Externalizing behaviors/violence exposure and suicide among US adolescents

Abstract

Introduction:

This study examines associations between externalizing behaviors/violence exposure and suicidal behavior among US high school students from 1991–2021

Methods:

Data for this cross-sectional study were drawn from the Youth Risk Behavior Survey and the total sample contained data on 234,588 adolescents. Logistic regression models were used to assess the relationship between externalizing behaviors/violence exposure and suicidal behavior. To assess trends over time, models were then assessed for multiplicative interactions between externalizing behaviors/violence exposure and time by sex. State-level trends were also assessed. All analyses were conducted in 2024

Results:

The prevalence of externalizing behaviors/violence exposure increased among youth with an injurious suicide attempt. Logistic regression models indicated statistically significant associations across suicidal behaviors with a higher magnitude of association observed among those with an injurious suicide attempt. For instance, compared to those who did not carry a gun, those who carried a gun had 6.32 (95% CI: 4.78, 8.36) times the odds of injurious suicide attempt vs. no attempt and 2.66 (95% CI: 2.00, 3.53) times the odds of non-injurious suicide attempt vs. no attempt. Stronger associations arose among male individuals. Among those with an injurious suicide attempt in 2021, state-level differences in weapon access emerged

Conclusions:

Adolescents with externalizing behaviors/violence exposure are at an increased risk of an injurious suicide attempt. The relationship varies over time and by sex. Culturally adaptive and structurally competent approaches to mental health and mechanisms to identify at-risk youth are imperative.

Introduction

Suicide remains a leading cause of death among adolescents (individuals aged 10 to 19) in the US^1,2 and is the second leading cause of death among individuals ages 15 to 19 years old.³ Suicidal behaviors include attempting suicide (a non-fatal, self-directed, potentially harmful behavior with intent to die),⁴ and ideation (thoughts, consideration or plan of self-harm or ending one’s life).⁵ Suicide attempts may be injurious (injurious behavior with an intent to die) or non-injurious (intentional self-harm does not result in an injury). In 2019, approximately one in five 9–12^th grade students seriously considered attempting suicide, one in eleven made an attempt, and one in forty made a suicide attempt requiring medical treatment.⁶ Additionally, over the last two decades, rates of suicidal behavior and fatal suicide death have increased. Among US individuals aged 10–24, suicide rates increased from 6.8 to 11.0 deaths per 100,000 between 2007 and 2021.⁷ Female adolescents have evidenced more rapid increases in suicide rates than males,^2,8 indicating a need for public health approaches to reduce risk. Although the prevalence of suicide attempts requiring medical treatment (in this paper referred as “injurious suicide attempt” [ISA]) is lower than attempts not requiring medical treatment,⁶ less is known on specific risk factors that lead to an injurious attempt.

Firearm-based suicide rates have been increasing among adolescents⁹ in recent years suggesting a growing urgency in examining dynamics surrounding firearms and their role pertaining to suicidal behavior. Access to lethal weapons is a strong predictor of suicide and case fatality among those who attempt. Indeed, firearms are among the most lethal methods of suicide death among individuals aged 10–24 years.¹⁰ In 2020, suicide deaths accounted for over half of all firearm related deaths in the US. Although access to firearms and other weapon use contributes to suicide risk, understanding the underlying risk for, and correlates of, firearm use among adolescents is understudied.

Carrying and using firearms and other weapons is situated within adolescents’ geographic and social context, and these behaviors are often associated with a broader constellation of externalizing behaviors/violence exposure. Weapon use,¹¹ feeling unsafe,¹² being threatened with a weapon,¹³ or in a physical fight¹⁴ and experiencing sexual assault^14,15 have all been associated with increased suicide risk. Yet, violence and suicide are often considered separately, despite their correlation. If adolescents carry weapons and use them during physical altercations, it may be the case that they are available for use during suicidal crises as well. Exposure to violence in the community and in school increases the risk of internalizing symptoms, and suicidal behaviors.^16–19

Access to firearms and suicide rates in the US vary by state. Suicide deaths are more prevalent among individuals who live in rural areas due in part to firearm access, isolation and limited access to mental health services.²⁰ State-level variations of suicide rates are heavily driven by suicide deaths involving firearms.²¹ For instance, in 2020, the suicide rate involving firearms was highest in states with the fewest gun laws (10.8 per 100,000) and lowest in states with the most gun laws (4.9 per 100,000).²⁰ Furthermore, the suicide rate involving non-firearm means was lower than the rate involving firearms except in states with the most gun laws. Suicide rates involving firearms are much higher than rates involving other means, with starker differences by state and firearm law provisions. Thus, understanding recent state-level variations in weapon use among US adolescents with a history of suicide attempt is imperative to inform where risk is most concentrated in the US.

