Adolescent suicide is a major public health problem and accounts for at least 100,000 annual deaths in young people worldwide (WHO, 2002). In the United States, suicide accounts for more adolescent deaths than all natural causes combined, with more than 2,000 youth dying by suicide per year (Anderson, 2002). Suicide ranked as the third leading cause of death among the 10- to 14-year-old and 15- to 19-year-old age groups in the United States in 2000, preceded only by accidents and homicide (Anderson, 2002). Nearly 20 percent of adolescents in the middle school and high school age groups report having seriously considered attempting suicide during the past year (Grunbaum et al., 2002). In the United States, the Centers for Disease Control and Prevention’s (CDC’s) large Youth Risk Behavior Surveillance (YRBS) survey also found that nearly 15% of adolescents had made a specific plan to attempt suicide, and that 8.8% of adolescents reported a suicide attempt; this 8.8% represented over 1 million teenagers, of whom approximately 700,000 received medical attention for their attempts (Grunbaum et al., 2002). These results are consistent with those cited in other epidemiological studies in the United States (Gould, Wallenstein, Kleinman, O’Carroll, & Mercy, 1998; Reynolds & Mazza, 1992; Roberts, Chen, & Roberts, 1997; Wichstrom, 2000; Windle, Miller-Tutzauer, & Domenico, 1992).
Although suicide attempts are less common before adolescence, they increase significantly during adolescence, with a peak between 16 and 18 years of age (Lewinsohn, Rohde, & Seeley, 1996). After age 18, there is a marked decline in frequency of suicide attempts, especially for young women (Kessler, Borges, & Walters, 1999; Lewinsohn, Rohde, Seeley, & Baldwin, 2001). As a result, the highest prevalence rate of suicide attempts across the life span exists during adolescence. For each youth suicide, there are approximately 100–200 suicide attempts (American Association on Suicidology, 2003). Researchers have found that between 31% and 50% of adolescent suicide attempters reattempt suicide (Shaffer & Piacentini, 1994), with 27% (males) and 21% (females) reattempting within 3 months of their first attempt (Lewinsohn et al., 1996). These data have major treatment implications, including the need for rapid intervention following the first attempt, as well as the need for treatment that targets both the internal and external conditions contributing to multiple attempts.
As the research reviewed in this chapter will show, the adolescents most at risk for multiple suicide attempts have multiple problems and meet criteria for at least one mental disorder. In an analysis of the 1999 YRBS data, Miller and Taylor (2005) found that the more problem behaviors an adolescent has, the greater his or her risk of suicidal behavior. “Problem behaviors” were defined as including violent behavior, binge drinking, cigarette smoking, high-risk sexual behavior, disturbed eating behavior, and illicit drug use. Compared to adolescents with zero problem behaviors, the odds of a medically treated suicide attempt were 2.3 times greater among respondents with one, 8.8 with two, 18.3 with three, 30.8 with four, 50.0 with five, and 277.3 with six problem behaviors (Miller & Taylor, 2005).
In contrast to suicide attempts, what we are calling “nonsuicidal self-injurious behavior” (NSIB) involves intentionally injuring oneself in a manner that often results in damage to body tissue, but without any conscious suicidal intent. Incidence of NSIB is increasing, especially among adolescents (Hawton & Fagg, 1992; Hawton, Harriss, Simkin, Bale, & Bond, 2004). Although the prevalence estimates need to be interpreted with caution, due to the limited number of studies, adolescents in community studies are reporting that they engage in NSIB at extremely high rates—between 18% and 15.9% (Muehlenkamp & Gutierrez, 2004 and CDC, 2004). College student populations engage in NSIB at similar rates, ranging from 12% (Favazza, 1998) to 35% (Gratz, 2001). Youth who cut themselves, especially repeatedly, have a significant risk of suicide (Cooper et al., 2005).
In the next part of this chapter, we define adolescent suicidal behaviors in more detail. We then examine the existing research on risk factors, highlighting specific behavioral and environmental conditions that appear to increase suicidal risk in adolescents, and comparing those with risk factors for adults. In Chapter 2 we review the research on existing treatments, including preliminary findings on DBT adaptations for suicidal adolescents. We describe DBT, with those adaptations that we feel are most effective for multiproblem suicidal adolescents, in subsequent chapters.
“Suicidal behaviors” include completed suicide, suicide attempts, and suicidal ideation. Because it can often be extremely difficult to assess the degree of suicide intent accompanying intentional self-injury, many suicidologists also consider NSIB as falling into the larger spectrum of adolescent suicidal behaviors, even though NSIB by definition involves no suicidal intent (Berman & Jobes, 1991; Brent et al., 1988; Lewinsohn et al., 1996; Reynolds & Mazza, 1994). We include NSIB within the general category of suicidal behaviors for a number of reasons. First, although some intentional self-injury is clearly without any suicide intent, intentional self-injury often occurs with enormous ambivalence or with swiftly changing intent, such that retrospective analyses of intent may be exceptionally difficult. Second, a behavior that starts as suicidal can evolve into a nonsuicidal act and vice versa. Third, intentional but nonsuicidal self-injury can itself be lethal. That is, it can unintentionally become a suicidal act. Some behaviors categorized as suicide may in fact have been nonsuicidal but none-the-less lethal self-injuries. Finally, and most important clinically, intentional self-injury even without suicide intent is a potent predictor of eventual suicide. It is a behavior with important overlapping characteristics. By excluding this behavior from the general category of suicidal behaviors it is extremely easy for both client and clinician to marginalize or trivialize the behavior. Indeed, many adolescent clients do engage in both suicide attempts and NSIB (Linehan, 1993a). Clearly, however, some youth engage only in NSIB and never in suicidal behavior, while others engage only in suicidal behavior and never in NSIB (Jacobson et al., 2006). Thus, the term “NSIB” helps distinguish not only those teens who deliberately self-injure with no intent to die from those who do have suicidal intent, but also the two types of self-injurious behavior in the same person. This can be critical when precipitants and consequences are very different for the two types of behavior. In this book, we define “completed suicide” as an intentional, self-inflicted death. A “suicide attempt” is self-injurious behavior with ambivalent or certain intent to die. “NSIB” is defined by the lack of an intent to die in the context of deliberate self-injury. “Suicidal ideation” consists of thoughts about being dead or killing oneself and can vary widely in clinical significance, depending on its qualities and context. To a large extent, these definitions hinge on the determination of an intent to die—but in the real world, intent is not always clear or easy to assess.
Suicidologists’ disagreements about how to define a suicide attempt generally revolve around the degree of intent necessary for a behavior to be considered suicidal (Linehan & Shearin, 1988; Farmer, 1988; Bille-Brahe et al., 2004). Some investigators infer or assume intent rather than measure it, and label all intentional self-injurious behavior not resulting in death as “suicide attempt.” Yet not all self-injury is intended to result in death. Brent, Perper, and Allman (1987) and Lewinsohn, Rohde, Seeley, and Klein (1997) reported that among adolescents who acted to self-inflict harm, approximately one-fourth reported no intent to die, and only about one-third of those seen in emergency rooms stated that they had wanted to die. Some investigators have attempted to provide a nomenclature for suicidal behavior that incorporates information about intent to die. O’Carroll et al. (1996), for example, classify a self-injurious behavior in which there is no intent to die but rather intent to communicate distress to someone else as “instrumental suicide-related behavior.” As we have noted in the Introduction, the WHO currently uses the term “suicide attempt” any time an individual does not die, regardless of whether suicidal intent is present (Bille-Brahe et al., 2004).
