The Concurrence of Sexual Violence and Physical Fight Among Adolescent Suicide Ideators and the Risk of Suicide Attempt

Experiences of sexual violence and physical ght had been identied as independent risk factors for suicidal behaviors among adolescents. The question raises whether the concurrence of these two risk factors further increases the risk of suicidal behaviors among ideators, and by how much the risk increases. This study analyzed data from the 2019 National Youth Risk Behavior Surveys. Students that reported had seriously considered attempting suicide in the past year were included (n=1755). The associations between physical ght/sexual violence status and the risk of suicide attempt/plan/injurious attempt were estimated. The concurrence of physical ght and sexual violence substantially increased the relative risks for attempted suicide (relative risk (RR)=2.01, 95%CI: 1.74, 2.31) and injurious suicide attempt (RR=4.03, 95%CI: 2.81, 5.78; subgroup analyses: among female, RR=3.25, 95%CI: 2.09, 5.04; among male, RR=6.19, 95%CI: 3.3, 12.14). Among students with concurrent physical ght and sexual violence, males reported more risk behaviors (median=14) than females (median=12)(P=0.0023). The concurrence of physical ght and sexual violence substantially increased the risks for attempted suicide among adolescent suicide ideators. A wide variety of risk behaviors clustered with the concurrence of physical ght and sexual violence, especially among males, which possibly lead to the higher rate of injurious suicide attempt among male students.


