The relationship of early alliance ruptures and their resolution to process and outcome in three time-limited psychotherapies for personality disorders.

Abstract

This study examined the relationship of early alliance ruptures and their resolution to process and outcome in a sample of 128 patients randomly assigned to 1 of 3 time-limited psychotherapies for personality disorders: cognitive-behavioral therapy, brief relational therapy, or short-term dynamic psychotherapy. Rupture intensity and resolution were assessed by patient- and therapist-report after each of the first 6 sessions. Results indicated that lower rupture intensity and higher rupture resolution were associated with better ratings of the alliance and session quality. Lower rupture intensity also predicted good outcome on measures of interpersonal functioning, while higher rupture resolution predicted better retention. Patients reported fewer ruptures than did therapists. In addition, fewer ruptures were reported in cognitive-behavioral therapy than in the other treatments. (PsycINFO Database Record (c) 2010 APA, all rights reserved).

Full text

Page 1

THE RELATIONSHIP OF EARLY ALLIANCE RUPTURES AND
THEIR RESOLUTION TO PROCESS AND OUTCOME IN
THREE TIME-LIMITED PSYCHOTHERAPIES FOR
PERSONALITY DISORDERS
J. CHRISTOPHER MURAN
Beth Israel Medical Center, Albert Einstein
College of Medicine
JEREMY D. SAFRAN
New School for Social Research, Beth Israel
Medical Center
BERNARD S. GORMAN
State University of New York, Beth Israel
Medical Center
LISA WALLNER SAMSTAG
Long Island University, Beth Israel Medical
Center
CATHERINE EUBANKS-CARTER
Beth Israel Medical Center
ARNOLD WINSTON
Beth Israel Medical Center, Albert Einstein
College of Medicine
This study examined the relationship of
early alliance ruptures and their reso-
lution to process and outcome in a
sample of 128 patients randomly as-
signed to 1 of 3 time-limited psycho-
therapies for personality disorders:
cognitive–behavioral therapy, brief re-
lational therapy, or short-term dynamic
psychotherapy. Rupture intensity and
resolution were assessed by patient-
and therapist-report after each of the
first 6 sessions. Results indicated that
lower rupture intensity and higher rup-
ture resolution were associated with
better ratings of the alliance and ses-
sion quality. Lower rupture intensity
also predicted good outcome on mea-
sures of interpersonal functioning,
while higher rupture resolution pre-
dicted better retention. Patients re-
ported fewer ruptures than did thera-
pists. In addition, fewer ruptures were
reported in cognitive–behavioral ther-
apy than in the other treatments.
Keywords: therapeutic alliance, rupture
and resolution, personality disorders,
time-limited psychotherapies, treatment
outcome, dropout status
Although promising psychotherapeutic inter-
ventions have been identified for a range of dif-
ferent psychological disorders (e.g., Lambert,
2004), substantial numbers of patients fail to ben-
J. Christopher Muran, Department of Psychiatry, Beth Is-
rael Medical Center and Albert Einstein College of Medicine;
Jeremy D. Safran, Department of Psychology, New School
for Social Research and Department of Psychiatry, Beth Israel
Medical Center; Bernard S. Gorman, Department of Psychol-
ogy, Nassau Community College of the State University of
New York, and Beth Israel Medical Center; Lisa Wallner
Samstag, Department of Psychology, Long Island University
at Brooklyn, and Department of Psychiatry, Beth Israel Med-
ical Center; Catherine Eubanks-Carter, Department of Psy-
chiatry, Beth Israel Medical Center; Arnold Winston, Depart-
ment of Psychiatry, Beth Israel Medical Center and Albert
Einstein College of Medicine.
The authors also wish to acknowledge the contributions of
all the clinical supervisors and research assistants, as well as
the therapists and patients who participated.
The research was supported in part by a grant from the
National Institute of Mental Health MH50246. Portions of this
research were presented at the annual meeting of the Society for
Psychotherapy Research (SPR), Madison, WI, June 2007.
Correspondence regarding this article should be addressed to
J. Christopher Muran, Beth Israel Medical Center, First Avenue
at 16th Street, New York, NY 10003. E-mail: jcmuran@
chpnet.org
Psychotherapy Theory, Research, Practice, Training
2009, Vol. 46, No. 2, 233–248
© 2009 American Psychological Association
0033-3204/09/$12.00DOI: 10.1037/a0016085
233

