SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Differentiating Suicide Attempters from Suicide Ideators
  4. Methodology
  5. Results
  6. Theoretical Discussion
  7. Conclusion
  8. References
  9. Appendix 1

Which factors distinguish suicide attempters from suicide ideators is a relatively neglected question in suicidology. Data from the 2001 Youth Risk Behavior Survey, encompassing 1,439 youth suicide ideators and 1,097 attempters, was used to explore which factors best differentiate suicide attempters from ideators, with a focus on violence involvement. Measures of violence include the contexts of fights, dating, and weapons carrying. Controls were incorporated for psychiatric disorders, risky sexual behavior, school integration, and demographics. Controlling for the other variables, violence differentiated attempts from ideation: fighting (OR = 2.18) and weapon carrying (OR = 1.13). Psychiatric factors that predicted attempts over ideation included major depression (OR = 1.86), use of cocaine (OR = 2.34), and having a suicide plan (OR = 2.69), while demographic factors included gender, age, residence in the Midwest, and Hispanic, African American, or Asian ethnicity. A supplementary analysis (N = 11,546) determined that violence also helped to differentiate suicide ideators from nonsuicidal youth. Four factors (including violence involvement, eating disorders, and gender consistently) differentiated both between suicide attempts and ideation, and also between suicide ideators and nonsuicidal youth. The link between violence involvement and suicidality is interpreted in terms of the capability for suicide from the interpersonal theory of suicide.

According to data on the Web of Science (accessed 11-27-12), there have been 46,162 publications on the topic of suicide published since 1980. Most of the work on predicting suicidality focuses on one aspect of suicidality: which factors predict suicide ideation, attempts, or completions (e.g., Anestis, Bagge, Tull, & Joiner, 2011; Duke, Pettingell, McMorris, & Borowsky, 2010; Eaton et al., 2011; Hinduja & Patchin, 2010; Rasmussen & Wingate, 2011; Swahn, Lubell, & Simon, 2004; Swahn et al., 2008; for exceptions see Gunn, Lester, & McSwain, 2011; Joiner et al., 2009). For example, a review of 100 consecutive empirical studies in MEDLINE (accessed 5-1-13) under the subject “suicide” determined that 15 articles dealt exclusively with suicide ideation, 37 exclusively with attempts, and 31 with only completions. Fifteen dealt with both suicide ideation and attempts, but each type of suicidality was analyzed separately. Finally, three dealt with completions and attempts. Only a single investigation sought to differentiate one form of suicide from another form: subway suicides from subway suicide attempts in Austria (Niederkratenthaler et al., 2012).

While the research on suicidality has offered extensive lists of risk and protective factors for suicide, there is little work on which, if any, factors can reliably distinguish those who attempt suicide from those who only consider suicide. Factors that may predict suicide attempts may not automatically distinguish attempts from ideation.

The present investigation sought to differentiate suicide ideators from attempters. What makes attempters different from ideators? The answers to this question can inform efforts at suicide attempt prevention. That is, for example, if violence involvement differentiates attempters from ideators, with attempters more engaged in such activities as dating violence, preparation for violence including weapon carrying, and physical assault, such behaviors can serve as warning signs for prevention. In addition, factors that fail to differentiate attempters from ideators might also inform those involved in suicide prevention regarding where to best place their limited resources.

First, the literature that has sought to differentiate attempters from ideators is briefly reviewed and some of its recurrent limitations noted. Next, the focus turns to violence involvement as a candidate for differentiating attempts from ideation. The model presented here includes 21 control variables representing four additional variable sets (psychiatric morbidity, sexual risk behaviors, school integration, and demographics) drawn from the existing literature on suicidality. The first analysis is based on over 2,000 suicidal subjects and presents strong, consistent findings on how violence involvement differentiates attempts from ideation. In an expanded analysis of over 12,000 subjects, violence is shown to also differentiate suicide ideators from nonsuicidal youth. The theoretical implications of these findings on violence are discussed in light of current work on the interpersonal theory of suicide (Joiner, 2005; Joiner et al., 2009).

Differentiating Suicide Attempters from Suicide Ideators

  1. Top of page
  2. Abstract
  3. Differentiating Suicide Attempters from Suicide Ideators
  4. Methodology
  5. Results
  6. Theoretical Discussion
  7. Conclusion
  8. References
  9. Appendix 1

Existing research on this issue tends to have one or more limitations, including being based on nonrepresentative samples of the population such as psychiatric patients (Paluszny, Davenport, & Kim, 1991; Wetzler et al., 1996), small samples of less than 100 cases (Paluszny et al., 1991), reliance on possible tautological arguments where one aspect of suicidality causes another aspect of suicidality (Joiner et al., 2009; Wetzler et al., 1996), and exploration of a limited number of predictor variables (Paluszny et al., 1991). There are also conflicting findings wherein, for example, some studies cite depression as a construct, which can differentiate attempters from ideators (Gunn et al., 2011), while others report that depression fails to differentiate attempters from ideators (Joiner et al., 2009).

