Show Summary Details
Page of

Defining Anxiety Disorders 

Defining Anxiety Disorders
Defining Anxiety Disorders
Page of

PRINTED FROM OXFORD MEDICINE ONLINE ( © Oxford University Press, 2022. 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: 01 July 2022

Before discussing the anxiety disorders, it is important to consider the concept of anxiety and its heterogeneity. Anxiety refers to multiple mental and physiological phenomena, including a person's conscious state of worry over a future unwanted event, or fear of an actual situation. Anxiety and fear are closely related. Some scholars view anxiety as a uniquely human emotion and fear as common to nonhuman species. Another distinction often made between fear and anxiety is that fear is an adaptive response to realistic threat, whereas anxiety is a diffuse emotion, sometimes an unreasonable or excessive reaction to current or future perceived threat.

Distinguishing Anxiety from Anxiety Disorders

Defining the boundaries between extremes of normal behavior and psychopathology is a dilemma that pervades all psychiatry. For some very extreme conditions, such as Downs syndrome, diagnostic decisions are straightforward. Milder forms, by contrast, present problems when one attempts to define the point at which “caseness” begins. A few symptoms escape this definitional conundrum by virtue of their being deviant, regardless of their severity. This applies to symptoms such as delusional beliefs or hallucinations. In the case of anxiety, however, it is especially problematic to establish the limits between normal behavior and pathology because when mild, anxiety plays an adaptive role in human development, signaling that self-protective action is required to ensure safety. Because anxiety can be rated on a continuum, some inves-tigators suggest that extreme anxiety represents only a severe expression of the trait, rather than a distinct or pathological state. Distributions may consist of distinct entities, however. For example, some cases of mental retardation, as caused by neurological injury, represent a quantal departure from factors influencing normal variations in intelligence. By analogy, the fact that anxiety falls on a continuum of severity does not preclude the presence of qualitatively distinct disorders at any point in the distribution (Klein & Pine, 2001).

Anxiety may become symptomatic at any age when it prevents or limits developmentally appropriate adaptive behavior (Klein & Pine, 2001). However, anxiety about particular circumstances may develop at one or another developmental stage, based on the typical age-related experiences that occur during this stage. For example, anxiety about separation represents a normal aspect of development that is experienced by many young children. Similarly, in adolescence, particular questions arise concerning anxiety about social situations, given changes in the social milieu that are experienced as stressful by many adolescents. A useful rule of thumb for determining the diagnostic threshold is the person's ability to recover from anxiety and to remain anxiety-free when the provoking situation is absent. For example, it is not necessarily deviant for adolescents to respond with acute discomfort or anxiety when meeting a peer that they find attractive. Such reactions reach clinical levels, however, when adolescents are unable to recover from the anxiety (as manifested by recurrent doubts or ruminations about how they behave), or when adolescents avoid such encounters on a consistent basis. Similarly, clinical anxiety in this situation might be characterized by the development of concerns about future meetings with unfamiliar peers or even avoidance of activities that might require peer interactions. Therefore, an adolescent's lack of flexibility in affective adaptation is an important pathological indicator. In addition, the degree of distress and dysfunction influences diagnostic decisions; these vary with developmental stage, as well as with cultural and familial standards. When anxiety symptoms are developmentally inappropriate, subjective distress is relatively more informative. For example, separation anxiety is developmentally more congruent with early childhood than with adolescence. In brief, three clinical features impinge on the definition of pathological anxiety. Two of these, distress and dysfunction, vary in importance as a function of developmental stage. The third, symptomatic inflexibility, is diagnostically relevant regardless of age.

The ability to draw firm conclusions on the ideal criteria for disorders will remain limited so long as signs and symptoms are the exclusive basis for establishing the presence of psychiatric disorders. Longitudinal research can provide some answers by identifying specific symptom patterns and thresholds that have long-term significance. In practice, however, such evidence has proved to be informative but rarely conclusive.

The past two decades have witnessed a great expansion in the study of anxiety disorders. An earlier emphasis on rating scales or interviews assessing multitudes of unrelated fears and worries has been replaced by an emphasis on the study of diagnostic groups that reflect explicit clinical criteria. Scale ratings can be grouped to generate overall scores of anxiety, or what has come to be called “internalizing” symptoms, such as in the widely used Child Behavior Checklist (CBCL; Achenbach, 1991), but as the evidence shows, scale ratings correspond poorly to clinical entities.

Difficulties in separating “normal” from “pathological” anxiety are clearly reflected in results from epidemiological studies, in which the prevalence of anxiety disorders changes markedly with relatively minor changes in the definition of impairment (reviewed Klein & Pine, 2001). However, adolescents with anxiety disorders who seek treatment typically suffer from markedly impairing anxiety, and there is little ambiguity about determining whether they have “normal” or abnormal levels of anxiety.

This challenge poses both practical and conceptual problems. The practical problem concerns the timing of treatment. Two mistakes are possible. An adolescent who needs treatment may fail to receive it if the threshold for diagnosing the disorder is set too high (“a false negative”). An adolescent whose anxiety reflects a reasonable response to adverse circumstances may receive unnecessary treatment (“a false positive”). The decision to treat versus not treat is linked to costs and benefits that inform decisions about each adolescent.

The conceptual problem concerns the need to provide a principled basis for distinguishing disorder from nondisorder beyond the current imperfect clinically based principles. In an ideal circumstance, these principles would be based on understandings of pathophysiology. Consistent with this perspective, some philosophers of medicine have attempted to provide objective, biological criteria for demarcating disorder (e.g., major depression) from distressing states that fall within the bounds of normal unhappiness (e.g., grief). Others have claimed that all ascriptions of disorder reflect nothing more than societally determined value judgments about undesirable states and behavior.

Merging these polarized views, Wakefield (1992) proposed a harmful dysfunction account of disorder, holding the position that disorder is a hybrid concept comprising a factual component and an evaluative component. The factual component specifies what is dysfunctional—a derangement in a psychobiological function—and the value component specifies the resultant harm—usally emotional suffering, social maladjustment, or both. Therefore, ascription of disorder requires that two interrelated criteria be met: a psychobiological mechanism is malfunc-tioning, and this underlying dysfunction results in suffering, maladaptation, or both.

Wakefield's criteria imply that a person may be characterized by internal dysfunction but not qualify as having a disorder because no resultant harm occurs. For example, some youngsters characterized by extreme shyness or behavioral inhibition may find niches for themselves that enable them to adapt without marked distress. Even though the dysfunction requirement is met, these children would not be considered disordered because their dysfunction does not result in suffering or maladaptation. Conversely, some youngsters, bullied by larger children for example, may experience chronic anxiety at school. But because their suffering does not arise from dysfunction in the psychobiological mechanisms for estimating threat, Wakefield's criteria would suggest that they not be diagnosed as disordered. Mechanisms for detecting threat work precisely as they are “designed” to work: the bullied youngsters experience chronic anxiety because they are continually exposed to threat, not because they have a mental disorder. It is important to note that this is only one definition of a “mental disorder.”

Wakefield's (1992) framework is not without its limitations. Attempts to elucidate a value-free perspective on function—especially when cast within an evolutionary framework (McNally, 2001a)—raises yet another set of thorny prob lems. Nevertheless, the harmful dysfunction provides a useful model for posing questions regarding the distinction between normal psychological distress and its pathological variants.


In the following sections we will describe each of the anxiety disorders listed in the Diagnostic and Statistical Manual of Mental Disorders, 4th ed., (DSM-IV; American Psychiatric Association, 1994). For an accompanying list of signs and symptoms see Table 9.1.

Table 9.1 Signs and Symptoms of Adolescent Anxiety Disorders


Key Diagnostic Feature

Other Criteria for the Disorder

Other Relevant Clinical Signs and Symptoms

Panic disorder

The occurrence of spontaneous panic attacks. These are paroxysms of fear or anxiety associated with somatic symptoms, such as palpitations or shortness of breath.

Panic attacks must be recurrent and must be associated with either concern about additional attacks, worry about the implication of the attacks, or changes in behavior. Panic disorder is frequently associated with agoraphobia, or anxiety about being in places where escape might be difficult.

Spontaneous panic attacks are very rare before puberty. Typical developmental course for progressive forms of the disorder involves initial development of isolated spontaneous panic attacks around puberty, followed by recurrent panic attacks, and then agoraphobia in adulthood. This process can take years to unfold across maturation from adolescence to adulthood.

Social anxiety disorder

The occurrence of extreme fear in social situations in which an individual is exposed to unfamiliar people

Exposure to social situations provokes anxiety that is associated with severe distress or impairment. The individual must show the capacity for age-appropriate social relationships.

