Flowchart of subject progress throughthe phases of the study. There were 295 subjects in the intention-to-treatanalysis and 211 subjects in the completer analysis. CCBT indicates comprehensivecognitive behavioral therapy; FLU, fluoxetine; PBO, placebo.
Mean Brief Social Phobia Scale (BSPS)score by treatment group (n = 295). Piecewise linear mixed-effectsmodels analysis. CCBT indicates comprehensive cognitive behavioral therapy;FLU, fluoxetine; PBO, placebo.
Davidson JRT, Foa EB, Huppert JD, Keefe FJ, Franklin ME, Compton JS, Zhao N, Connor KM, Lynch TR, Gadde KM. Fluoxetine, Comprehensive Cognitive Behavioral Therapy, and Placeboin Generalized Social Phobia. Arch Gen Psychiatry. 2004;61(10):1005-1013. doi:10.1001/archpsyc.61.10.1005
Copyright 2004 American Medical Association. All Rights Reserved.Applicable FARS/DFARS Restrictions Apply to Government Use.2004
Generalized social phobia is common, persistent, and disabling and is
often treated with selective serotonin reuptake inhibitor drugs or cognitive
We compared fluoxetine (FLU), comprehensive cognitive behavioral group
therapy (CCBT), placebo (PBO), and the combinations of CCBT/FLU and CCBT/PBO.
Randomized, double-blind, placebo-controlled trial.
Two academic outpatient psychiatric centers.
Subjects meeting a primary diagnosis of generalized social phobia were
recruited via advertisement. Seven hundred twenty-two were screened, and 295
were randomized and available for inclusion in an intention-to-treat efficacy
analysis; 156 (52.9%) were male, 226 (76.3%) were white, and mean age was
Treatment lasted for 14 weeks. Fluoxetine and PBO were administered
at doses from 10 mg/d to 60 mg/d (or equivalent). Group comprehensive cognitive
behavioral therapy was administered weekly for 14 sessions.
Main Outcome Measures
An independent blinded evaluator assessed response with the Brief Social
Phobia Scale and Clinical Global Impressions scales as primary outcomes. A
videotaped behavioral assessment served as a secondary outcome, using the
Subjective Units of Distress Scale. Adverse effects were measured by self-rating.
Each treatment was compared by means of χ2 tests and piecewise
linear mixed-effects models.
Clinical Global Impressions scales response rates in the intention-to-treat
sample were 29 (50.9%) (FLU), 31 (51.7%) (CCBT), 32 (54.2%) (CCBT/FLU), 30
(50.8%) (CCBT/PBO), and 19 (31.7%) (PBO), with all treatments being significantly
better than PBO. On the Brief Social Phobia Scale, all active treatments were
superior to PBO. In the linear mixed-effects models analysis, FLU was more
effective than CCBT/FLU, CCBT/PBO, and PBO at week 4; CCBT was also more effective
than CCBT/FLU and CCBT/PBO. By the final visit, all active treatments were
superior to PBO but did not differ from each other. Site effects were found
for the Subjective Units of Distress Scale assessment, with FLU and CCBT/FLU
superior to PBO at Duke University Medical Center, Durham, NC. Treatments
were well tolerated.
All active treatments were superior to PBO on primary outcomes. Combined
treatment did not yield any further advantage. Notwithstanding the benefits
of treatment, many patients remained symptomatic after 14 weeks.
