Preliminary Reliability and Validity of the Generalized

BEHAVIOR THERAPY

33,215-233,2002

Preliminary Reliability and Validity of the Generalized Anxiety Disorder Questionnaire-IV: A Revised Self-Report Diagnostic Measure of Generalized Anxiety Disorder MICHELLE G. NEWMAN A N D R E A R . ZUELLIG KEVIN E . KACHIN MICHAEL J. CONSTANTINO AMY PRZEWORSKI THANE ERICKSON LAURIE CASHMAN-MCGRATH

The Pennsylvania State University This study examined the Generalized Anxiety Disorder Questionnaire-IV (GAD-QIV), a revised self-report diagnostic measure of generalized anxiety disorder (GAD) based on the fourth edition of the Diagnostic and Statistical Manual. GAD-Q-IV diagnoses were compared to structured interview diagnoses of individuals with GAD, social phobia, panic disorder, and nonanxious controls. Using Receiver Operating Characteristics analyses, the GAD-Q-IV showed 89% specificity and 83% sensitivity. The GAD-Q-IV also demonstrated test-retest reliability, convergent and discriminant validity, and kappa agreement of .67 with a structured interview. Students diagnosed with GAD by the GAD-Q-IV were not significantly different on two measures than a GAD community sample, but both groups had significantly higher scores than students identified as not meeting criteria for GAD, demonstrating clinical validity of the GAD-Q-IV.

Generalized anxiety disorder (GAD) was first established as a diagnostic entry in the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III; American Psychiatric Association, 1980), when it was introduced as a residual category. It was distinguished as a full-fledged anxiety We would like to thank Louis G. Castonguay, Ph.D., for his helpful comments on an earlier draft of this manuscript. Preparation of this article was supported in part by National Institute of Mental Health Research Grant MH-58593. Address correspondence to Michelle G. Newman, Ph.D., Department of Psychology, Penn State University, 310 Moore Bldg., University Park, PA 16802-3103; e-maiD: mgnl @psu.edu. 215 005-7894/02/0215~)23351.00/0 Copyright 2002 by Associationfor Advancementof BehaviorTherapy All rights for reproductionin any formreserved.

216

N E W M A N ET A L .

•disorder in the revised third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R; American Psychiatric Association, 1987), and was further defined in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DMS-IV; American Psychiatric Association, 1994), in which it underwent many changes in symptom criteria. Thus, GAD is a relatively new diagnosis that has undergone some recent revisions. Research on the etiology, symptomatology, and treatment of GAD has increased substantially in the past several years. In conjunction with this research, a number of reliable and valid self-report measures of GAD-related features were developed. These measures include the Meta-Cognitions Questionnaire (MCQ; Cartwright-Hatton & Wells, 1997), the Intolerance of Uncertainty Scale (IU; Freeston, Rheaume, Letarte, Dugas, & Ladouceur, 1994), the Why Worry Scale (WW; Freeston et al., 1994), the Consequences of Worrying Questionnaire (CWQ; Davey, Tallis, & Capuzzo, 1996), the Reactions to Relaxation and Arousal Questionnaire (RRAQ; Heide & Borkovec, 1983), and the Penn State Worry Questionnaire (PSWQ; Meyer, Miller, Metzger, & Borkovec, 1990). The majority of these measures, however, were designed to assess aspects of functioning important to the development or maintenance of GAD rather than to directly assess G A D symptomatology. For example, the MCQ measures dimensions of positive and negative beliefs about worry (Cartwright-Hatton & Wells, 1997). In contrast, the IU measures emotional and behavioral reactions to ambiguous situations, the implications of being uncertain, and attempts to control the future. Further, whereas the WW assesses reasons why people worry, the CWQ assesses consequences associated with worrying. In addition, the RRAQ assesses a fear of relaxation. The PSWQ, on the other hand, measures the traitlike tendency to worry, a core GAD symptom. Although the PSWQ is often used as a screening device for high worriers and people who may have GAD, its failure to assess the full range of DSM diagnostic criteria precludes it from being considered a diagnostic measure. For example, this measure does not assess whether the worry is uncontrollable, the range of physical symptoms associated with worry, or the degree of impairment or distress from worry. In fact, the only published self-report measure that captures the full range of GAD symptoms is the Generalized Anxiety Disorder Questionnaire (GAD-Q; Roemer, Borkovec, Posa, & Borkovec, 1995). The GAD-Q was developed for use as an initial screening device to identify individuals with GAD. Using an undergraduate student sample, Roemer and colleagues (1995) found that it had a high rate of agreement with the Anxiety Disorders Interview Schedule Revised (Di Nardo & Barlow, 1988). Further, this study found that undergraduate students who met GAD criteria using this measure did not differ significantly from a clinical community sample on levels of questionnaire-assessed anxiety measured by the PSWQ and on the State Trait Anxiety Inventory (STAI; Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983). Additional studies using the GAD-Q to select GAD participants have found that such individuals had higher PSWQ scores,