This condition typically develops in late childhood or early adolescence. The disorder is associated with shyness or other subclinical behavioral features, such as certain temperamental types.

Separation anxiety disorder

Extreme anxiety about being separated from home or from an individual to whom a child is attached

The anxiety is associated with either distress upon separation, worry about harm to an attachment figure, avoidance of situations requiring separation, or physical complaints when separation is anticipated.

This condition is among the most prevalent mental disorders in children. The condition typically develops in early childhood, showing high rates of remission between childhood and adolescence. Some data suggest a familial or longitudinal association with panic disorder in adults.

Obsessive compulsive disorder (OCD)

Recurrent, persistent, intrusive, anxiety-provoking thoughts (obsessions) or repetitive acts (compulsions) that a person feels driven to perform

This pattern of intrusive thoughts or acts is recognized as unreasonable and consumes an hour of each day. These symptoms produce significant interference or distress.

This condition typically presents with stereotyped thoughts or acts. These might include concerns that the individual is in some way dirty or that the person has sinned. The disorder is frequently associated with tics and attention deficit disorder during childhood.

Posttraumatic stress disorder (PTSD)

Following exposure to trauma, in the form of a frightening event, an individual develops recurrent reexperiencing of the event, attempts to avoid stimuli associated with the event, and develops signs of increased arousal.

Reexperiencing can involve flashbacks, nightmares, or images. Avoidance can involve changes in behavior, changes in cognition, or new-onset feelings of detachment. Increased arousal can involve insomnia, exaggerated startle, or irritability.

PTSD is associated with many comorbid disorders, including major depression, other anxiety disorders, and behavior disorders. Different types of traumas may involve different symptomatic manifestations. For example, symptoms may differ in acute vs. chronic trauma.

Generalized anxiety disorder (GAD)

A pattern of excessive worry on most days for a period of 6 months. This worry is difficult to control.

Worry is characterized as apprehension when anticipating an upcoming feared event. Worry is associated with restlessness, fatigue, reduced concentration, or difficulty falling asleep.

GAD shows very high rates of comorbidity with a range of conditions, particularly anxiety disorders. In clinical settings, GAD virtually never presents as an isolated condition but is complicated by another comorbid disorder. Beyond the relationship with other anxiety disorders, GAD shows an unusually strong association with major depression.

Specific phobia

Marked and excessive fear of a specific object, such that exposure to the object precipitates extreme anxiety

The fear either causes avoidance that interferes with functioning or produces marked distress in the individual.

Specific phobia generally produces lower levels of impairment than other anxiety disorders. Phobias can be divided according to the nature of the feared object into various types, including animal type, natural-environment type, blood-injury type, or situational type.

Specific Phobia

A specific phobia is an extreme or unreasonable feeling of fear or anxiety linked to a specific animal, object, activity, or situation. Individuals usually experience anxiety when they encounter or think about the feared target. The diagnosis of phobic disorder requires that the person experience extreme distress and impairment in normal functioning. Although phobic disorders can begin at an early age, they often occur in childhood. Although there are many possible phobic targets, most children fear the same limited range of objects or events. Although most phobias do not have a distinct biological profile, blood phobia is an exception for it is accompanied by a sudden drop of blood pressure and heart rate and often fainting. This profile is not characteristic of phobic reactions to most other feared targets.

Separation Anxiety Disorder

Onset of separation anxiety disorder, defined by unrealistic worry accompanying separation from home or caretaker that interferes with appropriate behavior, usually occurs in late childhood but before adolescence. Indeed, this is the only disorder in which onset must occur before the person is 18 years old. Because separation anxiety disorder is accompanied by a reluctance to engage in activities that require separation from a caretaker, it can take the form of fear of school attendance. Although some adolescents develop separation anxiety disorder, refusal of school attendance among adolescents can occur because of social anxiety, rather than anxiety over separation.

Social Anxiety Disorder

The central clinical feature in social anxiety disorder, also called social phobia, is extreme worry over ridicule, humiliation, or embarrassment in a social situation that is not the result of a serious cognitive or physical impairment in the ability to interact with others. Havelock Ellis, writing a century ago, called such patients “modest” and regarded their fear as instinctual and possessing a sexual component (Ellis, 1899). Although these children avoid social situations, some are unable to articulate these concerns and simply feel uncomfortable in unfamiliar social settings. The diagnosis of social anxiety disorder requires that the child or adolescent experience distress with peers. There are two categories of social anxiety disorder. One type involves a restricted range of fears, limited to performance situations. A second category, called the generalized subtype, is applied when the anxiety is evoked in a majority of social settings. This definition typically encompasses situations feared by individuals with performance-limited social anxiety. This generalized form usually has an earlier onset, lasts for a longer time, and is often comorbid with other symptoms.

There is variability in the application of nosological definitions of social anxiety disorder in clinical settings. Some clinicians apply it to individuals who are anxious in one or two situations. Others apply the same diagnosis to patients who experience performance or test anxiety (worry over public speaking or examinations), or who feel anxiety when eating with strangers but not when interacting with others.

Obsessive-Compulsive Disorder

The diagnosis of obsessive-compulsive disorder (OCD) is made when the individual has recur rent obsessive thoughts or repetitive behaviors called compulsions that the individual feels he or she must perform. Young children are more often characterized by compulsions than by obsessions. Some children spend considerable time involved in elaborate rituals surrounding cleaning or checking routines, providing vague justifications for these behaviors. Adolescents are both more willing and better able to describe the obsessive thoughts that accompany their compulsions. Patients with OCD sometimes exhibit tics and attention deficit hyperactivity disorder (ADHD). Although OCD had been considered a rare condition, recent research suggests that as many as 1% of children exhibit this disorder.

Generalized Anxiety Disorder

A diagnosis of generalized anxiety disorder (GAD) is given to adolescents who worry about a variety of events or life circumstances—usually schoolwork, appearance, money, or their future. Age of onset of GAD is usually later than for most other disorders, although many patients report having been anxious for many years. Further, GAD is likely to be comorbid with other symptoms, but the primary symptom is a chronic state of worry, rather than chronically avoidant behavior.

Panic Disorder

The essential feature of panic disorder is the repeated experience of intense fear of impending doom or danger following the unprovoked experience of bodily symptoms, especially rapid heart rate, shortness of breath, choking sensations, and sweating, or a feeling of depersonalization. The onset of this disorder is usually in late adolescence or early adulthood. Some patients with panic attacks will develop a fear of leaving home to avoid a panic attack; these patients are called agoraphobic. Panic disorder usually begins with sporadic isolated episodes of anxiety that accompany a panic attack and becomes a full-blown panic disorder between adolescence and adulthood (Pine, Cohen, Gurlet, Brook, & Ma, 1998).

Although young children can have an occasional panic reaction, it is unclear whether these attacks are accompanied by thoughts of impending danger. Moreover, it is extremely rare for children to experience panic reactions spontaneously in the absence of a trigger. Spontaneous panic attacks are the essential element of the diagnosis. Hence, there is some controversy over the degree to which panic disorder occurs in preadolescents. If this disorder does occur in children, it is relatively infrequent. Some investigators believe that the essential missing component in early childhood is the unprovoked change in bodily sensations, rather than the ability to impose a catastrophic interpretation.

Posttraumatic Stress Disorder

Unlike the other anxiety disorders, a diagnosis of posttraumatic stress disorder (PTSD) rests on a clearer causal sequence in which a person is first exposed to a traumatic event, feels frightened because of the threat to personal integrity, and then develops the disorder. Patients with PTSD present three kinds of symptoms. First, the patient must suffer from episodes of reexperiencing the traumatic event, which can include recollections of the event manifested in flashbacks or recurring dreams. Second, patients must attempt to avoid any event or place associated with the trauma, and the avoidance is accompanied by feelings of numbness or reduced responsivity. Finally, PTSD patients must experience signs of increased physiological arousal, especially difficulty falling asleep, increased irritability, or exaggerated startle. As with panic disorder, there is some debate over whether children exposed to trauma manifest all three symptoms. Children are likely to display separation anxiety disorder following a traumatic event. Nonetheless, the diagnostic criteria for PTSD in the DSM-IV are the same for children and adults.