Many mechanisms connect externalizing behaviors/violence exposure to suicide attempts. Adolescents with externalizing symptoms, including those who carry and use weapons often have higher rates of internalizing symptoms such as depression and anxiety,²² which are well documented risk factors for suicide attempts.²² Exposure to violence through mechanisms such as feeling unsafe and being threatened with a weapon is associated with increased internalizing symptoms among adolescents²³. Bullying victimization may also result in weapon carrying and altercations²⁴ while also amplifying suicide risk, especially among adolescents who lack social support²⁵ and believe that they need to protect and defend themselves.²⁴ Moreover, adolescents with bullying victimization have 3 times the odds of suicidal behaviors compared to those without.^26,27 Thus, the increased risk that externalizing behaviors and experiencing violence pose to suicide attempts among adolescents warrants further investigation.

The present study examines the associations between externalizing behaviors/violence exposure, firearm carrying, and suicidal behavior to provide an assessment of the ways in which broader externalizing behaviors/violence exposure may increase the risk for self-harm. Given the noted impact of access to lethal means on suicide risk, the association between firearm carrying and injurious self-harm was of particular interest. This study centers on high school students from 1991–2021. There were three central aims. The first aim was to assess whether there were increases in the prevalence of externalizing behaviors/violence exposure across time by suicide attempt. The second aim was to test associations between externalizing behaviors/violence exposure and suicidal behaviors by sex and year, including whether these associations varied. The last aim was to understand state-level variations in weapon access among youth with an ISA in recent years.

Methods

Data were drawn from the Youth Risk Behavior Survey (YRBS)- a series of cross-sectional school-based surveys of representative samples of students in private and public high schools initiated in 1990 and released every two years. The YRBS is reviewed by the CDC’s IRB and is designed to protect students’ privacy. Participation in the survey is anonymous and voluntary. Response rates vary by year and state with an average overall response rate of 60%.²⁸ This study uses de-identified and publicly available data from the YRBS combined datasets which include nationwide surveys conducted from 1991–2021. The total sample contains data on 234,588 adolescents.

Measures

ISA was assessed from 1991–2021 using the survey item “If you attempted suicide during the past 12 months, did any attempt result in an injury, poisoning or overdose that had to be treated by a doctor or nurse?”.

To understand the magnitude of the association between externalizing behaviors/violence exposure and ISA, we also assessed this association with other suicidal indicators such as ideation. Three additional survey items (data from 1991–2021) were used to assess the additional suicidal behaviors: suicide attempt ((how many times did you actually attempt suicide?), considered suicide (did you ever seriously consider attempting suicide?), and made a suicide plan.

Multiple items were used to create 8 variables capturing externalizing behaviors/violence exposure: 1) carry a weapon at school (on how many days did you carry a weapon such as a gun, knife, or club on school property?; data from 1993–2021), 2) carry a gun (data from 2017–2021), 3) feel unsafe at school (on how many days did you not go to school because you felt you would be unsafe at school or on your way to or from school?; data from 1993–2021), 4) threaten at school with a weapon (data from 1993–2021), 5) physical fight (data from 1991–2021), 6) physical fight at school (data from 1993–2021), 7) sexual violence (how many times did anyone force you to do sexual things that you did not want to do?; data from 2017–2021), 8) sexual dating violence (data from 2013–2021), and 9) physical dating violence (data from 2013–2021).

The time frame for all variables was within the past 12 months. Additional information on study items can be found in Appendix A.

All variables were dichotomized to assess the strength of the associations between externalizing behaviors/violence exposure and suicidal behavior using odds ratios.

