Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Jan 1.
Published in final edited form as: J Clin Child Adolesc Psychol. 2016 Oct 12;47(1):131–141. doi: 10.1080/15374416.2016.1220313

Risk-Taking Behavior and Suicidality: The Unique Role of Adolescent Drug Use

Brooke A Ammerman 1, Laurence Steinberg 2, Michael S McCloskey 3
PMCID: PMC6082017  NIHMSID: NIHMS1500651  PMID: 27732082

Abstract

Objective:

Research has supported an association between suicidal thoughts / behaviors and risk-taking, which may be particularly strong during adolescence when risk-taking is known to increase. However, extant research has focused on individual risk-taking behaviors (e.g., alcohol use), limiting our ability to evaluate the unique association between different risk-taking behaviors and suicidal thoughts / behaviors. The current study aimed to fill this gap by examining the simultaneous influence of multiple risk-taking behaviors (i.e., risky sexual behavior, tobacco/alcohol use, illicit drug use, delinquent behavior, violent behavior) on adolescent suicidal thoughts / behaviors.

Method:

Data from the National Longitudinal Study of Adolescent Health was utilized. The sample consisted of 4,834 adolescents who completed home interviews at two time points. At the first time point, participants’ mean age was 15.15, with 48% (n = 2315) identifying as male. Participants provided information about suicidal thoughts / behaviors and multiple risk-taking behaviors at an initial interview and at a second interview, approximately 11 months later.

Results:

When independently examined, nearly all assessed risk-taking behaviors were independently associated with suicidal ideation concurrently and prospectively, and with suicide attempts concurrently. When all risk-taking behaviors were examined simultaneously, illicit drug use was the only significant concurrent and significant prospective, albeit negative, predictor of suicidal thoughts and only concurrent predictor of suicidal behavior.

Conclusions:

The current findings suggest that illicit drug use may have a stronger association with suicidal thoughts and behaviors than other risk-taking behavior. These findings have implications for prevention and intervention programs for adolescents.

Keywords: adolescents, risk-taking, suicide attempts, suicide ideation, drug use

Suicide is the second-leading cause of death among those aged 15–24, with nearly 5,500 youth aged 10–24 committing suicide each year (CDC, 2014). Further, it is believed that for every completed suicide in this age group there are between 100 and 200 suicide attempts (Goldsmith et al., 2002). One of the most well-cited risk factors for suicidal behavior is suicidal ideation; roughly one-third of those experiencing suicidal ideation go on to attempt suicide (Nock et al., 2009). Even more alarming is that these rates may be on the rise: according to the CDC’s Youth Risk Behavior Survey, the percentage of high school students who reported seriously considering suicide increased from 13.8% to 17.0% between 2009 and 2013, and the number who reported attempting suicide during the previous year increased from 6.3% to 8.0% during the same period (Centers for Disease Control and Prevention, n.d.).

Increasedsuicidal ideation and behavior is of concern not only due to the potential detrimental consequences, but also because adolescent suicidal ideation and behavior are associated with several other negative outcomes, both concurrently and prospectively. For example, at the time of suicidal thoughts and behaviors, 89% and 96% of individuals, respectively (Nock et al., 2013), meet criteria for at least one DSM mental disorder, primarily those related to elevated fear, anger, anxiety, depression, or substance use (Andover et al., 2012; Nock et al., 2013). Engaging insuicidal behaviors during adolescence is also predictive of later psychopathology, poorer adjustment, risky sex, and psychiatric treatment in adulthood (Briere et al., 2015). Given the highprevalence rates and related negative consequences, a rich literature has focused on identifying risk factors for suicide among adolescents. One area deserving of further attention is risk-taking.

Though several theories of adolescent risk-taking have been proposed (see Furby & Beyth-Maron, 1992; Millstein & Igra, 1995; Steinberg 2004), it is generally agreed that adolescence is a developmental period marked by the emergence and escalation of risk-taking (Collado et al., 2014; Gullo & Dawe, 2008; Smith et al., 2013; Steinberg, 2004), and that these behaviors increase from early to late adolescence (Arnett, 1992; Reyna & Farley, 2006). For example, increases in the use of tobacco, alcohol, and illicit drugs (Chen & Jacobson, 2012; National Institute on Drug Abuse, 2009), unprotected sex and sexually transmitted infections (CDC, 2012; Steinberg, 2008), driving crashes and fatalities (CDC, 2014a; National Highway Traffic Safety Administration, 2012), and violent and non-violent crimes (Pastore & Miguire, 2006; Piquero, 2008) have all been found during adolescence. Though some increase in risk-taking behavior in adolescence can be normative, high levels of risk-taking behaviorare associated with poorer academic achievement and role performance (Costa et al., 1995; Newcomb & McGee, 1991), increased psychopathology (Dishion, 2000; Tubman et al., 2003) and suicide risk (Pena, Matthieu, Zayas, Masyn, & Caine, 2012).

Extant research supports the association between suicidal behavior and several risk-taking behaviors during adolescence. Suicidal behavior in adolescence is linked to substance use, with adolescent girls beingthree times more likely and adolescent males being seventeen times more likely to attempt suicide while under the influence of alcohol (Groves, Stanley, & Sher, 2007; McManama et al., 2014; Wong, Zhou, Goebert, & Hishinuma, 2013). Adolescents who engage in risky acts of physical violence(Greening et al., 2010; Zhang et al., 2012) and/orcriminal behavior (Bauer et al., 2014; Bjorkenstam et al., 2011)are also at increased risk for suicidal behavior. Not surprisingly, two latent class analyses found that adolescents belonging to a higher risk-taking class, compared to adolescents in lower risk-taking classes showed an increased likelihood of attempting suicide (Sullivan, Childs, & O’Connell, 2010) as well as an increased likelihood of making multiple suicide attempts (Pena et al., 2012). Although these findings demonstrate a clear connection between risk-taking and suicidal thoughts and behaviors, the cross-sectional nature of the majority of findings leaves the directionality of these associations unclear.

Despite strong evidence of a link between risk-taking and suicidal behavior, little research has examined multiple risk-taking behaviors concurrently in the prediction of suicidal thoughts and behaviors. However, one study found that, among four latent classes of individuals, the class endorsing the greatest amount of risk-taking behaviors (i.e., drug use, drinking behavior, risky sexual behavior, violence perpetration) also reported the highest levels of suicide attempts (Thullen, Taliaferro, & Muehlenkamp, 2015). These risk-taking behaviors were more associatedwith suicide attempts thanother well-established correlates of suicidal thoughts and behaviors, such as maladaptive dieting and non-suicidal self-injury (Andover & Gibb, 2010). Thus, the consideration of multiple risk-taking behaviors may be integral to the prediction of suicidal thoughts and behaviors.

