did not have sufficient comprehension of English to engage in assessment; further research is required into developing approaches to risk assessment for those with cultural and linguistic barriers to communication (Sorge and Saita, 2021).
Finally, as is common in studies prospectively examining predictors of adverse events in clinical settings, the “risk paradox” requires consideration. If an individual is assessed as being at risk of suicide or self-harm, it would be ethically unsound not to intervene to manage this risk for the sake of determining predictive validity. As such, we must consider the possibility that the number of incidents might have been higher without intervention and that some of those in the “false positive” group represent correctly identified high-risk individuals who have benefitted from effective risk management. However, the study examined self-harm occurring up to 12 months post-reception to capture incidents of self-harm that did not occur in the immediate post-risk assessment period where effective risk management is most likely to occur.
Conclusion
Self-harm and suicide rates remain elevated in prison and there is a challenge in identifying those at risk who might benefit from targeted interventions. Screening for risk in the context of the universal health screening of prison entrants that is routine practice in many jurisdictions may form one component of a broader approach. Brief universal screening for self-harm/suicide risk at prison entry alone may be inadequate; there may be advantages in pairing screening with further triaging/assessment, along with universal preventative approaches (e.g. removing ligature points). Brief universal screening should incorporate both history of self-harm/suicidality and recent ideation, and consider whether approaches need to differ for men and women entering custody. Establishing the predictive validity of brief universal screening approaches will likely remain difficult and the question of whether implementation of screening leads to a reduction in self-harm/suicide events should be prioritised instead.
Sociodemographic, clinical and criminal justice characteristics of the sample and chi-square comparison by gender
| Sample characteristics | Total (N = 341) | Male (n = 229) | Female (n = 112) | χ2 (df) |
|---|---|---|---|---|
| Sociodemographic | ||||
| Median age [IQR] | 32 [25–40] | 33 [24–41] | 31.5 [25–38] | |
| Aboriginal and/or Torres Strait Islander status | 90 (26.4%) | 48 (21.0%) | 42 (37.5%) | 10.59 (1)* |
| Marital Status Single Married/Defacto Divorced/Separated/Widowed |
188 (55.3%) 134 (39.4%) 18 (5.3%) |
139 (61.0%) 79 (34.6%) 10 (4.4%) |
49 (43.8%) 55 (49.1%) 8 (7.1%) |
9.09 (2)* |
| Education <year 12 | 248 (73.8%) | 171 (74.6%) | 77 (68.8%) | 1.33 (1) |
