effect profile (e.g., not sedating) may be better tolerated by some patients than that of benzodiazepines.

Antidepressant medications have shown efficacy for GAD in controlled trials. Because of very high rates of comorbidity of GAD with depressive disorders and symptoms, antidepressant medications may provide additional benefits for GAD patients. However, most antidepressants are slower acting than benzodiazepines, often taking 3–4 weeks for benefits to become obvious. Tricyclic antidepressants (TCAs) such as imipramine demonstrated benefit over placebo in early controlled trials, yet recent controlled trials have shown benefits for newer antidepressants such as the SSRI paroxetine and the dual serotonin–norepinephrine reuptake inhibitor (SNRI) venlafaxine-XR. The latter is the first antidepressant to receive a U.S. Food and Drug Administration (FDA) indication for treatment of GAD. These antidepressants are well tolerated and lend to long-term treatment, and not surprisingly have quickly become commonplace for treatment of GAD. For a detailed review of pharmacotherapy for adult GAD see the report by Roy-Byrne and Cowley (2002).

In examining the relative and combined efficacy of CBT and medication, Power et al. (1990) randomly assigned 113 patients to one of five treatment conditions: CBT + diazepam (DZ), CBT + placebo (PBO), CBT alone, DZ alone, and PBO. Percent responders, defined as reduction of >2 standard deviations from pretreatment, at posttreatment were 90.5% for CBT + DZ, 83% for CBT + PBO, 86% for CBT, 68% for DZ, and 37% for PBO. All active treatments were superior to PBO, but did not differ from one another. Percent responders at 6-month follow-up were 71% for CBT + DZ, 67% for CBT + PBO, 71% for CBT, and 41% for DZ; chi-squared analysis revealed the superiority of treatments that included CBT treatments over medication (DZ) alone.

Generalized Anxiety Disorder is primarily excessive worry in one or more areas of one's life. Youth with GAD are often viewed by others as “little adults” because their worries may focus on keeping schedules, family finances, the environment, health issues, relationships, and perfectionism—themes that are more typically concerns for adults. Youth with GAD may not be disruptive or acting-out in their behavior, so their difficulties may go unnoticed by parents, family, and teachers. Nevertheless, their internal distress interferes with their overall functioning.

Preliminary data evidences developmental differences in symptom report between older and younger children with GAD (formerly overanxious disorder [OAD]), and it seems that with age, children's manifestation of GAD changes in both content and in symptom number (Kendall & Pimentel, 2003; Strauss, Lease, Last, & Francis, 1988). There also appear to be differences across GAD symptoms with respect to their efficiency in yielding a DSM-IV diagnosis (Pina, Silverman, Alfano, & Saavedra, 2002). In addition, older children, over the age of 12, present with a higher number of overall symptoms, as well as more anxiety about past events. This increase in symptom report may be somewhat GAD-specific, rather than a mere function of increased age, as findings with youth with SAD often reveal that younger children report more symptoms than do older children (Francis, Last, & Strauss, 1987).

Historically, a wide range of treatments has been used to treat youth with GAD, including various psychological approaches (e.g., behavioral, cognitive-behavioral, psychodynamic, family, play) as well as medications. Only a very small number of treatments have met the criteria for and have been assigned the label of “empirically supported.” With regard to GAD, much of this research has been with children, not with adolescents. Our descriptions here of the treatment of GAD in youth will place an emphasis on the treatments that have received empirical support.

The results of several Type 1 randomized clinical trials using CBT, conducted by different research groups, support the application of this ap

proach (Barrett, Rapee, Dadds, & Ryan, 1996; Flannery-Schroeder & Kendall, 2000; Kendall, 1994; Kendall et al., 1997; Mendlowitz et al., 1999). Evidence to date suggests that CBT for children and adolescents is effective compared to no-treatment control conditions (e.g., Barrett, Dadds, & Rapee, 1996; Kendall, 1994; Kendall et al., 1997) and that CBT for youth with anxiety is a “probably efficacious” treatment (Chambless & Hollon, 1998).

An initial Type 1 randomized clinical trial evaluated a CBT protocol with 8-to 13-year-old children (Kendall, 1994). The manual-based CBT intervention (called the “Coping Cat”) targets one or more of the following disorders; SAD, social phobia, and GAD. A large percentage of the cases in the report were highly comorbid and many had GAD as the primary disorder. Cognitive-behavioral therapy addresses the cognitive biases associated with anxiety through psychoeducation, cognitive restructuring to change self-talk, relaxation training, guided imagery, problem solving, and numerous graded exposure tasks. In this RCT, 47 children ages 8 to 13 years were randomized to either CBT or a wait-list condition. Children in the CBT condition demonstrated significant improvement from pre-to posttreatment on self-reported distress, parent-reported distress of child, and, importantly, diagnostic status. Specifically, 66% of treated children no longer met criteria for their principal anxiety diagnosis following treatment.

