Probing dual harm and non-violent misconduct among imprisoned adult men in Northern Ireland

Purpose: This study examines the prevalence of dual harm (i.e. self-harm and violence) among imprisoned adult men in Northern Ireland, the relationship between dual harm and non-violent misconduct, while controlling for other known risk factors for misconduct, as well as how those who engage in dual harm may differ from other groups. Methods: Using the administrative records of 892 adult men, descriptive statistics assessed the prevalence of dual harm. A negative binominal regression followed by predicted margins examined the relationship between dual harm and non-violent misconduct accounting for controls. Additionally, a multinomial logistic regression was utilised to identify if those engaged in dual harm differed from others in terms of their characteristics and in-prison experiences. Results: The findings indicate that 1-in-5 adult men were engaged in dual harm, with these men accounting for 72% of all non-violent misconduct incidents examined. Dual harm was significantly related to an increased involvement in non-violent misconduct compared to other harm histories (self-harm only, violence only, or no harm) even when other known risk factors were considered. Those engaged in dual harm were also discovered to possess a number of characteristics that differ significantly from other groups. Conclusion: These findings strengthen emerging research indicating those who engage in dual harm are a distinct group that can be challenging to manage due to their increased involvement in misconduct and their multiple needs, which existing services and supports may be ill suited to address. K E Y WO R D S corrections, dual harm, misconduct, Northern Ireland, prison


INTRODUCTION
Dual harm involves a coexisting history of violence towards others and self-harm (Slade et al., 2020).Studies suggests that dual harm is especially prevalent in prison (Slade, 2018(Slade, , 2019;;Slade et al., 2020).There is a nascent body of literature indicating those who dual harm in prison are a distinct group requiring specialist provision and are challenging to manage due to their increased involvement in misconduct (Kottler et al., 2018;Slade, 2019;Slade et al., 2020).Misconduct refers to behaviour that violates prison rules (DeLisi, 2003;Trulson et al., 2010).Research suggests that people in prison who engage in dual harm commit more misconduct than other groups, leading Slade et al. (2020, p. 193) to conclude that people who dual harm contribute 'greatly to the instability of the custodial setting'.However, research on dual harm in prison is limited and often does not control for known risk factors for misconduct.This study seeks to enhance our understanding of the relationship between dual harm and misconduct in three ways.Firstly, it examines the relationship between dual harm and non-violent misconduct, controlling for other known risk factors for misconduct.It focusses on non-violent misconduct as violent misconduct is used to identify those who dual harm while imprisoned.Secondly, it investigates the generalisability of this nascent research to Northern Ireland (NI).Thirdly, it addresses gaps in existing studies that are limited in their ability to assess how medical, conviction, and imprisonment histories may relate to dual harm.

Harm history and misconduct
Compared to those who do not self-harm, research indicates that people who self-harm are more likely to report heightened emotional reactivity, difficulties regulating emotions, more rumination, and greater use of strategies to regulate emotions, which fail to address the underlying issues (Dawkins et al., 2019;Glenn et al., 2011;Gratz & Roemer, 2008;Williams & Hasking, 2010).Studies suggest that self-harm can be used as a coping mechanism to distract from a tendency to ruminate on negative emotional experiences and can serve interpersonal functions (e.g.communicating distress, establishing a boundary) (Ehring & Ehlers, 2014;Gardner et al., 2016).Research also highlights that, alongside emotional processes, individual cognitions surrounding the use of self-harm, and the outcome expectancies associated with self-harm, can influence this behaviour (Dawkins et al., 2019).Unfortunately, self-harm is especially prevalent in prison, posing challenges for the wellbeing of those imprisoned and how this should be addressed (Hawton et al., 2014).