Page 2

efit from these treatments, as evidenced by sig-
nificant patient dropout rates. Estimates of patient
attrition rates average about 47% and range as
high as 67% (Wierzbicki & Pekarik, 1993;
Sledge, Moras, Hartley, & Levine, 1990). Despite
the many advances made in psychotherapy treat-
ments and techniques, as Barrett, Chua, Crits-
Christoph, Gibbons, and Thompson (2008) ob-
serve, these high dropout rates are comparable to
those found over 50 years ago. Even without
consideration of the longstanding problem of at-
trition, the evidence indicates that there is still con-
siderable room for improvement. For example, in
their meta-analysis of randomized clinical trials for
major depression, generalized anxiety disorder, and
panic disorder, Westen and Morrison (2001) found
that only 63% of panic disorder patients, 52% of
generalized anxiety disorder patients, and 54% of
depressed patients who completed treatment were
considered improved at termination. It is important
to note that these findings are based on rather le-
nient criteria for improvement.
It is also important to note that estimates of
retention and improvement are probably inflated
by the fact that many patients are screened out of
research protocols due to the presence of compli-
cated diagnostic profiles. In their critical review,
Westen and Morrison (2001) found that the ex-
clusion rates of patients who presented with mul-
tiple diagnoses typically ranged from 60% to
70%. The majority of patients who seek treatment
in our mental health clinics or private practices,
however, do present with multiple diagnoses,
with estimates of comorbidity ranging from 40%
to 70% (see Kessler et al., 1994). Patients who
present with comorbid personality disorders have
been found to be particularly treatment resistant
(Benjamin & Karpiak, 2001; Clarkin & Levy,
2004; Piper & Joyce, 2001). These challenging
patients are commonly encountered in outpatient
settings: the Rhode Island Methods to Improve
Diagnostic Assessment and Services (MIDAS)
project, the largest clinical epidemiological study
using semistructured interviews in an outpatient
setting, found that almost half (45.5%) of patients
presented with a personality disorder diagnosis
(Zimmerman, Chelminski, & Young, 2008). Sim-
ilarly, a study of patients seen by a community
mental health team in the United Kingdom found
that 52% met criteria for one or more personality
disorders (Keown, Holloway, & Kuipers, 2002).
Given this perspective on the effectiveness of
psychotherapy, it seems critical to identify vari-
ables that may mediate treatment outcome. One
such variable that has received considerable at-
tention is the therapeutic alliance, which has con-
sistently been shown to be a robust predictor of
outcome regardless of treatment modality (Hor-
vath & Symonds, 1991; Martin, Garske, & Davis,
2000). A number of clinical researchers have
noted that because personality-disordered pa-
tients present with longstanding and inflexible
patterns of emotional and interpersonal difficul-
ties (Benjamin, 1993; Livesley, 2001; Millon &
Davis, 1996), they invariably pose great chal-
lenges to therapists, especially with regard to the
therapeutic alliance (e.g., Benjamin & Karpiak,
2001; Muran, Segal, Samstag, & Crawford,
1994). Therapists are more likely to encounter
problems in the alliance with such patients, given
their emotional lability or constriction that makes
empathy difficult, as well as their restricted range
of interpersonal behavior that exerts a strong pull
for certain behavioral responses from therapists,
which in turn confirm and perpetuate patients’
pathogenic beliefs (Kiesler, 1996).
The research on problems, or ruptures, in the
therapeutic alliance is a still developing literature,
especially with regard to the prevalence of rup-
tures and the relevance of their resolution for
outcome (see Safran, Muran, Samstag, &
Stevens, 2002). There is, however, ample evi-
dence that weakened alliances are correlated with
unilateral termination by the patient (Samstag,
Batchelder, Muran, Safran, & Winston, 1998;
Tryon & Kane, 1990, 1993, 1995), as well as the
related finding that poor outcome cases show
greater negative interpersonal process (i.e., hos-
tile and complex interactions) than good outcome
cases (e.g., Coady, 1991; Henry, Schacht, &
Strupp, 1986; Samstag et al., 2008). These find-
ings suggest that the process of recognizing and
resolving ruptures in the therapeutic alliance may
play an important role in treatment retention and
outcome.
Several studies that have examined patient-
rated alliance patterns across time have also
yielded findings that suggest that rupture resolu-
tion is related to good outcome. Some studies
have found evidence of U-shaped (high-low-
high) patterns predicting good outcome (Kiv-
lighan & Shaugnessy, 2000; Patton, Kivlighan, &
Multon, 1997), while others have identified
briefer, more localized V-patterns related to good
outcome (Stiles et al., 2004; Strauss et al., 2006).
These efforts can be understood as indirect as-
Muran et al.
234