In addition to the definitional confusion across research studies internationally, clinical assessment of suicidal intent in adolescents and adults remains a challenge because self-reports of suicidal ideation, intent, and behavior are unreliable. For example, one study using both pencil-and-paper questionnaires and a semistructured interview format to evaluate adolescents’ self-reports of suicidality reported discrepancies between assessment modalities in 50% of the 48 adolescents (Velting, Rathus, & Asnis, 1998). Moreover, adolescents’ reports of attempts often indicate ambivalence, further complicating their accurate assessment (King et al., 1995). However, when interviewers are reliably trained and use a structured interview format, they can be reliable judges of suicidal intent with older adolescents and adults (Linehan, Heard, & Armstrong, 1993b).
In a recent review article on NSIB, Gratz (2003) offered the following definition: “Deliberate self-harm may be defined as the deliberate, direct destruction or alteration of body tissue, without conscious suicidal intent but resulting in injury severe enough for tissue damage to occur” (p. 192). NSIB has been associated with a number of mental disorders, including schizophrenia (Herpertz, 1995), trichotillomania, personality disorders, eating disorders, substance use disorders, PTSD, and intermittent explosive disorder (Favazza & Rosenthal, 1990; Zlotnick, Mattia, & Zimmerman, 1999), as well as mental retardation and a variety of neurological, developmental, and genetic disorders (Schroeder, Oster-Granite, & Thompson, 2002). In particular, these researchers and others have found impulsive NSIB to occur in up to 80% of clients with BPD, as compared to prevalence estimates of approximately 4% in the general population of mental health treatment seekers (Shearer, Peter, Quaytman, & Wadman, 1988). Among clients diagnosed with BPD, the onset is usually reported during early adolescence, whereas onset appears later during young adulthood among individuals without BPD (Symons, 2002).
There is growing evidence that suicidal behaviors and NSIB have different phenomenological pathways (Muehlenkamp & Gutierrez, 2004), functions (Brown, Comtois, & Linehan, 2002), and correlates (Boergers, Spirito, & Donaldson, 1998; Herpertz, 1995; Pattison & Kahan, 1983). As Groholt, Ekeberg, Wichstrom, and Haldorsen (2000) point out, suicidal acts pose greater risks and may require different interventions than repetitive NSIB. Furthermore, a client who repeatedly cuts one’s arm to temporarily relieve anger may have difficulty developing a therapeutic alliance with a therapist who focuses primarily on reasons for living instead of alternative methods of managing anger. Therefore, assessing intent allows us to discriminate behaviors that are suicide attempts (i.e., self-injurious behavior with varying levels of intent to die) from self-harming acts that involve no intent to die.
While the evidence to support the relationship between impulsivity and NSIB is mixed, the empirical research suggests that childhood sexual and physical abuse and emotional neglect account for significant variance in the risk for NSIB in adulthood (Gratz, 2003). Moreover, the function of NSIB is commonly conceptualized as a means of regulating emotions (Linehan, 1993a). Both clinical and empirical research suggests that NSIB functions as a form of emotional avoidance and escape from unwanted emotions (Gratz, 2003).
Although suicidal behaviors and NSIB should be considered clinically distinct entities, it is also important to remember that individuals can engage in both types of behaviors at different points in their lives. Lipschitz et al. (1999) found that adolescents hospitalized for suicidal behaviors were more likely also to exhibit NSIB than adolescents hospitalized for other problems were. In one study of adults, 37% of those with self-injury had histories of suicide attempts (Herpertz, 1995). Other studies have found that 10–27% of self-injuring adults eventually die from suicide (Stanley, Gameroff, Michalsen, & Mann, 2001; Cowmeadow, 1995).
As with the other terms discussed here, researchers have been unable to reach a consensus on a definition of “suicidal ideation,” and so comparing findings from different studies has been difficult (King, 1997). More recently, researchers have suggested conceptualizing suicidal ideation as occurring on a continuum of increasing clinical significance, and some have begun to operationalize definitions that help to overcome the difficulties (Lewinsohn et al., 1996; Roberts et al., 1997). Members of the Oregon Adolescent Depression Project (Lewinsohn et al., 1996) put forth the following categories to operationalize suicidal ideation, in order of increasing severity: “thoughts of death or dying,” “wishing to be dead,” “thought of hurting (or killing) self,” and “suicidal plan.”
Suicidal ideation most generally involves current thoughts of death, of killing oneself, or of being killed. Some adolescents may present with passive suicidal ideation (e.g., “I wish I were dead”) but report having no plan or intent to kill themselves. For a subgroup of these adolescents, the idea of actively taking their own lives is unfathomable. In contrast, some adolescents report active suicidal ideation that is more alarming to the clinician (e.g., “I feel like killing myself”). When asked, these clients may report having a specific plan to kill themselves. A suicidal plan involves identifying a specific method, and possibly a given time frame, in which an adolescent plans to kill him- or herself. Once an adolescent reports having a plan, the clinician must assess for suicidal intent. “Intent” characterizes the adolescent’s level of commitment in carrying out the plan. For some adolescents, suicidal intent may be clear and definite; however, many report ambivalence or minimal intent to die (Brent et al., 1993b; King et al., 1995). Hence adolescents may report having a specific plan but have no intent to die (e.g., “I thought about jumping off a bridge, but I would never do it”). Others may describe their intent as ambivalent (e.g., “I am thinking about taking an overdose, but I am not sure if I can go through with it”). Still others may have full intent to kill themselves (e.g., “I intend to shoot myself with my own gun this Sunday when my parents leave town”). Further complicating matters is the inconsistency across assessments in adolescents’ reports of their own suicidal behavior, as noted earlier (Velting et al., 1998).
Suicidal ideation is a strong predictor, if not one of the best predictors, of suicide attempts (Andrews & Lewinsohn, 1992; Kienhorst, DeWilde, van den Bout, Diekstra, & Wolters, 1990). Lewinsohn et al. (1996) found in their prospective study that two dimensions of suicidal ideation were highly correlated: severity and duration. Specifically, adolescents who spent more time thinking about suicide also tended to have more serious thoughts about suicide. More importantly, adolescents who indicated a greater intensity of suicidal ideation (i.e., by endorsing a greater number of items during the past week) were more likely to attempt suicide (Lewinsohn et al., 1996). In sum, as suicidal ideation becomes more frequent and intense, the risk of suicide attempts increases. Researchers have discovered that suicidal ideation is commonly associated with an Axis I disorder in adolescents; it is most strongly associated with depression, but also occurs in anxiety, disruptive behavior, and substance use disorders (Lewinsohn et al., 1996).
One of the greatest frustrations for clinicians, researchers, and family members is the inability to predict in advance which individual adolescents will attempt or complete suicide. It is unlikely that the state of the art will improve dramatically in the near future. Research is limited by both ethical problems and recording errors in determining whether predictions are accurate. There is also the more general problem of predicting infrequent events. The best that can be done is to describe the characteristics of subpopulations in which rates of suicide are higher than in the general population. Such a description can then be used to determine whether or not a given adolescent is at high risk for suicide.