RELIABILITY AND VALIDITY OF THE GAD-Q-IV

217

greater daily interference from worry (Roemer, Borkovec, Posa, & Lyonfields, 1991), significantly different attachment styles (Cassidy, 1992), and greater autonomic rigidity (Lyonfields, Borkovec, & Thayer, 1995) than nonanxious controls. In addition, studies have shown that individuals diagnosed with GAD based on the GAD-Q had deficiencies in parasympathetic tone (Lyonfields et al., 1995) and problematic attachment styles (Cassidy, 1992) similar to treatment-seeking GAD clients, and that such clients were not significantly different in their worry frequency and content (Roemer et al., 1991; Shadick, Roemer, Hopkins, & B orkovec, 1991) than treatment-seeking GAD clients. Despite its past utility, the GAD-Q was based on DSM-III-R symptomatology and a revised version was dictated by symptom changes described in DSM-IV. Whereas DSM-III-R required pathological worry to be unrealistic or excessive, DSM-IV requires such worry to be excessive and uncontrollable. In addition, whereas for DSM-III-R the person had to worry about two or more life circumstances, for DSM-IV the individual must worry about a variety of events or activities. Moreover, whereas DSM-III-R allowed for the presence of 6 out of 18 possible symptoms, for DSM-IV an individual must have at least 3 of 6 primary symptoms. This latter change in symptomatology reflects findings that people with GAD experience more symptoms of central nervous system hyperactivity, such as those in the motor tension and vigilance and scanning clusters, than symptoms triggered by the autonomic system (Anderson, Noyes, & Crowe, 1984; Marten et al., 1993; Noyes et al., 1992). Finally, whereas DSM-III-R required that the symptoms were "often present when anxious;' DSM-IV requires that the individual experience the symptoms "more days than not." As a result of the revisions in DSM criteria, the Generalized Anxiety Disorder Questionnaire-IV (GAD-Q-IV) was developed. This article presents the preliminary psychometric properties of the GADQ-IV. A reliable diagnostic self-report measure for GAD is particularly important because research has found that GAD is one of the least reliably diagnosed anxiety disorders (Brown, Di Nardo, Lehman, & Campbell, 2001; Di Nardo, Moras, Barlow, Rapee, & Brown, 1993). As a result, Borkovec (Borkovec & Newman, 1998; Borkovec & Whisman, 1996) has recommended the use of two diagnostic interviews to determine the presence of GAD. However, these interviews are often time-consuming and costly. Thus, a self-report measure for GAD, which could be used as an initial screening device, could cut research costs and time considerably by requiring only one follow-up structured interview. In addition, the use of such a measure as an initial assessment device might screen out individuals who would have been ruled out as a result of the more costly structured interview. Moreover, such a measure could be used to quickly select undergraduates meeting GAD symptom criteria for experimental studies before attempting to replicate findings in the more difficult-to-obtain clinical samples. To help simplify the use of this measure as a screening device, a scoring system was developed for the GAD-Q-IV, allowing for attainment of a total score. In addition, in contrast to the study by Roemer and colleagues (1995), the

218

NEWMAN ET AL.

current study examined the stability of this measure across a 2-week period and the ability of the GAD-Q-IV to discriminate GAD from other anxiety groups. In particular, the current study examined the ability of the GAD-Q-IV to discriminate the presence of GAD from panic disorder and social phobia. This comparison was important, as panic disorder and social phobia are common comorbid diagnoses with GAD, in addition to sharing symptoms with GAD.

Development of the GAD-Q-IV Initial construction of the GAD-Q-IV used the GAD-Q as a starting point, which took a form similar to that used in the ADIS-R (Di Nardo & Barlow, 1988) section for GAD. Thus, this self-report pencil-and-paper instrument includes a yes/no checklist of whether the participant has experienced excessive and unrealistic worry for at least 6 months, a section where the participant writes his or her worry topics, and a checklist of the 18 symptoms. A final question requires participants to rate the degree of interference resulting from their worry on an 8-point Likert scale (0 = none, 4 = moderately, and 8 = very severely). In revising the GAD-Q to create the GAD-Q-IV, we eliminated questions referring to unrealistic worry and instead included items assessing the occurrence of excessive and uncontrollable worry. In addition, we replaced the checklist of 18 DSM-III-R symptoms with the 6 D S M - I V symptoms. We also added a question assessing the degree to which the participant finds worry and the associated symptoms distressing. For increased clarity related to the interference question, we added intermediate intensity descriptions to the 8-point Likert scale such that 0 = none, 2 = mild, 4 = moderate, 6 = severe, and 8 = very severe. We also added a question to assess degree of distress from worry using the same scale. The goal of the present studies was to examine the psychometric properties of this new instrument. In Study 1, we examined the sensitivity and specificity of the GAD-Q-IV through comparison of GAD-Q-IV diagnosis to D S M - I V diagnosis using either the Anxiety Disorders Interview Schedule for DSM-IV, Lifetime Version (ADIS-IV-L; Di Nardo, Brown, & Barlow, 1994), or the Anxiety Disorders Interview Schedule for DSM-IV (ADIS-IV; Brown, Di Nardo, & Barlow, 1994). In Study 2, we examined concurrent validity, and in Study 3 we examined the stability of the measure across time, convergent and discriminant validity, and the clinical relevance of a group of student participants identified as having GAD by the GAD-Q-IV.

Study 1 : Sensitivity, Specificity, and Ability to Discriminate Known Groups Method Participants

Participants were 143 undergraduates recruited as part of two separate assessment studies. Of the 143 participants, 90 were individuals interested in


219

being assessed for an anxiety disorder with the potential of being referred for free treatment as part of separate treatment studies taking place at the Center for Treatment of Anxiety and Depression at The Pennsylvania State University. Thirty of the 143 participants met DSM-IV criteria for GAD as a primary or secondary diagnosis, Of those who did not meet criteria for GAD, 18 had primary panic disorder, 42 had primary social phobia, and 53 did not meet criteria for any anxiety diagnoses. Additional comorbid diagnoses of the 30 GAD participants included 1 with comorbid obsessivecompulsive disorder (OCD; 3%), 4 with major depression (13%), 8 with dysthymia (27%), 2 with posttraumatic stress disorder (PTSD; 7%), 10 with specific phobia (33%), 12 with panic disorder (40%), and 20 (67%) with comorbid social phobia. Within the primary social phobia group, none had OCD, 1 had panic disorder (2%), 2 had major depression (5%), 3 had PTSD (7%), 6 had specific phobia (14%), and 1 had dysthymia (2%). Within the panic disorder cohort, none had OCD or dysthymia, 4 had comorbid social phobia (22%), 3 had specific phobia (17%), 1 had PTSD (6%), and 3 had major depression (17%). One hundred thirteen (80%) participants were female and 29 (20%) were male. Six (4%) were African American, 9 (6%) were Asian American, 6 (4%) were Hispanic, 111 (78%) were Caucasian, and 10 (7%) identified themselves as other. The average age was 21, with a range of 18 to 41. Chi-square analyses showed no significant differences between the GAD and comparison groups on ethnicity or gender distribution. An analysis of variance (ANOVA) also found no differences in mean age. Measures The Panic Disorder Self-Report. The PDSR (Newman, Zuellig, Kachin, & Constantino, 1998) is a 22-item self-report measure designed to diagnose panic disorder based on DSM-IV criteria. The PDSR has demonstrated good test-retest reliability in a college sample. Both convergent and discriminant validity have also been demonstrated. The validity of the PDSR was supported by comparisons between PDSR diagnoses and clinician-based ADISIV-L diagnoses of individuals diagnosed with panic disorder, GAD, social phobia, and a nonanxious comparison group. The PDSR showed a 100% specificity and 89% sensitivity. Diagnoses made by the PDSR yielded a 0% false positive rate and a 11% false negative rate. This questionnaire was used as an initial screening measure to select individuals who might meet criteria for panic disorder and controls. The Social Phobia Questionnaire (SPQ; Newman, Kachin, Zuellig, & Constantino, 1997; Newman, Kachin, Zuellig, Constantino, & Cashman, 2001). The SPQ is a 10-item self-report measure designed to diagnose social phobia based on DSM-IV criteria. The validity of the SPQ was supported by comparisons between SPQ diagnoses and clinician-based structured interview diagnoses of individuals meeting criteria for social phobia, panic disorder, and a nonanxious comparison group. The SPQ showed a specificity of 82% and a sensitivity of 85%. Diagnoses made by SPQ yielded an 18% false positive rate

