youth suicidal behavior even after controlling for other contributory factors such as parental psychopathology (see Johnson et al., 2002; Wagner, 1997; Wagner et al., 2003).

Child sexual abuse has also been found to be associated with increased risk for suicidal behavior (Johnson et al., 2002; Silverman et al., 1996), as well as with many other adverse psychological outcomes (Fergusson, Horwood, & Lynskey, 1996.). Some of the suicide risk conferred by child sexual abuse is likely related more generally to parental psychopathology, although one longitudinal study that controlled for many other risk factors (Fergusson et al., 1996) identified a unique contribution of this variable to suicidality.

Stressful life events have been found to be associated with suicide deaths (Beautrais, 2001; Brent, Perper, & Moritz, 1993; Gould et al., 1996; Marttunen, Aro, & Lönnqvist, 1993) and suicide attempts (Beautrais, Joyce, & Mulder, 1997; Fergusson, Woodward, & Horwood, 2000; Lewinsohn et al., 1996). Studies of young suicide attempters suggest that the type of stressor associated with suicidal behavior is age related, with suicidal behavior in younger adolescents most frequently precipitated by conflicts with parents, and in older adolescents by interpersonal loss, in particular the loss of a romantic relationship (Brent et al., 1999; Groholt et al., 1998). Interpersonal loss has also been identified as a significant stressor among youth with substance abuse problems (Brent et al., 1993b; Gould, Greenberg, et al., 2003; Marttunen, Aro, Henriksson, & Lönnqvist, 1994).

Bullying has also been suggested as a precipitant, with at least one study finding both victims and perpetrators to be more likely to engage in suicidal behavior than youth not involved in bullying (Kaltiala-Heino, Rimpela, Marttunen, Rimpela, & Rantanen, 1999). There appear to be commonalities between bullying and other forms of social maligning, such as those reported by gay and lesbian youth. Although media accounts of suicide among young people frequently allude to bullying as a cause, scientific evidence for this is lacking. No studies of bullying have controlled for the presence of other risk factors in suicide victims, in particular psychopathology, and to date in the United States, no systematic suicide prevention efforts have targeted this factor.

Legal or disciplinary problems have been found to precipitate suicidal behavior in youth with conduct disorder and other disruptive disorders (Brent, Perper, & Moritz, 1993; Gould et al., 1996). Academic difficulties have also been found to be associated with increased risk for suicidal behavior. Several studies have demonstrated increased suicidal ideation or behavior among students at risk for dropping out of high school (Beautrais, Joyce, & Mulder, 1996; Thompson & Eggert, 1999; Wunderlich et al., 1998). Gould et al. (1996) also reported that school problems and not being in school or in a work situation pose considerable risk for suicide death.

There is evidence that young people are particularly vulnerable to the impact of suicide contagion, whether through media coverage or direct knowledge of a peer's suicide (Gould, 2001; Gould, Jamieson, & Romer, 2003). Research has described outbreaks or clusters of both suicide deaths (Gould, Wallenstein, & Kleinman, 1990; Gould, Wallenstein, Kleinman, O'Carroll, & Mercy, 1990) and attempted suicides (Gould, Petrie, Kleinman, & Wallenstein, 1994) among young people. Having a friend who has attempted suicide has been found to discriminate depressed adolescents who themselves make a suicide attempt from those who do not (Lewinsohn, Rohde, & Seeley, 1994). The causality of the association between a youngster's knowledge of a friend or family member who attempts suicide and the youth's subsequent suicidal behavior has been supported using two waves of data from the National Longitudinal Study of Adolescent Health (Add Health), a nationally representative study of youth in grades 7 through 12 (Cutler, Glaeser, & Norberg, 2001). Despite some research reporting no association between media reporting and subsequent suicides (Mercy et al., 2001), the evidence of the significant impact of media coverage on suicide is ample and continues to grow (see Gould, 2001; Pirkis & Blood, 2001a, 2001b; Schmidtke & Shaller, 2000; Stack, 2000).

