Show Summary Details
Page of

Prevention of Anxiety Disorders 

Prevention of Anxiety Disorders
Prevention of Anxiety Disorders
Page of

PRINTED FROM OXFORD MEDICINE ONLINE ( © Oxford University Press, 2021. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy and Legal Notice).

date: 08 December 2021


The case for efforts in preventing anxiety disorder in youth has been made elsewhere (e.g., Weissberg, Kumpfer, & Seligman, 2003), but warrants brief reiteration here: (1) anxiety disorders are common (Kessler, 1994); (2) pediatric onset is also common (March, 1995); (3) anxiety disorders are associated with significant morbidity and comorbidity that often extends into adulthood (Costello & Angold, 1995); (4) the economic burden of anxiety disorders in the United States is enormous ($42.3 billion in 1990; Greenberg et al., 1999); and (5) most pediatric suf-ferers do not receive adequate care (Kendall & Southam-Gerow, 1995). Prevention efforts should target both risk and protective factors associated with the etiology and maintenance of the disorders. Some risk and protective factors may be less modifiable than others (e.g., gender, familial factors), and thus the interventions need to target mediating variables. The issue of timing is also important to consider because certain risk and protective factors may be more likely to exert their influence during certain developmental periods relative to other periods. For example, it may be when an adolescent needs to make the transition from middle to high school that being behaviorally inhibited (described in Chapter 9 and briefly below) heightens the adolescent's risk for developing an anxiety disorder. Consequently, the development of effective prevention of anxiety disorders will require (1) comprehensive knowledge of the risk and protective factors as well as their complex interrelationships during different periods in development; (2) advances in methods to detect the presence and/or absence of these factors; and (3) interventions that increase protective factors and/or reduce risk factors, or both. The goal of such programs is to reduce the enormous individual and societal burdens imposed by anxiety disorders.

In reviewing the studies on prevention of anxiety disorder conducted to date with children and adolescents, we used the system advocated by the Institute of Medicine's Committee on Prevention of Mental Disorders (Mrazek & Haggerty, 1994; Munoz, Mrazek, & Haggerty, 1996) and adopted by several prevention experts (e.g., Craske & Zucker, 2001; Donovan & Spence, 2000; Winett, 1998a). According to this system, prevention programs are classified as (1) indicated prevention programs, which target at-risk individuals who already have symptoms and/or a biological marker but do not fully meet diagnostic criteria for the disorder; (2) selective prevention programs, which target individuals presumed to be at high risk for the development of a disorder (e.g., witnesses of violence); and (3) universal prevention programs, in which entire populations are targeted regardless of risk factors (e.g., third graders).

Before considering the intervention studies, it is important to briefly consider what is known about risk and protective factors at the individual, familial, and societal level, because it is knowledge of these factors and their interrelations that should inform the development of specific intervention strategies. Unfortunately, knowledge of such factors is limited, and perhaps the paucity of prevention studies in anxiety disorders is a direct result of this limited knowledge.

Of particular concern is the absence of evidence about protective factors that are specific to anxiety disorders. That is, although the youth resilience literature has generally underscored the importance of factors such as high IQ, self-esteem, social support, and positive coping in serving to protect young people from the development of psychopathology in general, there is a paucity of literature regarding whether any protective factor(s) may serve to protect against anxiety disorders in particular. Certainly, development of effective prevention programs will continue to be hampered until evidence-based knowledge has accumulated in this area. The summary below is thus reflective of this in proportion to the literature; that is, considerable more coverage is paid to risk factors than to protective factors.


Individual Psychological Characteristics

Elevated but Subsyndromal Anxiety Symptoms

Many children exhibit symptoms of anxiety at some time before adulthood, and two questions are of particular interest here: (1) Do children with elevated but subsyndromal levels of anxiety show greater than normal levels of impaired functioning in their roles at home, at school, or with peers? (2) Does subsyndromal anxiety predict later psychiatric disorder, whether an anxiety disorder or some other diagnosis? To answer both questions it is necessary to control for comorbidity with other symptoms and disorders; that is, impaired functioning or future anxiety disorder must be linked directly to the anxiety symptoms, not to other symptoms or disorders that may co-occur.

To address the question of whether adolescents with elevated but subsyndromal levels of anxiety show greater than normal levels of impaired functioning in their roles at home, at school, or with peers, it is helpful to draw upon data from the Great Smoky Mountains Study of youth aged 9 to 16 (Costello, Mustillo, Erkanli, Keeler, & Angold, 2004). In this study, in which 1,420 children and adolescents and their parents were interviewed annually, children and adolescents with an anxiety disorder but no other psychiatric diagnoses were twice as likely to show functional impairment as those with no disorder. Even among youths with no diagnoses, those with symptoms of anxiety were twice as likely to have impaired functioning compared to those with no symptoms. This was true of both pre-and postpubertal youths. Thus, in this population-based sample, subsyndromal anxiety symptoms were associated with youths' impaired ability to function well at home, at school, and with peers.

Regarding whether subsyndromal anxiety predict later psychiatric disorder, data from the same study were used to compare children and adolescents who had an anxiety disorder at least once in the 8-year period of observation with those who had never had an anxiety disorder. Youths who had an anxiety disorder at least once during that period had an average of two symptoms during the years when they did not have a diagnosis; the average for those who never had an anxiety disorder was 0.4 symptoms. This finding suggests that youths with a vulnerability to anxiety disorders show clinical symptoms even at times when they would not meet formal diagnostic criteria.

Among children and adolescents without a history of anxiety disorders, those who developed one disorder in any given year of the study had three times as many subsyndromal anxiety symptoms in the year before they developed a disorder compared to those who did not develop an anxiety disorder (2.0 vs. 0.7 symptoms). Almost half of the youths who developed a new anxiety disorder the following year had at least two clinically significant symptoms the previous year, compared with one in five youths who would not develop a disorder. This finding suggests that it should be possible to identify high-risk children and adolescents for prevention programs with a high degree of accuracy.

It is important to remember, however, that anxiety disorders are not all that common among children and adolescents. In the Great Smoky Mountains Study, although subsyndromal symptoms quadrupled the likelihood that a youth without a previous history of anxiety disorders would develop one, the likelihood was increased from 1% to 4% only. It follows that 96% of the children and adolescents with two or more anxiety symptoms did not develop a disorder within the next year. Overall, children and adolescents with the highest likelihood of an anxiety disorder were those with a past history of anxiety disorders (13%).

Autonomic Reactivity

Although research findings are consistent in showing that children and adolescents who display anxiety display alterations in autonomic reactivity, Sweeny and Pine (2004) have noted limits in studies that have relied on cardiovascular measures as indices of autonomic activity. These limits include the fact that cardiovascular measures are regulated by a wide variety of neural structures and thus provide relatively indirect information about the state of brain systems that might be implicated in anxiety disorders. In addition, abnormalities in cardiovascular control appear to occur in other conditions, and so they do not appear to be specific to anxiety disorders. The context in which cardiovascular measures are obtained can also influence any reactivity that might be observed on these measures, thereby raising a concern about whether such findings are actually epiphenomena (Sweeny & Pine, 2004).

Respiratory indexes, in contrast, are relatively free of the limits noted above with cardiovascular measures (Sweeny & Pine, 2004). Respiratory indexes that have been used in this area of research include minute ventilation (the amount of air breathed every minute), tidal volume (size of each breath), and respiratory rate. Guided by Klein's (1993) theory that panic attacks are a suffocation alarm triggered by cues of suffocation, most of the research using respiratory indexes has focused on using samples of patients with panic disorder or an anxiety disorder other than panic disorder as well as “normal” controls and have had the participants breathe air that has an increased concentration of carbon dioxide.

Although research findings generally show that patients with panic disorder experience high degrees of anxiety, panic attacks, and changes in respiratory parameters in response to carbon dioxide exposure whereas other patient groups and normal controls do not (e.g., Papp et al., 1993; Papp, Martinez, Klein, Coplan, & Gorman, 1995), these findings have not emerged in all studies (e.g., Rapee, Brown, Antony, & Barlow, 1992; Woods & Charney, 1998). Only one study (Pine, Cohen, Gurley, Brook, & Ma, 1998) has extended this work to young people (ages 7 to 17; mixed sample of anxiety disorders), but separate analyses were not conducted for the preadolescent versus adolescent subsamples. Pine et al.'s findings with this sample of youth paralleled the positive findings obtained with adults. In light of the paucity of research conducted with adolescent samples, considerable more research is needed before firm conclusions can be drawn about the influence of adolescents' autonomic reactivity as a risk factor for anxiety disorders.

Behavioral Inhibition

The detailed review of the temperamental vulnerability to behavioral inhibition was also presented earlier. It is currently known from two independent labs that children who were highly reactive to novel stimuli as infants were more likely than others to display extreme shyness, timidity, and restraint to unfamiliar people, situations, and objects when they were 2, 4, 7, and 11 years of age (Fox, Henderson, Rabin, Caikins, & Schmidt, 2001; Kagan, 2002), and were more likely to show biological differences that may implicate the amygdala. Although these children are at 3-to 4-fold increased risk for development of an anxiety disorder compared to other children, most do not go on to develop one. Thus, behavioral inhibition is not a strong predictor of later anxiety disorder. This finding points to the importance of identifying protective factors that limit the rate of later anxiety disorders in vulnerable individuals.