Accurate estimates of the prevalence of DSM anxiety disorders are only available for children older than 8 years of age because studies of younger children have failed to have population-based samples and clear diagnostic criteria (these studies often use symptom scales rather than diagnostic indices) or have failed both requirements. The best estimates of the prevalence of anxiety disorders in preschool children, based on a primary care clinic sample (Lavigne et al., 1996, 1998, 2001), were very low. The following sections summarize prevalence data based on samples from the general population and studies published over the past decade (Costello, Egger, & Angold, 2004). The prevalence of any anxiety disorder increases with the duration of time over which the symptom's presence is counted. Thus 3-month estimates range from 2.2% to 8.6%, 6-month estimates from 5.5% to 17.7%, 12-month estimates from 8.6% to 20.9%, and lifetime estimates from 8.3% to 27%.

Specific Anxiety Disorders

The most prevalent diagnoses are DSM-III-R overanxious disorder (OAD) (0.5% to 7.1% with a median of 3.6%), DSM-IV specific phobias (0.1% to 12.2% with a median of 3.5%), and social phobia (0.3% to 15.1% with a median of 3.5%). Panic disorder (0.1% to 3.1%, median of 1.1%) and OCD (0.1% to 7.1%, median of 0.6%) are far less common. The use of adult diagnostic instruments such as the Disorders Interview Schedule (DIS) produce higher rates of specific and social phobias, agoraphobia, and OCD. By contrast, instruments designed specifically for children yield lower prevalence rates. Thus caution should be exercised when adult diagnostic instruments are used with children.

Anxiety and Disability

The relation between the diagnosis and everyday functioning remains a focus of controversy. Health maintenance organizations, insurance companies, and governmental agencies are concerned with whether children diagnosed with anxiety disorder require treatment (Costello, Burns, Angold, & Leaf, 1993). One perspective requires that a child show significant impairment or disability to receive a diagnosis, in which disability can refer to a particular symptom or the entire syndrome. Clinicians could rate a child's psychological functioning but fail to make a clinical diagnosis (Hodges, Doucette-Gates, & Liao, 1999; Shaffer et al., 1983).

The prevalence of anxiety disorders varies according to whether disability is or is not part of the definition. If a child must meet the criteria for a diagnostically relevant symptom as well as impairment in everyday functioning, the prevalence of a diagnosis is reduced by two-thirds. Further, requiring both specific impairment as well as severe scores on the Children's Global Assessment Scale (CGAS; Shaffer, Fisher, Dulcan, & Davies, 1996) reduced the prevalence of disorder by almost 90%. The prevalence of simple phobia was affected most severely; the prevalence fell from 21.6% if no impairment was required to an estimate of only 0.7% when impairment in daily functioning was required. Thus, all estimates of the frequency of the anxiety disorders depend in a serious way on the source of evidence and the criteria adopted. There is no “correct” prevalence in the sense that there is a correct height, in meters, for the Empire State Building.

Gender and Age Differences in Prevalence

Most investigators report that girls are more likely than boys to have an anxiety disorder. For example, more girls than boys between ages 9 and 16 years participating in the Great Smoky Mountains Study had an anxiety disorder (12.1% vs. 7.7%; Costello et al., 2004). Three studies revealed more phobias in girls, two reported more panic disorder and agoraphobia in girls, and only one study found more separation anxiety disorder and OAD in girls than in boys. In one of the few studies that examined the potential confounding factors linked to gender, the excess of anxiety disorder in girls was not eliminated after controlling for 15 possible confounding factors (Lewinsohn, Gotlib, Lewinsohn, Seeley, & Allen, 1998). One confound was the fact that the child's age is often correlated with the time frame used to estimate symptoms. Investigators who used 3-month prevalence rates reported the lowest prevalence but studied the youngest subjects. By contrast, investigators using 12-month estimates had the highest prevalence but worked with the oldest children. In the Great Smoky Mountains Study the prevalence of separation anxiety decreased with age, whereas social phobia, agoraphobia, and panic disorder increased with age. It is difficult to draw conclusions about gender differences in the fears, worries, and anxieties of clinic-referred samples (either clinically anxious samples or other types of clinical samples) given the limited amount of research that has been conducted. Further research in this area is of critical importance (see Silverman & Carter, in press).


On any one day, between 3% and 5% of children and adolescents suffer from an anxiety disorder. Rates of GAD and specific phobias remain constant across childhood and adolescence. Although girls are more likely than boys to have an anxiety disorder, the gender difference is not as marked in the general population as it is in clinical samples, perhaps because boys are less likely to be referred for treatment. Numerous studies have demonstrated that girls who mature earlier than their peers exhibit higher rates of anxiety symptoms and disorders (e.g., Caspi & Moffitt, 1991); such findings have not been obtained with boys.


Comorbidity among the anxiety disorders has been a problem for nosology, epidemiology, diagnosis, and treatment. The high level of comorbidity in clinical samples is mirrored in community samples (Brady & Kendall, 1992; Kendall & Clarkin, 1992; Kendall, Kortlander, Chansky, & Brady, 1992; Table 9.2). A review by Costello et al. (2004) yielded equivocal results because not all diagnoses were present in every study, there was a lack of consensus regarding controls for comorbidity, and concurrent comorbidity and sequential comorbidity were not always distinguished.

Table 9.2 Summary of Comorbidity from Pediatric Samples

Anxiety Disorder

Community Samples

Clinical Samples

Social anxiety disorder, selective mutism

Specific phobias, separation anxiety disorder

Other anxiety disorders, major depression, substance abuse

Generalized anxiety disorder (formerly overanxious disorder)

Depression, possibly alcohol and other substance abuse

Separation anxiety disorder, specific phobia, social anxiety disorder

Separation anxiety disorder

OAD/GAD, speciflc phobia, social anxiety disorder, possibly subsequent panic disorder

OAD/GAD, specific phobia, social anxiety disorder

Specific phobias

Separation anxiety disorder, social anxiety disorder

Separation anxiety disorder, social anxiety disorder

Panic disorder

Possibly social anxiety disorder, specific phobia


Obsessive compulsive disorder

Depression, other anxiety disorders, tic disorders

Posttraumatic stress disorder

Depression, other anxiety disorders, externalizing disorders

Depression, panic disorder, social anxiety disorder, GAD, externalizing disorders

GAD, generalized anxiety disorder; OAD, overanxious disorder.

Generalized Anxiety Disorder with Overanxious Disorder

The new formulations in DSM-IV for GAD indicate that children who formerly received a diagnosis of OAD should be placed in a new category called generalized anxiety disorder, or GAD. The criteria for GAD are permissive, hence a child could receive this diagnosis if he or she displayed only one of the six critical symptoms (restlessness, fatigue, difficulty concentrating, irritability, muscle tension, or sleep disturbance). The GAD symptoms are different from those that define OAD (worry about the past or future, concern about one's competence, need for reassurance, somatic symptoms, excessive self-consciousness, and muscle tension). Further, the new criteria for GAD resemble those used to diagnose major depressive episodes; examination of the overlap between OAD and GAD should take into account the possibility of a correlation with depression.

The Great Smoky Mountains Study, involving 1,420 children, which examined comorbidity among OAD, GAD, and depression (Costello, Mustillo, Erklani, Keeler, & Angold, 2003), found that among children who were comorbid (5.4% of the entire sample or 47% of those with any of the three diagnoses), more than half had all three disorders, and only 12 children (16% of those with GAD or OAD) had both disorders but no signs of depression.

Comorbidity Among the Phobias

Most published studies confirm comorbidities among the three phobias: specific, social, and agoraphobia. However, none had any association with PTSD after investigators controlled for other anxious comorbidities. This finding suggests that PTSD is a distinct disorder (Pine & Grun, 1999).

Comorbidity Between Panic Disorder and Separation Anxiety Disorder

There is no significant concurrent comorbidity between panic disorder and separation anxiety, but this tendency does not rule out the possibility of sequential comorbidity. Early appearance of separation anxiety appears to predict panic disorder (Black, 1994; Klein, 1995; Silove, Manicavasagar, Curtis, & Blaszczynski, 1996), but no community studies have tested this hypothesis adequately. There is little association between separation anxiety and either phobias or OAD.

Comorbidity With Other Disorders

Comorbidity between any one anxiety disorder on the one hand, and ADHD, conduct disorder, depression, or substance abuse disorder on the other, reveals the highest level of comorbidity with depression, with a median odds ratio of 8.2 (95% confidence interval (CI) 5.8–12; Costello, Egger, et al., 2004).

There is also a sequential link between early anxiety and later depression (Costello et al., 2003); Orvaschel, Lewinsohn, & Seeley, 1995). It is not clear, however, whether depression or anxiety increases the subsequent risk for the complementary disorder or whether the natural sequence is from an initial anxiety disorder to the later development of depression.