The variation in the findings and some methodological differences across the studies are summarized in Table 1. Wetzler et al. (1996) explored the utility of five possible independent variables (depression, impulsivity, negative life events, positive life events, and aggression) for differentiating attempts from ideation among members of a sample of 225 depressed adolescent outpatients. Unfortunately, none of those variables differentiated the attempters from the ideators. Two additional–gender and physical health–variables, forming a possible tautology, did differentiate between the groups: lifetime suicide plans and recent active suicidal behavior. Another investigation, also based on a patient sample (N = 58) of pediatric patients, found that only two variables–genders and physical health–differentiated the 42 attempters from 16 ideators. (Paluszny et al., 1991).

Table 1. Previous Research Differentiating Suicide Attempters from Suicide Ideators: Author and Year, Sample, Sample Size, Factors Differentiating Attempts from Ideation, Factors Not Differentiating Attempts and Ideation
Author (Year)SampleSample sizeVariables differentiating SA and SIVariables not differentiating SA and SI
  1. SI, suicide ideators; SA, suicide attempters; SAM, suicide attempters requiring medical treatment; SAN, suicide attempters not requiring medical treatment.

Gunn et al. (2011)National representative, age: 18 and overSI 2,025, SA 310Fighting, depression, and marital status Alcohol abuse, anxiety, drug abuse, gender, hopelessness, recklessness, and social withdrawal
Joiner et al. (2009)Patients, with severe suicidality313Negative life events, past suicide attempts, (suicide attempts × burdensomeness × belongingness)Age, gender, family suicide, family depression, family manic, depression, hopelessness, borderline, past manic, current manic, depression, past depression, burdensomeness, belongingness (BUR × BELONG),(BELONG × SA), (BUR × SA)
Paluszny et al. (1991)Patients, age 12–28SA 42, SI 16Gender, physical illnessAge, constriction, depression, lack of insight, performance in school, family problems, girlfriend/boyfriend problems, acting out behavior, drug and alcohol abuse
Wetzler et al. (1996)Patients admitted to suicide and depression programSAM 84, SAN 57, SI 40SAM higher on lifetime suicide plans and actions; SAN higher on recent active suicidal behaviorAnger/aggressivity, depression, impulsivity, life events

Two investigations provide evidence that violence differentiates between ideators and attempters. One study was based on a large national representative sample. Data referred to 2,025 suicide ideators and 310 suicide attempters, all 18 and older, from the National Survey on Drug Abuse and Health. Gunn et al. (2011) determined that only three of nine variables were able to significantly differentiate suicide attempts from suicide ideation: aggression (fighting), marital status, and depression. Joiner et al. (2009) were able to differentiate recent attempters from ideators through a multiplicative interaction term containing a violence measure, the number of previous suicide attempts. However, this measure also raises some concern of a tautological argument in that a previous behavior is used to predict the same behavior in the present.

Following the leads in previous research differentiating attempts from ideation (Gunn et al., 2011; Joiner et al., 2009), the present inquiry focused on the link between violence and suicidality. A further reason for the focus on violence is that there is a substantial related literature that tends to find that involvement in violence contributes to suicidality (for reviews see Nock & Marzuk, 2000; Plutchik, 2000). More recently, Swahn et al. (2004) determined that involvement in physical fighting raised the odds of a suicide attempt among boys by 5.0 times and among girls by 4.8 times. A study based on a representative sample of 136,549 students in Minnesota determined that physical, nonsexual abuse by a household adult increased the odds of a suicide attempt by 5.07 times for girls and 4.85 times for boys (Duke et al., 2010). In a study of 1,963 middle school students, both bullying perpetrators and victims had a significant high risk of a suicide attempt (Hinduja & Patchin, 2010). The bulk of this work is descriptive, focusing on making generalizations on risk and protective factors for suicide. In addition, it does not test to see whether violence involvement can differentiate attempts from ideation. Nevertheless, it provides relatively consistent empirical generalizations regarding the association between suicidality and violence.

Other Predictor and Control Variables

In addition to violence involvement measures, the present investigation included additional predictor variables. The selection of factors was informed by previous investigations seeking to differentiate attempts from completions (Gunn et al., 2011; Joiner et al., 2009; Paluszny et al., 1991; Wetzler et al., 1996) and the more general literature on suicide attempts and ideation (e.g., Anestis et al., 2011; Duke et al., 2010; Eaton et al., 2011; Hinduja & Patchin, 2010; Rasmussen & Wingate, 2011; Swahn et al., 2004, 2008).