Statistical analysis

The prevalence of externalizing behaviors/violence exposure was visualized over time using locally estimated scatterplot smoothing (LOESS) regression curves²⁹, a non-parametric method used to model the correlation between two variables and fit a smooth curve through a set of data points in a scatterplot while also constructing confidence intervals (CIs) around the curves. LOESS curves can reveal trends in data that might be difficult to model with parametric curves.

Descriptive statistics were assessed for externalizing behaviors/violence exposure by ISA and other suicidal behaviors. Logistic regression models were used to assess the relationship between externalizing behaviors/violence exposure and suicidal behavior. Interactions and associations by sex were then evaluated. All models were adjusted for time (year). Interaction terms between time and externalizing behaviors/violence exposure were included in additional regression models. Significant interaction test results warranted further investigation stratified by year (2-year groups were created). Thus, logistic regression models assessed the relationship between externalizing behaviors/violence exposure and ISA among each 2-year group.

State-level analyses evaluated variations in the relationship between weapon access and ISA in recent years (the latest available data in 2021 was used). Due to limited data, only 16 states were included (16 states [ID, IL, IA, LA, MI, MS, MT, NE, NH, NM, NY, OK, PA, UT, VA, WV] had data on carrying a weapon and 14 states [IL, IA, KY, LA, MI, MS, MO, MT, NV, NM, PA, UT, VA, WV] had data on carrying a gun).

All analyses were conducted in 2024 using SAS 9.4³⁰ including strata, cluster, and weights statements per YRBS documentation. Missing data were excluded from analysis. A significance level of 0.05 was used for all statistical tests. R software, version 4.2 was used for visualizations.

Results

The sample contains data on 234,588 adolescents. Descriptive results among suicidal outcomes are shown in Table 1 (ISA), Appendix Table 1 (suicide attempts including and excluding ISAs), Appendix Table 2 (suicidal ideation/plan).

Table 1.

N (%) of study participants in each independent variable by suicide outcome (ISA), (N=206,868)

Characteristics	ISA, n (%)	No Attempt, n (%)
	5,171 (2.50)	201,697 (97.50)
Carry a weapon at school
No	3,458 (73.83%)	177,053 (94.46%)
Yes	1,226 (26.17%)	10,379 (5.54%)
Carry a gun
No	745 (81.52%)	32,444 (96.50%)
Yes	169 (18.48%)	1,176 (3.50%)
Feel unsafe at school
No	3,643 (73.55%)	181,302 (95.26%)
Yes	1,310 (26.45%)	9,010 (4.73%)
Threatened at school with a weapon
No	3,335 (67.70%)	177,378 (93.54%)
Yes	1,591 (32.30%)	12,247 (6.46%)
Physical fight
No	1,708 (35.66%)	137,283 (69.44%)
Yes	3,096 (64.44%)	60,412 (30.56%)
Physical fight at school
No	2,961 (62.94%)	168,410 (89.50%)
Yes	1,743 (37.06%)	19,755 (10.50%)
Sexual violence
No	427 (50.41%)	30,195 (90.36%)
Yes	420 (49.59%)	3,221 (9.64%)
Sexual dating violence
No	986 (65.58%)	54,421 (94.57%)
Yes	518 (34.42%)	3,126 (5.43%)
Physical dating violence
No	1,025 (64.91%)	55,589 (94.96%)
Yes	554 (35.09%)	2,949 (5.04%)

The first aim was to assess the prevalence of all predictors over time by suicide attempt. Although the prevalence of carrying a weapon at school decreased from 1993 to 2021, it increased between 2019 to 2021 from 11.13 to 14.99% among those with an ISA and 5.43 to 8.18% among those without an ISA. From 2017 to 2021, the prevalence of carrying a gun increased from 16.45 to 20.59% among those with an ISA (Appendix Figure 1).

Correlations between externalizing behaviors/violence exposure and suicidal behaviors were then evaluated. Logistic regression models indicated statistically significant associations across study outcomes, and odds ratios were consistently higher when examining ISA as an outcome compared with non-ISA and suicidal ideation as an outcome. For instance, compared to those who did not carry a gun, those who carried a gun had 6.32 (95% CI: 4.78, 8.36) times the odds of an ISA vs. no attempt and 2.66 (95% CI: 2.00, 3.53) times the odds of non-ISA vs. no attempt. All models were adjusted for year (Table 2).

