Introduction

Suicide is one of the leading causes of death worldwide [1], especially among adolescents of ages 12–18 [2]. Despite extensive literature identifying risk and protective factors of suicidal behaviors [3], prevalence of completed suicides among the adolescent population continues to rise, making it an ongoing public health crisis [4].

Suicide terminology is often heterogeneous in the literature; however, one standard nomenclature suggests that suicidal behavior exists on a spectrum of severity [5]. It includes suicidal ideations (SI) which describe a variety of thoughts ranging from contemplation and wishes to preoccupation with death [6]. These thoughts can manifest with varying levels of seriousness and intensity including intent which refers to the severity of the patient’s wish to terminate his or her life as well as a plan which alludes to a specific method of carrying out the act [6]. Identifying the risk factors underlying the transition from one level to the other would be vital in suicide prevention efforts, especially due to the fact that only a third of those with suicidal ideations eventually go on to make an attempt [7].

Some of the risk factors that are common to suicidal ideations and attempts among adolescents include: previous suicide attempts [4], bullying [8], mental illness [9] and chronic medical non-psychiatric illnesses [10]. On the other hand, factors that differentiate youth with suicide attempt from those with suicidal ideations include exposure to suicidal behavior in family or friends, depressive disorders, anxiety disorders, and smoking among others [11]. In the National Youth Risk Behavior surveys, among youth with suicidal ideations, those who experienced rape, physical fights, and used heroin were more likely to attempt suicide [12].

Another risk factor for suicide is the presence of antisocial behavior during adolescence. In fact, a Swedish National Cohort Study published in 2011, identified that both females and males with convictions faced elevated suicide risks, with more severe delinquency independently associated with heightened risk of suicide in young adulthood [13].

Youth who die by suicide are more likely to have a parent or sibling with severe mental illness, a parent who died early or who died by suicide and to have a mental illness themselves as compared to age- and gender-matched controls in large national registries [14].

Most classic suicide risk factors in the literature are usually related to suicidal behaviors and do not always specify the predictors of the transition from ideations to a plan or from a plan to an attempt [15]. Suicide is often hypothesized to progress in seriousness from death wishes to suicidal ideations, then to planning and finally to an attempt with an exponentially increasing risk [16]. While many studies investigated the transition from ideations to attempts, they did not specifically look at the factors underlying the transition from a suicide plan to a suicide attempt, creating a vast knowledge gap [17]. In fact, planning is a very important factor to consider in conducting suicide risk assessments as it has been frequently associated with an increased risk of suicide [3]. Among individuals who have attempted suicide, 60% had prior suicide plans [7]. Additionally, an epidemiological study of adolescents conducted in Germany in 2019 found that most (85.4%) but not all individuals eventually transitioned from a suicide plan to an attempt, often within a year [18]. Hence, prompt and timely recognition of risk among adolescents who report suicide plans is of utmost importance.

Given the aforementioned knowledge gap, we aim to investigate the different predictors of suicide attempts among adolescents with a suicide plan, using data from the National Survey on Drug Use and Health (NSDUH).

Methods

Statistical analysis

We completed our statistical analyses using the Complex Samples module in the Statistical Package for Social Sciences (SPSS) version 21. We adjusted sample weights through computing a variable (We) according to the following formula available in the survey methodology books [19]:

We conducted bivariate analyses for all variables with the dependent variable. We measured the associations using the adjusted F test, which is a variant of the Chi-Square test adjusted for complex samples. We set statistical significance at the alpha level of 5%. We also conducted Spearman’s correlation test among our main variables of interest to test for multicollinearity. We then conducted multivariate logistic regression analyses with the variables that were significant in bivariate analyses, and we adjusted the models to reach the most parsimonious model. Finally, we calculated the adjusted OR (aOR) and its corresponding 95% confidence interval (CI) for attempting suicide when having a positive self-report of engaging in antisocial behavior.

Results

Sample characteristics

Our total sample included 3,003 respondents aged 12–17. The sample is divided between 1780 suicide attempters (A) and 1223 non-attempters (NA). As shown in Table 1, there was a higher proportion of girls in the (A) group (80. 1% vs 75.7% in the NA group, p = 0.013) and adolescents in the (A) group came from families with lower income than those in the NA group (p = 0.035). The two groups did not differ on other demographic variables. Most adolescents in the (A) and (NA) groups were aged 15–17 years (69.3% in the A group and 66.1% in the NA group, p = 0.13) and were non-Hispanic white (54.6% in the A group and 58.8% in the NA group, p = 0.118).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Population characteristics.

https://doi.org/10.1371/journal.pone.0331261.t001

Although around 80% of adolescents in both groups had a major depressive episode (MDE) in the past year (p > 0.05), adolescents in the (A) group were more likely to have had a depressive episode with severe impairment than those in the (NA) group (71.5% vs 66.6%, p = 0.021). Adolescents who attempted suicide were more likely to have received treatment/counseling or medications for MDE in the past year (58.1%) than those who did not attempt suicide (45.3%) (p < 0.001).

