Sudden gains during therapy of social phobia.

Abstract

The present study investigated the phenomenon of sudden gains in 107 participants with social phobia (social anxiety disorder) who received either cognitive-behavioral group therapy or exposure group therapy without explicit cognitive interventions, which primarily used public speaking situations as exposure tasks. Twenty-two out of 967 session-to-session intervals met criteria for sudden gains, which most frequently occurred in Session 5. Individuals with sudden gains showed similar improvements in the 2 treatment groups. Although cognitive-behavioral therapy was associated with more cognitive changes than exposure therapy, cognitive changes did not precede sudden gains. In general, the results of this study question the clinical significance of sudden gains in social phobia treatment.

Full text

Page 1

Sudden Gains During Therapy of Social Phobia
Stefan G. Hofmann, Stefan M. Schulz, Alicia E. Meuret, David A. Moscovitch, and Michael Suvak
Boston University
The present study investigated the phenomenon of sudden gains in 107 participants with social phobia
(social anxiety disorder) who received either cognitive–behavioral group therapy or exposure group
therapy without explicit cognitive interventions, which primarily used public speaking situations as
exposure tasks. Twenty-two out of 967 session-to-session intervals met criteria for sudden gains, which
most frequently occurred in Session 5. Individuals with sudden gains showed similar improvements in
the 2 treatment groups. Although cognitive–behavioral therapy was associated with more cognitive
changes than exposure therapy, cognitive changes did not precede sudden gains. In general, the results
of this study question the clinical significance of sudden gains in social phobia treatment.
Keywords: sudden gains, social anxiety disorder, social phobia, cognitive–behavioral therapy, exposure
therapy
In recent years, psychotherapy researchers have investigated the
phenomenon of large, rapid, and stable decreases in symptomatol-
ogy during treatment, which has been referred to as sudden gains
(Tang & DeRubeis, 1999b). Sudden gains are typically defined by
a set of three quantitative criteria: (a) the sudden gain must be large
in absolute terms, (b) the sudden gain must represent at least a 25%
reduction from the level of symptomatology before the gain oc-
curred, and (c) the mean level of symptomatology in the three
therapy sessions preceding the gain must be significantly higher
than the mean level of symptomatology in the three postgain
sessions. Most studies of sudden gains have examined treatment
changes in depression with the Beck Depression Inventory (BDI;
Beck, Ward, Mendelson, Mock, & Erbaugh, 1961; but see also
Stiles et al., 2003; Vittengl, Clark, & Jarrett, 2005).
When applying these criteria, Tang and DeRubeis (1999b)
found that sudden gains occurred in more than 50% of individuals
who responded to cognitive–behavioral therapy for depression and
that these gains accounted for more than 50% of these individuals’
total improvement in depression. Compared with participants who
did not experience sudden gains at posttreatment, individuals with
sudden gains showed better outcome at posttreatment and at
6-month and 18-month follow-ups.
The initial study by Tang and DeRubeis (1999b) was followed
by a number of other studies examining sudden gains (Gaynor et
al., 2003; Hardy et al., 2005; Kelly, Roberts, & Ciesla, 2005; Tang,
DeRubeis, Beberman, & Pham, 2005; Tang, Luborsky, & An-
drusyna, 2002; Vittengl, Clark, & Jarrett, 2005). Interventions in
these studies ranged from traditional cognitive therapy to support-
ive expressive therapy or pharmacotherapy. Although sudden
gains were primarily assessed in efficacy trials of time-limited
individual treatments (e.g., Tang & DeRubeis, 1999b; Vittengl et
al., 2005), their incidence was also assessed in group therapy for
depression (Kelly et al., 2005) and individuals with mixed psychi-
atric diagnoses who were given routine treatments in outpatient
clinic settings (Stiles et al., 2003). Overall, results corroborated the
existence of sudden gains during the acute phase of treatment, as
defined by the criteria of Tang and DeRubeis (1999b), with the
median pregain session occurring at Session 5. The mean magni-
tude score reduction on the BDI varied from 10 to 13 points, with
17%–50% of clients experiencing sudden gains.
All studies showed larger overall posttreatment symptom reduc-
tions in patients who experienced sudden gains during acute treat-
ment compared with patients who did not. Predictions for long-
term maintenance of gains, however, were mixed. Some studies
showed better outcomes at follow-up for patients with sudden
gains (Gaynor et al., 2003; Hardy et al., 2005; Stiles et al., 2003;
Tang & DeRubeis, 1999b), whereas other studies found no differ-
ences at follow-up for patients with and without sudden gains
(Tang et al., 2002; Vittengl et al., 2005). Vittengl et al. (2005)
reported that patients with sudden gains during acute treatment
actually showed more depressive symptoms and negative failure
attributions at follow-up than those who had not experienced
sudden gains. The authors argued that the sudden gain treatment
response pathway might be more likely to occur for patients who
show higher pretreatment scores on self-report measures of
depression.
The studies examining sudden gains to date have exclusively
focused on depression. The present study is the first to investigate
the occurrence of sudden gains during psychological treatment of
social phobia (social anxiety disorder), the most common form of
anxiety disorder in the population (Kessler et al., 1994). The most
efficacious psychological interventions for social phobia are ex-
posure therapy and cognitive–behavioral treatments (Hofmann &
Barlow, 2002). Contemporary theories of social anxiety and social
phobia emphasize the role of cognitive processes, especially self-
Stefan G. Hofmann, Stefan M. Schulz, Alicia E. Meuret, David A.
Moscovitch, and Michael Suvak, Department of Psychology, Boston
University.
This research was supported by National Institute of Mental Health
Grant MH-57326 to Stefan G. Hofmann and a Gottlieb-Daimler- and Karl
Benz-Foundation grant to Stefan M. Schulz. We would like to thank Nicola
Stenzel for the help in the data reduction.
Correspondence concerning this article should be addressed to Stefan G.
Hofmann, Department of Psychology, Boston University, 648 Beacon
Street, 6th Floor, Boston, MA 02215. E-mail: shofmann@bu.edu
Journal of Consulting and Clinical Psychology
2006, Vol. 74, No. 4, 687–697
Copyright 2006 by the American Psychological Association
0022-006X/06/$12.00DOI: 10.1037/0022-006X.74.4.687
687