Introduction
Suicide is among the leading causes of death among adolescents worldwide 1 . Globally, suicide was the fourth leading cause of death among the 15-19-year-olds in 2019 1 . To take the challenge of reducing the suicide rate among adolescent, better understanding of the adolescents' suicide behaviors is of great importance. It has been noted that only a small subset of those with suicidal ideation will go on to attempt, and even fewer will die by suicide 2 . For example, according to the 2019 National Youth Risk Behavior Surveys (YRBS) from the United States, 18.8% of the investigated high school students reported having seriously considered suicide, 15.7% reported having made a suicide plan, 8.9% reported having attempted suicide, and 2.5% reported had made a suicide attempt requiring medical treatment 3 . As suicidal ideation alone is not an effective predictor of suicide attempts, identi cation of risk factors for suicidal behaviors among ideators is important for the development of intervention programmes. Moreover, as past suicide attempt is one of the most consistent risk factors for suicide 2 , and studies had found that a history of suicide attempt was associated with long-standing psychosocial impairment 4,5 , investigations of risk factors and behavior patterns associated with suicide attempts among adolescent suicide ideators could contribute to the health promotions for both youth and adult.
Previous studies had identi ed a wide range of risk factors associated with the transition from suicide ideators to attempters, ranging from biological to behavior and sociological factors 2,6−8 . However, when considered individually, many of the previously identi ed risk factors have shown poor prediction of suicidal behavior [9][10][11] . As individuals who die by suicide usually present with numerous risk factors, rather than a single risk factor in isolation, exploring the effects of combined risk factors might improve our understanding of pathways that may result in suicide attempts and provide novel information for distinguishing suicide attempters from suicide ideators 11 .
Experiences of sexual violence and physical ght had been identi ed as independent risk factors for suicidal behaviors [12][13][14][15] . Their importance in the pathways of adolescent suicidal behaviors is strengthened by the recent study that used the classi cation tree analysis to identify patterns of risk factors that differentiate adolescent suicide attempters and ideators 11 . The best-performing tree included three classi cation variables, and lifetime history of rape and past-year physical ght were the rst and second classi cation variable, respectively (the third variable was heroin use). The risk of being attempters was higher among youth characterized by past-year physical ghts and heroin use (78%), youth characterized by past-year physical ghts and none heroin use (around 42%), and youth who had experienced rape (58%), compared with youth endorsed none of these three variables (29%). Although the overall classi cation accuracy was modest (area under the curve 0.65), ndings of this study suggested that the association between sexual violence, physical ght, and suicidal behaviors among adolescent worth further investigation.
In our preliminary study, we noticed that there is an overlap between the population experienced sexual violence and the population reported past-year physical ghts. According to the interpersonal theory of suicide, for individuals with a suicidal desire, it is possible to acquire the capability for suicide 2 . And both the sexual violence and physical ghts could contribute to the capability for suicide. The question raises whether the concurrence of these two risk factors further increases the risk of suicidal behaviors among ideators, and by how much the risk increases. Answers to this question not only helps identify high risk populations, but also improves our understanding of how multiple factors may affect suicide behaviors among ideators. To our knowledge, this question has not yet been explored among adolescents.
With regards to the epidemiology of suicidal behavior, one of the most consistent ndings is the sex difference. Worldwide, male deaths by suicide outnumber female deaths, yet female nonlethal suicide attempts outnumber male attempts 16 . As pathways of suicidal behaviors might be different among two sexes, we speculate that the effects of sexual violence and physical ghts on the risk of suicidal behaviors could be different among two sexes.
Previous studies had shown sexual violence is associated with other health-risk behaviors, like cigarette smoking and marijuana use 17 ; and these health-risk behaviors also served as risk factors for suicide behaviors. Similar patterns also observed for physical ghts. For example, weapon-carrying is associated with increased risk in physical ght and suicide behaviors 18,19 . According to the problem behavior theory, when adolescents are not provided with a healthy environment, health-risk behaviors and other problem behaviors may cluster together 20 .
Although the mechanisms of how multiple factors interacted in the suicide pathways could be complex and di cult to conceptualize, based on the problem behavior theory, we propose that individuals exposed to concurrent sexual violence and physical ghts would have more health-risk behaviors than individuals with only one of the exposures, and the cluster of health-risk behaviors would further increase the risk of suicidal behaviors.
Drawing from previous study ndings and existing theories, we hypothesized that among adolescent suicide ideators: 1) the concurrency of sexual violence and physical ghts would increase the risk of suicidal behaviors; 2) the effect of concurrent sexual violence and physical ghts on suicidal behaviors would be different among sex; 3) individuals with the experience of concurrent sexual violence and physical ght would be clustered with more health-risk behaviors, and would result in more severe suicidal behaviors. To test our hypotheses, we used 2019 YRBS data and included students that reported had seriously considered attempting suicide in the past year. According to the responses for the queries on the past-year experience of physical ght and sexual violence, students were categorized into four groups: 1) no physical ght and no sexual violence; 2) sexual violence without physical ght; 3) physical ght without sexual violence; and 4) physical ght and sexual violence. We rst described the general characteristics, and the rates of suicide plan, attempted suicide, and injurious suicide attempt according to study groups; and then explored the associations between physical ght/sexual violence status and the risk of suicide attempt/plan/injurious attempt; subgroup analysis according to sex were also conducted; last, we analyzed the distribution of other suicide attempt-associated factors among four groups and compared the cumulative number of risk factors among four groups and different sexes.

Data source
Publicly available data from the 2019 YRBS were analyzed. YRBS is a cross-sectional, school-based survey administered in public and private schools in the U.S. by the Centers for Disease Control. Details on this survey can be found at 26 . One hundred and eighty-one schools were sampled in the 2019 YRBS. Students in grades 9 through 12 were included in the sampling frame. The overall response rate (school response rate multiples the student response rate) was 60.3% 26 . This study included students reporting past-year seriously considered attempting suicide (Question 26 in the questionnaire). Students seriously considered attempting suicide but had missing data for past-year suicide behaviors (skipped Question 28 for past-year attempted suicide) were excluded.
The national YRBS was approved by the Institutional Review Board (IRB) at the US Centers for Disease Control and Prevention. The participation of students in the YRBS is anonymous and voluntary. Before survey administration, informed consent of parents was obtained according to local parental permission procedures. During survey administration, students completed the self-administered questionnaire during one class period and recorded their responses directly on a computer-scannable booklet 26 . In this study, all YRBS data analyses methods were performed in accordance with the guidelines and regulations provided by the Centers for Disease Control and Prevention.