There is a dearth of literature examining the presence of multiple risk-taking behaviors as a predictor of suicidal behavior(e.g., Harel-Fisch, Abdeen, Walsh, Radwan, & Fogel-Grinvald, 2012; Thullen et al., 2015). This is highlighted by the fact that no study to date has simultaneously examined multiple risk-taking behavior. Thus, the current study aimed to simultaneously examine multiple risk-taking behaviors in association with suicidal thoughts and behaviors both contemporaneously and prospectively. It was expected that each type of risk-taking assessed in the present study (i.e., risky sexual behavior, tobacco and alcohol use, illicit drug use, delinquent behavior, violent behavior) will each be associated with the presence of suicidal ideation and the presence of suicide attempts. Such hypotheses are based on previous research suggesting that the aforementioned risk-taking behaviors are associated with attempted suicide (Thullen et al., 2015); given this, we would also expect such risk-taking behaviors to be associated with suicidal ideation. However, given the lack of previous research simultaneously examining risk-taking behaviors in relation to both suicidal ideation and attempts, no specific hypotheses were made when considering all risk-taking behaviors together.

Method

Sample

Data for the present analyses come fromthe National Longitudinal Study of Adolescent Health (Add Health;Harris et al., 2009), comprising 7th-12th grade adolescents from 134 U.S. schools. The study design consisted of four different waves; the present analyses only utilized data from Wave I and Wave II. The first wave of in-home computer-assisted interviews was conducted from April to December 1995 and completed by 20,745 students. From this in-home sample, 14,738 students completed the second wave of interviews, conducted between April and August 1996. The mean timebetween these two data collection points was 11 months. For the current analyses, the sample consisted only of adolescents who completed both the Wave I and Wave II in home interview (n= 14,738), and who had sample weights from Wave I (n= 4,834); this ensured that the final sample provided responses at both time points of interest (Wave I and Wave II), and, further, that we could obtain unbiased estimates of population parameters and standard errors from our analyses, reducing false-positive statistical test results (Chen & Chantala, 2014). The final sample consisted of 4,834 adolescents. In the final sample, participants’ ages at Wave I ranged from 11 to 21 years old, with a mean age of 15.15 (SD = 1.60), with 48% (n = 2315) identifying as male. The majority of the sample identified as Caucasian (67.5%), followed by African American (23.8%), Asian (4.0%), American Indian (3.8%), and “other” (6.3%;Participants were allowed to identify with more than one race.)Approximately 12% of the sample identified as Hispanic. At Wave I, 12.8% of participants endorsed having suicidal ideation and 3.7% endorsed having made a suicide attempt in the past 12 months. At Wave II, 10.8% of participants endorsed having suicidal ideation and 3.1% endorsed having made a suicide attempt in the past 12 months.

Measures

Risky sexual behavior.

The latent variable of risky sexual behavior comprised four different items assessing participant’s engagement in risky sexual activities: unprotected sexual intercourse with romantic partners, sexual intercourse withnonromanticand/or nonrelationship sexual partners, unprotected sexual intercourse with a nonromantic and/or nonrelationship partner, and engaging in solicited sexual intercourse. All items were coded dichotomously, where a score of 1 indicated the presence of the risky sexual behavior. The item of unprotected sex with a romantic partner was created through three questions. Participants were asked if they used birth control every time they had sexual intercourse with a romantic partner, allowing for identification of up to three romantic partners and asked about intercourse with each. If they denied using birth control with any partner, they received a score of 1. The same procedures were used to determine if the participant had engaged in unprotected sex with a nonromantic and/or nonrelationship partner. Participants were also asked if they had engaged in sexual intercourse with a nonromantic and/or nonrelationship partner, and if they had engaged in sexual intercourse in exchange for drugs or money. These items were coded as positive (1) if they reported ever engaging in the behavior. These questions were asked at both Wave I and Wave II and each assessed the lifetime occurrence of these behaviors.

Tobacco and alcohol use.

The latent variable of tobacco and alcohol use comprised five different items: two items were related to tobacco use and three items were related to alcohol use. All items were coded dichotomously, where a score of 1 indicated the presence of the behavior. With regard to tobacco use, participants were asked if they ever smoked cigarettes regularly (e.g., at least one cigarette every day for 30 days) and how old they were when they used chewing tobacco or snuff for the first time. The latter question was coded as positive (1) if the participant endorsed ever using chewing tobacco or snuff. With regard to alcohol use, participants were asked if they ever drink beer, wine, or liquor when they were not with their parents; on how many days in the past 12 months did they drink five drink or more in a row; and if they have ever driven while drunk. These three items were coded as positive if participants endorsed ever engaging in these behaviors. These questions were asked at both time points; the questions at Wave Iassessed the lifetime occurrenceof these behaviors, with the exception of having five or more drinks, and Wave II assessed occurrence since the month of the last interview (e.g., Wave I) or over the past 12 months.

Illicit drug use.

The latent variable of illicit drug use comprised five different items; all were coded dichotomously, where a score of 1 indicated the presence of the behavior. Participants were asked the number of times they had usedmarijuana, cocaine, inhalants, and other types of illegal drugs. If the participants endorsed ever using the drug, the category was coded as positive (1). Participants were also asked if they had ever driven while high on drugs. This item was coded as positive (1) if they reported ever engaging in the behavior, and was asked at both Wave I and Wave II. Questions at Wave I assessed the lifetime occurrence of these behaviors and Wave IIassessed occurrence since the month of the last interview (e.g., Wave I).

Delinquent behavior.

The latent variable of delinquent behavior comprised three items. All items were coded dichotomously, where a score of 1 indicated the presence of the behavior. Participants were asked how many times they engaged in the following behavior in the past 12 months: painted graffiti or signs on someone’s property or public place, deliberately damaged property that didn’t belong to them, and stole something worth more than $50. These questions were asked at both Wave I and Wave II; each time point assessed the occurrence of these behaviors in the previous 12 months.

Violent behavior.

The latent variable of violent behavior comprised four items. All items were coded dichotomously, where a score of 1 indicated the presence of the behavior. Participants were asked how many times they engaged in the following behaviors in the past 12 months: a physical fight, a physical fight where they were injured and had to be treated by a doctor or nurse, hurt someone else badly enough to need bandages or care from a doctor or nurse, and pulled a knife or a gun on someone. These questions were asked at both Wave I and Wave II; each time point assessed the occurrence of these behaviors in the previous 12 months.

Suicidal thoughts and behaviors.