Table 2.

Logistic regression models assessing the association between violence related predictors and suicide related outcomes among US adolescents by sex, adjusting for year

Predictors	ISA vs. no attempt, OR (95% C.I.)	Injurious and other suicide attempt vs. no attempt, OR (95% C.I.)	Non ISA vs. no attempt OR (95% C.I.)	ISA vs non ISA	Suicidal ideation/plan vs no attempt, OR (95% C.I.)
Carry a weapon at school
No	Ref	Ref	Ref	Ref	Ref
Yes (overall)	6.28 (5.59, 7.07)	6.95 (6.15, 7.84)	2.77 (2.48, 3.10)	2.47 (2.16, 2.82)	2.62 (2.45, 2.81)
Yes (female)	7.18 (6.07,8.51)	8.87 (7.47, 10.54)	3.98 (3.45, 4.59)	2.22 (1.82, 2.71)	3.99 (3.59, 4.44)
Yes (male)	9.69 (8.16,11.49)	10.74 (9.04, 12.77)	3.91 (3.36, 4.56)	2.70 (2.22, 3.29)	3.07 (2.83, 3.33)
Predictor*sex: F test (p-value)	4.84*	1.44		2.49	14.95**
Carry a gun
No	Ref	Ref	Ref	Ref	Ref
Yes (overall)	6.32 (4.78, 8.36)	6.92 (5.22, 9.18)	2.66 (2.00, 3.53)	2.63 (1.83, 3.77)	1.94 (1.68, 2.23)
Yes (female)	6.62 (4.32, 10.13)	8.26 (5.30, 12.86)	4.09 (2.57, 6.53)	2.02 (1.18, 3.48)	3.09 (2.24, 4.25)
Yes (male)	10.24 (6.89, 15.22)	11.01 (7.38, 16.42)	3.18 (2.32, 4.38)	3.48 (2.19, 5.52)	2.27 (1.87, 2.74)
Predictor*sex: F test (p-value)	2.10	0.86	1.04	2.40	2.18
Feel unsafe at school
No	Ref	Ref	Ref	Ref	Ref
Yes (overall)	7.28 (6.58, 8.06)	8.43 (7.59, 9.36)	3.91 (3.57, 4.28)	2.15 (1.91, 2.42)	2.91 (2.74, 3.11)
Yes (female)	4.84 (4.23, 5.53)	5.63 (4.92, 6.45)	3.18 (2.86, 3.54)	1.76 (1.51, 2.06)	2.65 (2.44, 2.88)
Yes (male)	12.32 (10.47, 14.50)	14.14 (11.96, 16.72)	5.15 (4.41, 6.01)	2.72 (2.24, 3.30)	3.18 (2.92, 3.47)
Predictor*sex: F test (p-value)	74.42***	69.83***	32.48***	10.23**	9.50**
Threatened at school with a weapon
No	Ref	Ref	Ref	Ref	Ref
Yes (overall)	6.95 (6.30, 7.67)	8.00 (7.24, 8.84)	3.88 (3.58, 4.21)	2.05 (1.83, 2.31)	3.28 (3.11, 3.46)
Yes (female)	5.55 (4.88, 6.32)	6.80 (5.96, 7.76)	4.02 (3.62, 4.47)	1.68 (1.44, 1.95)	3.64 (3.36, 3.95)
Yes (male)	11.39 (9.87, 13.14)	12.97 (11.24, 14.97)	5.27 (4.67, 5.95)	2.44 (2.05, 2.91)	3.78 (3.54, 4.04)
Predictor*sex: F test (p-value)	61.17***	47.75***	11.27**	12.20**	0.77
Physical fight
