Embed Size (px)
Citation preview
Tg
KD
a
ARRAA
KGECMEC
SActtgaSvfpeat2
F
Vf
h0
Journal of Anxiety Disorders 28 (2014) 559–569
Contents lists available at ScienceDirect
Journal of Anxiety Disorders
esting a procedural variant of written imaginal exposure foreneralized anxiety disorder
atie Fracalanza, Naomi Koerner ∗, Martin M. Antonyepartment of Psychology, Ryerson University, Canada
r t i c l e i n f o
rticle history:eceived 20 February 2014eceived in revised form 11 May 2014ccepted 30 May 2014vailable online 14 June 2014
eywords:eneralized anxiety disorderxposure
a b s t r a c t
This experiment examined the degree to which it is more beneficial for individuals with generalizedanxiety disorder (GAD) to engage in repeated exposure to mental imagery of the same feared scenarioversus varying the exposure content. On three consecutive days, individuals with GAD (N = 57) spent20 min writing about: (1) the same worst case scenario (consistent exposure; CE), (2) variations of theirworst case scenario (varied exposure; VE), or (3) a neutral topic (neutral control; NC). Participants inthe CE condition displayed significant decreases in worry, acute cognitive avoidance, and intoleranceof uncertainty from baseline to 1-week follow-up; participants in the VE and NC conditions did not.Initial activation of self-reported anxiety (observed in the CE and VE conditions) and between-session
ognitive processesental imagery
motional processing theoryognitive-behavioral therapy
reduction in anxiety (observed in the CE condition only) were associated with improvement in worry.Including more references to negative emotion and writing in the present tense were also associatedwith greater improvement in worry in the CE condition. These findings suggest that writing repeatedlyabout the same worst case scenario may benefit people with GAD. The study also provides informationon potential mechanisms of change.
© 2014 Published by Elsevier Ltd.
According to the most recent editions of the Diagnostic andtatistical Manual of Mental Disorders (American Psychiatricssociation, 2000, 2013), generalized anxiety disorder (GAD) is aommon and debilitating disorder marked by excessive and uncon-rollable worry and anxiety. Borkovec and colleagues suggestedhat individuals with GAD engage in cognitive avoidance strate-ies to dampen mental images of feared scenarios and the anxiousrousal brought on by these images (Sibrava & Borkovec, 2006;töber & Borkovec, 2002). Stöber (1998, 2000) proposed that indi-iduals with GAD achieve this dampening effect by thinking abouteared scenarios in an abstract manner, since abstract thinkingroduces mental images that are less vivid and emotionally-vocative than concrete thinking (Paivio & Marschark, 1991). Asconsequence, however, abstract thinking may prevent the “emo-
ional processing” of fears (Sibrava & Borkovec, 2006; Stöber, 1998,
000; Stöber & Borkovec, 2002).Emotional processing theory (EPT; Foa & Kozak, 1986; update,oa, Huppert, & Cahill, 2006) explains that for “emotional
∗ Corresponding author at: Department of Psychology, Ryerson University, 350ictoria Street, Toronto, Ontario, Canada M5B 2K3. Tel.: +1 416 979 5000x2151;
ax: +1 416 979 5273.E-mail address: [email protected] (N. Koerner).
ttp://dx.doi.org/10.1016/j.janxdis.2014.05.011887-6185/© 2014 Published by Elsevier Ltd.
processing” to occur, a person must engage in repeated, system-atic exposure to the object or situation that he or she is avoiding.During this exposure the full “fear structure” encompassing stim-ulus, response and meaning elements, must be activated. Anxiousarousal during the first exposure session is taken as an indicatorof such activation. In addition, new information that is incompat-ible with the fear structure must be integrated into it, which isthought to be shown by a reduction in anxious arousal across expo-sure sessions.1 Because abstract thinking dulls mental imagery, andconsequently, emotional arousal, it may prevent the complete acti-vation of fear structures and impede emotional processing (seeSibrava & Borkovec, 2006).
If avoidance of threatening mental images plays a role in main-taining chronic worry, then confronting these images and theanxious arousal that they evoke may be helpful to individuals with
GAD, as it may promote emotional processing (van der Heider &ten Broeke, 2009; Zinbarg, Craske, & Barlow, 2006). In many anxi-ety disorders, the feared stimuli are tangible objects or situations1 Within-session anxiety reduction was also initially proposed as an indicator ofemotional processing (Foa and Kozak, 1986); however, it currently is not consideredto be an indicator of emotional processing, as it has not been consistently linked tosymptom improvement (see Foa et al., 2006).
5 nxiet
(atsiWfi2fs
1
c2itPtspilasntl(nsiadatdicfiiaemtTcadt
gptSTo&lfgge
60 K. Fracalanza et al. / Journal of A
e.g., snakes, driving) that can be approached in vivo. However,ccording to Stöber and Borkovec (2002), for people with GADhe feared stimuli are mental images of hypothetical worst casecenarios (WCSs) and the anxious arousal associated with thesemages. Based on this, exposure for GAD involves recounting the
CS coming true in as much emotional detail as possible, in therst person, and in the present tense (see Dugas & Robichaud,007; Rygh & Sanderson, 2004), which is thought to help concretizeeared mental images and promote a sense of actually being in theituation.
. Previous research on imaginal exposure for GAD
Although imaginal exposure is included in some evidence-basedognitive-behavioral therapies (CBT) for GAD (Dugas & Robichaud,007; Zinbarg et al., 2006), only a few studies have examined
maginal exposure independently of other CBT techniques, andhe methods used in these studies have varied considerably.rovencher, Dugas, and Ladouceur (2004) used a case replica-ion design to investigate the effects of twelve 1-hour therapyessions of imaginal exposure or problem solving in a GAD sam-le. Provencher et al. found that both techniques led to large
mprovements in worry and GAD-associated symptoms from base-ine to posttreatment (ds = 1.42–1.63). Goldman, Dugas, Sexton,nd Gervais (2007) compared the impact of five 30-minute ses-ions of written imaginal exposure to five 30-minute sessions ofeutral writing in a sample of high worriers. They found that par-icipants who engaged in imaginal exposure showed significant,arge improvement in worry from baseline to 2-week follow-upd = 1.22), whereas participants who engaged in neutral writing didot (d = .45). They also found that participants in the imaginal expo-ure and neutral writing conditions both showed significant, largemprovements in GAD-associated symptoms over time (ds = 1.14nd .80, respectively). There were no significant between-groupifferences at 2-week follow-up on either worry (d = .40) or GAD-ssociated symptoms (d = .22). In addition, Goldman et al. foundhat intolerance of uncertainty (IU) scores at one time point pre-icted levels of GAD symptoms at a subsequent time point in their
maginal exposure condition. Finally, Hoyer and colleagues (2009)onducted a randomized controlled trial comparing the impact offteen 1-hour sessions of worry exposure (i.e., evoking and holding
n mind a mental image of a feared scenario) or applied relaxation towaitlist condition in a GAD sample. Hoyer et al. found that worryxposure and applied relaxation both led to significant improve-ents in worry, anxiety, negative beliefs about worry, and the
endency to suppress thoughts compared to the waitlist condition.hese effects ranged from small to large in the imaginal exposureondition (ds = .27–.89), and medium to large in the applied relax-tion condition (ds = .78–.97); however, there were no significantifferences between the treatment conditions on outcomes at post-reatment.
Overall, the limited evidence suggests that the delivery of ima-inal exposure modifies GAD symptoms as well as underpinningrocesses. However, there are a number of important questionshat still need to be addressed regarding the effects of exposure.pecifically, there is much to be learned about the procedure itself.here have been many investigations of the optimal parametersf exposure for other anxiety disorders (e.g., Abramowitz, Foa,
Franklin, 2003; Chambless, 1990); however such studies areacking in GAD. Studies comparing different exposure procedures
or GAD are important to the development of empirically-baseduidelines for conducting exposure in practice, and having suchuidelines may ultimately improve the therapeutic value ofxposure.y Disorders 28 (2014) 559–569
2. The present study
The present study sought to extend what is known aboutexposure for GAD by comparing the effects of repeated imaginalexposure to the same WCS (consistent exposure) to the effectsof exposure to different WCSs (varied exposure). Past researchwith individuals with specific phobias has suggested some bene-fit to varying the exposure stimulus (Lang & Craske, 2000; Rowe& Craske, 1998), as this ought to result in greater generalizationof learning from the exposure context to “real life” encounterswith the feared stimulus (Schmidt & Bjork, 1992). In contrast, astudy with people who had experienced multiple traumas foundthat consistent exposure to the same trauma memory on each of 3days was more beneficial in reducing posttraumatic stress disorder(PTSD) symptoms than was exposure to a different trauma mem-ory on each of 3 days, which was no more helpful than reflectingon nonemotional events (Sloan, Marx, & Epstein, 2005). This sug-gests that under certain circumstances, not varying the exposurecontent may be most beneficial. It is possible that the benefits ofvaried exposure may depend on the type of stimulus that one isbeing exposed to (e.g., an animal versus a complex mental image).
People with GAD, by definition, have concerns that traverseseveral domains (e.g., Dugas, Freeston, et al., 1998). Further, theytypically have multiple worries within the same domain (e.g., manyworries related to work). Thus, there are many possible expo-sure stimuli. The first aim of the present study was to investigatewhether consistent or varied imaginal exposure would be mostbeneficial in reducing GAD symptoms. It was hypothesized thatindividuals who engaged in consistent or varied exposure wouldreport greater improvement in GAD symptoms from baseline to 1-week follow-up compared to individuals who engaged in neutralwriting. Given that previous findings on the relative advantage ofconsistent versus varied exposure have been mixed, it was possiblethat either could facilitate greater improvement in GAD symptoms,thus no specific hypotheses were offered.