Page 2

perception, in the maintenance of the disorder (e.g., Clark & Wells,
1995; Leary & Kowalski, 1995; Rapee & Heimberg, 1997). It has
therefore been suggested that effective psychological treatment
changes the person’s representation of the self in a more positive
direction (e.g., Clark & Wells, 1995; Rapee & Heimberg, 1997).
Supporting evidence for this hypothesis comes from a number of
treatment studies suggesting that effective treatment of social
phobia specifically reduces negative self-perception (Hofmann,
Moscovitch, Kim, & Taylor, 2004; Woody, Chambless, and Glass,
1997).
The objective of the present study was to examine changes in
social anxiety and self-perception during the course of treatment
for social phobia. Treatment consisted of either cognitive–
behavioral group therapy (CBGT) or exposure group therapy
(EGT; without explicit cognitive interventions). On the basis of the
existing sudden gain literature, we predicted that we would be able
to identify sudden gains during both treatment modalities when
adopting the criteria to define sudden gains to the social phobia
treatments. We expected that individuals receiving CBGT would
show more frequent sudden gains and that these gains would be
preceded more often by changes in cognitions (i.e., a reduction in
negative self-statements) than individuals receiving EGT. Finally,
we predicted that individuals with sudden gains would show better
treatment response at the end of treatment and at follow-up than
individuals without such gains.
Method
Participants
Data were compiled from 107 patients who received a principal Diag-
nostic and Statistical Manual of Mental Disorders (4th ed.; DSM–IV;
American Psychiatric Association, 1994) diagnosis of social phobia on the
basis of an assessment with the Anxiety Disorders Interview Schedule for
DSM–IV: Lifetime Version (DiNardo, Brown, & Barlow, 1994). The study
was approved by the Institutional Review Board of the university. All
participants gave their consent by signing a consent form that was reviewed
with participants. Informed consent was obtained by a member of the
research staff at the initial meeting prior to the pretreatment assessment.
Exclusion criteria for this study included (a) prior nonresponse to
adequately delivered study treatment, (b) current diagnosis of psycho-
active substance abuse or dependence, (c) current active suicidal po-
tential, (d) current diagnosis of bipolar disorder, and (e) current diag-
noses of schizophrenia or other psychotic disorders. In addition,
participants had to report at least moderate public speaking anxiety,
which was defined by a self-report rating of 4 or greater on an 8-point
Likert scale. This inclusion criterion was chosen because repeated
exposure to speech situations was an important component of one of the
treatment protocols (EGT). Although this intervention used only public
speaking situations as in-session exposure practices, we found in pre-
vious studies that the treatment effects easily generalized to other social
situations (Hofmann, 2004). Most individuals with social phobia who
presented at our clinic for treatment (88.9%) met all study criteria and
were invited to participate in the study.
Participants were predominantly men (n ? 63; women, n ? 44) and
Caucasian (n ? 94). Non-Caucasian participants identified themselves as
African American (n ? 4), Asian (n ? 4), or Hispanic (n ? 2). The average
age at intake was 32.54 years (SD ? 9.86), with a mean of 16.01 (SD ?
2.46) years of education. They had a mean annual income of $45,762
(SD ? 47,968; Mode ? 30,000; Mdn ? 32,500; range ? 0–310,000).
Seventy-three percent of the patients (n ? 78) were diagnosed with the
generalized subtype of social phobia, including 1 patient with a principal
codiagnosis of avoidant personality disorder. Forty-eight percent of the
participants (n ? 52) received one, and 18% (n ? 19) more than one,
additional secondary clinical diagnosis. The most common comorbid Axis
I diagnoses were major depressive disorder (n ? 33), generalized anxiety
disorder (n ? 12), and dysthymia (n ? 7). A comparison between the two
treatment conditions in the demographic and clinical characteristics is
presented in Table 1.
Table 1
Comparison Between Treatment Conditions in Demographic and Clinical Characteristics
Variable
EGT (n ? 59)CBGT (n ? 48)
?2
tdfpNM SDNMSD
Subject characteristics
Gender
Men
Women
Age (years)
Ethnicity
White
African American
Asian
Hispanic
Other
Family and social background
Income (US $)
Married
Generalized subtype
Additional Axis I diagnoses
Self-report measures at pretest
LSAS
SPAI
0.021.89
35
24
28
20
32.089.78 33.10 10.040.53 105.60
.990.001
52 42
3
2
1
1
1
2
1
2
44,72642,922 47,118 54,4200.23 88
1
1
1
.82
.27
.34
.23
14
44
42
15
33
30
1.23
0.89
1.45
40.75
102.35
18.10
27.81
44.58
108.01
15.79
24.74
1.11
1.10
95.27
.28102
Note.
and Anxiety Inventory.
EGT ? exposure group therapy; CBGT ? cognitive–behavioral group therapy; LSAS ? Liebowitz Social Anxiety Scale; SPAI ? Social Phobia
688
HOFMANN, SCHULZ, MEURET, MOSCOVITCH, AND SUVAK

Page 3

Treatments
Participants attended at least 8 of 12 weekly sessions of CBGT (n ? 48)
or EGT (n ? 59). CBGT is a comprehensive cognitive–behavior treatment
protocol (Heimberg, 1991), whereas EGT is an exposure treatment without
explicit cognitive interventions (Hofmann, 1999). Some of the participants
were part of a previous study (Hofmann, 2004). As part of this study,
participants were randomly assigned to one of two treatment modalities
(CBGT vs. EGT).1Participants in the two groups did not differ in marital
status, gender, or education status (ps ? .2). However, participants in the
EGT group were slightly older (in years, M ? 34.39, SD ? 10.29) than
participants in the CBGT group (M ? 30.27; SD ? 8.89), t(105) ? 2.19,
p ? .03.
The treatment attrition rates were 16% (9 of 57) from CBGT and 22%
(17 of 76) from EGT. The two study treatments did not differ in their
attrition rates, ?2(1) ? 0.66, p ? .40. Furthermore, attrition from the study
was not associated with any demographic variables or Axis I symptom-
atology (ps ? .20, ds ? 0.17). More detailed results on these attrition data
will be reported elsewhere (Hofmann & Suvak, in press).
The two study treatments were conducted in a group format with two
therapists and 5–7 participants per group and delivered according to
structured treatment protocols. Nine therapists with comparable experience
had been trained to implement the two treatments. The training consisted
of (a) reading the detailed treatment protocol, (b) listening to an audiotaped
first treatment session (in which the treatment rationale was presented), (c)
attending weekly supervision meetings, and (d) coleading at least one
complete treatment group. All therapists were advanced doctoral students
in clinical psychology. Harlan Juster, a senior therapist and former collab-
orator of Richard G. Heimberg, the author of the CBGT protocol, provided
weekly telephone supervision for the CBGT groups. Stefan G. Hofmann,
who developed the EGT manual, supervised the EGT therapists on a
weekly basis. The EGT protocol included repeated in-session in-vivo
exposures to social performance situations, video feedback, didactic train-
ing, and weekly homework assignments. Although participants feared
numerous social situations, this intervention focused primarily on the
patients’ public speaking anxiety. We limited the exposure situations of the
EGT group to repeated public speaking situations because public speaking
is the most commonly feared social situation that can easily be created in
a group setting. Previous studies have shown that treatments that primarily
target public speaking anxiety generalize to other social fears in that they
have similar acute treatment effects on generalized social anxiety symp-
toms as more comprehensive treatments (e.g., Newman, Hofmann, Trabert,
Roth, & Taylor, 1994). Moreover, we modified the topic or setting of the
situation to elicit at least moderate anxiety even in participants who did not
rate public speaking as their main social fear. For example, a participant
who reported dating as his or her main fear might have been asked to give
a speech on “dating rules” or “my most embarrassing moment during
dating,” and a person who was concerned with assertiveness situations and
disagreeing with others might have been asked to give a speech on a
controversial topic, in which he or she argued against the majority opinion
of the group.
In contrast, patients in the CBGT groups were taught skills to identify
negative cognitions, observe the covariation between anxious mood and
cognitions, examine thinking errors—including overestimation of social
cost—and formulate rational alternatives to these errors. Exposure exer-
cises were conducted in session and assigned for homework. Exposures
were explained as a means by which patients could scientifically test the
validity of anxious predictions. In addition to the difference in the emphasis
on cognitive strategies, the two protocols also differed in the treatment
rationale, which was based on the cognitive model of anxiety in the case of
CBGT and on a basic habituation rationale in the case of EGT. The results
from previous meta-analyses (Feske & Chambless, 1995; Gould, Buck-
minster, Pollack, Otto, & Yap, 1997; Taylor, 1996) suggest that both
treatment protocols produce comparable short-term treatment effects.
Treatment Integrity
A random sample of 41 treatment sessions were audiotaped (19 different
CBGT sessions and 22 different EGT sessions) and selected to be evalu-
ated blindly for protocol adherence and therapist competence by a 5th-year
doctoral student in clinical psychology. This person had been trained in
various empirically supported interventions and closely supervised in her
clinical work by experienced and licensed clinicians. She was further
familiar with both treatment protocols involved in the study. To examine
the reliability of the rating system, a master’s student also blindly evaluated
a random subsample of 22 taped sessions (11 CBGT tapes and 11 EGT
tapes).
Therapist adherence to the two protocols was determined by rating the
following items on a 5-point Likert scale (1 ? not met, 3 ? partly met, 5 ?
fully met): “The therapists used a ‘purely’ behavioral model of social
phobia (i.e., without specific cognitive interventions)” and “The therapists
used cognitive restructuring techniques.” We estimated interrater agree-
ment by calculating intraclass correlation coefficients (ICC; Shrout and
Fleiss, 1979). Specifically, we used Shrout and Fleiss’s ICC (2, 1) equa-
tion, which assumes that the two raters are random effects. The results
suggest that the two raters showed high agreement in the degree to which
therapists used a “purely” behavioral model (ICC ? 0.82, p ? .0001).
Moreover, the two raters agreed in their assessment of the extent to which
therapists used cognitive restructuring techniques (ICC ? 0.995, p ?
.0001). The EGT therapists used a more behavioral model (M ? 4.77,
SD ? 0.87) relative to CBGT therapists (M ? 1.00, SD ? 0.00), t(21) ?
20.36, p ? .0001, d ? 5.97, whereas the CBGT therapists used relatively
more cognitive restructuring techniques (M ? 4.79, SD ? 0.92) relative to
EGT therapists (M ? 1.27, SD ? 0.55), t(39) ? 15.12, p ? .0001, d ?
4.49.
The senior rater further evaluated the 41 tapes for therapist competency
using an adaptation of the Vanderbilt Psychotherapy Process Scale (Suh,
Strupp, & O’Malley, 1986). Specifically, the rater evaluated each tape
using a 5-point Likert scale (1 ? not at all, 2 ? some, 3 ? fair amount, 4 ?
pretty much, and 5 ? great deal) to characterize the therapeutic relation-
ship and the therapist’s demeanor. The results showed that both treatment
groups were rated high on the general quality of the relationship between
the group members and the therapists (CBGT: M ? 4.53, SD ? 0.51; EGT:
M ? 4.59, SD ? 0.59), on their productivity (CBGT: M ? 4.57, SD ?
0.50; EGT: M ? 4.41, SD ? 0.50), and on how well the therapists and the
group members worked together (CBGT: M ? 4.74, SD ? 0.45; EGT:
M ? 4.82, SD ? 0.39).
1The randomized clinical trial was based on 90 participants. To maxi-
mize our sample size, we included additional individuals who were part of
the regular client flow. Those individuals were assigned, after a short
waiting period, to one of the two treatment groups (CBGT vs. EGT),
depending on therapist availability, which explains the unequal number of
participants in the two conditions. In addition, we included participants
who were part of the waiting list in the randomized, controlled trial.
Therefore, the present study deviates from a randomized clinical trial, and
we have insufficient data available to present a flow diagram of the
progress. A comparison between the 90 participants who were part of the
randomized trial and the 17 participants who were later added to the
analyses showed no differences in any demographic characteristics, includ-
ing gender, age, ethnicity, education, income, marital status, and comor-
bidity (ps ? .5). Furthermore, there was no difference in the Liebowitz
Social Anxiety Scale (LSAS; Liebowitz, 1987) at pretest (p ? .5). The
only difference was a slightly higher Social Phobia and Anxiety Inventory
(SPAI; Turner, Beidel, Dancu, & Stanley, 1989) difference score in par-
ticipants from the randomized trial at pretreatment, t(102) ? 2.10, p ? .04.
However, the results of the subsequent analyses on sudden gains did not
change when we excluded these 17 participants from the analyses.
689
SUDDEN GAINS IN SOCIAL PHOBIA