Cognitive Factors

There are several characteristics of the individual that have been linked with anxiety and its disorders in children and adolescents. In Chapter 9, mention was made of information-processing biases and anxiety sensitivity. In this section, two additional cognitive characteristics are indicated: coping skills and perceived control. Individuals' coping skills strategies, which refer to a variety of methods individuals employ in an attempt to cope with negative or aversive situations, may be categorized as (1) problem-focused, (2) avoidant, or (3) emotion-focused. Problem-focused coping refers to strategies that either directly address or minimize the effect of the problem. Avoidant coping focuses on either avoiding or escaping the problem. Emotion-focused coping is directed toward the subjective level of distress associated with the problem. There is research evidence that problem-focused methods such as actively seeking out information, positive self-talk, diversion of attention, relaxation, and thought-stopping are associated with reduced levels of anxiety and emotional distress in 8-to 18-year-olds (Brown, O'Keefe, Sanders, & Baker, 1986). Generally, children's and adolescents' use of problem-focused coping strategies has been found to be more associated with positive psychological adjustment than their use of emotion-focused coping strategies. Interestingly, adolescent use of avoidant coping has been found to be associated with high levels of depression in adolescence (Ebata & Moos, 1991). There has been little systematic research on the association between specific types of coping strategies and the development and maintenance of anxiety disorders in adolescence. There also has been little systematic research on which specific coping skills should be taught to adolescents across diverse anxiety-provoking situations. Research in this area is clearly of importance given that coping-skills training represents a major feature of cognitive-behavioral treatments (see Chapter 10). Another characteristic of the individual that relates to cognitions is individuals' perceived control. Specifically, Barlow (2001) has suggested that children who experience uncontrollable events early in life may develop a propensity to perceive or process events as not being under their control, which for some youngsters may serve as a risk for the development of anxiety and its disorders. Chorpita, Brown, and Barlow (1998) have presented some interesting data showing that perceived control may serve as a mediator of family environment among youths with anxiety disorders. Clearly, further research on the role of perceived control as a protective and risk factor in anxiety disorders is needed, particularly on its specificity (or lack thereof) to anxiety.


As discussed in greater detail in Chapter 9, genetic factors clearly influence the risk for anxiety disorders and, taken together, the epidemiological and genetic data imply distinct biological profiles for the varied anxiety disorders, many of which implicate neurochemical processes. A recent meta-analysis found only a modest genetic contribution to four anxiety categories, and no evidence for a significant effect of shared environment (Hettema, Neale, & Kendler, 2001). When the individual studies themselves are reviewed, however, inconsistencies emerge with respect to the degree to which genetics were implicated in transmission of anxiety disorders; rates appear to vary as a function of the site of the laboratory, as well as the informant supplying the relevant information. There is evidence for genetic contributions to personality traits such as neuroticism, introversion (Eaves, Eysenck, & Martin, 1989), shyness (Daniels & Plomin, 1985), and behavioral inhibition (DiLalla, Kagan, & Reznick, 1994; Kagan, 1994), each of which may increase risk for the subsequent development of anxiety disorder. In general, many studies of the genetics of anxiety disorders involving children and adolescents have substantive methodological limitations, thus there remains a great deal to discover in this area. Also, it is important to note that the presence of a genetic influence for anxiety disorders does not imply that the course of illness is immutable. From the perspective of prevention, it may be that studying other risk factors in youth at genetic risk for anxiety disorders may prove especially fruitful, and may suggest roads to interventions that reduce the genetic risk.


Familial Factors

Parent–Child Interaction and Attachment

All four of the attachment styles in children according to the classification by Ainsworth, Blehar, Waters, and Wall (1978) and by Main and Solomon (1990)—secure, insecure-avoidant, insecure-ambivalent, and insecure-disorganized —have been found to be represented in children with anxiety disorders. The highest risks for developing an anxiety disorder are associated with disorganized attachment, which is associated with unresolved trauma or loss, and ambivalent attachment (Cassidy, 1995; Manassis, Bradley, Goldberg, Hood, & Swinson, 1994; Warren, Huston, Egeland, & Sroufe, 1997). The specificity of an association between disorganized attachment in terms of its link with a specific type of anxiety disorder, such as separation anxiety disorder, has not been established.

Retrospective studies.

Lutz and Hock (1995) examined whether adult mental representations of attachment relationships and memories of childhood experiences with parents contributed to a mother's anxiety about separation from her own infant. Mothers with insecure attachment representations, when asked to remember details of their own childhood, reported more negative recollections of early parental caregiving, particularly rejection and discouragement of independence. Cassidy (1995) found that adolescents and adults with generalized anxiety disorder reported more caregiver unresponsiveness, role-reversal and enmeshment, and feelings of anger and vulnerability toward their mothers than controls. Systematic and formal assessments of the adolescents' and adult attachment styles were not conducted in this sample, however.

Prospective studies.

Manassis et al. (1994) examined adult attachment and mother–child attachment in 20 mother–child dyads (children ages 18 to 59 months) in which the mothers suffered from anxiety disorders. The mothers all had insecure adult attachments, and 80% also had insecure attachments with their children. Among the insecurely attached children, 3 of 16 met diagnostic criteria for anxiety disorders; none of the secure children did. Two had separation anxiety disorder (one with disorganized attachment, one with avoidant attachment) and one had avoidant disorder (with disorganized attachment). Insecure children also had higher internalizing scores on the Child Behavior Checklist than those of secure children. When the dyads who had been classified as disorganized and mothers who had been classified as unresolved were assigned their “best” alternate category, and combined with the remaining three attachment categories, a higher than expected rate of ambivalent/resistant attachment and a lower than expected rate of secure attachment were found.

Warren et al. (1997) studied 172 adolescents aged 17.5 years who had participated in assessments of mother–child attachment at 12 months of age. Of these 172 adolescents, 26 (15%) met diagnostic criteria for anxiety disorders. More of the disordered adolescents were classified as anxious/resistant in infancy than the adolescent without anxiety disorders. More adolescents diagnosed with other disorders (not anxiety) were, as infants, classified as avoidant. Furthermore, being classified as anxious/resistant attachment doubled the risk of subsequently developing an anxiety disorder and better predicted adolescent anxiety disorders than either maternal anxiety or child temperament. The interaction between anxious/resistant attachment and one aspect of temperament (slow habituation to stimuli) further increased the risk of a subsequent anxiety disorder. However, secure, insecure-avoidant, and insecure-resistant attachment were all represented among the adolescents with anxiety disorders (data on the insecure-disorganized classification were unavailable).

Linkages have also been found between attachment and subclinical levels of anxiety. Female undergraduates who were insecurely attached were perceived by their friends as being more anxious than their counterparts who were securely attached (Barnas, Pollina, & Cummings 1991). Crowell, O'Connor, Wollmers, and Sprafkin (1991) found that children with behavioral disturbances whose mothers were classified as secure on the Adult Attachment Interview rated themselves as less anxious and depressed than children with behavioral disturbances whose mothers were insecure-dismissing. Cassidy and Berlin (1994) reported increased fearfulness across several studies of insecure-ambivalent/resistant children.

Belsky and Rovine (1987) have suggested a potential linkage between attachment and anxiety when attachment is placed on a spectrum from the style associated with the most overt distress (ambivalent/resistant) to that associated with the least overt distress (avoidant). Secure individuals are in the middle of the spectrum, with some exhibiting relatively high distress and some exhibiting relatively low distress (Belsky & Rovine, 1987). Consistent with Belsky and Rovine (1987), 2.5-year-old children who were either insecure-ambivalent/resistant or secure with relatively high distress showed higher indices of fear and separation distress than children in the other attachment classifications (Stevenson-Hinde & Shouldice, 1990).

In summary, insecure attachment has been linked with both clinical and subclinical anxiety in children of different age ranges. The link may be stronger when the child also has temperamental vulnerability to anxiety, although the evidence for this is not as clear. Limitations of this research include paucity of prospective studies, the varying definitions of anxiety (e.g., anxiety symptoms, anxiety disorders) used across studies, and small sample sizes.


The research conducted on parenting has focused primarily on parental rearing styles, with the latter conceptualized along two orthogonal dimensions: warmth versus hostility, and control versus autonomy (Boer, 1998; Cassidy, 1995; Dadds, Barrett, Rapee, & Ryan, 1996; Lutz & Hock, 1995; Manassis et al., 1994; Rapee 1997; Siqueland, Kendall, & Steinberg, 1996; Warren et al., 1997).

Retrospective reports.

In a meta-analysis of five studies, with a total of 463 patients in the experimental groups, Gerlsma, Emmelkamp, and Arrindell (1990) found that adults with phobias reported a parental rearing style characterized by less affection and more control. Studies of adults meeting diagnostic criteria for panic disorder or social phobia/avoidant personality disorder have demonstrated a similar recollection of child-rearing patterns, in that these adults view their parents, and their relationship with them, as low in affection and overcontrolling (Rapee, 1997).

Empirical research has documented an influence of parental rearing styles on the development of anxiety (see Rapee, 1997, for review). Interestingly, adults with insecure-preoccupied attachments frequently report parental rejection and control (Main & Goldwyn, 1991), which suggests that parenting style may be related to adult attachment status.

Prospective reports and behavioral observations.

In an early study, Bush, Melamed, & Cockrell (1989), using a self-report measure of parental rearing patterns, found parental reported use of positive reinforcement, modeling, and persuasion was associated with lower levels of child anxiety during their child's undergoing of a fearful medical procedure; parental use of punishment, physical force, and reinforcement of dependency was associated with higher levels during the medical procedure. Siqueland et al. (1996) found that parents of children with anxiety disorders were rated by observers as less granting of psychological autonomy than were the parents of “normal” controls. In addition, children with anxiety disorders rated their mothers and fathers as less accepting and less granting of psychological autonomy compared to control children's ratings of their parents.

Direct observations of parent–child interactions have provided further evidence of family processes that may be specific to families of children with anxiety disorders, and these processes may serve to either bring out and/or maintain these disorders in children (e.g., Chorpita, Albano, & Barlow, 1996; Dadds et al., 1996; Ginsburg, Silverman, & Kurtines, 1995). For example, Dadds et al. (1996) studied specific sequences of communication exchanged between parents and children (ages 7 to 14) in a discussion of ambiguous hypothetical situations. Parents of children with anxiety disorders (n = 66) were less likely to grant and reward autonomy of thought and action than controls (n = 18). Dadds et al. also found that these parents fostered cautiousness and avoidance of taking a social risk by modeling caution, providing information about risk, expressing doubt about the child's competency, and rewarding the child for avoidance by expressing agreement and nurturance when the child decided he or she would not join in with the other children. Dadds et al. referred to this finding as the FEAR effect (Family Enhancement of Avoidant and Aggressive Responses).

In a study with 16 children (mean age = 11 years) of agoraphobic mothers and 16 children of mothers with no history of psychopathology matched by age, gender, and socioeconomic status (Capps, Sigman, Sena, & Henker, 1996), agoraphobic mothers reported more maternal separation anxiety with regard to their child than the control group. Maternal separation anxiety correlated negatively with children's perceived control (Capps et al., 1996). The effect probably is best interpreted as the result of a reciprocal relation between caregiver and child: when a child is more anxious, there may be greater cause for the parent's anxiety about separation.