The second objective of the present study was to examine theeffect of exposure on two processes involved in GAD – cognitiveavoidance and IU (Dugas, Freeston, et al., 1998; Dugas, Gagnon,Ladouceur, & Freeston, 1998; Sibrava & Borkovec, 2006). Theoreti-cal writings and past research suggest that these processes shouldimprove with imaginal exposure to threatening mental images(Dugas & Robichaud, 2007; Goldman et al., 2007; van der Heiderand ten Broeke, 2009). It was predicted that from baseline to 1-week follow-up, individuals assigned to either consistent or variedexposure would display a significant improvement in: the habitualtendency to engage in cognitive avoidance (from here on, chroniccognitive avoidance), fear and avoidance of an idiographic mentalimage of the WCS (from here on, acute cognitive avoidance), andIU, relative to individuals assigned to the neutral writing condi-tion. No hypotheses were advanced regarding differences betweenthe exposure conditions, given the mixed findings describedearlier.
The third goal of the present study was to test the notion thatthe activation of anxious arousal in the first exposure session and areduction in anxious arousal across exposure sessions predict out-comes following exposure. To our knowledge, these core tenetsof EPT as applied to exposure for GAD have not yet been tested.It was hypothesized that relative to participants in the neutralwriting condition, participants in the exposure conditions woulddisplay a significant increase in anxious arousal at the first session,and a significant reduction in peak anxious arousal across ses-sions. It was further predicted that these indicators of emotional
processing would be significantly associated with improvementsin GAD symptoms, chronic and acute cognitive avoidance, and IUin individuals who engaged in exposure. No hypotheses were pro-posed regarding differences between the exposure conditions.
nxiet
twiTmtGtcepMiwcwt
3
3
viw(wsitcdimsoewc1
RTUW1Ddpnot
Tmi(nawoms
K. Fracalanza et al. / Journal of A
The fourth purpose of this study was to explore the associa-ion between adherence to written exposure instructions, includingriting with reference to negative emotions, in the first person, and
n the present tense, and improvement in GAD-related outcomes.he abovementioned writing instructions are thought to promoteore effective exposure by reducing emotional and temporal dis-
ance from threatening mental imagery (Dugas & Robichaud, 2007;oldman et al., 2007), although this has not yet been empirically
ested in a GAD sample. So far, previous research has shown aonnection between greater inclusion of references to negativemotions during written exposure and greater improvement insychological outcomes in individuals with PTSD symptoms (Sloan,arx, Epstein, & Lexington, 2007). It was hypothesized that includ-
ng more references to negative emotions in accounts of the WCSould be positively associated with improvement in GAD out-
omes in the exposure conditions; however no other hypothesesere formulated given the dearth of existing research on this ques-
ion.
. Method
.1. Participants
Participants were recruited via advertisements at Ryerson Uni-ersity and in the Toronto community. Interested individuals werenvited to complete a telephone screen to determine eligibility,
hich included the Mini International Neuropsychiatric InterviewMINI) Screen, and corresponding modules of the MINI Interviewhen items from the MINI Screen were endorsed. The telephone
creen was used to determine if participants met the followingnclusion criteria: (1) 18–65 years of age; (2) endorsed symptomshat met the criteria for a principal diagnosis of DSM-IV GAD; (3) nourrent suicidality, alcohol or drug abuse, or symptoms of bipolarisorder; (4) no current psychotherapy; and (5) not currently tak-
ng psychotropic medication, or taking a stable dose of psychotropicedication for at least 12 weeks. In total, 165 individuals were
creened, and 65 met the inclusion criteria. Eight people droppedut, 2 from the consistent exposure condition, 4 from the variedxposure condition, and 2 from the control condition, but thereas no significant difference in the number of dropouts between
onditions, �2 = 0.85, p = .65. The final sample consisted of 57 adults,9 in each of the three conditions.
Approximately half of the sample were students recruited fromyerson University (47%), and the other half was recruited from theoronto community (53%). At baseline, participants from Ryersonniversity produced significantly higher scores on the Penn Stateorry Questionnaire (PSWQ; Meyer, Miller, Metzger, & Borkovec,
990) and the Cognitive Avoidance Questionnaire (CAQ; Sexton &ugas, 2008) than did participants from the community. Stu-ent participants were also significantly younger than communityarticipants. However, there was no significant difference in theumber of students and community members assigned to eachf the experimental conditions, and recruitment source was con-rolled for in all subsequent analyses.
Information about the sample characteristics can be found inable 1. The sample was predominantly (78.95%) female, with aean age of 33.72 years (SD = 13.77 years). The breakdown of self-
dentified ethnicity was as follows: White (50.88%), East Asian15.79%), Black (10.53%), Multiethnic (7.01%), Other (an ethnicityot listed; 5.26%), South Asian (3.51%), South East Asian (3.51%),nd Latin American (3.51%). Participants with comorbid diagnoses
ere included in the present study to increase the generalizabilityf the results. Most participants (38.60%) endorsed symptoms thatet the criteria for GAD only, 33.33% had one additional diagno-
is, and 28.07% had two or more additional diagnoses. The most
y Disorders 28 (2014) 559–569 561
common additional diagnosis was social anxiety disorder (38.60%of the sample), followed by major depressive disorder (22.81%),obsessive-compulsive disorder (17.54%), panic disorder (10.53%),and dysthymic disorder (3.51%). There were no significant differ-ences between conditions in the presence of the abovementionedadditional diagnoses (�2s = 0–3.60; ps > .17). The mean PSWQ score(M = 63.09, SD = 8.25) of participants in the current study was com-parable to the mean PSWQ scores of clinical GAD samples (e.g.,Dugas et al., 2010; Hoyer et al., 2009; Newman et al., 2011). Inaddition, most of the sample (98%) scored above the cut score of 45that is used to identify GAD on the PSWQ (Behar, Alcaine, Zuellig,& Borkovec, 2003), and 67% scored above the cut score of 65 thatis used to distinguish GAD from other anxiety disorders (Fresco,Mennin, Heimberg, & Turk, 2003).
3.2. Measures
The MINI (Sheehan et al., 1998) is a widely-used semistructureddiagnostic interview that assesses for many DSM-IV Axis I disor-ders. The MINI has very high interrater reliability (�s = .88–1.0),good test–retest reliability (rs = .76–.93), and high convergentvalidity with other clinical interviews (Lecrubier et al., 1997;Sheehan et al., 1998). The MINI Screen (Sheehan et al., 1998) can beused in conjunction with the MINI. It consists of 21 closed-endedgateway questions about current symptoms of Axis I disorders, andpositive responses on the screen prompt administration of the cor-responding module in the MINI. The MINI Screen has sensitivityranging from 0.63 to 0.82, and specificity ranging from 0.61 to 0.83(Alexander, Haugland, Lin, Bertollo, & McCorry, 2008).
The PSWQ is a 16-item self-report “gold standard” measure ofthe tendency to worry excessively. Examples of items are “My wor-ries overwhelm me” and “Many situations make me worry.” PSWQitems are rated on a 5-point Likert scale from 1 (not at all typicalof me) to 5 (very typical of me). Scores on the PSWQ range from 16to 80. The PSWQ has high internal consistency (˛s = .86–.91; Dearet al., 2011), and a range of test–retest reliabilities (rs = .74–.92; seeStartup & Erickson, 2006 for a review).
The Worry and Anxiety Questionnaire (WAQ; Dugas et al.,2001) measures the symptoms of DSM-IV-defined GAD. The associ-ated symptoms subscale of the WAQ (WAQ-A) is a 6-item subscalethat assesses the severity of the associated symptoms of GAD (e.g.,being easily fatigued, muscle tension). The WAQ-A has been usedin conjunction with the PSWQ in past research to assess the sever-ity of GAD symptoms (e.g., Dugas et al., 2003, 2010). The WAQ-Aasks respondents to rate the degree to which they have been both-ered by GAD-associated symptoms over the past six months on a5-point Likert scale from 1 (not at all) to 5 (very severely). Scoreson the WAQ-A range from 6 to 30. The WAQ has demonstrated ade-quate test–retest reliability and good convergent and discriminantvalidity (Dugas et al., 2001).
The CAQ is a 25-item self-report measure of the general ten-dency to engage in strategies to avoid distressing thoughts andmental images. Examples of items are “I have thoughts that I try toavoid” and “I often do things to distract myself from my thoughts.”CAQ items are rated on a 5-point Likert scale in terms of how typicalthey are of an individual from 1 (not at all typical) to 5 (completelytypical). Scores on the CAQ range from 25 to 125. The CAQ hasexcellent internal consistency (˛ = .95), good test–retest reliability(r = .85), and has demonstrated convergent and divergent validity(Sexton & Dugas, 2008).
The Modified Behavioral Approach Test (modified BAT) wasdeveloped for the present study to assess acute cognitive avoidance
of a mental image of a WCS. BATs are commonly used to assessfear and avoidance of tangible stimuli (e.g., spiders; Harrington &Antony, 2009). For the modified BAT, participants imagined theirWCS coming true for 30 s. Then they rated their fear in response
562 K. Fracalanza et al. / Journal of Anxiety Disorders 28 (2014) 559–569
Table 1Sample characteristics by condition (N = 57).