Social phobia, now often referred to as social anxiety disorder, iswidespread with prevalence rates as high as 14% in the United States,1 begins early in life, and rarely remits.2 Ofthe 2 types of social phobia, generalized social phobia (GSP) is believedto be more severe than nongeneralized.3 A recentstudy of GSP in a health maintenance organization setting showed an 8% prevalencerate, along with raised incidence of suicide attempts, greater health seeking,and reduced earning capacity; physician recognition of GSP was no more than0.5% of all cases.4
Controlled trials suggest benefit from medication and psychosocial approaches.Among medications, the selective serotonin reuptake inhibitor (SSRI) groupis the most extensively studied.5 Cognitivebehavioral treatments (CBTs) are also effective in social phobia, especiallya form of group CBT, which includes cognitive restructuring, exposure to simulatedsituations in sessions, and in vivo exposure homework assignments.6 Group CBT has been found to be more effective thanpsychoeducation,6 atenolol,7 andbuspirone hydrochloride8 and equivalent tophenelzine sulfate.6 Comprehensive cognitivebehavior therapy (CCBT) was developed by Foa et al (unpublished data, 1994) for GSP. In this group CBT, social skills training is added to exposuretherapy and cognitive restructuring, because social skills deficits are apertinent part of GSP and may not respond well to programs if skill trainingis absent.9
With the emergence of SSRIs as frontline and proven pharmacotherapyfor GSP, we considered it important to compare an SSRI with CCBT. To our knowledge,this is 1 of only 2 completed GSP studies to include a combination cell inwhich subjects received both CCBT and an SSRI.10 Ourproject had 5 overall goals: (1) to compare the effects of 14 weeks’treatment with fluoxetine (FLU) alone, group alone (CCBT), combined CCBT/FLU,CCBT/placebo (PBO) (to take into account nonspecific pill taking), and PBOalone; (2) to evaluate maintenance of treatment effects following completionof treatment; (3) to demonstrate transportability of treatments (ie, thatCCBT can be successfully implemented in a center specializing in pharmacotherapy[Duke University Medical Center, Durham, NC] and vice versa for a center specializingin medication [University of Pennsylvania, Philadelphia, Pa]); (4) to investigatemechanisms of therapeutic change by examining the relationship between cognitivedistortions and social skills deficits; and (5) to explore predictors of treatmentresponse. This report will focus on the first goal, short-term effects ofthe 5 treatments.
The study was conducted at 2 academic medical centers with outpatientprograms specializing in anxiety disorder research. At each research center,FLU, CCBT, CCBT/FLU, CCBT/PBO, and PBO were all compared, with FLU and PBObeing administered on a double-blind basis. An independent rater, blindedto treatment assignments, conducted the primary outcome assessments. Eligiblesubjects met DSM-IV criteria for primary GSP, assessedby the Structured Clinical Interview for DSM-IV.11 Subjects underwent psychiatric and medical evaluationto establish inclusion and exclusion criteria. Subjects were assigned to treatmentby block randomization, which was generated by computer program at Duke UniversityMedical Center, in groups of 10, with 2 subjects assigned to each of the 5conditions. There were some exceptions to the implementation of this processbecause of a small number of prerandomization dropouts. We balanced CCBT groupsto include at least 2 women and 2 men and typically had a male and a femaletherapist. The study coordinator at each site enrolled and allocated subjectsto their treatment groups. This individual was blind to the sequence priorto assignment.
Medication was administered and monitored by a psychiatrist. Medicationsessions were audiotaped, and tapes were audited at random by 1 of the investigatorsat each site. Pharmacotherapy was provided according to the study manual andadherence rated according to a standardized checklist (available on request)to assure that no CCBT was being conducted and that medication was being providedin a standard way. Medication visits occurred weekly for 4 weeks, then every2 weeks. Study investigators (E.B.F., F.J.K., and M.E.F.) videotaped and evaluatedCCBT at each site and provided feedback to therapists in weekly supervisionsessions (adherence rating scale available by request). Early in the trial,the sites provided feedback to therapists via weekly supervisory conferencecalls to ensure consistency of treatment across sites. Group CCBT was administeredat weekly intervals.
Enrollment began in early 1995 and continued until September 2001. Regularconference calls were held between the 2 sites throughout. The protocol wasapproved by the institutional review board at each site, and all subjectsprovided written informed consent.