Some articles are marked by model specification bias. For example, in an analysis of the national Youth Violence Survey, the only confounders that are controlled are gender, age, and ethnicity (Swahn et al., 2008). Psychiatric constructs including depression and substance abuse as well as other constructs and measures are not included, possibly due to lack of appropriate data in the data archives. The present review included psychiatric controls and measures of three additional variable sets: risky sexual behavior, school integration, and demographics.

Methodology

  1. Top of page
  2. Abstract
  3. Differentiating Suicide Attempters from Suicide Ideators
  4. Methodology
  5. Results
  6. Theoretical Discussion
  7. Conclusion
  8. References
  9. Appendix 1

All data are from the 2001 Youth Risk Behavior Survey (YRBS). The YRBS is based on a national representative sample of students enrolled in 199 high schools, selected through a multistage cluster sampling procedure. The student response rate was 89%. Unlike some years, the 2001 survey has some advantages for model specification. It contains a measure of major depression, an important psychiatric control. For further details on the YRBS, see Centers for Disease Control (2003).

In the first part of the analysis, the sample was restricted to subjects who reported a suicide attempt (n = 1,097) and those who reported suicide ideation (n = 1,439). A suicide attempt was measured using the following item: “During the past 12 months, how many times did you actually attempt suicide?” The item measuring suicide ideation was, “During the past 12 months, did you ever seriously consider attempting suicide?” where the responses are either yes or no. In this study, the dependent variable was constructed to differentiate suicide attempters from suicide ideators, where 1 = suicide attempter and 0 = suicide ideator. Some ideators were also attempters, and they are placed in the attempter group. Because the dependent variable is a dichotomy, logistic regression techniques were appropriate (Menard, 2002).

Violence Involvement

Three measures of the central construct of violence were employed. (1) Involvement in serious physical fighting was measured by the item: “During the past 12 months, how many times were you in a physical fight in which you were injured and had to be treated by a doctor or nurse?” where 1 = one or more times and 0 = never. (2) Dating violence was measured by the following item: “During the last 12 months did your boyfriend or girlfriend ever hit, slap, or physically hurt you on purpose?” where 1 = yes and 0 = no. (3) Violence preparedness was captured using a measure concerning possession of weapons: “During the last 30 days on how many days did you carry a weapon such as a gun, knife, or club?” Responses ranged from 0 through 6 or more days. Carrying a weapon signals an anticipation of violence either as an offender or victim or both (Stack, 2012; Wolfgang, 1958). A weapon is also an important symbol of the subculture of violence (Wolfgang, 1958).

Psychiatric Controls

To check on possible spuriousness, wherein depression and other psychiatric variables might contribute to both suicidality and violence, a set of controls were introduced for psychiatric conditions. The presence of depression is measured as a binary variable by the item: “During the past 12 months, did you ever feel so sad or hopeless almost every day for 2 weeks or more in a row that you stopped doing some usual activities?” where 1 = yes and 0 = no. The presence of an eating disorder was measured by an affirmative response to the item: “During the past 30 days, did you vomit or take laxatives to lose weight or to keep from gaining weight?” Three measures were employed to measure substance abuse: (1) marijuana use is tapped by the item: “During the last 30 days, how many times did you use marijuana?” where 1 = 3 or more times and 0 = all others; (2) ocaine use is captured by the item: “During the past 30 days, how many times did you use any form of cocaine, including powder, crack, or freebase” where 1 = 3 or more times and 0 = all others; and (3) Alcohol abuse was measured by the item: “During the past 30 days, how many days did you have five or more drinks of alcohol in a row, that is, within a couple of hours” where 1 = 2 or more days and 0 = all others. Finally, a control was introduced for having a suicide plan: “During the past 12 months, did you make a plan about how you would attempt suicide?” where 1 = yes and 0 = no.

Risky sexual behaviors were measured by three items. First, the number of reported sex partners during the last 30 days, responses ranged from zero through six or more; second, unsafe sex was measured using a response of no to the item: “The last time you had sexual intercourse, did you or your partner not use a condom?” Finally, teen pregnancy was measured with the question: “How many times have you been pregnant or gotten someone pregnant?” where 1 = one or more times and 0 = none.

School integration was captured by two items. Membership in a sports team (0, 1), and, in reverse, having a low grade point average (0,1). The latter was reporting that during the last year the subject received either “mostly D's” or “mostly F's” (1), and all other responses are 0's.