Research on self-harm in prison reveals that engagement in self-harm can be used to regulate emotions and/or distract from negative emotional experiences (Power et al., 2016).However, the nature of the prison environment is also believed to play a key role (Clemente-Faustino & de Guzman, 2022;Favril et al., 2020).After conducting a meta-analysis of the risk factors for self-harm in prison, Favril et al. (2020: 688-689) conclude both individual and environmental factors influence this behaviour, such that "prisoners might import a vulnerability for self-harm into prison […] that might interact with custody-specific stressors […] and thereby increase the likelihood of self-harming".Indeed, studies indicate that staff-prisoner relationships, staff perceptions, attitudes and knowledge about self-harm, as well as the availability of services and training can influence the quality of care received and amount of distressed experienced (Ekanem & Woods, 2022;Hewson et al., 2022).Consequently, while people may import risk factors for self-harm into prison, the nature of imprisonment and limitations in the ability of prisons to cope with this behaviour may significantly worsen this behaviour and the psychological distress experienced.
Studies have found that people who self-harm are at an increased risk of engaging in violence (O'Donnell et al., 2015;Richmond-Rakerd et al., 2019;Sahlin et al., 2017).Similar to self-harm in prison, involvement in prison violence has been linked to risk factors that individuals import into prison, feeling unsafe, the deprivations associated with imprisonment, as well as oppressive/unfair/unjust regimes (Bierie, 2013;Butler, 2008;Day et al., 2015;Edgar et al., 2003;Sykes, 1958).In this way, violence in prison is theorised as occurring when contextual factors trigger strain and when cognitions, past experiences, and beliefs lead people to view violence as an appropriate response to this strain (Blevins et al., 2010).
In theorising engagement in dual harm, Sahlin et al. (2017) propose that the bidirectional relationship witnessed between self-harm and violence may not reflect a causal unidirectional relationship, whereby one behaviour leads to another, but rather that these behaviours develop from common vulnerabilities.Shafti et al. (2021) propose that self-harm and violence share common biological and adverse environmental risk factors that can contribute to people developing a personality style/schema, predisposing them to engaging in these behaviours.Being placed in distressing situations may trigger harmful behaviour, with the particular behaviour engaged in depending on the emotional and functional purposes it is intended to serve and if past experience reinforces using that behaviour in that situation (Shafti et al., 2021).Accordingly, people who share the common biological and adverse environmental risk factors for self-harm and violence may engage in violence or self-harm 'interchangeably to serve the same purpose (e.g.emotional regulation or interpersonal goals)' with the specific behaviour chosen dependent on the context, situation, their cognitions and outcome expectancies (Shafti et al., 2021, p. 11).
This theory is supported by Pickering et al. (2022) who found men engaging in dual harm within prison experienced early adverse environments and trauma, with these experiences being interlinked with difficult and unpredictable environments, incoherent identities and negative emotional states.They argue that whether the men engaged in self-harm or violence was mediated by their immediate circumstances, opportunity and the extent pain distracted them from their negative emotional states (Pickering et al., 2022).Moreover, the men reported experiencing a dissonance between the identities they seek to project and the 'bad' identities others attempt to project onto them, with violence or self-harm being adopted depending on if they fought against this 'bad' identity or succumbed to it.Past research on prison violence has recognised that people can use violence to protect their self-identity from perceived threats/challenges (Butler, 2008).Yet, prisons generally respond to violence and self-harm using separate processes, with violence towards others often addressed with punishment and containment, while self-harm tends to elicit a greater emphasis on treatment and compassion (Slade, 2018(Slade, , 2019;;Slade et al., 2020).The separation of these processes has been argued to hinder efforts to provide a holistic understanding of this behaviour and hamper attempts to address the needs of those involved (Slade, 2018(Slade, , 2019;;Slade et al., 2020).
Common risk factors for dual harm compared to self-harm only, violence only, or no harm include having more negative childhood experiences, impulsivity, poor self-control, emotional dysregulation, genes related to dysfunctional serotongergic systems, and higher rates of substance dependence and psychotic symptoms (Durham, 2021;Richmond-Rakerd et al., 2019;Spaan et al., 2022;Wang et al., 2022).Research also suggests that people who dual harm are more likely to experience anger, psychopathology, as well as difficulties regulating emotions and behaviours (Durham, 2021;Spaan et al., 2022;Wang et al., 2022).These characteristics may help explain why people who dual harm may be found guilty of committing more misconduct than other groups in prison.For instance, if people who dual harm experience poorer self-control, higher rates of substance dependence, more impulsivity, anger and emotional dysregulation, as well as greater difficulties regulating their behaviour, they may struggle to conform to prison rules and be more likely to be caught engaging in misconduct.