Inclusion criteria were: (1) DSM-IV diagnosisof GSP; (2) age between 18 and 65 years; (3) fluency in English; and (4) provisionof written informed consent. Exclusion criteria were: (1) a primary comorbidanxiety disorder (defined by which disorder was the more debilitating andclinically salient); (2) lifetime history of schizophrenia, bipolar disorder,or organic brain syndrome; (3) major depression within the last 6 months;(4) substance abuse or dependence within the past year; (5) mental retardationor pervasive developmental disability; (6) unstable medical condition; (7)prior failure of response to fluoxetine at 60 mg/d for at least 4 weeks orto 12 weekly sessions of CCBT for GSP; (8) concurrent psychiatric treatmentor other psychoactive medications; (9) positive urine drug screen results;(10) inability to maintain 2 weeks’ psychotropic drug-free washout;and (11) pregnancy or lactation.
Table 1 presents demographic characteristicsby treatment group, showing no group differences. Between-site demographicdifferences were observed for ethnicity, marital status, and employment, withfewer African American subjects and more married and employed subjects inthe Duke University Medical Center site sample, as compared with the Universityof Pennsylvania site sample. Baseline symptom scores for each treatment groupare shown in Table 2.
Fluoxetine was started at 10 mg/d, increasing on day 8 to 20 mg/d, onday 15 to 30 mg/d, and on day 29 to 40 mg/d. Unless adverse effects becameproblematic, the goal was for subjects to reach 40 mg/d. At days 43 and 57,the dose could be raised to 50 mg/d and 60 mg/d, respectively, if subjectsfailed to achieve a Clinical Global Impressions (CGI) Improvement12 score of 1 or 2 and were tolerating medication. Compliancewas monitored by reviewing daily medication logs and pill counts at each visit.Normally, the dose was given in the morning.
Comprehensive cognitive behavioral therapy is a 14-week group treatmentthat combines in vivo exposure, cognitive restructuring, and social skillstraining. Derived in part from a treatment developed by Heimberg et al,13 CCBT differs from Heimberg group CBT in that CCBTincludes specific social skills training (eg, how to begin a conversationwith a stranger) and improves specific behaviors (eg, eye contact), the role-playsare much shorter, and CCBT is 2 sessions longer than group CBT. Two therapists(1 male, 1 female) who received extensive training prior to this study conductedthe treatment; each group consisted of 5 to 6 subjects. The first 2 sessionswere educational, with therapists presenting a cognitive behavioral modelof social anxiety and explaining treatment techniques. Sessions 3 and 4 weredevoted to social skills training; patients received instruction and role-playedshort (30-60 second), repeated (5-7 times) scenarios devoted to initiating,maintaining, and ending conversations, as well as compromise/negotiation.In sessions 5 through 13, patients participated in longer (3-4 minute) role-playstailored to their specific social concerns. Prior to each role-play, subjectsidentified a core dysfunctional thought associated with that situation anda relevant rational response to replace it. Next, social skills training instructionswere provided before the role-play, and specific aspects of each role-playwere repeated to facilitate skills acquisition. Between sessions, subjectscompleted homework assignments designed to help them confront fearful socialsituations using the techniques learned in therapy. Session 14 included adiscussion of treatment gains and recommendations for future practices.
Primary outcome assessments by the blinded independent evaluator (IE)were as follows: (1) CGI Improvement scale, a 7-point rating measured improvementwherein response is defined as having a CGI Improvement score of 1 (very muchimprovement) or 2 (much improvement), and the 7-point CGI Severity scale12 and (2) the Brief Social Phobia Scale (BSPS), an18-item scale comprised of fear, avoidance, and physiological symptoms.14 Independent evaluator ratings were conducted at baselineand at weeks 4, 8, and 14. Success of the blinding procedure was not evaluated.Altogether, 8 IEs were associated with this trial from start to finish (3at Duke University Medical Center and 5 at University of Pennsylvania). Intraclasscorrelation pairwise agreement coefficients for between raters ranged from0.77 to 0.98.