Page 4

Moreover, the therapists’ demeanor during the sessions was rated as
being involved (CBGT: M ? 5.00, SD ? 0.00; EGT: M ? 4.91, SD ?
0.29), optimistic (CBGT: M ? 4.53, SD ? 0.70; EGT: M ? 4.5, SD ?
0.51), and respectful (CBGT: M ? 4.95, SD ? 0.23; EGT: M ? 4.91,
SD ? 0.29). Therapists from neither group appeared to be annoyed
(CBGT: M ? 1.05, SD ? 0.23; EGT: M ? 1.00, SD ? 0.00), authorita-
tive (CBGT: M ? 1.26, SD ? 0.45; EGT: M ? 1.18, SD ? 0.39), defensive
(CBGT: M ? 1.00, SD ? 0.00; EGT: M ? 1.00, SD ? 0.00), or judgmental
(CBGT: M ? 1.11, SD ? 0.32; EGT: M ? 1.00, SD ? 0.00; all ps ? .1,
all ds ? 0.44). Because only the senior rater gave competency ratings, no
reliability data were available for these data.
Assessments
Before treatment, all participants received the lifetime version of the
Anxiety Disorders Interview Schedule for DSM–IV. This interview lasted
between 4 and 6 hr per participant and was conducted by advanced clinical
psychology doctoral students, who were unaware of the objective of this
study. The kappa coefficient between two independent raters at our Center
was .77 for social phobia as a principal diagnosis (Brown, DiNardo,
Lehman, & Campbell, 2001). The reliability coefficients (kappa) for the
other anxiety diagnoses ranged between .67 (generalized anxiety disorders)
and .86 (specific phobias) and between .22 (dysthymia) and .72 (major
depressive disorder) for mood disorders (Brown et al., 2001).
Additional assessments occurred immediately after treatment (posttest)
and at 6-month follow-up (without receiving any additional treatments
between the posttest and the follow-up assessment). All participants en-
tered into the study with the understanding that they were not permitted to
initiate any additional therapy for their social phobia for the duration of the
study. This information was included in the consent form. None of the
participants, including those who received a 6-month follow-up, initiated
any psychological or pharmacological treatments during the course of the
study.
At all assessment points, participants were asked to fill out the Social
Phobia and Anxiety Inventory (SPAI; Turner, Beidel, Dancu, & Stanley,
1989). The SPAI is a 109-item self-report instrument that has been widely
used to assess the cognitive, somatic, and behavioral dimensions of social
phobia. This measure is capable of discriminating socially phobic persons
from those with other anxiety disorders (Turner et al., 1989) and from
normal controls (Beidel, Borden, Turner, & Jacob, 1989). Internal consis-
tency estimates of the SPAI are high, ranging from .85 to .96 (Osman et al.,
1996; Turner et al., 1989). The internal consistency of the SPAI difference
score in the current sample was .99 at pretest, posttest, and follow-up.
Stability of the SPAI has also been found to be high, with Turner et al.
(1989) reporting a 2-week test–retest reliability of .86 on the SPAI differ-
ence score in a sample of college students. Compared with other popular
measures of social anxiety, the SPAI was among the most reliable instru-
ments and the most sensitive to treatment change and for distinguishing
social phobia subtypes (Ries et al., 1998). Therefore, the SPAI was chosen
as the primary outcome variable and measure of treatment change.
Prior to each session, participants’ level of social anxiety was measured
with a self-report version of the Liebowitz Social Anxiety Scale (LSAS;
Liebowitz, 1987). This is a 24-item scale to measure fear and avoidance in
24 different social situations. The clinician-administered scale is widely
used in studies of pharmacological treatment of social phobia. A recent
study using a large sample of individuals with social phobia examined the
psychometric characteristics of the self-report version of the LSAS (Baker,
Heinrichs, Kim, & Hofmann, 2002). The self-report version of the LSAS
was highly correlated with the clinician version, with correlations between
these versions ranging from .78 to .85, depending on the subscales. Baker
and colleagues (2002) also reported good 12-week test–retest reliability for
the total score (r ? .83). The internal consistency for the total score
(Cronbach’s ? ? .95) was high. The self-report version of the instrument
was also sensitive to treatment change. For the purpose of this study, we
averaged the ratings of all fear and avoidance items, which were scaled
from 0 to 100. The LSAS was chosen as a session-by-session measure of
treatment change.
In addition to these general measures of social anxiety, we also exam-
ined changes in cognitions related to social anxiety. For this purpose, we
administered the Self-Statement During Public Speaking Test (SSPS, Hof-
mann & DiBartolo, 2000) prior to each session. The SSPS is a 10-item
instrument that has two 5-item subscales to measure positive (SSPS-P) and
negative self-statements (SSPS-N) related to a public speaking situation.
The 2-factor structure of the measure was replicated in student and clinical
samples. The internal consistency in two large undergraduate student
samples ranged from .75 to .84 for the SSPS-P and from .83 to .86 for the
SSPS-N. Similar results were found in clinical samples (Hofmann &
DiBartolo, 2000). Clinical data showed that the SSPS-N (but not the
SSPS-P) was sensitive to treatment changes. Participants who scored high
on the SSPS-N had lower expectations for their success in a public
speaking task, reported more anxiety during the task, and were less satis-
fied with their performance than participants who showed low SSPS-N
scores (Hofmann & DiBartolo, 2000). We specifically examined the neg-
ative subscale scores of this scale as a measure of negative self-statements,
which appear to be a mediator of treatment change in social phobia (as
opposed to positive self-statements; Hofmann et al., 2004).
The internal consistencies (Cronbach’s ?) of the measures that were
administered to the current sample at each session were high for the LSAS
(.90–.96), SSPS-P (.86–.91), and SSPS-N (.83–.91).
Definition of Sudden Gains
Consistent with the terminology introduced by Tang and DeRubeis
(1999b), the session immediately preceding the sudden gain is referred to
as pregain session (N ? 1), and the session preceding the pregain session
is referred to as prepregain session (N ? 2). The session immediately after
the sudden gain is referred to as the aftergain session (N ? 1). The criteria
for sudden gains were as follows:
Criterion 1.
The gain between session N and session N ? 1 should be
large in absolute terms. Some studies (e.g., Hardy et al., 2005; Stiles et al.,
2003) defined the magnitude of the gain on the basis of the reliable change
index (Jacobson & Truax, 1991). Consistent with this methodology, we
derived the LSAS cutoff score for a sudden gain by dividing the LSAS
change score from Sessions 1 to 12 by the standard error of the LSAS
difference scores. This resulted in a critical LSAS difference score of
10.30. Tang and DeRubeis (1999b) reported that “for patients who expe-
rienced sudden gains, the sudden gains accounted for an average of 51% of
their total symptom reduction” (p. 896). When we used an LSAS cutoff
score of 10, the sudden gains accounted for an average of 50.27%. When
we used the next higher cutoff score (11), the sudden gains accounted for
56.79% of the total symptom reduction. Therefore, to approximate Tang
and DeRubeis’ (1999b) methodology and to be consistent with the previous
sudden gain literature, we chose an LSAS difference score of 10 as the
cutoff to define a sudden gain.2
2The choice of the cutoff criterion for a sudden gain is arbitrary, and
other strategies are possible. For example, one of the reviewers suggested
that we define the cutoff as the score that is one standard deviation from the
clinical sample mean. In fact, Tang and DeRubeis (1999b) chose 7 BDI
points as the cutoff score, which happens to be approximately one standard
deviation from the mean of depressed individuals. This strategy would
translate to an LSAS cutoff score of 16. When we used this cutoff score,
only 6 out of a total of 967 between-session intervals featured sudden
gains, too few to conduct any meaningful statistical analyses. Although this
is a plausible method, there is no precedence in the sudden gains literature
for choosing one standard deviation as the cutoff rule. More important, it
690
HOFMANN, SCHULZ, MEURET, MOSCOVITCH, AND SUVAK