More recently, Hudson and Rapee (2002) studied 57 children and adolescents (37 children with anxiety disorders, 20 nonclinic-referred children; aged 7 to 16 years) and found that mothers and fathers were overly involved not only with their anxiety-disordered child but also with the child's sibling (without anxiety disorders). The authors concluded that because parental overinvolvement does not occur exclusively in youths with anxiety disorders, it probably is not simply a response to difficulties with anxiety and coping that they have observed with the diagnosed youth. It also suggests that parental overinvolvement does not in and of itself cause anxiety disorders.

Anxious parents could increase their offsprings' risk of anxiety disorders by (1) having difficulty modeling appropriate coping strategies; (2) reacting to their children's fears negatively because they represent an aspect of themselves that they would rather deny; or (3) becoming overly concerned about their children's anxiety, resulting in overprotection and thus reducing opportunities for desensitization. The latter two reactions are consistent with dismissive and preoccupied adult attachment types, respectively. Anxious parents who are securely attached, by contrast, may be able to empathize with their children's fears, which may then be perceived as supportive. Thus, the transmission of parental anxiety may depend on the interaction between attachment and parental psychopathology (Radke-Yarrow, DeMulder, & Belmont, 1995).

Peer, School, and Community

The ecology of adolescent development and culture includes an expanded network of peer, school, and community affiliations. The transition to middle and high school constitutes a period of high developmental risk, in which there is an increased incidence of school truancy, failure and dropout, engagement in high-risk sexual and self-injurious behaviors, smoking and drug use, initiation into gangs, and contact with the juvenile justice system. It is also a period of increased exposure to interpersonal violence. For example, in 1999, almost 10% of 9th to 12th graders reported being hit or physically hurt by a boyfriend or girlfriend. In the sections below, particular high-risk activities engaged in by adolescents and their associated risk with anxiety are discussed.


Initiation into cigarette smoking in adolescents is recognized as a major public health problem. As summarized by Upadhyaya, Deas, Brady, and Kruesi (2002) from a number of national surveys, approximately 3,000 adolescents start smoking each day, resulting in about 21% of high school seniors smoking daily, and a total of 4 million adolescent smokers. Smoking prevention and early treatment are important components of universal and selective public health prevention strategies, especially given that the American Health Association estimates that addiction to tobacco during adolescence accounts for 80% of adult smokers. As Upadhyaya et al. (2002) discuss, there is continuing interest in the interaction between onset of adolescent psychiatric conditions and smoking behavior, including experimental smoking and cessation difficulty. Among the disorders studied, Johnson et al. (2000) report that heavy cigarette smoking (defined as over 20 cigarettes per day) is associated with higher rates of agoraphobia, anxiety, and panic disorders in adolescents. Other studies have reported an even stronger association of adolescent smoking with attention-deficit hyperactivity disorder (Johnson et. al., 2000) and major depressive disorder (Dierker et al., 2001). Most of these studies note the importance of the relationship between peer smoking influences and individual psychiatric vulnerabilities. The general conclusion is two-pronged: first, smoking prevention and cessation programs need to incorporate screening for adolescent psychiatric disorders, including anxiety disorders and, second, attention to adolescent anxiety and comorbid disorders need to include strategies to address risks of tobacco addiction.

Drug Use

Adolescence is a developmental period in which experimentation with alcohol and drugs is com mon. It also is a time of development risk for early onset of alcohol and substance abuse and dependence. Nelson and Wittchen (1998) found that among youth and young adults, the peak incidence of alcohol disorders occurred at 16–17 years of age. Alcohol and drug use problems in adolescents are a strong predictive factor of adult alcohol and drug dependence (Swadi, 1999). Studies of substance abuse and alcohol motivation in adolescents suggest a multifactorial explanatory framework. Among the many factors, Comeau, Stuart, and Loba (2001) found that high anxiety sensitivity predicts conformity motives for alcohol and marijuana use, whereas anxiety traits are associated with coping motives for alcohol and cigarette use. Zucker, Craske, Barrios, and Holguin (2002) reported that among young adults with panic disorder, up to one in five cases had an onset related to an adolescent experience with a psychoactive drug. In a review of studies of adolescent use of the recreational drug “ecstasy,” Montoya, Sorrentino, Lukas, and Price (2002) found a strong association between repeated drug use and anxiety disturbances, with potential neurobiological consequences that are of concern within this critical developmental stage.

Initiation and use of alcohol and drugs among adolescents are also related to life stresses, including traumatic events (Wills, Vaccaro, & McNammar, 1992). In one study, substance-abusing adolescents were found to be five times more likely to have a history of trauma and concurrent posttraumatic stress disorder (PTSD) compared to a community sample (Deykin & Buka, 1997). In a large study of adolescents enrolled in four drug treatment programs, a high positive correlation was found between severity of posttraumatic stress symptoms and higher levels of substance use and HIV risk behavior (Stevens, Murphy, & McNight, 2003).

As with cigarette addiction, prevention strategies for adolescent substance abuse need to include early intervention for anxiety vulnerable and traumatized youth and at the same time recognize that prevention or early intervention for adolescent substance abuse may also constitute an anxiety disorder prevention strategy.

Gang Affiliation and Other Criminal Behavior

Gang affiliation is a serious cultural problem in adolescence. There are an estimated 24,000 gangs, with over 772,000 members active across the United States (U.S. Department of Justice, 2002). There is a complexity to youth involvement in gangs. Many studies have examined the confluence of risk factors that predict gang membership, including neighborhood, family, school, peer group, and individual variables (Hill, Levermore, Twaite, & Jones, 1996). There is an emerging literature about the extent of trauma and loss exposure associated with gang membership and delinquent behavior more generally (Wood, Foy, Layne, Pynoos, & James, 2002). Despite high rates of trauma exposure prior to gang membership, commonly youth report that their worst traumatic experiences are gang related and the source of current PTSD symptoms (Wood et al., 2002). Ages 11–13 are primary years for solicitation and inculcation into gang affiliation and activities, contributing to years of increased trauma and loss exposure during adolescence. Consequently, intervention programs to prevent youth from becoming involved in gangs should be considered an adjunct prevention strategy for adolescent PTSD.

Adolescence is also the time when involvement in the justice system accelerates. Approximately 1.8 million youth go through the juvenile justice system each year, with over 360,000 detained and 176,000 incarcerated (U.S. Department of Justice, 2002). Recently, attention has turned to the high prevalence of adolescent psychiatric disorder present among juvenile justice detainees. Of importance, the rate of anxiety disorders is as high as one in three and equal to or exceeds those of mood disorder. Of the anxiety disorders, studies vary in the prevalence of specific anxiety disorders. Studies that have assessed PTSD have found it to be among the highest (Wasserman, McReynolds, Lucal, Fisher, & Santos, 2002). Interestingly, separation anxiety disorder among adolescents (an age-range where it is less expected) is surprisingly high among African-American and Hispanic and Latino detained youth (Teplin, Abram, McClelland, Dulcan, & Mericle, 2002). The juvenile justice contact provides a key opportunity for mental health intervention that can play a significant role in an overall public mental health approach to adolescent anxiety disorders and delinquency prevention programs.

Social Support

The adult literature is replete with studies that suggest the possible beneficial effects of social support following exposure to traumatic events, but less is known about its role in mitigating anxiety disorder symptoms outside the context of trauma, and even less about the influence of social support in adolescent anxiety disorders. Studies of veterans from the Vietnam, Gulf, and Lebanon wars have found that veterans' perceptions of poor social support are associated with worse PTSD symptoms; the relationship remains regardless of whether veterans report retrospectively about the support they received immediately after his or her return from duty (Barrett & Mizes, 1988; Fontana, Schwartz, & Rosenheck, 1997; Foy, Resnick, Sipprelle, & Carroll, 1987; Solomon, Mikulincer, & Avitzur, 1988; Stretch, 1985, 1989; Sutker, Davis, Uddo, & Ditta, 1995) and even when controlling for level of combat exposure, another robust predictor of PTSD symptoms among veterans (Boscarino, 1995; King, Leonard, & March, 1998). Among civilian victims of violence, poor social support also has been linked to PTSD symptoms in victims of violent nonsexual assault (Bisson & Shepherd, 1995), domestic violence (Astin, Lawrence, & Foy, 1993; Kemp, Green, Hovanitz, & Rawlings, 1995), and rape survivors (Resick, 1993; Steketee & Foa, 1987; Zoellner, Foa, & Brigidi, 1999). Moreover, Fontana and Rosenheck (1998) found that good postdischarge social support was strongly predictive of less PTSD in female veterans who were victims of sexual harassment, rape, or attempted rape. Social support is also associated with recovery among victims of “noninterpersonal” traumas, such as natural disasters (e.g., Madakasira & O'Brien, 1987), motor vehicle accidents (Buckley, Blanchard, & Hickling, 1996), and chronic, life-threatening illness, including patients treated for breast cancer (Andrykowski & Cordova, 1998), African-American women with HIV/AIDS (Myers & Durvasula, 1999), and survivors of childhood leukemia and their mothers and fathers (Kazak et al., 1997).

There are a number of shortcomings to the extant adult literature on social support in the wake of trauma. First, although a large body of research supports the conclusion that social support is associated with decreased PTSD symptomology (e.g., Greene, Grace, & Gleser, 1985; Keane, Zimering, & Caddell, 1985); most of these studies have relied on retrospective reports (some as many as 30 years after the fact) of social support. Second, the studies typically have aggregated and equally weighted the influence of friends, co-workers, and neighbors with that of immediate family, which may obscure the more influential effects for the latter (Griffith, 1985). Third, and perhaps most importantly for the purpose of considering prevention efforts, no studies have attempted to delineate the mechanism responsible for the apparent positive impact of social support on posttrauma recovery. Pennebaker and Seagal (1999) suggest that painful events, which have not been structured in a narrative format, may contribute to the continued experience of negative feelings and are more likely to remain in consciousness as unwanted thoughts (Wegner, 1989). Foa and Riggs (1993) suggest that trauma disclosure within naturally occurring social support systems provides three potential benefits: First, disclosure enables the trauma survivor to confront frightening memories in a relatively safe environment, allowing habituation of fear reactions, much as is accomplished in flooding treatment of PTSD (Foa et al., 1999; Foa, Feske, Murdock, Kozak, & McCarthy, 1991; Keane, Fairbank, Caddell, & Zimmering, 1989b; Richards, Lovell, & Marks, 1994). Second, given the observation that traumatic memories are often disjointed and confused, disclosure, particularly repeated disclosure, provides the survivor with an opportunity to create a more coherent memory. Finally, disclosure is thought to provide an opportunity for survivors to evaluate potentially mistaken cognitions regarding the impact on themselves (e.g., I am incompetent or worthless) or the world (e.g., the world is unpredictably dangerous). Herman (1992) suggests that disclosure may also serve to “reconnect” the trauma survivor to others within the social arena. That is, the act of disclosing the trauma to another person may provide an opportunity for the survivor to redevelop a sense of trust and attachment to others. Thus, disclosing the trauma to a supportive person may function in multiple ways to facilitate recovery.