Measures
The 2019 YRBS included four queries on suicide ideation and attempt: "Q26. During the past 12 months, did you ever seriously consider attempting suicide? "(answer: Yes/No; variable label: considered suicide); "Q27. During the past 12 months, did you make a plan about how you would attempt suicide? "(answer: Yes/No; variable label: made a suicide plan); "Q28. During the past 12 months, how many times did you actually attempt suicide? " (answer: 0 times/ 1 time/ 2 or 3 times/ 4 or 5 times/ 6 or more times; variable label: attempted suicide); and "Q29. 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?" (answer: I did not attempt suicide during the past 12 months/Yes/No; variable label: injurious suicide attempt).
The main outcome variable was attempted suicide (Q28). This variable were dichotomized such that responses with one or more suicide attempts were coded as suicide attempt (yes) and responses with zero suicide attempts were coded as suicide attempt (no). The secondary outcome variables were planned suicide (Q27) and injurious suicide attempt (Q29). Injurious suicide attempt were dichotomized such that responses of "I did not attempt suicide during the past 12 months" and "No" were coded as injurious suicide attempt (no) and responses of "Yes" was coded as injurious suicide attempt (yes).
Physical ght was measured using the following items: "Q17. During the past 12 months, how many times were you in a physical ght?"; and "Q18. During the past 12 months, how many times were you in a physical ght on school property?". A binary variable was created such that respondents who reported one or more times of physical ght and respondents who reported one or more times of physical ght on school property were coded as past-year physical ght (yes), and respondents reported neither of these experiences were coded as past- year physical ght (no). Sexual violence was measured using the following items: "Q19. Have you ever been physically forced to have sexual intercourse when you did not want to? "; "Q20. During the past 12 months, how many times did anyone force you to do sexual things that you did not want to do? (Count such things as kissing, touching, or being physically forced to have sexual intercourse.)"; and "Q21.
During the past 12 months, how many times did someone you were dating or going out with force you to do sexual things that you did not want to do? (Count such things as kissing, touching, or being physically forced to have sexual intercourse.) ". A binary variable was created such that all respondents with forced sexual intercourse (Q19), all respondents with one or more times of sexual violence (Q20), and all respondents with one or more times of sexual dating violence (Q21) were coded as past-year sexual violence (yes), and respondents reported none of these experiences were coded as past-year sexual violence (no).
The associations between 59 risk behaviors investigated in the 2019 YRBS and suicide attempt were screened by Chi-squared test or Fisher's exact test, and 42 risk behaviors were found to be positively associated with the risk of suicide attempt (P < 0.05), and these 42 behaviors were included in this study as suicide attempt-associated risk factors (Supplementary Table 1).

Statistical analysis
Categorical variables were presented as absolute values and percentages. Between-group comparisons for categorical variables were conducted using Chi-squared test or Fisher's exact test, as appropriate. Pairwise comparisons were conducted using Bonferroni correction (α'=0.05/ number of comparison sets). Log-binomial regression was used to assess the associations between the status of physical ght/sexual violence and suicide behaviors (attempted suicide, planned suicide, and injurious suicide attempt). Relative risk (RR) and 95%CI were calculated and were adjusted for demographic variables that found to be signi cantly associated with physical ght/sexual violence status. Subgroup analysis for the associations between the status of physical ght/sexual violence and suicide behaviors were conducted according to sex. Attributable risk percent (AR%) was calculated by (RR-1)/RR×100%. Survey weights were not used because our research questions focused on the associations between physical ght/sexual violence status and suicide attempts among those seriously consider attempting suicide, rather than estimating population prevalence. Missing data was not imputed. All tests were set two-tailed with a type 1 error rate xed at 5%. All statistical analyses were performed using SAS 9.3 (SAS Institute, Inc., Cary, NC).