Of those who attempted suicide, 17% had a substance use disorder versus 7.2% among those who did not attempt suicide (p < 0.001). 11.5% of the (A) group had alcohol use disorder compared to 5.0% of the (NA) group (p < 0.001) while 30% of the A group used tobacco in the previous year compared to 19.6% in the NA group (p < 0.001). Adolescents in the (A) reported being more likely to have ever been arrested than those in the NA group (p < 0.002). 3.9% in (A) group and 1.9% in NA group were on parole or probation in the past year (p = 0.041). When looking at the different violence characteristics between those who attempted suicide and those who didn’t, it is observed that 32.0% of (A) group vs 22.5% in (NA) group had one serious fight or more at school or work (p < 0.001). Few adolescents in both groups had carried a handgun (5.3% in (A) group and 4.3% in (NA) group, p = 0.386), sold illegal drugs (7.5% in (A) group and 4.0% in (NA) group, p = 0.005) or stole/tried to steal an item worth> $50 (9.2% in (A) group and 5.9% in (NA) group, p = 0.014). Moreover, of those who attempted suicide, 14.2% had attacked with intent to seriously harm once or more, while of those who didn’t attempt suicide, 9.6% did. This difference was statistically significant (p = 0.016).

Multinomial logistic regression (Table 2)

We did not find significant multicollinearity among our main variables of interest after running the Spearman’s correlation test. All categories of ASB were statistically significant in the unadjusted logistic regression model. When adjusting for age, sex, race/ethnicity and survey year, these results remained significant (Model 1). There is an incremental association with the number of reported ASB. In fact, in the second and third models, three or more reported ASB were significantly associated with increased odds of attempting suicide {(aOR 1.81 (1.11–2.96)}.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Odds Ratios from multinomial logistic regression analyses of suicide attempts.

https://doi.org/10.1371/journal.pone.0331261.t002

Discussion

Our study aimed to investigate factors that are associated with attempting suicide among adolescents with suicidal ideations and plans.

Our results show that those who attempted suicide in this subpopulation were more likely to be girls, from lower socioeconomic background, those who’ve had a major depressive episode with severe impairment and have received professional help for depression in the year preceding the assessment. In addition, tobacco use, substance and alcohol use disorders were all associated with suicide attempts. Finally, we found an incremental association between antisocial behavior and suicide attempts in this population of adolescents with suicidal ideations and plans.

In line with previous findings on predictors of suicide attempts among adolescents [20,21] our results show gender as a significant predictor among ideators with a plan, with higher rates among females than males. In general, research has shown that female adolescents have a higher life-time prevalence of suicide attempts [22,23]. Our findings also align with prior research indicating that major depressive disorder is strongly correlated with the likelihood of suicide attempts among adolescents [21,24]. More specifically, in 2006 Spann et al. suggested that depressive symptomatology characterized by hopelessness was significantly associated with attempts [25]. In addition, we found that adolescents who attempted suicide were more likely to have sought help recently. Similarly, studies have indicated elevated rates of contact between patients and healthcare professionals prior to suicidal attempts, both fatal and non-fatal [26].

In line with prior research [27,28], our findings indicate that substance use heightens the risk of suicide attempts in adolescents. A more recent meta-analysis further supports this association, showing that substance misuse—including alcohol, tobacco, cannabis, opioids, and stimulants—is significantly associated with increased suicide mortality, with particularly elevated risk observed among female adolescents. Indeed, substance use contributes to suicide attempts via multiple pathways. Acute pharmacological effects of drug intoxication may impair judgment, lower inhibitions, worsen impulse control, and affect specific neurotransmitter systems, which increase the likelihood of suicide attempts [29]. Substance use can also trigger suicidal behavior by compromising developmental tasks [27] and triggering changes in behavioral, affective, and cognitive processes leading to dysregulation of aggression and impulsivity [30].

Along the same lines, we found that violent behavior, particularly involvement in fights and attacking with intent to harm, increased the risk of suicide among adolescents with ideations and a plan. This finding aligns with John Mann’s stress-diathesis model, which underscores the significance of aggression in suicidal tendencies, particularly how disinhibition toward certain behaviors can lead to harm to oneself or others. [31]. Across numerous studies involving high-risk individuals, trait aggression has consistently emerged as a predictive factor for future suicide. A recent meta-analysis further supports this association, showing that trait aggression and impulsivity are significantly linked to suicidal behavior [32]. Additionally, research conducted in school settings indicates that adolescents contemplating suicide tend to display elevated levels of aggression, highlighting its relevance as an indicator of suicidal vulnerability [33]. Even though this understanding of increased impulsivity among suicide attempters has been demonstrated in the literature, whether through substance use or violent tendencies, ours is the first study to highlight it in a sample of adolescents with suicidal ideations and plan.

The association between antisocial behavior and suicide has been reported in previous studies. In a national survey of juvenile correctional facilities from 2000 to 2014, for instance, suicide rates were consistently 2–3 times higher for youths in custody than for those in the general population [34]. Another study found that antisocial behavior and juvenile delinquency predicted an increased risk for suicidal ideations and behaviors [16]. Our findings further supported the aforementioned association, but within a high-risk population. This association was incremental with a particularly pronounced increase in the risk of suicide attempts when three or more antisocial behaviors were present.