Page 5

Criterion 2.
severity before a sudden gain. Hence, the gain of the LSAS score at session
N ? 1 was required to be at least 25% of the score at session N.
Criterion 3.
Finally, an improvement should be large in relation to
symptom fluctuations before and after the gain. Tang and DeRubeis
(1999b) used an independent t test, comparing the mean score of three
therapy sessions before the gain (sessions N ? 2, N ? 1, and N) with the
mean score of the three therapy sessions after the gain (N ? 1, N ? 2, N ?
3). This criterion was subsequently criticized because of autocorrelation of
the data (e.g., Hardy et al., 2005). Therefore, many studies have used the
critical value, t(4) ? 2.78, as the cutoff criterion to identify a sudden gain
(e.g., Hardy et al., 2005; Vittengl et al., 2005).
We used both criteria and obtained identical results. Following the
recommendations by Tang and DeRubeis (1999b), changes in the first
session or before the last session were excluded from the analyses.
The improvement should be large relative to symptom
Handling of Missing Data
To establish Criteria 1 and 2, we compared two adjacent sessions. Of a
total of 1,284 adjacent sessions, 139 sessions had missing data, with
statistically equal numbers of missing data in each group (63 of 708
sessions in EGT and 76 of 576 in CBGT), ?2(1) ? 0.05, p ? .82.
Therefore, 210 of 1,177 session to session comparisons could not be
performed (93 of 210 in EGT and 117 of 210 in CBGT), ?2(1) ? 1.66, p ?
.20.
To establish Criterion 3, we conducted independent-samples t tests to
compare the mean of the three consecutive LSAS session scores before a
possible sudden gain with the mean of the three consecutive LSAS scores
after a possible sudden gain. Five comparisons in Criterion 3 had missing
data (EGT ? 2; CBGT ? 3). One comparison in the EGT group was
excluded because of significant missing data. The remaining 4 tests were
performed with at least one missing data point (the one test in the EGT
group had two missing data points, whereas each of the comparisons in the
CBGT group had only one missing data point). The number of comparisons
with missing data points did not differ between the two groups (CBGT: 3
of 112 vs. EGT: 1 of 27), ?2(1) ? 0.14, p ? .71. Results remained the same
regardless of whether these tests were performed with fixed degrees of
freedom (4) or whether degrees of freedom were adjusted to account for
missing data.
To examine whether missing data might have influenced our results, we
followed up each repeated measures analysis of variance (ANOVA) with
an analogous analysis in a multilevel regression framework using the
software program Hierarchical Linear Modeling–5 (HLM-5; Raudenbush,
Bryk, Cheong, & Congdon, 2001). Multilevel, or mixed-effects, regression
models have been increasingly used as an alternative technique to an
ANOVA approach for analyzing repeated measure data (e.g., Bryk &
Raudenbush, 1992; Raudenbush & Bryk, 2002). One advantage of this
approach is that it is more efficient in analyzing unbalanced designs (i.e.,
when the number of data points and/or the time between data points varies
from person to person). The results did not differ between the two methods.
Therefore, we reported the repeated measures ANOVA analyses to facil-
itate comparisons to existing studies.
Results
Differences in Treatment Groups
To examine the short-term treatment effects, we conducted a 2
(group: EGT vs. CBGT) ? 2 (time: pretreatment vs. posttreat-
ment) repeated measures ANOVA with the SPAI difference score
as the dependent variable. The results showed a significant time
effect, F(1, 81) ? 77.40, p ? .001, partial ?2? .49. The group
effect, F(1, 81) ? 0.03, p ? .86, partial ?2? .01, and Time ?
Group interaction effect, F(1, 81) ? 2.73, p ? .10, partial ?2?
.03, were not significant.
To examine the long-term treatment effects, we used a 2 (group:
EGT vs. CBGT) ? 2 (time: post vs. follow-up) repeated measures
ANOVA. The findings revealed no significant time effect, F(1,
49) ? 0.47, p ? .50, partial ?2? .01, group effect, F(1, 49) ?
1.18, p ? .28, partial ?2? .02, or Time ? Group interaction, F(1,
49) ? 2.22, p ? .14, partial ?2? .04.3The statistical assumptions
were met for these and all other tests.
Occurrence of Sudden Gains
A total of 20 of 107 patients (18.69%) experienced at least one
sudden gain. Two patients had two sudden gains. Therefore, only
22 of a total of 967 between-sessions intervals featured sudden
gains. Sudden gains were distributed throughout the whole course
of treatment. There were 7 individuals with sudden gains (41
without sudden gains) in the CBGT group and 13 individuals with
sudden gains (46 without) in the EGT group. This difference was
not statistically significant, ?2(1) ? 0.33, p ? .57. The sudden
gains occurred most commonly in Session 5 (4 sudden gains) in
the CBGT group and in Sessions 4 (2 sudden gains) and 11 (2
sudden gains) in the EGT group. When combining the two treat-
ment groups, the gains occurred most commonly in Session 5 (5
sudden gains).4
The mean magnitude of all sudden gains (Figure 1) was 15.78
LSAS points (SD ? 7.20). Individuals who experienced sudden
3Because the two treatment groups differed in age, we also conducted 2
(group) ? 2 (time) repeated measures analyses of covariance with age as
a covariate and SPAI difference score as the dependent variable. For the
pre–post comparison, the results showed again a significant time effect,
F(1, 80) ? 5.91, p ? .02, partial ?2? .07, but no Time ? Group
interaction effect, F(1, 80) ? 2.68, p ? .11, partial ?2? .03, or group
effect, F(1, 80) ? 0.02, p ? .90, partial ?2? .001. The age effect and
Time ? Age interaction effect were not statistically significant (ps ? .24).
For the post versus follow-up comparison, the results again showed no
significant time effect, F(1, 48) ? 0.01, p ? .97, partial ?2? .001, group
effect, F(1, 48) ? 1.27, p ? .27, partial ?2? .03, Time ? Group
interaction effect, F(1, 48) ? 2.16, p ? .15, partial ?2? .04, age effect, or
Time ? Age interaction effects (ps ? .14). Similar results were obtained
when choosing the LSAS as the dependent variable.
4To examine whether the group format of the treatment systematically
influenced the results (e.g., whether participants in the same treatment
group experienced sudden gains at the same time), we used HLM, a
Bernoulli regression model, by means of the software program HLM-5.
HLM has been developed for the analyses of nested data structures in
which 1 participant contributes multiple data points at various time points.
This procedure allows for unequal numbers of time series data points per
person and variable spacing of time points across individuals (e.g., Bryk &
Raudenbush, 1992; Raudenbush & Bryk, 2002). Level 1 of the multilevel
is questionable whether changes of one standard deviation in LSAS scores
are clinically comparable to changes of one standard deviation in BDI
scores in a different clinical sample, because of differences in the mea-
surement constructs (depression vs. social anxiety), the measurement in-
struments, and the sample characteristics. In fact, when we used an LSAS
cutoff score of 16, the sudden gains accounted for 63.57% (M ? 23.59,
SD ? 10.45) of the total symptom reduction (M ? 37.11, SD ? 17.52),
which is substantially higher than what has been reported in the sudden
gains literature on depression. Therefore, we adopted the strategies by
Hardy et al. (2005) and Stiles et al. (2003) for defining the cutoff score.
691
SUDDEN GAINS IN SOCIAL PHOBIA