Research has identified a number of risk factors for misconduct, including individual characteristics and experiences people import into prison, as well as prison conditions and the nature of the prison regime (Steiner et al., 2014).For instance, age, race/ethnicity, deprivation, history of imprisonment, convictions, substance use, head injury/epilepsy and impairments have been identified as possible risk factors for misconduct (Butler et al., 2021a;Steiner et al., 2014).Other risk factors include the submission of prison complaints, prison visitation, first experience of imprisonment, and using drugs (Butler et al., 2021a(Butler et al., , 2021b;;Steiner et al., 2014).Past studies indicate that a small number of people can be responsible for a disproportionate amount of misconduct (Butler, Caudill, et al., 2021;DeLisi, 2003;Trulson et al., 2010).Slade et al. (2020) found that the 11% of men who engaged in dual harm in prison accounted for 56% of misconduct incidents examined, arguing this over-representation may be due to difficulties experienced in self-regulating reactive impulsive responses to distressing situations.Estimated prevalence of dual harm in prison ranges from roughly 3% among females to 29% among males, with variations evident depending on the timeframes used (Kottler et al., 2018;Slade, 2018;Slade et al., 2020).Differences are also evident via gender, with imprisoned female dual harmers being more likely to be younger, British, and have more convictions and periods of imprisonment compared to those who self-harmed only, while being black was associated with a lower probability of engaging in dual harm compared to violence only (Kottler et al., 2018).For men, no difference was observed on age or conviction history, although those who dual harmed on average spent more time in prison and were younger than those who had not engaged in any self-harm or violence (Slade et al., 2020).
Significant gaps in knowledge remain, however, due to few studies being conducted on dual harm in prison as it is a recently proposed concept (Wang et al., 2022).For instance, little is known about how people who engage in dual harm may differ from others on their in-prison experiences and medical histories or how generalisable existing studies are to different cultural contexts.This is despite in-prison experiences and medical histories being a known risk factor for misconduct (Steiner et al., 2014).Most studies on dual harm in prison have been conducted in English and Wales medium security prisons (Kottler et al., 2018;Pickering et al., 2022;Slade, 2018;Slade et al., 2020).Yet, a study conducted in China suggests that cultural variations can exist (Wang et al., 2022).

The present study
This study addresses these gaps in knowledge by examining the prevalence of dual harm among imprisoned adult men in NI and if those engaged in dual harm are responsible for a disproportionate amount of non-violent misconduct.Moreover, if those engaged in dual harm differ from others in terms of their demographics, medical history, convictions and in-prison experiences is examined, before moving on to assess if dual harm is related to non-violent misconduct when accounting for other known risk factors for misconduct.The study focuses on non-violent misconduct as violent misconduct is used to identify engagement in dual harm while imprisoned.In this way, the study advances our knowledge of dual harm and misconduct in three ways.It investigates if a relationship continues to exist between dual harm and non-violent misconduct when controlling for known risk factors for misconduct.Secondly, it assesses the generalisability of known risk factors for dual harm in prison to NI. Thirdly, it enhances our knowledge of how involvement in dual harm may vary depending on medical history, convictions and in-prison experiences.

Data
The study utilised administrative data derived from the Northern Ireland Prison Service (NIPS) on all imprisoned men (n = 892) detained in Maghaberry prison on 22 November 2017.Maghaberry is one of two adult male prisons in NI and is the largest and sole high-security prison, holding both men who have been sentenced and remanded into custody.The sample represented about 69% of the total average adult male daily prison population (NISRA, 2018).The administrative data contained information on demographic characteristics, health and convictions, and recorded events within prison for their entirety of their imprisonment in NIPS.Therefore, while the sample was taken from Maghaberry prison, the administrative data contained information on their in-prison behaviour and experiences throughout their time in all NIPS prisons up until 22 November 2017.While administrative data can underestimate the true prevalence of misconduct and self-harm due to under-reporting and different recording and staff practices, administrative data has been found to be a useful and valid measure of these behaviours (Borschmann et al., 2017;Steiner & Wooldredge, 2014).Ethical approval for the research was obtained from Queen's University Belfast and NIPS, with NIPS providing access to the anonymised administrative data set.

