Co-occurrence of internalizing and externalizing behaviours appears to increase risk of suicidal ideation and behaviour [16, 17]. Consistent with this, we have previously noted that adolescents who engage in both NSSI and firesetting are at elevated risk [24, 25]. The current study builds on this previous work to examine factors that potentially confer this risk. We aimed to identify which psychosocial characteristics of a sub-group exhibiting both NSSI and firesetting differed across levels of suicidality. We expected indicators of psychosocial dysfunction to be greater across increasing levels of suicidality. Overall, support for this hypothesis was observed.
Suicidality and mental health difficulties
The rate of suicidal ideation (63%) and attempt (19%) among the current sample is higher than in incarcerated youth (e.g., 19.2% ideation, 8.4% attempt , and comparable to reports within adolescent psychiatric samples (e.g., 58% ideation, 29% attempt) . Our findings extend our previous work by suggesting that, in addition to a previous diagnosis of a mental health problem , adolescents engaging in both NSSI and firesetting and exhibiting suicidal tendencies are experiencing ongoing and current psychological distress. Collectively, these observations add to existing research on outcomes for multi-problem youth [13, 16] by demonstrating that co-occurring NSSI and firesetting is associated with significant psychological impairment and suicidality during adolescence. Findings highlight the importance of addressing mental health problems in suicide prevention efforts among this subgroup.
Negative life events and suicidality
Consistent with past research highlighting the role of life stressors in development of suicidal thoughts and behaviors , negative life events emerged as an indicator of suicidal ideation and attempt. Notably, interpersonal or violent victimization (i.e., experiences of serious physical or sexual assault) increased the likelihood of suicidal ideation among our sample. In addition to experiences of physical or sexual assault, adolescents attempting suicide were more likely to have been bullied, or lost a friend or family member to suicide, than those reporting suicidal thoughts in the absence of an attempt. Indeed, experiences of abuse, an inability to effectively handle interpersonal stressors such as bullying, and exposure to suicidal behavior have each been identified as “tipping points” for suicidal behavior . However, our findings extend this knowledge by suggesting that while experiences that threaten physical integrity or challenge one’s sense of safety and security (i.e., physical and sexual abuse) lead adolescents to consider ending their life, ongoing interpersonal difficulties (i.e., bullying victimization) and exposure to completed suicide in peer networks and family are factors that may prompt these individuals into action.
Psychological characteristics and suicidality
Personality characteristics related to anxiety (i.e., Behavioral Inhibition) and persistence in the pursuit of goals (i.e., drive) were differentially related to suicidality. These findings accord with prior research indicating adolescents who are sensitive to negative experiences, or those who perceive themselves as unable to pursue goals, to be at higher risk of suicidal behavior [50, 51]. Despite trends in the literature associating impulsivity-related personality traits and alcohol use with adolescent suicide [52, 53], we found no differences in these variables across levels of suicidality. This may be attributable to the measures of impulsivity and alcohol use employed in the current study. Although the Fun-Seeking subscale of the BIS/BAS is highly correlated with well-validated measures of impulsivity , other evidence suggests the scale to have greater specificity for measuring trait-like tendencies to pursue novelty and reward (e.g., sensation-seeking) . Tanner et al.  observed Fun-Seeking to predict co-occurring NSSI and firesetting, suggesting the sensation- seeking aspect of impulsivity may relate to engagement in problem behaviors, while other impulsivity-related traits (e.g., negative urgency, the tendency to act rashly when experiencing negative affect)  may be more salient in predicting suicidality among high-risk youth. Similarly, the AusAUDIT assesses the frequency and quantity of alcohol consumption rather than problematic use, thus, it is possible that while underage alcohol consumption relates to engagement in multiple problem behaviors, it is the problematic use of alcohol that elevates the risk of suicidal ideation and attempt among multi-problem youth . Additional research is required to examine these hypotheses in greater detail.
