Journal of Consulting and Clinical Psychology Copyright 2002 by the American Psychological Association, Inc.

2002, Vol. 70, No. 5, 1067–1074 0022-006X/02/$5.00 DOI: 10.1037//0022-006X.70.5.1067

Skills Training in Affective and Interpersonal Regulation

Followed by Exposure: A Phase-Based Treatment for PTSD
Related to Childhood Abuse

Marylene Cloitre Karestan C. Koenen

New York Presbyterian Hospital—Weill Medical College Columbia University
of Cornell University

Lisa R. Cohen Hyemee Han

St. Luke’s–Roosevelt Hospital Weill Medical College of Cornell University

Fifty-eight women with posttraumatic stress disorder (PTSD) related to childhood abuse were randomly
assigned to a 2-phase cognitive– behavioral treatment or a minimal attention wait list. Phase 1 of
treatment included 8 weekly sessions of skills training in affect and interpersonal regulation; Phase 2
included 8 sessions of modified prolonged exposure. Compared with those on wait list, participants in
active treatment showed significant improvement in affect regulation problems, interpersonal skills
deficits, and PTSD symptoms. Gains were maintained at 3- and 9-month follow-up. Phase 1 therapeutic
alliance and negative mood regulation skills predicted Phase 2 exposure success in reducing PTSD,
suggesting the value of establishing a strong therapeutic relationship and emotion regulation skills before
exposure work among chronic PTSD populations.

Posttraumatic stress disorder (PTSD) has an estimated lifetime ment considerations. The Diagnostic and Statistical Manual of
prevalence of between 5% and 10% in the general population, with Mental Disorders (4th ed.; DSM–IV; American Psychiatric Asso-
women being affected twice as often as men (Kessler, Sonnega, ciation, 1994) PTSD field trials (Roth, Newman, Pelcovitz, van der
Bromet, Hughes, & Nelson, 1995). Among women, the most Kolk, & Mandel, 1997) reported the lifetime prevalence of CA-
common traumas occur during the developmental years and con- related PTSD at 67%, making PTSD the leading Axis I disorder in
sist mainly of childhood sexual abuse and physical abuse. Twice as this population. Problems in emotion regulation and interpersonal
many women have experienced childhood abuse compared with functioning were identified as two additional symptom sets occur-
adult rape (e.g., Finkelhor, Hotaling, Lewis, & Smith, 1990), yet ring with equal if not greater frequency than the PTSD symptom
the development of empirically supported treatments for adult constellation. This study presents a randomized, controlled trial of
survivors of child abuse (CA) has lagged far behind that of adult a treatment developed to specifically address the three core prob-
rape. To date, there are two well-tested treatments for PTSD lems of the CA population: PTSD symptoms, emotion regulation
related to rape (Foa, Rothbaum, Riggs, & Murdoch, 1991; Resick problems, and interpersonal difficulties.
& Schnicke, 1992), but none for childhood abuse. One reason for The affect regulation and interpersonal disturbances of women
this is that the psychological sequelae of CA include symptoms with CA have been well documented. It has been argued that these
that extend beyond the PTSD diagnosis, leading to complex treat- problems are a relatively distinct feature of childhood trauma and
derive from the trauma’s disruptive impact on the achievement of
the developmental goals of affect regulation and interpersonal
Marylene Cloitre, Anxiety and Traumatic Stress Program, Payne Whit-
relatedness (van der Kolk, 1996). The most compelling support for
ney Clinic, New York Presbyterian Hospital—Weill Medical College of this view is provided by studies that have directly compared
Cornell University; Karestan C. Koenen, Department of Public Health, individuals with childhood onset trauma with individuals with
Columbia University; Lisa R. Cohen, Department of Psychiatry, St. adult onset trauma (e.g., rape victims, disaster victims) and found
Luke’s–Roosevelt Hospital, New York; Hyemee Han, Department of Psy- that CA survivors are consistently more troubled, particularly in
chiatry, Weill Medical College of Cornell University. the domains of affect modulation, anger management, and inter-
Marylene Cloitre is now at the Child Study Center, New York Univer- personal relationships (Cloitre, Scarvalone, & Difede, 1997; van
sity School of Medicine; Karestan C. Koenen is now at the National Center der Kolk, Roth, & Pelcovitz, 1993; Zlotnick et al., 1996).
for PTSD, Boston. Affect dysregulation is broadly defined as the tendency to have
This study was supported by Grant MH57883 from the National Insti-
low-threshold, high-intensity emotional reactions followed by
tute of Mental Health to Marylene Cloitre. We thank Laurie Alexander,
Karen Heffernan, Amy Kossoy, Kate Marcelli, and Carol Srinivasan for
slow return to baseline. It has recently been investigated in the
their assistance with the study. childhood abuse population (e.g., Rorty, 1996; Zlotnick, 1999),
Correspondence concerning this article should be addressed to Marylene most definitively in the DSM–IV PTSD field trials (van der Kolk
Cloitre, 418 East 59th Street, Apartment 25B, New York, New York et al., 1993). Over 70% of respondents endorsed problems in
10022. E-mail: mcloitre@med.cornell.edu getting upset easily, having trouble calming down, and letting go