Measures
The administrative data contained information on harm history, age, race/ethnicity, nationality, religion, neighbourhood deprivation, medical history, convictions, submission of prison complaints, visitation, prison drug tests taken and passed, periods of incarceration, days spent imprisoned and involvement in misconduct.
Harm history was determined by examining the extent to which participants had been found guilty of engaging in violent misconduct (i.e.behaving violent towards others) and had been referred under the Supporting Prisoners at Risk (SPAR) process.People were referred under the SPAR process if they engaged in serious self-harm, attempted to take their own life or if staff were concerned about the likelihood to seriously self-harm (Sudgen, 2016).Four categories were created.A 'No Harm' group consisting of men with no SPAR referrals and no record of violent misconduct (44.3%).A 'Self-Harm only' group who had no record of violent misconduct but a history of SPAR referrals (27.0%).A 'Violence only' group consisting of those found guilty of violent misconduct but no history of SPAR referrals (8.4%).Finally, a 'Dual harm' group who had both a record of violent misconduct and SPAR referrals (20.3%).
Age was assessed in years while race/ethnicity, nationality, and religion were categorised into groups reflective of NI socio-political importance.Race/ethnicity was measured as 'White, excluding Irish Travellers' (93%) or 'Non-White, including Irish Traveller' (7%) as Irish Travellers are recognised as a distinct racial group under the Race Relations (NI) Order 1997.Nationality had four groups: Northern Irish (60.8%),Irish (11.9%),British (15.3%), and other (12.1%).Religion had three categories: Catholic (52.2%),Protestant (34.3%), and other (13.5%).Neighbourhood deprivation was calculated by matching postcodes to NI Census ranking on this from 1 to 890 areas (NISRA, 2017).Deprivation was recoded to have higher rankings equate to greater deprivation levels (M = 609.1,SD = 220.9).Deprivation was missing for a small group with no NI postcode reported (due to no fixed abode, address missing, or living outside NI).For these cases, the mean deprivation was imputed and indicated with a missing flag variable with reason missing.
Medical and conviction histories are non-mutually exclusive dummy variables.The medical variables considered whether the men had self-declared a medical history of the following conditions: Head Injury/Epilepsy (15.7%);Impairments (11.1%); and Addiction (55.3%).Medical history was self-reported by the men and not verified by prison staff.Convictions were collapsed into four categories (violence (85.5%), property (52.5%), drugs (28.0%) and other offences (36.8%)) with the dummy variable indicating whether the men had a conviction for that type of offence.Measures of in-prison experiences included: number of prison complaints submitted; if prison drug tests had been taken and the proportion passed; number of prison visits attended; number of times incarcerated in NIPS; total days spent imprisoned in NIPS; amount of misconduct engaged in; and nature of that misconduct.

Analytical strategy
Descriptive statistics, negative binominal regression and multinomial logistic regression were used to address the study aims.Descriptive statistics examined prevalence of dual harm and if those engaged in dual harm were responsible for a disproportionate amount of non-violent misconduct.Non-violent misconduct included all other types of misconduct excluding violence due to violent misconduct being used to determine harm history.The negative binominal regression investigated if harm history was related to non-violent misconduct when controlling for other known risk factors for misconduct.The multinomial logistic regression identified if those engaged in dual harm differed from others in terms of their demographics, medical history, convictions and in-prison experiences.

FINDINGS
Table 1 presents the sample characteristics.
While most men (44.3%) were in the 'No Harm' group, 27.0% were in the 'Self-Harm only' group, 20.3% in the 'Dual Harm' group and 8.4% were in the 'Violence only' group.This indicates that 1-in-5 of the men had engaged in dual harm.Furthermore, these men were responsible for a disproportionate amount of non-violent misconduct, with the 20.3% of men engaged in dual harm accounting for 72.0% of non-violent misconduct.
A negative binominal regression was used to examine if harm history was related to non-violent misconduct, controlling for other known risk factors for misconduct.Variations in days spent imprisoned was controlled for using the exposure command in Stata, which accounts for varying time at risk but does not produce a coefficient for days spent imprisoned in the output.Table 2 shows dual harm is significantly related to involvement in non-violent misconduct net of controls.