The Social Phobia and Anxiety Inventory,15 whichis reliable, valid, and sensitive to treatment effects,16 servedas a secondary measure.
Symptoms possibly attributable to treatment were evaluated by meansof the Severity of Symptoms Scale,17 whichwas given to subjects at each visit, regardless of treatment assignment.
A role-play test evaluated social skills before and after treatment.The subject was videotaped in each of 3 social scenarios: initiating a conversationwith a stranger; expression of negative assertiveness; and an impromptu speech.Subjects were asked to indicate their level of anxiety using the SubjectiveUnits of Distress Scale (SUDS), an analogue scale ranging from 0 (relaxedand calm) to 100 (extremely fearful).The SUDS ratings were assessed 7 times:immediately after receiving role-play instructions (ie, at baseline); beforeeach of the 3 role-play scenarios (to measure anticipatory anxiety); and aftereach scenario (to measure “aftermath”). An experimenter and aconfederate, both blinded to the subject’s treatment status, were presentduring the test. Each scenario lasted 3 minutes.
The following hypotheses were tested: (1) all active treatments wouldbe superior to PBO and (2) combined CCBT/FLU treatment would be superior toother active treatments. Post hoc, exploratory, 2-tailed pairwise contrastswere performed to determine whether the active treatments differed from oneanother. Sample size was calculated based on pilot data obtained earlier.Three power estimates were made for the primary contrasts, with a single pooledestimate of the error term and 1-tailed hypothesis testing. Power to detecta difference for 60 subjects per group, with an effect size of 0.30 at α = .05,ranged from 0.7 to 0.9 in the combined site sample according to comparison.
First, to detect possible pretreatment differences among conditions,1-way analyses of variance were performed on pretreatment data for the BSPS,CGI, Social Phobia and Anxiety Inventory, and SUDS ratings. Second, the differentialefficacy of the conditions at 14 weeks or an earlier end point was analyzedusing (1) χ2 tests to compare the proportion of subjects whoachieved treatment response status within each treatment and (2) linear mixed-effectsmodels (LMM), from SAS procedure MIXED,18 toassess differences in course of treatment response. For these analyses, piecewiselinear growth curve models, with a change point at week 4, provided the bestfit for the data. An unstructured variance model best fit the data for allanalyses. We applied this piecewise LMM to each of the 3 continuous outcomemeasures. These analyses were followed by pairwise analyses to determine whethertreatments differed in how fast they induced change. In addition, piecewiseLMMs were applied to each outcome variable to determine whether each treatmentinduced change across the 14 treatment weeks. Site effects were examined byincluding a binary indicator and its interactions with treatment and timein the piecewise LMMs mentioned earlier. When the interaction of site by treatmentor site by time was significant, we conducted follow-up analyses to explainthese interactions. Sex, age, marital status, ethnicity, and employment statuswere all examined to determine whether they differentially affected treatmentoutcome. None of these variables had significant interactions with treatmentcondition by time and were therefore not included in final analyses.
Linear mixed-effect model analyses included all randomized subjectsand were conducted using pretreatment and posttreatment behavioral measures,with the behavioral measure as the dependent variable.
Responder analyses were conducted separately for all randomized subjectsfor whom any data were available (ITT) and subjects who completed 14 weeks(completer sample).
Effect size, along with 95% confidence intervals, was calculated accordingto the Cohen formula19 for a difference betweengroups of m1 (postactive treatment) – m2 (post-PBO)/pooled standarddeviation (m1m2), in order to determine effect sizes of active treatmentsat posttreatment.