The odds ratio for the comorbidity of anxiety with risk for conduct disorder/oppositional disorder is 3.1 (95% CI 2.2–4.6), and with ADHD it is 3.0 (95% CI 2.1–4.3). These confidence intervals imply a significant degree of comorbidity. Although the bivariate odds ratios that involve substance use or abuse were significant in some studies, an association between anxiety and substance abuse disappeared when comorbidity between anxiety and other psychiatric disorders was controlled (Costello et al., 2004).

Although there is little concurrent comorbidity for anxiety and substance abuse (Weissman et al., 1999), childhood onset of an anxiety disorder might predict either lower or higher rates of substance abuse in adolescence. Kaplow, Curran, Angold, and Costello (2001) reported that children with separation anxiety were less likely than others to begin drinking alcohol, and if they did, they did so at a later age than that of most youth. But children with GAD were more likely to begin drinking and abuse alcohol earlier in adolescence.

Onset and Course

The evidence does not permit a confident reply to the question of whether anxiety disorders in preschool children are precursors of similar disorders in adolescents. Retrospective data, which are always fallible, indicate that adolescents with anxiety disorders recalled their first onset of anxiety being at around 7 years of age (Costello, Erkanli, Federman, & Angold, 1999; Orvaschel et al., 1995). The Great Smoky Mountains Study revealed that specific phobias, GAD, separation anxiety, and social phobia all appeared around the time the child began school, while agoraphobia, OAD, and OCD appeared several years later, usually at 9 to 11 years of age (Costello et al., 2003).

Although early anxiety disorder appears to be a precursor of later depression (Alloy, Kelly, Mineka, & Clements, 1990; Breslau, Schultz, & Peterson, 1995; Kendler, Neale, Kessler, Heath, & Eaves, 1992; Lewinsohn, Zinbarg, Seeley, Lewinsohn, & Sack, 1997; Silberg, Rutter, & Eaves, 2001a, 2001b; Silberg, Rutter, Neale, & Eaves, 2001), we do not know the influence of anxiety on the timing or occurrence of other psychiatric disorders, with the sole exception that early separation anxiety and GAD have different predictive consequences for the later abuse of alcohol (Kaplow, Curran, Angold, & Costello, 2001).

Summary of Prevalence

Separation anxiety and phobic disorders are seen in early childhood but become rare by adolescence. However, panic disorder and agoraphobia have the opposite developmental profile; they are rare in childhood and increase in adolescence. We do not yet know whether some adolescent disorders are later manifestations of an underlying syndrome that was displayed earlier or whether they represent new forms of psychiatric illness. An answer to this question requires longitudinal research.


Learning Theories

Early behavioral models for the treatment of anxiety have been based on two primary suppositions. First, fears and phobias are acquired through classical conditioning, i.e., through the formation of association between a neutral stimulus and an aversive stimulus such that the for mer acquires the aversive properties of the latter. The neutral stimulus is then designated as a conditioned stimulus (CS) and the original aversive stimulus is called an unconditioned stimulus (US). Second, the acquired fears can be unlearned through extinction, i.e., through presentation of the CS in the absence of the US. This conceptualization gave rise to exposure therapy, in which patients are taught to systematically confront their feared situations, objects, responses (e.g., tachycardia), or memories, under safe circumstances with the goal of extinguishing their phobic fear. While there have been debates about the mechanisms through which exposure therapy reduces anxiety symptoms, the benefit of this therapy has been demonstrated by a large body of research (cf., Barlow, 2001; Ollendick & March, 2004).

Discontent with nonmediational (automatic) accounts for acquisition and extinction of pathological anxiety led to the development of theories that posited a pivotal role for cognitive factors in anxiety (e.g., Beck, Emory, & Greenberg, 1985). The assumption here is that it is not the events themselves but rather their threat “meaning” that is responsible for the evocation of anxiety. Meaning in these theories is often assumed to be represented in language. Accordingly, in cognitive therapy for anxiety disorders, verbal discourse is used to challenge the patient's threat interpretations of events and to help replace them with more realistic ones, especially so with adolescents. With young children, however, cognitive therapy frequently takes the simpler form of thought replacement such that fearful thoughts are replaced with brave ones (e.g., “I am a brave boy. I can handle this.”). The child is provided new thoughts to replace the old ones (see Ollendick & Cerny, 1981).

The focus on the meaning of events as accounting for pathological anxiety paralleled the reconceptualization of conditioning in learning theories. For example, Rescorla noted that “conditioning depends not on the contiguity between the CS and US but rather in the information that the CS provides about the US” (Rescorla, 1988, p. 153) and that the “organism is better seen as an information seeker using logical and perceptual relations among events along with its own pre-conception to form a sophisticated representation of its world” (Rescorla, 1988, p. 154). In the same vein, when discussing the phenomenon of extinction, Bouton (1994, 2000) stated that “in the Pavlovian conditioning situation, the signal winds up with two available `meanings'” (Bouton, 2000, p. 58). Obviously, for rats the meaning of events cannot be represented in verbal language; rather, it is represented as associations among stimuli, responses, and outcomes.

Advances in information processing theories of conditioning and of pathological anxiety (e.g., Lang, 1977) influenced conceptualizations of treatment for the anxiety disorders. One such conceptualization, emotional processing theory, was proposed by Foa and Kozak (1986). In this theory, fear is viewed as a cognitive structure in memory that serves as a blueprint for escaping or avoiding danger that contains information about the feared stimuli, fear responses, and the meaning of these stimuli and responses. When a person is faced with a realistically threatening situation (e.g., a car accelerating at you, a fierce-looking dog approaching you) the fear structure supports adaptive behavior (e.g., swerving away, running away). A fear structure becomes pathological when the associations among stimulus, response, and meaning representations do not accurately reflect reality; in this instance, harmless stimuli or responses assume threat meaning. In emotional processing theory, meaning is thought to be embedded in associations among stimuli, responses, and consequences (as in Rescorla, 1988), as well as in language, especially in the form of thoughts, beliefs, and evaluations (as in Beck, 1976).

Within emotional processing theory the anxiety disorders are thought to reflect the operation of specific pathological fear structures (cf. Foa & Kozak, 1985). For example, the fear structure of individuals with panic disorder is characterized by erroneous interpretations of physiological responses associated with their panic symptoms (e.g., tachycardia, difficulty breathing) as dangerous (e.g., leading to heart attack). As a result of this misinterpretation, individuals with panic disorder avoid locations where they anticipate experiencing panic attacks or similar bodily sensations, such as physical exertion. In contrast, the fear structure of individuals with OCD most often involves the erroneous interpretation of safe stimuli (e.g., brown spots) as dangerous (e.g., AIDS-contaminated blood). Accordingly, the core pathology in panic disorder lies in the erroneous meaning of physiological responses, whereas the core pathology of OCD lies in the erroneous meaning of external events. The supposition that inaccurate negative cognitions underlie the anxiety disorders has also been at the heart of theories posed by cognitive therapists (e.g., Clark, 1986; Rapee & Heimberg, 1997; Salkovskis, 1985).

If fear and avoidance reflect the activation of an underlying cognitive fear structure, then changes in the fear structure should result in corresponding changes in emotions and behavior. Indeed, Foa and Kozak (1986) proposed that psychological interventions known to reduce fear, such as exposure therapy, achieve their effects through modifying the fear structure. According to emotional processing theory two conditions are necessary for therapeutic fear-reduction to occur: first, the fear structure must be activated; second, information that is incompatible with the pathological aspects of the fear structure must be available and incorporated into the existing structure. Thus, within this framework, exposure therapy is thought to correct the erroneous cognitions that underlie the specific disorder (e.g., tachycardia equals heart attack). This is also the explicit mechanism by which cognitive therapy is thought to reduce fear. In this way the mechanisms that are thought to operate during exposure greatly overlap with those of cognitive therapy. Moreover, some cognitive therapists explicitly posit that fear activation is necessary to refute the patient's false interpretations, and cognitive therapy programs routinely include an exposure component in the form of “behavioral experiments.” It must be noted that the evidence for change in cognitions as the central mechanism in fear reduction is somewhat incomplete at this time; accordingly, additional work on the mediators or mechanisms of change in both the cognitive and behavior therapies is drastically needed. Such research is especially needed with children and adolescents, in whom the exact role of cognitions has been less frequently examined (Prins & Ollendick, 2003).