220

N E W M A N ET AL.

and a 15% false negative rate. This questionnaire was used as an initial screening measure to select individuals who might meet criteria for social phobia and controls. Procedure Participants received the GAD-Q-IV (see Appendix) as part of a battery of self-report measures administered during a mass-testing situation that also included the PDSR and the SPQ. Data were pooled from two separate studies; however, both studies took place during the same time period. Fifty-nine of the participants (20 nonanxious controls, 18 panic disordered participants, and 21 of the GAD participants) were recruited as part of an ongoing study examining childhood diagnoses in adult anxiety disorder clients (Zuellig & Newman, 1996). In this sample, individuals who endorsed several symptoms for GAD on the GAD-Q-IV, or who endorsed several symptoms of panic disorder on the PDSR, or who appeared not to meet criteria for any anxiety disorder were invited to participate in a face-to-face structured interview. The administration of the GAD-Q-IV and the interview always took place at least 10 days apart. The 59 participants were interviewed by a trained doctoral student using the ADIS-IV-L. All interviews were audiotaped. A randomly selected sample of taped interviews (45% of entire sample; 45% of GAD sample; 45% of panic disordered participants; 45% of controls) was assessed by a second trained, blind assessor. Interrater reliability revealed a 94% overall concordance rate for GAD, panic disorder, or an absence of these diagnoses. Given the high diagnostic concordance between the first interviewer and the second, blind rater, the diagnosis of the first interviewer was used to determine concordance estimates between the GAD-Q-IV and ADIS-IV-L. A recent study found that the ADIS-IV-L had good reliability (kappa = .65) in the diagnosis of primary or secondary GAD (Brown et al., 200l). The other 83 participants were recruited for a study examining interpersonal differences between social phobia subgroups (Kachin, Newman, & Pincus, 2001). In this study, participants who either endorsed several symptoms for social phobia on the SPQ or who appeared not to meet social phobia criteria were invited to participate in a structured interview. As with the first study, the administration of the GAD-Q-IV and the interview always took place at least 10 days apart. These participants (33 nonanxious controls, 42 socially phobic participants, and 8 GAD participants) were interviewed twice by independent assessors using the ADIS-IV. The first assessor served as a screener and any individuals not meeting criteria for social phobia (except those who demonstrated a clear absence of social phobia) were not invited back for a second interview. The second interviewer was uninformed of the diagnostic assignment of the first interviewer and was encouraged to use his own judgment about whether the individual met diagnostic criteria for any disorder. In cases where initial disagreement occurred (13 % of the cases), the two assessors met and diagnosis was determined by consensus. If consensus


221

could not be reached, the participant was excluded from the study. Reliability of the GAD-Q-IV was examined by comparing GAD-Q-IV criteria to the ADIS-IV consensus diagnosis. Although two different structured interviews were used in these studies, it is important to note that the interviews are identical in terms of assessment of current DSM-IV criteria. The only difference between the interviews is that the ADIS-IV-L queries about diagnoses throughout the individual's life whereas the ADIS-IV focuses more centrally on current diagnoses. The GAD-Q-IV was scored by using a sum total response. This scoring system was devised in an attempt to create a score that would best reflect the DSM-IV diagnostic threshold for the GAD diagnosis. To create a total score, all yes answers were coded as 1 and all no answers as 0 (e.g., Item 1: Do you experience excessive worry? Item 2: Is your worry excessive in intensity, frequency, or amount of distress it causes? Item 3: Do you find it difficult to control your worry? Item 4: Do you worry about minor things? and Item 6: Did you worry more days than not over the prior 6 months?). In addition, for Item 5, which asks for a list of most frequent worry topics, individuals were given 1 point for each topic listed up to 6, and this total was divided by 3. Similarly, for Item 7, participants were given 1 point for each physical symptom they experienced up to 6, and this total was divided by 3. Finally, the numbers circled for Items 8 and 9 (i.e., degree of distress and interference) were each divided by 4, and these numbers were added together. Because the questionnaire requests that individuals skip the remaining items when they do not endorse initial criteria, such skipped items were scored as 0. Total scores ranged from 0 to 12. Our justification for this scoring system was that it allowed for a maximum of 7 points for the first six items, which assessed the core GAD symptoms of excessive and uncontrollable worry. In addition, it allowed for 1 point for everyone who met the GAD diagnostic requirement for items related to associated symptoms (i.e., at least three physical symptoms, at least moderate severity, at least moderate distress). At the same time, someone who endorsed the highest possible score on items reflecting associated symptoms could earn a maximum of 2 points, and people who were subthreshold on these items could still receive a weighted score that varied dependent on how far they were below threshold. Results Chi-square analyses indicated no differences between the samples of the two studies in ethnicity or gender. ANOVAs indicated no difference between the studies in age of participants, in total GAD-Q-IV scores of persons diagnosed with GAD (i.e., M = 19.95, SD = 8.58 vs. M = 19.85, SD = 5.58), or in total GAD-Q-IV scores of individuals not meeting criteria for an anxiety disorder (i.e., M = 1.90, SD = 6.15 vs. M = 2.69, SD = 3.14). As a result, the data sets of the two studies were combined and analyzed jointly. In order to determine the ability of the GAD-Q-IV, to discriminate people

222

N E W M A N ET AL°

n 0.9,

a0

I"

0.8.

o

o o,

0,7

o o•

"ff .~_ 0 . 6

° o #"

8

°q

o.5

ij P

-

0#

- - D - - No discrimination GAD-Q-IV

t~

0.4

iO iI

0,3

¢0 jo i0

0.2 i I

iJ

0.1

i 0 If

0

~

0

0.2

0.4

0.6

0.8

1

1 - Specificity (false positives)

FIG. 1. R e c e i v e r O p e r a t i n g Characteristic curve o f the s e n s i t i v i t y and specificity o f the GAD-Q-IV.

with GAD from panic disordered, socially phobic, and control participants, Receiver Operating Characteristics (ROCs) of this measure were analyzed comparing the GAD-Q-IV to ADIS-IV diagnosis. All participants with anxiety disorders other than GAD as well as controls were grouped together and compared to individuals with either primary or secondary GAD. Figure 1 shows the resulting ROC curve. The area under the curve was .93 ( S E = .023), p < .001, suggesting that the probability is 93% that someone with GAD will have a higher score on the GAD-Q-IV than someone without GAD. The 95% confidence interval of the area under the curve ranged from .88 to .98.

Table 1 shows the sensitivity and specificity for the various cutoff points of the

RELIABILITYAND VALIDITYOF THE GAD-Q-IV

223

TABLE 1 SENSITMTYAND SPECIFICITYOF THE GAD-Q-IV

GADQ Score

Correctly Sensitivity Specificity Classified (%) (%) (%)

3.46 3.63 3.83 4.17 4.50 4.83 5.17 5.50 5.70 5.88 6.21 6.54 6.88 7.46 8.08 8.63 9.00 9.17

93 93 90 90 90 88 83 83 83 80 80 77 77 73 73 73 70 67

78 79 79 81 82 82 84 89 89 89 91 91 92 93 94 95 96 97

Positive Predictive Power (%)

Negative Predictive Power (%)

False Positive Rate (%)

False Negative Rate (%)

53 54 53 55 56 57 58 66 68 67 70 71 72 73 76 79 81 83

98 98 97 97 97 96 95 95 95 94 95 94 94 93 93 93 92 92

22 21 21 20 18 18 16 12 11 11 9 9 8 7 6 5 4 4

7 71 10 10 10 13 17 17 17 20 20 23 23 27 27 27 30 33

8t 82 81 83 84 83 84 87 88 87 88 89 89 89 90 90 90 90

Note. Numbers in boldface type represent values for the optimal cutoff point to select individuals with GAD using the GAD-Q-IV. GAD-Q-IV. The optimal balance between sensitivity and specificity is achieved with a cutoff point o f 5.7. This cutoff leads to a sensitivity o f 83% (25 o f 30) and a specificity o f 89% (101 o f 113). Thus, using this cutoff, the rate of false positive diagnoses b y the G A D - Q - I V was 11% and the rate o f false negative diagnoses assigned b y the G A D - Q - I V was 17%. K a p p a agreement between the A D I S - I V and the G A D - Q - I V was .67 with 88% o f participants correctly classified. If, on the other hand, a G A D - Q - I V user w o u l d prefer to be overinclusive rather than to miss a n y o n e w h o h a p p e n s to m e e t G A D criteria, a cutoff o f 4.5 y i e l d s a sensitivity o f 90% (27 o f 30) and a specificity o f 82% (93 o f 113) with a false positive rate o f 18% and a false negative rate o f 10%. In this case, k a p p a a g r e e m e n t with the A D I S - I V is .60.

Study 2: Concurrent Validity Method Participants Three hundred ninety-one undergraduate students (116 [30%] males and 275 [70%] f e m a l e s ) participated. A g e o f participants r a n g e d f r o m 17 to 30 years,

224

NEWMAN ET AL.

with a mean of 19. Ten (3%) of the participants defined themselves as African American, 16 (4%) as Asian American, 17 (4%) as Hispanic, 1 as Native American, and 347 (89%) as Caucasian. Individuals received class credit in exchange for their participation.

Procedure Participants attended a group questionnaire completion session where they filled out the battery of questionnaires listed below as well as the GAD-Q-IV. Questionnaires selected to assess discriminant validity were the Social Interaction Anxiety Scale (Mattick & Clarke, 1998) and the PTSD checklist (Weathers, Litz, Herman, Huska, & Keane, 1991). The questionnaire selected to assess convergent validity was the PSWQ. Measures Penn State Worry Questionnaire (PSWQ; Meyer et al., 1990). The PSWQ is a 16-item self-report measure of the frequency and intensity of worry. Factor analysis indicated that the PSWQ assesses a unidimensional construct with an internal consistency coefficient of .91 (Meyer et al.). High test-retest reliability was also demonstrated (Meyer et al.). The PSWQ has also been shown to distinguish individuals with GAD from each of the other anxiety disorder groups (Brown, Antony, & Barlow, 1992). Correlations between the PSWQ and measures of anxiety, depression, and emotional control supported the convergent and discriminant validity of the measure (Brown et al., 1992). In addition, this measure discriminated samples that (a) met all, some, or no DSM-III-R diagnostic criteria for GAD and (b) met criteria for GAD versus PTSD (Meyer et al.). The PSWQ has also demonstrated sensitivity to change in response to psychotherapy (Meyer et al.). The Social Interaction Anxiety Scale (SIAS; Mattick & Clarke, 1998). This 20-item scale assesses general fears of social interaction. The scale possesses high levels of internal consistency and test-retest reliability. It has also been shown to discriminate socially phobic participants from those with panic disorder with or without agoraphobia, simple phobia samples, and nonanxious samples (Mattick & Clarke; Peters, 2000). Further, the SIAS demonstrated strong convergent validity with self-report measures and with negative and positive thoughts in speech and conversation (Cox, Ross, Swinson, & Direnfeld, 1998; Ries et al., 1998). It also demonstrated discriminant validity when compared to established measures of depression, state and trait anxiety, and social desirability (Mattick & Clarke) and was shown to be sensitive to change arising from treatment (Cox et al., 1998; Mattick & Clarke; Ries et al.). PTSD Checklist (PCL; Weathers, Litz, et al., 1991). This scale was developed to assess PTSD symptoms in civilian populations and consists of 17 items that correspond to DSM-IV symptoms of PTSD. In a sample of Vietnam veterans, PCL scores demonstrated a coefficient alpha of .93 and convergent validity with other PTSD scales. The cutoff score for this scale has also dem-


225

onstrated a diagnostic sensitivity of .82 and a specificity of .83 (Weathers, Huska, & Keane, 1991). Interrater agreement with a structured interview for current PTSD was .74. Reliability and validity of this scale has also been demonstrated in patients with PTSD arising from treatment of breast cancer (Andrykowski, Cordova, Studts, & Miller, 1998) and from a motor vehicle accident (Blanchard, Jones-Alexander, Buckley, & Forneris, 1996).

Results Convergent and discriminant validity were examined using the 5.7 cutoff to determine presence versus absence of GAD and applying point biserial correlations. Results showed that the GAD-Q-IV was more highly correlated (r = .66) with the PSWQ than it was with the PCL (r = .45) or with the SIAS (r = .34). To determine whether the convergent validity correlation was significantly higher than the discriminant validity correlations, we applied the formula recommended by Cohen and Cohen (1975) to test the significance of the difference between two correlations from the same sample. Results showed that the correlation between the GAD-Q-IV and SIAS was significantly lower than the correlation between the GAD-Q-IV and PSWQ, t(389) = 7.65, p < .01. Similarly, the correlation between the GAD-Q-IV and PCL was lower than the correlation between the GAD-Q-IV and the PSWQ, t(389) = 5.56, p < .01.

Study 3: Test-Retest Reliability and Clinical Relevance of Groups Identified With the GAD-Q-IV Method Participants Two samples of participants were included in this study. The first sample consisted of 148 undergraduates (102 [69%] females and 46 [31%] males). Age ranged from 18 to 40 years, with a mean of 20. Four (3%) of the participants defined themselves as African American, 7 (5%) as Asian American, 6 (4%) as other, and 130 (88%) as Caucasian. These individuals received class credit in exchange for their participation. The second sample consisted of 69 clinical community participants from a prior GAD therapy outcome study (Borkovec, Newman, Pincus, & Lytle, 2002). These clients averaged 37 years of age (SD = 12), and duration of the GAD diagnosis averaged 13 years (SD = 12). Ethnicity was represented by 62 Caucasian, 2 African American, 3 Hispanic, and 2 Middle Eastern clients. Forty-five clients were women. Measures The Zung Self-Rating Depression Scale (SDS; Zung, 1965). This 20-item scale was developed to measure three basic aspects of depression: pervasive affect, physiological concomitants, and psychological concomitants. The test-retest reliability of the SDS was shown to be .61 in the current sample. The SDS has good known groups validity in distinguishing between depressed

226

N E W M A N ET A L .

and nondepressed samples and has good concurrent validity when compared with other depression measures (Zung, 1965). The SDS has also demonstrated sensitivity to clinical changes in depression (Rehm, 1981). The Panic Disorder Severity Scale (PDSS; Shear et al., 1997). This 7-item scale was developed for the overall assessment of panic disorder severity. Each item is rated on a 5-point Likert scale ranging from none to extreme. Internal consistency, convergent and discriminant validy, and sensitivity to change have been demonstrated (Shear et al.). Although this measure was originally designed as a clinician-administered device, it has subsequently exhibited sensitivity to change when used as a self-report measure (Penava, Otto, Maki, & Pollack, 1998). In the current sample, test-retest reliability was .84 over a 2-week period and discriminant validity was demonstrated with the PSWQ (r = .21) and SDS (r = .14). Reactions to Relaxation and Arousal Questionnaire (RRAQ; Heide & Borkovec, 1983). This is a 9-item, factor-analytically derived measure of fear of relaxation, a phenomenon common to persons with GAD. Each item is rated on a 5-point scale from not at all to very much so. An example of an item is, I'm apprehensive when I'm sitting around doing nothing. The maximum possible score that can be attained is 45 and the minimum is 0. Analyses of data from the current study showed that the RRAQ had a Cronbach's alpha of .85. Test-retest reliability of this scale was .83. This questionnaire also showed good convergent and discriminant validity, with significantly higher correlations with the PSWQ than with the SDS and the PDSS. The RRAQ has also been shown to discriminate persons with GAD from those without GAD and to be sensitive to change associated with psychotherapy for GAD (Borkovec & Inz, 1990). Procedure Undergraduate students attended two questionnaire completion sessions exactly 2 weeks apart where they completed the GAD-Q-IV, SDS, PDSS (Shear et al., 1997), RRAQ, and the PSWQ. Questionnaires selected to assess discriminant validity included the SDS and the PDSS. Questionnaires selected to assess convergent validity were the PSWQ and RRAQ. The clinical community sample was diagnosed based on a modified version of the ADIS-III-R (Di Nardo & Barlow, 1988), which included additional questions in the GAD section corresponding to two criteria being proposed at the time of study initiation by the DSM GAD subcommittee (i.e., uncontrollable worrying, and three out of six somatic symptoms). A second ADIS to reduce the likelihood of false positive cases was given within 2 weeks by the therapist who would see the client in therapy upon acceptance into the trial. Each of the four therapists ruled out one or more clients at this point (total = 9). Only clients for whom both interviewers achieved consensus on GAD diagnosis were included in the study. All but 2 of the 69 clients (97.1%) met both DMS-III-R and DSM-IV GAD criteria. Participants completed a battery of questionnaires, including the PSWQ and RRAQ, prior to receiving therapy.


227

Results

Test-retest reliability was examined by using the 5.7 cutoff score suggested by the ROC analysis. Undergraduates were categorized into those who did and those who did not meet GAD criteria at Times 1 and 2. Using Logistic Regression, we then examined the likelihood of remaining stable across time. Results showed that the GAD-Q-IV at Time 2 was reliably predicted by Time 1 score X2(1, N = 148) = 42.1 ,p < .001, and that 92% (136/148) of the sample showed stability across time in terms of their categorization. Odds ratios indicated that someone classified as meeting criteria for GAD at Time 1 is 50 times more likely to be classified as meeting criteria for GAD at Time 2 than someone not classified as meeting criteria for GAD at Time 1. Kappa agreement between Time 1 and Time 2 was .64. Using the 5.7 cutoff to determine presence versus absence of GAD, point biserial correlations were conducted to ascertain concurrent validity in the undergraduate sample. Based on the formula recommended by Cohen and Cohen (1975), results showed that the correlation between the GAD-Q-IV and PDSS (r = .30) was significantly lower than the correlation between the GAD-Q-IV and PSWQ (r = .55), t(145) = 2.27, p < .05, and was also lower than the correlation between the GAD-Q-IV and RRAQ (r = .58), t(145) = 2.60, p < .05. Moreover, the correlation between the GAD-Q-IV and SDS (r = .26) was significantly lower than the correlation between the GAD-Q-IV and PSWQ (r = .55), t(145) = 2.59, p < .05, and was lower than the correlation between the GAD-Q-IV and RRAQ (r = .58), t(145) = 3.25, p < .01. To test the clinical relevance of student participants who met or failed to meet GAD criteria based on the GAD-Q-IV (using the cutoff of 5.7), comparisons of PSWQ and RRAQ scores were made between GAD (n = 21) and non-GAD (n = 127) undergraduates in this study and pre-therapy scores of the 69 clinical community participants. Univariate analyses of variance showed significant main effects on both the PSWQ, F(2, 211) = 150.22,p < .001, and the RRAQ, F(2, 211) = 129.53, p < .001. Post hoc Bonfen'oni corrections indicated that whereas the GAD-Q-IV-identified non-GAD group differed significantly from both the student GAD ~ o u p and the clinical community cohort (all ps < .001), the GAD-Q-IV-identified student GAD group was not significantly different from the clinical community group on either of the measures (Table 2). General Discussion

These results provide preliminary evidence that the GAD-Q-IV may be an effective way to initially screen for the presence of diagnosable GAD and the absence of GAD. The false positive rates found in the current study were lower than what Roemer and colleagues (1995) found in their study of the reliability of the GAD-Q (i.e., 11% versus 20%). Further, the addition of groups of individuals diagnosed with panic disorder and with social phobia did not increase the rate of false positive diagnoses beyond what was found by Roemer et al. Moreover, the overall rate of agreement between the GAD-Q-IV and the ADIS-

228

NEWMAN ET AL. TABLE 2 MEANS AND STANDARDDEVIATIONS OF R R A Q AND PSWQ SCORESFOR CLINICAL COMMIJNrrY G A D CLIENTS, G A D - Q - I V IDENTIFIED G A D PARTICIPANTS,AND G A D - Q - I V IDENTIFIED NoN-GAD PARTICIPANTS

Measure RRAQ PSWQ Note.

GAD-Q-IV Selected Non-GAD (N = 127)

GAD-Q-IV Selected GAD (N = 21)

Clinical Community GAD (N = 69)

11.03 ~ (0.65) 40.78 a (0.98)

24.6 b (1.62) 63.55 b (2.44)

27.86 b (0.87) 67.88 b (1.31)

Differing superscripts across rows indicates significant differences between means with allps < .001.

IV was comparable to what has been reported by Brown et al. (2001) in their description of the GAD diagnostic agreement between two independent raters using the ADIS-IV (i.e., kappa = .65) with a mixed patient sample. One potential critique of using the 5.7 cutoff is that it allows for the possibility that participants can be selected as meeting criteria for GAD without endorsing all of the diagnostic criteria on the GAD-Q-IV. In fact, we initially scored the GAD-Q-IV by requiring participants to endorse all of the GAD criteria. However, even though this scoring scheme led to 96% specificity, sensitivity was only 67%, meaning that roughly a third of the participants who met GAD criteria were missed. Given that the GAD-Q-IV is meant to be a screening measure, it was as important not to miss people who met GAD criteria as it was to be sure that only people who met full criteria were identified. Thus, it seemed appropriate that we revisit our scoring scheme in an attempt to create a better balance between sensitivity and specificity. The 5.7 cutoff achieves this balance by producing a low false positive rate and a fairly low false negative rate. Nonetheless, if someone desired greater certainty that only persons with GAD were included in their sample, they might choose a higher cutoff (e.g., 9). In addition to showing good rates of agreement with an ADIS interview, diagnostic status based on the GAD-Q-IV was found to be stable over a 2week period. Moreover, the measure demonstrated good convergent and discriminant validity. Thus, there is preliminary evidence for use of this measure as an initial screening device. The use of this questionnaire may reduce the number of individuals ruled out following the more costly structured interview. Moreover, this questionnaire may be valuable as a means to quickly select undergraduates meeting GAD symptom criteria for experimental studies examining the processes related to GAD before attempting to replicate findings in more difficult-to-obtain clinical community samples. Given the high rates of comorbidity and substantial overlap between GAD, social phobia, and panic disorder (Barlow, 1988; Borkovec, Abel, & Newman, 1995; Newman, Castonguay, Borkovec, & Molnar, in press; Okasha et al., 1994; Roemer, Molina, & Borkovec, 1997; Sanderson & Barlow, 1990), it


229

is especially encouraging that the GAD-Q-IV successfully discriminated persons with GAD from persons with these conditions and with no mental disorder. It is also important to note that like most clinical samples, the GAD, panic disordered, and socially phobic students in the current study had a number of overlapping diagnoses. The fact that the GAD-Q-IV was able to distinguish those with primary or secondary GAD from those with mixed diagnoses other than GAD provides additional support for its validity. Nonetheless, it would be important for future studies to demonstrate the ability of the GAD-Q-IV to discriminate GAD from samples in which mood disorders, somatoform disorders, and PTSD are more highly represented than was the case with the undergraduate sample in this study. Even though prior research as well as the current study found that undergraduate participants selected with the GAD-Q or the GAD-Q-IV were similar to clinical community samples on a variety of externally valid criteria (Cassidy, 1992; Lyonfields et al., 1995; Roemer et al., 1991; Shadick et al., 1991), the sensitivity and specificity rates found here may not apply to community clinic samples that are likely to have higher rates of PTSD, mood, or somatoform disorders. A high rate of these disorders may make discrimination of GAD more difficult due to symptom overlap and because the GAD-Q-IV does not contain questions that would eliminate people whose symptoms occurred exclusively during the course of a mood disorder or PTSD. Several additional limitations should be noted. First, interrater agreement related to the ADIS-IV-L diagnosis of 59 participants in Study 1 was based on a second rater listening to an audiotape of the structured interview. A more stringent means to determine the accuracy of the diagnosis would have been to use a second independent rater as was done for the other 83 participants in Study 1. The study samples were also limited by a lack of ethnic diversity. Future research should attempt to replicate these findings with a more ethnically diverse sample. In addition, given the extensive psychometric data available on the PSWQ, future studies should compare the sensitivity and specificity of the GAD-Q-IV to the sensitivity and specificity of the PSWQ. Future studies should also determine whether the GAD-Q-IV discriminates GAD participants from groups of individuals with a specific phobia and OCD. It should also be noted that the GAD-Q-IV should be used only as a measure to determine presence or absence of GAD and should not be used to ascertain dimensional severity. The reason for this is that if someone responds negatively to the first three items, they are directed to discontinue responding. Therefore, by design the last three items are dependent on the first three items. Further, the skip-out creates a skewed distribution because a large number of participants cannot score above a certain number (those who skipped out), whereas other participants score very high because they hit the last items. Thus, the measure does not operate on a continuous dimension due to this skip-out. Further, analyses such as internal consistency and item-total correlations are not appropriate as the skip-out instruction set inflates internal consistency estimates and raises the item-total correlations of the "skipped"

230

NEWMAN

ET AL.

items. Moreover, because of the skip-out, the last three items have not been fully tested in the open population. Future refinements of this questionnaire should include changing the response format such that all of the items are Likert scaled and the questionnaire contains no skip-out. In this way, the GAD-Q-IV can be fully tested as a measure of dimensional severity. Our research group has begun testing such a questionnaire.