Participant’s suicidal ideation and suicide attempts were assessed. All items were coded dichotomously, where 1 indicated the presence of the behavior. Participants were asked whether they had ever seriously thought about committing suicide in the past 12 months and how many times they had attempted suicide in the past 12 months. The items were coded as 1 if they endorsed ever contemplating or attempting suicide. These questions were asked at both Wave I and Wave II; each time assessing these thoughts / behaviors in the previous 12 months.

Data Analysis

Prior to running the primary analyses, a factor analysis was conducted to assess fit of the observed variables on the five proposed latent variables (e.g., risky sexual behavior, tobacco and alcohol use, drug use, delinquency behavior, violent behavior). Weighted least squares estimation was used because(a) the five latent variables were constructed with non-normally distributed items, and (b) the suicidal ideation and attempt outcomes variables were non-normally distributed (Muthen & Muthen, 1998–2012). Latent auto-regressive models were utilized to examine the influence of each risk-takinglatent variable, independent of other risk-takinglatent variables, on suicidal ideation and attempts at Wave Iand Wave II. Two models for each risk-taking latent variable were conducted, one examining suicidal ideation and the other examining suicide attempts. This resulted in 10 total models. For example, in the model predicting attempted suicide, the risk-taking latent variable at Wave I and Wave II (allowing for covariation) and attempted suicide at Wave I and Wave II (allowing for covariation) were included in the model. See Figure 1 for an example model. Thisallowed for consideration of the unique effect of the type of risk-taking on both suicidal ideation and attempted suicide (for overview see Geiser, 2012). Following this, structural equation models (SEM) predicting either suicidal ideation or attempted suicide were conducted including all risk-takinglatent variables, which were allowed to covary. Model fit was based on a root mean square error of approximation (RMSEA) below 0.06 and a comparative fit index (CFI) greater than 0.90 (Hu & Bentler, 1999; Steiger, 2007). Additionally, SEM models included the cross sectional grand sample weights from the Add Health Wave I data, based on recommendations put forth by Chen & Chantala (2014). SEM was also used to conduct exploratory analyses of separate models for females and males given differences in suicidal ideation and attempted suicide in preliminary analyses. All analyses were conducted in Mplus Version 7.0 (Muthen & Muthen, 1998–2012). By default in Mplus, robust estimation of standard errors and robust chi-square tests of model fit are provided (Muthen & Muthen, 1998–2012). An alpha level of .01 was utilized given the number of analyses conducted.

Figure 1.

Figure 1

Open in a new tab

ExampleLatent Auto-regressive Model

Results

Preliminary Results

Item endorsement data for all individual risk-taking behavior questions is presented in Table 1. Participant endorsement ranged from 1% (e.g., unprotected sex with nonromantic partner; solicited sex) to 37% (e.g., drinking without parents) at Wave I and from 2% (e.g., inhalant use) to 37% (e.g., drinking without parents) at Wave II. Suicidal ideation (SI) at Wave I [t (4789) = 2.04, p = .02], attempted suicide (SA) at Wave I [t (5829) = 1.47, p = .14], SI at Wave II [t (4797) = −0.74, p = .46], and SA at Wave II [t (5829) = −1.51, p = .13] did not differ by participant age. More females reported SI at Wave I, χ2 (1, N = 4791) = 44.11, p <.001, and Wave II, χ2 (1, N = 4799) = 45.16, p <.001, reported SA at Wave I, χ2 (1, N = 4831) = 27.43, p < .001. However, no differences were found for SA at Wave II, χ2 (1, N = 4831) = .87, p =.35.

Table 1.

Item Endorsement for Risk-Taking Behaviors at Wave I and (Wave II)

Risky Sexual Behavior Tobacco/Alcohol Use Illicit Drug Use Delinquent Behavior Violent Behavior
Unprotected Sex RP .05 (.14)
Number NRP .23 (.17)
Unprotected Sex NRP .01 (.04)
Solicited Sex .01 (.03)
Reg. Cigarette Use .18 (.21)
Tobacco Use .06 (.07)
Drink W/o Parent .37 (.37)
Drink 5+ .24 (.28)
Drive Drunk .05 (.05)
Marijuana Use .23 (.25)
Cocaine Use .03 (.03)
Inhalant Use .06 (.02)
Other Drug Use .08 (.06)
Drive High .06 (.06)
Graffiti .09 (.07)
Steal worth $50+ .05 (.05)
Damage Property .19 (.14)
Physical Fight .32 (.21)
Injure Someone .18 (.08)
Injure Self .09 (.04)
Pull Knife/Gun .05 (.05)
Open in a new tab

Note: Item endorsement represents the percentage of participants who endorsed each item at each Wave;Wave II item endorsement is presented in parentheses; RP = romantic partner; NRP = nonrelationship and/or nonromantic partner; Reg. = regular; W/o = without; 5+ = five or more drinks

Factor loadings.

Two separate factor analyses (Wave I and Wave II) were conducted to confirm that the observed variables loaded onto the theorizedrisk-taking latent variables. The model examining Wave I variables demonstrated good fit, χ2 = 903.19, p <.001, CFI = .97, RMSEA = .02. All items loaded above .30 on their specified latent variable and below .30 on all other latent variables. A similar pattern of results was found when examining Wave II variables. The model demonstrated good fit, χ2 = 1332.95, p <.001, CFI = .95, RMSEA = .03. All items loaded above .30 on their specified latent variables and below .30 on all other latent variables. See Table 2 for Wave I and Wave II factor loadings.

Table 2.

Standardized Factor Loadings for Risk Behaviors at Wave I(and Wave II)

Risky Sexual Behavior Tobacco/Alcohol Use Illicit Drug Use Delinquent Behavior Violent Behavior
Unprotected Sex RP .56 (.58)
Number NRP .74 (.94)
Unprotected Sex NRP .59 (.73)
Solicited Sex .61 (.33)
Reg. Cigarette Use .82 (.84)
Tobacco Use .44 (.36)
Drink W/o Parent .85 (.81)
Drink 5+ .89 (.93)
Drive Drunk .78 (.82)
Marijuana Use .88 (.92)
Cocaine Use .82 (.86)
Inhalant Use .60 (.51)
Other Drug Use .83 (.86)
Drive High .87 (.93)
Graffiti .80 (.84)
Steal worth $50+ .83 (.78)
Damage Property .73 (.86)
Physical Fight .72 (.76)
Injure Someone .75 (.84)
Injure Self .49 (.63)
Pull Knife/Gun .77 (.88)
Open in a new tab

Note: Wave II factor loadings presented in parentheses; RP = romantic partner; NRP = nonrelationship and/or nonromantic partner; Reg. = regular; W/o = without; 5+ = five or more drinks

Study variable associations.