Cognitive Correlates of Anxiety Disorders

The cognitive approach to anxiety disorders comprises two research traditions (McNally, 2001b). In one tradition, researchers assume that introspective self-reports of anxious individuals can reveal aberrant cognition underlying symptom expression. These scholars administer questionnaires and conduct interviews to ascertain, for example, the intensity, frequency, and content of the worries and fears of children and adolescents. One such study revealed that school-age children worry most about school, health, and personal harm, especially the latter (Silverman, La Greca, & Wassertein, 1995). Another indicated that children and adolescents suffering from anxiety disorders report the same kinds of worries as those of their healthy counterparts, but that the intensity (not the number) of worries distinguished youngsters with anxiety disorders from those without anxiety disorders (Weems, Silverman, & La Greca, 2000). Researchers in this tradition have also studied the fear of anxiety symptoms (i.e., anxiety sensitivity; Reiss & McNally, 1985). Silverman and colleagues have developed the Childhood Anxiety Sensitivity Index (CASI; Silverman, Fleisig, Rabian, & Peterson, 1991; Silverman & Weems, 1999) to investigate this phenomenon.

In the second tradition, researchers eschew self-report as insufficiently sensitive to measure abnormalities in cognitive mechanisms that often operate rapidly, and outside of awareness. These scientists apply the methods of experimental cognitive psychology to elucidate biases favoring processing of threat-related information in anxiety-disordered patients (McNally, 1996; Williams, Watts, MacLeod, & Mathews, 1997). In this section, we review experiments on information-processing biases in anxious children and adolescents (see also Vasey, Dalgleish, & Silverman, 2003; Vasey & MacLeod, 2001).

Attentional Bias for Threat

Because attentional capacity is limited, people can attend only to certain stimuli at a given time, and any bias for selectively attending to threat-related stimuli should increase a person's likelihood of experiencing anxiety. Two experimental tasks have confirmed that adults with anxiety disorders are characterized by an attentional bias for processing information about threat. In the emotional Stroop task (Williams, Mathews, & MacLeod, 1996), subjects are shown words of varying emotional significance and are asked to name the colors in which the words appear while ignoring the meanings of the words. Delays in color-naming (“Stroop Interference”) occur when the meaning of the word automatically captures the subject's attention despite the subject's effort to attend to the color in which the word is printed. Most studies have shown that patients with anxiety disorders take longer to name the colors of words related to their threat-related concerns than to name the colors of other emotional or neutral words, and take longer to name the colors of threat words than do healthy subjects.

Although the emotional Stroop task has been traditionally interpreted as tapping an attentional bias for threat, debate continues about the mechanisms underlying the effect (Williams et al., 1996). For example, the emotional Stroop may reflect an inhibitory problem rather than an attentional one. That is, delayed color-naming of trauma-related words may reflect difficulty suppressing the meaning of trauma-related concepts once they are activated rather than selective attention per se (McNally, 2003, pp. 301–302).

Studies on the emotional Stroop in children have revealed mixed results. Relative to control subjects, spider-fearful children take longer to name the colors of spider words (Martin, Horder, & Jones, 1992) and colors of line drawings of spiders (Martin & Jones, 1995). Adolescents who developed PTSD after having survived a shipwreck exhibited Stroop interference for trauma-related words (Thrasher, Dalgleish, & Yule, 1994). Children who developed PTSD after being either physically or sexually abused exhibited similar patterns of trauma-related Stroop interference (Dubner & Motta, 1999). Relative to control subjects, children and adolescents (aged 9–17 years) with PTSD arising from either road traffic accidents or exposure to violence exhibited greater interference for trauma words than for neutral words (Moradi, Taghavi, Neshat-Doost, Yule, & Dalgleish, 1999). The magnitude of this trauma-related interference effect was unrelated to the age of the patients.

However, not all Stroop studies have confirmed an anxiety-linked attentional bias for threat cues in youngsters. For example, nonanxious as well as anxious children have exhibited delayed color-naming of threat words (Kindt, Bierman, & Brosschot, 1997; Kindt, Brosschot, & Everaerd, 1997). A pictorial version of the spider Stroop (naming colors of background against which spider pictures appeared) did not reveal a fear-related effect in children (ages 8–11; Kindt, van den Hout, de Jong, & Hoekzema, 2000).

A second paradigm provides a much less controversial measure of attentional bias. In the Dot Probe Attention Allocation Task (MacLeod, Mathews, & Tata, 1986), subjects view two words on a computer screen, one appearing above the other. On some trials, one word is threat related, whereas the other is not. After the words disappear, a small dot replaces one of the words. Subjects press a button as soon as they detect the dot. Relative to healthy control subjects, patients with anxiety disorders are faster to respond when the dot replaces a threat word than when it replaces a neutral word. Because threat cues capture attention in anxious patients, these individuals are especially quick to respond to a neutral cue that follows a threat cue.

Using this task, Vasey, Daleiden, Williams, and Brown (1995) found that children (ages 9–14 years) with anxiety disorders exhibited an attentional bias for threat, whereas control children did not. The attentional bias increased with age and with reading ability. Relative to their nonanxious counterparts, test-anxious school children (ages 11–14 years) exhibited an attentional bias for threat words (both socially and physically threatening; Vasey, El-Hag, & Daleiden, 1996). Patients with GAD (ages 9–18 years) exhibited an attentional bias for threat words, whereas patients with mixed anxiety and depression or healthy control subjects did not (Taghavi, Neshat-Doost, Moradi, Yule, & Dalgleish, 1999). The GAD patients did not show an attentional bias for depression-related words, and the attentional bias for threat words was unrelated to the age of the subject. Finally, patients with PTSD (ages 9–17 years) arising from either nondomestic violence or road traffic accidents ex hibited an attentional bias for social threat words (but not physical threat words), whereas control subjects did not (Dalgleish, Moradi, Taghavi, Neshat-Doost, & Yule, 2001).

Interpretive Bias for Threat

Anxious children tend to interpret ambiguous information in a threatening fashion. In one study, children (ages 7–9 years) heard homophones (e.g., whipping) that could be interpreted in either a threatening or a nonthreatening fashion (Hadwin, Frost, French, & Richards, 1997). The higher a child's self-reported trait anxiety, the more likely the child selected threatening pictures (e.g., rope) over nonthreatening pictures (e.g., cream) that made the homophones (e.g., whipping) unambiguous. In another study, GAD patients (ages 8–17 years) and healthy control children were shown homographs (e.g., hang), each possessing a threatening and a nonthreatening meaning (Taghavi, Moradi, Neshat-Doost, Yule, & Dalgleish, 2000). They were asked to construct a sentence including the homograph. Relative to the sentences constructed by control children, the anxious children more often constructed sentences incorporating the threatening interpretation of the homograph, implying that they had interpreted the ambiguous word in terms of its threatening meaning. This interpretive bias was unrelated to the age of the subjects.

Researchers asked anxious and nonanxious children to provide interpretations of ambiguous scenarios. Anxious fourth and fifth graders were more likely than their nonanxious peers to interpret nonhostile scenarios in a threatening fashion, whereas both groups interpreted ambiguous scenarios in a hostile fashion (Bell-Dolan, 1995). Patients with anxiety disorders (ages 9–13 years) exhibited a bias for interpreting ambiguous scenarios in a threatening manner, and this effect was strongly predicted by level of trait anxiety (Chorpita, Albano, & Barlow, 1996). Relative to healthy control children, patients ranging in age from 7 to 14 years who had anxiety disorders (overanxious, separation anxiety, social phobia, simple phobia), exhibited a bias for interpreting ambiguous scenarios in a threatening fashion (Barrett, Rapee, Dadds, & Ryan, 1996). This bias, however, was even more pronounced in patients with oppositional-defiant disorder.


Anxious children and adolescents exhibit threat-related attentional and interpretive biases that resemble those exhibited by anxious adults. Moreover, within most studies, the extent of bias did not vary as a function of the child's age. Still, questions remain. In one study, the responses of anxious children to two measures of attentional bias (dot probe and emotional Stroop) were uncorrelated, indicating that these tasks tap distinct constructs (Dalgleish et al., 2003). Further, researchers have yet to test whether these biases disappear following successful psychological or pharmacological treatment. A more detailed critique of information processing in adolescent psychopathology is available elsewhere (Vasey et al., 2003).

Biological Features of Adolescents and Anxiety States

Although the exact timing of puberty is not easy to specify, scientists agree that the psychological and biological features of the era called adolescence are influenced by cultural setting. Some cultures, like our own, delay the assumption of adult roles; others require a clear transition, with or without a rite-of-passage ceremony (Schlegel & Barry, 1991). Nonetheless, there is an identifiable period between 12 and 18 years, ubiquitous across societies, characterized by changes in hormones, brain structure, and behavior. These properties may have been conserved over evolution to promote autonomy and to foster dispersal of some individuals from the natal territory to another in order to avoid inbreeding (Schlegel & Barry, 1991; Spear, 2000).