Over half of the 4,000 youths aged 15–24 years who died by suicide in 2002 used firearms (CDC, n.d.a), and there is strong evidence that firearms used for both suicides and unintentional injuries by adolescents are mainly from the home environment (Bailey et al., 1997; Beautrais et al., 1996; Brent et al., 1993d; Grossman, Reay, & Baker 1999; Shah, Hoffman, Wake, & Marine, 2000).

Family firearm ownership has been found to correlate with the youth suicide rate for 15-to 24-year-olds (Birckmayer & Hemenway, 2001). Almost three quarters of youth suicides and suicide attempts have been found to involve the use of a firearm belonging to a household member; in more than half of the cases, a parent's gun was involved (Grossman et al., 1999; Reza, Modzeleski, Feucht, Anderson, & Barrios, 2003). A disproportionate number of parents with guns in the household have been found to leave their guns loaded and not locked up (Coyne-Beasley, McGee, Johnson, & Bordley, 2002; Coyne-Beasley, Schoenback, & Johnson, 2001).

Guns have been estimated to be four to five times more prevalent in the homes of suicide victims compared to controls (Brent et al., 1993d; Kellermann et al., 1992; Shah et al., 2000). There appears to be a gradient of risk, with loaded guns and handguns posing the greatest risk. In a recent review of case–control studies, Brent and Bridge (2003) reported that the odds of a youth dying by suicide was 31.3–107.9 times higher in homes where a gun was present than in homes without guns.

Of note, youth who use firearms for suicide reportedly have fewer identifiable risk factors, such as expressing suicidal thoughts, suicidal intent, psychopathology, and substance abuse, compared to those using other means (Azrael, 2001; Brent et al., 1999; Groholt et al., 1998), and firearm suicides appear to be more impulsive and spontaneous (Azrael, 2001). Thus, to at least some extent, means availability appears to function as a contributing factor to youth suicide, independent of other factors. Among the approximately 10% of youth who died by suicide and had no clear psychiatric diagnosis, the only factor found to discriminate this group from nor mal controls was the presence of a loaded gun in the home (Brent et al., 1993d; Kellerman et al., 1992).

Despite strong evidence for the role of firearms in youth suicide, it should be noted that restricting access to guns may not always result in a decrease in overall suicide deaths. A study that examined suicide methods used by Australian males between 1975 and 1998 (De Leo, Dwyer, Firman, & Neulinger, 2003) reported a declining rate of firearm suicide among males overall as well as among a subset of males aged 15–24, attributed in part to increased restrictions on weapons purchases. In both samples, however, these declines coincided with an increase in the rate of suicide by hanging. Among young males, the increased rate of hanging coincided with an increase in the overall suicide rate.

Researchers have found that students who describe their families as emotionally involved and supportive were much less likely to report suicidal behavior than students who described their families as less supportive and involved (McKeown et al., 1998; Resnick et al., 1997; Rubenstein, Halton, Kasten, Rubin, & Stechler, 1998, Rubenstein, Heeren, Housman, Rubin, & Stechler, 1989; Zhang & Jin, 1996). There is limited

evidence that family cohesion is independent of the adolescent's levels of depression or life stress (Rubenstein et al., 1989, 1998).

The National Longitudinal Study on Adolescent Health (Resnick et al., 1997) reported that student connectedness to school was also a primary protective factor. Similarly, there is limited evidence that perceived suicide risk is inversely related to religious orthodoxy, particularly among racial and ethnic minorities (Greening & Stoppelbein, 2002).

Jessor (1991) suggested that a range of social skills, including decision making and problem solving, may be protective factors for suicidal behavior, and this relationship has received considerable attention among school administrators. The impact of social skills on youth suicidal behavior has not been directly tested, however, and there is no evidence that students with good decision-making or problem-solving skills are overall less suicidal. Although some studies have found that at-risk students who participated in interventions designed to improve social skills showed decreased depression and suicidal behavior (LaFramboise & Howard-Pitney, 1995; Thompson, Eggert, & Herting, 2000; Thompson, Eggert, Randell, & Pike, 2001), it is not clear that these decreases were accounted for by any increase in acquired social skills.