| Unemployed at time of offence | 252 (73.9%) | 153 (66.8%) | 99 (88.4%) | 18.16 (1)** |
| Clinical | ||||
| Ever treated for mental health problem | 228 (66.9%) | 136 (59.4%) | 92 (82.1%) | 17.58 (1)** |
| Current psychiatric symptom/s | 290 (85.0%) | 188 (82.1%) | 102 (91.1%) | 4.76 (1)* |
| Criminal justice | ||||
| Legal status – on remand | 223 (65.4%) | 162 (70.7%) | 74 (66.1%) | 0.11 (1) |
| First time in custody | 92 (27.0%) | 65 (28.4%) | 44 (39.3%) | 3.72 (1) |
*p < 0.05; **p < 0.001
Results of self-harm risk screening at prison entry and chi-square comparison by gender
| Screening variables | Total (N = 341) | Male (n = 229) | Female (n = 112) | χ2 (df) |
|---|---|---|---|---|
| Screening questions: historical behaviour | ||||
| History of self-harm (excluding suicide attempts) | 81 (23.8%) | 46 (20.1%) | 35 (31.3%) | 5.18 (1)** |
| Self-harm in past 12 months | 31 (9.1%) | 17 (7.4%) | 14 (12.5%) | 2.35 (1) |
| History of suicide attempt(s) | 112 (32.8%) | 70 (30.6%) | 42 (37.5%) | 1.64 (1) |
| Suicide attempt in past 12 months | 30 (8.8%) | 20 (8.7%) | 10 (8.9%) | 0.00 (1) |
| Screening questions: suicidal ideation in past week | ||||
| Thoughts that they would be better off dead | 64 (18.8%) | 36 (15.7%) | 28 (25.2%) | 4.42 (1)** |
| Thoughts that life not worth living | 64 (18.8%) | 33 (14.4%) | 31 (27.9%) | 8.94 (1)** |
| Thoughts of hurting or killing self | 17 (5.0%) | 12 (5.2%) | 5 (4.5%) | 0.10 (1) |
| Thoughts about how they might end their life (suicide plan) | 13 (3.8%) | 10 (4.4%) | 3 (2.7%) | 0.59 (1) |
| Screening scores (sum of all eight screening items) | ||||
| 0 | 182 (53.4%) | 138 (60.3%) | 44 (39.3%) | 13.30 (1)*** |
| ≥1 | 159 (46.6%) | 91 (39.7%) | 68 (60.7%) | 13.30 (1)*** |
| ≥2 | 106 (31.1%) | 59 (25.8%) | 47 (42.0%) | 9.21 (1)** |
| ≥3 | 57 (16.7%) | 37 (16.2%) | 20 (17.9%) | 0.16 (1) |
| ≥4 | 41 (12.0%) | 26 (11.4%) | 15 (13.4%) | 0.30 (1) |
| ≥5 | 25 (7.3%) | 16 (7.0%) | 9 (8.0%) | 0.12 (1) |
| ≥6 | 19 (5.6%) | 13 (5.7%) | 6 (5.4%) | 0.02 (1) |
| ≥7 | 4 (1.2%) | 2 (0.9%) | 2 (1.8%) | |
| 8 | 1 (0.3%) | 0 (0.0%) | 1 (0.9%) | |
| Variable combinations | ||||
| Any history of self-harm/suicide attempts | 132 (38.7%) | 82 (35.8%) | 50 (44.6%) | 2.48 (1) |
| Self-harm or suicide attempt in past 12 months | 51 (15.0%) | 32 (14.0%) | 19 (17.0%) | 0.53 (1) |
| Any suicidal ideation in the past week | 76 (22.3%) | 40 (17.5%) | 36 (32.1%) | 9.35 (1)** |
| Recent history and recent active ideation* | 13 (3.8%) | 9 (3.9%) | 4 (3.6%) | 0.03 (1) |
*Self-harm or suicide attempt in past 12 months, and recent thoughts of hurting or killing self or suicide plan. **p < 0.05; ***p < 0.001
Predictive validity of screening items: results of logistic regression and ROC analyses for suicide/self-harm history and suicidal ideation questions
| Screening variables | Sens. (%) | Spec. (%) | PPV (%) | NPV (%) | Adjusted OR [95% CI] | p-value | AUC [95% CI] |