Results
Study groups and suicide behaviors Figure 1 shows the ow chart of the study population. Among the 13677 respondents in the 2019 YRBS, 2633 reported past-year seriously consider attempting suicide. Among these, 538 respondents skipped the query of suicide attempt, 340 respondents skipped the queries for physical ght and/or sexual violence, therefore were excluded from this study. The nal analyses included 1755 students. These students were categorized into four groups according to the physical ght/sexual violence status: the (physical ght -/ sexual violence -) group (n = 827, 47.12%); the (physical ght -/ sexual violence +) group (n = 360, 20.51%); the (physical ght +/ sexual violence -) group (n = 318, 18.12%); and the (physical ght +/ sexual violence +) group (n = 250, 14.25%). Table 1 shows the distribution of demographic characteristics and the rates of past-year suicide attempts among four groups. The distribution of sex, sex identity, and grade were signi cantly differed among groups (P < 0.05). The rates of suicide plan, suicide attempt, and injurious suicide attempt were signi cantly different among four groups (P < 0.05). The (physical ght +/ sexual violence +) group had the highest rates of suicide plan (75.5%), suicide attempt (63.6%), and injurious suicide attempt (28.4%). Associations between physical ght/sexual violence status and suicide behaviors  Suicide attempt-associated factors Table 3 shows