Page 6

gains improved by 31.39 points (SD ? 13.31) from pre- to post-
treatment. Therefore, sudden gains accounted for 50.27% of par-
ticipants’ overall improvement. In contrast, individuals who did
not experience a sudden gain only improved by 14.96 LSAS points
(SD ? 14.32). This difference was statistically significant, t(23) ?
4.03, p ? .001, d ? 1.19. Patients experiencing sudden gains had
significantly higher LSAS scores at pretreatment (M ? 51.07,
SD ? 11.00) than patients without sudden gains (M ? 40.39, SD ?
17.53), t(73) ? 2.25, p ? .03, d ? 0.73. Table 2 shows the
comparison of patients with sudden gains in the two treatment
groups.
Reversal of Sudden Gains
Consistent with the sudden gains literature, we examined how
often participants gave up 50% or more of the gain achieved
between Sessions N and N ? 1 (i.e., a score greater than LSASN ? 1
? .5 [LSASN– LSASN ? 1] for one or more sessions following the
sudden gain session). When this criterion was used, 2 patients
showed a reversal of their sudden gains.
Sudden Gains and Treatment Groups
To examine whether sudden gainers showed the same degree of
change in social anxiety (as measured with the SPAI) during
treatment as individuals without sudden gains, we used a 2 (group:
individuals with vs. without sudden gains) ? 2 (time: pre vs. post)
repeated measures ANOVA with the SPAI difference score as the
dependent variable. The results revealed a significant time effect,
F(1, 81) ? 70.50, p ? .001, partial ?2? .47, and a significant
Time ? Group interaction effect, F(1, 81) ? 4.91, p ? .03, partial
?2? .06. Sudden gainers improved more during the course of
treatment (M ? 34.18, SD ? 25.65) than patients without sudden
gains (M ? 19.90, SD ? 23.17), t(81) ? 2.22, p ? .03, d ? 0.58
(Figure 2). The group effect was not statistically significant, F(1,
81) ? 1.80, p ? .18, partial ?2? .02.
To examine longer-term changes, we used a 2 (group: individ-
uals with sudden gains vs. those without) ? 2 (time: post vs.
6-month follow-up) repeated measures ANOVA with the SPAI
difference score as the dependent variable. The results showed
no significant time effect, F(1, 49) ? 0.14, p ? .71, partial ?2?
.003, group effect, F(1, 49) ? 1.15, p ? .29, partial ?2? .02,
or Time ? Group interaction effect, F(1, 49) ? 0.003, p ? .96,
partial ?2? .001.
Figure 1.
(LSAS) shown for Sessions N ? 2, N ? 1, N, N ? 1, N ? 2, and N ? 3 are the means (with error bars
representing standard errors) of the corresponding sessions from the 20 sudden gains in our sample, with session
N as the pregain session. The figure also shows the first and last sessions of all patients who experienced sudden
gains.
Mean sudden gains in the total sample. The total scores of the Liebowitz Social Anxiety Scale
model consisted of individual participants who were nested within the
treatment groups, the Level 2 component of the model. For the current
analyses, 107 participants were nested within 24 groups. Examining the
variance component of the unconditional model revealed that the proba-
bility of experiencing a sudden gain did not significantly vary across
groups, ?2(23) ? 26.17, p ? .29. Because there was no effect of the nested
structure of the data (i.e., individuals nested within groups) on the proba-
bility of experiencing a sudden gain, all subsequent analyses did not model
treatment group-level variance.
692
HOFMANN, SCHULZ, MEURET, MOSCOVITCH, AND SUVAK

Page 7

To further examine whether individuals with sudden gains in the
two treatment groups differed in the magnitude of the LSAS
treatment change, we conducted a 2 (time: pre vs. post) ? 2
(group: EGT vs. CBGT) ? 2 (sudden gains: individuals with
sudden gains vs. those without) repeated measure ANOVA with
the LSAS as the dependent variable. Results showed the expected
time effect, F(1, 71) ? 113.34, p ? .001, partial ?2? .62, Time ?
Group effect, F(1, 71) ? 4.68, p ? .03, partial ?2? .06, and
Time ? Sudden Gain effect, F(1, 71) ? 11.56, p ? .001, partial
?2? .14. The group effect, F(1, 71) ? 1.74, p ? .19, partial ?2?
.02, sudden gain effect, F(1, 71) ? 0.64, p ? .43, partial ?2? .01,
Group ? Sudden Gain interaction, F(1, 71) ? 0.04, p ? .84,
partial ?2? .001, and Time ? Group ? Sudden Gain interaction,
F(1, 71) ? 0.50, p ? .48, partial ?2? .01, were not statistically
significant. The latter result suggests that individuals with sudden
gains showed similar improvements in the two treatment groups.
Because the sudden gain phenomenon has been studied primar-
ily in the depression literature, we further examined whether the
sudden gains were specifically associated with major depressive
disorder (MDD), which was the most common comorbid condition
in the present sample. Of the 33 participants who had a comorbid
diagnosis of MDD, 9 (27%) experienced a sudden gain, whereas of
Figure 2.
Inventory (SPAI) scores from pretreatment (pre), to posttreatment (post), and to 6-month follow-up (FU) among
individuals who showed sudden gains (black circles) and those who did not (white circles).
Mean change scores (with error bars representing standard errors) in Social Phobia and Anxiety
Table 2
Comparison of Patients With Sudden Gains in EGT and CBGT
VariableEGT (n ? 59)CBGT (n ? 48)
?2
t dfp
Individuals with sudden gains (N)
Mode session of sudden gain
Median session of sudden gain
Average magnitude (% total
improvement)
Magnitude of first sudden gains
M
SD
Range
Total improvement of patients
experiencing sudden gains
M
SD
Range
13
5
7
70.541 .46
4 and 11
8
49.2860.00
0.98 18.34
17.44
8.94
14.03
2.18
10.46–44.38 11.25–17.08
1.7713 .10
35.39
14.20
23.38
6.83
16.66–63.32 15.84–29.40
Note.
EGT ? exposure group therapy; CBGT ? cognitive–behavioral group therapy.
693
SUDDEN GAINS IN SOCIAL PHOBIA