|---|---|---|---|---|---|---|---|
| Screening questions: historical behaviour | |||||||
| History of self-harm (excluding suicide attempts) | 46.7 | 77.3 | 8.6 | 96.9 | 3.29 [1.14, 9.54] | 0.028 | 0.62 [0.47, 0.78] |
| Self-harm in past 12 months | 26.7 | 91.7 | 12.9 | 96.5 | 4.98 [1.41, 17.59] | 0.013 | 0.59 [0.43, 0.76] |
| History of suicide attempt(s) | 66.7 | 68.7 | 8.9 | 97.8 | 4.65 [1.54, 14.05] | 0.006 | 0.68 [0.54, 0.82] |
| Suicide attempt in past 12 months | 40.0 | 92.6 | 20.0 | 97.1 | 9.24 [2.97, 28.79] | <0.001 | 0.66 [0.50, 0.83] |
| Screening questions: suicidal ideation in past week | |||||||
| Thoughts they would be better off dead | 33.3 | 81.8 | 7.8 | 96.4 | 2.34 [0.77, 7.15] | 0.134 | 0.58 [0.42, 0.73] |
| Thoughts that life not worth living | 40.0 | 82.2 | 9.4 | 96.7 | 3.47 [1.16, 10.38] | 0.026 | 0.61 [0.45, 0.77] |
| Thoughts of hurting or killing self | 20.0 | 95.7 | 17.6 | 96.3 | 6.07 [1.51, 24.42] | 0.011 | 0.58 [0.42, 0.74] |
| Thoughts about how they might end their life (suicide plan) | 20.0 | 96.9 | 23.1 | 96.3 | 7.64 [1.85, 31.59] | 0.005 | 0.59 [0.42, 0.75] |
| Screening scores (sum of all eight screening items) | |||||||
| 0 | 26.7 | 45.4 | 2.2 | 93.1 | 0.29 [0.09, 0.94] | 0.038 | 0.36 [0.22, 0.50] |
| ≥1 | 73.3 | 54.6 | 6.9 | 97.8 | 3.44 [1.07, 11.06] | 0.038 | 0.64 [0.50, 0.78] |
| ≥2 | 66.7 | 70.6 | 9.4 | 97.9 | 5.55 [1.81, 17.05] | 0.003 | 0.69 [0.55, 0.83] |
| ≥3 | 53.3 | 85.0 | 14.0 | 97.5 | 7.79 [2.60, 2.35] | <0.001 | 0.69 [0.54, 0.85] |
| ≥4 | 40.0 | 89.3 | 14.6 | 97.0 | 6.35 [2.07, 19.45] | 0.001 | 0.65 [0.48, 0.81] |
| ≥5 | 26.7 | 93.6 | 16.0 | 96.5 | 5.88 [1.69, 20.54] | 0.005 | 0.61 [0.44, 0.77] |
| ≥6 | 26.7 | 95.4 | 21.1 | 96.6 | 8.25 [2.30, 29.63] | 0.001 | 0.61 [0.45, 0.78] |
| ≥7 | 6.7 | 99.1 | 25.0 | 95.8 | 7.13 [0.68, 74.38] | 0.100 | 0.53 [0.37, 0.69] |
| 8 | 0.0 | 99.7 | 0.0 | 95.6 | 0 | 0.49 [0.35, 0.65] | |
| Variable combinations | |||||||
| Any history self-harm/suicide attempts | 66.7 | 62.6 | 7.6 | 97.6 | 3.54 [1.18, 10.69] | 0.025 | 0.65 [0.51, 0.79] |
| Self-harm/suicide attempt in past 12 months | 53.3 | 86.8 | 15.7 | 97.6 | 8.92 [2.97, 26.83] | <0.001 | 0.70 [0.55, 0.86] |
| Any suicidal ideation in past week | 46.7 | 78.8 | 9.2 | 97.0 | 3.47 [1.21, 9.99] | 0.021 | 0.63 [0.47, 0.78] |
| Recent history and recent active ideation* | 20 | 96.9 | 23.1 | 96.3 | 8.26 [1.98, 34.39] | 0.004 | 0.56 [0.42–0.75] |
*Self-harm or suicide attempt in past 12 months, and recent thoughts of hurting/killing self or suicide plan
Comparison of predictive validity of screening items for men and women
| Screening variables | AUC – male [95% CI] | AUC – female [95% CI] | Diff. between AUCs | SE | 95% CI | Z | p-value |
|---|---|---|---|---|---|---|---|
| Screening questions: historical behaviour | |||||||