Adolescence is marked by a reactivation of the hypothalamic–pituitary–gonadal axis, development of secondary sexual characteristics, and the onset of reproductive capacity, even though the increased circulation of sex hormones does not account for much of the variance in the behavior of adolescents (Brooks-Gunn, Graber, & Paikoff, 1994). The timing of pubertal signs is influenced by gender and environment; onset of puberty may be influenced more strongly by environmental stressors in girls than in boys (Moffitt, Caspi, Belsky, & Silva, 1992).

Brain Changes

The brain undergoes changes throughout life (Eriksson et al., 1998), with intervals of modest change punctuated by periods of more rapid transformation (Spear, 2000). Periods of more dramatic change include not only pre-and early postnatal eras but also adolescence (Spear, 2000). Rakic, Bourgeios, and Goldman-Rakic (1994) estimate that up to 30,000 cortical synapses are lost every second during portions of the pubertal period in nonhuman primates, resulting in a decline of nearly 50% in the average number of synaptic contacts per neuron, compared with the number prior to puberty. There is a similar loss of synapses in the human brain betwen 7 and 16 years of age (Huttenlocher, 1979), but the scarcity of human postmortem tissue makes it difficult to provide a more detailed description of this phenomenon. Although the implications of the massive pruning remain speculative, it is likely that it reflects active restructuring of connections and the promotion of more mature patterns. Some forms of mental retardation are associated with unusually high density of synapses (Goldman-Rakic, Isseroff, Schwartz, & Bugbee, 1983).

The elimination of synapses that are presumed to be excitatory, accompanied by a reduction in brain energy utilization, transform the adolescent brain into one that is more efficient and less energy consuming (Chugani, 1996; Rakic et al., 1994). These changes could permit more selective reactions to stimuli that in younger children activate broader cortical regions (Casey, Geidd, & Thomas, 2000).

Adolescence is also marked by changes in the relative volume and level of activity in different brain regions. For example, there is an increase in cortical white matter density (reflected in myelinated fiber tracts) and a corresponding decrease in gray matter, especially in frontal and prefrontal regions (Giedd et al., 1999; Sowell et al., 1999a, 1999b). The overall result of these varied changes is net decrease in the volume of the prefrontal cortex (Sowell et al., 1999b, van Eden, Kros, & Uylings, 1990). In the hippocampus and the amygdala however, gray matter volumes continue to increase during late childhood and adolescence (Giedd et al., 1997; Yurgelun-Todd, Killgrove, & Cintron, 2003).

There are also developmental shifts in patterns of innervation, including the circuits involved in the recognition and expression of fear, anxiety, and other emotions (Charney & Deutsch, 1996). The responsiveness of the cortical gamma-aminobutyric acid (GABA)–benzodiazepine receptor complex to challenge increases as animals approach puberty (Kellogg, 1998), and there are maturational changes in the hippocampus in humans as well as in animals (Benes, 1989; Wolfer & Lipp, 1995), especially increases in GABA transmission (Nurse & Lacaille, 1999). Further, pubescent animals show lower utilization rates of serotonin in the nucleus accumbens than younger or older animals (Teicher & Andersen, 1999).

Developmental increases in amygdala–prefrontal cortex (PFC) connectivity are seen during adolescence in work conducted in laboratory animals (Cunningham, Bhattacharyya, & Benes, 2002), along with alterations in amygdala activation (Terasawa & Timiras, 1968) and the processing of emotional and stressful stimuli. Lesions of the amygdala have opposite effects on fearfulness to social stimuli when those lesions are in infant versus adult monkeys (Prather et al., 2001). Although levels of negative affect and anxiety have been correlated with amygdalar activity in adults (Davidson, Abercrombie, Nitschke, & Putnam, 1999), recent studies using functional magnetic resonance imaging (fMRI) to examine amygdalar activation in response to emotionally expressive faces in younger individuals have yielded a varying mosaic of evidence (Killgore, Oki, & Yurgelun-Todd, 2001; Pine, Grun, et al., 2001).

Maturational changes in the cerebellum, and the circuitry connecting the cerebellum to the prefrontal cortex, continue through adolescence. Lesions of the adult cerebellum disrupt the regulation of emotion and interfere with performance on tasks requiring executive functions (Schmahmann & Sherman, 1998), although this is less apparent in those younger than 16 years (Levisohn, Cronin-Golomb, & Schmahmann, 2000).

One consequence of this restructuring of the brain during adolescence is that early developmental compromises might be exposed. That is, brain regions vulnerable to dysfunction, due either to genetics or to adverse early experience, might be unmasked by the combination of brain restructuring and stressful life experiences (Goldman-Rakic et al., 1983; Hughes & Sparber, 1978).


There is great interest in detecting the biological variables that might distinguish anxious from nonanxious patients. Many, but not all, of these biological measures are influenced directly or indirectly by activity of the amygdala, bed nucleus, and their projections to the brain stem, autonomic nervous system, endocrine targets, cortex, and central gray matter (Pine, 1999, 2001, 2002; Pine, Cohen, & Brook, 2001; Pine, Fyer, et al., 2001; Pine, Grun, et al., 2001). It is relevant that connectivity between the amygdala and prefrontal cortex, along with level of amygdalar activation, increases during adolescence (Cunningham, Bhattacharyya, & Benes, 2002; Terasawa & Timiras, 1968).

Does the Amygdala Respond to Threat or Novelty?

There is debate over whether the amygdala is activated primarily by events that are potentially harmful or events that are unexpected or discrepant. Support for the former, more popular, position comes from the elegant research of LeDoux (1996, 1998, 2000) and Davis (1992, 1998) who have shown that acquisition of conditioned body immobility or bodily startle in rats, via Pavlovian conditioning with electric shock as the unconditioned stimulus, requires the integrity of the amygdala. But the amygdala also responds to discrepant and unexpected events that are harmless. Select neurons in the amygdala, as well as in the bed nucleus, hippocampus, and brainstem sites, reliably respond to unexpected or discrepant events, regardless of whether they are threatening or harmful (Wilson & Rolls, 1993). And the reactivity of amygdalar neurons to unexpected or discrepant events habituates, often rapidly, as the event becomes expected and loses its surprise value (La Bar, Gatenby, Gore, Le Doux, & Phelps, 1998).

Nonetheless, Ohman and Mineka (2001) argue that the amygdala reacts primarily to signs of danger rather than to novelty. They suggest that all animals inherit a fear module, located in the amgydala, that reacts without conscious awareness and free of cognitive control to events that pose a threat to the integrity of the body (confrontation with snakes and spiders are classic examples of fear-evoking events). There are serious problems with this theoretical position.

First, the behavioral reactions of monkeys, chimpanzees, and human infants to a snake are no different from their reactions to discrepant events that are harmless (for example, a tortoise or seaweed). The British psychiatrist Isaac Marks (1987) described the terror his 21 2 -year-old son displayed when he first saw thousands of dried skeins of seaweed. However, the boy lost his fear following repeated exposures to these stimuli.

Only 30% of monkeys born and reared in the laboratory showed more prolonged withdrawal to a live snake than to blue masking tape (Nelson, Shelton, & Kalin, 2003). If snakes were a biologically potent incentive for fear, a majority of monkeys should have shown an immediate withdrawal reaction.

It is relevant that discrepant events that pose no danger can produce the same level of amygdalar activation as dangerous ones. Adults in an fMRI scanner looking at faces with neutral expressions showed amygdalar activation to new, compared with familiar faces, even though no face had a fearful, disgusting, or threatening expression (Schwartz et al., 2003).

These data suggest that the amygdala is biologically prepared to react to unexpected or discrepant events, a hypothesis supported by Cahill and McGaugh (1990), who believe that a primary function of the amygdala is to initiate a cascade of physiological reactions to novel or unexpected events. The degree of activation of the amygdala is correlated with the degree of arousal produced by the unexpected event, and not with its potential for danger or level of aversiveness. Although some might argue that every unexpected event also elicits a fear state this assumption seems a bit difficult to defend when the discrepant event is a neutral face, unexpected food for a rat, or the sudden appearance of a smiling parent from behind a screen of closed hands saying “peek a boo.”

The phenomena of conditioned freezing or bodily startle in an animal, proposed as a model for human anxiety disorders, have so deeply penetrated contemporary thought that many clinicians and researchers have forgotten that humans do not become fearful or anxious to events qua events, but to the symbolic interpretations imposed on them. An anthropologist who studied the Ojibwa Indians of Northern Canada over 60 years ago (Hallowell, 1955) observed that adults report a state of fear when they fall chronically ill because prolonged illness means that a sorcerer has cast a spell on them. The symbolic meaning of the illness, not the somatic distress of being sick, is the origin of the adult fear.