Characteristic Consistent exposure (CE)(n = 19)
Varied exposure (VE)(n = 19)
Neutral control (NC)(n = 19)
Age in yearsa – M (SD) 28.35 (10.06) 33.79 (14.36) 39.00 (14.91)Sex – frequency (%)
Female 15 (78.95%) 15 (78.95%) 15 (78.95%)Male 4 (21.05%) 4 (21.05%) 4 (21.05%)
Race/ethnicity – frequency (%)White 10 (52.63%) 6 (31.58%) 13 (68.43%)East Asian 4 (21.06%) 5 (26.32%) 0 (0.00%)Black 2 (10.53%) 3 (15.79%) 1 (5.26%)Mixed Race 0 (0.00%) 1 (5.26%) 3 (15.79%)Other Ethnicity 0 (0.00%) 2 (10.53%) 1 (5.26%)South Asian 1 (5.26%) 0 (0.00%) 1 (5.26%)South East Asian 1 (5.26%) 1 (5.26%) 0 (0.00%)Latin American 1 (5.26%) 1 (5.26%) 0 (0.00%)
Highest educationb – frequency (%)High school diploma 5 (26.32%) 2 (10.53%) 5 (27.78%)College diploma 1 (5.26%) 3 (15.79%) 5 (27.78%)Bachelor’s degree 9 (47.37%) 10 (52.63%) 8 (44.44%)Master’s degree 3 (15.79%) 3 (15.79%) 0 (0.0%)Doctorate degree 1 (5.26%) 1 (5.26%) 0 (0.0%)
Employment status – frequency (%)Not working (including student) 9 (47.37%) 7 (36.84%) 6 (31.58%)Employed part-time 4 (21.05%) 8 (42.11%) 7 (36.84%)Employed full-time 6 (31.58%) 4 (21.05%) 6 (31.58%)
Marital status – frequency (%)Single 11 (57.89%) 13 (68.42%) 14 (73.68%)Married/common-law 6 (31.58%) 5 (26.32%) 3 (15.79%)Divorced/widowed 2 (10.53%) 1 (5.26%) 2 (10.53%)
Recruitment source – frequency (%)Toronto community 8 (42.11%) 10 (52.63%) 12 (63.16%)Ryerson University community 11 (57.89%) 9 (47.37%) 7 (36.84%)
Number of additional diagnoses M (SD) 1.11 (0.81) 0.79 (0.92) 1.00 (1.16)
Note: There were no statistically significant differences between the CE, VE, or NC conditions on any of the variables reported above.a hose n
level.
taadi
iigiit2((
(nctTt
Pcnwpbti
Two participants in the CE condition and two participants in the NC condition cb One participant in the NC condition chose not to report their highest education
o the mental image on an 11-point Likert scale from 0 (no feart all) to 10 (the most fear I have ever felt), and their desire tovoid imagining the WCS on an 11-point Likert scale from 0 (noesire to avoid imagining) to 10 (the highest possible desire to avoid
magining).The Intolerance of Uncertainty Scale (IUS; Buhr & Dugas, 2002)
s a 27-item measure of negative beliefs about uncertainty and itsmplications. Examples of items are: “When I am uncertain I can’to forward” and “Uncertainty keeps me from living a full life.” IUStems are rated on a 5-point Likert scale in terms of how character-stic they are of an individual from 1 (not at all characteristic of me)o 5 (entirely characteristic of me). Scores on the IUS range from7 to 135. The IUS has demonstrated excellent internal consistency˛ = .95), good test–retest reliability (r = .74), and construct validitySexton & Dugas, 2009).
The Self-Assessment Manikin – Anxious Arousal SubscaleSAM-A; Bradley & Lang, 1994) is a pictorial measure that depictsine manikins ranging in appearance on a continuum from veryalm (coded as 1) to very anxious (coded as 9). Respondents circlehe manikin that reflects their level of subjective anxious arousal.he SAM-A is sensitive to changes in anxious arousal between writ-en disclosure sessions (e.g., Sloan et al., 2005, 2007).
Linguistic Inquiry and Word Count, 2007 (LIWC2007;ennebaker, Booth, & Francis, 2007) is software that analyzes theontent of written scripts. For example, LIWC2007 calculates theumber of words in selected categories (e.g., negative emotionords, first person words, present tense words), and provides the
roportion of words in specific categories relative to the total num-er of words in a script. LIWC2007 was used in the present studyo assess the degree to which participants followed the writingnstructions of their assigned condition.ot to report their age.
3.3. Procedure
3.3.1. ScreeningEligibility was assessed over the telephone using the MINI
Screen, modules of the MINI Interview, and questions that assessedthe other eligibility criteria (see Section 3.1).
3.3.2. BaselineParticipants completed the PSWQ, WAQ-A, CAQ, and IUS. The
experimenter worked with each participant to identify their WCSusing the downward arrow technique, a method of identifyingcore fears (Burns, 1980). For example, if a participant indicatedthat he/she worried most about school, then he/she was asked toarticulate what he/she feared most about school. The participant’sresponse was then followed with the question, “What would beso bad about that?” until the WCS was identified. Participants werethen asked several questions to concretize the scenario (e.g., “Whatdo you see and hear?”). Next, participants practiced imagining aneutral stimulus, after which they completed the modified BAT andSAM-A.
3.3.3. Random assignment to experimental conditionsAfter the baseline assessment, participants were randomly
assigned to one of three conditions: consistent exposure (CE), var-ied exposure (VE), or neutral control (NC). The writing instructionsused in the present study were based on Goldman et al.’s (2007)modified version of the original written disclosure procedure by
Pennebaker and Beall (1986). Goldman et al.’s instructions dif-fered from the original written disclosure paradigm in that theyasked participants to write about a hypothetical WCS in the presenttense, as opposed to a past stressful or traumatic event, so that the
nxiet
pstshsvwiPbtR
3acutt(2
3atssAtidfrt
3ahnie
3
ta
4
4
wetptpttcwOfw
K. Fracalanza et al. / Journal of A
rocedure resembled exposure practices for GAD. In the presenttudy, participants wrote continuously for 20 min per session onhree consecutive days, based on a meta-analysis showing thattudies using at least 15 min of writing on three or more daysad larger effects on psychological outcomes than studies using amaller “dose” of written disclosure (Frattaroli, 2006). Indeed, pre-ious studies have shown that using three 20-minute sessions ofritten disclosure can produce significant and large improvement
n psychological symptoms (e.g., ds = 1.11–1.17; Sloan et al., 2007).articipants also completed the SAM-A three times per session:efore writing, midway through writing, and after writing. Par-icipants completed each writing session alone in a laboratory atyerson University.
.3.3.1. Consistent exposure (CE). At each session, participants weresked to write a story about the same WCS (determined at baseline)oming true. As per theory on imaginal exposure and CBT man-als for GAD, participants were asked to write with reference toheir emotional experience, in the first person, and in the presentense to reduce emotional and temporal distance from the WCSsee Cuthbert et al., 2003; Dugas & Robichaud, 2007; Goldman et al.,007).
.3.3.2. Varied exposure (VE). Participants were asked to writebout three different scenarios within one life domain (e.g., health)o make the procedure comparable to past studies on in vivo expo-ure, wherein participants were exposed to multiple forms of theame stimulus (e.g., several different spiders; Rowe & Craske, 1998).t the first session, participants were asked to write a story about
heir WCS coming true. At the second and third sessions, new WCSsn the same life domain as the first WCS were identified using theownward arrow technique, and participants wrote about a dif-erent WCS at each session. Participants were asked to write witheference to their emotional experience, in the first person, and inhe present tense.
.3.3.3. Neutral control (NC). At each session, participants weresked to write about what they would do if they found out that theyad a day off. They were asked to write in a factual manner, witho references to emotions or opinions, as per the control procedure
n other written exposure studies (e.g., Frattaroli, 2006; Goldmant al., 2007).
.3.4. 1-Week follow-upOne week after the third writing session, participants completed
he same questionnaires and modified BAT that they had completedt baseline.
. Results
.1. Preliminary analyses
A two-part manipulation check was performed to determinehether participants followed the writing instructions. First, the
xperimenter read each script to check if participants wrote abouthe topic that they were assigned to, and found that each partici-ant did. Second, each participant’s three scripts were submittedogether to LIWC2007. This allowed for an assessment of the pro-ortion of negative emotion words, first person words, and presentense words to total words used by participants in each condi-ion, which was of interest because participants in the exposureonditions were instructed to include these elements in their
riting whereas participants in the control condition were not.ne-way analyses of variance (ANOVAs) showed a significant dif-erence between conditions in the proportion of negative emotionords to total words, F(2, 54) = 13.45, p < .01, �2
p = .33. Bonferroni
y Disorders 28 (2014) 559–569 563
posthoc tests (used in all pairwise contrasts in the present study)showed that participants in the CE (M = 3.42, SD = 2.85) and VE(M = 3.30, SD = 1.71) conditions used significantly more negativeemotion words than did participants in the NC condition (M = 0.50,SD = 0.75), ps < .05, with no significant difference between the CEand VE conditions. Participants in the CE (M = 8.30, SD = 2.24) andVE (M = 8.38, SD = 3.18) conditions also used a greater proportionof first person words to total words than did participants in theNC condition (M = 6.48, SD = 2.95), although these differences werenot statistically significant, F(2, 54) = 2.75, p = .07, �2
p = .09. Simi-larly, participants in the CE (M = 3.05, SD = 2.76) and VE (M = 3.13,SD = 1.11) conditions used a greater proportion of present tensewords to total words compared to participants in the NC condi-tion (M = 2.06, SD = 0.94), but these differences also did not reachstatistical significance, F(2, 54) = 2.09, p = .13, �2
p = .07.The three conditions were compared on demographic charac-
teristics and outcome measures at baseline (see Tables 1 and 2).ANOVAs and chi-square tests showed that there were no base-line differences between conditions on any variable, except for theWAQ-A, F(2, 53) = 4.75, p = .01, �2
p = .15. Participants in the VE con-dition produced significantly lower WAQ-A scores at baseline thandid participants in the CE or NC conditions, ps < .05; the scores ofCE and NC participants did not differ significantly (see Table 2). Theinternal consistencies of all outcome measures used in the presentstudy were calculated at baseline and at follow-up and were foundto be high (˛s = .70–.96).