APPENDIX GAD-Q-IV 1. Do you experience excessive worry?

Yes

No

2. Is your worry excessive in intensity, frequency, or amount of distress it causes'?

Yes

No

3. Do you find it difficult to control your worry (or stop worrying) once it starts?

Yes

No

4. Do you worry excessively and uncontrollably about minor things such as being late for an appointment, minor repairs, homework, etc.?

Yes

No

5. Please list the most frequent topics about which you worry excessively and uncontrollably: a.

d.

b.

e.

c.

f.

6. During the last six months~ have you been bothered by excessive and uncontrollable worries more days than not? Yes No IF YES, CONTINUE. IF NO, SKIP REMAINING QUESTIONS. 7. During the past six months, have you often been bothered by any of the following symptoms? Place a check next to each symptom that you have had more days than not: _ _ Irritability

_ _ Restlessness or feeling keyed up or on edge __

Difficulty falling/staying asleep or restless/unsatisfying sleep

_ _ Difficulty concentrating or mind going blank

__

Being easily fatigued

__

Muscle tension

8. How much do worry and physical symptoms interfere with your life, work, social activities, family, etc.? Circle one number:

0

1

2

3

4

5

6

7

8

/

/

/

/

/

/

/

/

/

None

Mildly

Moderately

Severely

Very Severely

9. How much are you b6thered by worry and physical symptoms (how much distress does it cause you)? Circle one number: 0 / No distress

I /

2 / Mild distress

3 /

4 / Moderate distress

5 /

6 / Severe distress

7 /

8 / Very Severe Distress


231

References American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author. American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed). Washington, DC: Author. Anderson, D. J., Noyes, R., & Crowe, R. R. (1984). A comparison of panic disorder and generalized anxiety disorder. American Journal of Psychiatry, 141,572-575. Andrykowski, M. A., Cordova, M. J., Studts, J. L., & Miller, T. W. (1998). Posttraumatic stress disorder after treatment for breast cancer: Prevalence of diagnosis and use of the PTSD Checklist-Civilian Version (PCL-C) as a screening instrument. Journal of Consulting and Clinical Psychology, 66,586-590. Barlow, D. H. (1988). Anxiety and its disorders: The nature and treatment of anxiety and panic. New York: Guilford Press. Blanchard, E. B., Jones-Alexander, J., Buckley, T. C., & Forneris, C. A. (1996). Psychometric properties of the PTSD checklist (PCL). Behaviour Research and Therapy, 34,669-673. Borkovec, T. D., Abel, J. L., & Newman, H. (1995). The effects of therapy on comorbid conditions in generalized anxiety disorder. Journal of Consulting and Clinical Psychology, 63, 479-483. Borkovec, T. D., & Inz, J. (1990). The nature of worry in generalized anxiety disorder: A predominance of thought activity. Behaviour Research and Therapy, 28,153-158. Borkovec, T. D., & Newman, M. G. (1998). Worry and generalized anxiety disorder. In P. Salkovskis, A. S. Bellack, & M. Hersen (Eds.), Comprehensive clinical psychology (Vol. 6, pp. 439-459). New York: Pergamon Press. Borkovec, T. D., Newman, M. G., Pincus, A. L., & Lytle, R. (2002). A component analysis of cognitive behavioral therapy for generalized anxiety disorder and the role of interpersonal problems. Journal of Consulting and Clinical Psychology, 70,288-298. Borkovec, T. D., & Whisman, M. A. (1996). Psychological treatment for generalized anxiety disorder. In M. R. Mavissakalian & R. F. Prien (Eds.), Long-term treatments of anxiety disorders (pp. 171-199). Washington, DC: American Psychiatric Association. Brown, T. A., Antony, M. M., & Barlow, D. H. (1992). Psychometric properties of the Penn State Worry Questionnaire in a clinical anxiety disorders sample. Behaviour Research and Therapy, 30, 33-38. Brown, T. A., Di Nardo, P. A., & Barlow, D. H. (1994). Anxiety Disorders h~terview Schedule for DSM-IE. San Antonio, TX: The Psychological Corporation. Brown, T. A., Di Nardo, P. A., Lehman, C. L., & Campbell, L. A. (2001). Reliability of DSM-IV anxiety and mood disorders: Implications for the classification of emotional disorders. Journal of Abnormal Psychology, 110, 49-58. Cartwright-Hatton, S., & Wells, A. (1997). Beliefs about worry and intrusions. The MetaCognitions Questionnaire and its correlates. Journal of Anxiety Disorders, 11,279-296. Cassidy, J. A. (1992). Generalized anxiety disorder and attachment: Emotion and cognition. Paper presented at the Rochester Developmental Psychopathology Symposium, University of Rochester. Cohen, J., & Cohen, P. (t975). Applied multiple regression~correlation analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum. Cox, B. J., Ross, L., Swinson, R. E, & Direnfeld, D. M. (1998). A comparison of social phobia outcome measures in cognitive-behavioral group therapy. Behavior Modification, 22,285297. Davey, G. C. L., Tallis, F., & Capuzzo, N. (1996). Beliefs about the consequences of worrying. Cognitive Therapy and Research, 20,499-520.

232

NEWMAN ET AL.