In a second Type 1 randomized clinical trial (Kendall et al., 1997), 94 children aged 9 to 13 years were randomized to the CBT protocol or a wait-list condition. Over 50% of treated youth were free of their principal diagnosis at posttreatment, with significant reductions in disorder severity even for the youth remaining symptomatic. Other controlled trials (e.g., Barrett, 1998) support the efficacy of CBT in childhood anxiety (for review see Kazdin & Weisz, 1998; Ollendick & King, 2000) and CBT protocols have also been adapted to the group format (Flannery-Schroeder & Kendall, 2000; Manassis et al., 2002; Silverman et al., 1999a).

When considering the treatment of adolescents with GAD, one must take into account that the literature on the treatment of “child and ad-olescent” anxiety is based primarily on studies that have evaluated treatments for youth ages 7 to 14, with a mean age around 10.6 (e.g., Barrett, Dadds et al., 1996; Kendall, 1994; Kendall et al., 1997). Moreover, in one of these studies (Barrett, Dadds, et al., 1996), the treatment effects were not the same across age ranges, showing that a combined CBT and family intervention was significantly more effective than CBT alone for 7-to 10-year-old children, but not for 11-to 14-year-old children. This finding suggests that older children might respond differently than younger children to the same treatment. More research on this important topic is sorely needed.

Most of the CBT treatments that have been empirically supported for children have substantial room for improvement when applied to adolescents. First, we cannot assume that previous findings from children ages 7 to 14 automatically apply to all adolescents. Second, adolescence is a transitional period, characterized by more biological, physical, psychological, and social role changes than any other stage except infancy (Kendall & Williams, 1986). However, these developmental differences among children and adolescents have been largely ignored in treatment outcome studies of anxious youth. Holmbeck and colleagues (2000) note, “although many authors advocate for adaptations of treatment manuals in ways that take development into account, few provide methods for doing so” (Holmbeck et al., 2000, p. 339). Similarly, Weisz and Hawley (2002) assert that one of the long-term goals for the field of treatment outcome research should be to “develop an array of developmentally tailored treatments that are effective with clinically referred teens” (p. 21). Because treatments have not fully addressed the developmental changes of adolescence, one could argue that the impact of the treatment on the adolescent's anxiety disorder has not yet been examined empirically.

To date, no controlled studies have investigated CBT for adolescent GAD. It is important for such studies to be conducted because GAD is prevalent in adolescents (see Merikangas & Avenevoli, 2002).

There are limited data on pharmacotherapy of GAD, and anxiety disorders in youth in general, with the exception of OCD (March, Biederman, et al., 1998). This situation exists despite the fact that there has been an increase in the use of psychotropic medications in child and adolescent samples (Pincus et al., 1998; Zito et al., 2000). There are few adequately designed studies available that establish the safety and efficacy of any class of medication for childhood GAD (Allen, Leonard, & Swedo, 1995; Bernstein, Borchardt, & Perwein, 1996). Medication options for the treatment of GAD in children and adolescents would appear to include the benzodiazepines, the nonbenzodiazepine anxiolytic buspirone, the SSRIs, and the TCAs. These medications have efficacy in adult GAD (Rickels, Downing, Schweizer, & Hassman, 1993) and venlafaxine has also been shown to be effective for adult GAD (Derivan, Entsuah, Haskins, Rudolph, & Aguiar, 1997; Rickels, Pollack, Sheehan, & Haskins, 2000). However, at this juncture there are not adequate data with regard to the use of medications with adolescents.

Although there exists a literature on the effectiveness of benzodiazepines in adult anxiety disorders (Rickels et al., 1983; Rickels, Schweizer, Case, & Greenblatt, 1990), there have been only a few studies, with small sample sizes, to examine the use of benzodiazepines in childhood anxiety disorders (Biederman, 1987; Graae, Milner, Rizzotto, & Klein, 1994; Simeon et al., 1992). Most of the studies involve children and adolescents with additional comorbidities such as major depression, panic disorder, and school refusal (Bernstein, Garfinkel, & Borchardt, 1990; Kutcher & Mackenzie, 1988). The results of these studies are mixed, with some finding positive results and others finding limited efficacy for benzodiazepines. Because of the inconsistency in results, reports of potentially serious side effects (Rickels et al., 1990, 1993), and risk of dependency and withdrawal, the benzodiazepines cannot be considered a front-line treatment.