1067
1068 CLOITRE, KOENEN, COHEN, AND HAN

of upsetting things. Other types of emotion management problems severity at a 3-month follow-up, whereas a supportive-counseling
were indicated, with the two most prominent being (a) fear of and a symptom-focused cognitive– behavioral treatment produced
experiencing anger and difficulty appropriately expressing it and no further changes (Foa et al., 1991).
(b) transient experiences of dissociation. Rather than reject the use of exposure and its potential long-term
In addition to producing acute distress, affect dysregulation benefits, it was proposed that women with CA PTSD could benefit
plays a significant role in the interpersonal difficulties of women from exposure if they were provided with and learned skills to
with CA. Typical interpersonal difficulties arise in emotion-laden reduce trauma-related characteristics associated with poor out-
situations that involve the management of conflict and the effec- come. Given all of the above considerations, the treatment devel-
tive negotiation of the power dynamics of relationships. According oped for CA-PTSD women was conceptualized as a sequentially
to the DSM–IV field trials, 91% of CA victims with PTSD en- based treatment organized into two phases. The first phase of
dorsed problems with sensitivity to criticism, inability to hear other treatment focused exclusively on skills training in affect and
viewpoints, difficulty in standing up for themselves, and a ten- interpersonal regulation (STAIR). The goals of the first phase of
dency to quit jobs and relationships without negotiation. Func- treatment were twofold: to directly and vigorously address prob-
tional impairments in the interpersonal domain are pervasive, lems in affect and interpersonal regulation as they negatively
extending across many life roles. Women with CA have reported impacted on day-to-day functioning and to prepare the client for
less satisfaction with dating and marriage partners, difficulties with the effective and successful use of the exposure treatment. The
parenting activities, problems in functioning at work, greater social second phase of treatment introduced and implemented the emo-
isolation, and poorer social adjustment than women without a tional processing of the trauma using a modified version of pro-
history of CA (Briere, 1988). The frequency and extent of affective longed exposure (PE) to resolve PTSD symptoms. We hypothe-
and interpersonal difficulties as well as their impact on functional sized that the implementation of skills training before exposure
capacity strongly indicate the need for interventions specific to would facilitate effective use of exposure by providing time to
these problem domains. establish a therapeutic alliance and by the development of affect
In addition to the adverse impact these types of problems have regulation skills. We also hypothesized that the treatment as a
on day-to-day life, they raise concerns about the use of exposure- whole would provide significant improvement in PTSD symp-
based treatments for CA survivors. Although the emotional pro- toms, emotion regulation problems, and interpersonal skills
cessing of traumatic material is largely agreed to be a critical deficits.
ingredient to the resolution of PTSD symptoms, there is substan-
tial, primarily clinical, literature indicating that exposure interven- Method
tions can be problematic. Symptom exacerbation, high drop-out
rates, and compliance problems have been associated with this Design
emotionally intensive form of treatment (Burnstein, 1986;
McDonough-Coyle et al., 2000; Pitman et al., 1991; Scott & This was a randomized clinical trial in which all potential participants
Stradling, 1997; Tarrier et al., 1999). were self-referred by means of advertisements in the community or word-
of-mouth. Following a brief phone screen, those found eligible for the
Unfortunately, the patient characteristics associated with poorer
study underwent the full assessment procedure. The assessment was com-
outcome in exposure therapy are typical of CA survivors. Specif- pleted in two visits. Visit 1 comprised a description of the study; signing
ically, patients who do not fare well in exposure-based treatments of informed consent; self-report questionnaires; and clinical interviews
show (a) difficulty tolerating distress and managing feelings such concerning trauma history, medical history, and health status. Visit 2
as anger and anxiety, (b) vulnerability to dissociation under stress, comprised clinician-administered diagnostic interviews to assess for Axis I
and (c) difficulty maintaining a good working relationship with a disorders and Axis II borderline personality disorder.
therapist (Chemtob, Novaco, Hamada, Gross, & Smith, 1997; Inclusion criteria required the presence of DSM–IV-defined diagnosis of
Cloitre & Koenen, 2001; Jaycox & Foa, 1996). Trauma survivors PTSD related to childhood sexual abuse, physical abuse, or both (DSM–
have been noted by many clinicians to have difficulty in tolerating IV). Sexual abuse was defined as at least one episode of sexual contact
the interpersonal nature of therapy, particularly “the [need] to trust (fondling, attempted or completed vaginal, oral, or anal intercourse) initi-
ated by a caregiver or individual in a position of authority to the participant
another person with his or her pain” (Turner, McFarlane & van der
when she was under the age of 18. The perpetrator must have been at
Kolk, 1996, p. 538). This difficulty would seem further exacer- least 5 years older than the participant, unless the participant experienced
bated in exposure treatment, which requires significant and sus- the sexual contact with this person as against her will. Childhood physical
tained verbal disclosure of deeply distressing events. abuse was defined as an action by a parent or other adult in charge of the
There is evidence, however, that when a client has successfully participant when she was under the age of 18 in which the adult purpose-
engaged in exposure therapy, the long-term benefits are superior to fully hit, pushed, punched, or cut the participant leaving bruises, scratches,
those found in other treatments. A recent study in which CA broken bones or teeth, or making her bleed. In addition, study participants
survivors with PTSD were randomized to three treatment condi- were required to always have had at least one clear memory of the abuse.
tions found that whereas an exposure-based treatment had higher Participants were required to be between 18 and 65 years of age and to plan
drop-out rates (41%) than either a present-centered treatment (9%) on residing in the area for the duration of the treatment. Exclusion criteria
included current diagnosis of organic or psychotic mental disorders, sub-
or a wait list (13%), treatment completers showed better mainte-
stance dependence, eating disorder, dissociative disorder, Bipolar I disor-
nance in PTSD symptom reduction compared with the present- der or borderline personality disorder, and the presence of suicide attempt
centered treatment at 3- and 6-month follow-up (McDonough- or psychiatric hospitalization within the last 3 months.
Coyle et al., 2000). This is consistent with results from a study of Eligible participants were randomized into one of two conditions: a free
women with rape-related PTSD that found that exposure therapy 12-week, 16-session active treatment (STAIR–modified PE) or a 12-week
tended to produce continuing improvements in PTSD symptom minimal attention wait list. Active treatment participants and their thera-
 TREATMENT FOR CHILD ABUSE-RELATED PTSD 1069