| Any history of self-harm (excluding suicide attempts) | 0.63 [0.45, 0.82] | 0.60 [0.30, 0.89] | 0.04 | 0.18 | −0.31, 0.38 | 0.20 | 0.839 |
| Self-harm in past 12 months | 0.60 [0.41, 0.80] | 0.57 [0.26, 0.87] | 0.04 | 0.19 | −0.32, 0.40 | 0.21 | 0.831 |
| Previous suicide attempt | 0.67 [0.51, 0.84] | 0.69 [0.44, 0.85] | −0.02 | 0.16 | −0.33, 0.28 | −0.13 | 0.894 |
| Suicide attempt in past 12 months | 0.69 [0.50, 0.88] | 0.58 [0.27, 0.90] | 0.11 | 0.19 | −0.26, 0.48 | 0.59 | 0.559 |
| Screening questions: suicidal ideation in past week | |||||||
| Thoughts they would be better off dead | 0.66 [0.47, 0.84] | 0.37 [0.15, 0.59] | 0.29 | 0.15 | −0.00, 0.58 | 1.95 | 0.051 |
| Thoughts that life not worth living | 0.66 [0.48, 0.85] | 0.49 [0.20, 0.77] | 0.18 | 0.17 | −0.16, 0.52 | 1.02 | 0.306 |
| Thoughts of hurting or killing self | 0.62 [0.43, 0.82] | 0.48 [0.20, 0.75] | 0.14 | 0.17 | −0.20, 0.48 | 0.81 | 0.421 |
| Thoughts about how they might end their life (suicide plan) | 0.62 [0.43, 0.82] | 0.49 [0.20, 0.77] | 0.13 | 0.18 | −0.21, 0.48 | 0.77 | 0.443 |
| Screening scores (sum of all eight screening items) | |||||||
| 0 | 0.33 [0.17, 0.49] | 0.43 [0.15, 0.70] | −0.10 | 0.16 | −0.41, 0.22 | −0.62 | 0.536 |
| ≥1 | 0.67 [0.52, 0.83] | 0.57 [0.30, 0.85] | 0.10 | 0.16 | −0.22, 0.41 | 0.62 | 0.536 |
| ≥2 | 0.70 [0.53, 0.87] | 0.67 [0.42, 0.93] | 0.03 | 0.16 | −0.28, 0.34 | 0.18 | 0.860 |
| ≥3 | 0.70 [0.52, 0.88] | 0.67 [0.36, 0.97] | 0.04 | 0.18 | −0.32, 0.39 | 0.19 | 0.846 |
| ≥4 | 0.73 [0.54, 0.91] | 0.43 [0.18, 0.68] | 0.30 | 0.16 | −0.02, 0.61 | 1.86 | 0.062 |
| ≥5 | 0.65 [0.46, 0.85] | 0.46 [0.19, 0.73] | 0.20 | 0.17 | −0.13, 0.53 | 1.16 | 0.245 |
| ≥6 | 0.66 [0.47, 0.86] | 0.47 [0.20, 0.75] | 0.19 | 0.17 | −0.15, 0.53 | 1.10 | 0.272 |
| ≥7 | 0.54 [0.36, 0.73] | 0.49 [0.21, 0.78] | 0.05 | 0.17 | −0.29, 0.39 | 0.30 | 0.763 |
| 8 | 0.50 [0.33, 0.68] | 0.49 [0.21, 0.78] | 0.01 | 0.17 | −0.33, 0.34 | 0.03 | 0.978 |
| Variable combinations | |||||||
| Any history self-harm/suicide attempts | 0.65 [0.48, 0.82] | 0.66 [0.40, 0.92] | −0.01 | 0.16 | −0.32, 0.30 | −0.07 | 0.943 |
| Self-harm/suicide attempt in past 12 months | 0.71 [0.53, 0.90] | 0.67 [0.37, 0.98] | 0.04 | 0.18 | −0.31, 0.40 | 0.23 | 0.817 |
| Any suicidal ideation in past week | 0.70 [0.52, 0.88] | 0.46 [0.18, 0.74] | 0.23 | 0.17 | −0.10, 0.57 | 1.36 | 0.172 |
| Recent history and recent active ideation* | 0.62 [0.43, 0.82] | 0.48 [0.20, 0.76] | 0.14 | 0.17 | −0.20, 0.48 | 0.81 | 0.418 |
*Self-harm or suicide attempt in past 12 months, and recent thoughts of hurting/killing self or suicide plan
References
Australian Bureau of Statistics (2021), “2016 Census QuickStats”, available at: https://quickstatswww.censusdata.abs.gov.au/census_services/getproduct/census/2016/quickstat/036 (accessed 1 November 2021).