Americans who report a fear of snakes are aware of the discrepant features of this species. Snakes have unusual skin covering, a typical ratio of head to body, and locomote in an unusual way. Seventeen of 22 adults with an animal phobia reported that it was the discrepant form of the animal's locomotion or appearance that upset them. No phobic patient had experienced any harm as a result of encountering the feared animal (McNally & Steketee, 1985).

An American woman with a phobia of birds supports this argument. The woman dates the origin of her fear to an afternoon when, as a seven year old, she was watching Hitchcock's film, “The Birds,” which showed large flocks of birds attacking humans. The woman remembers feeling very surprised by the fact that birds, which she had regarded as benevolent and beautiful, could be aggressive to humans. The sharp disconfirmation of her childhood belief could have activated the amygdala. Because the film displayed birds attacking humans, the idea of harm became associated with amygdalar activation and the accompanying somatic consequences produced by the feeling of surprise. If the girl had not been surprised by the birds' behavior, it is likely that the phobia for birds would not have developed.

However, because some unexpected events are potential threats—the attack of a large dog or a scorpion on the bedspread—it is likely that some anxiety disorders, but probably not most, are the result of Pavlovian conditioning mechanisms.

Biological Correlates of Adolescent Anxiety Disorders

Advances in genetics, imaging, and cognitive neuroscience provide the opportunity to combine discoveries in neuroscience with insights from clinical psychobiology. Current views of adolescent anxiety disorders are influenced by two limiting facts. The first is that the research on adolescents has been modeled on investigations of adults; the second is that all current anxiety disorders are heterogeneous in their origin. This second fact means that investigators would profit from using biological variables to distinguish between patients with transient symptoms and those with more persistent disorders (Merikangas, Avenevoli, Dierker, & Grillon, 1999; Pine, Wasserman, & Workman, 1999).

Autonomic Nervous System

Many, but not all, adults with anxiety disorders show abnormalities of autonomic regulation, especially lability of the cardiovascular system. This feature is most common among adults with panic disorder, social anxiety, and GAD (Gorman & Sloan, 2000). These abnormalities occur in both the sympathetic and parasympathetic systems and probably contribute to the association between anxiety disorder and cardiovascular mortality (Gorman & Sloan, 2000). Although children at risk for one or more anxiety disorders, because of a temperamental bias, show high sympathetic tone in the cardiovascular system (Kagan, Snidman, McManis, & Woodward, 2001), this relation is not robust and children with different disorders often display similar autonomic profiles (Pine et al., 1998). One mechanism that ties autonomic regulation to psychology is the result of peripheral feedback from the cardiovascular system to the brain. If this so matic activity pierces consciousness, the person might conclude that a threat is imminent (Moss & Damasio, 2001). Perturbations in respiratory function are characteristic of panic disorder (Pine, 1999) and lead panic patients to experience a heightened feeling of anxiety (Coryell, Fyer, Pine, Martinez, & Arndt, 2001; Pine et al., 2000).

Hypothalamic–Pituitary–Adrenal Axis

Patients with an anxiety disorder often show perturbations in the HPA axis. Further, both rodents and nonhuman primates show changes in the hypothalamic–pituitary–adrenal (HPA) axis during acute stress, as well as after a stress experienced early in life (Essex, Klein, Cho, & Kalin, 2002; Kaufman, Plotsky, Nemeroff, & Charney, 2000; Meaney, 2001; Monk, Pine, & Charney, 2002). The strongest association between activation in the HPA axis and anxiety disorder is seen in PTSD (Bremner, 1999; Bremner et al., 1999; Yehuda, 2002). Although enhanced feedback sensitivity in the HPA axis is often associated with an anxiety disorder, unfortunately, some children with an anxiety disorder exhibit the opposite pattern of reduced feedback sensitivity (Coplan et al., 2002; De Bellis, 2001; Heim & Nemeroff, 2002).


Brain chemistry can affect the excitability of a particular brain region in diverse ways. Neurochemical regulation in adult anxiety disorders is studied most often with pharmacological challenges, positron emission tomography, or measurement of peripheral neurochemical metabolites. Because the first two techniques are invasive, data on adolescents are restricted primarily to peripheral measures.

Adults with anxiety often show enhanced activity in the neurons of the locus ceruleus (Coplan et al., 1997; Sullivan, Coplan, & Gorman, 1998; Sullivan, Coplan, Kent, & Gorman, 1999). For example, adults with panic disorder and children with separation anxiety disorder show an abnormal response to the administration of yohimbine (Sallee, Sethuraman, Sine, & Liu, 2000). However, children and adults with a diagnosis of OCD show an abnormal, neurohormonal response to clonidine (Sallee et al., 1998). There is also an association between environmental stress and a prolactin response to serotonergic probes (Heim & Nemeroff, 2002), and adults with anxiety disorders show abnormalities in serotonergic regulation.


A dramatic indication of a relation between immunology and anxiety disorder comes from studies of children with OCD. Earlier work had found a specific association between OCD and neurological conditions affecting the basal ganglia, including pediatric Snydenham's chorea. This discovery led to the recognition of a specific form of OCD, called Pediatric Autoimmune Neuropsychiatric Disorder Associated with Streptococcus (PANDAS; Swedo, 2002), marked by anxiety, OCD, and motor tics that emerge following infection with group A ß-hemolytic streptococcus. This syndrome reflects an immunological reaction within underlying fronto-striatal-thalamo-cortical-circuitry. It may be relevant that the offspring of adults with panic disorder show selected allergic disorders reflecting anomalies in the immune system (Kagan et al., 2001; Slattery et al., 2002).

Brain Imaging

A variety of techniques have been used to study anxiety disorder. These include MRI, fMRI, magnetic resonance spectroscopy (MRS), and electrophysiology.

Morphometric MRI evidence, which provides information on brain structure, reveals that OCD adults have abnormalities in the circuit involving the prefrontal cortex, basal ganglia, and thalamus (Rauch, Savage, Alpert, Fischman, & Jenicke, 1997). Some of these abnormalities have been observed in children and adolescents with OCD (Rosenberg & Hanna, 2000; Rosenberg, MacMillan, & Moore, 2001). Adults with PTSD have reduced volume in the hippocampus; but children with PTSD do not show these specific reductions, even though they have a smaller brain volume (De Bellis et al., 1999). Children with GAD show increased volume of the amyg dala and superior temporal gyrus of the right hemisphere (De Bellis et al., 2002).

Functional magnetic resonance imaging quantifies brain activity. Despite these advantages it relies on measures of blood flow and therefore is an indirect index of neuronal events. Moreover, fMRI does not measure absolute amount of blood flow, but differences in changes in blood flow during an experimental task compared with a control task.

Despite these caveats, adults with PTSD show enhanced amygdalar activation (Rauch et al., 2000), and children with anxiety disorders show activation to faces with fearful expressions (Thomas et al., 2001a, 2001b). However, this latter response could be due to the surprise of seeing fearful faces, because healthy children show enhanced amygdalar activation to neutral faces (Thomas et al., 2001a, 2001b).

Magnetic resonance spectroscopy (MRS) is a noninvasive technique that can reveal aspects of brain neurochemistry. One study with MRS found a reduction in levels of N-acetylaspartate in the cingulate gyrus of children who had PTSD (De Bellis, Keshavan, Spencer, & Hall, 2000).

Electroencephalogram Activity

The electroencephalogram (EEG) represents the synchronized activity of large numbers of cortical pyramidal neurons which, at any moment, have a dominant frequency of oscillation at particular sites. A state of mental and physical relaxation is usually but not always associated with more power in the alpha frequency band (8–13 Hz) in frontal areas. A state of psychological arousal is associated with greater power in the higher-frequency beta band (14–30 Hz). The change to higher frequencies could be the result of more intense volleys from the amygdala to the cortex.

In addition, there are usually small hemispheric differences in the amount of alpha power on the right, compared with the left, side at frontal and parietal sites. Because alpha frequencies are associated with a relaxed psychological state, the less alpha power at a particular site, the more likely that site is neuronally active. The technical term for loss of alpha power is desynchronized, and investigators assume that desynchronization of alpha frequencies is a sign that the individual has moved to a more aroused state.