There is also evidence that social support may mitigate the impact of negative life events in children and adolescents whose parents are divorcing (Cowen, Wyman, Work, & Parker, 1990) and in those who have been exposed to community violence (Berman, Kurtines, Silverman, & Serafini, 1996; Hill, Levermore, Twaite, & Jones, 1996; White, Bruce, Farrell, & Kliewer, 1998) and hurricanes (e.g., La Greca, Silverman, Vernberg, & Prinstein, 1996; Vernberg, La Greca, Silverman, & Prinstein, 1996). For example, White and colleagues (1998) found a strong neg-ative relation between anxiety level and family social support in a longitudinal study investigating the effects of family social support on anxiety in 11-to 14-year-olds exposed to community violence. Unfortunately, little else has been done to explore the specific role of this potentially important protective factor. Perhaps acceptance into a supportive social network attenuates the effects of the putative anxiety disorder risk factors described earlier. Thus, social support serves as one possible explanation for so many children and adolescents elevated on these risk factors (e.g., behavioral inhibition) not going on to develop full-blown disorders. The importance of the adolescent's peer group suggests that social support may be particularly relevant during this period (see Table 11.1 for a summary of putative risk factors).

Table 11.1 Who May Be at Risk?



Individual Factors

Elevated but subsyndromal anxiety symptoms

Increased risk of developing full-blown disorder in next 2 years if elevated symptoms are already present

Behavioral inhibition (temperament)

Tendency to avoid novel stimuli and experiences; excessive shyness in response to new people

Anxiety sensitivity

Tendency to interpret physiological sensations of anxiety as threatening in and of themselves

Cognitive factors

Avoidant coping style, low perceived control

Family Factors


Insecure attachment, possibly interacting with behaviorally inhibited temperament

Parent–child interactions

Parental tendency to suggest avoidant problem-solving strategies; overinvolvement and overprotection in response to child's fears; poor modeling of coping responses

Peer, School, and Community Factors


Association with panic disorder in particular

Alcohol and other drug use

Elevates other risk factors (e.g., MVAs), may also elevate risk in and of itself

Gang affiliation and criminal behavior

Exposure to traumatic events, commission of interpersonal violence

Trauma exposure

Criterion A trauma increases the risk for PTSD and other anxiety symptoms, perhaps especially in those who are already vulnerable or in response to certain traumas regardless (e.g., sexual assault)

Poor social support

Associated with more symptoms and poorer outcomes in adults, possibly a mediating factor

MVA, motor vehicle accident; PTSD, posttraumatic stress syndrome.

Early Detection and Screening

The success of prevention intervention programs for anxiety disorders in adolescents depends a great deal on having early detection and screening strategies in place at key access points where youths might be identified. The types of early detection and screening strategies are likely to vary with the type of preventive intervention program being implemented (universal, selective, indicated). In this section, key access points are identified and specific types of screens that might be administered, depending on the type of preventive intervention strategy, are summarized.

Before proceeding with this discussion, a general point is first worth noting. For the majority of access points or settings where early detection and screening strategies might be conducted, some type of rating scale is recommended for initial use. Because of their objective scoring procedure, rating scales minimize the role of clinical inference and interpretation. As a result, there is no need to use highly trained staff for administration and scoring. In addition, most rating scales contain questions that would be of clear concern to non-mental health professionals, such as school board institutional review board members, because the scales contain items that are face valid. Finally, a wide range of rating scales is available for administration to various informants, including children and adolescents, as well as parents, teachers, and clinicians. Consequently, information can be obtained from either a single source (e.g., adolescent only) or multiple sources (e.g., adolescent, parent) depending on one's available resources. If resources are limited, the consensus in the field is that information from youths themselves should be obtained for screening and assessing for internalizing problems, including anxiety (Loeber, Green, & Lahey, 1990).

Despite the advantages in using rating scales for the purpose of early detection and screening, several caveats are worth noting. One is that although the measures mentioned in this section all possess adequate psychometric properties in terms of reliability and validity, their actual utility for screening purposes awaits further empirical evaluation. For example, data on the measures' sensitivity (the percentage of individuals who receive the diagnosis who were positively identified by the rating scale; true positives) and specificity (the percentage of individuals who do not receive the diagnosis and who are not identified by the rating scale as anxious; true negatives) (Vecchio, 1966) are scarce when it comes to child and adolescent samples, particularly nonwhite samples. Second, currently available rating scales are likely to select more false positives than true positives (Costello & Angold, 1988). That is, youths identified as anxious at an initial screen are likely not to be anxious or depressed at the second stage of an investigation. Consequently, a useful and cost-efficient approach for early detection and screening for indicated intervention programs would employ a multistage sampling design (e.g., Ialongo, Edelsohn, Werthamer-Larsson, Crockett, & Kellam, 1993; Kendall, Cantwell, & Kazdin, 1989; Roberts, Lewinsohn, & Seeley, 1991). At the first stage, a rating scale would be administered to informants to identify youths who score 1 or 2 standard deviations from the sample mean or who deviate from normative data. These identified cases would then undergo more precise and comprehensive assessments (e.g., structured diagnostic interviews) at the second or third stage of the research.

Potential Access Points for Early Detection and Screening


The school setting is an obvious access point for early detection and screening of anxiety and its disorders because this is where the children and adolescents are. If a preventive interventionist is interested in developing and implementing an indicated prevention program, there are several rating scales that can be administered to target high-risk children and adolescents who may demonstrate minimal but detectable symptoms of anxiety and/or anxiety disorders. In general, most of the research studies that have used rating scales for screening anxiety symptoms and disorders have used largely preadolescent samples of children. There is a paucity of work in which the study's samples involved adolescents specifically.

For anxiety symptoms, the most widely used child and adolescent self-rating scale measure is the Revised Children's Manifest Anxiety Scale (RCMAS; Reynolds & Richmond, 1978), which was used as an initial screen in the Queensland Early Intervention and Prevention of Anxiety Project (Dadds et al., 1999; Dadds, Spence, Holland, Barrett, & Laurens, 1997), described below. The RCMAS is a 37-item scale: 28 items are summed yielding a Total Anxiety score; the other nine items are summed to yield a Lie score. Youths respond either yes or no to all 37 items. Factor analytic studies have also provided support for the RCMAS's three-factor subscale structure (Physiological, Worry/Oversensitivity, and Concentration) as well as the Lie scale (e.g., Reynolds & Richmond, 1978). Positive scale convergence between the RCMAS and other widely used child self-rating scales of anxiety and related constructs (trait anxiety, fear, depression) in community samples have also been found (e.g., Muris, Merckelbach, Ollendick, King, & Bogie, 2002). In addition, there is considerable evidence supporting the reliability of the RCMAS and its subscales in community samples. For example, Pela and Reynolds (1982) reported test–retest reliability of r = .98 of the Total Anxiety scale using a 3-week interval. Internal consistency of the RCMAS is also excellent, with estimates ranging from .82 to .85.

For anxiety symptoms linked more directly to DSM-IV anxiety disorders, the Multidimensional Anxiety Scale for Children (MASC; March, Parker, Sullivan, & Stallings, 1997), the Screen for Child Anxiety Related Emotional Disorders (Birmaher et al., 1997), and the Spence Children's Anxiety Scale may be useful. The MASC, for example, is a 45-item scale that yields a Total Anxiety Disorder Index and four main factor scores: Social Anxiety (with performance anxiety and humiliation as subfactors), Physical Symptoms, (with tension-restlessness, somatic-autonomic arousal as subfactors), Harm/Avoidance (with perfectionism, anxious coping as subfactors), and Separation/Panic. In addition, six items yield an Inconsistency Index to identify careless or contradictory responses. Youths may be identified on the basis of either specific subscale scores on these measures or the total score.

For social anxiety, the Social Anxiety Scale for Children–Revised (La Greca & Stone, 1993) and the adolescent version (La Greca & Lopez, 1998) as well as the Social Phobia and Anxiety Inventory for Children (Beidel, Turner, & Morris, 1995) have been found to be helpful in identifying highly social anxious children (Epkins, 2002; Morris & Masia, 1998), although variations in the two measures' classification correspondence indicated variation with sample, age, and gender. In light of this variation, coupled with the fact that both the Epkins (2002) and Morris and Masia (1998) studies did not sample adolescents (Epkins's sample was 8 to 12 years; Morris and Masia's was 9 to 12 years), additional research on the utility of these measures for screening among adolescents is needed. Moreover, research on all of these scales' utility in the context of prevention is lacking.

The Childhood Anxiety Sensitivity Index (CASI; Silverman, Fleisig, Rabian, & Peterson, 1991) has been used in a number of studies and appears useful as a screen for adolescents who may be at risk for displaying panic attacks and panic disorder (Hayward, Killen, Wilson, & Hammer, 1997; Weems, Hayward, Killen, & Taylor, 2002). In fact, it appears to be the only measure that has been used as a screen for adolescents specifically. Consequently, further description of the scale might be in order.

The CASI consists of 18 items that assess the extent to which children and adolescents believe the experience of anxiety will result in negative consequences. Sample items include: “It scares me when I feel like I am going to throw up” and “It scares me when my heart beats fast.” Youths respond to each item using a 3-point scale: none (1), some (2), or a lot (3). The CASI yields a total score by summing the ratings across all items. The CASI scores can range from 18 to 54, with higher scores reflecting higher levels of anxiety sensitivity. Silverman et al. (1991) reported internal consistency estimates (coefficient alphas) of .87 for both a nonclinical and clinical sample, and test–retest reliability estimates (using a retest interval of 2 weeks) of .76 and .79 for the nonclinical and clinical samples, respectively.