Suicidologists have worked vigorously to identify risk factors for adolescent suicidal behaviors. The research has clearly demonstrated that certain distal and proximal risk factors, when combined, increase the probability of suicidal behavior. The most important distal factors linked with adolescent vulnerability to suicide are prior suicidal behaviors, mental disorders, chronic family disturbance, gender, homosexual or bisexual orientation, and ethnicity. Important proximal risk factors are stressful life events, sexual and physical abuse, academic difficulties, functional impairment from physical disease or injury, suicide in the social milieu, and access to suicidal means. Suicide risk increases when proximal risk factors occur in the context of distal risk factors. Hence, for example, Lewinsohn et al. (1996) hasten to point out that stressful life events should be considered “red flags” for clinicians. By the same token, however, a young client’s experiencing one or more stressful life events per se should not alarm the clinician, since many adolescents experience such events without suicide as a consequence. Thus Lewinsohn, Rohde, and Seeley (1994) make two important points for clinicians to remember:
A proximal risk factor in combination with one or more distal risk factors is what heightens the risk of suicide.
Past suicidal behaviors are stronger predictors of future suicidal behaviors than stressful life events or other proximal risk factors.
However, assessment of risk is complicated by the fact that some risk factors can be considered both distal and proximal. For example, social and environmental factors such as family conflict and parental psychopathology can function as proximal factors that cause or exacerbate existing mental disorders or psychological pain, which in turn increase the risk of suicide. Table 1.1 summarizes the key distal and proximal risk factors for suicide in adolescents that have been identified in the research. Below we review these risk factors in more detail.
The best predictors of future suicidal behaviors among adolescents and adults are past suicidal behaviors. In particular, it has become well established that a prior suicide attempt is one of the single most important predictors of completed suicide (Gould, Greenberg, Velting, & Shaffer, 2003; Shafii, Carrigan, Whittinghill, & Derrick, 1985), with a 30-fold increased risk for boys and a 3-fold increased risk for girls (Shaffer et al., 1996). Numerous “psychological autopsy” studies of adolescents who complete suicide have found high rates of previous suicide attempts, ranging between 10% and 44% (Brent et al., 1988; Marttunen, Aro, & Lönnqvist, 1992; Shafii et al., 1985).
Several prospective studies followed adolescents admitted to psychiatric inpatient units for suicidal behaviors from 4 to 15 years postadmission. Documented suicide rates of these clients ranged from 9% to 11.3% (Motto, 1984; Otto, 1972). In a follow-up study of adolescents evaluated in an emergency room after suicide attempts, researchers discovered that 8.7% of the males and 1.2% of the females committed suicide within 5 years (Kotila, 1992). Gender differences are consistent across studies.
Clinical researchers agree that suicidal behaviors among adolescents are clearly associated with diagnosable mental disorders (Andrews & Lewinsohn, 1992; Brent et al., 1988; Kovacs, Goldston, & Gatsonis, 1993; Lewinsohn et al., 1996; Rich, Young, & Fowler, 1986; Shaffer, Garland, Gould, Fisher, & Trautman, 1988; Shaffi et al., 1985). Psychological autopsy studies have reported that over 90% of adolescents completing suicide had a mental illness at the time of their death, although younger adolescents completing suicide tend to have lower rates of mental illness, averaging around 60% (Beautrais, 2001; Brent, Baugher, Bridge, Chen, & Chiappetta, 1999a; Groholt, Ekeberg, Wichstrom, & Haldorsen, 1998; Shaffer et al., 1996). Although various mental disorders have been found among these adolescents, three classes of Axis I disorders characterize most attempted and completed suicides among adolescents internationally: depressive and anxiety disorders; impulsive, disruptive, and antisocial behavior disorder; and substance-related disorders (Andrews & Lewinsohn, 1992; Berman & Jobes, 1991; Brent et al., 1993b; Fergusson & Lynskey, 1995; Gould et al., 2003; Marttunen, Aro, Henriksson, & Lönngvist, 1991; Pfeffer, Newcorn, Kaplan, Mizruchi, & Plutchik, 1988; Shaffer et al., 1996; Sigurdson, Staley, Matas, Hildahl, & Squair, 1994). In addition, personality disorders (especially BPD) are increasingly being linked with adolescent suicidality, as is comorbidity of both Axis I and Axis II disorders.
TABLE 1.1. General Risk Factors for Adolescent Suicidal Behavior
| Distal risk factors | Proximal risk factors |
|
|
Depressive disorders have been reported among adolescents attempting and completing suicide, at rates ranging from 49% to 64% (Brent et al., 1993; Marttunen et al., 1991; Rich et al., 1986; Shaffer et al., 1996), with the highest rates among psychiatric inpatients (Spirito, Brown, Overholser, & Fritz, 1989). In Finland, Marttunen et al. (1991) found depression to be the most prevalent diagnosis among their adolescents who completed suicide, with half of the boys and two-thirds of the girls meeting diagnostic criteria. Suicidal behaviors are common among adolescents with early-onset depressive disorders (Brent et al., 1993b; Kovacs et al., 1993; Pfeffer et al., 1991). Indeed, Kovacs et al. (1993) found a four- to fivefold increase in suicidal ideation and behavior among adolescents with depressive disorders as compared to adolescents with other mental disorders.
These statistics are noteworthy, since the risk of developing a depressive disorder increases as one gets older, but rises dramatically between the ages of 9 and 19 (King, 1997). At any one time, approximately 1 in 20 children and adolescents suffer from major depressive disorder (Essau & Dobson, 1999). Bipolar disorders, while less prevalent among adolescents, have been considered a significant risk factor in some studies (Brent et al., 1988, 1993) but not in others (Marttunen et al., 1991; Shaffer, Gould, & Hicks, 1994).
High rates of comorbidity between anxiety and depression have been documented in children and adolescents (Brady & Kendall, 1992). Among adults, there is increasing evidence not only that anxiety symptoms are associated with the severity of the depression, but that severe anxiety may be a risk factor for suicide, especially when it is associated with depression (Fawcett, 1997). Lewinsohn et al. (1996) identified anxiety disorders as a risk factor for suicidal behavior among adolescents. More recently, Goldston et al. (1999) reported trait anxiety to be predictive of posthospitalization suicide attempts, independent of mental disorder. In another study, investigators found that adolescents with a history of panic attacks were three times more likely to express suicidal ideation and two times more likely to report suicide attempts than those without a history of panic attacks (Pilowsky, Wu, & Anthony, 1999). These findings were true even after the investigators controlled for depression, drug and alcohol use, gender, and ethnicity/race. PTSD has also been associated with adolescent suicidal behavior (Giaconia et al., 1995; Mazza, 2000).
Other researchers have found evidence of anxiety and perfectionism among a subgroup of adolescents completing suicide (Shaffer et al., 1988). In two other studies of such adolescents, researchers found many of their sample characterized by withdrawn, lonely, supersensitive, and inhibited personality styles (Hoberman & Garfinkel, 1988; Shafii et al., 1985).