A total of 4306 subjects were screened by telephone, of whom 722 wereinvited for a full study assessment. The most common reasons for exclusionat this stage were the presence of major depression, social phobia being nongeneralized,presence of another primary anxiety disorder, and substance abuse. Of these,295 were randomized to 1 of the 5 treatments, with the following dispositions:16 (5%) dropped out prior to receiving treatment; 68 (23%) dropped out betweenweek 1 and week 14; and 211 (72%) completed 14 weeks of treatment. Randomizedsubjects who were analyzed composed the ITT (n = 265) and completer(n = 211) samples, respectively, as follows: FLU (n = 57,39), CCBT (n = 60, 48), CCBT/FLU (n = 59, 42), CCBT/PBO(n = 59, 46), and PBO (n = 60, 36). Figure 1 provides details of subject disposition. The overall significancefor rate of dropout by treatment type was not statistically significant (χ24 = 8.22; P = .08).However, in pairwise contrasts, the PBO group had significantly more dropoutsthan the CCBT group (χ21 = 6.53; P = .01) and the CCBT/PBO group (χ21 = 4.48; P = .03),respectively.
The proportion of subjects who discontinued prematurely did not differstatistically across site (χ28 = 9.67; P = .29). Reasons for early discontinuation areprovided in Figure 1.
There were no significant differences among groups in respect of meanmedication doses. For the combined PBO groups (CCBT/PBO and PBO), mean (SD)maximum and final attained visit doses were 49.3 (15.2) mg equivalents perday and 46.4 (17.9) mg equivalents per day, respectively. For the combinedFLU groups (CCBT/FLU and FLU), they were 47.3 (13.9) mg/d and 43.6 (16.4)mg/d, respectively.
The ITT response rates determined by a CGI Improvement score of 1 or2 were as follows: FLU, 50.9%; CCBT, 51.7%; CCBT/FLU, 54.2%; CCBT/PBO, 50.8%;and PBO, 31.7%. The overall χ2 test showed a trend towardsignificance (χ24 = 8.03; P = .09), and pairwise significance was found for all treatmentcomparisons against PBO: FLU vs PBO (χ21 = 4.46; P = .03); CCBT vs PBO (χ21 = 4.94; P = .03); CCBT/FLUvs PBO (χ21 = 6.19; P = .01); and CCBT/PBO vs PBO (χ21 = 4.52; P = .03). For the completer sample, responserates were as follows: FLU, 64.1%; CCBT, 64.6%; CCBT/PBO, 59.6%; CCBT/FLU,66.7%; and PBO, 40.5%. The overall χ2 test showed a trendtoward significance (χ24 = 8.09; P = .09). Pairwise significance was found forthe following treatment comparisons against PBO: FLU vs PBO (χ21 = 4.77; P = .03);CCBT vs PBO (χ21 = 5.46; P = .02); and CCBT/FLU vs PBO (χ21 = 6.02; P = .01); CCBT/PBO vs PBO showed a trend towardsignificance (χ21 = 3.17; P = .08). There were more subjects in the responder samplein the CCBT/PBO group at Duke University Medical Center than at the Universityof Pennsylvania. No other site differences were detected.
Effect sizes (95% confidence interval) for each comparison vs PBO, basedon the ITT analysis of variance, provided the following results on the BSPS:0.40 (0.02 to 0.77) for FLU; 0.30 (−0.07 to 0.66) for CCBT; 0.24 (−0.13to 0.60) for CCBT/FLU, and 0.52 (0.14 to 0.89) for CCBT/PBO vs PBO.
For the CGI Severity scale, effect size (95% confidence interval) resultsfor each treatment relative to PBO were as follows: 0.42 (0.04 to 0.80) forFLU; 0.27 (−0.10 to 0.64) for CCBT; 0.30 (−0.07 to 0.67) for CCBT/FLU;and 0.42 (0.04 to 0.79) for CCBT/PBO.
At week 14, significant differences existed in favor of all active treatmentscompared with PBO on the BSPS and Social Phobia and Anxiety Inventory (P<.05) (Table 2).The CGI Severity scale evidenced the same pattern, with FLU and CCBT/FLU bothbeing superior to PBO (P = .01) but notreaching significance for CCBT or CCBT/PBO (P<.10).