Although buspirone has been shown to have both anxiolytic and antidepressants affects in adults (Rickels et al., 1990, 1991), it does not appear to be a highly effective or broad-spectrum anxiolytic in youth (Pohl, Balon, Yergani, & Gershon, 1989; Sheehan, Raj, Sheehan, & Soto, 1990).

Most of the clinical trials in which TCAs were used did not have children and adolescents with GAD only, but rather the much more complicated comorbid group of “school-refusing” children. The several Type 1 placebo-controlled studies of TCAs for children with anxiety-based school refusal provide conflicting results (Berney et al., 1991; Bernstein et al., 1990; Gittelman-Klein & Klein, 1973; Klein, Manuzza, Chapman, & Fyer, 1992). None of these studies were designed to examine children with the primary diagnosis of GAD, so no conclusions can be drawn as to the efficacy of TCAs for this anxiety disorder. Another issue of concern with TCA use in children is associated with a growing recognition of cardiac risk (Popper & Ziminitzky, 1995; Riddle, Geller, & Ryan, 1993; Riddle, Nelson, et al., 1991; Varley & McClellan, 1997). Given the uncertain clinical efficacy of TCAs for anxiety-disordered children (including GAD) plus the significant side effects, this class of medication is not a first choice.

The SSRI antidepressants are potential candidates for the treatment of childhood GAD. The safety of the SSRIs recommends them, as does their effectiveness in treating depression, which may be comorbid with childhood GAD. The SSRIs have shown preliminary efficacy in the treatment of adult GAD (Pohl, Wolkow, & Clary, 1998) and there is some evidence to suggest the use of SSRIs to treat childhood GAD. Unfortunately, the majority of studies include patients with GAD, SAD, social phobia, selective mutism, and anxiety disorder not otherwise specified. In Type 3 open trials, fluoxetine has shown some preliminary benefit for overanxious disorder, SAD, and social phobia (Birmaher et al., 1994; Fairbanks et al., 1997; Manassis & Bradley, 1994). In a Type 1 study, Rynn, Siqueland, and Rickels (2001) randomized (n = 22) children meeting the diagnosis of GAD to receive either placebo or 50 mg (maximum dose) of sertraline for 9 weeks, with positive results. The main side effects of

the sertraline-treated children, as compared to the placebo-treated children, were dry mouth, drowsiness, leg spasms, and restlessness. The RUPP study (2001), mentioned earlier, also provides supportive evidence.

There have been previous reports of activation or agitation with the SSRIs, and it appears to be dose related (Apter et al., 1994; Riddle, Harin, & King, 1990; Riddle, King, et al., 1991). With these recent positive studies, it appears that SSRIs are emerging as the first line medication treatment for childhood anxiety disorders. Because of withdrawal symptoms being reported with the discontinuation of SSRIs, including nausea, headache, dizziness, and agitation (Labellarte, Walkup, & Riddle, 1998), these medications should not be abruptly discontinued.

The combination of medications and CBT for the treatment of GAD has several potential merits and a few potential enigmas. First, the GAD youth who is nonresponsive to one of the approaches may be responsive to the other or to their combination. Second, on the optimistic side, it may be that the combined treatment effect is additive—medications reduce initial distress and prepare the client for a more active and involved participation in CBT treatment. A less optimistic view of the combined treatment is that a medication may detract from the efficacy of CBT. That is, if exposure to the situation and the facing of the unwanted emotional distress in that situation contribute to the efficacy of CBT, then it is possible that the medication-produced nonanxious condition will prevent the CBT participant from experiencing the anxiety during exposure to in vivo tasks. The effectiveness of the medications could undermine one of the thought-to-be-active aspects of CBT (habituation to anxiety in the situation). Clearly, research on this question is very much needed.