Several researchers have suggested that most completed suicides by adolescents are impulsive acts, with only about 25% providing evidence of planning (Hoberman & Garfinkel, 1988; Shaffer et al., 1988). Aggression with impulsivity has also been linked with suicidal behavior in both children and adolescents (Apter et al., 1995; Brent et al., 1994; Garfinkel, Froese, & Hood, 1982; Grosz et al., 1994; Inamdar, Lewis, Siomopoulos, Shanok, & Lamela, 1982; Pfeffer et al., 1988; Plutchik, van Praag, & Conte, 1989). In a study examining the prevalence of suicidal and violent behaviors in a sample of 51 hospitalized adolescents, Inamdar et al. (1982) found that 67% had been violent, 43% had been suicidal, and 28% had been both. Plutchik et al. (1989) have theorized that suicide risk is increased when aggressive impulses are triggered, then amplified by forces such as substance abuse, and not reduced by opposing forces such as appeasement from others. A study of suicidal adults suggested that a personality style marked by pronounced impulsivity and aggression characterized individuals at risk for suicide attempts, regardless of Axis I mental disorder (Mann, Waternauz, Haas, & Malone, 1999).
It should not then be a surprise that disruptive behavior disorders are common diagnoses found among suicidal adolescents (Kovacs et al., 1993; Marttunen, Aro, Henriksson, & Lönnqvist, 1994; Shafii et al., 1985), especially males (Brent et al., 1993; Shaffer et al., 1996). In a longitudinal study, Kovacs et al. (1993) found that 45% of youth diagnosed with conduct disorders, substance use disorders, and depressive disorders made suicide attempts, compared to only 22% of youth with depressive disorders only and 10% of youth with no depressive disorders. Other researchers have suggested that conduct disorders in adolescents may play a role equal to or even larger than that of depression in adolescent suicidal behaviors. Apter, Bleich, Plutchik, Mendelsohn, and Tyano (1988) found higher scale scores for suicidality on the Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS) for adolescents with conduct disorder than for those with major depressive disorder, even though those adolescents diagnosed with conduct disorder were less depressed. Apter et al. (1995) have suggested that aggression, a large component of conduct disorder, may be as important a risk factor as depression in some kinds of suicidal behavior. These researchers hypothesize two classes of suicidal behavior during adolescence: a wish to die (depression) and a wish not to be here for a time (poor impulse control). Apter and colleagues elaborate by stating that the first type of suicidal behavior characterizes disorders with prominent depression (such as major depressive disorder and anorexia nervosa), and that the second characterizes disorders of impulse control (such as conduct disorder).
Substance use and abuse have been found with great frequency among adolescents attempting and completing suicide, and are therefore considered primary risk factors for adolescent suicidal behavior (Brent, Perper, & Allman, 1987; Crumley, 1979; Hoberman & Garfinkel, 1988; Marttunen et al., 1995; Rich et al., 1986; Shaffer et al., 1996; Shafii et al., 1985; Sigurdson et al., 1994). In studies of substance-using adolescents, suicide attempts occur at rates threefold those of non-substance-using adolescents, with the “wish to die” increasing dramatically after the onset of substance use (Berman & Schwartz, 1990).
In studies conducted internationally of completed suicide among adolescents and young adults, evidence of alcohol or other substance abuse was found in 28–54% of cases (Brent et al., 1987; Hawton, Fagg, & McKeown, 1989; Hoberman & Garfinkel, 1988; McKenry, Tishler, & Kelley, 1982; Shaffi et al., 1985; Marttunen et al., 1991; Rich et al., 1986). In a psychological autopsy study of 120 individuals under 20 years of age who completed suicide, Shaffer et al. (1996) reported drug and/or alcohol abuse as a risk factor for older adolescent males. Other studies to date have not highlighted such significant gender differences with regard to this risk factor. One recent study evaluated 89 consecutive admissions to a specialty outpatient clinic for depressed and suicidal inner-city teens. Of the 49 subjects with histories of self-injurious behavior, 18.4% met diagnostic criteria for cannabis abuse or dependence (Velting & Miller, 1999).
Due to the commonly held belief among mental health professionals that personality is still evolving during adolescence, there has been a reluctance to diagnose personality disorders among this age group. Yet, as we will discuss in more detail later, there is considerable overlap between the characteristics of those at high risk for suicidal behavior and those diagnosed with a personality disorder, especially BPD. Linehan, Rizvi, Welch, and Page (in press) reviewed diagnoses given to individuals who completed suicide over 14 research samples. They found personality disorder rates of 40–53% in these individuals, and concluded that these disorders are as great a risk factor for suicidal behaviors as major depression and schizophrenia are. In three out of four adolescent and youth suicide samples, they found similar high rates (see later discussion).
The relationship between suicidal behavior and personality disorders (particularly BPD) in adolescents has been recognized for over two decades (Brent et al., 1994; Clarkin, Friedman, Hurt, Corn, & Aronoff, 1984; Crumley, 1979; Marton et al., 1989; Marttunen et al., 1995; McManus, Lerner, Robbins, & Barbour, 1984; Pfeffer et al., 1988; Runeson & Beskow, 1991). Personality disorders and the tendency to engage in impulsive violence have become critical risk factors for completed suicide among adolescents (Brent et al., 1994). Brent et al. (1993a) compared adolescent inpatients who had attempted suicide with never-suicidal inpatient controls and found that the suicidal clients were more likely than the controls to have personality disorders or features of such disorders (81% vs. 58%, respectively), particularly those of the borderline type (32% vs. 10%). Those who had attempted suicide showed greater severity of borderline behavioral criteria even after suicidality was removed as a criterion. One report from Sweden described a similar rate of 33% for BPD among adolescents and young adults completing suicide (Runeson & Beskow, 1991). In a Finnish study of females ages 13–22 years who completed suicide, Marttunen et al. (1995) found that 26% of their 1,397 subjects met criteria for BPD. Velting, Rathus, and Miller (2000) found that American adolescents attempting suicide had higher levels of borderline behavioral criteria on the Millon Adolescent Clinical Inventory than nonsuicidal outpatient controls. As with other types of psychopathology, comorbidity of BPD with major depression and substance use among suicidal adolescents heightens the suicide risk (Marttunen et al., 1995; see “Comorbidity,” below). It has been suggested that the co-occurrence of mood and personality disorders represents a particularly significant risk factor for suicidal behavior (Blumenthal & Kupfer, 1986).
Comorbidity of mental disorders is the rule rather than the exception among adolescents (Volkmar & Woolstorn, 1997), and comorbid disorders are often present in adolescents who commit suicide (Rich et al., 1986; Shafii et al, 1988). Although suicidal adolescents may abuse substances in the absence of other Axis I disorders, substance-related disorders often coexist in the presence of depression and/or disruptive behavior disorders (Berman & Jobes, 1991; Bukstein, Glancy, & Kaminer, 1992; Lewinsohn et al., 1996; Shaffer et al., 1996). Depression, conduct disorder, and substance abuse frequently present concurrently, with the frequency and lethality of attempts increasing with the degree of comorbidity (Frances & Blumenthal, 1989). Depression comorbid with alcohol/substance abuse, conduct problems, and/or BPD represents a particularly high-risk profile for completed suicide and other suicidal behaviors among teenagers (Brent et al., 1993b; Kovacs et al., 1993; Marttunen et al., 1995; Shafii et al., 1985).
Although most adolescents who make a suicide attempt have a diagnosable mental disorder, it is important to note that most adolescents with a mental disorder do not make a suicide attempt (Lewinsohn et al., 1996). Adolescents at highest risk for suicide tend to have high rates of comorbidity of both Axis I and Axis II disorders. As should be clear by now, suicidal behaviors span diagnostic categories. As we will show a bit later, assessing risk by identifying specific diagnostic categories is not as clinically helpful as looking for specific clusters of affective, cognitive, and behavioral characteristics across diagnostic categories. We continue now with our review of distal risk factors for adolescent suicidal behaviors.