At Wave I, correlations between risk-taking latent variables were all significant at p <.001, as were all correlations between risk domains atWave II. See Table 3. There were significant relations between each Wave Irisk-taking latent variable and the corresponding risk-taking latent variable at Wave II, ranging from b = .77 (risky sexual behavior at Wave I and Wave II) to b = .95 (violent behavior at Wave I and Wave II)(all p’s <.001). SI at Wave I predicted SI at Wave I (p <.001). However, neither SIat Wave I (p = .93) nor the presence of a SA(p = .25) at Wave Ipredicted the occurrence of SA at Wave II. SA at Wave I did not predict the presence of SI at Wave II (p = .96). See Table 4.

Table 3.

Correlations for Latent Variables

Risky Sex 1 Tob/Alc 1 Drug 1 Delinquent 1
Risky Sex 1 --
Tob/Alc Use 1 .64* --
Drug Use 1 .66* .83* --
Delinquent 1 .36* .48* .60* --
Violent 1 .49* .41* .46* .63*
Risky Sex 2 Tob / Alc 2 Drug 2 Delinquent 2
Risky Sex 2 --
Tob/Alc Use 2 .44* --
Drug Use 2 .44* .81* --
Delinquent 2 .18* .37* .48* --
Violent 2 .44* .46* .50* .66*
Open in a new tab

Note:

*

p<.001; risky sex = risky sexual behavior; tob/alc use = tobacco and alcohol use; drug use = illicit drug use; delinquent = delinquent behavior; violent = violent behavior

Table 4.

Predictive Relationships Between Study Variables

Regression Relationship b (SE) p-value
Risky Sex T1 → Risky Sex T2 .77 (.03) <.001
Tob/Alc Use T1 → Tob/Alc Use T2 .89 (.01) <.001
Drug Use T1 → Drug Use T2 .90 (.01) <.001
Delinquent T1 → Delinquent T2 .78 (.03) <.001
Violent T1 → Violent T2 .95 (.03) <.001
S. Ideation T1 → S. Ideation T2 .36 (.06) <.001
S. IdeationT1 → S. Attempt T2 .003 (.17) .93
S. Attempt T1 → S. Attempt T2 .04 (.10) .25
S. Attempt T1 → S. Ideation T2 −.01 (.22) .96
Open in a new tab

Note: Risky sex = risky sexual behavior; tob/alc use = tobacco and alcohol use; drug use = illicit drug use; delinquent = delinquent behavior; violent = violent behavior

Predicting Suicidal Ideation

Risky sexual behavior (RSB).

RSB at Wave I correlatedwith SI at Wave I, b = .22, SE = .03, OR = 1.30, p <.001, but did not predictat Wave II,b = .16, SE = .08, OR = 1.07,p = .04. At Wave IIRSBwas not associated withWave II SI, b = −.04, SE = .07, OR = 0.95, p = .56.

Tobacco and alcohol use (TA).

TA at Wave I correlated with Wave I SI, b = .32, SE = .03, OR = 1.51, p<.001, but did not predict Wave II SI, b = −.10, SE = .11, OR = 0.97,p =.34. At Wave II TAcorrelatedwith Wave IISI, b = .35, SE = .10, OR = 1.23,p =.001.

Illicit drug use (ID).

ID at Wave Icorrelatedwith SI at Wave I, b = .44, SE = .03, OR = 1.29, p<.001, butnegatively predicted SI atWave II, b= −.22, SE = .09, OR = 0.96,p = .01. Wave II IDcorrelatedwith SI at Wave II, b = .55, SE = .09, OR = 1.20,p<.001.

Delinquent behavior (DB).

DB at Wave Icorrelated with SI at Wave I, b = .32, SE = .04, OR = 1.3,p<.001, but did not predict SI at Wave II, b = .01, SE = .09 OR = 1.07,p = .88. At Wave IIDB correlated withSI at Wave II, b = .25, SE = .09, OR = 1.16p = .003.

Violent behavior (VB).

VB at Wave Icorrelated with SI at Wave I, b = .31, SE = .03, OR = 1.18, p<.001, and predicted SI at Wave II, b = −.51, SE = .13, OR = 0.87, p<.001. At Wave IIVB correlated with SI at Wave II, b = .71, SE = .13, OR = 1.72, p<.001.

Combined risk model.

ASEM model including all risk-takinglatent variables at Wave I and Wave II, and SI at Wave I and Wave IIdemonstrated good fit, χ2 = 49006.00, p <.001, CFI = .94, RMSEA = .03. SI at Wave Ipositively predicted SI at Wave II. Illicit drug use at Wave Iwas positively associated SI at Wave I and negatively predicted SI at Wave II. Illicit drug use and tobacco and alcohol use at Wave IIwere bothpositively associated with SI at Wave II. Once the impact of illicit drug use and tobacco and alcohol use on SI was taken into account, the other risk-takinglatent variables didnot predict SI. The combined model is presented in Figure 2. This model was also run for males and females separately, each producing a similar pattern of results(data available upon request). The one gender difference was thatfor males, Wave I illicit drug use was only positively associated SI at Wave I, whereas for females, illicit drug use positively predicted SI at Wave I and Wave II. Tobacco and alcohol use was not associated with SI.

Figure 2.

Figure 2

Open in a new tab

Predicting Suicidal Ideation Note: All values presented are standardized; Not all estimated parameters are represented in diagram due to space; ** p<.001; *, p<.01; T1 = Wave I; T2 = Wave II; risky sex = risky sexual behavior; tob&alc = tobacco and alcohol use; drug = illicit drug use; delinquency = delinquent behavior; violence = violent behavior

Predicting Attempted Suicide

Risky sexual behavior (RSB).

Wave I RSBcorrelated with Wave I SA, b = .28, SE = .05, OR = 1.71,p<.001, but did not predict SA at Wave II, b = −.09, SE = .13, OR = 0.89, p=.50. At Wave II RSB was notassociated with Wave II SA, b=.11, SE=.13, OR=1.0, p = .42.

Tobacco and alcohol use (TA).

Wave ITA correlated with SA at Wave I, b = .32, SE = .04, OR = 1.54,p<.001, but did not predict SA at Wave II, b = −.03, SE = .17, OR = 0.96,p = .86. At Wave II TAand SAwere not associated, b = −.11, SE = .16, OR = 0.93,p = .49.

Illicit drug use (ID).

Wave IID correlated with SA at Wave I, b = .44, SE = .04, OR = 1.32, p< .001, but did not predict SA at Wave II, b = .22, SE = .19, OR = 1.03, p = .26. At Wave II IDwas notassociatedwith SA at Wave II, b = −.26, SE = .19, OR = 0.96,p = .16.