Corben, S. and Tang, H. (2019), “NSW inmate census 2018”, NSW: Corrective Services NSW, available at: www.correctiveserviceswww.dcj.nsw.gov.au/documents/research-and-statistics/sp47-nsw-inmate-census-2018.pdf (accessed 1 November 2021).
Daigle, M.S. and Côté, G. (2006), “Nonfatal suicide-related behavior among inmates: testing for gender and type differences”, Suicide and Life-Threatening Behavior, Vol. 36 No. 6, pp. 670-681, doi: 10.1521/suli.2006.36.6.670.
Favril, L., O’Connor, R.C., Hawton, K. and Vander Laenen, F. (2020a), “Factors associated with the transition from suicidal ideation to suicide attempt in prison”, European Psychiatry, Vol. 63 No. 1, pp. 1-7, doi: 10.1192/j.eurpsy.2020.101.
Favril, L., Yu, R., Hawton, K. and Fazel, S. (2020b), “Risk factors for self-harm in prison: a systematic review and meta-analysis”, The Lancet Psychiatry, Vol. 7 No. 8, pp. 682-691, doi: 10.1016/S2215-0366(20)30190-5.
Fazel, S. and Benning, R. (2009), “Suicides in female prisoners in England and Wales, 1978–2004”, British Journal of Psychiatry, Vol. 194 No. 2, pp. 183-184, doi: 10.1192/bjp.bp.107.046490.
Fazel, S., Hayes, A.J., Bartellas, K., Clerici, M. and Trestman, R. (2016), “Mental health of prisoners: prevalence, adverse outcomes, and interventions”, The Lancet Psychiatry, Vol. 3 No. 9, pp. 871-881, doi: 10.1016/S2215-0366(16)30142-0.
Fleming, J., Gately, N. and Kraemer, S. (2012), “Creating HoPE: mental health in Western Australian maximum security prisons”, Psychiatry, Psychology and Law, Vol. 19 No. 1, pp. 60-74, doi: 10.1080/13218719.2010.543405.
Franklin, J.C., Ribeiro, J.D., Fox, K.R., Bentley, K.H., Kleiman, E.M., Huang, X., Musacchio, K.M., Jaroszewski, A.C., Chang, B.P. and Nock, M.K. (2017), “Risk factors for suicidal thoughts and behaviors: a meta-analysis of 50 years of research”, Psychological Bulletin, Vol. 143 No. 2, pp. 187-232, doi: 10.1037/bul0000084.
Gallagher, C.A. and Dobrin, A. (2005), “The association between suicide screening practices and attempts requiring emergency care in juvenile justice facilities”, Journal of the American Academy of Child & Adolescent Psychiatry, Vol. 44 No. 5, pp. 485-493, doi: 10.1097/01.chi.0000156281.07858.52.
Gottfried, E.D. and Christopher, S.C. (2017), “Mental disorders among criminal offenders: a review of the literature”, Journal of Correctional Health Care, Vol. 23 No. 3, pp. 336-346, doi: 10.1177/1078345817716180.
Gould, C., McGeorge, T. and Slade, K. (2018), “Suicide screening tools for use in incarcerated offenders: a systematic review”, Archives of Suicide Research, Vol. 22 No. 3, pp. 345-364, doi: 10.1080/13811118.2017.1334611.
Hawton, K., Linsell, L., Adeniji, T., Sariaslan, A. and Fazel, S. (2014), “Self-harm in prisons in England and Wales: an epidemiological study of prevalence, risk factors, clustering, and subsequent ‘suicide”, The Lancet, Vol. 383 No. 9923, pp. 1147-1154, doi: 10.1016/S0140-6736(13)62118-2.