pists completed a brief measure of therapeutic alliance following each of study. All treatment sessions were audiotaped, and sessions were moni-
the 16 sessions. All study participants completed the clinician-administered tored for adherence. Treatment consisted of 16 sessions delivered over a
and self-report measures at pre- and posttreatment and a subset of self- 12-week period and was organized into two phases. The first, Skills
report measures (PTSD and affect-regulation measures) at midtreatment. Training in Affect and Interpersonal Regulation, consisted of eight weekly
Those in the STAIR–modified PE condition received 3-month and 9-month 1-hr sessions. The second phase, modified Prolonged Exposure, consisted
posttreatment follow-ups. Those completing the wait-list condition were of twice weekly 1.5-hr sessions. Following is a brief description of the
offered a free 12-week treatment of either STAIR–modified PE or any treatment. For more detailed information please see the STAIR–modified
other clinically relevant treatment in our clinic of comparable duration and PE manual available on request from Marylene Cloitre.
number of sessions. STAIR. STAIR is a cognitive– behavioral treatment that targets the
development of emotion management and interpersonal skills. The inter-
ventions were derived from generic cognitive– behavioral and dialectical
Sample behavior therapy (Linehan, 1993) strategies that were adapted to the needs
of the CA trauma population. Each session focuses on a particular skills
During an 18-month period, 207 women were scheduled for initial
deficit understood within the context of the experience of CA trauma and
evaluations. Approximately 50% did not show or cancelled without re-
its typical consequences. Session-by-session topics are as follows: (1)
scheduling their assessments. Of the 103 who completed the evaluation, 58
labeling and identifying feelings, (2) emotion management (particularly
were found eligible for and entered into treatment (31 in the active
anger and anxiety), (3) distress tolerance, (4) acceptance of feelings and
treatment condition and 27 in the minimal attention wait-list condition).
enhanced experiencing of positive emotions, (5) identification of trauma-
Reasons for not being eligible for the study included not meeting criteria
based interpersonal schemas and their enactment in day-to-day life, (6)
for full PTSD (43%), current substance-dependence disorder (18%), and
identification of conflict between trauma-generated feelings and current
borderline personality disorder (14%). Other reasons represented less than
interpersonal goals, (7) role plays related to issues of power and control,
5% of the rule outs (e.g., recent hospitalization, thought disorder). Of
and (8) role plays related to developing flexibility in interpersonal situa-
the 58 women who entered treatment, 12 dropped out: 9 from the active
tions involving power differentials. The role plays highlighted the presence
treatment (29%) and 3 from the wait list (11%). There were no sociode-
and expression of emotion. This included role plays of clients’ typical
mographic, clinical, or symptom differences between completers and
problematic interpersonal behaviors and new, alternative behaviors. All
dropouts.
STAIR sessions had the same format and structure. They began with
Sociodemographic characteristics. The average age of the women
psychoeducation about the rationale and goals of the interventions, fol-
was 34 years (SD ⫽ 7.22). Ethnicity breakdown for the sample revealed
lowed by skills acquisition, and skills application and practice. Between-
that 46% were Caucasian, 20% were African American, 15% were His-
sessions work was assigned and consisted of application of the skills to
panic, and 19% were other ethnicities including Asian, Caribbean, and
current life difficulties.
American Indian. A majority of the sample was either single (42%) or
Modified PE. Phase 2 used the prolonged imaginal exposure technique
separated or divorced (24%); the remainder were either married or living
described by Foa and Rothbaum (1998) in which clients repeatedly de-
with a significant other (34%). Fifty-two percent had completed college or
scribe their traumatic events in a detailed and emotionally engaged fashion.
more, 37% had some college, and 11% had a high school education or less.
PE, developed for rape victims, was modified for this population in several
Forty-one percent of the sample worked full time; 35% were either part-
ways. The in vivo exposure to rape-related cues was eliminated. Three
time workers, students, or both; and 24% were either homemakers, unem-
other components were added. First, a postexposure stabilization check
ployed, or disabled. Twenty-six percent had an annual personal income of
was included that guided the participant in using coping skills to modulate
$30,000 or more, 43% had an annual income of $15,000 to $30,000, and
her feeling states to ensure postexposure emotional stability and orientation
31% had $15,000 or less.
to the present. Second, a postexposure emotion-focused processing inter-
Abuse characteristics. Forty-eight percent (48%) of the sample had
vention was included in which the participant identified the presence and
experienced both sexual and physical abuse, 39% had experienced sexual
intensity of fear, anxiety, dissociation, and sadness during the exposure.
abuse only, and 13% had experienced physical abuse only. There were no
Last, the participant was asked to identify negative interpersonal schemas
sociodemographic or clinical differences across the women with different
embedded in the narrative. Therapist and participant contrasted these
types of childhood abuse.
abuse-related schemas with the more adaptive schemas generated during
Comorbidity. Forty-five percent (45%) of the participants had current
STAIR to highlight differences between the past and present circumstances
major depression, with a further 35% having past major depression.
and personal resources. The final component of the session entailed a
Seventy-nine percent (79%) were diagnosed with some type of anxiety
review of applying coping skills to current life problems and application of
disorder, with generalized anxiety disorder (48%) being the most common.
new interpersonal schemas to current relationships. Between-sessions work
Twenty-five percent (25%) met criteria for a past substance abuse disorder,
included listening to the taped narratives at least once a day.
and 16% had a past eating disorder. Almost half (48%) had a history of
Minimal attention wait list. Participants were informed that they could
suicide attempts or engaging in self-mutilating behavior (e.g., cutting or
receive treatment in 12 weeks. They were monitored through weekly
burning). Twenty-five percent (25%) had received a minimum of 10
15-min phone sessions with the clinical coordinator.
outpatient visits for psychotropic medications, psychotherapy, or both in
the past year, and 29% had used the psychiatry emergency room in the past
year.
Assessment

Treatment Trauma history was assessed at pretreatment using two clinician-

administered instruments, the Childhood Maltreatment Interview Schedule
Treatment sessions were conducted by five female doctoral-level clinical (Briere, 1992) and the Sexual Assault and Additional Interpersonal Vio-
psychologists, each of whom treated between 3 and 6 participants. Ther- lence Schedule (Resick & Schnicke, 1992). PTSD was diagnosed with the
apists were trained using manuals with treatment guidelines and received Clinician-Administered PTSD Scale (CAPS; Blake et al., 1995), which
weekly supervision from Marylene Cloitre. Therapists also attended a measures both the frequency and intensity of the 17 DSM–IV symptoms
full-day, intensive workshop on PE supervised by Edna B. Foa in the first (range ⫽ 0 –136). All other Axis I diagnoses were assessed with the
year of the study. A second full-day supervisory and review workshop was Structured Clinical Interview for the DSM–IV (SCID-I; Spitzer, Williams,
conducted with Edna B. Foa at the beginning of the second year of the Gibbon, & First, 1994), and borderline personality disorder (BPD) status
1070 CLOITRE, KOENEN, COHEN, AND HAN