Demographic Controls

Gender was measured as a dichotomous variable where 1 = male and 0 = female. Age was measured in years. A series of binary variables measured ethnicity: Asian (0, 1), African American (0, 1), Hispanic (0, 1), and other non-White (0, 1). Caucasians serve as the reference category. Region of the nation was measured in terms of three binary variables with the Northeast as the reference category. Urbanity was measured by two binary variables: city (0, 1) and suburbs (0, 1), with rural residence as the benchmark category.

In the second part of the analysis, the sample was expanded to include all nonsuicidal youth plus only the suicide ideators (N = 11,546). All suicide attempters are excluded from this second analysis. In this supplemental analysis, the dependent variable was a dichotomy where 1 = suicide ideation and 0 = all other youth. The five variable sets from the first part of the analysis were employed as predictor variables. A goal was to locate consistent predictors, ones that cannot only differentiate between attempts and ideation, but also differentiate suicide ideators from nonsuicidal subjects.

Results

  1. Top of page
  2. Abstract
  3. Differentiating Suicide Attempters from Suicide Ideators
  4. Methodology
  5. Results
  6. Theoretical Discussion
  7. Conclusion
  8. References
  9. Appendix 1

For a list of variable means and a matrix of correlation coefficients see Appendix 1. The results of two multivariate logistic analyses are provided in Table 2. In columns 2 and 3, the analysis incorporates only the three measures of violence. Controlling for the other two measures of violence, fighting differentiated attempts from ideation. The odds of reporting a suicide attempt are 155% higher for those subjects reporting having been in a physical fight requiring medical attention for injuries sustained. In like manner, dating violence and the number of days carrying a weapon also differentiated between attempts and ideation.

Table 2. Constructs Differentiating and Not Differentiating Between Suicide Ideators (=0) and Suicide Attempters (=1), Multiple Logistic Regression Analysis (N = 2,536 youth, Youth Risk Behavior Survey, 2001)
VariableLogistic coefficientOdds ratioLogistic coefficientOdds ratioZ-test slopes
  1. Z-test in column 6 refers to different in slopes test comparing regression slope in column 2 with regression slope in column 4 (Paternoster, Brame, Mazerole, & Piquero, 1997). GPA, grade point average.

  2. *p < .05, **p < .1.

Violence
Fighting, with injuries requiring medical attention.93*2.55.78*2.180.63
Dating violence.46*1.58.17**1.191.73
Number of days carried weapon.13*1.14.12*1.13.15
Psychiatric controls
Major depression  .62*1.86 
Eating disorder  .52*1.69 
Marijuana use  −.17.84 
Cocaine use  .85*2.34 
Binge drinker  −.02.97 
Suicide plan  .99*2.70 
Risky sexual behavior
Number of sex partners  .07*1.08 
Unsafe sexual intercourse  .051.05 
Pregnant/impregnated someone  .26**1.29 
School integration
Sports team member  −.04.95 
Low GPA   .151.17 
Demographic controls
Male (=1)  −.50*.60 
Age  −.17*.84 
Race (White = reference)
African American  .38*1.47 
Hispanic  .76*2.15 
Asian  .63*1.89 
Other race (non-White)  .59*1.81 
Region (Northeast = reference)
South region  −.31**.73 
West region  −.22.79 
Midwest region  −.49*.61 
Urbanity (rural = reference)
City resident  .111.11 
Suburban resident  .00081.0008 
Constant term−.70.92  
Model chi-squared115.73* 443.39*  
Nagelkerke R-squared.063 .235  
% cases correctly classified61.53% 68.78%  

Columns 4 and 5 of Table 2 show the results for the full model, including controls for variables from the psychiatric, risky sexual behavior, school integration, and demographic perspectives. Controlling for the 23 other predictor variables, the odds of reporting a suicide attempt increased 118% for those persons having been in a physical fight with injuries that required medical attention. Fighting still differentiated the suicide attempts from the suicide ideators. Further, from the odds ratio for weapons, each additional day the subject carried a weapon increased the odds of an attempt by 13%. There were mixed findings concerning variables from the psychiatric perspective. Both depression and an eating disorder differentiated the attempters from the ideators. The odds of a suicide attempt increased 86% for persons reporting depression. The presence of an eating disorder increased the odds of a suicide attempt by 69%. Only one of the three measures of substance abuse (cocaine user) differentiated between attempters and ideators. Finally, the presence of a suicide plan distinguished the attempters from ideators. Those subjects reporting a plan for suicide were fully 170% more apt to have attempted suicide than those who had no such plan.

The results for measures of risky sexual behavior and school integration were mixed. Only the number of sex partners differentiated between attempters and ideators (odds ratio = 1.08).