Some studies have investigated individual or specific antisocial behaviors. For instance, one study found that driving while intoxicated was an independent risk factor for suicidal ideations and attempts [35]. Our study aligns with this finding, revealing that having a Driving Under the Influence (DUI) charge is associated with elevated rates of suicide attempts (p = 0.005).

Moreover, our study revealed a significant association between being arrested, on parole, or on probation and an elevated risk of suicide. Existing literature confirms that contact with the justice system, at any juncture, constitutes a risk factor for suicide among delinquent youth in both community and incarcerated settings. Among incarcerated youths, prevalence rates of recent suicide attempts varied from 3.0% to 8.5%, with lifetime attempts ranging from 11.0% to 29.4% [36]. Additionally, in a sample of young African-American adults, lifetime arrests were associated with lifetime suicidal ideation [37]. Conversely, we found no association between possessing a firearm and suicide attempts. A reason for this could be that firearms are fatal means of suicide and are typically associated with a higher risk of completed suicide [38]. The sample in our study only included survivors of suicide attempts.

While it is important to note that our findings do not imply a causal relationship, gaining a nuanced understanding of the theories influencing the transition from planning to attempting suicide is crucial. In fact, recent theories on suicide align with the ‘ideation to action’ framework [17]. This framework suggests that the emergence of suicidal thoughts and the transition from thoughts to attempts are distinct processes with different risk factors. It also emphasizes the role of “suicide capability” in this transition [3] which is believed to grow by experiencing painful and provocative events, resulting in an increased ability to tolerate pain, fear, and death. One aspect of this theory is the association between substance use and the increased risk of suicide attempts demonstrated in our study. In fact, engaging in substance use may correlate with a reduced fear of death [39] due to its potentially fatal consequences. This would consequently increase an individual’s acquired capability of suicidal behavior [40]. Along the same lines, youths who engage in potentially harmful or “risk-taking” behaviors such as antisocial behaviors would be expected to have increased capabilities to attempt suicide especially in the context of suicidal ideations as they have a high ability to tolerate painful and fear-inducing events [41].

While our findings offer valuable insights into suicide predictors, they also come with several limitations. Firstly, the use of a self-report questionnaire complicates the determination of whether these adolescents have been convicted of crimes or placed in correctional facilities. It also exposes the study to recall bias and potential misreporting of certain information.

Secondly, the cross-sectional design of our study prevents the establishment of a clear time frame for the occurrence of suicide attempts. In other words, the temporality of the attempts cannot be established, nor whether the reported attempts were related to the reported plans. Moreover, it is important to note that, due to the cross-sectional nature of the analysis, the non-attempters remain a high-risk group, and a proportion may “convert” to becoming attempters in the future.

Additionally, the assessment of risk relied solely on two questions, thus hindering our ability to ascertain the accuracy of suicide risk assessment. While one question assessed lifetime suicide attempts and the other focused on suicidal ideation during the worst time period, this discrepancy introduces temporal ambiguity, making it unclear whether the plan and the attempt occurred during the same episode. Moreover, factors such as social context, premeditation, impulsivity, and the circumstances surrounding the suicide attempt itself were not evaluated. The dataset also lacked measures of childhood abuse and neglect—established risk factors for suicidal behavior—limiting our ability to control for these important confounders. While protective factors are also critical in understanding suicide risk, they were not available in the NSDUH dataset and thus could not be included in our analysis. Similarly, although other psychiatric conditions are relevant, the NSDUH provides limited diagnostic data; we focused on major depressive episodes as the most consistently assessed and available condition. Lastly, because the NSDUH excludes institutionalized youth (e.g., those who are hospitalized or incarcerated), the findings may not generalize to higher-risk populations who are overrepresented in these settings. Our study has several notable strengths. Firstly, we utilized a large, nationally representative sample, which enhances the generalizability of our findings. Secondly, in contrast to the majority of literature examining the correlation between antisocial behaviors and suicidality, our study involved a non-institutionalized sample. This approach allows for a more accurate representation of the effect of the antisocial behaviors themselves, rather than the impact of involvement in the justice system. Finally, to our knowledge, ours is the first study to investigate the predictors of the transition from suicidal ideations and planning to attempting.

Conclusion

This study aimed to identify factors among adolescents associated with transitioning from suicidal ideations with a plan to suicide attempt. Our findings revealed significant associations between suicide attempts and gender, socioeconomic status, major depressive disorder, substance use disorders, and antisocial behaviors. Particularly noteworthy was the pronounced increase in suicide risk associated with engaging in three or more antisocial behaviors. These findings underscore the importance of addressing mental health issues, substance use, and antisocial behaviors in suicide prevention efforts targeted at at-risk adolescents. Moreover, our study highlights the need for comprehensive risk assessment protocols that consider both individual and contextual factors. In conclusion, while providing valuable clinical insights, future research should explore these associations further and develop interventions targeting modifiable risk factors to prevent suicide attempts among vulnerable adolescents.