4.2. Objective 1: Effects of written exposure on GAD symptoms
Table 2 displays the means and standard deviations for meas-ures of GAD symptoms and GAD-relevant processes separated bycondition. It was hypothesized that participants in the CE andVE conditions would display greater improvement in GAD symp-toms than would participants in the NC condition. A mixed ANOVAon PSWQ scores showed no main effect of Time, F(1, 53) = 2.11,p = .15, �2
p = .04, and no main effect of Condition, F(2, 53) = .15,p = .86, �2
p = .01. However the interaction of Time and Conditionwas significant, F(2, 53) = 4.84, p = .01, �2
p = .15. Participants in theCE condition displayed a significant reduction in PSWQ scores frombaseline to follow-up, F(1, 53) = 19.32, p < .01, �2
p = .27 (d = .91);whereas participants in the VE and NC conditions did not, F(1,53) = 0.03, p = .87, �2
p < .01 (d = .04), and F(1, 53) = 1.50, p = .23, �2p =
.03 (d = .35), respectively. There were no between-group differ-ences in PSWQ scores at follow-up, F(2, 53) = 0.78, p = .46, �2
p = .03.A mixed ANOVA on WAQ-A scores showed no main effect of
Time, F(1, 53) = 0.83, p = .37, �2p = .02, and no main effect of Condi-
tion, F(2, 53) = 0.87, p = .42, �2p = .03, but a significant interaction,
F(2, 53) = 7.25, p < .01, �2p = .22. Participants in the CE and NC con-
ditions displayed a significant reduction in WAQ-A scores frombaseline to follow-up, F(1, 53) = 19.59, p < .01, �2
p = .27 (d = .84), andF(1, 53) = 9.47, p < .01, �2
p = .15 (d = .79), respectively; whereas par-ticipants in the VE condition did not, F(1, 53) = 0.54, p = .46, �2
p =.01 (d = .25). There were no between-group differences in WAQ-Ascores at follow-up, F(2, 53) = 0.42, p = .66, �2
p = .02.
4.3. Objective 2: Effects of written exposure on GAD-relatedprocesses
It was hypothesized that participants in the CE and VE condi-tions would show greater improvement in GAD-related processesthan would participants in the NC condition. A mixed ANOVA
on CAQ scores revealed no main effect of Time, F(1, 53) = 1.98,p = .17, �2p = .04, no main effect of Condition, F(2, 53) = 0.34, p = .71,�2
p = .01, and no interaction effect, F(2, 53) = 0.07, p = .93, �2p < .01.
In the present study, even when omnibus interaction effects were
564 K. Fracalanza et al. / Journal of Anxiety Disorders 28 (2014) 559–569
Table 2Means and standard deviations on GAD symptoms and processes by condition (N = 57).
Measure Time Consistent exposure (CE)(n = 19)
Varied exposure (VE)(n = 19)
Neutral control (NC)(n = 19)
PSWQ Baseline 66.27 (7.87) 61.74 (8.31) 61.26 (8.04)Follow-up 58.74 (7.93) 61.45 (8.18) 59.05 (9.54)
WAQ-Aa Baseline 22.05 (4.06) 18.21 (4.40) 21.63 (4.34)Follow-up 17.71 (4.90) 18.92 (4.18) 18.68 (6.17)
IUS Baseline 81.56 (17.02) 85.25 (21.42) 85.30 (23.54)Follow-up 70.00 (19.92) 85.33 (22.48) 80.79 (24.15)
CAQ Baseline 76.00 (15.28) 74.05 (14.99) 70.47 (20.67)Follow-up 75.32 (13.28) 75.45 (13.22) 70.84 (20.69)
BAT-fear Baseline 7.97 (1.57) 6.68 (2.99) 7.24 (2.37)Follow-up 5.90 (2.55) 5.47 (3.08) 6.45 (2.02)
BAT-avoidance Baseline 8.11 (2.06) 7.76 (3.72) 8.26 (1.84)Follow-up 5.21 (3.29) 6.45 (3.50) 7.24 (1.86)
N naireI AT = M
nHtc
sSaip�Hnfi�ri
Wnipd5aawt
F5Ftft�ra
4e
4
t
ote: GAD = generalized anxiety disorder; PSWQ = Penn State Worry QuestionUS = Intolerance of Uncertainty Scale; CAQ = Cognitive Avoidance Questionnaire; B
a Significant differences between scores at baseline.
onsignificant, they were followed-up with pairwise contrasts (seeancock & Klockars, 1996 for a detailed rationale). Pairwise con-
rasts of the interaction effect on CAQ scores showed no significanthange in any condition over time.
Acute cognitive avoidance was conceptualized in the presenttudy as fear and avoidance of an imagined WCS during the BAT.ince fear and avoidance were each rated separately, they werenalyzed using two discrete ANOVAs. A mixed ANOVA on fear ofmagining the WCS showed no main effect of Time, F(1, 53) = 0.23,= .63, �2
p < .01, no main effect of Condition, F(2, 53) = 1.11, p = .34,2p = .04, and no interaction effect, F(2, 53) = 0.67, p = .52, �2
p = .03.owever, only participants in the CE condition reported a sig-ificant decrease in fear of imagining the WCS from baseline to
ollow-up, F(1, 53) = 7.59, p < .01, �2p = .13 (d = .97), whereas partic-
pants in the VE and NC conditions did not, F(1, 53) = 2.74, p = .10,2p = .05 (d = .29), and F(1, 54) = 1.30, p = .26, �2
p = .02 (d = .29),espectively. There were no between-group differences on fear ofmagining the WCS at follow-up, F(2, 53) = 0.63, p = .54, �2
p = .02.A mixed ANOVA on self-rated desire to avoid thinking about the
CS showed no main effect of Time, F(1, 53) = 2.12, p = .15, �2p = .04,
o main effect of Condition, F(2, 53) = 0.81, p = .45, �2p = .03, and no
nteraction effect, F(2, 53) = 1.72, p = .19, �2p = .06. However, only
articipants in the CE condition reported a significant decrease inesire to avoid imagining the WCS from baseline to follow-up, F(1,3) = 14.03, p < .01, �2
p = .21 (d = .80), whereas participants in the VEnd NC conditions did not, F(1, 53) = 2.90, p = .10, �2
p = .05 (d = .33),nd F(1, 53) = 1.66, p = .20, �2
p = .03 (d = .55), respectively. Thereere no between-group differences on desire to avoid imagining
he WCS at follow-up, F(2, 53) = 1.91, p = .16, �2p = .07.
A mixed ANOVA on IUS scores showed no main effect of Time,(1, 53) = 1.84, p = .18, �2
p = .03, no main effect of Condition, F(2,3) = 1.12, p = .33, �2
p = .04, and a nonsignificant interaction effect,(2, 53) = 3.03, p = .057, �2
p = .10. Only participants in the CE condi-ion displayed a significant decrease in IUS scores from baseline toollow-up, F(1, 53) = 11.65, p < .01, �2
p = .18 (d = .70); whereas par-icipants in the VE and NC conditions did not, F(1, 53) < 0.01, p = .98,2p < .01 (d < .01), and F(1, 53) = 1.54, p = .22, �2
p = .03 (d = .40),espectively. There were no between-group differences on the IUSt follow-up, F(2, 53) = 2.36, p = .10, �2
p = .08.
.4. Objective 3: Effects of written exposure on indicators ofmotional processing
.4.1. Initial anxious arousalSee Table 3 for SAM-A scores at each session separated by condi-
ion. It was hypothesized that during the first session, individuals
; WAQ-A = Worry and Anxiety Questionnaire-Associated Symptoms Subscale;odified Behavioral Approach Test.
in the CE and VE conditions would show an increase in anxiousarousal from baseline to peak anxious arousal, whereas individualsin the NC condition would not. Based on previous research (e.g.,van Minnen & Foa, 2006), peak anxious arousal was operational-ized as the higher of the two SAM-A scores following baseline. Themixed ANOVA on initial anxious arousal showed no main effect ofTime, F(1, 53) = 3.13, p = .08, �2
p = .06, and no main effect of Con-dition, F(2, 53) = 2.35, p = .11, �2
p = .08, but a significant interactioneffect, F(2, 53) = 14.82, p < .01, �2
p = .36. Participants in the CE andVE conditions displayed a significant increase in SAM-A scores frombaseline to peak, F(1, 53) = 20.87, p < .01, �2
p = .28 (d = 1.05), andF(1, 53) = 19.60, p < .01, �2
p = .27 (d = 1.14), respectively. In contrast,participants in the NC condition showed a significant decrease inSAM-A scores from baseline to peak during the first writing session,F(1, 53) = 4.89, p = .03, �2
p = .08 (d = .47).