Page 8

the 74 participants who did not have a MDD diagnosis, 11 (15%)
experienced sudden gains. This difference was not statistically
significant, ?2(1) ? 2.31, p ? .13. Of the 59 participants assigned
to the EGT group, 20 (34%) had a comorbid MDD diagnosis,
whereas of 48 participants assigned to the CBGT group, 13 (27%)
had a comorbid MDD. This difference was not statistically signif-
icant, ?2(1) ? 0.58, p ? .45.
Cognitive Changes Preceding Sudden Gains
Changes in cognitions were measured with the SSPS-N (Hof-
mann & DiBartolo, 2000), which was administered together with
the LSAS at the beginning of each session.5We selected the
SSPS-N because studies have shown that self-focused attention
(Woody & Rodriguez, 2000; Woody et al., 1997) and in particular
negative self-focused attention (Hofmann, 2000; Hofmann et al.,
2004) enhances social anxiety and decreases with successful
treatment.
To examine cognitive change prior to the sudden gains, we
computed change scores of the cognitive measure from the begin-
ning of the prepregain session (N ? 1) to the beginning of the
pregain session (N). We refer to this as changes before pregain
sessions. We used the cognitive change from the session before the
prepregain session (N ? 2) to the beginning of the pregain session
(N ? 1) as a control interval (prepregain change).
A repeated measures ANOVA with group (EGT vs. CBGT) as
the between factor and time (changes before pregain vs. changes
before prepregain change) as the within-subjects factor revealed a
significant main effect of group, F(1, 16) ? 6.25, p ? .024, partial
?2? .28. CBGT was associated with greater cognitive change
overall. However, the main effect of time, F(1, 16) ? 0.02, p ?
.88, partial ?2? .00, and the Time ? Group interaction were not
significant, F(1, 16) ? 0.50, p ? .49, partial ?2? .03.
To explore whether sudden gains were accompanied by cogni-
tive changes in the pregain session and the interval between the
pregain session and the aftergain session, we computed a repeated
measures ANOVA with group (EGT vs. CBGT) as the between
factor, and time (beginning of pregain session vs. beginning of
aftergain session) as the within-subjects factor for the SSPS-N.
This analysis showed no significant main effect of group, F(1,
17) ? 1.54, p ? .23, partial ?2? .08, time, F(1, 17) ? 2.73, p ?
.117, partial ?2? .14, or Time ? Group interaction, F(1, 17) ?
0.62, p ? .44, partial ?2? .04.
In addition to the SSPS-N, we also examined the ratio SSPS-P/
(SSPS-P ? SSPS-N). This ratio is derived from the states of mind
model (Schwartz & Garamoni, 1989) and measures the balance be-
tween positive and negative thoughts. The results showed, again, a
significant main group effect, F(1, 16) ? 5.86, p ? .028, partial ?2?
.27, but no time effect, F(1, 16) ? 1.88, p ? .19, partial ?2? .11, or
Time ? Group interaction, F(1, 16) ? 0.38, p ? .549, partial ?2?
.02. As expected, participants in the CBGT group showed greater
treatment changes (M ? 0.43, SD ? 0.13) than participants in the
EGT group (M ? 0.34, SD ? 0.29). Analyses of the SSPS-P revealed
no significant main effect of group, F(1, 16) ? 3.23, p ? .091, partial
?2? .17, time, F(1, 16) ? 2.41, p ? .140, partial ?2? .13, or
Time ? Group interaction, F(1, 16) ? 1.13, p ? .304, partial ?2?
.07. Similar results were found when examining changes in the
SSPS-P and the ratio SSPS-P/(SSPS-P ? SSPS-N) from the begin-
ning of the pregain to the beginning of the aftergain session.
Discussion
Sudden gains are enduring reductions in symptom intensity
from one session to the next, which have, thus far, exclusively
been studied in the treatment of depression (Gaynor et al., 2003;
Hardy et al., 2005; Tang & DeRubeis, 1999b; Tang et al., 2002;
Tang et al., 2005; Vittengl et al., 2005). It remains uncertain
whether the sudden gain phenomenon can also be observed in
disorders other than depression. Moreover, the factors that deter-
mine these gains remain largely unexplored. Some authors assume
that the therapeutic strategies in cognitive–behavioral therapy that
encourage cognitive change are responsible for sudden gains in
treatment (e.g., Hollon, 1999; Tang & DeRubeis, 1999a). This is
consistent with the cognitive model, which assumes that changes
in cognitions precede symptom improvements in therapy. In con-
trast, other authors assume that common factors, rather than spe-
cific cognitive factors, are responsible for sudden gains (Ilardi &
Craighead, 1994, 1999; Lambert, 2005). For example, Ilardi and
Craighead (1994) argued that most of the symptom improvement
in cognitive–behavioral therapy for depression (60%–70%) ap-
pears to occur during the first 4 weeks of treatment and cannot,
therefore, be explained by cognitive modification. The objective of
this study was to examine the phenomenon of sudden gains during
cognitive–behavioral therapy and exposure therapy for social pho-
bia. We expected that individuals receiving cognitive–behavioral
therapy would show more frequent sudden gains. We further
predicted that these gains would be preceded more often by
changes in cognition in individuals receiving cognitive–behavioral
therapy compared with those receiving exposure therapy without
explicit cognitive interventions.
Consistent with the depression literature, our results showed that
some treatment sessions are characterized by relatively large im-
provements in social anxiety from one session to the next in a
subgroup of individuals. However, sudden gains were not more
commonly associated with cognitive–behavioral intervention and
changes in cognitions. It should be noted, however, that these
results cannot rule out the possibility that important cognitive
changes did occur during the pregain sessions. The measurement
interval for cognitive changes in our study was from the beginning
of the prepregain session to the beginning of the pregain session.
In contrast, Tang and DeRubeis (1999b) emphasized the impor-
tance of examining in-session cognitive changes within the pregain
session. Moreover, the degree of cognitive intervention was not the
only distinguishing characteristic between the two study groups.
The treatments also differed in the treatment rationale participants
received and the type of exposure situations participants under-
went (EGT exposures consisted of public speaking tasks, whereas
CBGT exposures consisted of a number of different situations).
Therefore, future studies should examine cognitive changes within
the pregain session, perhaps by designing a questionnaire that asks
patients about cognitive changes within that session and uses an
additive design to examine the specific effects of cognitive inter-
ventions on sudden gains.
To define, identify, and investigate sudden gains, we closely
followed the methodology by Tang and DeRubeis (1999b), who
5This procedure to measure cognitive change differed from the method
used by Tang and DeRubeis (1999b), who measured cognitive change by
rating audiotaped recordings of the pregain sessions.
694
HOFMANN, SCHULZ, MEURET, MOSCOVITCH, AND SUVAK