Discussion
The concurrence of physical ght and sexual violence substantially increased the relative risks for attempted suicide and injurious suicide attempt among adolescent suicide ideators. The pattern of associations were different among two sexes: among female, the association between physical ght and/or sexual violence and suicide behaviors were more prevalent; in males, the concurrence of physical ght and sexual violence was associated with remarkably higher risk for injurious suicide attempt. High rates of suicide-attempt associated risk behaviors, as well as a high cumulative number of risk behaviors were observed among the population with concurrent physical ght and sexual violence, especially among males.
Both physical ght and sexual violence had been known as risk factors/predictors for suicidal behaviors among adolescents and adults [12][13][14][15] . For example, the YRBS data showed that high school students who reported attempting suicide during the preceding 12 months were more likely also to have reported ghting than those who reported not attempting suicide (61.5% vs 30.3%) 15 . Stack et al. reported that involvement in physical ghting raised the odds of a suicide attempt by 2.18 times 13 . A experience of sexual violence was found to be associated with increased risk of suicide attempt among both high school male and female students (OR = 11.3 and 4.54, respectively; P < 0.05) 14 . Consistent with previous ndings, our analyses showed both factors were associated with increased risk of suicide attempt. The potential role of physical ght and sexual violence in the suicidal behavior trajectory may be understood within the context of the interpersonal theory of suicide 2 . According to the interpersonal theory, three constructs, namely thwarted belongingness, perceived burdensomeness, and the acquired capability for suicide, are needed for an individual to attempt suicide. The capability for suicide could be acquired by both increased physical pain tolerance and reduced fear of death through habituation to the physically painful and fearful aspects of self-harm 2 . Drawing on the interpersonal theory of suicide, physical ghts may serve as a risk factor for suicidal behaviors by increasing the individual's acquired capability for suicide. In support of this explanation, one study had reported that individuals who reported engaging in physical ghts had higher acquired capability scores 21  Our study observed associations between physical ght/sexual violence and a wide variety of health-risk behaviors, many of which had been reported in previous studies 17,19,22,23 . For example, weapon-carrying was found to be associated with increased involvement in physical ghting 19 , and early alcohol use initiation and heavy drinking was associated with comorbid ght and suicide attempts among adolescent 23 ; a history of adolescent sexual victimization was found to be associated with increased likelihood of cigarette smoking, marijuana use, multiple sexual partners, and early sexual intercourse among college women 17 . Considering that the risk behavior we investigated were selected from factors that associated with the risk of suicide attempt, the observed association between physical ght and/or sexual violence and the risk of suicide attempt could be a result of combined effects of the clustered risk factors. The clustered factors could interact as both cause and effect, and in uenced the suicidal behavior through complex pathways, therefore isolating the effects of single factors might not practical. Also, interventions targeting single factors might not powerful enough for the prevention of suicidal attempts. The problem behavior theory suggested that the risk-health behaviors might cluster together because they serve similar psychological and social developmental functions 20 . Therefore, for the purpose of developing intervention plans, it is important to identify the key factors that leads to subsequent risk behaviors and to identify reversible factors that could serve as intervention targets.
Among both sexes, the concurrency of physical ght and sexual violence was associated with a high risk for injurious suicide attempt.
26.5% of female students with concurrent physical ght and sexual violence reported injurious suicide attempt; among male students, the rate was 32.8%. The RR for injurious suicide attempt was remarkably higher among male (RR = 6.19, 95%CI: 3.3, 12.14) than female (RR = 3.25, 95%CI: 2.09, 5.04), using the risk in the (physical ght -/ sexual violence -) group in the same sex as the reference. Sex difference in the suicidal behavior has consistently been observed. In most part of the world, females are more likely than men to attempt suicide, and they are less likely to die by suicide 16 . One explanation from the interpersonal theory of suicide is that females are less likely to develop the acquired capability for suicidal behavior that are men, because females generally experiences fewer events that could reduce their fear of self-injury through habituation (e.g., exposure to guns, physical ghts, violent sports, etc.), and females have lower pain and fear tolerance than men 2 . Our nding supported this explanation in that among the youth with concurrent physical ght and sexual violence, males reported more health-risk behaviors than females (male: median = 14, range: 2-40; female: median = 12, range: 0-33; P = 0.0023). This nding adds empirical evidence for the understanding of sex difference in suicidal behaviors, and again strengths the importance of examining an individual's risk factor pro le when evaluating the suicide risk.
Strengths of this study include the novel investigation of the association between concurrent physical ght and sexual violence and the risk of suicide attempt among adolescent suicide ideators, and the cumulative number of suicide risk factors among several sub-populations.
In searching for risk factors that could differentiate attempters from ideators, most of previous studies examined the effects of single factors. The ndings of our study emphases that in both researches and intervention practices, the individual's risk of suicide behaviors should be conceptualized as a combined effects of multiple factors. The cumulative number of suicide risk factors also helps understand the remarkable higher rates of injurious suicide attempt among adolescent males with concurrent physical ght and sexual violence.
Besides, when estimating the strength of association between risk factors and suicide behaviors, unlike studies that used logistic regression to estimate the OR, we used log binomial regression to estimate the RR and AR which are more intuitive and can be better interpreted 24 .
This study has several limitations. First, the cross-sectional design of the study limits our ability to make causal inferences. It is possible that some of the reported experiences of physical ght and sexual violence were occurred after the suicide attempt. Nevertheless, the identi ed patterns could still be informative for the understanding of how the suicide-related risk behaviors clustered and for the identi cation of high risk populations that need multiple behavior interventions. Longitudinal studies are needed to test the causality relationship between sexual violence, physical ght, and suicide behaviors. Second, the YRBS data are collected from self-reported questionnaires, the answers would subject to recall bias and classi cation error which was caused by individual's different understanding of the queries. Third, the YRBS is a school based survey, therefore may not representative for adolescents who did not attend school. Lastly, ndings of this study are draw from the adolescents from the United States, the combined effects of could be different under different culture and social-development conditions 25 . To improve the understanding of the suicide behaviors and promote effective interventions, future research is needed to validate these ndings in different counties and populations.
To conclude, the concurrence of physical ght and sexual violence substantially increased the risks for attempted suicide and injurious suicide attempt among adolescent suicide ideators. A wide variety of risk behaviors clustered with the concurrence of physical ght and sexual violence, especially among males, which possibly lead to the higher rate of injurious suicide attempt among male students. Further study is needed to explore the mechanism of how combined risk factors works in the suicide pathways. And to guide the development of suicide intervention strategies, it is important to identify the key factors that leads to subsequent risk behaviors and to identify reversible factors that could serve as intervention targets.

Declarations
Author contributions X.L. contributed to the conception of the work. X.L., S.X. and J.D. analyzed the data and wrote the main manuscript text. All authors reviewed the manuscript.

Additional Information
Competing interests Figure 1 Flow diagram of study population