4.4.2. Between-session anxious arousalIt was also predicted that individuals in the CE and VE condi-
tions would show a decrease in peak anxious arousal from the firstto third session, whereas individuals in the NC condition would not.There was no main effect of Time, F(2, 106) = 0.74, p = .48, �2
p = .01,but a significant effect of Condition, F(2, 53) = 8.38, p < .01, �2
p = .24,such that on average participants in the CE and VE conditions dis-played higher peak anxious arousal than did participants in the NCcondition, ps < .01, with no difference between the exposure con-ditions. The interaction effect was not significant, F(4, 106) = 1.52,p = .20, �2
p = .05. Although all participants displayed a decline inpeak anxious arousal across writing sessions, this decline wasonly significant for participants in the CE condition, F(2, 52) = 5.82,p < .01, �2
p = .18 (d = .92), and was not significant for participants inthe VE condition, F(2, 52) = 2.61, p = .08, �2
p = .09 (d = .49), or the NCcondition, F(2, 52) = 0.62, p = .54, �2
p = .02 (d = .03).
4.4.3. Associations between emotional processing and GADoutcomes
It was hypothesized that indicators of emotional processingwould be associated with improvement on GAD symptoms andrelated processes in the exposure conditions. Given that there wereno a priori hypotheses about differences between the exposure con-ditions, correlations were performed in a sample combining bothexposure conditions (see Table 4). In addition, given that individ-uals in the CE condition showed improvement on most outcomes
whereas individuals in the VE condition did not, correlations wereperformed within the CE and VE conditions separately.Pearson correlations between change scores on indicators ofemotional processing and change scores on GAD symptoms and
K. Fracalanza et al. / Journal of Anxiety Disorders 28 (2014) 559–569 565
Table 3Means and standard deviations on subjective anxious arousal by condition (N = 57).
Writing session Measurement Consistent exposure (CE)(n = 19)
Varied exposure (VE)(n = 19)
Neutral control (NC)(n = 19)
First Baseline 3.89 (1.60) 3.79 (1.55) 4.42 (2.27)Peak intensity 6.11 (1.70) 5.95 (2.07) 3.37 (2.11)
Second Baseline 3.47 (2.27) 4.05 (1.62) 3.68 (1.97)Peak intensity 4.95 (2.09) 5.26 (2.02) 3.05 (2.09)
Third Baseline 2.84 (1.39) 3.21 (1.78) 3.63 (1.86)Peak intensity 4.74 (1.82) 5.00 (2.13) 3.32 (2.06)
Note: Subjective anxious arousal was measured with the Self-Assessment Manikin-Anxious Arousal Scale (SAM-A). SAM-A ratings range from 1 (very calm) to 9 (very anxious).Peak intensity = the highest level of anxiety indicated by participants on the SAM-A after baseline (either midwriting or postwriting).
Table 4Correlations between study measures in the exposure conditions (N = 38).
Initial Anx BtwSes Anx PSWQ WAQ-A CAQ BAT-F BAT-A IUS Neg Emo First person PresTen
InitialAnx 1.00 −.36* −.34* −.13 .03 −.22 −.02 −.17 −.39** −.38** −.37*
BtwSesAnx 1.00 .27* .20 −.08 −.06 .07 .15 .47** .40** .53**
PSWQ 1.00 .57** −.26 .42** .50** .52** .31* .18 .27WAQ-A 1.00 −.07 .36** .41** .42** .17 .01 .15CAQ 1.00 −.14 −.34* −.19 .19 −.20 .22BAT-F 1.00 .63** .20 .07 .01 .05BAT-A 1.00 .32* −.02 −.08 .01IUS 1.00 .16 −.13 .16NegEmo 1.00 .49** .85**
First person 1.00 .48**
PresTen 1.00
Note: InitialAnx = change in initial anxious arousal on the Self-Assessment Manikin-Anxious Arousal Subscale (SAM-A); BtwSesAnx = change in anxious arousal across ses-sions on the SAM-A; PSWQ = change on the Penn State Worry Questionnaire; WAQ-A = change on the Worry and Anxiety Questionnaire-Associated Symptoms Subscale;CAQ = change on the Cognitive Avoidance Questionnaire; BAT-F = change in fear in response to imagining a threatening scenario on a Modified Behavioral Approach Test;BAT-A = change in desire to avoid imagining a threatening scenario on a Modified Behavioral Approach Test; IUS = change on the Intolerance of Uncertainty Scale; NegEmo = theproportion of negative emotion words to total words; First person = the proportion of first person words to total words; PresTen = the proportion of present tense words tot
pditpbae(BaWaIrowt
4i
to
ifaws
otal words.* p < .05.
** p < .01.
rocesses were calculated.2 Greater activation of anxious arousaluring the first session was significantly associated with greater
mprovement in worry in both exposure conditions examinedogether (r = −.34, p = .02) and in the CE condition alone (r = −.53,= .01), but not in the VE condition alone (r = −.18, p = .23). Likewise,etween-session decreases in anxious arousal were significantlyssociated with decreases in worry in both exposure conditionsxamined together (r = .27, p = .05) and in the CE condition aloner = .38, p = .05), but not in the VE condition alone (r = .13, p = .30).etween-session decreases in anxious arousal were also moder-tely correlated with improvement in desire to avoid imagining theCS (r = .39, p = .05) and IU (r = .34, p = .08) in the CE condition alone,
lthough the latter association was not statistically significant.ndicators of emotional processing were not associated with GAD-elated processes when both exposure conditions were combinedr in the VE condition alone. Indicators of emotional processingere also not associated with changes in GAD-associated symp-
oms in any subsample of participants.
.5. Objective 4: Associations between adherence to writingnstructions and GAD outcomes
To explore the extent to which adherence to writing instruc-ions for exposure was related to improvement on GAD-relatedutcomes, Pearson correlations were calculated between the
2 Change scores for initial activation were calculated by subtracting peak anx-ous arousal from baseline anxious arousal during the first session. Change scoresor between-session anxious arousal were calculated by subtracting peak anxiousrousal during the third writing session from peak anxious arousal during the firstriting session. Change scores for GAD symptoms and processes were calculated by
ubtracting scores at 1-week follow-up from scores at baseline on each measure.
writing adherence indicators determined by LIWC2007 (proportionof negative emotion words to total words, first person words to totalwords, and present tense words to total words) and change scoreson GAD symptoms and processes (see Table 4). As with objective 3,these analyses were run in both exposure conditions combined andeach exposure condition separately. Greater reference to negativeemotions was associated with greater improvement in worry whenboth exposure conditions were examined together (r = .31, p = .05),and in the CE condition alone (r = .47, p < .05), but not in the VE con-dition alone (r = .09, p = .35). In addition, greater use of present tensewords was significantly associated with improvement in worry inthe CE condition alone (r = .50, p = .01), but not in the VE conditionalone (r = −.18, p = .24). Use of first person language was not signif-icantly related to GAD outcomes in any subsample. In addition, nowriting adherence indicator was significantly associated with anyGAD outcome other than worry.
5. Discussion
The first objective of the present study was to examine theeffects of consistent exposure, varied exposure, and a control con-dition on GAD symptoms using an adapted written disclosureprocedure consisting of three 20-minute sessions. We found thatconsistent exposure produced large significant decreases in worry,the cardinal feature of GAD, but varied exposure and neutral controlwriting did not. Like our study, Goldman et al. (2007) used a briefwriting procedure to facilitate exposure, and found that writingrepeatedly about the same feared scenario led to large improve-
ments in worry from baseline to 2-week follow-up. In addition,studies that have examined longer courses of imaginal exposurefor GAD have shown that imaginal exposure improves worry frompre- to posttreatment (Provencher et al., 2004) and compared to
5 nxiet
a2pfitwt
ws(adeecricpapwc“is
si1cpiioGbtepadaMtcmls&otfMioe
cGviass
66 K. Fracalanza et al. / Journal of A
waitlist condition (Hoyer et al., 2009; Wolitzky-Taylor & Telch,010). Similarly, Sloan and colleagues’ (2005) results from a sam-le who had experienced trauma are congruent with the presentndings. They showed that writing about the same trauma memoryhree times led to a large significant reduction in PTSD symptomshereas writing about different trauma memories or a neutral topic
hree times did not.On the other hand, there are some differences between previous
ork and our finding that worry improved over time with con-istent exposure. First, the abovementioned study by Sloan et al.2005) found between-group differences in PTSD symptoms at 4-nd 8-week follow-ups, whereas we did not detect between-groupifferences in worry at 1-week follow-up. This may be due to Sloant al.’s larger sample and subsequent greater power to detect differ-nces between groups at follow-up assessments. Our finding thatonsistent exposure improved worry over time also differs fromesults suggesting an advantage to varying the exposure stimulusn specific phobia samples (e.g., Rowe & Craske, 1998). The dis-repancy between the present results and the findings on specifichobia could be because the situations that people with GAD fearre more complex and typically have many different meanings anderceived consequences (e.g., Lang, 1979; Wells & Carter, 2006);hereas, the fears of people with specific phobias tend to be more
ircumscribed (Hood & Antony, 2012). Thus, in order to optimizeemotional processing” (Foa et al., 2006) for individuals with GAD,t may be necessary to focus on a single feared scenario for severalessions before targeting a different one.