Delinquent behavior (DB).

Wave I DBcorrelated with Wave I SA, b = .26, SE = .00, OR = 1.24,p<.001, but did not predict Wave II SA, b = .08, SE = .12, OR = 1.08,p = .68. At Wave II DB andSA were not associated, b = −.01, SE = .12, OR = 0.98, p = .92.

Violent behavior (VB).

Wave I VBcorrelated with SA at Wave I, b = .24, SE = .05, OR = 1.24, p<.001, but did not predict SA at Wave II, b = .01, SE = .16, OR – 1.07, p = .95. At Wave II VB did notcorrelatewith Wave II, b = −.01, SE = .18, OR = 0.96, p = .95.

Combined risk model.

ASEM model including all risk-takinglatent variables at Wave I and Wave II, and SA at Wave I and Wave II was then tested. This model demonstrated good fit, χ2 = 48217.00, p <.001, CFI = .94, RMSEA = .03. Illicit drug use at Wave Iwas positively associatedwith SA at Wave I. No other risk-takinglatent variables predictedSA at Wave I or Wave II once illicit drug use was taken into account. The combined model is presented in Figure 3. This model was also run for males and females separately, with each gender producing a similar pattern of results (data available upon request). The one gender difference was that illicit drug use at Wave I was positively associated with SA at Wave I only for males.

Figure 3.

Figure 3

Open in a new tab

Predicting Suicide Attempts Note: All values presented are standardized; Not all estimated parameters are represented in diagram due to space; ** p<.001; *, p<.01; T1 = Wave I; T2 = Wave II; risky sex = risky sexual behavior; tob&alc = tobacco and alcohol use; drug = illicit drug use; delinquency = delinquent behavior; violence = violent behavior

Discussion

The purpose of the current study was to examine the influence of risk-taking behaviors on suicidal thoughts and behaviors in adolescence. Further, we assessed these relationships when considering multiple risk-taking behaviors across two time points. Our hypotheses were partially supported. Nearly all risk-taking behaviors, when examined individually, were associated with (a) concurrent suicidal ideation at each time point, and (b) the presence of a suicide attemptat the first time point. However, when risk-taking behaviors were examined simultaneously, illicit drug use and tobacco and alcohol use werethe only concurrent predictors and illicit drug use the only prospective predictorassociated with suicidal ideation; illicit drug use was the only concurrent predictor of attempted suicide and no risk-taking behaviors were prospective predictors of attempted suicide.

Illicit drug use was independently related to suicidal ideation and attemptsacross both time points. Although illicit drug use was positively associated with suicidal ideation concurrently, contrary to our predictions, illicit drug use at Wave I was a negative prospective predictor of suicidal ideation at Wave II in both the independent and combined models. That is, greater illicit drug use led to lower suicidal ideation one year later. One possible explanation of this finding is that those individuals with greater illicit drug use (and related suicidal ideation) at Wave I were more likely to receive psychological treatment prior to Wave II. Supporting this notion, those with suicidal ideation at Wave I were in fact more likely to report having received counseling at Wave II. Examination of this unexpected relationship should be the focus of attention in future research.

On the other hand, after considering other risk-taking behaviors in the model, illicit drug use was the only risk-taking behavior to concurrently predict attempted suicide. This supports a strong literature on the association between drug use and suicide attempts and completed suicides (Liu, Case, & Spirit, 2014; Wilcox, Conner, & Caine, 2004). The current findings further suggest that illicit drug use may have a stronger association with suicide attempts than other risk-taking behavior, which is in line with past research proposing that illicit drug use may be more important than tobacco and alcohol use in suicidal behavior(Wong et al., 2013). The current study partially supports this observation. It was found that, when examined independently, tobacco and alcohol use was concurrently related to suicide attempts at Wave I; however, after considering the influence of illicit drug use, tobacco and alcohol use was no longer related to suicide attempts. There are a few possible explanations of these findings. One theory of suicide proposes that suicide is most likely to occur when an individual both desires to commit suicide and has the capability to do so (e.g., increased tolerance of pain and reduced fear of death), often achieved through prior exposure to painful and provocative experiences (Joiner, 2005). Although intravenous drug use may have a more direct relationship to increased pain tolerance and suicidal behavior (Liu et al., 2014) as compared to other illicit drug use, it was not directly assessed in the current study. However, illicit drug use as assessed may still be associated with attenuated fear about death, given its potentially lethal direct and indirect consequences (e.g., overdose, car accidents). If so, it is possible that habituation to fear of death through illicit drug use increases one’s acquired capability for suicidal behavior (Riberio et al., 2014). If acquired capability were the driving factor for the observed connection between illicit drug use and suicidal behavior, however, we would also expect that other potentially detrimental / painful risk-taking behaviors, such as violent behavior, would have demonstrated a similar pattern of results. It is possible that the internalizing nature of illicit drug use may be an important characteristic. Both illicit and licit drug use have been implicated in the self-medication hypothesis, suggesting that substance use may be used to self-soothe in the face of distressful psychological states (Khantzian, 2003). There is a long line of research supporting the role of substance use in depression (e.g., Markou, Kosten & Koob, 1998), and it is possible this extends to suicidal behavior. Future research is needed to better understand these potential relationships and to test the aforementioned theoretical frameworks.

When examining risk-taking behaviors independently, all risk-taking behaviors were concurrently associated with suicidal ideation at Wave I and Wave II (excluding risky sexual behavior). In addition to illicit drug use and tobacco and alcohol use, the presence of violent behavior at Wave Ialsoindependently, prospectively predicted suicidal ideation at Wave II. These findings are not surprising. Risk-taking behaviors have been previously found as risk factors for suicidal behaviors (e.g. Pena et al., 2012, Sullivan et al., 2010, Thullen et al., 2015), so it would be expected that the same behaviors would be associated with suicidal ideation. Similarly, all risk-taking behaviors concurrently predicted attempted suicide at Wave I; however, no risk-taking behaviors concurrently or prospectively predicted attempted suicide at Wave II.