Parents, by the nature of their roles, have direct and long-standing influences on the health of their children. When parents manifest their own serious problems, which may result in conflictual relations with their adolescents, the question becomes this: How and to what extent do parents (and other family members) affect adolescents’ suicidal behavior? Various theories coupled with research data suggest that family functioning plays an important role in the etiology and maintenance of adolescent suicidal behavior (Adams, Overholser, & Lehnert, 1994; Berman & Jobes, 1991; King, Segal, Naylor, & Evans, 1993). Research has found that when family processes are disturbed, there is an increased risk of suicidal ideation and attempts among adolescents (Pfeffer, 1989).
A family history of suicidal behavior significantly increases the risk of completed suicide (Brent et al., 1988; Gould, Fisher, Flory, & Shaffer, 1996; Shafii et al., 1985) and attempted suicide (Bridge, Brent, Johnson, & Connolly, 1997; Johnson, Brent, Bridge, & Connolly, 1998). Agerbo, Nordentoft, and Mortensen’s (2002) Danish Registry Study found youth suicide to be nearly five times more likely in the offspring of mothers who had completed suicide and twice as common in the offspring of fathers, even after adjustments for parental mental disorders.
Parental mental disorders, particularly depression and substance abuse, have been associated with suicidal ideation, attempts, and completed suicide in adolescents (Brent et al., 1988; Bukstein et al., 1993; Gould et al., 1996; Kashani, Goddard, & Reid, 1989). Impaired parent-child communication and low levels of emotional support and expressiveness are also associated with adolescent suicidal behavior (Campbell, Milling, Laughlin, & Bush, 1993; Garber, Little, Hilsman, & Weaver, 1998; Keitner et al., 1990; King et al., 1993; Martin & Waite, 1994; Pfeffer, 1989; Wagner, 1997).
Although some researchers suggest that a disproportionate number of adolescents attempting suicide do not live in stable, intact homes (Beautrais, 2001; Brent et al., 1993c; Groholt et al., 1998), the association between separation/divorce and suicide decreases when parental mental disorders are accounted for (Gould et al., 1996; King et al., 1993). King (1997) asserts that while these findings do not support the specific link to suicidal behavior, family loss and instability are important as risk factors for multiple poor outcomes. Interestingly, Lewinsohn et al. (1996) found higher rates of multiple suicide attempts among those adolescents who had a parent die before an adolescent was 12 years of age. These researchers suggest that parental loss may be an important yet underinvestigated suicide risk factor for predicting repeated suicide attempts in those with a history of one such attempt. Thus disturbances in family functioning appear to be important, but the extent to which disturbances in family functioning affect adolescent suicidal behavior remains unclear. Nevertheless, these findings support the need to evaluate and treat the suicidal adolescent within the context of his or her family system.
Over the past 30 years, the incidence of completed suicide and suicide attempts in older adolescents (i.e., ages 15–19) has shown significant gender and ethnic variations. Whereas suicidal ideation and attempts are more common among females in the United States (Gould et al., 1998; Grunbaum et al., 2002; Lewinsohn et al., 1996), completed suicide is five times more common among 15- to 19-year-old males (Anderson, 2002). While these gender differences are found to be similar in Western Europe, New Zealand, Australia, and North America, they are not consistent around the world. In fact, completed suicide rates for males and females are equal in some Asian countries and are higher among females in China (WHO, 2002).
Studies have consistently found gender differences among adolescents who attempt suicide as well (Gould et al., 2003). Approximately 10–20% of girls versus 4–10% of boys report having made a suicide attempt during their lifetime. Hence girls report attempting suicide two to four times as frequently as boys. In fact, King (1997) suggests that in a typical high school classroom, it is likely that two girls and one boy have made a suicide attempt during the past year.
Several factors may explain why females make more suicide attempts while males have a greater frequency of completed suicides. First, females have higher rates of mood disorders, which are associated with suicidal behavior (Brent et al., 1999a; Shaffer et al., 1996). Males have higher rates of aggressive behavior and substance abuse, which are often associated with completed suicide (Shaffer et al., 1994). Second, females in the United States choose less lethal methods, such as overdoses (females, 30%; males, 6.7%) (Anderson, 2002). Third, females often experience higher rates of sexual abuse, which is correlated with suicidal behavior as well (Friedman et al., 1982). Fourth, according to Linehan’s (1973) study of American college students, older adolescents and young adults perceive nonfatal suicidal behavior as more “feminine” and less potent than killing oneself. Others have described nonfatal suicidal behavior as “feminine” because it is interpreted as a call for help—a behavior that is expected of women (Suter, 1976). In contrast, females who complete suicide are viewed more negatively and the behavior is perceived as more unacceptable than in males, since suicide involves a degree of self-determination that may be considered incompatible with femininity (Canetto, 1997b). In the United States, cultural scripts of gender and suicidal behavior are likely to influence adolescents’ decisions about suicidal behavior (Canetto, 1997a). Adolescents are quite sensitive and responsive to cultural messages—even more so than adults, given that they are in the midst of defining their identities. Thus the influence of “gender-appropriate” ideas of suicidal behavior may be significant and requires further evaluation.
Cross-sectional and longitudinal epidemiological studies have found homosexual adolescents of both sexes to be two to six times more likely to attempt suicide than their heterosexual peers (Blake et al., 2001; Garofolo, Wolf, Wissow, Woods, & Goodman, 1999; Harry, 1989; Remafedi, French, Story, Resnick, & Blum, 1998; Russell & Joyner, 2001). According to Harry’s (1989) review, risk for attempts is typically heightened at about 18–19 years of age, when a teen is “coming out.”
In a study of 137 gay and bisexual males between the ages of 14 and 21 (drawn from a nonclinical sample), nearly one-third of the subjects reported at least one suicide attempt, and almost half of them had repeatedly attempted suicide, with 54% of the attempts considered moderate to high in lethality risk (Remafedi, Farrow, & Deisher, 1993). One-third of the attempts occurred in the same years subjects identified themselves as homosexual or bisexual, yet “suicide attempts were not explained by experiences with discrimination, violence, loss of friendship, or current personal attitudes toward homosexuality” (Remafedi et al., 1993, p. 495). Rather, gender nonconformity and precocious psychosexual development were predictive of suicidal behavior: The younger these subjects were when they identified themselves as homosexual/bisexual, the more likely they were to report suicidal behavior. The authors suggested as a possible reason for this that early and middle adolescents may be less able to cope with the isolation and stigma associated with a homosexual identity than older adolescents, who may have better-developed coping skills.
Remafedi et al. (1998) examined the relationship between sexual orientation and suicide risk in a population-based sample of adolescents. They conducted a cross-sectional statewide survey of public school students in grades 7–12. Among the 394 students who described themselves as bisexual/homosexual and 336 gender-matched heterosexual students, suicide attempts were reported by 28.1% of bisexual/homosexual males, 20.5% of bisexual/homosexual females, 14.5% of heterosexual females, and 4.2% of heterosexual males. Thus for males, but not for females, bisexual/homosexual orientation was significantly associated with suicide attempts.
In another population-based sample of 3,365 public high school students in grades 9–12, Garofalo et al. (1999) found that self-identified gay, lesbian, and “not-sure” youth were 3.41 times more likely to report a suicide attempt than their peers. Similar to Remafedi et al.’s (1998) results, Garofalo et al. (1999) found sexual orientation to have an independent association with suicide attempts for males. For females, the association of sexual orientation with suicidality may be mediated by drug use and violence/victimization behaviors. It may also be lesbians are less likely to be identified by others as gay during adolescence, and therefore their sexual orientation may not be independently associated with suicidal behavior (Downey, 1994).