Another finding relevant to our first objective was that con-istent exposure and neutral control writing led to significantmprovements in GAD-associated symptoms from baseline to-week follow-up, whereas GAD-associated symptoms did nothange in the varied exposure condition. At a first glance, theositive effect of the control writing procedure is counterintu-
tive; however, like the present study, Goldman et al. (2007) alsonstructed participants to write about what they would do on a dayff as a control condition and found that the control task reducedAD-associated symptoms from baseline to 2-week follow-up. Toetter understand this result we examined the content of the scriptshat participants in our control condition wrote. We found thatvery participant in the control condition described engaging inleasurable activities (e.g., watching movies, socializing), and 47%lso wrote about engaging in goal-directed activities (e.g., cleaning,oing schoolwork). Past research has shown that partaking in suchctivities can reduce depressive symptoms (Dimidjian, Barrera,artell, Munoz, & Lewinsohn, 2011), some of which overlap with
he associated symptoms of GAD (e.g., difficulty sleeping, difficultyoncentrating, fatigue). It may be that writing about pleasure orastery activities improved GAD-associated symptoms because it
ed people to do some of the things that they wrote about. Indeed,tarting activities is easier after visualizing oneself doing so (PhamTaylor, 1999). In addition, some studies have compared the effects
f writing about traumatic events to writing about more positiveopics (e.g., accomplishing life goals) and have found that bothorms of writing produce equal health benefits (King, 2001; King &
iner, 2000). King and colleagues theorized that writing about top-cs like meeting goals may be helpful because it increases awarenessf goals and promotes a sense of accomplishment–both possiblexplanations for the positive effect of our control writing task.
The second purpose of the present study was to examinehanges in GAD-related processes that may occur with exposure.iven that the writing procedure was designed to expose indi-iduals to vivid mental images of their worst fears coming true,
t was hypothesized that such exposure would decrease chronicnd acute cognitive avoidance. Neither consistent nor varied expo-ure improved chronic cognitive avoidance. However, a largeignificant decrease in acute cognitive avoidance was observed iny Disorders 28 (2014) 559–569
individuals who engaged in consistent exposure. Specifically, thisgroup reported less fear and avoidance of imagining their worstcase scenario coming true following three sessions of writtenexposure to it; effects that were not observed in the other con-ditions. This finding supports the notion that exposure improves atleast one form of cognitive avoidance (Dugas & Robichaud, 2007;Sibrava & Borkovec, 2006), although more exposure sessions maybe needed to modify chronic cognitive avoidance.
Individuals who engaged in consistent exposure also displayeda large significant improvement in IU from baseline to 1-weekfollow-up, even though the procedure was not designed to explic-itly expose individuals to uncertainty. Dugas and Robichaud (2007)proposed that imaginal exposure may modify a person’s generaltolerance for uncertainty through exposure to the possibility that afeared scenario could happen, or through exposure to the uncer-tain elements contained within a feared scenario. Participants mayalso have benefitted from repeatedly confronting uncertainty abouthow they would feel upon deliberately elaborating on their worstfear coming true.
Improvement in cognitive avoidance and IU may be mechanismsof the therapeutic effects of exposure for GAD. In a posthoc analy-sis, we found that improvement in acute cognitive avoidance and IUwere each moderately correlated with improvement in worry andGAD-associated symptoms. This was true when correlations wereexamined in both exposure conditions together, and in the consis-tent exposure condition alone. Although the directionality of theassociation between cognitive processes and GAD symptoms couldnot be established with the present study’s design, future studiesshould seek to determine whether changes in cognitive avoidanceand IU precede changes in GAD symptoms during exposure.
The third aim of this study was to examine the degree to whichtenets of emotional processing theory (EPT) are supported in thecontext of exposure for GAD. During the first session, partici-pants in the exposure conditions showed an increase in anxiousarousal, suggesting that writing about a feared scenario success-fully activated fear structures. Higher levels of initial activationwere associated with greater improvements in worry, which is con-sistent with EPT and previous research (see Foa et al., 2006 fora review). On the other hand, participants in the control condi-tion showed a decrease in anxious arousal during the first session,which may have been produced by first asking them to thinkabout their worst case scenario for the baseline BAT, and thenasking them to write about a neutral topic. It also was predictedthat those in the exposure conditions would display a decrease inpeak anxious arousal across exposure sessions relative to those inthe control condition. This was observed only in individuals whowrote about the same feared scenario at each session, but not inindividuals who wrote about a different feared scenario at eachsession, or individuals in the control condition. In line with EPT,greater between-session reductions in anxious arousal coincidedwith greater improvement in worry.
These findings suggest that emotional processing was more pro-nounced for individuals who engaged in consistent exposure thanfor individuals who engaged in varied exposure. It may be thatconsidering the same worst case scenario several times gave peoplethe opportunity to arrive at a new and potentially less threateningunderstanding of their greatest fear. Indeed, obtaining a new per-spective on a feared scenario is a goal of exposure for GAD (Zinbarget al., 2006). On the other hand, individuals who were asked towrite about a new feared scenario at each exposure session maynot have benefitted because they did not have enough time tochange their perspective on any one scenario. A separate expla-
nation is that additional time spent writing about the same worstcase scenario may have allowed people to examine this situation ina more detailed and specific (i.e., concrete) manner. Concrete think-ing evokes emotion-laden mental imagery (Paivio & Marschark,
nxiet
11ae&(acg
bGefiftlorm(taImbnotwWfcfprtt
inetaiemZimtadRtaptalcs
w
K. Fracalanza et al. / Journal of A
991), which is thought to encourage emotional processing (Stöber,998, 2000). Further, training people to think concretely (versusbstractly) about situations has been found to promote moreffective solutions to interpersonal problems (Pawluk, Koerner,Antony, 2011), and to reduce emotional reactivity to stressors
Watkins, Moberly, & Moulds, 2008). In sum, writing consistentlybout the same scenario may have allowed individuals to betteroncretize their fear, thereby promoting emotional processing, theeneration of solutions, and improved emotional responding.
The fourth goal of this study was to examine the associationetween adherence to writing instructions and improvement inAD outcomes. We found that greater use of references to negativemotions was associated with larger improvement in worry. Thisnding is consistent with Sloan et al.’s (2007) experiment, which
ound that: (1) writing about a trauma with as much referenceo emotion as possible led to greater improvement on psycho-ogical outcomes than did writing about a trauma with a focusn meaning-making, or writing about a day off, and (2) greatereferences to negative emotions significantly predicted improve-ent on psychological outcomes. Greenberg, Wortman, and Stone
1996) theorized that experiencing negative emotions and comingo see oneself as capable of managing these emotions is a mech-nism through which written exposure improves mental health.ndeed, a sense of mastery over one’s ability to handle emotions
ay be especially important for individuals with GAD, who haveeen found to experience more intense emotions and to react moreegatively to their own emotional experience than people with-ut GAD (Mennin, Heimberg, Turk, & Fresco, 2005). We also foundhat the use of present tense language was positively associatedith improvement in worry in the consistent exposure condition.riting in the present tense may reduce temporal distance from
eared mental imagery, potentially facilitating the benefits of con-rete thinking described earlier. In addition, given that worry is aorm of future-oriented thinking (Brown, O’Leary, & Barlow, 1993),ractice using the present-moment perspective may be helpful foreducing it. Indeed, GAD treatments that promote present-orientedhinking through strategies such as mindfulness have been foundo improve worry (e.g., Roemer & Orsillo, 2007).
With respect to clinical implications, our findings suggest thatndividuals with GAD should confront one feared worst case sce-ario at least three times before changing the focus of imaginalxposure. Practitioners should also instruct individuals with GADo reference their emotional reactions during imaginal exposurend to describe their feared scenario like it is really happeningn the here-and-now. These suggestions are in line with imaginalxposure instructions that are outlined in several GAD treatmentanuals (Dugas & Robichaud, 2007; Rygh & Sanderson, 2004;
inbarg et al., 2006); our findings simply provide empirical back-ng for these recommendations. It is of note that GAD treatment
anuals typically position imaginal exposure in the latter part ofreatment, after individuals have learned other strategies to man-ge worry, and suggest that people with GAD need to understand aetailed rationale for exposure prior to engaging in it (e.g., Dugas &obichaud, 2007; Zinbarg et al., 2006). In contrast, our results pointo the possibilities that individuals with GAD may be willing andble to start exposure early in treatment, and may not need amplereparation to benefit from it. Indeed, we found that dropout fromhe present study was very small and comparable across conditions,nd individuals engaged in exposure during their first visit to the
ab after receiving minimal rationale for doing so.3 In addition, theonsiderable age range of participants in our sample (18–61 years)uggests that writing repeatedly about a threatening scenario may3 Participants in all conditions were told that previous research suggests thatriting can have positive effects on health, as per Goldman et al. (2007).
y Disorders 28 (2014) 559–569 567
be a useful treatment strategy for younger and older individualswith GAD. Given that imaginal exposure is a component of someGAD protocols for adolescents (Hudson, Hughes, & Kendall, 2004)and older adults (Wetherell, Gatz, & Craske, 2003), future researchshould test the efficacy of imaginal exposure in these populations.