Seven of the eleven demographic control variables significantly differentiated between the attempters and ideators. As anticipated, males were less apt than females to be attempters. From the odds ratio of .60, we can calculate [(1−.6)*100] that males are 40% less apt to be suicide attempters (vs. ideators) than females. Suicide attempts become less common than ideation with age. From the odds ratio [(1−.84)*100], each year decreases the risk of an attempt by 16%. All four measures of ethnicity enhanced the risk of an attempt compared with the reference group of Caucasians. The odds ratio shows that the risk of an attempt was between 47% greater in the case of African Americans and 115% greater in the case of Hispanics. Finally, residence in the Midwest differentiated the attempts from the ideators, but the measures of urbanity did not. The full model explained 23.5% of the variability in the dependent variable and correctly classified 68.78% of the cases as either attempts or ideation.

The results of the multiple logistic regression analysis based on the expanded YRBS sample including suicide ideators and nonsuicidal youth are presented in Table 3. Herein the goal was to differentiate the suicide ideators from the nonsuicidal youth.

Table 3. Constructs Differentiating Suicide Ideators (1) from Nonsuicidal Youth (0). Multiple Logistic Regression Analysis (N = 11,546, Youth Risk Behavior Survey)
VariableLogistic coefficientOdds ratioLogistic coefficientOdds ratioZ-test slopes
  1. Z-test in column 6 refers to different in slopes test comparing regression slope in column 2 with regression slope in column 4 (Paternoster et al., 1997). GPA, grade point average.

  2. *p < .05; **p < .1.

Violence
Fighting, with injuries requiring medical attention.42*1.53.181.201.12
Dating violence.75*2.11.34*1.403.24*
Number of days carried weapon.03**1.04.031.03.08
Psychiatric controls
Major depression  1.64*5.19 
Eating disorder  .66*1.94 
Marijuana use  .24*1.28 
Cocaine use  −.44.64 
Binge drinker  .13**1.14 
Risky sexual behavior
Number of sex partners  .011.01 
Unsafe sexual intercourse  .15*1.16 
Pregnant/impregnated  .18**1.20 
School integration
Sports team member  −.16*.84 
Low GPA   .27*1.31 
Demographic controls
Male (=1)  −.33*.71 
Age  −.01.99 
Race (White = reference)
African American  −.65*.52 
Hispanic  −.61*.54 
Asian  −.29**.74 
Other race (non-White)  −.16**.84 
Region (N.E. = reference)
South region  .121.13 
West region  .17**1.19 
Midwest region  .29**1.34 
Urbanity (rural = reference)
City resident  .081.09 
Suburban resident  .081.08 
Constant term−2.19* −2.60*  
Model chi-squared91.92* 1048.8*  
Nagelkerke R-squared.015 .17  
% cases correctly classified88.5% 88.6%  

The results of two multivariate logistic analyses are provided in Table 3. In columns 2 and 3, the analysis incorporates only the three measures of violence.

Violence involvement, measured in terms of dating violence, predicted the dependent variable, herein suicide ideation versus the absence of suicidality. Columns 4 and 5 of Table 3 report the results for the full model including controls for variables from the psychiatric, risky sexual behavior, school integration, and demographic perspectives. Controlling for the 23 other predictor variables, having been a victim of dating violence raises the odds of suicidality by 40%. While there are some shifts in what is significant, two measures of psychiatric morbidity remain significant predictor variables: depression and presence of an eating disorder. Both measures of school integration are significant, reversing the results from Table 2. With respect to the demographic predictors, male gender remains a protective factor. With respect to race, Black and Hispanic are protective factors, changing the results from Table 2. Urbanity differentiates between suicidal and nonsuicidal youth. The full model explained 17.0% of the variability in the dependent variable and correctly classified 88.6% of the cases as either suicidal or nonsuicidal. We turn to the interpretation of the study's findings on violence.

Theoretical Discussion

  1. Top of page
  2. Abstract
  3. Differentiating Suicide Attempters from Suicide Ideators
  4. Methodology
  5. Results
  6. Theoretical Discussion
  7. Conclusion
  8. References
  9. Appendix 1

The findings on violence involvement and suicidality have theoretical implications for extending Joiner's (2005) interpersonal theory of suicide (IPTS) as it involves the capability for suicide. In an early exposition of IPTS, it was contended that experience with pain, violence, and provocative events build the “capability for suicide” (Joiner, 2005). Past experiences with pain and violence can desensitize the individual to violence against oneself (suicide). Such experiences are seen as necessary to overcome a natural instinct for self-preservation. For example, enduring surgeries, injuries in contact sports, victimization by bullies, victimization or perpetration of violence in such contexts as domestic violence, dating violence, intimate partner violence, gang violence, and robberies can serve to make one more capable of suicidal behavior (e.g., Joiner, 2005).