Page 3

sessments of ruptures and their resolution over
the course of treatment by means of examining
levels and patterns in repeated self-report mea-
sures of the quality of the therapeutic alliance.
By contrast, the present study examined the
relationship of ruptures and their resolution to
outcome by means of direct assessments: specif-
ically, patient- and therapist-report measures de-
signed to assess ruptures and their resolution
within treatment sessions. Our aim was to exam-
ine the relation of these measures to standard
psychotherapy process measures, such as those
assessing therapeutic alliance and session quality,
and to treatment outcome, including dropout sta-
tus. The focus was on early psychotherapy
process—postsession ratings of the first six ses-
sions of 30-session treatment protocols—because
the quality of the alliance as measured early in
treatment has been shown to be the most predic-
tive of outcome (Horvath & Symonds, 1991) and
because the preponderance of dropouts occur
within the first few sessions of treatment, with
several studies finding a median length of treat-
ment of about six sessions (Garfield, 1994).
Given the challenges noted above in developing a
good alliance with personality-disordered pa-
tients, the treatment cases in this study involved
patients with personality disorder diagnoses.
Analyses also included examination of differ-
ences between patient and therapist perspectives
of ruptures and their resolution, as well as differ-
ences in rupture intensity and resolution among
three time-limited treatment conditions: a short-
term dynamic psychotherapy, cognitive–
behavioral therapy (CBT), and an integrative re-
lational model. This data set was previously
examined with regard to relative treatment effi-
cacy, and the three treatments were generally
found to be equally effective (Muran, Safran,
Samstag, & Winston, 2005).
Method
Participants
Both patients and therapists provided informed
consent with respect to the parameters of the
research protocol. Patient and therapist demo-
graphics are provided in brief below (see Muran
et al., 2005, for elaboration). Patients paid a nom-
inal fee per session based on an income-sensitive
sliding scale in order to approximate a naturalis-
tic treatment setting; these fees ranged from $20
to $80 per session.
Patients.
The patients included 60 men and
68 women (N ? 128), ranging in age from 21 to
65 years (M ? 41.33, SD ? 10.52). Fifty percent
of the patients were single, never married; 28%
married or remarried; 21% divorced or separated;
and 1% widowed. Nineteen percent were high
school graduates, 50% college graduates, and
31% had graduate degrees. Eighty-one percent
were employed. Ninety percent were White, 1%
Black, 5% Latino, and 4% other. Thirty-seven
percent were Jewish, 34% Christian, and 29%
other. The patients presented with a myriad of
longstanding difficulties related to depression,
anxiety, and interpersonal functioning. Fifty-five
percent met criteria for a current primary diagno-
sis of Mood Disorder, 28% Anxiety Disorder,
13% V-Code, and 4% Adjustment Disorder on
Axis I of DSM–IV (American Psychiatric Asso-
ciation, 1994); and 35% met criteria for multiple
Axis I diagnoses. The principal inclusion crite-
rion was a diagnosis of Personality Disorder (PD)
Cluster C or Not Otherwise Specified on Axis II.
We focused on Cluster C patients rather than
Cluster A or B personality-disordered patients,
because of a concern that patients meeting crite-
ria for Cluster A or B diagnoses would be more
likely to need a longer term therapeutic approach.
Sixty-six percent met criteria for a diagnosis of
PD NOS, 22% Avoidant PD, 10% Obsessive-
Compulsive PD, and 2% Dependent PD; and
19% met criteria for multiple noncluster A or B
diagnoses. Patient diagnoses were reliably estab-
lished with the Structured Interview for DSM–IV-
Axis I & II (SCID: First, Spitzer, Gibbon, &
Williams, 1995) administered by trained research
assistants (see Muran et al., 2005, for details).
Therapists.
Seventy therapists (36% men and
64% women) treated the 128 patients in this
study. Therapists ranged in age from 25 to 65
years (M ? 38.14, SD ? 8.89) and included
clinical psychologists (41%), psychiatry attend-
ing physicians (12%) and residents (7%), social
workers (11%), and psychology interns (17%),
and externs (12%) in a psychiatry department at a
major metropolitan medical center. Clinical ex-
perience ranged from one to 35 years, with an
average experience level of 4.77 years (SD ?
7.50). Ninety-six percent of the therapists were
White, 2% Asian, and 2% Black or Latino. Forty-
five percent were single, never married; 50%
married or remarried; and 5% divorced or sepa-
Alliance Ruptures and Resolution
235