As expected, suicidal ideation predicted the presence of suicidal ideation approximately one year later. Unexpectedly, however, neither suicidal ideation nor suicide attempts predicted suicide attempts one year later. These findings, particularly those related to suicide attempts, are inconsistent withpreviousliterature that suggested past (suicidal) behavior is the best predictor of future behavior (e.g., Goldston, Daniel, Reboussin, Reboussin, Frazier, & Kelley, 1999; Lewinsohn, Rohde, & Seely, 1994). It is possible that given the age range of the current sample, and likelihood a caregiver is still involved in their daily living, that these suicide attempts had been brought to the attention of the caregiver. This may have led to the adolescent receiving help for their thoughts and behaviors, thus, potentially mitigating future suicidal behavior. As previously noted, those who reported a suicide attempt at Wave I were more likely to have reported receiving counseling at Wave II. It is also possible that the lack of relationship between suicide attempts at the two time points is due to the short time-span (11 months) between data collections, which may not have allowed enough time to pass for the behavior to re-occur.

It was also surprising that gender differences were not present in relationships between the assessed risk-taking behaviors and suicidal ideation or attempts. Not only does previous research suggest that males are more likely to exhibit specific risk-taking behavior, such as violent behavior (Bjorkqvist, 1994), but the relationship between alcohol use and suicide attempts may be stronger among males (Groves, Stanley, & Sher, 2007; McManama et al., 2014; Wong, Zhou, Goebert, & Hishinuma, 2013). As such, we would have anticipated gender differences to emerge. It is possible that the current study did not find such gender differences due to the consideration of multiple risk-taking behaviors in the model, negating potential differences in any one behavior. Another potential explanation is that differences may emerge in the pattern of results if the frequency or lethality of suicide attempts was examined, especially given that males are more likely to use more lethal methods (Beautrias, 2003). The current findings do suggest, however, the relationship between illicit drug use and suicide attemptsis important to consider among both male and female adolescents.

There are a few limitations of note in the current study. First, not all risk-taking behaviors were assessed across the same time period. For example, risky sexual activity was assessed across the lifetime at both time points whereas other behaviors, such as drug use, were assessed in the 12 months prior to the interview. This may have influenced the pattern of findings not only for risky sexual behavior (which may have evidenced stronger associations if examined in 12-month increments), but also in relation to other risk-taking behavior. Although the pattern of findings for the combined models did not differ after removing risky sexual behavior from the model, this limitation should be considered in future research. Further, not all risk-taking behavior was assessed along the same metric; cigarette use was assessed as daily use whereas other tobacco products were assessed for lifetime use. Future research should better assess for and differentiate among those risk-taking behaviors occurring across the lifetimeand those within the time period during which suicidal ideation and attempts are assessed, in addition to using continuous (as opposed to dichotomous) measures of risk-taking behavior. Further, it is possible that individuals may have endorsed engaging in a risk-taking behavior (e.g., illicit drug use) as a means to attempt suicide. Such relationships should be further explored as they may represent a unique risk for suicidal behavior. The current data utilized only single-item self-report measures of suicidal ideation and attempts, and, further, did not consider suicidal ideation severity or multiple suicide attempts, an area necessary for future research. Additionally, there may have been a third variable at play (i.e., depression) influencing the assessed relationships that will be important for future research to consider. Given the long-standing link between depression and suicidal behavior (e.g., Minkoff, Begman, Beck & Beck, 1973) and risk-taking behavior (e.g., Richardson, Radziszewka, Dent, & Flay, 1993), depression may mediate the associations observed here. Finally, it should be noted that the current dataset was collected in the mid-1990’sin a community sample; the findings presented should be replicated in an updated sample and further explored in clinical populations.

Despite these limitations, the current study has significant strengths, including the longitudinal design and the use of a nationally representative sample of adolescents. Further, the findings have important implications for suicide risk assessment during this developmental period; for example, it may be important to assess for current illicit drug use, as the findings highlight this risk-taking behavior in current risk for suicidal behavior. However, the unexpected finding between illicit drug use and suicidal ideation should be examined in future research, particularly as it relates to suicide risk. Overall, the findings emphasize the importance of considering all types of risk-taking in the assessment of suicidal thoughts and attempts.

Acknowledgements

This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health website (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01-HD31921 for this analysis.

FUNDING

This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01-HD31921 for this analysis. Brooke A. Ammerman was supported by National Research Service Award F31MH107156 from the National Institute of Mental Health.

Contributor Information

Brooke A. Ammerman, Temple University

Laurence Steinberg, Temple University.