No	Ref	Ref	Ref	Ref	Ref
Yes (overall)	4.34 (3.96, 4.75)	4.64 (4.23, 5.09)	2.61 (2.46, 2.77)	1.76 (1.59, 1.94)	2.00 (1.93, 2.07)
Yes (female)	5.01 (4.49, 5.60)	5.66 (5.06, 6.33)	3.33 (3.09, 3.58)	1.68 (1.48, 1.90)	2.63 (2.51, 2.76)
Yes (male)	5.69 (4.67, 6.92)	6.05 (4.97, 7.37)	3.14 (2.84, 3.47)	1.90 (1.54, 2.34)	2.08 (1.97, 2.20)
Predictor*sex: F test (p-value)	1.47	0.48	2.92	1.85	34.42***
Physical fight at school
No	Ref	Ref	Ref	Ref	Ref
Yes (overall)	5.18 (4.69, 5.72)	5.57 (5.03, 6.16)	2.35 (2.17, 2.54)	2.35 (2.10, 2.63)	1.93 (1.84, 2.03)
Yes (female)	5.65 (4.94, 6.46)	6.51 (5.67, 7.48)	3.07 (2.76, 3.42)	2.11 (1.81, 2.45)	2.41 (2.23, 2.61)
Yes (male)	7.48 (6.38, 8.78)	8.03 (6.84, 9.42)	3.04 (2.70, 3.43)	2.61 (2.16, 3.14)	2.22 (2.09, 2.37)
Predictor*sex: F test (p-value)	7.27*	3.97*	0.61	4.45*	1.72
Sexual violence
No	Ref	Ref	Ref	Ref	Ref
Yes (overall)	9.21 (7.45, 11.40)	11.30 (9.15, 13.94)	5.91 (5.06, 6.90)	1.90 (1.48, 2.44)	5.39 (4.85, 6.00)
Yes (female)	5.53 (4.26, 7.17)	6.75 (5.20, 8.76)	4.74 (4.02, 5.60)	1.41 (1.06, 1.88)	4.25 (3.77, 4.79)
Yes (male)	22.86 (16.56, 31.55)	27.73 (20.09, 38.26)	7.29 (5.31, 10.01)	3.80 (2.45, 5.90)	5.65 (4.54, 7.02)
Predictor*sex: F test (p-value)	47.19***	43.83***	6.67*	16.07**	5.39*
Sexual dating violence
No	Ref	Ref	Ref	Ref	Ref
Yes (overall)	9.16 (7.61, 11.03)	11.18 (9.29, 13.45)	5.69 (4.98, 6.49)	1.99 (1.59, 2.49)	5.12 (4.63, 5.65)
Yes (female)	5.65 (4.56, 6.99)	6.81 (5.49, 8.45)	4.20 (3.58, 4.94)	1.62 (1.25, 2.09)	4.02 (3.63, 4.45)
Yes (male)	21.25 (15.86, 28.46)	26.15 (19.50, 35.07)	8.17 (6.34, 10.52)	3.32 (2.29, 4.82)	5.59 (4.58, 6.82)
Predictor*sex: F test (p-value)	57.53***	56.53***	18.12***	11.16**	9.52**
Physical dating violence
No	Ref	Ref	Ref	Ref	Ref
Yes (overall)	10.30 (8.62, 12.31)	12.65 (10.59, 15.10)	5.47 (4.71, 6.35)	2.32 (1.87, 2.88)	4.52 (4.11, 4.99)
Yes (female)	7.70 (6.25, 9.50)	9.66 (7.81, 11.94)	4.66 (3.93, 5.53)	2.07 (1.62, 2.64)	4.13 (3.66, 4.66)
Yes (male)	15.65 (11.75, 20.84)	18.33 (13.74, 24.46)	6.24 (4.97, 7.84)	2.97 (2.10, 4.21)	4.59 (3.96, 5.33)
Predictor*sex: F test (p-value)	18.25***	14.30**	5.49 **	3.49	1.76