For the BSPS, treatment by time models (ie, slope or rate of change)was significant for weeks 0 to 4 (P = .002)and for weeks 4 to 14 (P = .02) (Figure 2). Pairwise analysis suggested that FLUalone produced more change than PBO, CCBT/FLU, and CCBT/PBO from weeks 0 to4 (P<.01), while CCBT evidenced more change thanCCBT/FLU and CCBT/PBO (P<.05) and a trend towardsuperiority over PBO (P = .10). From weeks4 to 14, CCBT/FLU and CCBT/PBO conditions produced more change than PBO (P<.05), and CCBT evidenced a trend in the same direction(P = .09). Furthermore, CCBT/FLU showedmore change than FLU in the last 10 weeks (P = .03),and CCBT/PBO showed a trend in the same direction (P = .07).Similar results were found for the CGI Severity scale. In summary, FLU demonstratedthe fastest onset of action followed by CCBT. By the final visit, combinedtreatments were not different from monotherapies, and all active treatmentswere superior to PBO.
Posttreatment SUDS ratings are presented in Table 3. The multiple tests were combined to produce average ratingsfor all 3 scenarios. For each treatment, we provide the average postinstruction,anticipatory, and aftermath scores for all 3 scenarios. Significant time ×treatment × site interactions emerged for all 3 analyses (P<.05), leading us to break down the treatment comparisons by site.At the University of Pennsylvania, no comparisons were significant. At DukeUniversity Medical Center, subjects in the FLU group reported significantlyless anxiety than the PBO group on all behavioral measures. Comprehensivecognitive behavioral therapy/fluoxetine was superior to PBO in anticipatoryanxiety before scenarios but not in baseline anxiety. Fluoxetine was associatedwith significantly lower anticipatory distress than CCBT, and CCBT/PBO wasassociated with less distress than CCBT at anticipatory and posttest tasks.
Further examination of Table 3 indicatesa difference in outcome between those who received CCBT in any form and thosewho did not. While there were no differences in outcome per se, the patternof SUDS scores from postinstruction to anticipatory to aftermath is consistentat week 0 across conditions; at each step, there is an increase in anxiety.However, at week 14, subjects who participated in CCBT evidenced, on average,no change or even a decrease in SUDS score from anticipatory to aftermath,while the PBO and FLU subjects continued to show an increase in anxiety fromanticipatory to aftermath. This difference in SUDS ratings was significant(P<.05). These data provide some evidence thatexposure during group therapy did have an effect on behavioral ratings.
Significantly higher rates of treatment emergent events were noted on4 symptoms (Table 4). Insomnia and headacheboth occurred more often in the PBO and CCBT/PBO groups relative to CCBT aloneand in the CCBT/FLU and FLU groups relative to CCBT alone. Nausea occurredmore frequently in the PBO and CCBT/PBO groups relative to the group undergoingCCBT alone and in the FLU group relative to the CCBT and CCBT/PBO groups.Relative to the PBO group, anorgasmia was more commonly seen in the CCBT/PBO,CCBT/FLU, and FLU groups, as well as in the FLU vs CCBT/PBO groups.
In adults with GSP, this study demonstrated efficacy for FLU and CCBTrelative to PBO but no evidence for greater benefit of combined treatmentover monotherapies. Response rates indicated a treatment vs placebo differenceranging from 15% to 24%. A comparison of our findings with the study of phenelzineand group CBT by Heimberg et al6 indicatesthat the 2 studies have a similar PBO response rate (31% and 33%, respectively)and broadly comparable rates of response to group therapy (52% and 58%, respectively).Response rates to FLU and phenelzine were 51% and 65%, respectively. Givena very broad spectrum of activity for monoamine oxidase inhibitors, it isconceivable that phenelzine carries greater benefit than an SSRI, a hypothesisconsistent with a recent meta-analysis by Hidalgo et al.5
One SSRI, sertraline hydrochloride, has been studied in conjunctionwith CBT in a Swedish primary care setting.10,20 Sertraline/CBTand sertraline alone, but not CBT/PBO, exceeded the response from PBO alone.Combined sertraline/CBT tended to produce greater benefits. However, on 28-weekposttreatment follow-up, initial CBT alone was associated with further gain,whereas sertraline alone or combined with CBT was associated with deterioration.This study differed from ours in sample characteristics, type of CBT, andlack of a CBT-alone group, which may account for some of the different findings.