Compared to the 'Dual Harm' group, the 'No Harm' group (B = −1.234,IRR = 0.291, p < .001), the 'Self-Harm only' (B = −0.862,IRR = 0.423, p < .001)and the 'Violence only' group (B = −0.456,IRR = 0.634, p = .001)all had a lower risk of non-violent misconduct.To illustrate this relationship, a predicted count was executed (using Stata margins command), which estimates non-violent misconduct count for harm history groups, while holding all controls at the mean.The results shown in Figure 1 reveal that on average the 'Dual Harm' group had 7.3 non-violent misconduct incidents, while the 'No Harm' had 2.1, the 'Self-Harm only' group had 3.1 and the 'Violence only' group had 4.6.
Table 2 further shows that age was significantly associated with non-violent misconduct, with younger men being at a greater risk of non-violent misconduct (B = −0.048,IRR = 0.953, p < .001).Compared to those reporting a Northern Irish nationality, 'other' nationalities were less at risk of non-violent misconduct (B = −0.512,IRR = 0.599, p = .022).In contrast to those with an NI address, men whose address was not known or had an address outside of NI were at an increased risk of non-violent misconduct (B = 0.487, IRR = 1.627, p = .007and B = 0.701, IRR = 2.016, p = .020,respectively).Men with a history of impairments also had an increased risk of non-violent misconduct compared to those without this history (B = 0.334, IRR = 1.397, p = .009),as well as those reporting a history of addiction (B = 0.243, IRR = 1.276, p = .012).Similarly, those with a conviction for property offences were an increased risk of non-violent misconduct compared to those without such a history (B = 0.360, IRR = 1.434, p < .001).On the other hand, more visits and a higher proportion of prison drug tests passed were associated with a reduced risk of non-violent misconduct (B = −0.001,IRR = 0.999, p = .050and B = −2.036,IRR = 0.131, p < .001,respectively).Compared to having taken a prison drugs test, having not yet taken one was associated with an increased risk of non-violent misconduct (B = 0.939, IRR = 2.558, p < .001).Lastly, more incarcerations were associated with an increased risk of non-violent misconduct (B = 0.020, IRR = 1.020, p = .009).
To shed light on whether those engaged in dual harm differed from others in terms of their demographics, medical history, convictions and in-prison experiences, a multinomial logistic regression was conducted comparing those in the 'Dual Harm' group to the other harm history groups (see Table 3).
In comparison to the 'Dual Harm' group age, number of incarcerations and the number of days spent imprisoned was statistically significant across all groups.In other words, older men were more likely to be in the 'Self-Harm only' group (B = 0.084, RRR = 1.088, p < .001),'Violence only' group (B = 0.072, RRR = 1.072, p = .001),and the 'No Harm' group (B = 0.106, RRR = 1.112, p < .001)than the 'Dual Harm' group.Those with more incarcerations were less likely to be in the 'Self-Harm only' group (B = −0.129,RRR = 0.879, p < .001),'Violence only' group (B = −0.095,RRR = 0.910, p = .008)or 'No Harm' group (B = −0.169,RRR = 0.845, p < .001)compared to the 'Dual Harm' group.Similarly, in comparison to the 'Dual Harm' group, the other harm history groups were associated with spending fewer days imprisoned (see Table 3).
Additionally, those with higher proportion of prison drug tests passed had a higher risk of being in both the 'Self-harm only' group (B = 1.478,RRR = 4.385, p = .002)and 'No Harm' group (B = 1.696,RRR = 5.452, p < .001)compared to the 'Dual Harm' group, with no statistical differences with the 'Violence only' group.No other characteristics in the model reached statistical differences with the 'Self-harm' group compared to the 'Dual Harm' group.Furthermore, those with impairments had a lower risk of being in the 'Violence only' group (B = −1.738,RRR = 0.176, p = .028)and 'No Harm' group (B = −0.947,RRR = 0.388, p = .020)compared to the 'Dual Harm' group.Visitation was uniquely associated with the 'Violence only' group, with those receiving more prison visits being more likely to be in the 'Violence only' group compared to the 'Dual Harm' group (B = 0.004, RRR = 1.004, p = .034).