was determined with the BPD section of the SCID-II (First, Spitzer, STAIR session, the therapist allowed more than 20 min of non-
Gibbon, Williams, & Benjamin, 1994). The CAPS and SCID-I were protocol talk to elapse without redirecting participant back to
implemented at pre-, post-, and follow-up assessments. Clinician raters session agenda.
were blind to treatment condition at pre- and posttreatment.
Self-report questionnaires were administered for each of the three prob-
lem domains. PTSD symptomatology was assessed with the Modified Immediate Effects of Treatment
Posttraumatic Stress Disorder Symptom Scale (MPSS-SR; Falsetti,
Resnick, Resick, & Kilpatrick, 1993), a 34-item measure that separately Comparison of group means at pre-, mid-, and posttreatment.
assesses the frequency and severity of each of the 17 symptoms of PTSD We conducted 2 (group: STAIR–modified PE vs. wait list) ⫻ 2
(range ⫽ 0 –119). Emotion related problems were assessed with six mea-
(time: pre- vs. posttreatment) multivariate analyses of variance
sures. Capacity to regulate any negative mood was assessed with the
General Expectancy for Negative Mood Regulation Scale (NMR; Cantan- (MANOVAs) or 2 (group) ⫻ 3 (pre- vs. mid- vs. posttreatment)
zaro & Mearns, 1990), a 30-item measure, with higher scores indicating MANOVAs, depending on the number of assessment points, for
better mood regulation (range ⫽ 30 –150). Problems with anger were each of the three conceptually grouped symptom domains: (a)
assessed with the Anger Expression subscale (Ax/Ex), a 24-item measure PTSD symptoms, (b) affect regulation difficulties, and (c) inter-
from the State–Trait Anger Expression Inventory (Speilberger, 1991), in personal problems for the completers sample. Group was a
which higher scores indicate more frequent experiences of anger (range ⫽ between-subjects variable (STAIR–modified PE and wait list),
0 –72). Ability to identify and label feeling states was measured with the whereas time was a within-subject variable. A MANOVA was
Toronto Alexythimia Scale—20-item version (TAS–20; Bagby, Parker & followed by analyses of variance (ANOVAs) for each of the
Taylor, 1993), with higher scores indicating greater difficulty (range ⫽
individual measures only when the Wilks’s Lambda for the inter-
20 –100). Dissociation, generally understood as a protective reaction
against painful affects associated with trauma, was assessed with the
action term (Group ⫻ Time) was significant. Analyses were con-
14-item Dissociation Scale (DISS; range ⫽ 0 –36) developed by Briere and ducted on study completers for active (n ⫽ 22) and wait-list (n ⫽
Runtz (1990). Emotions of depression and anxiety were assessed, respec- 24) conditions and also on the intent-to-treat sample (active con-
tively, with the 21-item Beck Depression Inventory (BDI; Beck, Ward, dition n ⫽ 31 vs. wait list n ⫽ 27) where end-point or last available
Mendelson, Mock & Erbaugh, 1981), with a score range of 0 – 63, and the ratings were carried forward to the next assessment point.
20-item State subscale of the State–Trait Anxiety Inventory (STAI–S; The Group ⫻ Time interaction effects in the MANOVAs were
Spielberger, 1983), with a score range of 0 – 80. significant for all three symptom domains: PTSD symptom mea-
Interpersonal and functional impairment were assessed with the 127- sures (Wilks’s ⌳ ⫽ 12.61, p ⬍ .01), the affect regulation measures
item Inventory of Interpersonal Problems (IIP; Horowitz, Rosenberg, Baer,
(Wilks’s ⌳ ⫽ 2.85, p ⬍ .01), and the interpersonal measures
Ureno, & Villasenor, 1988), in which the total score is an average of items
rated on a 5-point scale ranging from 0 to 4; the Social Adjustment
(Wilks’s ⌳ ⫽ 4.68, p ⬍ .01). Repeated measures ANOVAs were
Scale–Self Report (SAS-SR; Weissman & Bothell, 1976), which measures conducted for each dependent measure and the interaction effects
functioning in family, work and social functioning, where the total score is are presented in Table 1. All treatment measures show a significant
an average of items rated on a 1–5-point scale; and the 48-item Interper- decrease between pre- and posttreatment for STAIR–modified PE
sonal Support Evaluation List (ISEL; Cohen & Hoberman, 1983), which compared with the wait list. Simple comparisons indicated that for
measures social support (range ⫽ 0 – 40), where higher scores indicate the wait-list group there were no changes on any measure from
more social support. The therapeutic relationship was assessed using the pre- to midtreatment or from mid- to posttreatment. Simple com-
12-item Working Alliance Inventory (WAI; Tracey & Kokotovic, 1998), in parisons for the STAIR–modified PE group indicated there were
which the total score is an average of items rated on an 8-point scale
significant pre-to-midtreatment improvements in the NMR, Ax/
ranging from 0 to 7.
Ex, BDI, and STAI (all ps ⬍ .02) but not in the MPSS-SR, DISS,
or TAS–20. There were, however, significant mid-to-posttreatment
Results reductions in the MPSS-SR, DISS, and the TAS–20 (all ps ⬍ .01),
as well as further significant improvements in the BDI and STAI
Treatment Adherence
(all ps ⬍ .01). There were no further improvements in the NMR
Audiotapes of 44 therapy sessions (11% of 408 sessions) were and Ax/Ex. Intent-to-treat analyses revealed the same measures as
rated. Two STAIR–modified PE sessions from each client were having significant interaction terms and the same significant pair-
selected; one was randomly selected from the STAIR phase of the wise comparisons, with the additional finding of continued im-
treatment, the other randomly selected from the modified PE provement from mid-to-posttreatment for the NMR.
phase. The number of components within a session ranged from 7 Phase 1 predictors of Phase 2 PTSD symptom reduction. Par-
to 14, with an average of 8 (SD ⫽ 2.3). Raters were familiar with tial correlations between Phase 1 change scores in symptom mea-
the treatment program but had not treated any participants in the sures and Phase 2 PTSD symptom reduction were assessed, con-
study. They reviewed audiotapes and rated each component as trolling for PTSD symptoms at the beginning of Phase 2. In
present or absent. Interrater reliability was assessed by randomly addition, Phase 1 therapeutic alliance, as measured by the averaged
selecting four tapes (10% of rated tapes) and comparing ratings for WAI scores for Sessions 3, 4, and 5, was assessed as a predictor of
all components (n ⫽ 37). There was perfect agreement on the Phase 2 PTSD reduction, also controlling for PTSD symptoms at
ratings (␬ ⫽ 1.00). the beginning of Phase 2. Predictors of Phase 2 improvement were
Of a total of 333 components reviewed across all sessions therapeutic alliance (r ⫽ –.62, p ⬍ .03) and improvement in
sampled, 316 were rated as completed (95%). Of the 44 sessions negative mood regulation as measured by the NMR (r ⫽ –.47, p ⬍
reviewed, 29 (66%) had all components completed, 14 (32%) had .03). No other changes in Phase 1, including the significant reduc-
one component missing, and 1 (2%) had three components miss- tions in depression and anxiety, were associated with Phase 2
ing. Only one deviation from protocol was detected: In a final PTSD reduction.
 TREATMENT FOR CHILD ABUSE-RELATED PTSD 1071