Recent research provides strong support for a hierarchical model for anxiety sensitivity as represented in the CASI with a single second-order factor and four facets (first-order factors) labeled Disease Concerns, Unsteady Concerns, Mental Incapacitation Concerns, and Social Concerns (Silverman, Goedhart, Barrett, & Turner, 2003). This type of research on the facets of anxiety sensitivity has the potential to further our understanding about the prevention of anxiety problems to the extent that the specific facets of anxiety sensitivity may be more related to a given pathological condition than the total anxiety sensitivity construct as recent research suggests (Joiner et al., 2002).

Silverman et al. (2003) also tested for factorial invariance (equal factor loadings) of the completed four-factor model across gender and age (two age groups: children, ages 7 to 11 years, and adolescents, ages 12 to 17 years). The hypothesis of equal factor loadings on the four factors across age and gender and across samples could not be rejected, thereby indicating factor stability. The only age difference to emerge in this study was that children displayed a higher level of Unsteady and Disease Concerns (and consequently a higher level of Total Concerns) than adolescents, a finding that is consistent with age differences in frequency and intensity of common fears (Gullone, 2000). Perhaps a higher level of fearfulness and less knowledge of physical processes in children than in adolescents constitutes a higher level of worrying among children about their physical condition. Whether the emergence in children of higher levels of Unsteady and Disease Concerns is something that would be worth assessing early on by means of the CASI, with possible intervenion as a preventative step, would be an interesting avenue to pursue.

In addition, evidence indicates that a large proportion of children and adolescents who display school refusal behavior are likely suffering from some type of anxiety disorder, particularly separation anxiety disorder in young children and social anxiety, panic, or generalized anxiety disorder in older children and adolescents (Kearney & Silverman, 1997). Thus it is critical that school counselors and psychologists be informed and educated about the nature of school refusal behavior so that they can help detect such cases and refer students for appropriate therapeutic, rather than disciplinary, action.

If a preventive interventionist is interested in developing and implementing a selective prevention program in a school setting, specific groups or individuals considered to be at risk for developing anxiety and its disorders need to be identified. At the preschool level, Rapee (2002) used a mother-completed rating scale of the child's temperament, followed by a laboratory observation of behavioral inhibition, as a screen for selecting youngsters in the Macquarie University Preschool Intervention Program (described below).

In light of the high rates of traumatic exposure among young people, particularly adolescents, youths who have been exposed to traumatic events are another group that should be considered for early detection and screening in the school setting, with focus particularly on posttraumatic stress and anxiety reactions. Successful efforts in such screening, using most frequently the Reaction Index (Frederick, Pynoos, & Nader, 1992), have appeared in the area of community violence (e.g., adolescent sample: Berman, Kurtines, Silverman, & Serafini, 1996), natural disasters (e.g., child sample; La Greca et al., 1996; Vernberg et al., 1996); and sniper shootings (e.g., child sample; Pynoos, Frederick, Nader, & Arroyo, 1987). March, Amaya-Jackson, Terry, and Costanzo (1997) and, more recently, Foa, Johnson, Feeny, and Treadwell (2001) have developed and conducted psychometric evaluation of the Child and Adolescent Trauma Survey (CATS) and the Child PTSD Symptom Scale (CPSS), respectively; both have been found to be psychometrically sound.

The CPSS, for example, assesses traumatic stress symptoms in children and adolescents, 8 to 18 years of age. The CPSS items assess all 17 DSM-IV symptom criteria for PTSD and yield a Total Severity Score (17 items) and three empirically derived factor scale scores representing DSM-IV clusters B (Re-experiencing), C (Avoidance), and D (Arousal). The CPSS also includes a seven-item impairment rating scale to assess functioning in such domains as family, peers, and school. Evidence has indicated moderate to excellent internal consistency, retest reliability, and concurrent validity as well as excellent sensitivity and specificity (Foa, Cashman, Jaycox, & Perry, 1997). All together, these PTSD scales have the potential for use in early detection and screening of youths at high risk for developing posttraumatic stress and anxiety reactions from their exposure to traumatic events. Further evaluative research regarding their utility for such purposes is needed, however. In addition, because many adolescent-onset problems such as cigarette smoking frequently co-occur with anxiety and its disorders, as noted earlier in this section, screening for anxiety with one of the anxiety symptoms scales (e.g., RCMAS, MASC) may be worthwhile to include, whenever beginning work with adolescents for any such problems.

Finally, Beidel and colleagues (Beidel & Turner, 1988; Beidel, Turner, & Trager, 1994) conducted a series of studies showing that the Test Anxiety Scale for Children (Saranson, Davidson, Lighthall, & Waite, 1958) could serve as a useful screen in identifying children who may show detectable symptoms of anxiety disorders, including social anxiety disorder, specific phobia, and generalized anxiety disorder. Clearly, as “high-stakes” testing (e.g., SATs, ACTs) becomes more of a stressor with adolescence, the potential utility of test anxiety as a marker, and the Test Anxiety Scale as a screen among adolescent samples, deserves scrutiny.

Health-care settings.

There are multiple access points for early detection and screening in healthcare settings, particularly in pediatrics, obstetrics-gynecology (ob-gyn), and psychiatry. The pediatric setting, for example, is the natural site for early detection and screening of young children with pediatric onset of either obsessive-compulsive disorder (OCD) or a tic disorder following an abrupt onset of symptoms after a group A β‎-hemolytic streptococcal infection, which is referred to as PANDAS (see Chapter 10). The MASC Obsessive-Compulsive Screener (March, Parker, et al., 1997), together with the Conners-March Developmental Questionnaire, may identify such cases; the former could be used to identify OCD symptoms in both pediatricians' and dermatologists' offices. Indeed, for a large proportion of families, the pediatrician office is the “first gate” they enter when their child or adolescent begins to show disturbances associated with anxiety and its disorders, such as somatic complaints and panic attack symptoms. It would thus seem critical for primary care physicians to have understanding of and knowledge about anxiety disorders so that they could inquire about the presence or absence of key symptoms of the various disorders (described in Chapter 9), and refer, as necessary, to a mental health professional for further evaluation based on the results of these initial queries.

Given the preponderance of female cases of anxiety disorders relative to male cases particularly from adolescence and beyond, ob-gyn settings represent yet another potentially useful and critically important access point for early detection and screening. Studies have demonstrated that pubertal maturation in adolescent girls, particularly early onset, may constitute a risk factor for developing anxiety symptoms and disorders (e.g., Caspi & Moffit, 1991; Graber, Brooks-Gunn, Paikoff, & Warren, 1994), particularly panic attacks (Hayward, Killen, Kraemer, et al., 1997). Such findings suggest the potential utility of educating physicians in the ob-gyn area about the risks of anxiety problems in their young adolescent female patients. The manner in which such young patients may become overly sensitive to the physical changes that occur during the menstrual cycle (i.e., high anxiety sensitivity) might be carefully considered and even assessed using the CASI (Silverman et al., 1991). Relatedly, research findings, albeit sparse, suggest that hormonal fluctuations during the female reproductive cycle may serve to either exacerbate or reduce anxiety symptoms and disorders. For example, among some women, the postpartum period may be a risk for the onset and exacerbation of anxiety symptoms and disorders (March & Yonkers, 2001). Also among some women, pregnancy may be a period when past episodes of panic disorder improves (Shear & Oommen, 1995). In light of such findings, it seems critically important that ob-gyn physicians carefully consider their female patients' emotional states during regularly scheduled appointments. Adult anxiety rating scales, such as the Hamilton Scales, may be worth administering as a potential screen for the presence of anxiety symptoms in these patients.

Finally, psychiatry departments housed in medical settings, community mental health settings, or in private practice represent yet another important, though again, largely untapped access point by which to conduct early detection and screening. In light of the strong evidence for familial transmission of anxiety disorders, adult patients who present with anxiety disorders, depressive disorders, or both (Weissman, 1988) should be carefully queried about the functioning of their children and adolescents. For such purposes, parent rating scales such as the Child Behavior Checklist (CBCL; Achenbach, 1991) and the Connors Rating Scales (Conners, 1997) could be administered. Although parents with anxiety problems are likely to endorse high levels of internalizing problems in their offspring with the CBCL (e.g., Silverman, Cerny, Nelles, & Burke, 1988), some of which might be due to the parent's own pathology, as noted earlier, this initial step is a screen. Further follow-up would then be conducted with the children themselves by means of structured interview schedules, such as the Anxiety Disorders Interview Schedule for Children (ADIS-C): DSM-IV (Silverman & Albano, 1996). This interview schedule is the one most widely used in child and adolescent anxiety disorders and includes a child and a parent version. Previous research demonstrates good to excellent test-retest reliability for the diagnosis of anxiety disorders (e.g., κ‎ = .63 to .83 for the ADIS-C child version, κ‎ = .65 to .88 for the ADIS-C parent version, and κ‎ = .80 to .92 for the composite diagnosis; Silverman, Saavedra, & Pina, 2001).

Web-based surveys.

Finally, recent survey studies involving adults' reactions to the terrorist attacks of September 11, 2001 employed Web-based measures (e.g., Schuster et al., 2001), which allowed for efficient data collection from large numbers of subjects. The method is not without its shortcomings, however, the most prominent of which is the possibility of sampling bias: (1) those most interested in the topic and affected by the events might be most likely to participate, which will perhaps overestimate effects; and (2) despite the appearance of ubiquity in our culture, it is estimated that fewer than 50% of urban homes have personal computers, and thus Web-based surveys are likely to exclude large numbers of participants. It may be that use of Web-based surveys in the school context may be a useful way to proceed, however: it would avert a problem of having little unscheduled time during the school day to conduct research studies, and may improve participation among adolescents, who are increasingly computer-savvy. The problem of excluding youngsters who do not have a computer in their home could be addressed by allowing the use of school and/or researcher-provided laptops for this purpose.