The use of maladaptive coping strategies has long been implicated in suicidal thoughts and behaviors , thus, it is interesting that in the current study a lack of problem solving, rather than the use of non-productive coping strategies (i.e., avoidance or disengagement), was related to suicidality. However, this finding is consistent with research identifying that an inability to generate solutions in the context of life stressors or psychological distress is a key deficit among individuals who have considered or attempted suicide . It must also be noted that although differences in non-productive and reference to others (i.e., use of external supports) coping did not reach statistical significance in the present study, inspection of mean scores indicated a greater use of non-productive strategies, and a lower use of reference to others coping, as level of suicidality increased. It is possible that significant differences would emerge in replication studies with larger samples. However, our current findings regarding coping beliefs (discussed below) may provide an alternative explanation for these findings.
Adolescents with a past suicide attempt reported more severe self-injury (i.e., greater resultant harm, such as requiring medical attention) than those reporting suicidal ideation only, an observation lending support to the habituation hypothesis (i.e., that repeated NSSI desensitizes….; Joiner ). Interestingly, frequency of NSSI did not differ significantly between adolescents reporting suicidal ideation and those who had attempted suicide (M
episodes = 5.8 vs. 4.8, respectively). Several researchers have found frequency of self-injury to predict suicide attempts , others failing to observe similar relationships . Given the current sample comprised youth with NSSI and firesetting, it is possible that frequency of NSSI is only predictive of suicide attempts so far as it increases acquired capability for suicide, but once this capability is established (i.e., in subgroups of adolescents engaging in multiple problem behaviors), the salience of NSSI frequency in predicting suicide diminishes. It is also possible that the more medically severe self-injury reported by adolescents with a prior suicide attempt in the current study represents ‘trialing’ of suicidal behavior (i.e., an episode of self-injury with ambiguous intent) when existing attempts to manage distressing experiences (e.g., engagement in multiple problem behaviors) are no longer effective. Research examining the role of frequency versus severity in the NSSI/suicidality nexus, which also clarifies self-injurious intent, may assist in addressing these hypotheses.
Reasons for living as protective factors against suicidality
Finally, the current study examined potential protective factors against suicidality (i.e., reasons for living) among adolescents exhibiting co-occurring NSSI and firesetting. Although it is essential to consider both risk and protective factors in order to accurately evaluate suicide risk , the majority of research efforts to date have focused on identifying risk factors for suicidality; thus, our finding that survival and coping beliefs may buffer the risk of suicide attempt among at-risk adolescents represents an important addition to existing suicide prevention literature. Notably, when considered alongside our results regarding coping style, which implicated a lack of problem-solving skills rather than use of avoidant coping strategies, the current findings suggest that an adolescent’s perception of their ability to cope with or generate solutions to problems (i.e., self-efficacy related to coping) may play a more salient role in protecting against suicidality. Consistent with this hypothesis, recent development of the Self-Efficacy to Avoid Suicidal Action scale (SEASA) , is an important step in predicting suicide attempts. Considered in this context, our current findings suggest the study of coping-related self-efficacy may add to understanding of modifiable factors to inform interventions with suicidal adolescents.
Although tentative given the small sample and the limitations noted below, a number of our findings support the habituation hypothesis  in explaining links between NSSI and suicidality, namely, (a) the high rate of suicidality among a subgroup of adolescents engaging in problem behaviors likely to involve pain and/or fear, (b) the implication of negative life events involving physical pain and fear (i.e., physical abuse, sexual assault, and bullying victimization) and (c) the increase in medical severity of NSSI observed among adolescents with a suicide attempt.
Taken together, with results of our previous work [24, 25], results of our study provide valuable insights into clinical suicide risk assessment among subgroups of adolescent self- injurers. Specifically, our findings indicate a number of commonly cited suicide risk factors—impulsivity, substance use, and maladaptive coping—may not be reliable predictors of suicidal thoughts or attempts among most at-risk adolescents. Similarly, while research examining NSSI characteristics and acquired capability for suicide has focused on frequency of self- injury , current findings indicate that NSSI severity may be better at identifying which adolescents are most likely to act on thoughts of suicide. We recommend mental health professionals enquire about seriousness of wounds following self-injury when assessing risk for suicide.
Our findings indicate that while experiences of victimization and psychological distress are observed among adolescents with suicidal ideation, ongoing interpersonal difficulties, threats to physical integrity (i.e., more severe NSSI and physical abuse), and exposure to suicide in close relationships may help differentiate adolescents more likely to act on thoughts of suicide. These factors may assist clinicians in identifying adolescents requiring a thorough risk assessment and suicide prevention plan.