Horton, M.C., Dyer, W., Tennant, A. and Wright, N.M. (2018), “Assessing the predictability of self-harm in a high-risk adult prisoner population: a prospective cohort study”, Health and Justice, Vol. 6 No. 1, pp. 1-13, doi: 10.1186/s40352-018-0076-3.
Justice Health and Forensic Mental Health Network, JHFMHN (2017), “2015 network patient health survey report”, Sydney, Australia: Author, available at: www.justicehealthwww.nsw.gov.au/publications/2015_NHPS_FINALREPORT.pdf (accessed 1 November 2021).
Korobanova, D., Spencer, S.J. and Dean, K. (2021), “Prevalence of mental health problems in men and women in an Australian prison sample: comparing psychiatric history taking and symptom screening approaches”, International Journal of Forensic Mental Health, Vol. 21 No. 1, pp. 1-17, doi: 10.1080/14999013.2021.1895379.
Large, M., Corderoy, A. and McHugh, C. (2021), “Is suicidal behaviour a stronger predictor of later suicide than suicidal ideation? A systematic review and meta-analysis”, Australian and New Zealand Journal of Psychiatry, Vol. 55 No. 3, pp. 254-267, doi: 10.1177/0004867420931161.
Large, M., Kaneson, M., Myles, N., Myles, H., Gunaratne, P. and Ryan, C. (2016), “Meta-analysis of longitudinal cohort studies of suicide risk assessment among psychiatric patients: heterogeneity in results and lack of improvement over time”, Plos One, Vol. 11 No. 6, p. e0156322, doi: 10.1371/journal.pone.0156322.
Larney, S., Topp, L., Indig, D., O’Driscoll, C. and Greenberg, D. (2012), “A cross-sectional survey of prevalence and correlates of suicidal ideation and suicide attempts among prisoners in New South Wales, Australia”, BMC Public Health, Vol. 12 No. 1, pp. 1-7, doi: 10.1186/1471-2458-12-14.
Miranda-Mendizabal, A., Castellví, P., Parés-Badell, O., Alayo, I., Almenara, J., Alonso, I., Blasco, M.J., Cebriá, A., Gabilondo, A., Gili, M., Lagares, C., Piqueras, J.A., Rodríguez-Jiménez, T., Rodríguez-Marín, J., Roca, M., Soto-Sanz, V., Vilagut, G. and Alonso, J. (2019), “Gender differences in suicidal behavior in adolescents and young adults: systematic review and meta-analysis of longitudinal studies”, International Journal of Public Health, Vol. 64 No. 2, pp. 265-283, doi: 10.1007/s00038-018-1196-1.
Naud, H. and Daigle, M.S. (2010), “Predictive validity of the suicide probability scale in a male inmate population”, Journal of Psychopathology and Behavioral Assessment, Vol. 32 No. 3, pp. 333-342, doi: 10.1007/s10862-009-9159-8.
O’Shea, L.E. and Dickens, G.L. (2015), “Predictive validity of the short-term assessment of risk and treatability (START) for aggression and self-harm in a secure mental health service: gender differences”, International Journal of Forensic Mental Health, Vol. 14 No. 2, pp. 132-146, doi: 10.1080/14999013.2015.1033112.
O’Shea, L.E., Picchioni, M.M., Mason, F.L., Sugarman, P.A. and Dickens, G.L. (2014), “Predictive validity of the HCR-20 for inpatient self-harm”, Comprehensive Psychiatry, Vol. 55 No. 8, pp. 1937-1949, doi: 10.1016/j.comppsych.2014.07.010.
Perry, A.E., Marandos, R., Coulton, S. and Johnson, M. (2010), “Screening tools assessing risk of suicide and self-harm in adult offenders: a systematic review”, International Journal of Offender Therapy and Comparative Criminology, Vol. 54 No. 5, pp. 803-828, doi: 10.1177/0306624X09359757.