Page 9

first described this phenomenon. For this reason, we primarily
limit the comparison between our results and the findings from the
depression literature to the original Tang and DeRubeis study. The
cutoff score to define a sudden gain in Tang and DeRubeis’ study
accounted for an average of 51% of the total symptom reduction in
LSAS scores. Similarly, the cutoff score we chose to define a
sudden gain accounted for 50.27% overall improvement among
those participants who experienced sudden gains. We observed 22
sudden gains out of 967 between-session intervals (2.23%). This is
consistent with Tang and DeRubeis’ study, which found 29 sudden
gains out of 927 between-session intervals (3.13%). Tang and
DeRubeis further reported that the median pregain session of a
16–20 session course of individual cognitive–behavioral therapy
for depression occurred in Session 5. Similarly, we observed that
most sudden gains of a 12-session group treatment occurred in
Session 5. In case this finding is reliable, future research should
explore whether the fifth session contains any unique characteris-
tics. Finally, consistent with the study by Tang and DeRubeis,
sudden gains were not associated with better outcomes at
follow-up.
In addition to these similarities, we also observed a number of
important differences in our results compared with those reported
by Tang and DeRubeis (1999b). First, the percentage of patients
experiencing a sudden gain in our sample was lower than what was
found in the Tang and DeRubeis study (18.67% vs. 39.34%).
However, both percentages are within the range of what would be
expected on the basis of the sudden gains literature, which showed
that between 17% and 50% of clients experience sudden gains
(Gaynor et al., 2003; Hardy et al., 2005; Stiles et al., 2003; Tang
et al., 2002, 2005; Vittengl et al., 2005). Second, we only identified
2 patients (1.9%) with a reversal of sudden gains. In contrast, Tang
and DeRubeis found that 17% of their sample experienced a
reversal in sudden gains. Other studies identified between 9%
(Gaynor et al., 2003) and 47% (Tang et al., 2002) of individuals
with reversed sudden gains. This may suggest that individuals with
social phobia show a more stable rate of improvement as compared
with individuals with depression. It is further possible that the
group format in which the social phobia treatments were con-
ducted had a moderating and stabilizing effect on the treatment
gains. It should be noted, however, that the probability of experi-
encing a sudden gain did not significantly vary across the treat-
ment groups. Third, Tang and DeRubeis (1999a) reported that
participants with sudden gains had better outcome at posttreat-
ment, 6-month, and 18-month follow-up assessments. In contrast,
we observed no differences in outcome between individuals with
and without sudden gains at the posttreatment or 6-month
follow-up assessments. This finding is in line with some studies
(Tang et al., 2002; Vittengl et al., 2005) but at odds with other
reports (Gaynor et al., 2003; Hardy et al., 2005).
It is interesting that we found that participants with sudden gains
had higher levels of pretreatment social anxiety. Similarly, previ-
ous studies on sudden gains during the treatment for depression
also observed higher levels of pretreatment symptomatology in
individuals who later experienced sudden gains (e.g., Vittengl et
al., 2005). This may provide an alternative explanation for the
nature of the sudden gain phenomenon. Specifically, it is possible
that our (or any other) methodology to identify participants with
sudden gains simply selects individuals who are more severe at
pretreatment. Because of regression to the mean, those individuals
are expected to show greater gains before the end of treatment than
the less severe participants.
In sum, the results of this study revealed a number of similarities
to the depression literature with regards to the magnitude and
frequency of sudden gains. However, the sudden gains in this
study did not predict better short-term or long-term treatment
outcome, were not correlated with a particular treatment modality,
and were not associated with measurable changes in cognitions.
These findings question the external validity of the sudden gain
phenomenon. In fact, the most parsimonious explanation for the
phenomenon in the present study is that participants simply show
clinically nonsignificant fluctuations in their symptomatology dur-
ing therapy.
Limitations of the study include the sole reliance on self-report
measures to assess cognitive changes and social anxiety. Further-
more, we only examined cognitions pertaining to negative self-
perception. Although there is some justification in the social pho-
bia literature to primarily concentrate on self-perception, other
cognitions, especially those related to estimated social cost (Foa,
Franklin, Perry, & Herbert, 1996; Hofmann, 2004; Wilson &
Rapee, 2005) and emotional control (Hofmann, 2005), may prove
to be more important than changes in self-perception. Finally, the
study was not a randomized, controlled trial and participants were
primarily Caucasians with relatively high socioeconomic status,
which limits the generalizability of the study findings. Finally, one
of the most significant problems in comparing our study with
previous work in sudden gains involves the cutoff score used to
define a sudden gain (Criterion 1 of the sudden gain criteria). This
is an arbitrary score that is dependent on the measuring construct,
the measuring instrument, and the sample characteristics. How-
ever, it is not unusual in the clinical literature to choose different
cutoff scores of a construct for different patient groups. For ex-
ample, the expressed emotions literature accepts different cutoffs
for high and low expressed emotions in individuals with depres-
sion and schizophrenia.
Despite these limitations, this study points to an important area
of future inquiry: the empirical analysis of the psychotherapy
process. Consistent with the depression literature, sudden gains in
individuals who underwent group therapy for social phobia were
identified. Although CBGT was associated with more cognitive
changes and lower levels of social anxiety at follow-up than EGT,
the session-by-session data provided little support for the notion
that cognitive changes produce sudden gains. Therefore, the cog-
nitive model is either inadequate to explain sudden gains during
treatment for social phobia and/or sudden gains are not associated
with cognitive changes. Alternatively, it is possible that changes in
cognitions do, in fact, precede sudden gains. However, this effect
might be delayed, and cognitions immediately preceding the sud-
den gain only poorly predict those gains. In fact, Tang and DeRu-
beis (1999b) suggested that the pregain session is the critical
session in which the therapeutic breakthrough occurs. Further-
more, it is possible that changes in cognitions did precede the
sudden gains in the present study but that the instruments to assess
these changes were inadequate. In either case, the validity of the
sudden gain phenomenon remains questionable unless it can be
demonstrated that this rare occurrence has any clear clinical
significance.
695
SUDDEN GAINS IN SOCIAL PHOBIA