The present study had a number of noteworthy strengths. First,diagnostic interviews were administered to ensure that the find-ings would be applicable to clinical samples. Second, the studyanswers questions about the potential therapeutic effects of expo-sure on processes that are known to underpin GAD. Related to this,we developed a behavioral approach test that assessed fear andavoidance of a mental image of a personal worst case scenario forthe present study, and this represents a novel measure of acutecognitive avoidance that can be used in future research. Third, thestudy provided preliminary answers to questions about the extentto which the tenets of EPT apply to exposure for GAD. In termsof limitations, the observed power to detect differences betweenconditions on GAD symptoms and processes at follow-up was low;thus, although significant and large decreases were observed in theconsistent exposure condition, 1 week later, their scores were com-parable to those of participants assigned to the other conditions.In addition, the finding that participants in the consistent expo-sure and control conditions showed a reduction in GAD-associatedsymptoms from baseline to 1-week follow-up must be interpretedcautiously, as these participants began the experiment with higherGAD-associated symptoms than did participants in the varied expo-sure condition. Also, about half of the sample (47%) comprisedundergraduate students and most of the sample (79%) was female,thus, results may not generalize to all individuals with GAD. Thebrevity of the written exposure procedure used in the present studymay also be viewed as a limitation. However, the present study wasdesigned as a proof-of-concept experiment, which is an efficientway to address novel research questions before resources are allo-cated toward testing full-scale interventions (Clark-Carter, 2010).Further, there is considerable evidence that improvement in psy-chological symptoms can occur after only 3 brief sessions of writtendisclosure (Frattaroli, 2006).
An imperative future research direction is to conduct testsof other procedural variants of exposure for GAD. For example,Wolitzky-Taylor and Telch (2010) examined the degree to whichit is better for students with academic anxiety to write narrativesabout their fears or to audio-record these narratives, and foundthat both led to significant improvement in the general tendency toworry, although the effect was larger for audio-recorded narratives.This parameter and others, such as the duration of exposure, num-ber of exposures, and space between exposures should be tested inGAD samples to help identify the optimal exposure procedures forthis population. It is also important to consider that although theexposure target in the present study was mental imagery of fearedscenarios and the anxious arousal that accompanies this imagery(e.g., Stöber & Borkovec, 2002; Zinbarg et al., 2006); other expo-sure targets have been proposed for GAD. For instance, the ContrastAvoidance Model of Worry suggests that a core fear of individualswith GAD is experiencing a sudden change from a positive or neu-tral emotional state to a negative emotional state, or a negativeemotional contrast (Newman & Llera, 2011). This model is groundedin research suggesting that worry may promote and sustain a nega-tive emotional state (e.g., Brosschot, Gerin, & Thayer, 2006; Llera &Newman, 2010), and in this way, may help people to avoid suddenlyexperiencing negative emotions (Newman & Llera, 2011). In lightof this, future experiments may wish to examine whether writtendisclosure can be used to promote exposure to negative emotional
contrasts.In conclusion, although numerous studies have tested best prac-tices for the delivery of exposure for other forms of anxiety, weknow very little about how best to conduct imaginal exposure with
5 nxiet
pesociospth
A
R
R
A
A
A
A
B
B
B
B
B
B
C
C
C
D
D
D
D
D
D
D
D
F
414.
68 K. Fracalanza et al. / Journal of A
eople who have GAD. These individuals have typically not experi-nced the scenarios that they fear, and cannot plan to confront thesecenarios in vivo due to their catastrophic nature (e.g., losing a lovedne, being fired). For these reasons, imaginal exposure for GAD isomplex and often challenging to conduct. The present experimentnvestigated one procedural variant that could improve the effectf imaginal exposure for GAD, and tested theory about how expo-ure for GAD may have its effects. Additional studies on the optimalarameters of exposure for GAD and on its mechanisms will helpo bridge the gap between research and practice in this area, andave the potential to improve the effectiveness of GAD treatment.
cknowledgements
This research was funded by the Canadian Institutes of Healthesearch (CIHR).
eferences
bramowitz, J. S., Foa, E. B., & Franklin, M. E. (2003). Exposure and ritual preven-tion for obsessive–compulsive disorder: effects of intensive versus twice-weeklysessions. Journal of Consulting and Clinical Psychology, 71, 394–398.
lexander, M. J., Haugland, G., Lin, S. P., Bertollo, D. N., & McCorry, F. A. (2008). Mentalhealth screening in addiction, corrections and social service settings: validatingthe MMS. International Journal of Mental Health and Addiction, 6, 105–119.
merican Psychiatric Association. (2000). Diagnostic and statistical manual of mentaldisorders, 4th ed., text revision (DSM-IV-TR). Arlington, VA: Author.
merican Psychiatric Association. (2013). Diagnostic and statistical manual of mentaldisorders: fifth edition DSM-5. Arlington, VA: Author.
ehar, E., Alcaine, O., Zuellig, A. R., & Borkovec, T. D. (2003). Screening for generalizedanxiety disorder using the PSWQ: a receiver operating characteristic analysis.Journal of Behavior Therapy and Experimental Psychiatry, 34, 25–43.
radley, M. M., & Lang, P. J. (1994). Measuring emotion: the Self-Assessment Manikinand the semantic differential. Journal of Behavior Therapy and Experimental Psy-chiatry, 25, 49–59.
rosschot, J. F., Gerin, W., & Thayer, J. F. (2006). The perseverative cognition hypoth-esis: a review of worry, prolonged stress-related physiological activation, andhealth. Journal of Psychosomatic Research, 60, 113–124.
rown, T. A., O’Leary, T. A., & Barlow, D. H. (1993). Generalized anxiety disorder.In: D. H. Barlow (Ed.), Clinical handbook of psychological disorders: a step-by-steptreatment manual. (pp. 137–188). New York, NY: Guilford Press.
uhr, K., & Dugas, M. J. (2002). The Intolerance of Uncertainty Scale: psychometricproperties of the English version. Behaviour Research and Therapy, 40, 931–946.
urns, D. D. (1980). Feeling good: the new mood therapy. New York, NY: Morrow andCo.
lark-Carter, D. (2010). Quantitative Psychological Research (3rd ed.). New York, NY:Taylor & Francis.
hambless, D. L. (1990). Spacing of exposure sessions in treatment of agoraphobiaand simple phobia. Behavior Therapy, 21, 217–229.
uthbert, B. N., Lang, P. J., Strauss, C., Drobes, D., Patrick, C. J., & Bradley, M. M. (2003).The psychophysiology of anxiety disorder: fear memory imagery. Psychophysi-ology, 40, 407–422.
ear, B. F., Titov, N., Sunderland, M., McMillan, D., Anderson, T., Lorian, C., et al.(2011). Psychometric comparison of the GAD-7 and the PSWQ for measuringresponse during treatment of generalised anxiety disorder. Cognitive BehaviourTherapy, 40, 216–227.
imidjian, S., Barrera, M., Martell, C., Munoz, R. F., & Lewinsohn, P. M. (2011). Theorigins and current status of behavioral activation treatments for depression.Annual Review of Clinical Psychology, 7, 1–38.
ugas, M. J., Brillon, P., Savard, P., Turcotte, J., Gaudet, A., Ladouceur, R., et al. (2010). Arandomized clinical trial of cognitive-behavioral therapy and applied relaxationfor adults with generalized anxiety disorder. Behavior Therapy, 41, 46–58.
ugas, M. J., Freeston, M. H., Ladouceur, R., Rhéaume, J., Provencher, M., & Boisvert,J. (1998). Worry themes in primary GAD, secondary GAD, and other anxietydisorders. Journal of Anxiety Disorders, 12, 253–261.
ugas, M. J., Freeston, M. H., Provencher, M. D., Lachance, S., Ladouceur, R., & Gosselin,P. (2001). The Worry and Anxiety Questionnaire: validation in non-clinical andclinical samples. Journal de Thérapie Comportementale et Cognitive, 11, 31–36.
ugas, M. J., Gagnon, F., Ladouceur, R., & Freeston, M. H. (1998). Generalized anxi-ety disorder: a preliminary test of a conceptual model. Behaviour Research andTherapy, 36, 215–226.
ugas, M. J., Ladouceur, R., Léger, E., Freeston, M. H., Langolis, F., Provencher, M. D.,et al. (2003). Group cognitive-behavioral therapy for generalized anxiety dis-order: treatment outcome and long-term follow-up. Journal of Consulting andClinical Psychology, 71, 821–825.
ugas, M. J., & Robichaud, M. (2007). Cognitive-behavioral treatment for generalizedanxiety disorder: from science to practice. New York, NY: Routledge.
oa, E. B., Huppert, J. D., & Cahill, S. P. (2006). Emotional processing theory: an update.In: B. O. Rothbaum (Ed.), Pathological anxiety: emotional processing in etiology andtreatment (pp. 3–24). New York, NY: Guilford Press.
y Disorders 28 (2014) 559–569
Foa, E. B., & Kozak, M. J. (1986). Emotional processing of fear: exposure to correctiveinformation. Psychological Bulletin, 99, 20–35.
Frattaroli, J. (2006). Experimental disclosure and its moderators: a meta-analysis.Psychological Bulletin, 132, 823–865.
Fresco, D. M., Mennin, D. S., Heimberg, R. G., & Turk, C. L. (2003). Using the PennState Worry Questionnaire to identify individuals with generalized anxiety dis-order: a receiver operating characteristic analysis. Journal of Behavior Therapyand Experimental Psychiatry, 34, 283–291.
Greenberg, M. A., Wortman, C. B., & Stone, A. A. (1996). Emotional expression andphysical health: revising traumatic memories or fostering self-regulation? Jour-nal of Personality and Social Psychology, 71, 588–602.