In the United States, youth suicide is most common among Native Americans (Anderson, 2002; Middlebrook, LeMaster, Beals, Novins, & Manson, 2001). White youth have higher rates than African American youth, with Asians/Pacific Islanders having the lowest rates. Based on a significant increase in suicide among African American males between 1986 and 1994, the long-standing difference between African Americans and white has declined. Importantly, since the mid-1990s, there has been a gradual decline in suicide rates among both white and African American males and females (Gould et al., 2003). There are no clear explanations for this decline during the past decade. Hispanics have a relatively low suicide completion rate, but are significantly more likely than either white or African American adolescents to report suicidal ideation and to have made a suicide attempt (Kann, Kinchen, Williams, & Ross, 2000; Reynolds & Mazza, 1992; Tortolero & Roberts, 2001).
In one study of suicidal behavior among inner-city Hispanic adolescent females, researchers compared 33 subjects and their mothers with a matched nonsuicidal control group, to begin to generate a set of hypotheses to explain this behavior (Razin et al., 1991). The findings were as follows: Attempts were nearly always impulsive and nonlethal, though often with a stated wish to die. Nearly all were overdoses, and were precipitated by conflicts with mothers or boyfriends. Attempting girls’ parents were less often born in the United States; their mothers seemed medically less healthy; and their extended families were more often supported by public assistance and had higher rates of criminal and mental problems. School performance was poorer among attempting girls, who often had suffered more and earlier losses—especially of biological fathers, with whom few had ongoing relationships. They had also recently lost friends and expressed a mistrustful stance toward friendships. Similarly, their mothers had fewer friends and more often expressed a mistrustful stance. Relationships with mothers seemed more intense, desperate, and even violent, and attempting girls were much more “parentified” (i.e., mothering their mothers). Although knowledge of suicidal models was common in both groups, the mothers of attempting girls knew even more models among family, friends, and neighbors than did their daughters or the nonsuicidal subjects or their mothers. More of the suicidal girls’ mothers had themselves made attempts. Families of most attempting girls were usually mobilized by the attempts. This list of findings serves as an initial profile of risk factors for suicidal behavior among inner-city Hispanic adolescent females. However, given the small sample size and the fact that some reported findings did not reach statistical significance, the study’s conclusiveness is limited.
The data are mixed regarding the effect of SES and suicide. Several studies have found youth attempting suicide to have lower SES than community controls, even after other social and mental health risk factors are controlled for (Beautrais, Joyce, & Mulder, 1996; Fergusson, Woodward, & Horwood, 2000; Wunderlich, Bronish, & Wiichen, 1998). Interestingly, Gould et al. (1996) reported that African American youth who had completed suicide had significantly higher SES than their general population counterparts, but no such effect was found among white or Hispanic youth completing suicide. Other studies have found little effect of low SES on suicide completion generally, after controlling for family history of mental illness or suicide (Agerbo et al., 2002; Brent et al., 1988).
In most cases, a distal risk factor is not sufficient in itself to precipitate suicidal behavior. Most adolescent suicidal behavior is triggered when a proximal risk factor, such as a stressful life event, is combined with a distal risk factor (Lewinsohn et al., 1996; Shaffer et al., 1988). Below we review some of the common proximal risk factors associated with adolescent suicidal behavior.
Historically, interpersonal conflicts and separations are considered the most common precipitants to adolescent suicide (Marttunen, Aro, & Lönngvist, 1993; Spirito et al., 1989). Breakup of a romantic relationship, disciplinary crises or legal problems, humiliation, and arguments are some of the stressful life events identified in attempted and completed suicides of youth around the world, even after controlling for psychopathology, family, and personality factors (Beautrais, 2001; Brent et al., 1993c; Gould et al., 1996; Lewinsohn et al., 1996; Marttunen et al., 1993; Runeson, Beskow, & Waern, 1996; Shaffer et al., 1988). Specific stressors may vary with age. For example, romantic difficulties are more common precipitants among older adolescents, while parent–child conflicts are more common among younger adolescents (Brent et al., 1999).
Researchers have found that both childhood sexual abuse and physical abuse are also associated with suicidal behavior in adolescents, even after controlling for a variety of potentially confounding variables, including an adolescent’s psychopathology, parental psychopathology, and demographics (Brent et al., 1993c, 1999; Fergusson, Horwood, & Lynskey, 1996; Gould et al., 2003; Johnson et al, 2002).
School difficulties, not working or attending school, and dropping out of high school (without attempting to earn a general equivalency diploma go on to college), have been identified as risk factors for attempted and completed suicide in several countries, even after controlling for psychopathology and social risk factors (Beautrais et al., 1996; Gould et al., 1996; Wunderlich et al., 1998).
Physical diseases and injuries, to the extent that they result in functional impairment, have also been found to increase the risk of future suicide attempts in adolescents (Lewinsohn et al., 1996). Being diagnosed as having AIDS or as being HIV-positive, while considered a more definitive risk factor among adults, has not received adequate empirical study among adolescents. Prior studies of adults diagnosed with AIDS report a 7- to 36-fold increased risk of suicide (Kizer, Green, Perkins, Coebbert, & Hughes, 1988; Cote, Biggar, & Dannengerg, 1992). To date, no studies have examined this question among adolescents diagnosed with AIDS. Although suicidal ideation and other types of psychiatric morbidity in HIV-infected people have been described in several reports (Lyketsos & Federman, 1995; McKegney & O’Dowd, 1992), few definitive data exist examining the risk of suicide among individuals found positive for the HIV infection. One of the few studies examining this question among 4,147 HIV-seropositive military service applicants, including older adolescents and adults, reported no significant increased risk of death by suicide in the months following HIV screening (Dannenberg, McNeil, Brundage, & Brookmeyer, 1996). These investigators point out that because suicide risk is reported to be greatly increased after symptomatic HIV disease is present, clinicians should carefully assess persons with HIV infection for suicidal risk factors during initial counseling and subsequent counseling and medical care.
Exposure to the suicidal behavior of others can precipitate imitative suicidal behavior, at least in some individuals (Velting & Gould, 1997). Adolescents are highly susceptible to suggestion and imitative behavior, as these are primary modes of social learning and identity formation. Velting and Gould (1997) propose that modeling cues through personal acquaintance, community exposures, and exposure to media coverage may all play a role in imitative suicidal behavior. Numerous studies have found that significantly more peers, friends, and/or family members had attempted or completed suicide in the social networks or families of persons with suicide ideation, attempts, and completions than in control groups (Brent, Bridge, Johnson, & Connolly, 1996a; Brent, Moritz, Bridge, Perper, & Canobbio, 1996b; Garfinkel et al., 1982; Gould et al., 1996; Harkavy-Friedman, Asnis, Boeck, & Difiore, 1987; Shafii et al., 1985; Smith & Crawford, 1986). In addition to increased rates of suicidal behaviors in these relatives and friends, suicidologists have examined suicide clusters and the influence of the media on adolescents.