% Mean (SD) Min Max
Harm Several other characteristics uniquely differed from the 'No Harm' group compared to the 'Dual Harm' group.Those with an 'other' nationality compared to a Northern Irish nationality were more likely to be in the 'No Harm' group than the 'Dual Harm' group (B = 1.922,RRR = 6.834, p = .020).Neighbourhood deprivation was also associated with no harm, with men from more deprived neighbourhoods being more likely to be in the 'No Harm' group than the 'Dual Harm' group (B = 0.001, RRR = 1.001, p = .044).Individuals with no fixed abode and address not known (compared to those with an NI address) were also less likely to be in the 'No Harm' group than the 'Dual Harm' group (B = −0.756,RRR = 0.470, p = .043and B = −1.387,RRR = 0.250, p = .027,respectively).Lastly, those convicted of violent and property offences had a lower risk of being in the 'No Harm' group compared to the 'Dual Harm' group (see Table 3).

DISCUSSION
This study establishes an estimated prevalence of 20.3% of men in NI prisons had engaged in dual harm.Moreover, the 20.3% of men who engaged in dual harmed accounted for a disproportionate amount of non-violent misconduct (72.0%), which strengthens the nascent literature suggesting those who dual harm in prison are a distinct group who can be challenging to manage due to their increased involvement in misconduct.These findings further validated the emerging literature demonstrating a relationship between dual harm and misconduct, while substantiating and strengthening the generalisability of this evidence base by demonstrating these patterns extend to the cultural context of NI and a high-security prison.It also bolstered our understanding by verifying the association between dual harm and misconduct remained significant when considering other known risk factors for misconduct, addressing a gap in previous studies (Kottler et al., 2018;Slade, 2018;Slade et al., 2020).Similar to previous research, the findings indicate that people who engage in dual harm in prison tend to be younger, spend more days in prison and are imprisoned more often (Kottler et al., 2018;Slade et al., 2020).On the other hand, increased visitation was associated with an increased likelihood of being in the 'Violence only' group compared to the 'Dual Harm' group.This may reflect the unique cultural context of conflict in NI, with people imprisoned for politically motivated violence having had access to extra visitation (Butler, 2020;McEvoy, 2001).
This study also extends our knowledge by highlighting how those in the 'Dual Harm' group differed significantly from the 'No Harm' group on nationality, neighbourhood deprivation, address, impairments, convictions for violent and property offences, as well as the proportion of prison drug tests passed.Excluding history of impairments and the proportion of prison drug tests passed, the rest of the varia- Houston & Butler, 2019).The finding that those who passed a higher proportion of prison drug tests were more likely to be in the 'Self-Harm only' and 'No Harm' groups than the 'Dual Harm' group also corresponds to past studies linking dual harm and substance use in the general population (Durham, 2021;Richmond-Rakerd et al., 2019;Spaan et al., 2022;Wang et al., 2022).The present study then confirms the generalisability of some of the risk factors for dual harm in prison identified in previous research to NI, as well as the relevance of some of the risk factors associated with dual harm among the general population to the prison context.It also expands our knowledge by identifying new risk factors for dual harm in prison that have not been considered by previous studies (e.g.neighbourhood deprivation, address and impairments).Taken together, these findings indicate that those engaged in dual harm in NI prison are younger, may experience housing instability and impairments, are incarcerated more often as well as engage in more drug use and non-violent misconduct.The finding that those engaged in dual harm are more frequently imprisoned, fail more prison drug tests and amass more non-violent misconduct than others also suggests that these men may be caught in a cycle of behaviour, which existing services and supports struggle to address.As highlighted by Slade (2018Slade ( , 2019) ) and Slade et al. (2020), violence and self-harm in prison tend to be addressed using separate processes and procedures, with this schism compounded by the allocation of responsibility for policy and practice divided across government departments of health and justice.While a punishment and containment approach is frequently used to address misconduct and violence towards others, self-harm often prompts treatment and compassion (Slade, 2018(Slade, , 2019;;Slade et al., 2020).Yet, a punishment and containment approach tends to rely on segregation, which can increase self-harm (Kaba et al., 2014;Lanes, 2009).Consequently, using separate processes and procedures to address these behaviours may not only fail to meet the needs of those involved but may risk worsening self-harm.Another concern is while rehabilitative services and supports are available in prison, access to these programmes depend on eligibility criteria and perceptions of manageability (Bosma et al., 2018;Butler et al., 2021a).Bosma et al. (2018) argue that those who are more challenging to manage may be less likely to get access to these programmes.Furthermore, eligibility criteria for rehabilitation programmes can exclude people with co-occurring mental health and substance use issues, or impairments (PBNI, 2016).Consequently, existing services may not be able to meet the needs of those who dual harm, demonstrating the need for a more joined-up approach, offering specialist services, and supports, which can address co-occurring mental health and substance use issues, impairments and challenging behaviour, as well as reduce the situational factors that can worsen these behaviours.As misconduct has been linked to reconviction (Cochran et al., 2014;DeLisi et al., 2020;Mooney & Daffern, 2015), providing specialist support to those who dual harm may not only help to improve order and stability of custodial establishments but also reduce reconviction upon release.