Note: Interaction test results by sex are also shown; Asterisks are only used to designate multiple p-value limits (

^*p<0.05,

^**p<0.01,

^***p<0.005)

The second aim was to evaluate associations by sex and year. First, interactions between predictor variables and sex were assessed. Results showed significant interactions indicating that the association between externalizing behaviors/violence exposure and suicide varies by sex. Indeed, models by sex showed a stronger association among those who identify as male. For instance, compared to males who felt safe at school, those who felt unsafe at school had 12.32 (95% CI: 10.47, 14.50) times the odds of an ISA vs. no attempt whereas female adolescents who felt unsafe at school had 4.84 (95% CI: 4.23, 5.53) times the odds of an ISA vs. no attempt (Table 2).

Variations over time in the association between externalizing behaviors/violence exposure and suicide attempt were then assessed. Due to the high magnitudes of association observed in Table 2, subsequent analyses focused on ISAs. Significant joint test results were observed for those who carried a weapon at school (F value: 2.37, p=0.02) and felt unsafe at school (F value: 2.97, p<0.01). Although significant interactions between externalizing behaviors/violence exposure and time were not observed for each indicator, the odds of an ISA vs. no attempt were assessed among each predictor. Between 1991–1993, those who carried a weapon at school had 3.96 (95% CI: 2.74, 5.70) times the odds of an ISA vs. no attempt whereas between 2019–2021, they had 6.84 (95% CI: 4.79, 9.76) times the odds of an ISA vs. no attempt compared to those who did not carry a weapon at school. (Figure 1).

Figure 1.

Odds ratio of injurious suicide attempt among US youth by year and violence exposure variables

*Note: Interaction tests results for violence exposure variables by time (joint test F value (p-value): carry a weapon at school: 2.37 (0.02), carry a gun: 2.44 (0.12), feel unsafe at school: 2.97 (<.01), threatened at school with a weapon: 1.25 (0.27), physical fight: 0.78 (0.61), physical fight at school: 1.91 (0.06), sexual violence: 0.31 (0.58), sexual dating violence: 0.80 (0.45), physical dating violence: 0.06 (0.94).

Associations overtime by sex were further evaluated. For instance, among female youth, those who carried a gun had 4.58 (95% CI: 2.37, 8.88) times the odds of an ISA vs. no attempt in 2015–2017 and 7.72 (95% CI: 4.52, 13.17) times the odds of an ISA in 2019–2021 compared to those who did not (Appendix Figure 2). Among male youth, those who carried a gun had 7.60 (95% CI: 3.27, 17.63) times the odds of ISA vs. no attempt in 2015–2017 and 11.66 (95% CI: 7.53, 18.04) times the odds of ISA in 2019–2021 compared to those who did not (Appendix Figure 3).

The last aim was to assess state-level trends in weapon access among youth with an ISA in 2021. The prevalence of carrying a weapon at school was 23.36% in Mississippi, and 30.83% in New York. The prevalence was lower in states such as Montana (6.98%) and West Virginia (5.66%). Moreover, the prevalence of carrying a gun was 14.94% in Louisiana, and 17.51% in Nevada. In states such as Illinois (4.46%), and Missouri (3.64%), the prevalence of carrying a gun was much lower (Figure 2).

Figure 2.

Prevalence of carrying a weapon at school and a gun by state among US youth with an injurious suicide attempt, 2021

Discussion

The present study evaluated associations between externalizing behaviors/violence exposure and suicidal behavior among US adolescents. Three findings emerged. 1) Suicidal behavior was significantly associated with externalizing behaviors/violence exposure, with higher magnitudes of associations among those with an ISA; 2) The association between externalizing behaviors/violence exposure and ISA varied overtime and by sex; 3) Among those with an ISA in 2021, state-level differences in weapon access emerged.

Findings indicate that ISAs³¹ and suicide attempts/ideation are much more frequent among youth with access to weapons. Over the past two decades, the prevalence of carrying a handgun among adolescents aged 12–17 increased from 3.3% to 4.6% between 2002–2006 and 2015–2019.³² Household firearms are associated with an elevated risk of firearm death to occupants in the home³³ and the risk of self-harm among adolescents should also be considered.