Table 1
Analyses of Variance Means and Standard Deviations Pre- and Posttreatment
for STAIR-Modified PE Compared With Wait List

Treatment group

STAIR-modified
PE Wait list
(n ⫽ 22) (n ⫽ 24) Interaction

Measure and time M SD M SD F(1, 44) p

PTSD measures
MPSS-SR 5.39 ⬍.01
Pretreatment 69 16.6 73 18.6
Midtreatment 62 17.6 67 24.6
Posttreatment 29 27.6 58 28.6
CAPS 25.19 ⬍.01
Pretreatment 69 16.3 69 16.6
Midtreatment — — — —
Posttreatment 31 25.2 62 22.7
Affect regulation measures
NMR 8.82 ⬍.01
Pretreatment 85 15.6 84 17.9
Midtreatment 100 14.2 89 18.2
Posttreatment 110 19.5 85 18.6
Ax/Exa 4.07 .03
Pretreatment 32 8.9 35 7.1
Midtreatment 28 7.4 35 8.0
Posttreatment 24 9.4 36 7.1
DISSb 5.25 ⬍.01
Pretreatment 26 12.1 21 14.8
Midtreatment 20 10.8 19 14.2
Posttreatment 9 8.2 18 16.1
TAS-20c 5.60 ⬍.01
Pretreatment 56 10.8 55 12.9
Midtreatment 52 14.1 54 12.0
Posttreatment 43 13.3 53 14.0
BDI 5.89 ⬍.01
Pretreatment 25 10.6 23 9.0
Midtreatment 19 9.8 22 11.3
Posttreatment 8 7.8 20 11.4
STAI–S 11.98 ⬍.01
Pretreatment 57 9.6 53 15.6
Midtreatment 50 8.2 55 14.9
Posttreatment 36 8.6 55 14.9
Interpersonal and functional
impairment
IIP 13.73 .01
Pretreatment 1.88 0.57 1.70 0.46
Midtreatment — — — —
Posttreatment 1.06 0.46 1.60 0.66
SAS-SR 6.11 .02
Pretreatment 2.44 0.29 2.57 0.42
Midtreatment — — — —
Posttreatment 2.06 0.40 2.47 0.53
ISEL 9.70 .01
Pretreatment 24 8.1 23 8.8
Midtreatment — — — —
Posttreatment 30 7.6 23 9.5

Note. STAIR-modified PE ⫽ skills training in affect and interpersonal regulation and prolonged exposure
group; PTSD ⫽ posttraumatic stress disorder; MPSS-SR ⫽ Modified PTSD Symptom Scale–Self Report;
CAPS ⫽ Clinician-Administered PTSD Scale; NMR ⫽ General Expectancy for Negative Mood Regulation
Scale; BDI ⫽ Beck Depression Inventory; STAI–S ⫽ State subscale of the State–Trait Anxiety Inventory; IIP ⫽
Inventory of Interpersonal Problems; SAS-SR ⫽ Social Adjustment Scale–Self Report; ISEL ⫽ Interpersonal
Support Evaluation List.
a
Ax/Ex ⫽ Anger Expression subscale of the State–Trait Anger Expression Inventory. b DISS ⫽ Dissociation
scale. c TAS-20 ⫽ Toronto Alexithymia Scale–20-item version.
1072 CLOITRE, KOENEN, COHEN, AND HAN