Michael S. McCloskey, Temple University

References

  1. Adams EH, Blanken AJ, Ferguson LEX, & Kopstein A (1990). Overview of selected drug trends. Rockville, MD: National Institute on Drug Abuse. [Google Scholar]
  2. Arnett J (1992). Reckless behavior in adolescence: a developmental perspective. Developmental Review, 12, 339–373. [Google Scholar]
  3. Andover MS, & Gibb BE (2010). Non-suicidal self-injury, attempted suicide, and suicidal intent among psychiatric inpatients. Psychiatry Research,178, 101–105. [DOI] [PubMed] [Google Scholar]
  4. Andover MS, Morris BW, Wren A, & Bruzzese ME (2012). The co-occurrence of non-suicidal self-injury and attempted suicide among adolescents: distinguishing risk factors and psychosocial correlates. Child and Adolescent Psychiatry and Mental Health, 6, 1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bauer RL, Chesin MS, & Jeglic EL (2014). Depression, delinquency, and suicidal behaviors among college students. Crisis: The Journal of Crisis Intervention and Suicide Prevention, 35, 36. [DOI] [PubMed] [Google Scholar]
  6. Beatrais AL (2003). Suicide and serious suicide attempts in youth: A multiple-group comparison study. The American Journal of Psychiatry, 160, 1093 –1099. [DOI] [PubMed] [Google Scholar]
  7. Björkqvist K (1994). Sex differences in physical, verbal, and indirect aggression: A review of recent research. Sex Roles, 30, 177–188. [Google Scholar]
  8. Björkenstam E, Björkenstam C, Vinnerljung B, Hallqvist J, & Ljung R (2011). Juvenile delinquency, social background and suicide—a Swedish national cohort study of 992 881 young adults. International Journal of Epidemiology, 40, 1585–1592. [DOI] [PubMed] [Google Scholar]
  9. Brière FN, Rohde P, Seeley JR, Klein D, & Lewinsohn PM (2014). Adolescent suicide attempts and adult adjustment. Depression and Anxiety, 4, 270–276. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Centers for Disease Control and Prevention. (2014). National Center for Injury Prevention and Control. Web-based Injury Statistics Query and Reporting System (WISQARS) [online]. [cited 2016 July]. Available fromwww.cdc.gov/injury/wisqars/index.html
  11. Centers for Disease Control and Prevention. (2012). Youth risk behavior surveillance—United States, 2011. MMWR. Surveillance Summaries 2012:61 (no.SS-4). Available from www.cdc.gov/mmwr/pdf/ss/ss6104.pdf [PubMed] [Google Scholar]
  12. Centers for Disease Control and Prevention (CDC). (N.d.) Trends in the prevalence of suicide-related behaviors: National YRBS 1991–2009. Retrieved April 2015, from http://www.cdc.gov/HealthyYouth/yrbs/pdf/us_suicide_trend_yrbs.pdf
  13. Centers for Disease Control and Prevention (CDC). (2014a). Youth Risk Behavior Surveillance—United States, 2013. Morbidity and Mortality Weekly Report, 63, 1–172.24402465 [Google Scholar]
  14. Center for Disease Control and Prevention (CDC). (2014b). Teen drivers. Injury prevention &control: Motor vehicle safety. Retrieved from: http://www.cdc.gov/motorvehiclesafety/teen_drivers/index.html
  15. Center for Disease Control and Prevention (CDC). (2013). Ten leading causes of death and injury. Web-based Injury Statistics Query and Reporting System (WISQARS) [online]. Available from: http://www.cdc.gov/injury/wisqars/leadingcauses.html.
  16. Chen P & Chantala K (2014). Guidelines for analyzing Add Health data. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill. [Google Scholar]
  17. Chen P, & Jacobson KC (2012). Developmental trajectories of substance use from early adolescence to young adulthood: Gender and racial/ethnic differences. Journal of Adolescent Health, 50, 154–163. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Collado A, MacPherson L, Kurdziel G, Rosenberg LA, & Lejuez CW (2014). The relationship between puberty and risk taking in the real world and in the laboratory. Personality and individual differences, 68, 143–148. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Costa FM, Jessor R, Donovan JE, & Fortenberry JD (1995). Early initiation of sexual intercourse: The influence of psychosocial unconventionality. Journal of Research in Crime and Delinquency, 5, 93–121. [Google Scholar]
  20. Dishion TJ (2000). Cross-setting consistency in early adolescent psychopathology: Deviant friendships and problem behavior sequalae. Journal of Personality, 68, 1109–1126 [DOI] [PubMed] [Google Scholar]
  21. Farrell AD, Danish SJ, & Howard CW (1992). Relationship between drug use and other problem behavior sin urban adolescents. Journal of Consulting and Clinical Psychology, 60, 705–712. [DOI] [PubMed] [Google Scholar]
  22. Furby L, & Beyth-Marom R (1992). Risk taking in adolescence: A decision-making perspective. Developmental Review, 12, 1–44. [Google Scholar]
  23. Geiser C (2012). Data analysis with Mplus Guilford Press.Goldsmith SK, Pellmar TC, Kleinman AM, & Bunney WE (2002). Reducing suicide: A national imperative. Washington, DC: National Academy Press. [PubMed] [Google Scholar]
  24. Goldston DB, Daniel SS, Reboussin DM, Reboussin BA, Frazier PH, & Kelley AE (1999). Suicide attempts among formerly hospitalized adolescents: A prospective naturalistic study of risk during the first 5 years after discharge. Journal of the American Academy of Child & Adolescent Psychiatry, 38, 660–671. [DOI] [PubMed] [Google Scholar]
  25. Greening L, Stoppelbein L, Luebbe A, & Fite PJ (2010). Aggression and the risk for suicidal behaviors among children. Suicide and Life-Threatening Behavior, 40, 337–345. [DOI] [PubMed] [Google Scholar]
  26. Groves SA, Stanley BH, & Sher L (2007). Ethnicity and the relationship between adolescent alcohol use and suicidal behavior. International Journal of Adolescent Medicine and Health, 19, 19–25. [DOI] [PubMed] [Google Scholar]
  27. Gullo MJ, & Dawe S (2008). Impulsivity and adolescent substance use: Rashly dismissed as ‘all-bad?’. Neuroscience and Biobehavioral Reviews, 32, 1507–1518. [DOI] [PubMed] [Google Scholar]
  28. Harris KM, Halpern CT, Whitsel E, Hussey J, Tabor J, Entzel P, and Udry JR.(2009). The National Longitudinal Study of Adolescent to Adult Health: Research Design [WWW document]. URL: http://www.cpc.unc.edu/projects/addhealth/design.
  29. Hu LT & Bentler PM (1999). Cutoff Criteria for Fit Indexes in Covariance Structure Analysis: Conventional Criteria Versus New Alternatives. Structural Equation Modeling, 6, 1–55. [Google Scholar]
  30. Joiner T (2005). Why people die by suicide. Cambridge, MA, US: Harvard University Press. [Google Scholar]
  31. Khantzian EJ (2003). Understanding addictive vulnerability. Neuro-Psychoanalysis, 5, 5–12. [Google Scholar]
  32. Klonsky ED, & May AM(2015). The Three-Step Theory (3ST): A new theory of suicide rooted in the “ideation-to-action” framework. International Journal of Cognitive Therapy, 8, 114–129. [Google Scholar]
  33. Lewinsohn PM, Rohde P, & Seeley JR (1994). Psychosocial risk factors for future adolescent suicide attempts. Journal of Consulting and Clinical Psychology, 62, 297–305. [DOI] [PubMed] [Google Scholar]
  34. Lowry R, Crosby A, Brener N, & Kann L (2014). Suicidal thoughts and attempts among U.S. high school students: Trends and associated health-risk behaviors, 1991–2011. Journal of Adolescent Health, 54, 100–108. [DOI] [PubMed] [Google Scholar]