The IE and self-report analyses consistently demonstrated an advantageof all active treatments over PBO, other than the CGI Severity scale, on whichonly FLU and CCBT/FLU were superior to PBO at end point. Effect sizes (whichwere not derived from the LMM analyses) indicate a moderate effect of alltreatments, with CCBT/PBO and FLU tending to be the highest. The fact thatCCBT/PBO had a large effect size on the BSPS but only a trend toward superiorityover PBO on the CGI measures suggests that the CGI scale may be somewhat lesssensitive to changes in social anxiety symptoms. Alternatively, the BSPS maydetect subtle changes in social anxiety that do not reflect the global impairmentas well as the CGI scale.
Measures of social interaction were evaluated in vivo, in the form ofa series of videotaped social interactions and speech delivery. These assessmentsproved generally robust in picking up treatment effects relative to PBO atDuke University Medical Center but not at the University of Pennsylvania.This is because at the University of Pennsylvania, but not at Duke UniversityMedical Center, the patients treated with PBO responded well to the behavioraltask, even though their symptoms did not change substantially. We do not understandwhy CCBT alone should have failed to distinguish from PBO, although a numericallysuperior effect was observed in the posttest evaluation.
Mixed-effects models have become well accepted for interpretation ofclinical trial data and include all data without imputing missing values.21 We used a 2-level approach (piecewise linear growthmodel) and found FLU generated a faster response than the other treatments;at week 4, FLU showed superiority to CCBT/FLU, CCBT/PBO, and PBO. However,the degree of change at posttreatment did not differ between FLU and CCBT.Thus, our findings are similar to Heimberg et al,6 whoreported an earlier advantage for phenelzine over CBT, with the 2 treatmentsbeing comparable by week 12. Such a finding suggests that greater advantagewould accrue from a strategy of initial treatment with an SSRI, followed byaugmentation with psychosocial treatment after 4 to 8 weeks. This indeed wasfound in a study of posttraumatic stress disorder,22 andwe are now planning such a study in social anxiety disorder. Besides takingadvantage of the therapeutic impact of an SSRI, when psychosocial treatmentis added to established SSRI therapy, its effects would not be obfuscatedby issues of pill taking and adverse effects or pseudo adverse effects thatmay appear early in the course of treatment. Our findings about adverse effectsare interesting in this regard, because they indicate that pill taking itself(whether drug or PBO) is associated with a high rate of headaches and insomnia,which occurred in 27% to 42% of subjects. Thus, a person who is in therapyand taking a pill may be initially distracted by somatic concerns such asimpaired sleep and headaches. The same was true to a lesser extent for erectiledysfunction and nausea. Anorgasmia occurred more frequently in patients receivingFLU than PBO; we do not know if this in any way compromised the integrityof the double-blind design. Although the IEs were instructed not to discussadverse effects and to remind subjects only to address symptoms of socialphobia, adverse effects were discussed in the medication management sessions.
Similar to the findings of Heimberg et al,6 wedid not detect a site × treatment effect, indicating that medicationand psychosocial treatments, as well as PBO, can be given in a consistentmanner at sites that differ in their therapeutic allegiance. It is possiblethat the between-site differences on the behavioral test were related to thefact that less attention was devoted to intersite standardization of testdelivery, relative to medication and CBT.
Some limitations of our study need to be acknowledged. First, grouptreatment restricted flexibility of schedule for potential subjects, someof whom may have been deterred through fear of the group itself. Second, somesubjects declined to take part in treatment after being randomized, sincethey did not want to receive medication alone. This issue has been addressedin a separate report by our group23 and ispartly taken care of in the ITT analysis. It could be questioned whether FLUwas the most appropriate drug, particularly since there are 2 reports demonstratingno difference between FLU and PBO.24,25 Giventhat SSRIs as a class have very broad-spectrum anxiolytic properties but arenot invariably positive in all trials, we are as yet unpersuaded that thisargument has great merit. However, we do not know if other SSRIs might havebeen more effective (eg, paroxetine hydrochloride or sertraline).10,26- 30 Nopublished head-to-head comparisons of SSRIs exist to our knowledge. As inmost social phobia/social anxiety disorder trials to date, we excluded subjectswith a diagnosis of major depression. Data from a subsection of the subjectscreens administered at the University of Pennsylvania suggest that depressionwas the most frequent reason for study exclusions, comprising approximatelyone third of the patients excluded. It is essential that the next generationof studies determine effective treatment strategies for subjects with comorbidsocial phobia and depression, especially given their impairing effects.23 We did not assess the success of IE binding, as recommendedby the CONSORT guidelines.31 As such, thiscould be seen as a limitation. On the other hand, our experience from a similarprevious trial suggests that “guessing” increases attention and/orsensitivity to the matter of what treatment is being administered. As to theincreased anorgasmia in subjects receiving FLU, the IE was instructed notto discuss or encourage reports of adverse events, which formed part of thepharmacotherapy assessment.
To conduct and complete a trial of this magnitude is a considerableundertaking, representing approximately 10 years of work from inception ofthe idea, to acquisition of funding, recruitment into the study, and subsequentanalysis of data. Such trials can be successfully accomplished, and this onehas demonstrated that widely used treatments (ie, CBT and FLU) are effectivefor individuals with generalized social phobia, many of whom are significantlyimpaired. Although FLU and CCBT were more successful than PBO, substantialsymptoms still remained after 14 weeks’ treatment. We therefore wonderwhether longer-term pharmacological treatment is necessary and if changesin the delivery of CCBT would improve results. Furthermore, recent evidencesuggests that individual CBT produces better results than group CBT in GSP.32 Finally, combining FLU with CCBT did not provideany greater therapeutic benefit. One possible contributing reason may be thedistracting physical complaints that arise early with medication and may bea reason for sequential treatment rather than simultaneous initiation.
Correspondence: Jonathan Davidson, MD, Box3812, Duke University Medical Center, Durham, NC 27710 (email@example.com).
Submitted for Publication: December 29, 2003;final revision received April 12, 2004; accepted April 20, 2004.
Funding/Support: This study was supported bygrant R10-MH49339-05A1 from the National Institute of Mental Health, Bethesda,Md (Drs Davidson and Foa).
Additional Information: Medication and matchingplacebo were provided by Eli Lilly, Indianapolis, Ind, who have reviewed themanuscript. They were uninvolved in study design, data analysis, or manuscriptpreparation.
Acknowledgment: We thank Alan Bellack, PhD,and Richard Heimberg, PhD, for significant assistance in study design and/ortraining. Michael J. Kozak, PhD, and Donald J. Bux, PhD, served as coordinatorsat the University of Pennsylvania, Philadelphia, Pa. Nader Amir, PhD; NorahFeeny, PhD; Lisa Jaycox, PhD; Simon Rego, PhD; David Tolin, PhD; and ElnaYadin, PhD, served as therapists at the University of Pennsylvania/EasternPennsylvania Psychiatric Institute. Pat Segee, PhD, and Rita Davison, BA,served as study coordinators at Duke University Medical Center. Nicholas L.S. Potts, MD; Suzanne M. Sutherland, MD; Stewart D. Barnett, MD; John Travers,MD; Cherri Miner, MD; Brian A. Stabler, PhD; Ayesha Chaudhary, MD; SusmitaKashikar-Zuck, PhD; and Marlene Sandstrom, PhD, served as therapists at DukeUniversity Medical Center. L. Erik Churchill, MA, provided data managementsupport.