A “suicide cluster” may be defined as a group of suicide attempts that occur closer together in time and space than would normally be expected in a given community (CDC, 1988). In a review of the literature on suicide clusters, Velting and Gould (1997) argue that suicide contagion is a real effect, even though it appears to be a less potent risk factor than other psychiatric and psychosocial risk factors for suicide. Of all age groups, adolescents are at highest risk for clustering of attempted and completed suicides, with only minimal effects beyond 24 years of age (Brent et al., 1989; Gould et al., 1990, 1996); therefore, this age group should be allotted the greatest amount of resources for prevention and postvention work (Velting & Gould, 1997).
Regarding media influence, there has been a marked increase in studies examining the impact on suicide rates around the world of media-covered nonfictional and fictional suicides. Studies have begun to emphasize characteristics such as age, gender, and performance as important to the modeling effect, since perceived similarity between observer and model appears to facilitate imitation. In an example from a fictional 6-week TV serial broadcast in Germany, a 19-year-old male was portrayed as committing suicide by jumping in front of a train (Schmidtke & Hafner, 1988). Results revealed a subsequent 86% increase in the number of railway-related suicides among 15- to 29-year-old males, and a 147% increase among 15- to 19-year-old males. Although this example involved a fictional suicide, research has shown that real suicides covered in the media result even more clearly in subsequent suicides (especially for teenagers), and that the magnitude of the suicide increase is proportional to the amount, duration, and prominence of media coverage (Gould, 2001; Velting & Gould, 1997).
Accessibility to the means of suicide (e.g., firearms) is a significant proximal risk factor. The most common method of suicide in the United States, regardless of age, race, or gender, is the use of firearms. According to the National Center for Health Statistics (1996), 67.5% of the total number (3,344) of young persons committing suicide in 1994 used firearms. The probability of suicide increases fivefold when a firearm is kept in the home (Brent et al., 1991; Rosenberg, Mercy, & Houk, 1991). One study has found that the availability of guns in the home contributes more heavily to the population attributable risk percentage for suicide among adolescents under the age of 16 than does psychopathology (Brent, 1999). Other common methods used by males for completed suicides in the United States include jumping, hanging, and carbon monoxide poisoning. For females, the next most frequent methods include overdosing on pills or ingesting solid and liquid poisons (Minino, Arias, Kochanek, Murphy, & Smith, 2002).
Worldwide, however, hanging is the most common method of suicide. For example, in New Zealand and Australia, 54% and 36%, respectively, of youth suicides were accounted for by hanging (Berman, Jobes, & Silverman, 2006). In other countries with large agrarian societies, the use of pesticides is the most common method of suicide (Eddleston & Phillips, 2004).
The overwhelming majority of adolescent suicide attempts in the United States and the United Kingdom involve intentional overdose (Berman et al., 2006). In one large community study in the United States, ingestion and cutting accounted for 86% of the suicide attempts reported by girls and 45% of those reported by boys (Lewinsohn et al., 1996). In addition to ingestion (20%) and self-cutting (25%), Lewinsohn et al. (1996) found that other common methods used in attempts by boys were firearms (15%), hanging (11%), and “other” (22%), which included activities such as shooting air into one’s veins and running into traffic. When overdosing, adolescents most often use analgesics and prescribed medications (for themselves or their parents), such as antidepressants and tranquilizers (Worden, 1989).
Berman and Jobes (1991) highlight a number of factors that have been identified to explain the choice of method used:
Availability and accessibility (i.e., ease to obtain)
Sociocultural acceptance (i.e., normative use)
Knowledgeability (i.e., familiarity with use)
Social or behavior suggestion (e.g., modeling)
Saliency (e.g., suggested by publicity)
Personal, symbolic meaning of act or setting (e.g., a landmark jumping site such as the Golden Gate Bridge)
Intentionality and rescue-ability (i.e., if intent is high, methods of choice will be those most lethal, most efficient, and least likely to be interfered with). (p. 105)
Given the high number of impulsive suicides among adolescents, clinicians must take into account in their assessments the aforementioned factors, especially availability and accessibility. In addition, substance use at the time of suicidal behavior has been found to be related to the lethality of the method used (Brent et al., 1987).
Medical lethality of method and suicide intent have been found to be highly correlated, although they are certainly not synonymous. Robbins and Alessi (1985) found a high correlation between adolescent inpatients’ suicidal intent and the medical lethality of their suicide attempts. In contrast to these results, however, other researchers have discovered that the medical lethality of the chosen method does not always match the adolescents’ intent to die. For example, Harris and Myers (1997) found that adolescents who overdosed without intent to die (i.e., their intent was to cause drowsiness or unconsciousness) seriously underestimated the dangerousness of their actions. Specifically, 42% of an adolescent sample underestimated the dose of acetaminophen that could cause harm, and 50% underestimated the dose that could cause death. Thus many adolescents seriously underestimate the dangerousness of acetaminophen in overdose and lack knowledge regarding side effects of overdose, including toxicity. These findings indicate that a clinician must assess suicidal intent apart from medical lethality when ascertaining the seriousness of an attempt.
Table 1.2 presents a detailed list of the evidence-based risk factors for suicidal attempts and completions. Many of these behaviors apply to both suicidal adolescents and adults; each factor specific to adolescents has been marked with an asterisk (*). The same demographic and environmental risk factors appear in both Tables 1.1 and 1.2. The second table, however, offers a more detailed picture of personal, environmental, and behavioral characteristics. Those adolescents at high risk tend to have multiple problems across cognitive, emotional, interpersonal, and behavioral domains of functioning. It’s important to note that the factors putting adolescents most at risk for suicidal behaviors overlap significantly with the behavioral criteria for BPD (see the next section). The multiple risk factors for suicide found in these youth also include substance-related and other Axis I mental disorders, comorbid psychopathology, and familial dysfunction, to name a few. These persistent difficulties impair social, school, and occupational functioning, and such impairment in turn increases the likelihood of future suicide attempts. Follow-up studies indicate that about 10–50% of adolescents who attempt suicide make future attempts (Spirito et al., 1989), and that up to about 11% of attempting adolescents eventually die by suicide (Shaffer & Piacentini, 1994). Thus the high rate of continued psychological disturbance exhibited by adolescents who attempt suicide indicates the pressing need for effective psychological interventions for this group. Studies with adults who attempt suicide suggest that treatment may reduce repeated attempts and enhance social adjustment (Linehan, Armstrong, Suarez, Allmon, & Heard, 1991; Shaffer & Piacentini, 1994). Yet as many as 50% of adolescents who attempt suicide fail to receive any follow-up mental health treatment (Spirito et al., 1989), and of those who do receive such treatment, up to 77% do not attend therapy appointments or fail to complete treatment (Trautman et al., 1993). These high rates of noncompliance with treatment further hinder efforts to develop and evaluate psychological interventions for this group. In the next chapter we turn to the treatment outcome research for suicidal behaviors.
TABLE 1.2. Specific Risk Factors for Suicide Attempts and Completed Suicides in Adults and Adolescents
As noted earlier, the diagnosis of BPD in adolescents has historically been a controversial issue. In the recent empirical literature, however, there appears to be initial support for the existence of BPD in adolescents. According to the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text revision (DSM-IV-TR), “recurrent suicidal behavior…or self-mutilating behavior” (American Psychiatric Association, [APA], 2000, p. 710) is the fifth diagnostic criterion for BPD. Completed suicides occur in 8–10% of persons diagnosed with BPD, and self-mutilative acts and suicide threats and attempts are extremely common (APA, 2000).