  35. Liu RT, Case BG, & Spirito A (2014). Injection drug use is associated with suicide attempts but not ideation or plans in a sample of adolescents with depressive symptoms. Journal of PsychiatricResearch, 56, 65–71. [DOI] [PubMed] [Google Scholar]
  36. Markou A, Kosten TR, & Koob GF (1998). Neurobiological similarities in depression and drug dependence: A self-medication hypothesis. Neuropsychopharmacology, 18, 135–174. [DOI] [PubMed] [Google Scholar]
  37. McManama O’Brien KH, Becker SJ, Spirito A, Simon V, & Prinstein MJ (2014). Differentiating adolescent suicide attempters from ideators: Examining the interaction between depression severity and alcohol use.Suicide and Life-Threatening Behavior, 44, 23–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Millstein SG, & Igra VI (1995). Theoretical models of adolescent risk-taking behavior. Adolescent Health Problems: Behavioral Perspectives, 52–71. [Google Scholar]
  39. Minkoff K, Bergman E, Beck AT, & Beck R (1973). Hopelessness, depression, and attempted suicide. American Journal of Psychiatry, 130, 455–459. [DOI] [PubMed] [Google Scholar]
  40. Muehlenkamp JJ, & Gutierrez PM (2007). Risk for suicide attempts among adolescents who engage in non-suicidal self-injury. Archives of Suicide Research, 11, 69–82. [DOI] [PubMed] [Google Scholar]
  41. Muthén LK & Muthén BO (1998–2012). Mplus User’s Guide. Seventh Edition. Los Angeles, CA: Muthén & Muthén [Google Scholar]
  42. National Institute on Drug Abuse (2009). Monitoring the Future National Results on Adolescent Drug Use: Overview of Key Findings Bethesda, MD. [Google Scholar]
  43. National Highway Traffic Safety Administration (2012). Traffic safety facts: Young drivers (technical report no DOT HS 811 622) Washington, DC: National Highway Traffic Safety. [Google Scholar]
  44. Newcomb MD, & McGee L (1991). Influence of sensation seeking on general deviance and specific problem behaviors from adolescence to young adulthood. Journal of Personality and Social Psychology, 61, 614–628. [DOI] [PubMed] [Google Scholar]
  45. Nock MK, Green JG, Hwang I, McLaughlin KA, Sampson NA, Zaslavsky AM, &Kessler RC (2013). Prevalence, correlates, and treatment of lifetime suicidal behavior among adolescents: results from the National Comorbidity Survey Replication Adolescent Supplement. JAMA Psychiatry, 70, 300–310. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Nock MK, Hwang I, Sampson N, Kessler RC, Angermeyer M, Beautrais A,... Williams DR (2009). Cross-national analysis of the associations among mental disorders and suicidal behavior: Findings from the WHO World Mental Health Surveys. PLoS Med, 6, 1–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. O’Connor RC (2011). Towards an integrated motivational–volitional model of suicidal behaviour In: O’Connor RC, Platt S, Gordon J (Eds,), International Handbook of Suicide Prevention: Research, Policy and Practice.Wiley-Blackwell, Hoboken, NJ, pp. 181–198. [Google Scholar]
  48. Pastore AL, &Maguire K (2006). Estimated percent distribution of U.S. resident population and persons arrested for all offenses. [Accessed April, 2015];Sourcebook of Criminal Justice Statistics Online, Table 4.4.2006. 2006 Available at:http://www.albany.edu/sourcebook/pdf/t442006.pdf.
  49. Pena JB, Matthieu MM, Zayas LH, Masyn KE, & Caine ED (2012). Co-occurring risk behaviors among White, Black, and Hispanic US high school adolescents with suicide attempts requiring medical attention, 1999–2007: Implications for future prevention initiatives. Social psychiatry and psychiatric epidemiology, 47, 29–42. [DOI] [PubMed] [Google Scholar]
  50. Piquero AR (2008). Taking stock of developmental trajectories of criminal activity over the life course In The long view of crime: A synthesis of longitudinal research (pp. 23–78). Springer New York. [Google Scholar]
  51. Richardson JL, Radziszewska B, Dent CW, & Flay BR (1993). Relationship between after-school care of adolescents and substance use, risk taking, depressed mood, and academic achievement. Pediatrics, 92, 32–38. [PubMed] [Google Scholar]
  52. Reyna VF, & Farley F (2006). Risk and rationality in adolescent decision making: Implications for theory, practice, and public policy. Psychological Science in the Public Interest, 7, 1–44. [DOI] [PubMed] [Google Scholar]
  53. Rutter P, & Soucar E (2002). Youth suicide risk and sexual orientation. Adolescence, 37, 289–299. [PubMed] [Google Scholar]
  54. Shaffer D, & Craft L (1999). Methods of adolescent suicide prevention. Journal of Clinical Psychiatry, 60, 70–74. [PubMed] [Google Scholar]
  55. Smith AR, Chein J, & Steinberg L (2013). Impact of socio-emotional context, brain development, and pubertal maturation on adolescent risk-taking. Hormones and behavior, 64, 323–332. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Steiger JH (2007). Understanding the limitations of global fit assessment in structural equation modeling. Personality and Individual Differences, 42, 893–98. [Google Scholar]
  57. Steinberg L (2004). Risk taking in adolescence: what changes, and why?.Annuals of the New York Academy of Sciences, 1021, 51–58. [DOI] [PubMed] [Google Scholar]
  58. Steinberg L (2008) A social neuroscience perspective on adolescent risk-taking. Developmental Review, 28, 78–106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Steinberg L (2010). A dual systems model of adolescent risk‐taking. Developmental psychobiology, 52, 216–224. [DOI] [PubMed] [Google Scholar]
  60. Sullivan CJ, Childs KK, & O’Connell D (2010). Adolescent risk behavior subgroups: An empirical assessment. Journal of youth and adolescence, 39, 541–562. [DOI] [PubMed] [Google Scholar]
  61. Thullen MJ, Taliaferro LA, & Muehlenkamp JJ (In Press). Suicide Ideation and Attempts Among Adolescents Engaged in Risk Behaviors: A Latent Class Analysis. Journal of Research on Adolescence. [DOI] [PubMed] [Google Scholar]
  62. Tubman JG, Gil AG, Wagner EE, & Artigues H (2003). Patterns of sexual risk behaviors and psychiatric disorders in a community sample of young adults. Journal of Behavioral Medicine, 26, 473–500. [DOI] [PubMed] [Google Scholar]
  63. Van Orden KA, Witte TK, Gordon KH, Bender TW, & Joiner TE Jr (2008). Suicidal desire and the capability for suicide: tests of the interpersonal-psychological theory of suicidal behavior among adults.Journal of consulting and clinical psychology, 76, 72. [DOI] [PubMed] [Google Scholar]
  64. Wilcox HC, Conner KR, & Caine ED (2004). Association of alcohol and drug use disorders and completed suicide: an empirical review of cohort studies. Drug and alcohol dependence, 76, S11–S19. [DOI] [PubMed] [Google Scholar]
  65. Willis LA, Coombs DW, Cockerham WC, & Frison SL (2002). Ready to die: A postmodern interpretation of the increase of African-American adolescent male suicide. Social Science Medicine, 55, 907–920. [DOI] [PubMed] [Google Scholar]
  66. Wong SS, Zhou B, Goebert D, & Hishinuma ES (2013). The risk of adolescent suicide across patterns of drug use: a nationally representative study of high school students in the United States from 1999 to 2009. Social psychiatry and psychiatric epidemiology, 48, 1611–1620. [DOI] [PubMed] [Google Scholar]
  67. Zhang P, Roberts RE, Liu Z, Meng X, Tang J, Sun L, & Yu Y (2012). Hostility, physical aggression and trait anger as predictors for suicidal behavior in Chinese adolescents: A school-based study. PLoS One, 7, 1–5. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES