Generalized anxiety disorder: a preliminary test of a conceptual model

BEHAVIOUR RESEARCH A N D THERAPY PERGAMON

Behaviour Research and Therapy 36 (1998) 215-226

Generalized anxiety disorder: a preliminary test of a conceptual model Michel J. Dugas a,,, Fabien Gagnon b, Robert Ladouceur a, Mark H. Freeston a aEcole de psychologie, Universit~ Laval, Quebec, Canada G1K 7P4 bDdpartdment de psychiatrie, CHUQ-CHUL, Quebec, Canada GIK 4GS

Received 3 June 1997

Abstract

This study presents a preliminary test of a conceptual model of Generalized Anxiety Disorder (GAD) which is theoretically driven and has clear clinical implications. The model's main features are intolerance of uncertainty, beliefs about worry, poor problem orientation and cognitive avoidance. Subjects were 24 GAD patients and 20 non clinical control subjects. The results show that all main components of the model were highly related to the discriminant function and that intolerance of uncertainty was pivotal in distinguishing GAD patients from non clinical subjects. Further, the discriminant function derived from these four process variables was very effective for classifying GAD patients and non clinical subjects into their respective groups. Overall, 82% of subjects were correctly classified as 18 of 24 subjects in the GAD group and 18 of 20 subjects in the non clinical group were properly identified. The results are discussed in terms of the proposed model of GAD and its clinical implications. © 1998 Elsevier Science Ltd. All rights reserved Key words: Generalized Anxiety Disorder, Model, Treatment

1. Introduction

Although there has been a dramatic increase of articles on anxiety disorders over the past two decades, Generalized Anxiety Disorder ( G A D ) remains infrequently studied (Norton et al., 1995). F o r instance, between 1990 and 1992, only 7.2% of published anxiety disorder studies involved G A D (Cox et al., 1995). Our research team has been working to develop a cognitivebehavioral model o f G A D which is theory driven, empirically supported, and has clear clinical

* Author for correspondence. 0005-7967/98/$19.00 © 1998 Elsevier Science Ltd. All rights reserved PII: S0005-7967(97)00070-3

216

M. J. Dugas et al./ Behaviour Research and Therapy 36 (1998) 215-226

Situation lntolel Intolc lntole lntole lntole Intole Intole

~

+ Whati f . . . ~

ainty race of unce/%~.,,.m,~rance of unce: ainty . 1 ~ ~ y ~ Intolerance of uncel ainty ~ a i n t y ~ Intolerance of unce ainty ~ t o l e r a n c e of unce ainty. m cel amty ~ e f s # a b o u t worry~c~ 1 ainty

( W o rry) [°e' ....

,ntolelnt°l~ , n Intole m ce ainty Mood Intole: m ~ I ) c e ainty state Intole: race ~ uncertainty II £ntoleranc~l unce: ainty lntol . . . . . . t . . . . . tainty I lntol . . . . . elof . . . . . . inty lntole: m c e t uncertainty II lntolerancelof unce~ ainty lntol . . . . . . ~ . . . . tainty I lntol . . . . . ~ f . . . . . . inty ............

-'-" ...........

Life events

)1

oP°°~le,,, ~ Cog.nitive /I °nentat;°n~ i i i ' ~ ' i ~ ! ~ i ~ av°ldance J ~ Intolerance o! uncertainty

Intolerance ot uncertainty 1 )

Fig. I. Conceptual model of GAD. implications. The model, in its present state, is shown in Fig. 1. For the present purposes, only the main features of the model will be addressed; non specific features such as mood state and life events will therefore not be discussed. The first feature of the model is the central role given to intolerance of uncertainty, defined as the way an individual perceives information in uncertain or ambiguous situations and responds to this information with a set of cognitive, emotional and behavioral reactions (Ladouceur et al. 1997). It is our position that intolerance of uncertainty is a key process variable in GAD which can exacerbate initial 'what if...?' questions and even generate these questions in the absence of an immediate stimulus. Recent studies have shown that intolerance of uncertainty: (1) is highly related to worry, irrespective of anxiety and depression levels (Dugas et al., 1997); (2) distinguishes non clinical worriers meeting GAD criteria from those who do not (Freeston et al., 1994b); and (3) discriminates GAD patients from non clinical moderate worriers (Ladouceur et al., in press). Independent studies have also shown that although worriers and non worriers respond similarly to unambiguous behavioral tasks, worriers show disrupted responding as stimulus ambiguity increases (Metzger et al., 1990; Tallis, 1989; Eysenck and Mathews, 1991). In other words, when completing behavioral tasks, worriers appear to have a lower threshold of tolerance for uncertainty than non worriers. Interestingly, although the ambiguity of the task at hand (for instance, when there is no correct response) distinguishes worriers from non-worriers, level of difficulty does not (Ladouceur et al., 1997). The model also underscores the contribution of beliefs about worry to GAD. Examples of these beliefs include "worrying helps avoid disappointment", "worrying protects loved ones", "worrying helps find a better way of doing things" and "worrying can stop bad things from happening" (Freeston et al., 1994b). Recent findings indicate that beliefs about worry are related to level of worry (Freeston et al., 1994b) and that compared to non clinical moderate worriers, GAD patients believe that worrying is more useful in helping to find solutions and

M. J. Dugas et al. / Behaviour Research and Therapy 36 (1998) 215-226

217

preventing negative outcomes (Ladouceur et al., in press). Further, Davey, Tallis and Capuzzo (1996) have demonstrated that individuals who hold more negative and positive beliefs about the consequences of worrying have a greater tendency to worry. Borkovec and Roemer (1995) also offer support for the importance of beliefs about worry: they found that ratings of worry as a distraction from more emotional topics discriminate non clinical worriers meeting GAD criteria from non worried anxious and non anxious Ss. Clinician reports also indicate that GAD patients often claim that worrying helps them to be prepared for negative outcomes, even if these outcomes are generally improbable (Brown et al., 1993; Roemer and Borkovec, 1993). Obviously, beliefs such as these may be negatively reinforced by the non occurrence of the feared event. The third main feature of the model is poor problem orientation. Problem orientation may be defined as a set of metacognitive processes that reflect awareness and appraisal of everyday problems and one's own problem-solving ability (Maydeu-Olivares and D'Zurilla, 1996). Problem orientation includes problem perception, problem attribution, problem appraisal, personal control beliefs and emotional responses (Maguth et al., 1996). Davey (1994) was the first to suggest that worriers have poor problem orientation. He observed that worry was associated with poor problem-solving confidence and poor perceived control over the problem-solving process, both of which are indicators of poor problem orientation. In fact, with trait anxiety partialed out, worry is related to positive problem-solving behaviors such as problem-focused and information-seeking strategies; yet it remains associated with manifestations of poor problem orientation such as the tendency to define events as threats (Davey et al., 1992). A subsequent study carried out by Davey indicates that changes in problem-solving confidence can have a causal effect on catastrophic worrying (Davey et al., 1996). Our research team has shown that when mood state is statistically controlled, worry is significantly related to poor problem orientation yet unrelated to problem-solving skills (Dugas et al., 1995c; Dugas et al., 1997). It has also been shown that GAD patients have poorer problem orientation (yet similar knowledge of problem-solving skills) than non clinical moderate worriers (Ladouceur et al., in press). The final main feature of the model is the identification of cognitive avoidance as an important process variable in GAD. Researchers at Penn State University showed that worry is primarily made up of semantic cognitive activity or 'verbal' thought rather than mental images (Borkovec and Inz, 1990). Freeston et al. (1996) replicated this finding by showing that worry is composed predominantly of thoughts rather than images and that high worriers report a higher percentage of thoughts compared to moderate worriers. Borkovec and Hu (1990) also demonstrated that S s who worry, display less heart rate response to subsequent phobic imagery than Ss who either engage in relaxed or neutral thinking. They conclude that semantic cognitive activity, as found in worry, has the effect of decreasing somatic activity resulting from fearful imagery (Borkovec and Lyonfields, 1993). This has led the Penn State group to propose that the verbal content of worry represents avoidance of fear provoking imagery, and that worry is negatively reinforced by a decrease in aversive somatic activation. The avoidance of mental images and peripheral physiological activation leads to a decrease in emotional processing of the threatening material, which further maintains worry (Butler et al., 1995; Wells and Papageorgiou, 1995). Thus cognitive avoidance, or specifically the avoidance of threatening

218

M. J. Dugas et al./ Behaviour Research and Therapy 36 (1998) 215-226

mental images, also appears to be a key component for cognitive-behavioral models of G A D worry. The goal of this study is to carry out a test of the conceptual model of G A D presented above. Although previous studies have examined the different components of the model (mostly in non clinical populations), all main features of the model have yet to be included in the same study. By examining all components simultaneously in a G A D patient sample, the present study will help establish the relative importance of each of the model's main features. The hypotheses are the following: (1) intolerance of uncertainty, beliefs about worry, poor problem orientation and cognitive avoidance will discriminate G A D patients from non clinical control Ss; and (2) intolerance of uncertainty will be the most important variable in explaining differences between both groups.

2. Method Subjects

Twenty-four G A D patients and 20 non clinical control Ss participated in the study. The G A D group, which consisted of 17 females and 7 males, had a mean age of 38.8 yr (SD = 11.4) whereas the control group, which was made up of 14 females and 6 males, had a mean age of 34.9 yr (SD = 7.4). Secondary diagnoses in the G A D group were the following: Social Phobia (N = 13), Major Depressive Disorder ( N = 4), Panic Disorder with Agoraphobia (N = 3), Obsessive-Compulsive Disorder (N = 3), Dysthymic Disorder (N = 3), Panic Disorder without Agoraphobia (N = 2) and Post-traumatic Stress Disorder (N = 1). Instruments

2.1. Diagnostic and symptom measures 2.1.1. Anxiety disorders interview schedule for D S M - I V (ADIS-IV; Brown et al., 1994) The ADIS-IV, which is designed for the anxiety disorders, also contains items which screen for mood disorders, somatoform disorders, psychoactive substance use disorders, psychotic disorders and medical problems. The G A D section includes items which cover DSM-IV diagnostic criteria as well as other items about worry themes, percentage of the day spent worrying, duration of the disorder, etc. The interview yields information on the presence of Axis I disorders with severity ratings. 2.1.2. Penn state worry questionnaire (PSWQ; Meyer et al., 1990; translation." Ladouceur et al. 1992). The PSWQ consists of 16 items that measure the tendency to worry. The questionnaire is unifactorial, has high internal consistency and test-retest reliability, as well as adequate convergent and discriminant validity (Meyer et al., 1990). The French translation is unifactorial, has excellent test-retest reliability (r = 0.81), internal consistency (~ = 0.91), and convergent validity with other measures of worry and anxiety (Ladouceur et al., 1992).

M. J. Dugas et al./ Behaviour Researchand Therapy36 (1998) 215-226

219

2.1.3. Worry and anxiety questionnaire (WAQ; Dugas et al., 1995a) Derived from the Generalized Anxiety Disorder Questionnaire (GADQ; Roemer et al., 1995), the WAQ contains 16 items about DSM-IV diagnostic criteria for GAD and current research questions about worry. Only the GAD physical symptom items from the WAQ were retained for this study (i.e. restlessness or feeling keyed up or on edge, being easily fatigued, difficulty concentrating or mind going blank, irritability, muscle tension, and sleep disturbance). These items were used to complement the measure of worry (PSWQ) and thus obtain a more comprehensive assessment of GAD symptoms. 2.1.4. Beck anxiety inventory (BAI; Beck et al., 1988; translation: Freeston et al., 1994a) The BAI is a 21-item state anxiety scale measuring the intensity of cognitive, affective, and somatic anxious symptoms experienced during the past week. The French translation has good internal consistency (~ = 0.93) and appropriate test-retest reliability (r = 0.63). The factorial, convergent, and discriminant validity of the French version have been studied on non clinical and outpatient samples (Freeston et al., 1994a). 2.1.5. Beck depression inventory--abridged ( BDI-A; Beck and Beck, 1972; translation: Bourque and Beaudette, 1982) The abridged version of the BDI includes 13 of the original 21 items which measure the main depressive symptoms. The metric properties of the BDI-A are similar to those of the BDI (Reynolds and Gould, 1981), which have been shown to be excellent (Beck et al., 1988). The French translation has very good metric properties (Bourque and Beaudette, 1982). 2.2. Process measures 2.2.1. Intolerance of Uncertainty (IU; Freeston et al., 1994b) The IU questionnaire consists of 27 items about uncertainty, emotional and behavioral reactions to ambiguous situations, implications of being uncertain, and attempts to control the future. The relationship between the IU and measures of worry not accounted for by shared variance with negative effect (Dugas et al., 1997). The internal consistency of the IU is excellent (~ = 0.91) and it has shown evidence of criterion-related, convergent and discriminant validity (Freeston et al., 1994b). 2.2.2. Why worry? (WW; Freeston et al., 1994b) The WW questionnaire consists of 20 items measuring beliefs about worry. Two types of beliefs are assessed by the WW: (1) worrying can prevent negative outcomes from happening or provide distraction from fearful images or from thinking about worse things; and (2) worrying has positive effects such as finding a better way of doing things, increasing control, and finding solutions. The questionnaire has demonstrated criterion-related, convergent and discriminant validity (Freeston et al., 1994b) and adequate test-retest reliability at 5 weeks (r = 0.71) (Dugas et al., 1995b).

220

M. J. Dugas et al. / Behaviour Research and Therapy 36 (1998) 215-226

2.2.3. White bear suppression inventory ( WBSI; Wegner and Zanakos, 1994; translation. Letarte et al., in press) The WBSI is made up of 15 items measuring the tendency to suppress unwanted thoughts. It shows evidence of convergent and divergent validity and has adequate test-retest reliability (r = 0.69-0.92) (Wegner and Zanakos, 1992). The French translation shows good internal consistency (~ = 0.87). Although the WBSI was originally conceived as a unifactorial measure of thought suppression, recent analyses suggest the presence of two factors: (1) actual thought suppression; and (2) lack of control over thoughts. Examination of the individual items of the WBSI clearly supports this distinction. Because of lack of control over thoughts may be confounded with G A D symptoms (i.e. worry), only the first factor, actual thought suppression, was retained for the analyses. This decision allowed the role of thought suppression to be better assessed without overestimating its importance because of a symptom confound.

2.2.4. Social problem-solving inventory--abridged (SPSL D'Zurilla and Nezu, 1990) ( SPSI-A; Dugas et al., 1996) The SPSI-A, which measures social problem-solving ability, was developed from the SPSI and contains 35 of the original questionnaire's 70 items. The SPSI-A maintains both major scales and all seven subscales of the SPSI, but each subscale contains five rather than 10 items. The major scales are Problem Orientation (PO) and Problem-Solving Skills (PSS). The SPSI-A scales are highly correlated with the corresponding SPSI scales (r = 0.90-0.98), have good internal consistency (~ = 0.69-0.92) and test-retest reliability (r = 0.59-0.78) (Dugas et al., 1996). Considering that previous studies indicate that problem orientation is related to worry whereas knowledge of problem-solving skills is not (Davey, 1994; Dugas et al., 1995c), the SPSI-A was divided into its two major scales (PO, PSS) for the purposes of this study (i.e. a preliminary test of the G A D model).

2.3. Procedure Recruitment for the present study and a second study on Obsessive-Compulsive Disorder (OCD) were carried out simultaneously. Clinical patients were initially screened by psychiatrists and general practitioners affiliated with our research team for presence of G A D or OCD. Patients were referred to the first author, a clinical psychologist with 10 yr experience, for further evaluation with the ADIS-IV. All patients with a primary diagnosis of G A D were included in this study. The ADIS-IV interview was tape recorded and a second clinician independently confirmed the primary diagnosis by listening to the recording. When diagnostic disagreement occurred, a conference was held with other members of our research team to arrive at a definitive diagnosis. If disagreement persisted, the S was excluded from the study. Non clinical Ss were recruited through local advertisements. The first author also administered the ADIS-IV to all non clinical Ss to ensure that they did not meet DSM-IV diagnostic criteria and a second clinician listened to a tape recording of the interview to confirm their non clinical status. Of the 22 non clinical Ss who were interviewed, two were excluded from the study because they met DSM-IV diagnostic criteria (GAD and Panic Disorder, respectively). All remaining Ss were asked to respond to the battery of questionnaires described above. The

M. J. Dugas et al. / Behaviour Research and Therapy 36 (1998) 215-226

221

Table 1 Symptom measure means and standard deviations for both groups Group

PSWQ WAQ BAI BDI-A

GAD Mean

SD

Non clinical Mean

SD

62.9 35.8 20.8 11.5

9.5 4.9 10.4 6.5

38.2 13.9 5.7 1.4

9.7 12.5 7.6 2.6

Note: PSWQ: Penn State Worry Questionnaire; WAQ: Worry and Anxiety Questionnaire (GAD physical symptoms); BAI: Beck Anxiety Inventory; BDI-A: Beck Depression Inventory--Abridged.

means and standard deviations on the symptom measures for both groups are presented in Table 1.

3. Results

A multivariate analysis of variance was carried out to examine group differences on any combination of the following variables: intolerance of uncertainty (IU), beliefs about worry (WW), thought suppression (WBSI), problem orientation (PO subscale of the SPSI-A), and problem-solving skills (PSS subscale of the SPSI-A). The result indicates that the groups differ on a combination of these variables [F(5,34) = 7.93, P < 0.0001]. To determine the relative contribution of each variable to group membership (GAD vs non clinical), subsequent discriminant analyses were carried out. To better ascertain the role played by each variable, both the total canonical structure and the standardized canonical coefficients were examined. The results show that for the total canonical structure, all variables except problem-solving skills were highly related to the discriminant function. In other words, with the exception of problem-solving skills, the variables made an important contribution to group separation. However, examination of the standardized canonical coefficients reveals that intolerance of uncertainty (0.91) was the most important variable in explaining group differences. Therefore, intolerance of uncertainty was pivotal to the discriminant function. The results for the total canonical structure and the standardized canonical coefficients are presented in Table 2. The discriminant validity of the four key process variables (intolerance of uncertainty, beliefs about worry, problem orientation and cognitive avoidance) was further examined by attempting to classify Ss into their respective groups (GAD vs non-clinical) by using the discriminant function derived from these four variables. The resulting classification matrix was then compared to the matrix obtained when using the discriminant function derived from measures of G A D symptoms (PSWQ, WAQ), anxiety (BAI) and depression (BDI-A). Because the same sample was used to determine the discriminant function and classify Ss, a jackknife cross-validation technique was used (where the S being classified is excluded when calculating the discriminant function). When using the discriminant function derived from the process variables, 82% of S s were correctly classified (K = 0.64). Specifically, 18 of 24 Ss in the G A D group and 18 of 20 Ss in the non clinical group were properly identified. When using the discriminant

222

M. J. Dugas et al. / Behaviour Research and Therapy 36 (1998) 215-226

Table 2 Discriminant analysis: intolerance of uncertainty, beliefs about worry, cognitive avoidance, problem orientation and problem-solving skills Variables Intolerance of uncertainty Beliefs about worry Cognitive avoidance Problem orientation Problem-solving skills

Total canonical structure 0.94 0.69 0.80 -0.85 -0.12

Standardized canonical coefficients 0.91 -0.08 0.54 -0.22 -0.22

Note: Intolerance of uncertainty: Intolerance of Uncertainty questionnaire; Beliefs about worry: Why Worry? questionnaire; Cognitive avoidance: White Bear Suppression Inventory (items 1, 8, 10, 11, 12, 13, 15); Problem orientation: Social Problem-Solving Inventory (Problem Orientation Scale); Problem-solving skills: Social ProblemSolving Inventory (Problem-Solving Skills Scale).

function derived from the symptom measures, 91% of S s were correctly classified (• = 0.82). In this case, 22 of 24 G A D patients and 18 of 20 non clinical Ss were properly classified. Therefore, the process variables were as effective as the symptoms for classifying non clinical Ss but not quite as effective for classifying G A D patients.

4. Discussion

The results of the present study show that intolerance of uncertainty, beliefs about worry, poor problem orientation and cognitive avoidance were highly related to the discriminant function. Further, the discriminant function derived from these four process variables was very effective for classifying G A D patients and non clinical Ss into their respective groups. Thus, the first hypothesis was supported as these four process variables discriminated G A D patients from non clinical control Ss. The second hypothesis was also confirmed as intolerance of uncertainty was pivotal to the discriminant function. In other words, intolerance of uncertainty appears to be the key process variable in GAD. Considering that diagnostic criteria were used for group formation, it is not surprising that symptom measures were generally more effective than process variable measures in correctly classifying Ss. On the contrary, it is quite impressive that the process variables were as effective as they were in correctly identifying Ss. Recall that the process variables and symptoms were equally efficient in classifying non clinical Ss (18 of 20) and that the process variables remained highly effective in correctly identifying G A D patients (18 of 24). Therefore, when G A D patients are compared to non clinical Ss, the variables identified in the model presented in Fig. 1 appear to be both sensitive and specific to GAD. Because G A D patients were not compared to clinical control patients (i.e. other anxiety disorder patients) in the present study, further research is needed to confirm the specificity of the model's main components. However, the goal of the present study was not to establish the specificity of the process variables but rather to examine their relative importance in differentiating G A D patients from non clinical controls. It should be mentioned that recent studies have begun to confirm the specificity of the model's components by showing that compared to

M. J. Dugas et al. / Behaviour Research and Therapy 36 (1998) 215-226

223

other anxiety disorder patients, GAD patients: (1) are more intolerant of uncertainty (Ladouceur et al., 1995); (2) have poorer problem orientation; and (3) have superior knowledge of problem-solving skills (Ladouceur et al., 1996). A second factor which should be taken into consideration when interpreting the results of this study is the high rate of comorbidity of the GAD patients (17 of 24 had one or more secondary diagnoses). In particular, 13 of 24 GAD patients had a secondary diagnosis of Social Phobia. Could it be that the secondary diagnoses contributed in some important way to the discriminant capacity of the process variables? Considering that the severity of the secondary diagnoses was generally mild, their contribution to the discriminant capacity of the process variables is most likely negligible. In fact, the average severity of all secondary diagnoses was 3.1/8, which is in the mild to moderate range and considered below the clinical threshold on the ADIS-IV (Brown et al., 1994). As for the 13 patients with a secondary diagnosis of Social Phobia, the average severity of Social Phobia was 2.7/8. Thus, the generally mild severity of secondary diagnoses in the GAD sample suggests that the discriminant capacity of the process variables resulted from comparing GAD to the absence of pathology. Another factor which should be considered in interpreting the study's results is that GAD patients were more depressed than non clinical controls, as evidenced by their scores on the BDI-A (Table 1). This finding was expected as elevated levels of depressive symptoms are not uncommon in GAD patients (Butler et al., 1991). The question arises as to whether level of depression may explain the discriminant ability of the process variables. This does not seem to be the case as recent studies indicate that although GAD patients are more depressed than nonclinical high worriers (Dugas et al., 1994), they have similar scores on measures of intolerance of uncertainty, beliefs about worry and problem orientation (Ladouceur et al., in press). Therefore, level of depression does not appear to be a key factor in explaining the discriminant ability of the process variables. This study is the first to establish the relative importance of each of the model's main features and to show that intolerance of uncertainty is clearly the most important component in distinguishing GAD patients from non clinical controls. Considering the magnitude of the contribution of intolerance of uncertainty to the discriminant function in this study and previous findings showing strong correlations between intolerance of uncertainty and worry (r -- 0.70, Dugas et al., 1997; r = 0.63, Freeston et al., 1994b), it must be established that intolerance of uncertainty and worry are two distinct constructs. If intolerance of uncertainty is simply another way of describing worry, then the proposed model's relevance can be seriously called into question. Using a series of behavioral tasks, Ladouceur et al. (1997) have shown that intolerance of uncertainty and worry are indeed distinct constructs. Their results indicate that although intolerance of uncertainty is related to perception of performance on ambiguous tasks, worry is not. Further, it should be noted that other researchers have proposed similar constructs to intolerance of uncertainty as process variables for worry and GAD. For instance, Tallis et al. (1991) refer to elevated evidence requirements as an important process variable for worry and GAD. In fact, Tallis and Eysenck (1994) have proposed an interesting model of worry which includes elevated evidence requirements as a main feature. Unfortunately, specific clinical guidelines ensuing from the model have yet to be delineated. The clinical implications of the model proposed in this study will now briefly be discussed. First~ therapists should help their GAD patients become more tolerant of uncertainty. In ad-

224

M. J. Dugas et al. / Behaviour Research and Therapy 36 (1998) 215-226

dition to directly challenging assumptions that uncertainty can and should be avoided, intolerance of uncertainty may be decreased in several other ways. The reevaluation of beliefs about worry, problem orientation training and cognitive exposure may all help patients become more tolerant of uncertainty. Because GAD patients tend to overestimate the advantages (Ladouceur et al., in press) and underestimate the disadvantages of worrying (Brown et al., 1993), beliefs about worry should be examined and corrected when necessary. Cognitive techniques (e.g. Socratic questioning, historical tests, etc.) and behavioral strategies (e.g. hypothesis testing, prediction logs, etc.) may be used to reevaluate faulty beliefs about worry. The reevaluation of beliefs about worry also helps decrease intolerance of uncertainty because patients recognize and 'let go' of futile attempts to control the future by using worry; instead learning to cope with the uncertainty of future events. Our research team has recently reported on the importance of identifying two types of worry when treating GAD patients: (1) worries about immediate problems (e.g. meeting deadlines at work); and (2) worries about improbable future events (e.g. one's child dying in a car accident) (Dugas et al., 1997). For worries about immediate problems, GAD patients need to improve their problem orientation to facilitate problem solving. Considering that intolerance of uncertainty is a major contributor to poor problem orientation (Dugas et al., 1997), both these process variables should be addressed simultaneously. GAD patients need to learn to stay focused on the key elements of the problem situation while ignoring associated minor details. Intolerance of uncertainty may lead to excessive preoccupation with the minor details of the problem situation (in an effort to eliminate the uncertainty related to the problem situation and the problem-solving process). Once the key elements have been identified, patients should proceed with the problem-solving process even if they are uncertain of its outcome beforehand (thus targeting both poor problem orientation and intolerance of uncertainty). Therefore, GAD patients should seek a compromise between trying to avoid the problem situation and attempting to gather excessive amounts of information about the situation, both of which interfere with effective problem solving and prolong worry. Finally, for worries about improbable future events, cognitive exposure may be used to address cognitive avoidance, or more specifically the avoidance of threatening mental images. To maximize the benefits of exposure, GAD patients should be exposed to their worrisome images while using covert response prevention, i.e. the proscription of all voluntary activity used to neutralize the image. Cognitive exposure also helps decrease intolerance of uncertainty by changing the meaning given to threatening future events. If the events are considered less threatening following exposure, intolerance of uncertainty will be reduced because the possibility of these future events occurring will be deemed more acceptable. In summary, the process variables studied in this paper increase our understanding of GAD and point to specific interventions which should increase our capacity to help individuals suffering from this anxiety disorder.

Acknowledgements The research described in this article was financially supported by grants from the Medical Research Council of Canada and the Fonds de la Recherche en Sant~ du Quebec.

M. J. Dugas et al. / Behaviour Research and Therapy 36 (1998) 215-226

225

References Beck, A. T., & Beck, R. W. (1972). Screening depressed patients in family practice: A rapid technique. Postgraduate Medecine, 52, 8185. Beck, A. T., Epstein, N., Brown, G., & Steer, R. A. (1988). An inventory for measuring clinical anxiety: Psychometric properties. Journal of Consulting and Clinical Psychology, 56, 893-897. Beck, A. T,, Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8, 77 100. Borkovec, T. D., & Hu, S. (1990). The effect of worry on cardiovascular response to phobic imagery. Behaviour Research and Therapy. 28, 69-73. Borkovec, T. D., & lnz, 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. & Lyonfields, J. D. (1993). Worry: Thought suppression of emotional processing. In H. W. Krohne (Eds.), Attention and avoidance (pp. 101-118). Seattle: Hogrefe & Huber Publishers. Borkovec, T. D., & Roemer, L. (1995). Perceived functions of worry among generalized anxiety disorder subjects: Distraction from more emotionally distressing topics? Journal of Behavior Therapy and Experimental Psychiatry, 26, 25-30. Bourque, P., & Beaudette, D. (1982). l~tude psychom+trique du questionnaire de d~pression de Beck auprds d'un 6chantillon d'~tudiants universitaires francophones (Psychometric study of the Beck Depression Inventory with French-Canadian university students). Revue Canadienne des Sciences de Comportement, 14, 211-218. Brown, T. A., Di Nardo, P. A. & Barlow, D. H. (1994). Anxiety Disorders Interview ScheduleJbr DSM-IV. Albany, New York: Graywind Publications. Brown, T. A., O'Leary, T. A. & Barlow, D. H. (1993). Generalized Anxiety Disorder. In D. H. Barlow (Eds.), Clinical handbook of psychological disorders (pp. 137-189). New York: Guilford Press. Butler, G., Fennel1, M., Robson, P., & Gelder, M. (1991). A comparison of behavior therapy and cognitive behavior therapy in the treatment of generalized anxiety disorder. Journal of Consulting and Clinical Psychology, 59, 167-175. Butler, G., Wells, A., & Dewick, H. (1995). Differential effects of worry and imagery after exposure to a stressful stimulus: A pilot study. Behavioural and Cognitive Psychotherapy, 23, 45 56. Cox, B. J., Wessel, I., Norton, G. R., Swinson, R. P., & Direnfeld, D. M. (1995). Publication trends in anxiety disorders research: 1990-1992. Journal o[ Anxiety Disorders, 9, 531-538. D'Zurilla, T. J., & Nezu, A. M. (1990). Development and preliminary evaluation of the Social Problem-Solving Inventory. Psychological Assessment, 2, 156-163. Davey, G. C. L. (1994). Worrying, social problem-solving abilities, and problem-solving confidence. Behaviour Research and Therapy, 32, 327 330. Davey, G. C. L., Hamptom, J., Farrell, J., & Davidson, S. (1992). Some characteristics of worrying: Evidence for worrying and anxiety as separate constructs. Personality and Individual Differences, 13, 133-147. Davey, G. C. L., Jubb, M., & Cameron, C. (1996). Catastrophic worrying as a function of changes in problem-solving confidence. Cognitive Therapy and Research, 20, 333 344. Davey, G. C. L., Tallis, F., & Capuzzo, N. (1996). Beliefs about the consequences of worrying. Cognitive Therapy and Research, 5,499520. Dugas, M. J., Freeston, M. H., Blais, F. & Ladouceur, R. (1994, November). Anxiety and depression in GAD patients, high and moderate worriers. Poster presented at the Annual Convention of the Association for Advancement of Behavior Therapy, San Diego, CA. Dugas, M. J., Freeston, M. H., Lachance, S., Provencher, M. & Ladouceur, R. (1995a, July). The Worry and Anxiety Questionnaire." Initial validation in nonclinical and clinical samples. Poster presented at the World Congress of Behavioural and Cognitive Therapies, Copenhagen, Denmark. Dugas, M. J., Freeston, M. H. & Ladouceur, R. (1995b, October). Validation de mesures des m~;canismes lies ~i l'inqui&ude (Validation of measures of the mechanisms related to worry). Poster presented at the Annual Convention of the Soci6t+ Qu+becoise pour la Recherche en Psychologie, Ottawa, Canada. Dugas, M. J., Freeston, M. H., & Ladouceur, R. (1997). Intolerance of uncertainty and problem orientation in worry. Cognitive Therapy and Research, 21, 593-606. Dugas, M. J., Ladouceur, R., & Freeston, M. H. (1996). Version abr~g~e del'Inventaire de r+solution de probl6mes sociaux (Abbreviated version of the Social Problem Solving Inventory). Journal de ThOrapie Comportementale et Cognitive, 6, 59-62. Dugas, M. J., Letarte, H., Rh+aume, J., Freeston, M. H., & Ladouceur, R. (1995c). Worry and problem solving: Evidence of a specific relationship. Cognitive Therapy and Research, 19, 109-120. Freeston, M. H., Dugas, M. J., & Ladouceur, R. (1996). Thoughts, images, worry and anxiety. Cognitive Therapy and Research, 20, 265-273.

226

M. J. Dugas et al. / Behaviour Research and Therapy 36 (1998) 215-226

Freeston, M. H., Ladouceur, R., Thibodeau, N., Gagnon, F., & Rh~aume, J. (1994a). L'inventaire d'anxi&6 de Beck: Propri&+s psychom+triques d'une traduction franqaise (The Beck Anxiety Inventory: Psychometric properties of a French translation). L'Enc~phale, XX, 47-55. Freeston, M. H., Rheaume, J., Letarte, H., Dugas, M. H., & Ladouceur, R. (1994b). Why do people worry? Personality and Individual Differences, 17, 791-802. Ladouceur, R., Blais, F., Freeston, M. H. & Dugas, M. J. (in press). Problem solving and problem orientation in generalized anxiety disorder. Journal of Anxiety Disorders. Ladouceur, R., Dugas, M. J., Freeston, M. H., Provencher, M., Rh~aume, J., Gagnon, F., Thibodeau, N. & Blais, F. (1996, November). Problem solving in GAD patients, other anxiety disorder patients and nonclinical subjects. Poster presented at the Annual Convention of the Association for Advancement of Behavior Therapy, New York, NY. Ladouceur, R., Freeston, M. H., Dugas, M. J., Rh~aume, J., Gagnon, F., Thibodeau, N., Boisvert, J-M., Provencher, M. & Blais, F. (1995, November). Specific association between Generalized Anxiety Disorder and intolerance of uncertainty among anxiety disorder patients. Poster presented at the Annual Convention of the Association for Advancement of Behavior Therapy, Washington, DC. Ladouceur, R., Freeston, M. H., Dumont, J., Letarte, H., Rh~aume, J., Gagnon, F., & Thibodeau, N. (1992). Penn State Worry Questionnaire: Validity and reliability of a French translation. Canadian Psychology. 33, 236. Ladouceur, R., Talbot, F., & Dugas, M. J. (1997). Behavioral expressions of intolerance of uncertainty in worry: Experimental findings. Behavior Modification, 21,355-371. Letarte, H., Ladouceur, R., Freeston, M. H., & Rh6aume, J. (in). Incentive to suppress a neutral thought. Behavioural and Cognitive Psychotherapy, press, . Maguth, Nezu C., Nezu, A. M., D'Zurilla, T. J. & Friedman, S. H. ( 1996, November). Problem-solving assessment and therapy: Clinical practice applications. Workshop presented at the Annual Convention of the Association for Advancement of Behavior Therapy, New York, NY. Maydeu-Olivares, A., & D'Zurilla, T. J. (1996). A factor analysis of the Social Problem-Solving Inventory using polychoric correlations. European Journal of Psychological Assessment, 11, 98-107. Metzger, R. L., Miller, M. L., Cohen, M., Sofka, M., & Borkovec, T. D. (1990). Worry changes decision making: The effect of negative thoughts on cognitive processing. Journal of Clinical Psychology, 46, 78-88. 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, 487~496. Norton, G. R., Cox, B. J., Asmundson, G. J. G., & Maser, J. D. (1995). The growth of research on anxiety disorders during the 1980 s. Journal of Anxiety Disorders, 9, 75-85. Reynolds, W. M., & Gould, J. W. (1981). A psychometric investigation of the standard and short form Beck Depression Inventory. Journal of Consulting and Clinical Psychology. 49, 306-307. Roemer, L. & Borkovec, T. D. (1993) Worry: Unwanted cognitive activity that controls unwanted somatic experience. In D. M. Wegner & J. W. Pennebaker, (Eds 3 Handbook of mental control. Englewood Cliffs, NJ: Prentice-Hall. Roemer, L., Borkovec, M., Posa, S., & Borkovec, T. D. (1995). A self-report diagnostic measure of generalized anxiety disorder. Behaviour Research and Therapy, 26, 345-350. Tallis, F. (1989). Worrt': A cognitive analysis. Unpublished doctoral dissertation, University of London. Tallis, F., & Eysenck, M. H. (1994). Worry: Mechanisms and modulating influences. Behavioural and Cognitive Psychotherapy, 22, 3756. Tallis, F., Eysenck, M., & Mathews, A. (1991). Elevated evidence requirements and worry. Personality and Individual Diff~,rences, 12, 21 27. Wegner, D. M., & Zanakos, S. (1994). Chronic though suppression. Journal of Personality, 62, 615-640. Wells, A., & Papageorgiou, C. (1995). Worry and the incubation of intrusive images following stress. Behaviour Research and Therapy, 33, 579-583.