Effect size. We calculated Cohen’s d (1992) statistics to com- differences were observed between post- and 9-month assessment
pare STAIR–modified PE with the wait list at posttreatment. The measures.
effect size for the PTSD symptoms as measured by the CAPS
was 1.30 and by the MPSS-SR, 1.03. The effect sizes for affect Discussion
regulation and related problems were as follows: NMR ⫽ 1.32,
Ax/Ex ⫽ 1.46, DISS ⫽ 0.73, TAS–20 ⫽ 0.70, BDI ⫽ 1.24, and The primary goal of this study was to test the efficacy of a new
STAI–S ⫽ 1.60. The effect sizes for interpersonal functioning treatment, STAIR–modified PE, as compared with a wait-list
were as follows: IIP ⫽ .96, SAS-SR ⫽ .87, and ISEL ⫽ .82. control for women with PTSD related to childhood abuse. Relative
According to Cohen (1992), effect sizes for differences between to the women on wait list, those who received STAIR–modified
two independent means (i.e., clinically meaningful, usually ob- PE showed significant improvement in three specifically targeted
servable differences between the groups) are considered small for problem domains: affect regulation problems, interpersonal skills
values of .20 or less, medium for .50 or more, and large for .80 or deficits and PTSD symptoms. These gains were maintained and
more. some were enhanced at 3- and 9-month follow-up. In addition, we
End-state functioning. Good end-state functioning, following hypothesized that the inclusion of a skills-training phase prior to
Foa et al. (1999), was defined as being an MPSS-SR score ⬍ 20, conducting exposure would facilitate effective use of the exposure.
an STAI–S score ⬍ 40, and a BDI score ⬍ 10. These cutoffs are Development of a positive therapeutic alliance during Phase 1 and
identical to or close to mean scores found in normative female improvement in negative mood regulation were significant predic-
samples. Using this criterion, we found that 46% of the partici- tors of PTSD reduction during Phase 2 exposure.
pants in the STAIR–modified PE condition achieved good end- The changes observed in various symptom domains across the
state functioning in comparison with only 4% of wait-list partici- two treatment phases were consistent with the treatment rationale.
pants, ␹2(1, N ⫽ 46) ⫽ 10.74, p ⬍ .01. The diagnostic status of Simple comparisons for the STAIR–modified PE group indicated
participants after treatment was computed as another measure of that Phase 1 produced significant reductions in negative mood
good end-state functioning. In the STAIR–modified PE condition, regulation and anger expression but not in PTSD symptoms,
23% of participants retained their PTSD status as assessed by the whereas Phase 2 exposure work produced significant reductions in
CAPS compared with 75% of the wait-list condition participants, PTSD symptoms but not in negative mood regulation or anger
␹2(1, N ⫽ 46) ⫽ 12.28, p ⬍ .01. expression. Although there was obviously some symptom change
Symptom worsening. Following Tarrier et al. (1999), we iden- across both treatment phases for all study measures, these results
tified patients who showed symptom worsening posttreatment as suggest the relative specificity of the symptom reduction associ-
measured by an increase in CAPS total severity score compared ated with each treatment phase.
with baseline. One participant (4.5%) in STAIR–modified PE Many aspects of the findings point to the value of implementing
and 6 (25%) in the wait-list condition experienced symptom a phase-based approach to the treatment of chronic PTSD. First,
worsening. the organization of the treatment provided the opportunity for
clients to develop emotion regulation skills without being bur-
dened by the demands of exposure work. Studies of adult-onset
Follow-Up Analyses traumas that have implemented cognitive– behavioral and expo-
sure treatment simultaneously (Foa et al., 1999; Marks, Lovell,
Long-term effects were assessed for the active treatment Noshirvani, Livanou, & Thrasher, 1998) have found that improve-
(STAIR–modified PE). We completed two sets of pairwise t tests. ments in PTSD and trauma-related symptoms were not greater
The first analyses assessed all measures at posttreatment versus than in treatments that focused on only one or the other type of
3-month follow-up; data for 20 of 22 completers were obtained. intervention. Foa et al. (1999) suggested that the simultaneous
The second analyses assessed posttreatment versus 9-month fol- presentation of the interventions might have produced “informa-
low-up; data for 17 of 22 completers were obtained. tion overload” so that the participants learned none of the inter-
CAPS total scores were significantly lower at 3 months ventions particularly well. The sequential organization of STAIR–
(M ⫽ 26, SD ⫽ 17.4) compared with immediately posttreatment, modified PE allows for skills consolidation and the effective
t(29) ⫽ 2.23, p ⫽ .04, indicating continuing improvement in the application of these skills in day-to-day life.
STAIR–modified PE completers. There were no other differences Second, the sequential approach allowed for a “preparatory”
between post- and 3-month follow-up assessment measures, indi- phase of treatment in which both therapist and participant could
cating that all other STAIR–modified PE treatment gains were assess the strengths and weakness of the participant in engaging in
maintained. At 9 months, the CAPS total score (M ⫽ 22, exposure work. The value of skills training as a preparatory phase
SD ⫽ 14.5) was lower than that observed at 3 months and, as to emotional processing work is reflected in the data, demonstrat-
expected, was significantly lower than the posttreatment score, ing that the development of negative mood regulation skills sig-
t(16) ⫽ 2.82, p ⫽ .01, indicating that the additional posttreatment nificantly contributed to the successful outcome of the exposure
improvement observed at 3 months was maintained at 9 months. work.
Significant improvements were observed in all of the interpersonal Third, the phase-based treatment allowed the opportunity for the
and functional measures at 9 months: the IIP (M ⫽ .84, SD ⫽ .54), development of a good therapeutic relationship. Many trauma
t(15) ⫽ 2.40, p ⫽ .03; the SAS-SR (M ⫽ 1.83, SD ⫽ .48), theorists and researchers have suggested that the therapeutic rela-
t(16) ⫽ 2.21, p ⫽ .04; and the ISEL (M ⫽ 34, SD ⫽ 8.1), t(16) ⫽ tionship is a critical component to successful trauma work. This is
3.29, p ⫽ .01, indicating additional posttreatment improvement the first study to provide empirical evidence of the contribution of
that had not been seen at the 3-month assessment. No other the therapy relationship to the efficacy of a trauma-focused treat-
 TREATMENT FOR CHILD ABUSE-RELATED PTSD 1073