In reviewing four studies of adolescents and young adults who completed suicide, Linehan et al. (in press) found that with one exception (Rich & Runeson, 1992), the results were remarkably consistent in indicating high rates of personality disorders among these young persons when compared to matched-pair community control groups (Brent et al., 1994; Lesage et al., 1994) and to a sample of adults over 30 years of age who completed suicide (Rich et al., 1986). As noted earlier, it has been suggested that the co-occurrence of mood and personality disorders represents a particularly significant risk factor for suicidal behaviors (Blumenthal & Kupfer, 1986).
Numerous studies suggest that BPD can be validly and reliably diagnosed in adolescents (Bernstein et al., 1993; Marton et al., 1989; Chanen et al., 2004; Bradley, Jenei, & Weston, 2005). In fact, in a large epidemiological study still underway, Zanarini (2003) reports that 3.3% of 10,000 children age 11 assessed in Great Britain met full diagnostic criteria for BPD. These numbers are higher than the estimated 2% prevalence rate for BPD in adults in the general population (APA, 2000); they thus imply the potential for some children and adolescents to “mature out” of the BPD diagnosis within adulthood, especially if treatment is sought. Unfortunately, however, some researchers suggest that for a subgroup of adolescents, the BPD diagnosis is stable.
Crawford, Cohen, and Brook (2001) examined the dimensional stability of behavioral criteria for DSM-IV Cluster B personality disorders over an 8-year period in a sample of 408 community adolescents who were not receiving treatment. The results indicated that these criteria had moderate stability (.63 for boys, .69 for girls) across time. Interestingly, the stability estimates for Cluster B behavioral criteria were drastically reduced when assessed as categorical diagnoses; this suggests that the stability of personality dysfunction in adolescents may be better detected via a dimensional approach, due to the fact that adolescence is a fluid developmental period. Chanen et al. (2004) assessed the stability of personality disorder diagnoses among adolescent outpatients (n = 101) over 2 years; they used the Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II; First, Gibbon, Spitzer, Williams, & Benjamin, 1997) which allows for both categorical and dimensional assessment of personality disorders, in general and BPD more specifically. Eleven of 101 participants met criteria for BPD at baseline, and 12 of 96 participants met criteria for BPD at follow-up. Consistent with Crawford et al.’s work, Cluster B behavioral criteria showed moderate stability across time (intraclass correlation coefficient = .61, r = .63).
Researchers exploring the validity of BPD in adolescents point to the consistent relationships found between BPD and associated areas of dysfunction and distress as evidence of diagnostic validity. Levy et al. (1999) investigated both the concurrent and predictive validity of personality disorder diagnoses in adolescents via baseline and 2-year follow-up assessments of 142 inpatient adolescents on various clinician-rated and self-rated measures of distress and dysfunction. From the total sample, 86 participants were diagnosed with a personality disorder; 71 (89%) of these were diagnosed with BPD, so we assume that the results pertain most closely to a diagnosis of BPD. Consistent with other research, support was found for the concurrent validity of a personality disorder diagnosis in adolescent inpatients. In other words, at baseline, adolescents with a personality disorder (most often BPD) were more functionally impaired as measured by Global Assessment of Functioning Scale scores than those without a personality disorder. In addition, adolescents with a personality disorder scored significantly higher on 10 of the 12 Symptom Checklist-90—Revised (SCL-90-R; Derogatis, 1994) sub-scales, indicating that such a diagnosis is associated with functional impairments. Lastly, those with a personality disorder showed significantly greater dysfunction in the areas of drug use and further psychiatric hospitalizations at the 2-year follow-up than those without personality disorder (Levy et al., 1999). Kasen, Cohen, Skodol, Johnson, and Brook (1999) found that the odds of a personality disorder in young adulthood increased, given a personality disorder during adolescence in the same cluster. Hence, these patients are at high risk for ongoing problems in multiple domains of functioning. In sum, the presence of a BPD diagnosis in adolescence is associated with significant functional impairment and poor prognosis.
In a recent study examining gender differences among adolescents diagnosed with BPD, Bradley et al. (2005) found that the behavioral criteria and phenomenology of adolescent girls with BPD were similar to those of adults. In contrast, adolescent boys meeting BPD criteria had a more aggressive, disruptive, antisocial presentation. These results require further investigation.
However, personality disorders tend not to be diagnosed in multiproblem suicidal adolescents. Reasons that clinicians may not apply the diagnosis to teens include, but are not limited to, the following:
Reasons of training (many child and adolescent mental health professionals are not trained to assess personality disorders in adolescents, due to the general belief that they do not exist that early).
Questions regarding the reliability and the validity of the diagnosis in adolescents.
Reasons of perceived competence (many mental health professionals do not believe that they are competent to treat personality disorders in any age group).
The wish to maintain a sense of hope about a young person’s prognosis (since personality disorders have historically been recalcitrant to standard therapies).
Concerns about stigmatizing an adolescent with a personality disorder diagnosis.
Fiscal reasons (insurance companies often will not reimburse treatment for personality disorders).
The belief that the DSM-IV-TR system is nondevelopmental and thereby does not take into account childhood traits and behavior problems that are continuous with adult personality disorders (Kernberg, Weiner, & Bardenstein, 2000; Miller, Muehlenkamp, & Jacobson, 2006).
Despite all these doubts and beliefs, the pattern of results in the empirical literature indicates that the prevalence, reliability, and validity of BPD in adolescent samples are largely comparable to those found among adults with BPD (Miller et al., 2006). This comparability in and of itself suggests that BPD appears to operate in a similar fashion and has a similar course, regardless of age and developmental period. Studies also clearly indicate that while there is a legitimate subgroup of severely affected adolescents for whom the diagnosis remains stable over time, there appears to be a less severely affected subgroup that moves in and out of the diagnosis. Hence there is clinical relevance in identifing those for whom the diagnosis is stable, so as to provide appropriate treatments. Regardless of the presence of a full-fledged disorder, BPD behavioral criteria in an adolescent (even if fewer than five are present) may indeed accurately reflect significant distress and dysfunction (e.g., suicidality, self-cutting, identity disturbance, academic failure, social dysfunction, disturbed eating, and substance abuse) that require intervention. If more clinicians assess for and consider the diagnosis of BPD, many more adolescents will be appropriately assessed and treated for their BPD behavioral criteria; as a result, fewer will develop an ingrained and refractory pattern of dysfunctional behaviors, and fewer will be at heightened risk for suicidal behaviors and NSIB (Miller et al., 2006).
With the emerging empirically supported treatment for BPD (Lieb, Zanarini, Schmal, Lineban, & Bohus, 2004), coupled with the findings from longitudinal studies that BPD remits in large numbers of treated adults (Paris & Zweig-Frank, 2001; Zanarini, Frankenburg, Hennen, & Silk, 2003), the BPD diagnosis should no longer be considered “hopeless.” Furthermore, insurance is increasingly reimbursing for DBT as treatment for a BPD diagnosis in adults. Given the current situation, lack of appropriate diagnosis can function as a substantial barrier to effective treatment. At the same time, more research is indicated to further clarify the issues pertaining to the diagnosis of BPD in adolescents.
The factors that put adolescents most at risk for suicidal behaviors significantly overlap with the behavioral criteria for BPD. The multiple risk factors for suicide found in these youth have been reviewed in this chapter. The next chapter reviews the existing treatments for multiproblem suicidal adolescents.