Unlike the findings in the present study, which emphasize personal involvement in violence, findings in the empirical research on the capability for suicide, which typically omit measures of personal involvement with violent actions, have been mixed (Joiner, 2005; Witte, Didie, Menard, & Phillips, 2012; for a review see Ribeiro & Joiner, 2011). For example, painful restrictive food intake practices, a measure of capability, predicted suicide attempts (Witte et al., 2012). However, a 25-item measure of capability was not predictive of suicide attempts in a study based on 283 undergraduates (Anestis et al., 2011), and a 20-item measure of capability did not predict suicide ideation in a sample of 483 undergraduates (Rasmussen & Wingate, 2011). However, past suicide attempts combined with burdensomeness and thwarted belongingness exerted a significant multiplicative impact on current suicide attempts (Joiner et al., 2009).

The mixed results on the capability to suicide in the IPTS may be a function of measurement issues in the scales of capability, not the essence of the theoretical argument. For example, unlike the present study, the 20 items in the original Joiner scale did not include any questions on violence perpetration or victimization such as actual dating violence or fighting (items included in the present study). Instead they dealt with attitudes toward spectator violence, fear of death, and other related matters: “The best part of hockey games are the fights,” “The fact that I am going to die does not bother me,” and “The sight of my own blood does not bother me.” Possibly alternative specific measures emphasizing actual personal involvement in violence and painful experiences might provide more consistent findings.

The present study adds to the literature on IPTS by employing specific behavioral measures of violence. It finds relationships between actual violent behavior and suicidality. This suggests that the addition of such measures of behavioral violence perpetration, victimization, and preparedness (carrying weapons) might strengthen the measurement of the capability for suicide. More consistent findings for IPTS might emerge with such modifications.

Conclusion

  1. Top of page
  2. Abstract
  3. Differentiating Suicide Attempters from Suicide Ideators
  4. Methodology
  5. Results
  6. Theoretical Discussion
  7. Conclusion
  8. References
  9. Appendix 1

There is relatively little known about what theoretical constructs might help us to understand what differentiates individuals at one level of suicidality such as attempts, from another level such as ideation (e.g., Anestis et al., 2011; Duke et al., 2010; Eaton et al., 2011; Hinduja & Patchin, 2010; Rasmussen & Wingate, 2011; for reviews, see Lester, 2000; O'Connor, Platt, & Gordon, 2011; Stack, 2000; Wasserman & Wasserman, 2009). The present article addresses this question for a national representative sample of youth.

Findings from previous research seeking to differentiate attempts from ideation are split when it comes to violence involvement, with two studies reporting positive effects (Gunn et al., 2011; Joiner et al., 2009) and one study finding that aggressivity failed to differentiate attempts from ideation (Wetzler et al., 1996). The results of the present study showed that violence involvement successfully differentiated between attempts and ideation. Violence involvement continued to help differentiate ideation from the absence of suicidal behavior with controls for psychiatric morbidity, school integration, risky sexual behavior, and demographics. The full model was able to correctly classify 69% of the cases as attempts or ideation. Violence involvement was also useful in differentiating suicidal from nonsuicidal youth, the full model correctly classifying 89% of the cases.

The ability for violence to predict suicidality can be interpreted through a construct from Joiner's (2005) IPTS: the capability for suicide. Experience with actual violence can contribute to suicide through such mechanisms as promoting a familiarity with pain (including car accidents and surgery), desensitization, and promotion of fearlessness. The findings from previous investigations concerning capability are quite mixed, but tend to be based on attitudinal measures of violence. However, the findings in the present investigation suggest that the link between suicidality and violence would be strengthened through the inclusion of actual experiences with violence including various forms of assault both as a perpetrator and victim.

The present inquiry has its limitations. First, all suicidality items in the present study refer to a window of 12 months. Future research is needed using data sets that have measures of lifetime suicide attempts, plans, and/or ideation. Further, it is not clear if the results based on youth can be generalized to nonyouthful populations.

Caution should be exercised in generalizing the results of the present study, which links violence involvement with suicide attempts to the suicidal modality of suicide completions. The present study had no data on completions. However, in other studies, violence involvement has often been shown to be associated with suicide completions (e.g., Stack, 2012; Webb et al., 2012). In a large Danish population study of 27,000 suicides and over a half million living control subjects, it was determined that engagement in serious violence increased the risk of death by suicide by twelvefold (Webb et al., 2012). Stack (2012) determined that persons dying by suicide were significantly more supportive of the norms governing the use of violence (verbal violence, violence against objects, and physical assaults against people) than persons dying natural deaths. This was true even after controlling for a host of other variables including psychiatric predictors.

Available estimates indicate that most suicide completions do not involve a previous suicide attempt: between 59% and 73% of people who die through suicide do so without a history of any previous suicide attempt (Witte et al., 2012). Following the leads of the present study, future work is needed to ascertain whether measures such as fighting, dating/domestic violence, and violence preparedness (weapon carrying) can distinguish suicide attempts from suicide completions.