ment or, more specifically, to its role in effective exposure work. schemas reflected in a range of day-to-day situations and inter-
Future research is required to determine the extent to which the vened via role plays on those most relevant to current life. Finally,
skills-training component, as compared with other preexposure to minimize anticipatory anxiety and potential dropout, we used a
interventions, facilitates the development and impact of the ther- more intensive form of exposure in which the participant received
apeutic alliance on the exposure work. two sessions of exposure per week for 4 weeks rather than one
STAIR–modified PE provided skills training in interpersonal session a week for 8 weeks.
functioning during the first phase of treatment and continued Exposure studies reporting problematic outcomes among those
practice of these skills throughout the second phase. Consistent with chronic PTSD have led to the suggestion that exposure may
with this effort, treatment outcome was associated with significant be contraindicated for the CA population. The results of this study
improvement in interpersonal skills (IIP), role functioning (SAS- suggest that this conclusion is premature and too general. The
SR), and social support (ISEL). These findings are important inclusion of skills training in addition to exposure provides an
because they address a central concern of CA survivors who report alternative approach for patients like CA survivors who, because
significantly impaired functional capacity and impoverished qual- of affective and interpersonal regulation difficulties, might other-
ity of life, especially in regard to their social environment. In wise be inappropriate candidates for exposure-alone treatment.
addition, there is evidence that complaints concerning interper- The study results support a widely advocated model of interven-
sonal and social functioning rather than disorder-specific symp- tion for CA survivors (e.g., Herman, 1992) that includes three
toms are the predominant predictors of use of mental health principal components: the strengthening of self-management and
services among chronic PTSD patients (Ford, Fisher, & Larson, interpersonal effectiveness, an established therapeutic relationship,
1997), suggesting the importance of such outcome measures in and emotional processing of the trauma memories. The positive
assessing treatment effectiveness. results of the study provide the rationale for further research on
The participants treated in this study were representative of the STAIR–modified PE, especially as compared with nonexposure
CA population and, as expected, carried severe comorbid psycho- approaches for CA-related PTSD, and across other settings (e.g.
pathology, reflected in high rates of Axis I comorbidities, histories community services) and more diversified trauma populations.
of suicide attempts, self-harm behaviors, and crisis (emergency
room) interventions. Despite the relatively impaired sample treated
References
in this study, treatment outcome was generally good. The drop-out
rate was 29%, a proportion very similar to those found in studies American Psychiatric Association. (1994). Diagnostic and statistical man-
of adult-onset trauma populations (Marks et al., 1998; Resick, ual of mental disorders (4th ed.). Washington, DC: Author.
Nishith, Weaver, & Astin, 1999; Tarrier et al., 1999). Negative Bagby, M. R., Parker, J. D. A., & Taylor, G. (1993). The twenty-item
outcome was limited, with 1 participant (4.5%) experiencing wors- Toronto Alexithymia Scale: I. Item selection and cross-validation of the
ening of PTSD symptoms after treatment compared with baseline. factor structure. Journal of Psychosomatic Research, 38, 23–32.
This compares favorably with the results found by Tarrier et al. Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1981).
(1999), where 9 (31%) participants in an imaginal exposure treat- An inventory for measuring depression. Archives of General Psychia-
try, 4, 561–571.
ment and 3 (10%) in cognitive therapy were found to have PTSD
Blake, D. D., Weather, F. W., Nagy, L. M., Kaloupek, D. G., Gusman,
symptom exacerbation. In addition, STAIR–modified PE com- F. D., Charney, D. S., & Keane, T. M. (1995). The development of a
pleters experienced continued PTSD symptom reduction posttreat- clinician-administered PTSD scale. Journal of Traumatic Stress, 8,
ment, a finding similar to that observed among rape victims in PE 75–90.
treatment (Foa et al., 1991). At 9 months, completers showed a Briere, J. (1988). The long-term clinical correlates of childhood sexual
significant posttreatment improvement in interpersonal function- victimization. Annals of the New York Academy of Sciences, 528, 327–
ing, an outcome that had not emerged at 3 months. 334.
Some aspects of the study should be kept in mind. The small Briere, J. (1992). Child abuse trauma: Theory and treatment of the lasting
sample size, focus on women, and exclusion criteria limit gener- effects. Newbury Park, CA: Sage.
alizability of the findings. In addition, for practical and ethical Briere, J., & Runtz, M. (1990). Augmenting Hopkins SCL Scales to
measure dissociative symptoms: Data from two nonclinical samples.
reasons, we did not conduct 3- and 9- month follow-ups with the
Journal of Personality Assessment, 55, 376 –379.
wait-list group. Thus, the possibility that the maintenance in im- Burnstein, A. (1986). Treatment noncompliance in patients with posttrau-
provement seen in those who participated in STAIR–modified PE matic stress disorder. Psychosomatics, 27, 37– 40.
was a result of the passage of time or other factors not related to Cantanzaro, S. J., & Mearns, J. (1990). Measuring generalized expectations
the treatment cannot be discounted. Last, it should be noted that for negative mood regulation: Initial scale development and implica-
although this application of PE is consistent with the principles and tions. Journal of Personality Assessment, 54, 546 –563.
goals of exposure, it is not identical to the PE treatment used in Chemtob, C. M., Novaco, R. W., Hamada, R. S., Gross, D. M., & Smith,
other studies (Foa et al., 1991, 1999). First, in recognition of the G. (1997). Anger regulation deficits in combat-related posttraumatic
vulnerabilities of the CA population, we included an emotion- stress disorder. Journal of Traumatic Stress, 10, 17–35.
focused stabilization period after the exposure. Second, we used Cloitre, M., & Koenen, K. (2001). Interpersonal group process treatment
for CSA-related PTSD: A comparison study of the impact of borderline
imaginal exposure only and eliminated the situation-specific in
personality disorder on outcome. International Journal of Group Psy-
vivo exposure. We viewed current interpersonal difficulties as chotherapy, 51, 379 –398.
resulting not only from trauma-specific fear (i.e., a phobic re- Cloitre, M., Scarvalone, P., & Difede, J. (1997). Posttraumatic stress
sponse) but also from maladaptive interpersonal schemas that were disorder, self- and interpersonal dysfunction among sexually retrauma-
demonstrated in numerous and varied behaviors and situations. For tized women. Journal of Traumatic Stress, 10, 435– 450.
this reason, we encouraged the identification of core interpersonal Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159.
1074 CLOITRE, KOENEN, COHEN, AND HAN