Indirect information on psychosocial treatment comes from a Type 1 controlled study of behavior therapy combined with either imipramine or a placebo in 45 6-to 14-year-olds with school refusal specifically due to separation anxiety (Gittelman-Klein & Klein, 1971, 1973). To enter the drug study, children had to fail 4 weeks of vigorous behavioral therapy (Gittelman-Klein, 1975)—half (50%) of the children responded. With another 6 weeks of behavior therapy, now combined with a placebo, another 20% improved; thus 70% of severely separation-anxious children derived marked benefit from 10 weeks of behavior therapy. The comprehensive behavioral treatment in the Gittelman-Klein report had beneficial impact, but its precise efficacy is unknown because there were no comparison conditions. A Type 3 naturalistic study of school refusers reported superior efficacy of behavior therapy compared to inpatient care and to home schooling (Blagg & Yule, 1984). The lack of randomization weakens the conclusions.

Type 1 studies of CBT for children with school refusal behavior have also been conducted (King, Leonard, & March, 1998; Last, Hansen, & Franco, 1998). King et al. (1998) found that 4 weeks of CBT was superior to a wait-list control. School attendance was higher in the treated children than in the untreated children (89% vs. 60%), and ratings of anxiety also showed meaningful differences. In contrast, Last et al. (1998) found no significant difference in the improvement rate between CBT and standardized educational support treatment: mean percentages of

school attendance were 67% and 60%, respectively. The more stringent criterion of 95% school attendance was met by 22% of the CBT group and 21% of those receiving educational support. In both treatments, a high rate of children judged themselves to be improved, as did parents and clinicians (90% to 100%), but considerably less were considered well by independent raters.

Other studies of CBT have not focused solely on SAD but have included youth with this disorder in addition to other anxiety disorders. These studies have reported efficacy in treating SAD (e.g., Kendall, 1994; Kendall et al., 1997). Another Type 1 study (Silverman et al., 1999a) focused mostly on specific fears, the majority (67%) consisting of anxiety at bedtime requiring the presence of a parent, a cardinal symptom of separation anxiety. Following 10 weeks, children showed similar patterns of improvement with CBT, behavior therapy focusing on training parents in contingency management, or educational support.

Some investigators have examined whether parental involvement in treatment contributes to childrens' improvement. Of all the childhood anxiety disorders, SAD can be said to have the most direct impact on the family. Consequently, there is a strong rationale for parental involvement in treatment. A study of anxious school refusers reported significant superiority for the condition involving parents and teachers in the children's treatment, compared to treatment without this component (Heyne et al., 2002), but the advantage was not sustained over a 5-month follow-up. In a recent Type 1 study, group CBT was compared to a group family anxiety management program and to a wait-list control for youth with mixed anxiety disorders, 43% of whom had separation anxiety (Barrett, 1998). The treatments were superior to the wait-list condition, but did not differ from each other on most outcome measures.

The first use of a pharmacological agent, imipramine, in SAD derived from a postulated re lationship between this childhood condition and adult panic disorder (Klein, 1964; Klein & Fink, 1962). Marked clinical benefit for imipramine (mean 125 mg/day) versus a placebo was reported in a Type 1 study of 45 children with school phobia and separation anxiety (Gittelman-Klein & Klein, 1971). A smaller study (N = 20) failed to replicate earlier positive findings (Klein, Koplewicz, & Kanner, 1992). These two treatment studies targeted SAD specifically. Another study failed to observe a significant advantage in school-phobic children treated with relatively low doses of clomipramine (75 mg/day) (Berney et al., 1981). More recently, a Type 1 controlled trial in adolescent school refusers reported efficacy for imipramine (Bernstein et al., 2000), but interpretation is complicated by the comorbidity of major depression. The well-documented cardiotoxic effects of high doses of TCAs have limited their usefulness, especially since SSRIs do not present such a risk.

Clinical reports have claimed benefit from high-potency benzodiazepines (i.e., clonazepam) in mixed anxiety, but the only Type 1 controlled trial did not confirm these clinical observations (Graae et al., 1994). The design of this trial was less than rigorous, however. Benzodiazepines can induce behavioral disinhibition in children, which limits their application.

As noted earlier, a Type 1 multisite, placebo-controlled study (N = 128) of fluvoxamine (up to 300 mg/day) in children with separation anxiety, social anxiety, and generalized anxiety disorders reported a marked advantage for the medication (RUPP, 2001) with an overall improvement rate of 78% for fluvoxamine vs. 29% for placebo. There were no interactions for diagnosis and treatment outcome, except for the lesser response rate among patients with social phobia than that among those with other diagnoses (RUPP, 2003). Therefore, it may be surmised that SAD responded to fluvoxamine.

At this time, SSRIs may be the first-line pharmacological treatment for childhood SAD, but we are in need of information specifically regarding adolescents. On the whole, the SSRIs are well tolerated and have a favorable side effect profile, and some have demonstrated efficacy.