Page 4

rated. Fifty-nine percent of therapists were Jew-
ish, 18% Christian, 8% other, and 15% reported
no religious affiliation. All therapists attended a
90-min weekly case seminar throughout their
participation in the study. They also received 1-hr
weekly individual supervision for their first
case, and continued in individual supervision if
they were unlicensed. In preliminary analyses
(Muran et al., 2005), it was determined that
there was a statistically significant difference
among the treatment conditions with respect to
training degrees, but this had no significant
relation to treatment outcome, including drop-
out status; in addition, there was no statistically
significant therapist effect on outcome (see
Muran et al., 2005 for details regarding case
assignments by therapist for each treatment
condition). The therapists selected which treat-
ment condition they would participate in; their
choices were generally consistent with their
own preexisting orientations.
The 128 patients were admitted into this study
during an 8-year period from 1992 to 2000 and
did not include 18 patients (13 in the short-term
dynamic therapy and 5 in the CBT from an orig-
inal overall sample of 146) who were determined
to be at risk for treatment failure based on post-
session ratings completed by patients and thera-
pists on a number of dimensions, and were
offered the opportunity to be reassigned to
another treatment condition as part of another
study (Safran, Muran, Samstag, & Winston,
2005). Of the 128 patients in this study, 84
completed the treatment protocol and 44 did
not, terminating prematurely.
Treatment Models and Training Procedure
The three treatment models were manualized
and designed to treat personality disordered pa-
tients in a fixed 30-session, one-session-per-week
format. All sessions were videotaped and con-
ducted at an outpatient mental clinic in a general
hospital setting. Fifty-six (44%) of the 128 cases
were randomly sampled to evaluate treatment
adherence (i.e., the extent to which the therapists
conducted the treatments according to the respec-
tive manuals); adequate adherence was estab-
lished (see Muran et al., 2005, for details). Brief
summaries of the models are provided below (see
Safran et al., 2005, for elaboration). Previously
reported findings generally indicated that the
three treatments were equally effective, but there
was a significant difference regarding dropout
rates, favoring Brief Relational Therapy (Muran
et al., 2005).
Short-term dynamic psychotherapy.
Term Dynamic Psychotherapy (STDP: Pollack,
Flegenheimer, Kaufman, & Sadow, 1992; also
referred to as Brief Adaptive Psychotherapy) is a
time-limited dynamic model similar in many re-
spects to the approaches of Strupp and Binder
(1984) and Luborsky (1984). The general ap-
proach to the technique is one in which therapists
help patients gain insight into maladaptive trans-
actional patterns or core conflictual relationship
themes through interpretation. The treatment pro-
cess begins with the establishment of a case for-
mulation and identification of a major maladap-
tive pattern that the therapist and patient contract
to make the focus of treatment. The balance
of treatment is marked by interpretation of patient
transference material, exploring the details of the
pattern, and making links to both in-session and
extrasession material. The treatment goal is the
resolution of the conflict inherent in the pattern.
CBT.
The CBT treatment employed in this
study (Turner & Muran, 1992) is a schema-
focused model, in that personality and behavior is
understood as being organized by underlying be-
lief systems about the self that have become
“structuralized.” The general technical principle
involves constructing differential learning expe-
riences that challenge the content of maladaptive
schemas. Like STDP, the treatment process be-
gins with establishing a case formulation, which
includes defining a problem list and clarifying
core belief systems (Persons, 1989). The course
of treatment then involves the application of var-
ious cognitive and behavioral tasks, including
those assigned as homework, to challenge and
correct the patient’s irrational or dysfunctional
beliefs. The therapeutic relationship is founded
on the principle of “collaborative empiricism”
(Beck, Rush, Shaw, & Emery, 1979), whereby
the patient and therapist collaborate to test the
validity and viability of the patient’s beliefs.
Brief relational therapy.
therapy (BRT: Muran & Safran, 2002; Safran &
Muran, 2000; Safran, 2002) is a model that inte-
grates principles emerging out of contemporary
relational developments of psychoanalysis (see
Mitchell & Aron, 1999) with findings derived
from our research on rupture resolution (Safran &
Muran, 1996). A central assumption in BRT is
that therapists can never stand completely outside
Short-
Brief relational
Muran et al.
236