Model Programs

The outcome literature on the prevention of anxiety disorders is insufficiently developed to describe any existing program as a “model” and reflects the fact that the complex interplay among the many putative risk and protective factors for anxiety disorders is not well understood. Nevertheless, several studies have been undertaken and are described below. Most of these studies have been conducted using adult samples; however, a handful of studies have employed pediatric samples. Moreover, as is the case with anxiety disorder treatment studies, no adolescent-specific prevention intervention protocols have been developed as yet. Thus, although there are limitations inherent to extrapolating findings from studies comprised of adult samples to adolescent populations, the paucity of research on prevention of anxiety disorders in adolescents requires greater reliance on the adult literature. Efforts at anxiety disorders prevention are also hampered by insufficient and sometimes inconsistent information about the longitudinal course of disorders, the efficacy of procedures designed to reduce modifiable risk factors (e.g., anxiety sensitivity), the influence of protective factors, and the possible additive if not multiplicative effects of multiple risk factors. Further, as indicated earlier, immersion in adolescent culture represents a time of increased risk for a variety of negative life experiences and, as such, adolescents are at increased risk for the development of at least certain anxiety disorders. Thus, the research literature in this area remains underdeveloped, and addressing these critical gaps will be important in developing adolescent-specific prevention programs.


We have organized the review of the extant literature using the categories recommended by the Institute of Medicine's Committee on Pre vention of Mental Disorders: (1) indicated prevention programs, (2) selective prevention programs, and (3) universal prevention programs. Notably, some prevention intervention programs have been developed to target general psychopathology risk factors (e.g., children whose parents recently divorced; Pedro-Carroll & Cowen, 1985). However, because the link between these broader risk factors and the development of anxiety disorders is even more tenuous than the link between anxiety disorders and the specific anxiety disorder risk factors described above, we limit our discussion here to those studies that focused more specifically on prevention of anxiety symptoms and anxiety disorders. A comprehensive review of these studies is available elsewhere (Hudson, Flannery-Schroeder, & Kendall, 2004).

Indicated prevention programs.

These programs are most similar to the treatments for fully syndromal individuals with which the field is most familiar, in that patients are already experiencing symptoms of a disorder and are at high risk for the development of the full-blown syndrome. Harvey and Bryant (1998) reported that 78% of adult motor vehicle accident survivors who met criteria for acute stress disorder (ASD) suffered from PTSD 6 months after the trauma, compared to only 4% of survivors who did not meet criteria for ASD. Thus, ASD is considered a major risk factor for PTSD. It is important to note, however, that there are many shared features between the two disorders and this overlap may be the main contributor to the predictive relationship between ASD and PTSD. Indeed, severity of PTSD symptoms shortly after the traumatic event is a very strong predictor of PTSD severity later on. Foa, Hearst-Ikeda, and Perry (1995) conducted the first PTSD prevention study. Women who were recent victims of sexual and nonsexual assault and who met symptom criteria for PTSD received either a brief cognitive-behavioral therapy (CBT) program consisting of four weekly 1.5-hr sessions or four weekly assessments of their PTSD related symptoms. At 2 months postintervention assessment, the CBT group had a recovery rate of 10% for PTSD vs. 70% in the assessment control group. Using five sessions of Foa et al.'s prevention program (adding one additional session), Bryant, Harvey, Sackville, Dang, and Basten (1998) compared it to supportive counseling (SC) for male and female survivors of motor vehicle and industrial accidents who met diagnostic criteria for ASD. At posttreatment, only 8% of CBT participants met criteria for PTSD, compared to 83% of the SC patients. Although rates of PTSD increased over the course of a 6-month follow-up, CBT remained superior (17% PTSD incidence) to SC (67% PTSD incidence). In a subsequent study, Bryant, Sackville, Dangh, Moulds, and Guthrie (1999) modified the brief CBT by limiting it to psychoeducation and exposure, eliminating anxiety management (e.g., relaxation training) and cognitive restructuring, and compared this modified protocol to the full protocol and to SC. At posttreatment, 20% of participants in the full treatment program and 14% of participants in the brief CBT met criteria for PTSD, in comparison to 56% of participants in the SC condition. At 6 months, incidence of PTSD was 23%, 15%, and 67% for the full treatment program, exposure, and SC, respectively. In a more recent study, Foa, Zoellner, and Feeny (2004) again found that CBT accelerates recovery after sexual assault, compared to SC.

Another group presumably at risk for the development of an anxiety disorder is adults who present to emergency rooms with panic attack symptoms. Swinson, Soulios, Cox, and Kuch (1992) conducted an intervention study with such individuals, 40% of whom met full symptom criteria for panic disorder, thus rendering this study an indicated prevention/treatment hybrid. Nevertheless, at 6 months follow-up, participants randomized to a 1-hr exposure-based condition were improved on panic and anxiety measures, whereas those assigned to a 1-hr reassurance control intervention were no better. Subgroup analyses examining outcome for those with full syndromal panic disorder and those who were subthreshold were not reported, however. Gardenschwartz and Craske (2001) also targeted the prevention of panic disorder, but recruited college students who had experienced a panic attack within the last year, evidenced elevated anxiety sensitivity, and did not meet DSM criteria for panic disorder. Participants were randomly assigned to either a wait list or a day-long CBT workshop that included psychoeducation about agoraphobia and panic, behavioral and cognitive strategies, and interoceptive exposure. At 6 months follow-up, 14% of the wait-list group had gone on to develop fully syndromal panic disorder, compared with only 2% of the workshop participants; significant effects were also seen on other relevant indices (e.g., panic attack frequency × intensity index). As the authors note, a longer follow-up period may not have yielded similar outcome, as the workshop's effects may have been transient.

Only two randomized control trials (RCTs) involving indicated prevention programs have been conducted using pediatric samples. LaFreniere and Capuano (1997) examined the effects of a program directed at mothers of preschool children (N = 43) exhibiting anxious or withdrawn behavior, comparing it to no treatment. The intervention lasted for 6 months and consisted of four phases: (1) assessment; (2) education of the parents about their child's developmental needs; (3) determination of specific objectives for the family; and (4) implementation of the intervention during 11 home visits with child-directed interaction, modification of behavior problems, training in parenting skills, and enhancement of the effectiveness of social support systems. Given the age of the children, outcome variables included teacher ratings in social competence within the preschool setting and cooperation and enthusiasm during a problem-solving task rather than symptoms of a specific anxiety disorder. Results indicated that maternal stress was reduced and anxious–withdrawn behavior of the child was significantly lower at posttreatment in both conditions, although the social competence of children whose mothers received the intervention was greater prior to intervention than that of those whose mother received no treatment. The relatively brief follow-up period and the lack of information about anxiety disorder symptoms limit the utility of the findings. Nevertheless, the study offers preliminary and encouraging findings about the potential benefit of such programs for behaviorally inhibited young children.

The second pediatric study, known as the Queensland Early Intervention and Prevention of Anxiety Project, constitutes the most comprehensive effort made thus far in evaluating the efficacy of a prevention program for children and adolescents (Dadds et al., 1997, 1999). As in the Swinson et al. (1992) adult panic disorder prevention study described above, Dadds et al.'s study can be better characterized as a hybrid-indicated prevention and early-intervention study because 55% of the selected children met diagnostic criteria for at least one anxiety disorder. A total of 1,786 children (ages 7 to 14 years) were screened for anxiety problems by use of teacher nominations and children's self-ratings. After initial diagnostic interviews, 128 children were selected and randomly assigned to either a 10-week school-based psychosocial intervention based on Kendall's Coping Cat protocol (2000a), or to a monitoring group. The intervention was conducted over 10 weekly, 1-to 2-hr sessions at each intervention school. Group sizes ranged from 5 to 12 children. Parental sessions were conducted at the intervention schools in weeks 3, 6, and 9. Anxiety disorder diagnostic status was assessed at posttreatment and at 6-month, 12-month, and 24-month follow-up and yielded interesting results: the CBT and control groups differed significantly with respect to anxiety disorder diagnostic status at 6 months (27% vs. 57%) and at 24 months (20% vs. 39%), but did not differ at 12-month follow-up (37% vs. 42%). Notably, treatment benefits were most evident for those children who initially had moderate to severe clinician ratings of severity, with approximately 50% of these children retaining a clinical diagnosis at the 2-year follow-up, if they did not receive the intervention. For those children who initially showed symptoms of anxiety but who did not have a clinically significant anxiety disorder, there was minimal difference between the preventive intervention and the monitoring-only condition at 24 months follow-up, with 11% showing an anxiety disorder in the intervention group and 16% in the monitoring condition. In other words, children with subclinical anxiety problems did not appear to be at high risk of developing a more severe anxiety disorder if left untreated and benefited only minimally from the intervention.

Selective prevention programs.

Selective prevention intervention programs are delivered to individuals or groups considered to be at high risk for anxiety disorders yet are not evidencing anxiety disorder symptoms. There is a paucity of studies on these programs. To date, most selective prevention intervention programs have targeted individuals or groups exposed to stressful life events such as parental divorce (e.g., Alpert-Gillis, Pedro-Carroll, & Cowen, 1989; Hightower & Braden, 1991; Hodges, 1991; Short, 1998; Zubernis, Cassidy, Gillham, Reivich, & Jaycox, 1999); transition between primary and secondary school, which can be associated with a number of psychological difficulties (e.g., peer relationships, school refusal behavior, substance use; Felner & Adan, 1988); medical and dental procedures (Peterson & Shigetomi, 1981); and having a chronically ill sibling (e.g., Bendor, 1990). Although the findings from these studies generally yield positive effects, their direct relevance to preventing anxiety disorders in adolescents is unclear.

The only selective prevention intervention program designed specifically for anxiety disorders is the Macquarie University Preschool Intervention Program (Rapee, 2002). Children (ages 3.5 to 4.5 years) were recruited mainly via questionnaires distributed to preschools. Inclusion in the study was based on mother-completed ratings on the Australian version of the Childhood Temperament Scale (Sansan, Pedlow, Cann, Prior, & Oberklaid, 1996), followed up by laboratory observation of behavioral inhibition. Behaviorally inhibited children were randomly assigned to either an intervention condition or a monitoring condition. The intervention was conducted with parents only and focused on education about the nature of withdrawal and anxiety, parental anxiety management strategies, information about the importance of modeling competence and promoting independence, development of exposure hierarchies for the children and practice of graded exposure, and discussion of future development. The intervention was conducted in groups of six families and lasted for six sessions. Results at 12 months revealed that mothers in the intervention condition had self-ratings that indicated significantly greater decreases in their child's inhibited temperament as well as in the number of child anxiety diagnoses compared to mothers in the control condition. However, laboratory observations indicated that children in both groups had reduced behavioral inhibition with no significant differences between the two groups.

Universal prevention programs.