Finally, this study supports the importance of incorporating resilience factors into suicide risk assessment, as well as early intervention and prevention efforts. Developing coping and problem-solving ability, and possibly more importantly, addressing beliefs regarding an individual’s ability to cope with distressing experiences (e.g., cognitive restructuring that directly targets suicidal ideation)  appear to be promising interventions to reduce suicide attempts among at-risk adolescents. While it must be noted that the aforementioned findings relate specifically to youth engaging in NSSI and firesetting, future research is encouraged to examine whether similar relationships exist among adolescents engaging in NSSI and other externalizing behaviors (e.g., violence, substance use, etc.).
The cross-sectional nature of this study precludes conclusions regarding causality. It is possible that suicidal ideation or an attempt preceded engagement in NSSI and/or firesetting; these behaviors may represent alternate expressions of psychological distress or attempts to distract from suicidal tendencies . Although most research suggests NSSI precedes suicidal behavior [3, 5], temporal analyses could elucidate the direction of relationships between problem behaviors, psychosocial variables, and suicidality. Researchers are encouraged to conduct ongoing longitudinal examination of the aforementioned findings in order to further our understanding of the relationship between co-occurring problem behaviors and suicidality.
The number of participants reporting suicidal ideation (n = 34) or attempt was too small (n = 15) to reliably conduct more complex analyses, and explore more intricate associations between factors of interest and suicidality. As noted earlier, we chose to focus on a small, select, group of young people we previously observed to be at heightened risk of suicidal thoughts and behavior, with a view to differentiating those who report suicidal thoughts and behavior from those who do not. However, the inclusion of a large number of variables within such a small sample reduces power and necessitates the use of caution in interpreting the current findings. In future, researchers may benefit from oversampling within this population in order to obtain the required power to conduct more complex statistical analyses. Further, it would be interesting to explore reasons for living in a larger sample of youth reporting suicidal ideation, but no attempt, to ascertain which might be protective factors among young people contemplating suicide.
Related to this, in order to increase our sample size we included adolescents who had only engaged in NSSI once within our sample. Previous work suggests young people who engage in NSSI at least 4 times are most likely to report adverse outcomes , consistent with proposed DSM criteria for NSSI (NSSI on at least 5 days in the last year) . Yet, while adolescents exhibiting fewer than four episodes of NSSI might be considered to engage in relatively mild NSSI, our data suggest that if they also engage in repetitive fire-setting their risk of suicidal behavior is elevated. Consequently, we cautiously suggest that assessment of behavioral issues, such as firesetting, be conducted even when only mild forms of NSSI are exhibited. Still, restricting the definition of NSSI, yet oversampling to recruit a larger total sample, and particularly a larger sample reporting suicidal thoughts and behaviors, would enable the inclusion of additional analyses, such as the relevance of the frequency and severity of both NSSI and firesetting to suicidality.
An additional limitation of the present study involved reliance on self-report assessments to measure engagement in problem behaviors and suicidality. In particular, previous studies have highlighted challenges in regards to the validity of self-reported suicidal behavior . Future research utilizing multi-informant methods might offer additional utility in examining the relationship between problem behaviors and suicidality among youth. In addition, inclusion of a more detailed measure of suicidal ideation (e.g., the Scale for Suicidal Ideation)  would enable a more nuanced examination of the presence of suicidal ideation among this subgroup, such as the distinction between passive desire and specific plans for suicide.
Although a small study with noted limitations, the present study is the first to examine suicidality among a subgroup of school-based adolescents engaging in both NSSI and firesetting. While not a high-prevalence group of young people, clinicians can be mindful of elevated suicide risk among this select sub-group, and factors which might exacerbate or mitigate this risk. Findings suggest that exposure to experiences involving pain and fear (e.g., problem behaviors, NSSI of increasing severity, and personal victimization) might underlie the relationship between NSSI and suicidality in adolescence, but further work is required to test this proposition. The role of mental health problems and self-perceived ability to cope may also be implicated in the development of suicidality among multi-problem youth. Further exploration of the nuances of these relationships would be assisted by subsequent research with larger samples, with the goal of identifying and developing suicide prevention initiatives among select subsets of self-injurers.