Page 5

of the interpersonal field and look at the patient
objectively, and that to various degrees they un-
wittingly participate in vicious interpersonal cy-
cles with their patients. A key technical principle
is therapeutic metacommunication, which is an
attempt to disembed from these cycles by com-
municating about the communication process
(Kiesler, 1996). In contrast to the other two treat-
ment models, BRT places greater emphasis on
exploring interpersonal process and eschews es-
tablishing a case formulation early in treatment. It
is oriented toward cultivating awareness of the
self in relation to the other, rather than resolving
a central conflict or correcting an irrational belief.
Its primary task is to track alliance ruptures as
markers of vicious cycles and to engage the pa-
tient in a collaborative inquiry about these cycles.
BRT is essentially based on a social construction-
ist model of the therapeutic relationship, whereby
ruptures and their resolution are understood as
coparticipatory processes involving both patient
and therapist.
Measures and Assessment Procedure
Patients and therapists were asked to complete
various measures in the research program that
can be considered indices of treatment process
and outcome (see Muran, 2002; Muran et al.,
2005). Some of these measures are described
below because of their relevance for this study.
Treatment process.
After every session, pa-
tients and therapists were asked to complete par-
allel forms of a postsession questionnaire (PSQ:
Muran, Safran, Samstag, & Winston, 1992),
which consists of several measures assessing ses-
sion impact and the therapeutic alliance. Patients
were assured that their therapists would not have
access to their responses to the PSQ. They were
also provided with identification numbers, a pri-
vate area to complete the measure, and a deposit
system of locked mailboxes for completed ques-
tionnaires in order to ensure the confidentiality of
their reports and to increase the probability of
accurate reporting. PSQ data collected from the
first six sessions were used in this study as evi-
dence of early psychotherapy process. Comple-
tion rates for the PSQ across the first six therapy
sessions ranged from 70–86% for patients and
74–92% for therapists.
The PSQ includes two measures that have been
widely used in psychotherapy research and have
demonstrated sound psychometric properties, in-
cluding internal consistency and predictive valid-
ity with regard to a variety of overall outcome
indices. One is the 12-item version of the Work-
ing Alliance Inventory (WAI: Tracey & Koko-
tovic, 1989), from which an overall mean score
can be calculated. The other is the 12-item Ses-
sion Evaluation Questionnaire (SEQ: Stiles,
1980), which is scaled according to a semantic
differential and yields two subscales regarding
session smoothness and session depth of explo-
ration. The overall mean of the WAI, and the
SEQ smoothness and depth subscales (SEQ/S
and SEQ/D), were used as standard measures of
psychotherapy process in this study. These mea-
sures were averaged across the six sessions.
The PSQ has other measures that have more
limited psychometric support. These include
three direct questions regarding ruptures and their
resolution, all scaled in a five-point Likert-type
format, plus an open-ended description: Did you
experience any tension or problem, any misun-
derstanding, conflict or disagreement, in your
relationship with your _____ [therapist/patient]
during the session (Rupture Presence)? If yes,
please rate how tense or upset you felt about the
problem during the session (Rupture Intensity)?
Please describe the problem (Rupture Descrip-
tion). To what degree do you feel this problem
was resolved by the end of the session (Rupture
Resolution)?
The following Rupture Descriptions, provided
by patients in this sample, illustrate the range of
problems that patients reported as ruptures:
“Toward the end of the session, when I seemed
to run out of things to say, there were periods of
silence, and I began to feel intimidated because of
the silence. I also felt judged because she just
stared at me and didn’t say anything.” (STDP)
“When I was asked about what it would take
for me to feel comfortable with her and I said
that it would depend on if she liked me. I felt
vulnerable.” (BRT)
“I felt she was preaching to me and not letting
me discover things myself. I needed time to ab-
sorb new ideas.” (CBT)
In this study, Rupture Presence (RPP) was
examined as part of preliminary analyses to de-
scribe our data set. Patients’ Likert scale ratings
of RPP were reduced to two categories: no rup-
ture (ratings of 1 on the Likert scale) or rupture
(ratings of 2 through 5). Rupture Intensity (RPI)
was averaged across the six sessions, (zero was
recorded if no rupture was reported). Patient-
Alliance Ruptures and Resolution
237