As an extension of the indicated prevention work conducted in Australia and reviewed above, encouraging preliminary data have emerged from a school-based FRIENDS program for children ages 10–12 years (Barrett & Turner, 2001) and a second sample of children ages 10–13 years (Lowry-Webster, Barrett, & Dadds, 2001). The FRIENDS acronym stands for Feeling worried; Relax and feel good; Inner thoughts; Explore plans of action; Nice work, reward yourself; Don't forget to practice; and Stay cool. The program is cognitive-behavioral in orientation, can be delivered by teachers or psychologists, is conducted weekly for 10 weeks and followed by booster sessions, and has been compared to an untreated control group. Children who received the intervention had lower self-rated anxiety levels than those of controls at posttreatment; moreover, no statistical differences were found in outcome when FRIENDS was delivered by either teachers or psychologists (Barrett & Turner, 2001). These preliminary reports are encouraging, although the full sample has yet to be accrued, and the follow-up period (6 months) is insufficient to determine whether participation in the program affects anxiety symptoms and the development of anxiety disorders in the long run, especially for those anxiety disorders (e.g., panic disorder) that are more likely to develop in adolescence than in childhood.

Lowry-Webster et al. (2001) randomly assigned 594 children (10 to 13 years) within different schools to receive either the FRIENDS program or assessment only. The intervention was implemented by trained classroom teachers and three separate sessions for parents were also conducted. Pre-to postintervention changes were examined universally and for children who scored above the clinical cutoff for anxiety at pretest. Results revealed that children in the FRIENDS intervention condition reported less anxiety symptoms, regardless of their risk status, relative to the comparison condition. Notably, those who were already in the clinically anxious range on the Spence Children's Anxiety Scale fared better in the FRIENDS program than in the wait-list condition, which, as found by Dadds et al. (1997), suggests again that the FRIENDS program may be a useful intervention for children who are already experiencing significant problems with anxiety. Further research evaluating its efficacy with older adolescent samples is needed, however, before conclusions for its utility with this population can be drawn.

The FRIENDS program was also conducted and evaluated with culturally diverse migrant groups of non-English-speaking background (Yugoslavian, Chinese, and mixed-ethnic; N = 121 in Australia (Barrett, Sonderegger, & Sonderegger, 2001). The sample consisted of 106 primary and 98 high school students and were randomly assigned to either the FRIENDS program or a wait-list control condition (10 weeks wait). Participants in the FRIENDS program exhibited lower anxiety and a more positive future outlook than wait-list participants at posttest. Although the findings suggest the potential of the FRIENDS program in reducing anxiety associated with cultural change, the long-term effects of the program for this purpose have yet to be determined. Also unclear is the extent to which the youths involved in this program were actually suffering from clinically significant anxiety or were undergoing transitory duress due to their being of new immigration status.

As noted by Winett (1998a) and by Donovan and Spence (2000), an infrastructure capable of supporting such large-scale projects must be in place before a universal prevention program can be conducted, and the necessary resources can only be marshaled if anxiety disorders prevention is given a significant level of priority on a societal, school, and community level. It is notable that the only study of this kind specifically targeting reduction of anxiety symptoms was conducted with children rather than with adults, perhaps because schools contain the infrastructure that is needed for the implementation of universal prevention programs. Another reason to focus on children in prevention programs is the belief that early intervention will prevent the vulnerable individual from having experiences that will increase risk for developing an anxiety disorder (e.g., being bullied by peers). One advantage of universal prevention is that it does not label any individuals as being “at risk,” a process that may serve to increase anxiety about anxiety and initiate avoidant coping.

Evaluating and Measuring Success

What Constitutes Success?

As noted above, the small number of intervention studies that have been conducted thus far have focused primarily on the reduction of anxious symptoms as measured by self-report (e.g., Lowry-Webster et al., 2001), although some studies examined whether patients met diagnostic status for an anxiety disorder at both post-intervention and, perhaps more importantly, at follow-up (e.g., Dadds et al., 1997). Both symptoms and diagnostic status are relevant, although the latter is less sensitive because participants can lose the diagnosis at a particular assessment point yet still remain elevated symptomatically and thus presumably also remain at risk for increased symptoms down the road. No studies have examined anxious symptoms and diagnostic status longitudinally through adolescence and into early adulthood. Such studies would yield data that truly test the success of prevention interventions. Yet another kind of successful outcome for universal intervention programs would be destigmatization of anxiety as a character flaw or weakness. Clearly, most youth who participate in universal prevention programs are at low risk for the development of anxiety disorders, yet informing this majority of the nature and impact of anxiety may help reduce problems that often face anxious youth in the social context, such as ostracism, teasing, and peer rejection.

What Are the Active Ingredients?

The few prevention intervention studies that have been conducted thus far have not shed light on the mechanisms involved because most of the designs used have compared active treatment packages to repeated assessment only. The superiority of the CBT packages examined thus far could therefore be attributable to a wide variety of nonspecific factors, such as treatment credibility and therapist contact. Dismantling studies typically follow the establishment of ef ficacy (e.g., Schmidt et al., 2000), and thus the field may be a long time from discovering the impact of specific treatment interventions and their underlying mechanisms.

What Outcomes Are Targeted?

As noted above, in the small number of studies conducted to date, the outcomes targeted have focused primarily on anxiety symptoms and disorders. It might be worthwhile for future research to move beyond symptoms and diagnosis and pay increased attention to whether functional impairment has improvement. For example, are there improvements in the adolescent's grades or in his or her peer relationships? These are the outcomes that would seem to matter most and should be seriously considered in the design and evaluation of future prevention studies.

In addition, the potential of “positive psychology” has yet to be seriously considered in the context of preventing anxiety disorders in adolescents and targeting outcomes. Positive psychology is devoted to creating a science of human strengths that act as buffers against mental illness, including anxiety (Seligman, 2002). Dick-Niederhauser and Silverman (2003) have adapted positive psychology principles and have suggested their utility in serving as outcome targets for anxiety prevention studies. Thus, potential outcome targets might include the instilling of hope and the active pursuit of goals in young people, which in turn have been linked to the development of courage. Courage in turn has been linked with increased optimistic cognitive processing, a sense of self-efficacy, and skillful coping. Although measures exist to assess some positive psychological concepts, further instrument development and evaluation is needed for positive psychology principles to be fully implemented and studied in the context of anxiety prevention research.

Conceptual, Methodological, and Practical Issues

Methodological and conceptual issues vary across types of prevention program. That is, the methodological concerns arising in universal prevention that target the broad population of adolescents are different from those in selected and indicated prevention intervention. The former requires more streamlined assessments that would increase participation and compliance (thus ensuring sample representativeness) and reduce cost (thus assuring feasibility). Accordingly, a major issue in universal prevention program research is finding the best ways to prompt adolescents to participate in a study that addresses a problem that they probably do not have. Universal prevention programs are especially likely to be conducted in conjunction with school administrators, thus capitalizing on the schools' past successes in encouraging student participation will be important. As noted earlier, a brief survey conducted via a Web site might capture the interest of teens in particular, thus computer technologies may prove essential in this kind of work. The costs of universal programs may ultimately require a political commitment on the level of state or federal government. For the selective and indicated programs, the primary methodological concern is how to encourage participation while at the same time protecting student confidentiality. This may be especially important if the intervention itself is conducted at school and during school hours, when absence might be conspicuous; negative social costs both real and imagined may impact participation. Moreover, if the intervention is conducted in groups, confidentiality among group members needs to be considered. Students who have been identified for intervention participation because of having experienced a trauma or for being excessively shy might be reluctant to share their experiences if they do not have assurance that what is discussed in session will not be discussed outside with nonmembers. The provision of sufficient time to foster group cohesion to alleviate this concern would therefore be important in any selective and indicated prevention effort that involves discussion of personal matters in a group setting.

Another issue that warrants consideration is when to intervene. As discussed above in relation to trauma exposure, immediate intervention provided to all individuals exposed to the trauma has not been found helpful with adults and thus should probably be avoided when conducting interventions with youth who have been exposed to a traumatic event, such as shootings at the school.

Yet another issue is the match between the type of intervention and the developmental stage of the individuals. Perhaps group interventions can be particularly successful in adolescence when the value of the peer group is quite powerful and, if properly harnessed, may enhance the efficacy of the intervention. Interventions with younger children, by contrast, need to be targeted at parents, particularly interventions that target the parental risk factors thought to be associated with increased risk for anxiety disorders.

The choice of place for conducting prevention interventions is another important consideration. Often the school would be the most practical place and would allow for the accrual of the kinds of large sample sizes required to detect effects for intervention programs. The pediatrician's office can also be a useful place to conduct certain indicated prevention program interventions, such as providing psychoeducation about anxiety and anxiety disorders to youngsters who present frequently for treatment of gastrointestinal problems. Many of these youngsters may have subclinical anxiety symptoms and thus they may constitute an appropriate population for psychopathology and intervention research. Swinson et al.'s (1992) hybrid treatment/indicated prevention study on the use of psychoeducation for adults who presented to emergency rooms with panic symptoms serves as a model for this sort of study. The provision of prevention intervention in the medical context raises questions of feasibility of delivery: managed care has minimized the amount of time available in a visit to discuss seemingly peripheral issues such as anxiety symptoms. Thus the development of brochures, self-help programs, or interventions that can be delivered by support staff should be considered.

Prevention research by its nature requires longitudinal follow-up. One major issue is how best to retain participation in the study and how to guard against attrition over time. Here again informed consent from the student and family and active collaboration with the school will be helpful, but it is important to keep in mind that the most valuable assessment points for prevention programs take place years after the intervention is delivered. Thus it is imperative that studies be funded in a manner that will ensure the collection of data well into the future; inadequate participation in follow-up for these kinds of studies imperils the entire enterprise, as detection of sampling bias (e.g., better follow-up with less impaired participants or vice versa) threatens to compromise conclusions that could be drawn about the efficacy of intervention. Treatment studies have had to address this problem and have requested that the family provide the names of family and friends who will know how to contact them in the future if they move, Social Security numbers, and other such information to facilitate participation in long-term follow-ups.

A final issue that affects all prevention programs involves the ongoing assessment of risk, and responsibility for risk. Those at risk for anxiety may also be at increased risk for other psychiatric comorbidity, thus procedures must be enacted within prevention intervention programs to manage clinical emergencies. Moreover, the role of the parent in these programs must be considered: if the child or adolescent is found to be at increased risk for anxiety disorders upon screening and is then eligible to participate in the program, how much or how little the parent should be involved or have access to the information discussed in assessment and/or treatment needs to be specified up front, as it will certainly affect both entry and active participation.