References

  1. Top of page
  2. Abstract
  3. Differentiating Suicide Attempters from Suicide Ideators
  4. Methodology
  5. Results
  6. Theoretical Discussion
  7. Conclusion
  8. References
  9. Appendix 1
  • Anestis, M. D., Bagge, C. L., Tull, M. T., & Joiner, T. E. (2011). Clarifying the role of emotion dysregulation in the interpersonal-psychological theory of suicidal behavior in an undergraduate sample. Journal of Psychiatric Research, 45, 603611.
  • Centers for Disease Control. (2003). 2001 National school based youth risk behavior survey: Public use data documentation. Atlanta, GA: Author.
  • Duke, N. N., Pettingell, S. L., McMorris, B., & Borowsky, I. W. (2010). Adolescent violence perpetration: Association with multiple types of adverse childhood experiences. Pediatrics, 125, e778e786.
  • Eaton, D. K., Foti, K., Brener, N. D., Crosby, A. E., Flores, G., & Kann, L. (2011). Associations between risk behaviors and suicide attempts. Archives of Suicide Research, 15, 113126.
  • Gunn, J. F., Lester, D., & McSwain, S. (2011). Testing the warning signs of suicidal behavior among suicide ideators using the 2009 national survey on drug abuse and health. International Journal of Emergency Mental Health, 13, 147154.
  • Hinduja, S., & Patchin, J. W. (2010). Bullying, cyber bullying, and suicide. Archives of Suicide Research, 14, 206221.
  • Joiner, T. E. (2005). Why people die by suicide. Cambridge: Harvard University Press.
  • Joiner, T. E., Van Orden, K. A., Witte, T. K., Selby, E. A., Ribeiro, J. D., Lewis, R., et al. (2009). Main predictions of the interpersonal -psychological theory of suicidal behavior. Journal of Abnormal Psychology, 118, 634646.
  • Lester, D. (2000). Why people kill themselves. Springfield, IL: Charles Thomas.
  • Menard, S. (2002). Applied logistic regression analysis. Thousand Oaks, CA: Sage.
  • Niederkratenthaler, T., Sonneck, G., Dervic, K., Dervic, K., Nader, I. W., Voracek, M., et al. (2012). Predictors of suicide and suicide attempt in subway stations. Journal of Urban Health, 89, 339353.
  • Nock, M. K., & Marzuk, P. M. (2000). Suicide & violence. In K. Hawton & K. van Heeringen (Eds.), The international handbook of suicide and attempted suicide (pp. 437456). New York: John Wiley.
  • O'Connor, R. C., Platt, S., & Gordon, J. (Eds.) (2011). international handbook of suicide prevention. New York: Wiley-Blackwell.
  • Paluszny, M., Davenport, C., & Kim, W. J. (1991). Suicide attempts and ideation: Adolescents evaluated on a pediatric ward. Adolescence, 26, 209215.
  • Paternoster, R., Brame, R., Mazerole, P., & Piquero, A. (1997). Using the correct statistical test for the equality of regression coefficients. Criminology, 36, 859866.
  • Plutchik, R. (2000). Aggression, violence, & suicide. In R. Maris, A. Berman, & M. M. Silverman (Eds.), Comprehensive textbook of suicidology (pp. 407426). New York: Guilford.
  • Rasmussen, K. A., & Wingate, L. R. (2011). The role of optimism in the interpersonal-psychological theory of suicidal behavior. Suicide and Life-Threatening Behavior, 41, 137148.
  • Ribeiro, J. D., & Joiner, T. E. (2011). Present status and future prospects take up the interpersonal-psychological theory of suicidal behavior. In R. C. O'Connor, S. Platt, & J. Gordon (Eds.), International handbook of suicide prevention (pp. 169179). New York: Wiley-Blackwell.
  • Stack, S. (2000). Suicide: A 15-year review of the sociological literature. Suicide and Life-Threatening Behavior, 30, 145162.
  • Stack, S. (2012). The culture of violence and suicidality. In G. R. Hayes & M. H. Bryant (Eds.), Psychology of culture (pp. 201212). New York: Nova Science.
  • Swahn, M. H., Lubell, K. M., & Simon, T. R. (2004). Suicide attempts and physical fighting among high school students- United States- 2001. Morbidity and Mortality Weekly Report, 53, 474476.
  • Swahn, M. H., Simon, T. R., Hertz, M. F., Arias, I., Bossarte, R., Ross, J. G., et al. (2008). Linking dating violence, peer violence, and suicidal behavior among high risk youth. American Journal of Preventive Medicine, 34, 3038.
  • Wasserman, D., & Wasserman, C. (2009). Oxford textbook of suicidology and suicide prevention. New York: Oxford University Pres.
  • Webb, R., Shaw, J., Stevens, H., Mortensen, P. B., Appleby, L., et al. (2012). Suicide risk among violent and sexual criminal offenders. Journal of Interpersonal Violence, 27, 34053424.
  • Wetzler, S., Asnis, G. M., Hyman, R. B., Virtue, C., Zimmerman, J., & Rathus, J. H. (1996). Characteristics of suicidality among adolescents. Suicide and Life-Threatening Behavior, 26, 3745.
  • Witte, T. K., Didie, E. R., Menard, W., & Phillips, K. A. (2012). The relationship between body dysmorphic disorder behaviors and the acquired capability for suicide. Suicide and Life-Threatening Behavior, 43, 318331.
  • Wolfgang, M. (1958). Patterns in criminal homicide. Philadelphia: University of Pennsylvania Press.