Ryland, H., Gould, C., McGeorge, T., Hawton, K. and Fazel, S. (2020), “Predicting self harm in prisoners: risk factors and a prognostic model in a cohort of 542 prison entrants”, European Psychiatry, Vol. 63 No. 1, pp. 1-8, doi: 10.1192/j.eurpsy.2020.40.
Sorge, A. and Saita, E. (2021), “Assessment of suicide and self-harm risk in foreign offenders. Evaluating the use of tree-drawing test”, Mediterranean Journal of Clinical Psychology, Vol. 9 No. 3, pp. 1-18, doi: 10.13129/2282-1619/mjcp-3024.
Tyler, N., Miles, H.L., Karadag, B. and Rogers, G. (2019), “An updated picture of the mental health needs of male and female prisoners in the UK: prevalence, comorbidity, and gender differences”, Social Psychiatry and Psychiatric Epidemiology, Vol. 54 No. 9, pp. 1143-1152, doi: 10.1007/s00127-019-01690-1.
Walker, J., Illingworth, C., Canning, A., Garner, E., Woolley, J., Taylor, P. and Amos, T. (2014), “Changes in mental state associated with prison environments: a systematic review”, Acta Psychiatrica Scandinavica, Vol. 129 No. 6, pp. 427-436, doi: 10.1111/acps.12221.
Willis, M., Baker, A., Cussen, T. and Patterson, E. (2016), “Self-inflicted deaths in Australian prisons”, Trends and Issues in Crime and Criminal Justice no. 513. Canberra: Australian Institute of Criminology, available at: www.aicwww.gov.au/publications/tandi/tandi513 (accessed 1 November 2021).
Young, J.T., van Dooren, K., Borschmann, R. and Kinner, S.A. (2017), “ACT detainee health and wellbeing survey 2016: summary results”, ACT Government: Canberra.
Zhong, S., Senior, M., Yu, R., Perry, A., Hawton, K., Shaw, J. and Fazel, S. (2021), “Risk factors for suicide in prisons: a systematic review and meta-analysis”, The Lancet Public Health, Vol. 6 No. 3, pp. E164-E174, doi: 10.1016/S2468-2667(20)30233-4.
Acknowledgements
This research was supported by: National Health and Medical Research Council (NHMRC) Grant: “The Australian Centre for Research Excellence in Offender Health”. NHMRC Investigator Grant: “Improving the mental health of people in contact, or at risk of contact, with the criminal justice system”. Suicide Prevention Research Fund Innovation Grant: “Reducing self-harm and suicidality in vulnerable prisoners: piloting a structured approach to risk assessment and intervention”. Justice Health and Forensic Mental Health Network (JHFMHN) funded the participant payments and provided in-kind support of JHFMHN clinicians and researchers involved in the project.
Corresponding author
About the authors
Christie Browne is based at Justice Health and Forensic Mental Health Network, Malabar, Australia and Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia
Prabin Chemjong is based at Justice Health and Forensic Mental Health Network, Malabar, Australia; Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia and Western Sydney Local Health District, Blacktown, Australia
Daria Korobanova is based at Justice Health and Forensic Mental Health Network, Malabar, Australia; Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia and Western Sydney Local Health District, Blacktown, Australia
Seyoung Jang is based at the Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia
Natalia Yee is based at Justice Health and Forensic Mental Health Network, Malabar, Australia and Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia
Carey Marr is based at the Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia
Natasha Rae is based at Justice Health and Forensic Mental Health Network, Malabar, Australia; Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia and Western New South Wales Local Health District, Dubbo, Australia
Trevor Ma is based at Justice Health and Forensic Mental Health Network, Malabar, Australia and Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia
Sarah-Jane Spencer is based at Justice Health and Forensic Mental Health Network, Malabar, Australia, and Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia
Kimberlie Dean is based at Justice Health and Forensic Mental Health Network, Malabar, Australia and Discipline of Psychiatry and Mental Health, Faculty of Medicine and Health, University of New South Wales, Sydney, Australia