Goldman, N., Dugas, M. J., Sexton, K. A., & Gervais, N. J. (2007). The impact of writ-ten exposure on worry: a preliminary investigation. Behavior Modification, 31,512–538.
Hancock, G. R., & Klockars, A. J. (1996). The quest for �: developments in multiplecomparison procedures in the quarter century since Games (1971). Review ofEducational Research, 66, 269–306.
Harrington, J., & Antony, M. M. (2009). Assessment of anxiety disorders. In: M. M.Antony, & M. B. Stein (Eds.), Oxford handbook of anxiety and related disorders (pp.277–291). New York, NY: Oxford University Press.
Hood, H. K., & Antony, M. M. (2012). Evidence-based assessment and treatmentof specific phobias in adults. In: T. E. Davis, T. H. Ollendick, & L.-G. Öst (Eds.),Intensive one-session treatment of specific phobias (pp. 19–42). New York, NY:Springer.
Hoyer, J., Beesdo, K., Gloster, A. T., Runge, J., Höfler, M., & Becker, E. S. (2009). Worryexposure versus applied relaxation in the treatment of generalized anxiety dis-order. Psychotherapy and Psychosomatics, 78, 106–115.
Hudson, J. L., Hughes, A. A., & Kendall, P. C. (2004). Treatment of generalized anxi-ety disorder in children and adolescents. In: P. Barrett, & T. H. Ollendick (Eds.),Handbook of interventions that work with children and adolescents (pp. 115–144).New York, NY: John Wiley & Sons.
King, L. A. (2001). The health benefits of writing about life goals. Personality andSocial Psychology Bulletin, 27, 798–807.
King, L. A., & Miner, K. N. (2000). Writing about the perceived benefits of trau-matic events: implications for physical health. Personality and Social PsychologyBulletin, 26, 220–230.
Lang, P. J. (1979). A bio-informational theory of emotional imagery. Psychophysiology,16, 495–512.
Lang, A. J., & Craske, M. G. (2000). Manipulations of exposure-based therapy to reducereturn of fear: a replication. Behaviour Research and Therapy, 38, 1–12.
Lecrubier, Y., Sheehan, D. V., Weiller, E., Amorim, P., Bonora, I., Harnett-Sheehan, K.,et al. (1997). The Mini International Neuropsychiatric Interview (MINI): a shortdiagnostic structured interview: reliability and validity according to the CIDI.European Psychiatry, 12, 224–231.
Llera, S. J., & Newman, M. G. (2010). Effects of worry on physiological and subjectivereactivity to emotional stimuli in generalized anxiety disorder and nonanxiouscontrol participants. Emotion, 10, 640–650.
Mennin, D. S., Heimberg, R. G., Turk, C. L., & Fresco, D. M. (2005). Preliminary evidencefor an emotion dysregulation model of generalized anxiety disorder. BehaviourResearch and Therapy, 43, 1281–1310.
Meyer, T. J., Miller, M. L., Metzger, R. L., & Borkovec, T. D. (1990). Developmentand validation of the Penn State Worry Questionnaire. Behaviour Research andTherapy, 28, 487–495.
Newman, M. G., Castonguay, L. G., Borkovec, T. D., Fisher, A. J., Boswell, J. F., Szkodny, L.E., et al. (2011). A randomized controlled trial of cognitive-behavioral therapy forgeneralized anxiety disorder with integrated techniques from emotion-focusedand interpersonal therapies. Journal of Consulting and Clinical Psychology, 79,171–181.
Newman, M. G., & Llera, S. J. (2011). A novel theory of experiential avoidancein generalized anxiety disorder: a review and synthesis of research suppor-ting a contrast avoidance model of worry. Clinical Psychology Review, 31, 371–382.
Paivio, A., & Marschark, M. (1991). Integrative processing of concrete and abstractsentences. In: A. Paivio (Ed.), Images in the mind: the evolution of a theory (pp.134–154). New York, NY: Harvester Wheatsheaf.
Pawluk, E. J., Koerner, N., & Antony, M. (2011, November). Impact of concretenesstraining on worry, intolerance of uncertainty, and problem solving. In Posterpresented at the meeting of the Association for Cognitive and Behavioral TherapiesToronto, ON.
Pennebaker, J. W., & Beall, K. (1986). Confronting a traumatic event: toward anunderstanding of inhibition and disease. Journal of Abnormal Psychology, 95,274–281.
Pennebaker, J. W., Booth, R. J., & Francis, M. E. (2007). Linguistic Inquiry and WordCount (LIWC): LIWC2007 [Computer software]. Mahwah, NJ: Lawrence ErlbaumAssociates.
Pham, L. B., & Taylor, S. E. (1999). From thought to action: effects of process- versusoutcome-based mental simulations on performance. Personality and Social Psy-chology Bulletin, 25, 250–260.
Provencher, M. D., Dugas, M. J., & Ladouceur, R. (2004). Efficacy of problem-solvingtraining and cognitive exposure in the treatment of generalized anxiety dis-order: a case replication series. Cognitive and Behavioral Practice, 11, 404–
Roemer, L., & Orsillo, S. M. (2007). An open trial of an acceptance-based behaviortherapy for generalized anxiety disorder. Behavior Therapy, 38, 72–85.
Rowe, M. K., & Craske, M. G. (1998). Effects of varied-stimulus exposure training onfear reduction and return of fear. Behaviour Research and Therapy, 36, 719–734.
nxiet
R
S
S
S
S
S
S
S
S
K. Fracalanza et al. / Journal of A
ygh, J. L., & Sanderson, W. C. (2004). Treating generalized anxiety disorder: evidence-based strategies, tools, and techniques. New York, NY: Guilford Press.
exton, K. A., & Dugas, M. J. (2009). Defining distinct negative beliefs about uncer-tainty: validating the factor structure of the intolerance of uncertainty scale.Psychological Assessment, 21, 176–186.
exton, K. A., & Dugas, M. J. (2008). The Cognitive Avoidance Questionnaire: vali-dation of the English translation. Journal of Anxiety Disorders, 22, 355–370.
chmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice: commonprinciples in three paradigms suggest new concepts for training. PsychologicalScience, 3, 207–217.
heehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, P., Janavs, J., Weiller, E.,et al. (1998). The Mini-International Neuropsychiatric Interview (M.I.N.I): thedevelopment and validation of a structured diagnostic psychiatric interview forDSM-IV and ICD-10. Journal of Clinical Psychiatry, 59, 22–33.
ibrava, N. J., & Borkovec, T. D. (2006). The cognitive avoidance theory of worry. In:G. C. L. Davey, & A. Wells (Eds.), Worry and its psychological disorders: theory,assessment and treatment (pp. 239–256). Hoboken, NJ: John Wiley and Sons.
loan, D. M., Marx, B. P., & Epstein, E. M. (2005). Further examination of the expo-sure model underlying the efficacy of written emotional disclosure. Journal ofConsulting and Clinical Psychology, 73, 549–554.
loan, D. M., Marx, B. P., Epstein, E. M., & Lexington, J. M. (2007). Does altering
the writing instructions influence outcome associated with written disclosure?Behavior Therapy, 38, 155–168.tartup, H. M., & Erickson, T. M. (2006). The Penn State Worry Questionnaire (PSWQ).In: G. C. L. Davey, & A. Wells (Eds.), Worry and its psychological disorders: theory,assessment and treatment (pp. 101–119). Hoboken, NJ: John Wiley and Sons.
y Disorders 28 (2014) 559–569 569
Stöber, J. (1998). Worry, problem elaboration and suppression of imagery: the roleof concreteness. Behaviour Research and Therapy, 36, 751–756.
Stöber, J. (2000). Worry, thoughts, and images: a new conceptualization. In: U.von Hecker, S. Dutke, & G. Sedek (Eds.), Generative mental processes and cog-nitive resources: integrative research on adaptation and control (pp. 223–244).Dordrecht, NL: Kluwer.
Stöber, J., & Borkovec, T. D. (2002). Reduced concreteness of worry in generalizedanxiety disorder: findings from a therapy study. Cognitive Therapy and Research,26, 89–96.
van der Heider, C., & ten Broeke, E. (2009). The when, why, and how of worryexposure. Cognitive and Behavioral Practice, 16, 386–393.
van Minnen, A., & Foa, E. B. (2006). The effect of imaginal exposure length on outcomeof treatment for PTSD. Journal of Traumatic Stress, 19, 427–438.
Watkins, E., Moberly, N. J., & Moulds, M. L. (2008). Processing mode causally influ-ences emotional reactivity: distinct effects of abstract versus concrete construalon emotional response. Emotion, 8, 364–378.
Wells, A., & Carter, K. (2006). Generalised anxiety disorder. In: A. Carr, & M. McNulty(Eds.), Handbook of adult clinical psychology: an evidence based practice approach(pp. 423–457). East Sussex, UK: Routledge.
Wetherell, J. L., Gatz, M., & Craske, M. G. (2003). Treatment of generalized anxietydisorder in older adults. Journal of Consulting and Clinical Psychology, 71, 31–40.
Wolitzky-Taylor, K., & Telch, M. J. (2010). Efficacy of self-administered treatments forpathological academic worry: a randomized controlled trial. Behaviour Researchand Therapy, 48, 840–850.
Zinbarg, R. E., Craske, M. G., & Barlow, D. H. (2006). Mastery of your anxiety and worry(therapist guide) (2nd ed.). New York, NY: Oxford University Press.