Cohen, S., & Hoberman, H. (1983). Positive events and social supports as Resick, P. A., & Schnicke, M. K. (1992). Cognitive processing therapy for
buffers of life change stress. Journal of Applied Social Psychology, 13, sexual assault victims. Journal of Consulting and Clinical Psychol-
99 –125. ogy, 60, 748 –756.
Falsetti, S. A., Resnick, H. S., Resick, P. A., & Kilpatrick, D. G. (1993). Rorty, M. (1996). Histories of childhood trauma and complex posttrau-
The modified PTSD Symptom Scale: A brief self-report measure of matic sequelae in women with eating disorders. Psychiatric Clinics of
posttraumatic stress disorder. Behavior Therapist, 16, 161–162. North America, 19, 773–791.
Finkelhor, D., Hotaling, G., Lewis, I. A., & Smith, C. (1990). Sexual abuse Roth, S., Newman, E., Pelcovitz, D., van der Kolk, B., & Mandel, F.
in a national survey of adult men and women: Prevalence, characteristics (1997). Complex PTSD in victims exposed to sexual and physical abuse:
and risk factors. Child Abuse and Neglect, 14, 19 –28. Results from the DSM–IV field trial for posttraumatic stress disorder.
First, M. B., Spitzer, R. L., Gibbon, M., Williams, J. B. W., & Benjamin, Journal of Traumatic Stress, 10, 539 –555.
L. (1994). Structured Clinical Interview for DSM–IV Axis II Personality Scott, M. J., & Stradling, S. G. (1997). Client compliance with exposure
Disorders (SCID-II). New York: New York State Psychiatric Institute, treatments for posttraumatic stress disorder. Journal of Traumatic
Biometrics Research Department. Stress, 10, 523–526.
Foa, E. B., Dancu, C. V., Hembree, E. A., Jaycox, L. H., Meadows, E. A., Speilberger, C. D. (1983). Manual for the State–Trait Anxiety Inventory
& Street, G. (1999). A comparison of exposure therapy, stress inocula- (Form Y; Self-Evaluation Questionnaire). Palo Alto, CA: Consulting
tion training and their combination for reducing posttraumatic stress Psychologists Press.
disorder in female assault victims. Journal of Consulting and Clinical Speilberger, C. D. (1991). State–Trait Anger Expression Inventory (Rev.
Psychology, 67, 194-200. research ed., professional manual). Odessa, FL: Psychological Assess-
Foa, E. B., & Rothbaum, B. O. (1998). Treating the trauma of rape. New ment Resources.
York: Guilford Press. Spitzer, R. L., Willliams, J. B., Gibbon, M., & First, M. B. (1994).
Foa, E. B., Rothbaum, B. O., Riggs, D. S., & Murdock, T. B. (1991). Structured Clinical Interview for DSM–IV—Patient edition. New York:
Treatment of posttraumatic stress disorder in rape victims: A comparison New York State Psychiatric Institute, Biometrics Research Department.
between cognitive– behavioral procedures and counseling. Journal of Tarrier, N., Pilgrim, H., Sommerfield, C., Faragher, B., Reynolds, M.,
Consulting and Clinical Psychology, 59, 715–723. Graham, E., & Barrowclough, C. (1999). A randomized trial of cognitive
Ford, J., Fisher, P., & Larson, L. (1997). Object relations as a predictor of therapy and imaginal exposure in the treatment of chronic posttraumatic
treatment outcome with chronic posttraumatic stress disorder. Journal of stress disorder. Journal of Consulting and Clinical Psychology, 67,
Consulting and Clinical Psychology, 65, 547–559. 13–18.
Herman, J. L. (1992). Trauma and recovery. New York: Basic Books. Tracey, T. J., & Kokotovic, A. M. (1998). Factors structure of the Working
Horowitz, L. M., Rosenberg, S. E., Baer, B. A., Ureno, G., & Villasenor, Alliance Inventory. Psychological Assessment, 10, 207–210.
V. S. (1988). Inventory of Interpersonal Problems: Psychometric prop- Turner, S. W., McFarlane, A. C., & van der Kolk, B. A. (1996). The
erties and clinical applications. Journal of Consulting and Clinical therapeutic environment and new explorations in the treatment of post-
Psychology, 56, 885– 892. traumatic stress disorder. In B. A. van der Kolk, A. C. McFarlane, & L.
Jaycox, L. H., & Foa, E. (1996). Obstacles in implementing exposure Weisaeth (Eds.), Traumatic stress: The effects of overwhelming experi-
therapy for PTSD: Case discussions and practical solutions. Clinical ence on mind, body, and society (pp. 537–596). New York: Guilford
Psychology and Psychotherapy, 3, 176 –184. Press.
Kessler, R. C., Sonnega, A., Bromet, E., Hughes, M., & Nelson, C. (1995). van der Kolk, B. A. (1996). The complexity of adaptation to trauma:
Posttraumatic stress disorder in the National Comorbidity Survey. Ar- Self-regulation, stimulus, discrimination, and characterological develop-
chives of General Psychiatry, 52, 1048 –1060. ment. In B. A. van der Kolk, A. C. McFarlane, & L. Weisaeth (Eds.),
Linehan, M. M. (1993). Cognitive– behavioral treatment for borderline Traumatic stress: The effects of overwhelming experience on mind,
personality disorder. New York: Guilford Press. body, and society (pp. 182–213). New York: Guilford Press.
Marks, I., Lovell, K., Noshirvani, H., Livanou, M., & Thrasher, S. (1998). van der Kolk, B. A., Roth, S., & Pelcovitz, D. (1993). Complex PTSD:
Treatment of posttraumatic stress disorder by exposure and/or cognitive Results of the PTSD field trials for DSM–IV. Washington, DC: Amer-
restructuring. Archives of General Psychiatry, 55, 317–325. ican Psychiatric Association.
McDonough-Coyle, A., Friedman, M. J., McHugo, G., Ford, J., Mueser, Weissman, E., & Bothell, S. (1976). Assessment of patient social adjust-
K., Demment, C. C., et al. (2000, November). Cognitive restructuring ment by patient self-report. Archives of General Psychiatry, 33, 1111–
and exposure treatment for CSA survivors with PTSD. In M. Cloitre 1115.
(Chair), Empirically based treatments for childhood abuse and the Zlotnick, C. (1999). Antisocial personality disorder, affect dysregulation
multiply traumatized. Symposium conducted at the meeting of the In- and childhood abuse among incarcerated women. Journal of Personality
ternational Society for Traumatic Stress Studies, Miami, FL. Disorder, 13, 90 –95.
Pitman, R., Altman, B., Greenwald, E., Longre, R. E., Macklin, M. L., Zlotnick, C., Shea, M. T., Pearlstein, T., Begin, A., Simpson, E., &
Poire, R. E., & Steketee, G. (1991). Psychiatric complications during Costello, E. (1996). Differences in dissociative experiences between
flooding therapy for posttraumatic stress disorder. Journal of Clinical survivors of childhood incest and survivors of assault in adulthood. The
Psychiatry, 52, 17–20. Journal of Nervous and Mental Disease, 184, 52–54.
Resick, P. A., Nishith, P., Weaver, T., & Astin, T. (1999, November).
Preliminary findings of a controlled trial comparing cognitive processing
therapy and prolonged exposure. In B. Rothbaum (Chair), Recent devel- Received March 25, 2001
opments in the treatment of PTSD. Symposium conducted at meeting of Revision received September 30, 2001
the International Society for Traumatic Stress Studies, Washington, DC. Accepted October 1, 2001 䡲