Di Nardo, R A., & Barlow, D. H. (1988). Anxiety Disorders Interview Schedule-Revised (ADIS-R). Albany: Center for Stress and Anxiety Disorders. Di Nardo, E A., Brown, T. A., & Barlow, D. H. (1994). Anxiety Disorders Interview Schedulefor DSM-1V." Lifetime Version (ADIS-IV-L). San Antonio, TX: The Psychological Corporation. Di Nardo, P. A., Moras, K., Barlow, D. H., Rapee, R. M., & Brown, T. A. (1993). Reliability of DSM-IIISR anxiety disorder categories: Using the Anxiety Disorders Interview ScheduleRevised (ADIS-R), Archives of General Psychiatry, 50,251-256. Freeston, M. H., Rheaume, J., Letarte, H., Dugas, M. J., & Ladouceur, R. (1994). Why do people worry? Personality and Individual Differences, 17, 791 - 802. Heide, F. J., & Borkovec, T. D. (1983). Relaxation-induced anxiety: Paradoxical anxiety enhancement due to relaxation training. Journal of Consulting and Clinical Psychology, 51,171-182. Kachin, K. E., Newman, M. G., & Pincus, A. L. (2001). An interpersonal approach to the classification of social phobia subtypes. Behavior Therapy, 32,479-201. Lyonfields, J. D., Borkovec, T. D., & Thayer, J. F. (1995). Vagal tone in generalized anxiety disorder and the effects of aversive imagery and worrisome thinking. Behavior Therapy, 26, 457-466. Marten, R A., Brown, T. A., Barlow, D. H., Borkovec, T. D., Shear, K. M., & Lydiard, R. B. (1993). Evaluation of the ratings comprising the associated symptom criterion of DSM-IIIR Generalized Anxiety Disorder. Journal of Nervous and Mental Disease, 181,676-682. Mattick, R. R, & Clarke, J. C. (1998). Development and validation of measures of social phobia scrutiny fear and social interactions anxiety. Behaviour Research and Therapy, 36, 455470. Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Development and validation of the Penn State Worry Questionnaire. Behaviour Research and Therapy, 28, 487495. Newman, M. G., Castonguay, L. G., Borkovec, T. D., & Molnar, C. (in press). Integrative psychotherapy for generalized anxiety disorder. In R. Heimberg, D. Mennin, & C. Turk (Eds.), The nature and treatment of generalized anxiety disorder. New York: Guilford Press. Newman, M. G., Kachin, K. E., Zuellig, A. R., & Constantino, M. J. (1997, November). The reliability and validity of the social phobia questionnaire: A new self-report measure for social phobia. Paper presented at the 31 st annual meeting of the Association for Advancement of Behavior Therapy, Miami, FL. Newman, M. G., Kachin, K. E., Zuellig, A. R., Constantino, M. J., & Cashman, L. (2001). The social phobia questionnaire: Preliminary validation of a new self-report diagnostic measure of social phobia. Manuscript submitted for publication. Newman, M. G., Zuellig, A. R., Kachin, K. E., & Constantino, M. J. (1998, November). Examination of the reliability and validity of the PDSR: A self-report measure of panic disorder. Paper presented at the 32nd annual meeting of the Association for Advancement of Behavior Therapy, Washington, DC. Noyes, R., Woodman, C., Garvey, M. J., Cook, B. L., Suelzer, M., Chancy, J., & Anderson, D. J. (1992). Generalized anxiety disorder vs. panic disorder: Distinguishing characteristics and patterns of eomorbidity. Journal of Nervous and Mental Disease, 180,369-379. Okasha, A., Bishry, Z., Khalil, A. H., Darwish, T. A., Seif El Dawla, A., & Shohdy, A. (1994). Panic disorder: An overlapping or independent entity? British Journal of Psychiatry, 164, 818-825. Penava, S. J., Otto, M. W., Maki, K. M., & Pollack, M. H. (1998). Rate of improvement during cognitive-behavioral group treatment for panic disorder. Behaviour Research and Therapy, 36, 665-673. Peters, L. (2000). Discrirninant validity of the Social Phobia and Anxiety Inventory (SPAI), the Social Phobia Scale (SPS) and the Social Interaction Anxiety Scale (S/AS). Behaviour Research and Therapy, 38, 943-950.

RELIABILITYAND VALIDITYOF THE GAD-Q-IV

233

Rehm, L. R (1981). Assessment of depression. In M. Hersen & A. S. Bellack (Eds.), Handbook of behavioral assessment: A practical handbook (pp. 246-295). New York: Pergamon Press. Ries, B. J., McNeil, D. W., Boone, M. L., Turk, C. L., Carter, L. E., & Heimberg, R. G. (1998). Assessment of contemporary social phobia verbal report instruments. Behaviour Research and Therapy, 36, 983-994. Roemer, L., Borkovec, M., Posa, S., & Borkovec, T. D. (1995). A self-report diagnostic measure of generalized anxiety disorder. Journal of Behavior Therapy and Experimental Psychiatry, 26, 345-350. Roemer, L., Borkovec, M., Posa, S., & Lyonfields, J. D. (1991, November). Generalized anxiety disorder in an anologue population: The role of past trauma. Paper presented at the annual meeting of the Association for Advancement of Behavior Therapy, New York. Roemer, L., Molina, S., & Borkovec, T. D. (1997). An investigation of worry content among generally anxious individuals. Journal of Nervous and Mental Disease, 185, 314-319. Sanderson, W. C., & Barlow, D. H. (1990). A description of patients diagnosed with DSM-III revised generalized anxiety disorder. Journal of Nervous and Mental Disease, 178, 588591. Shadick, R. N., Roemer, L., Hopkins, M. B., & Borkovec, T. D. (1991, November). The nature of worrisome thoughts. Paper presented at the annual meeting of the Association for Advancement of Behavior Therapy, New York. Shear, M. K., Brown, T. A., Barlow, D. H., Money, R., Sholomskas, D. E., Woods, S. W., Gorman, J. M., & Papp, L. A. (1997). Multicenter collaborative Panic Disorder Severity Scale. American Journal of Psychiatry, 154, 1571-1575. Spielberger, C. D. Gorsuch, R. L., Lushene, R., Vagg, R R., & Jacobs, G. A. (1983). Manual for the State-Trait Anxiety Inventory STA1 (Form Y). Palo Alto, CA: Mind Garden, Inc. Weathers, F. W., Huska, J. A., & Keane, T. M. (1991). The PTSD Checklist-Civilian version (PCL-C). Available from F. W. Weathers, National Center for PTSD, Boston Veterans Affairs Medical Center, 150 S. Huntington Avenue, Boston, MA 02130. Weathers, F. W., Litz, B. T., Herman, D. S., Huska, J. A., & Keane, T. M. (1991). The PTSD checklist (PCL): Reliability, validity and diagnostic utility. Paper presented at the International Society for Traumatic Stress Studies, San Antonio, TX. Zuellig, A. R., & Newman, M. G. (1996, November). Childhood anxiety disorders in adults diagnosed with generalized anxiety disorder or panic disorder. Paper presented at the 30th annual meeting of the Association for Advancement of Behavior Therapy, New York. Zung, W. W. (1965). A self-rating depression scale. Archives of General Psychiatry, 12, 63-70. RECEIVED: February 7, 2001 ACCEPTED: November 29, 2001