Problem of Sustainability (Boosters, Ongoing Programs, Training)

Little is known about the long-term effects of prevention intervention methods for anxiety disorders in youth, as the longest follow-up period reported on thus far is 2 years (Dadds et al., 1997). The studies that have included follow-up have generally suggested maintenance of the in tervention effects, but here again these studies have focused on young or very young children and thus cannot inform the field about retention of benefits into and through adolescence and adulthood. A related question is whether booster sessions are needed to retain the gains from preventions programs, since the fairly predictable stressful life events that face young children growing into adolescence might compromise long-term maintenance. For example, studies discussed earlier suggest that young children with elevated but subclinical anxiety disorders may benefit most from prevention programs; transition from middle to high school may threaten these gains, and thus it may be reasonable to reinstitute the intervention during this transition. It is unknown whether this is the case, but the relation between loss of gains and stressful life events constitutes an especially important area for future study.

Problem of Contagion

There is no evidence of these undesirable effects from the studies of group treatment of youth that have been conducted thus far (e.g., Kendall et al., 1997), but the possibility for such effects remains. Although it is possible that discussion of anxiety themes may activate new fears in those who are already vulnerable, especially if the interventions involve group discussions, there is no evidence of these effects from group treatment studies (e.g., Kendall et al., 1997; Silverman et al., 1999a), including in groups in which the patients involved were very heterogeneous with respect to both age (i.e., child and adolescent patients) and primary anxiety diagnosis (e.g., OCD, specific phobia) and other clinical features (e.g., presence or absence of school refusal behavior; Lumpkin, Silverman, Weems, Markham, & Kurtines, 2002). However, the example of Critical Incident Stress Debriefing (CISD) suggests that the long-term recovery of certain adults who have experienced a trauma and have attended group meetings may be impeded by participating, and possibly the mechanism by which this effect is realized involves exposure to other participants' narratives of the traumatic event (e.g., Mayou, Bryant, & Ehlers, 2001). Provided that secondary gains (e.g., missing trigonometry class) for attending prevention intervention sessions are minimized, there is little reason for concern that students without anxious symptoms or risk factors would feign such problems.

Age-Appropriate Interventions (Developmental Approach)

The prevention intervention programs that have been evaluated thus far in research (e.g., Lowry-Webster et al., 2001) were designed for children rather than for adolescents. Adaptations to accommodate the developmental needs of adolescents should be made with specific attention to possible age-related increases in physiological functioning, emotional vulnerability, social and peer pressures, and comorbid conditions, as well as any other changes that these youngsters may be experiencing. In particular, prevention programs must consider the importance of the peer group within the intervention program itself, but also with an eye towards the social implications of participating in the program among nonparticipating students, especially if the program is either selective or indicated. Insufficient attention to these factors may reduce the number of teens willing to enter the program altogether, and can limit active participation within the program itself if a group format is implemented. One way to address this potentially important concern is to incorporate program graduates who may serve as role models for new participants, as a way to alleviate concerns that program participants may not be perceived as “cool” among the larger student population. Another way to make participation more palatable is to present information about adult role models who have struggled with anxiety and have openly discussed their difficulties, such as television host Mark Summers. However this is accomplished within a protocol, it is imperative that the culture of adolescence and the importance of the peer group be taken into consideration when developing appropriate interventions for teens.

Ethical Issues

A Note of Caution from Adult Early-Intervention Research

Longitudinal studies of trauma survivors (e.g., Riggs, Rothbaum, & Foa, 1995; Rothbaum et al., 1992) indicate that most individuals experience elevated levels of PTSD symptoms shortly after the traumatic event. In addition, elevated levels of depression and general anxiety often accompany PTSD symptoms. For most trauma survivors, however, these symptoms decline significantly over the ensuing 3 months without any professional intervention. That said, a significant minority of trauma survivors continues to experience high levels of posttrauma distress that, without professional treatment, may persist for months or years (Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995).

As discussed above, it is now well established that various forms of CBT are effective in reducing PTSD symptom severity as well as associated anxiety and depression (e.g., Foa et al., 1999). While there are effective treatments for individuals suffering from chronic PTSD, many sufferers either do not seek treatment for their trauma-related symptoms or do not have access to treatment. As a consequence, individuals' suffering and their inability to function can be prolonged. They are also vulnerable to associated comorbidity such as substance abuse. Such considerations have prompted trauma therapists to develop brief interventions applied shortly after the traumatic event to facilitate recovery and thereby prevent the development of chronic PTSD. Two approaches to facilitating recovery following a traumatic event have been researched. Abbreviated CBT packages such as those developed by Foa et al. (1995) and adopted by Bryant and colleagues (1998, 1999) have been found to be efficacious in accelerating recovery and reducing the likelihood of chronic PTSD. The other approach involves psychological debriefing (PD). Such programs typically comprise one session and are applied shortly after a traumatic event (frequently within 48–72 hr). In this session (which can be conducted in groups or individually), participants are encouraged to describe the traumatic event, including their thoughts, impressions, and emotional reactions. The session also includes normalization of the trauma survivors' reactions and planning for coping with the trauma and its sequelae. Results of RCTs for PD are somewhat mixed, but an important pattern is emerging. In general, participants in PD studies subjectively find the intervention to be helpful (i.e., high consumer satisfaction), yet objective measures of specific posttrauma symptoms typically yield no differences between those receiving PD and those who do not. Thus, the improvement typically observed following PD is better attributed to natural recovery, rather than to an active ingredient of the intervention (e.g., Bisson, 2003; Rose, Brewin, Andrews, & Kirk, 1999). Moreover, a few studies have found PD actually interfered with natural recovery (e.g., Mayou et al., 2000). The results of PD studies highlight the need for caution in using one-session interventions conducted shortly after traumatic events involving children and adolescents.


As noted earlier, unlike universal prevention programs, selective and indicated prevention programs specifically select participants on the basis of elevations of anxious symptoms or putative anxiety disorder risk factors. In the school context, where most prevention interventions are likely to take place, the latter program types require identification of a subgroup of participants from among the broader population who will be either encouraged or required to participate. The potential negative implications of this strategy have already been considered in the academic context with respect to educational issues, and have led to the gradual reduction of labeling for academic tracking systems (e.g., honors, regents, and basic classes) and to increased mainstreaming of special education students. Similar problems may be encountered in identifying already anxious or anxiety-vulnerable students for special attention or services. As discussed above, adolescence is a stage in life when similarity with the relevant peer group is valued, and intervention efforts that do not deal sensitively with this issue may be poorly attended or, worse yet, yield unintended negative consequences. Little has been written about this issue in the context of anxiety prevention programs implemented thus far, but methods to prevent such unintended consequences should be carefully considered.

Postvention, Follow-up, Dissemination (Monitoring Dissemination)

The studies conducted thus far have involved acute treatment and, at least in some studies, follow-up assessment only. It is unknown how to encourage ongoing use of skills learned in the prevention programs, nor is it known how best to encourage participation in follow-up assessments. Because the primary dependent variable of interest in prevention programs must be measured years later than the intervention was conducted, it is imperative to develop methods that encourage cooperation with long-term follow-up. Because most prevention interventions will likely be conducted in the school context, active collaboration with school administration will be critical to promote collection of these data. Families may also be able to facilitate participation, thus direct contact with families may be advisable. However, this raises issues with respect to confidentiality and the need to discuss up front with the young participant what will and will not be shared with parents and/or guardians.

The preliminary success of the FRIENDS program in the hands of teachers bodes well for transportability of this program to treatment providers other than mental health professionals with expertise in CBT. Clearly the implementation of CBT-oriented prevention programs cannot realistically be limited to Ph.D.-level psychologists, and a multidisciplinary approach may be the best way to proceed. This raises interesting questions about the best way of disseminating CBT and the degree of expert supervision needed in the short and long run to optimize treatment delivery; these questions touch on cost-effectiveness of prevention programs. Research on the treatment of adult PTSD suggests that masters-level counselors can be trained to successfully deliver prolonged exposure for women who have been sexually assaulted; indeed, these counselors were as effective in delivering prolonged exposure as were CBT experts (Foa et al., 2002). However, the counselors received weekly supervision from CBT experts, which increased the cost of the program. The next research question being currently examined is whether reduction in the amount of expert supervision for the counselors will reduce their success. A similar line of research needs to be pursued in adolescent anxiety disorder prevention, since the broad application of such interventions appears to be dependent on successful training of school personnel to implement these programs in the school context.

Impediments to Prevention

The first set of impediments to developing successful prevention intervention is the lack of knowledge about the complex interrelations among the various risk and protective factors for the development of anxiety disorders. Much is known about some specific factors but little is known about how they interact, which leaves the field bereft of a strong theoretical foundation upon which to build prevention programs. This may be the reason why prevention research has languished relative to treatment research: the factors associated with etiology may not be the same as those associated with maintenance, and thus comprehensive knowledge about the latter will allow for the development of treatment interventions even in the relative absence of the former.

Practical considerations have stunted the development of prevention programs as well. Prevention efforts are likely to be costly, as they necessarily involve collection of data from large samples over a long period of time. Large samples are needed because of the relatively low base rates of anxiety disorder in the population of interest, and because there is insufficient information about who will actually go on to develop an anxiety disorder. Consequently, it is important to conduct broad screens to obtain sufficient numbers of vulnerable children and adolescents for inclusion in the studies. For example, most behaviorally inhibited infants do not develop an anxiety disorder later in life, and thus a large sample of inhibited children would be needed to detect the efficacy of a prevention program targeting behavioral inhibition. Further, because the relevant outcome is the future development of anxiety disorders and perhaps of subsequent comorbid conditions (e.g., depression, substance abuse), prevention studies require data collection for years after the intervention to determine its ultimate impact. The need to conduct longitudinal follow-ups of these large numbers for long periods of time renders the study of prevention programs impractical, especially when the primary funding sources for anxiety disorders research (e.g., National Institute of Mental Health) typically favor shorter studies with more tangible impact. Thus, new sources of funding must be identified to generate knowledge that will inform the development of anxiety disorder prevention programs. Given the potential for anxiety disorders to derail adolescent development and thereby result in substantial personal and economic impact, these programs should be a major priority for a society that promises to leave no child behind.