Subjects reporting higher anxiety tend to have greater activation in the right frontal area than the left, whereas normal controls show more activation in the left frontal area. A preference for display of right versus left frontal activation could reflect either a stable trait or a transient state. It appears that a stable preference for right or left frontal activation can be influenced by an individual's temperament and, therefore, could reflect a stable property (Fox, Henderson, Rabin, Caikins, & Schmidt, 2001). McManis, Kagan, Snidman, and Woodward (2002) have found that 11-year-old children who had been highly reactive infants and fearful toddlers were likely to show right frontal activation under resting conditions. However, an asymmetry of activation can also reflect a transient state. Infants watching the approach of a stranger showed greater right frontal activation during that brief period of time (Fox & Bell, 1990). Hagemann and colleagues, who gathered EEG data on four separate occasions on a sample of 59 adults, concluded that 60% of the variance in asymmetry of activation reflected a stable trait while 40% was attributable to the specific occasion of testing (Hagemann, Naumann, Thayer, & Bartussek, 2002).

The event-related potential is a time-locked, post-synaptic potential generated by large numbers of cortical pyramidal neurons to a specific stimulus. The first waveform that represents the detection of a discrepancy is called N2 because it usually peaks at about 200 msec to an unexpected event. The two most frequently studied waveforms, P3 and N4, appear a bit later with peak voltages at about 400 msec, and are prominent at frontal sites when the subject is passive and has no task to perform. Kagan and colleagues have unpublished data indicating that 11-year-old children who had been highly reactive infants and fearful toddlers showed a larger negative waveform at 400 msec to nonthreatening discrepant scenes. Although this research is preliminary, it suggests that future investigators should examine EEG profiles and event-related potentials in their study of anxiety and anxiety disorders.


Years of work have affirmed that genetic factors influence the risk for anxiety disorders. One study of adults found modest heritability for GAD for both men (15%) and women (20%) and no effect of shared environment (Hettema, Prescott, & Kendler, 2001; see Table 9.3). Other research affirms the heritability of panic disorder (Crowe, 1985; Gorwood, Feingold, & Ades, 1999; Marks, 1986; Skre, Onstad, Torgersen, Lygren, & Kringlen, 1993). Merikangas and Risch (2003) suggest heritability estimates of 50% to 60% for adult panic disorder, with risk ratios ranging from 3 to 8 for first-degree relatives of adult probands with panic disorder. A meta-analysis by Hettema, Neale, and Kendler (2001) uncovered a modest genetic contribution to four anxiety categories and little or no effect of shared environment. One of the most extensive explorations of the contribution of genes to anxiety disorders is the Virginia Twin Study of Adolescent Behavioral Development (VTSABD; Eaves et al., 1997). This corpus, which relies primarily on self-report data, discovered strong additive genetic effects for OAD in both boys and girls (37%), with little effect of shared environment (Topolski et al., 1997). Silberg and colleagues (Silberg, Rutter, Neale, & Eaves, 2001) reported that 12% to 14% of the variance in OAD in girls was attributable to genes, and most of the remaining variance to nonshared environment.

Table 9.3 Genetics of Anxiety Disorders: Result of Meta-Analysis of Studies of Adults


Odds Ratioa


Shared Environment

Nonshared Environment

Panic disorder




Generalized anxiety disorder










Obsessive-compulsive disorder


a Ratio of prevalence in relatives of probands to that of relatives of comparison subjects

Source: From Hettema, Neale, and Kendler (2001).

The Virginia Twin Study corpus indicated a smaller genetic contribution to separation anxiety (only 4%), but large nonshared environmental effects (40% and 56%). The data for girls revealed minimal genetic effects on separation anxiety and a greater contribution of shared environment (11% for children 8 to 12 years old, 23% for children 14 to 17 years old; Silberg, Rutter, Neale et al., 2001). However, parent-report checklists from an Australian national twin registry found a higher genetic loading for separation anxiety symptoms in girls (50%) and a much lower one for boys (14%) (Feigon, Waldman, Irwin, Levy, & Hay, 2001).

The Virginia Study indicated that about 9% to 10% of the variance in phobic symptoms was genetic in girls; the remainder was attributable to nonshared environment (Topolski et al., 1997). However, a Swedish study found that shared environmental factors explained considerably more of the variance for fears of animals, unfamiliar situations, and mutilations than non-shared environment (Lichtenstein & Annas, 2000). Thus, it is important to appreciate that conclusions based on twin studies can vary markedly as a function of the site of the laboratory, as well as the informant supplying the relevant information. When mothers reported on separation anxiety disorder in a population-based sample of female twins living in Missouri, heritability estimates were high (62%) and there was only a modest effect of shared environment.

Weissman (1988) argued almost 20 years ago that high rates of separation anxiety in children of parents who were comorbid for panic and depression disorder implied an association between separation anxiety disorder in childhood and the later development of panic disorder. There was a fairly specific association between separation anxiety in children who had been brought to clinics and separation anxiety in the parents when they were children years earlier (Manicavasagar, Silove, Rapee, Waters, & Momartin, 2001).

In addition, there is evidence for genetic contributions to personality traits such as neuroticism and introversion (Eaves, Eysenck, & Martin, 1989), shyness (Daniels & Plomin, 1985), and behavioral inhibition (DiLalla, Kagan, & Reznick, 1994; Kagan, 1994). A group of very shy 7-year-old Israeli children were more likely than others to inherit the long form of the allele for the serotonin transporter promoter region polymorphism (Arbelle et al., 2003); however, not all studies have found this association.

Despite these findings, many studies fail to meet the highest research standards, which include the following: (1) clearly operationalized diagnostic criteria; (2) systematic ascertainment of probands and relatives; (3) direct interviews with a majority of subjects; (4) diagnostic assessment of relatives by investigators blind to the proband's status; and (5) family studies with inclusion of comparison groups (Hettema, Neale, & Kendler, 2001). These standards are occasionally met in studies with adults, but rarely in studies with children and adolescents.

Genes are only expressed within a certain envelope of environments and individuals both shape and select their environments (Rutter, Silberg, O'Connor, & Siminoff, 1999a, 1999b). Finally, it should be appreciated that the attribution of a genetic risk to an individual should not invite fatalism (Rutter et al., 1990). Some heritable conditions can be treated and a few can be controlled. The classic example is phenylketonuria, for which the cognitive impairment is caused by an inherited metabolic defect that can be controlled by restricting the child's diet.

By developing personalized treatment plans, the revolution in molecular genetics promises to transform the identification and treatment of anxiety disorders across the lifespan. Two complementary approaches are described. Pharmacogenomic studies use genomic technologies to identify chromosomal areas of interest and, hence, potential drug targets (see, for example, Arbelle et al. 2003; Smoller et al., 2003); pharmacogenetic studies identify candidate genes that moderate drug response (see, for example, Basile, Masellis, Potkin, & Kennedy, 2002), or adverse event profile (see, for example, Murphy, Kremer, Rodrigues, & Schatzberg, 2003). Identified difference may interact with age, gender, race and ethnicity (Lin, 2001).

In the adult literature on genetic factors, the most robust findings involve polymorphisms in the serotonin transporter (Weizman and Weizman, 2000). In comparison to progress in ADHD (Rohde, Roman, & Hutz, 2003), however, little is known about pharmacogenetic or pharmacogenomic approaches to anxiety disorders in the pediatric population. Shyness (Arbelle et al., 2003) and behavioral inhibition (Smoller et al., 2003) but not internalizing symptoms (Young, Smolen, Stallings, Corley, & Hewitt, 2003) all have been linked to candidate gene variation, illustrating how lack of consistency in phenotypic identification among other factors limits progress despite clear evidence from statistical genetic methods regarding the importance of genetic factors (Stein, Chavira, & Jang, 2001).

Future progress will depend on an improved understanding of the nature and identification of disease states and their natural course, which in turn will allow the development of more specific treatments, better risk prediction, and the implementation of preventive strategies based in pharmacogenomic and pharmacogenetic approaches (Gottesman and Gould, 2003; Pickar, 2003).


The research of the past few decades has expanded our understanding of the phenomena linked to the concepts of anxiety and anxiety disorder. A comparison of contemporary reports with those of the last half century provides reason for optimism, for we have learned several important facts.

First, the state we call anxiety in humans is not unitary in origin or consequence and can be the result of living with realistic threat, past history, conditioning, or a temperamental bias for unexpected somatic sensations that are interpreted as meaning one is anxious. Second, epidemiological and genetic data imply distinct bi ological profiles for the varied anxiety disorders, many of which implicate neurochemical processes. Finally, clinicians and investigators now have an initial set of cognitive and biological procedures that promise to aid differential diagnosis of individuals who report anxiety. Major advances will occur when investigators and clinicians add these procedures to their interview data. The results of this work will permit the parsing of individuals who have a particular diagnosis into subgroups with more homogeneous biological and psychological features. This knowledge should lead to a more fruitful set of psychiatric classifications.