Increases in the prevalence of externalizing behaviors/violence exposure among adolescents with a history of suicidal behaviors, especially ISA were observed. Potential explanations include underlying risk factors such as family instability. Family structure has been associated with determining parents’ ability to consistently be involved in their adolescents’ lives, neighborhoods, environments that they live in as well as the levels of violence that they may become exposed to.^34,35 Adverse childhood experiences such as violence exposure specifically occurring in the home, school and community have been associated with the increased desensitization to community as well as real-life violence³⁶ and likelihood of developing violent behaviors^37,38 and internalizing behaviors such as anxiety and depression,³⁹ which are risk factors for suicide.⁴⁰ Exposure to violence is seldom accounted for in healthcare settings. Thus, efforts aimed at identifying violence exposure are necessary to provide appropriate clinical care.⁴¹

Stronger associations between externalizing behaviors/violence exposure and suicidal behaviors among males were also apparent. Although suicidal ideation is higher among individuals classified as female,⁴² findings are in line with evidence indicating that fatal suicide is higher among males,⁴² and those with weapon access.⁴³ Stronger associations in recent time periods indicate the urgent need to intervene. Besides restricted access to weapons, interventions can include protective factors among adolescents exposed to community violence and can consist of family,⁴⁴ and school support.⁴⁵

State-level differences in weapon access among adolescents with an ISA were also observed. Generally, the prevalence of suicide involving firearms is higher in the West.⁴⁶ Due to variations in suicide rates, gun laws and weapon access in the US, state-level assessments of suicide risk and interventions are essential. Effective interventions that result in decreases in firearm related suicide deaths among adolescents such as child-access prevention (CAP) laws,⁴⁷ Extreme Risk or “Red Flag” laws, and safer storage recommendations also vary by state. Thus, additional gun safety policies that aim to implement evidence-based strategies to limit gun access to individuals at risk of suicide at the state-level are imperative.

Analyses included data up to 2021 thereby including the start of the 2019 novel coronavirus (COVID-19) pandemic. The pandemic had a significant impact on adolescent health.^48–51 Factors such as school closures, and quarantine contributed to poorer mental health outcomes (increases in anxiety, loneliness, and depression) and increases in suicidal behavior.⁵² Specifically, Black and Asian adolescents not only experienced the adverse mental health outcomes generally shared by adolescents,^48–51 but they were also exposed to messages around racism while witnessing murders of Black individuals by the police (i.e., George Floyd, Breonna Taylor) and increases in anti-Asian hate crimes that also influenced mental health outcomes.^53–55 Thus, future research assessing the relationship between externalizing behaviors/violence exposure and suicidal behaviors focused on minoritized youth is needed.

This study has several strengths. To the authors’ knowledge, it is the first to assess the association between various externalizing behaviors/violence exposure and suicidal behaviors, including ISAs, using a wide range of years among a large sample of US adolescents. It is also the first to assess this association at the state-level. While all suicide attempts warrant focus, ISAs are a particularly severe outcome. This study specifically identified associations between various externalizing behaviors/violence exposure and ISAs including dynamics around ISAs within the broader context of overall suicide attempts. An extensive array of measures was used to thoroughly define externalizing behaviors/violence exposure. Furthermore, the need for suicide related interventions aimed at limiting firearm access at the state-level and reducing suicidal urges among adolescents exhibiting externalizing behaviors and exposed to violence is highlighted.

Limitations

This study also has limitations. First, it uses survey data. Thus, responder bias may contribute to overreporting or underreporting of results (i.e. study predictors may further increase or decrease suicide risk). Furthermore, due to its cross-sectional design, predictors were assessed at the time of the survey, not right after the ISA thereby limiting the ability to establish temporality. Second, the YRBS administers surveys to students in private and public high schools. Therefore, adolescents who are not enrolled in high schools are not included in the sample. Third, this study only includes data on cisgender categories. Future studies should consider evaluating associations among various gender identities, races/ethnicities, and sexual orientations, especially with differing trends in firearm-based suicide rates across various demographic factors. Fourth, sparse data at the state-level hindered the assessment of trends overtime in every state.

Conclusions

This study’s findings confirm the association between externalizing behaviors/violence exposure and suicidal behaviors among adolescents. This association varies overtime and by sex. Structurally competent approaches to mental health and mechanisms to identify at-risk youth are imperative. Future research should consider assessing the relationship between externalizing behaviors/violence exposure and suicidal behaviors among adolescents utilizing comprehensive state-level data.

Abbreviations:

CI

Confidence Interval

ISA

Injurious Suicide Attempt

LOESS

Locally Estimated Scatterplot Smoothing Regression

US

United States

YRBS

Youth Risk Behavior Survey