Appendix 1

  1. Top of page
  2. Abstract
  3. Differentiating Suicide Attempters from Suicide Ideators
  4. Methodology
  5. Results
  6. Theoretical Discussion
  7. Conclusion
  8. References
  9. Appendix 1

Correlations Matrix. Pearson r correlations. Suicide Ideators and Attempters. Youth Risk Behavior Survey, 2001 N = 2,536

  YX1X2X3X4X5X6X7X8X9X10X11X12X13X14X15X16X17X18X19X20X21X22X23X24X25X26
  1. Correlations > .039, p < .05, two-tailed test.

  2. Y = Suicide Attempt. Violence Involvement: X1 = Fighting, X2 = Dating Violence, X3 = Weapon Carrying. Psychiatric Morbidity: X4 = Depression, X5 = Eating Disorder, X6 = Marijuana Use, X7 = Cocaine Use, X8 = Binge Drinking, X9 = Suicide plan. Risky Sexual Behavior: X10 = Number Sex Partners, X11 = Unsafe Sex, X12 = Pregnancy. School Integration: X13 = Sports Team, X14 = Low Grade Point Average. Demographic Controls: X15 = Male, X16 = Age. Race: X17 = African American, X18 = Hispanic, X19 = Asian, X20 = Other. Region, X21 = South, X22 = West, X23 = Midwest, X24 = Urban, X25 = Suburban.

Y1.0                          
X1.181.0                         
X2.13.211.0                        
X3.14.30.221.0                       
X4.17.05.10.061.0                      
X5.15.07.18.07.081.0                     
X6.08.21.16.31.05.091.0                    
X7.16.22.20.30.07.20.351.0                   
X8.06.14.14.24.07.10.37.311.0                  
X9.24.10.10.13.18.08.07.07.061.0                 
X10.16.26.30.33.08.20.37.42.31.891.0                
X11.08.06.16.11.08.08.13.16.18.06.401.0               
X12.12.16.21.19.09.09.19.23.16.04.37.261.0              
X13.00.04.01.02−07.02−04.00−01.00.00−06−.011.0             
X14.05.09.08.10.05.04.18.13.1100.14.06.07−.071.0            
X15−05.13.04.33−.08−.09.15.10.07.00.13−.03.00.08.101.0           
X16−0700.05.01.00−.04.07.00−07.00.15.16.09−.08−.02.091.0          
X17.02.04.04.00−.02−.04−01−03−11−03.08−.02.14.01.00.00.051.0         
X18.12.00−02−04.03.04−03.05.00−01.04.06.01−.02.02−.02.00−.221.0        
X19.01.00−02−02.00.00−03.00−02.01−.03−.02−.02−03.01.01.05−.07−.081.0       
X20.05.01.02.05.02.00.05.01.00.00.01.01−.01.04.01.00−.05−.14−.17−.051.0      
X21.00.00.04.03.01.01.01.03.01−03.09.05.05.00−.04.01.02.1400−.06−.051.0     
X22.02−02−03−01.00−.01−01.00.00.00−.04−.06−.06−.03.03.02−.01−19.19.08.08−.631.0    
X23−05.00−02−03.00.01.01−03.02.03−.08−.01−.01.04.00−.03.00.02−.19−.01−02−.38−.291.0   
X24.04.04.01−02.00−.02−01.00−03−02.01−.01.02.00−.02.02.04.10.16.00.00.00−.04.021.0  
X25−03−02−01.00.00.00.00−01.01.00.00.02.00.00.02.00−.01−.01−.09.03.00.00.07−.07−.811.0 
Mean.43.09.201.7.66.13.16.05.19.582.4.26.10.46.10.3716.1.15.20.03.10.44.33.15.38.511.0