Page 10

References
American Psychiatric Association. (1994). Diagnostic and statistical man-
ual of mental disorders (4th ed.). Washington, DC: Author.
Baker, S. L., Heinrichs, N., Kim, H.-J., & Hofmann, S. G. (2002). The
Liebowitz Social Anxiety Scale as a self-report instrument: A prelimi-
nary psychometric analysis. Behaviour Research and Therapy, 40, 701–
715.
Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961).
An inventory for measuring depression. Archives of General Psychiatry,
4, 561–571.
Beidel, D. C., Borden, J. W., Turner, S. M., & Jacob, R. G. (1989). The
Social Phobia and Anxiety Inventory: Concurrent validity with a clinic
sample. Behaviour Research and Therapy, 27, 573–576.
Brown, T. A., DiNardo, P. A., Lehman, C. L., & Campbell, L. A. (2001).
Reliability of DSM–IV anxiety and mood disorders: Implications for the
classification of emotional disorders. Journal of Abnormal Psychology,
110, 49–58.
Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models:
Applications and data analysis methods. Newbury Park, CA: Sage.
Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. In
R. G. Heimberg, M. R. Liebowitz, D. A. Hope, & F. R. Schneier (Eds.),
Social phobia: Diagnosis, assessment and treatment (pp. 69–93). New
York: Guilford Press.
DiNardo, P. A., Brown, T. A., & Barlow, D. H. (1994). Anxiety Disorders
Interview Schedule for DSM–IV: Lifetime Version (ADIS-IV-L). Albany,
NY: Graywind Publications.
Feske, U., & Chambless, D. L. (1995). Cognitive behavioral versus expo-
sure only treatment for social phobia: A meta-analysis. Behavior Ther-
apy, 26, 695–720.
Foa, E. B., Franklin, M. E., Perry, K. J., & Herbert, J. D. (1996). Cognitive
biases in generalized social phobia. Journal of Abnormal Psychology,
105, 433–439.
Gaynor, S. T., Weersing, V. R., Kolko, D. J., Birmaher, B., Heo, J., &
Brent, D. A. (2003). The prevalence and impact of large sudden im-
provements during adolescent therapy for depression: A comparison
across cognitive–behavioral, family, and supportive therapy. Journal of
Consulting and Clinical Psychology, 71, 386–393.
Gould, R. A., Buckminster, S., Pollack, M. H., Otto, M. W., & Yap, L.
(1997). Cognitive–behavioral and pharmacological treatment for social
phobia: A meta-analysis. Clinical Psychology: Science and Practice, 4,
291–306.
Hardy, G. E., Cahill, J., Stiles, W. B., Ispan, C., Macaskill, N., & Barkham,
M. (2005). Sudden gains in cognitive therapy for depression: A repli-
cation and extension. Journal of Consulting and Clinical Psychology,
73, 59–67.
Heimberg, R. G. (1991). Cognitive behavioral treatment of social phobia
in a group setting: A treatment manual. Unpublished treatment manual,
Center for Stress and Anxiety Disorders, University at Albany, State
University of New York.
Hofmann, S. G. (1999). Behavior group therapy for social phobia. Un-
published treatment manual, Boston University.
Hofmann, S. G. (2000). Self-focused attention before and after treatment of
social phobia. Behaviour Research and Therapy, 38, 717–725.
Hofmann, S. G. (2004). Cognitive mediation of treatment change in social
phobia. Journal of Consulting and Clinical Psychology, 72, 392–399.
Hofmann, S. G. (2005). Perception of control over anxiety mediates the
relation between catastrophic thinking and social anxiety in social pho-
bia. Behaviour Research and Therapy, 43, 885–895.
Hofmann, S. G., & Barlow, D. H. (2002). Social phobia (social anxiety
disorder). In D. H. Barlow (Ed.), Anxiety and its disorders: The nature
and treatment of anxiety and panic (2nd ed., pp. 454–476). New York:
Guilford Press.
Hofmann, S. G., & DiBartolo, P. M. (2000). An instrument to assess
self-statements during public speaking: Scale development and prelim-
inary psychometric properties. Behavior Therapy, 31, 499–515.
Hofmann, S. G., Moscovitch, D. A., Kim, H.-J., & Taylor, A. N. (2004).
Changes in self-perception during treatment of social phobia. Journal of
Consulting and Clinical Psychology, 72, 588–596.
Hofmann, S. G., & Suvak, M. (in press). Treatment attrition during group
therapy for social phobia. Journal of Anxiety Disorders.
Hollon, S. D. (1999). Rapid early response in cognitive behavior therapy:
A commentary. Clinical Psychology: Science and Practice, 6, 305–309.
Ilardi, S. S., & Craighead, W. E. (1994). The role of nonspecific factors in
cognitive–behavior therapy for depression. Clinical Psychology: Sci-
ence and Practice, 1, 138–156.
Ilardi, S. S., & Craighead, W. E. (1999). Rapid early response, cognitive
modification, and nonspecific factors in cognitive behavior therapy for
depression: A reply to Tang and DeRubeis. Clinical Psychology: Science
and Practice, 6, 295–299.
Jacobson, N. S., & Truax, P. (1991). Clinical significance: A statistical
approach to defining meaningful change in psychotherapy research.
Journal of Consulting and Clinical Psychology, 59, 12–19.
Kelly, M. A. R., Roberts, J. E., & Ciesla, J. A. (2005). Sudden gains in
cognitive behavioral treatment for depression: When do they occur and
do they matter? Behaviour Research and Therapy, 43, 703–714.
Kessler, R. C., McGonagle, K. A., Shanyang, Z., Nelson, C. B., Hughes,
M., Eshleman, S., et al. (1994). Lifetime and 12-month prevalence of
DSM–III–R psychiatric disorders in the United States. Archives of Gen-
eral Psychiatry, 51, 8–19.
Lambert, M. J. (2005). Early response in psychotherapy: Further evidence
for the importance of common factors rather than “placebo effects.”
Journal of Clinical Psychology, 61, 855–869.
Leary, M. R., & Kowalski, R. M. (1995). Social anxiety. New York:
Guilford Press.
Liebowitz, M. R. (1987). Social phobia. Modern Problems of Pharma-
copsychiatry, 22, 141–173.
Newman, M. G., Hofmann, S. G., Trabert, W., Roth, W. T., & Taylor,
C. B. (1994). Does behavioral treatment of social phobia lead to cogni-
tive changes? Behavior Therapy, 25, 503–517.
Osman, A., Barrios, F. X., Haupt, D., King, K., Osman, J. R., & Slavens,
S. (1996). The Social Phobia and Anxiety Inventory: Further validation
in two nonclinical samples. Journal of Psychopathology and Behav-
ioural Assessment, 18, 35–47.
Rapee, R. N., & Heimberg, R. G. (1997). A cognitive–behavioral model of
anxiety in social phobia. Behaviour Research and Therapy, 35, 741–756.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models:
Applications and data analysis methods. Thousand Oaks, CA: Sage.
Raudenbush, S. W., Bryk, A. S., Cheong, Y., & Congdon, R. T. (2001).
HLM 5: Hierarchical linear and nonlinear modeling. Chicago: Scien-
tific Software International.
Ries, B. J., McNeil, D. W., Boone, M. L., Turk, C. L., Carter, L. E., &
Heimberg, R. G. (1998). Assessment of contemporary social phobia
verbal report instrument. Behaviour Research and Therapy, 36, 983–
994.
Schwartz, R. M., & Garamoni, G. L. (1989). Cognitive balance and
psychopathology: Evaluation of an information processing model of
positive and negative states of mind. Clinical Psychology Review, 9,
271–294.
Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in
assessing rater reliability. Psychological Bulletin, 86, 420–428.
Stiles, W. B., Leach, C., Barkham, M., Lucock, M., Iveson, S., Shapiro,
D. A., et al. (2003). Early sudden gains in psychotherapy under routine
clinic conditions: Practice-based evidence. Journal of Consulting and
Clinical Psychology, 71, 14–21.
Suh, C. S., Strupp, H. H., & O’Malley, S. S. (1986). The Vanderbilt
process measure: The Psychotherapy Process Scale (VPPS) and the
Negative Indicators Scale (VNIS). In L. S. Greenberg & W. M. Pinsof
696
HOFMANN, SCHULZ, MEURET, MOSCOVITCH, AND SUVAK

Page 11

(Eds.), The Psychotherapeutic Process: A Research Handbook (pp.
285–323). New York, NY: Guilford Publications.
Tang, T. Z., & DeRubeis, R. J. (1999a). Reconsidering rapid early response
in cognitive behavioral therapy for depression. Clinical Psychology:
Science and Practice, 6, 283–288.
Tang, T. Z., & DeRubeis, R. J. (1999b). Sudden gains and critical sessions
in cognitive–behavioral therapy for depression. Journal of Consulting
and Clinical Psychology, 67, 894–904.
Tang, T. Z., DeRubeis, R. J., Beberman, R., & Pham, T. (2005). Cognitive
changes, critical sessions, and sudden gains in cognitive–behavioral
therapy for depression. Journal of Consulting and Clinical Psychology,
73, 168–172.
Tang, T. Z., Luborsky, L., & Andrusyna, T. (2002). Sudden gains in
recovering from depression: Are they also found in psychotherapies
other than cognitive–behavioral therapy? Journal of Consulting and
Clinical Psychology, 70, 444–447.
Taylor, S. (1996). Meta-analysis of cognitive–behavioral treatments for
social phobia. Journal of Behavior Therapy and Experimental Psychi-
atry, 27, 1–9.
Turner, S. M., Beidel, D. C., Dancu, C. V., & Stanley, M. A. (1989). An
empirically derived inventory to measure social fears and anxiety: The
Social Phobia and Anxiety Inventory. Psychological Assessment, 1,
35–40.
Vittengl, J. R., Clark, L. A., & Jarrett, R. B. (2005). Validity of sudden
gains in acute phase treatment of depression. Journal of Consulting and
Clinical Psychology, 73, 173–182.
Wilson, J. K., & Rapee, R. M. (2005). The interpretation of negative social
events in social phobia: Changes during treatment and relationship to
outcome. Behaviour Research and Therapy, 43, 373–389.
Woody, S. R., Chambless, D. L., & Glass, C. R. (1997). Self-focused
attention in the treatment of social phobia. Behaviour Research and
Therapy, 35, 117–129.
Woody, S. R., & Rodriguez, B. F. (2000). Self-focused attention and social
anxiety in social phobics and normal controls. Cognitive Therapy and
Research, 24, 437–488.
Received September 12, 2005
Revision received March 6, 2006
Accepted April 3, 2006 ?
Call for Papers: Special Section on Suicide and Self-Harm Behaviors
The Journal of Consulting and Clinical Psychology is requesting submissions of empirical papers
that focus on suicide and self-harm behaviors, including non-suicidal self-injury. In particular,
submissions are requested that may address one of the following topics; (1) beyond the identifica-
tion of broad biopsychosocial risk factors, what are possible specific mechanisms that promote
self-harm behavior, and might be addressed in prevention/intervention efforts? (2) how might
cross-disciplinary theoretical perspectives (e.g., biological, interpersonal) be integrated to under-
stand or treat self-harm behavior? (3) what are some innovative methodological paradigms for
investigating self-harm behaviors? (4) randomized clinical trial data on preventions/ interventions
designed to reduce self-harm behavior. The papers must present original empirical findings. The
goal of this special section is to have a set of papers that represent the lifespan.
The deadline for submissions of manuscripts is February 1, 2007. Final editorial decisions will
be made by late 2007, with an anticipated publication date of early 2008. All submissions should
be entered through the main submission portal for the journal (www.apa.org/journals/ccp.html).
Authors should indicate in their accompanying cover letter that the paper is to be considered for the
special section on “suicide and self-harm.” All submitted papers must be in APA format and
conform to the all the guidelines for submission for JCCP (see www.apa.org/journals/ccp).
Questions or inquiries regarding the special section should be directed to the section editor, Mitch
Prinstein (mitch.prinstein@unc.edu).
697
SUDDEN GAINS IN SOCIAL PHOBIA