A study limitation of this research is that it only examined men imprisoned in NI; therefore, its generalisability to other groups and cultural contexts may be limited.The time-frame used to assess 'Dual Harm' group membership was lifetime involvement in violent misconduct and SPAR referrals as of 22 November 2017, which might vary from other timeframes used.Also, the study drew on official figures, which are known to underestimate violence and misconduct in prison (Steiner & Wooldredge, 2014).Similarly, reliance on SPAR referrals to assess self-harm may have underestimated self-harm, as only serious cases would receive a referral (Borschmann et al., 2017).Another limitation is that this study was cross-sectional in nature, limiting the ability to draw temporal or causal conclusions about the nature of the relationships observed.Future research on this area should seek to use a longitudinal research design and self-report measures.
Nevertheless, the strengths of this research include the use of administrative data that covered a number of variables not previously studied in regards to dual harm in prison.Furthermore, the study drew from a large sample that included men on sentence and remand drawn from across the different security classifications.Moreover, this is the first study to provide a profile of those who dual harm in NI and engaged in misconduct.In conclusion, this research adds to the literature arguing those who dual harm is a unique group with distinct characteristics and behaviours who are at higher risk of engaging in misconduct.While some similarities exist between this group and those who exclusively engage in Sample descriptives n = 892 20448333, 0, Downloaded from https://bpspsychub.onlinelibrary.wiley.com/doi/10.1111/lcrp.12234 by Test, Wiley Online Library on [23/11/2022].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Not mutually exclusive -Individuals can have multiple medical conditions or multiple types of offences.Negative binominal regression for non-violent misconduct n = 892 20448333, 0, Downloaded from https://bpspsychub.onlinelibrary.wiley.com/doi/10.1111/lcrp.12234 by Test, Wiley Online Library on [23/11/2022].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License a T A B L E 2

harm only vs. dual harm Violent only vs. dual harm No harm vs. dual harm
(Bryan, 2004;om https://bpspsychub.onlinelibrary.wiley.com/doi/10.1111/lcrp.12234byTest,WileyOnlineLibrary on [23/11/2022].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)onWileyOnline Library for rules of use; OA articles are governed by the applicable Creative Commons License bles were uniquely associated with being in the 'Dual Harm' group.Those reporting a history of impairments were less likely to be in the 'Violence only' group and the 'No Harm' compared to the 'Dual Harm' group.Past studies have argued that the limitations of criminal justice organisations to ensure those with impairments (e.g.speech, language, communication and learning difficulties) are understood and can express themselves can lead them to feel frustrated and lash out at themselves and others(Bryan, 2004;Not mutually exclusive -Individuals can have multiple medical conditions or multiple types of offences.Multinomial logistic regression on harm history n = 892 20448333, 0, Downloaded from https://bpspsychub.onlinelibrary.wiley.com/doi/10.1111/lcrp.12234 by Test, Wiley Online Library on [23/11/2022].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License a T A B L E 3