Page 6

rated RPP in our dataset was significantly related
to RPI, r(124) ? .84, p ? .001; and therapist-
rated RPP was significantly related to RPI,
r(128) ? .55, p ? .001. In order to limit the
experiment-wise error rate, we used these large
and significant correlations to justify focusing
exclusively on RPI as the rupture index in our
main analyses. Rupture Resolution (RES) was
also examined as a predictor variable. For each
case, patient-rated RES was summed across the
six sessions and then divided by the number of
sessions in which the patient reported a rupture.
Treatment outcome.
assessed on multiple dimensions, including mea-
sures of subjective distress or symptomatology,
level of adaptive functioning, and interpersonal
or personality style, which is consistent with rec-
ommendations regarding the assessment of
change in personality disorders (Shea, 1997). Pa-
tients completed a battery of measures that
tapped these dimensions at different intervals
during the treatment, including intake and termi-
nation. Therapists also completed a number of
outcome measures after the third session of treat-
ment (in order to establish intake levels of patient
functioning) and then again at termination. Com-
pletion rates, for patient-rated and therapist-rated
outcome measures were 86% and 95%, respec-
tively, in this study. These outcome measures
included the following:
The Symptom Checklist-90 Revised (SCL-90R:
Derogatis, 1983) is a patient-rated self-report in-
ventory developed to assess general psychiatric
symptomatology. It consists of 90-items scaled in
a Likert-type format on degree of severity. Nor-
mative data and adequate psychometric proper-
ties have been reported. In this study, the Global
Severity Index (GSI), which is an overall mean
score, was used.
The Target Complaints (TC: Battle et al.,
1966) measure is an idiographic instrument de-
veloped to assess patients’ presenting problems.
Space is provided for three problems per patient,
and each problem is rated on a Likert-type scale
in terms of degree of severity. Patients identifythe
problems, and then both patients and therapists in-
dependently rate problem severity. In this study,
patients’ ratings of the three problems were aver-
aged for an overall patient target complaint (PTC)
index, and therapists’ ratings were averaged for an
overall therapist target complaint (TTC) index.
The Global Assessment Scale (GAS: Endicott,
Spitzer, Fleiss, & Cohen, 1976) is a clinician-
Treatment outcome was
rated scale for evaluating the overall mental
health of a patient. It involves a single rating on
a continuum ranging from 1, which represents
the hypothetically sickest individual, to 100,
the hypothetically healthiest. All therapists
were trained to reliable standards (i.e., intra-
class correlation ?.90).
The Inventory of Interpersonal Problems (IIP:
Horowitz, Alden, Wiggins, & Pincus, 2000) is an
inventory developed to assess patient social ad-
justment and interpersonal difficulties. A short-
form to be rated by the patient was developed
from factor analytic procedures. It consists of 64
items scaled in a Likert-type format on degree of
distress. Normative data and adequate psycho-
metric properties have been reported. In this
study, the overall mean score was used to deter-
mine outcome.
The Wisconsin Personality Inventory (WISPI:
Klein et al., 1993) is a 214-item self-report ques-
tionnaire scaled in a Likert-type format and de-
rived from an interpersonal perspective on the
DSM-IIIR model of PDs. It includes 11 PD sub-
scales. Some normative data and adequate psy-
chometric properties have been reported. In this
study, the overall mean score was used to deter-
mine outcome.
In order to reduce the number of statistical
tests and experiment-wise error rate, as well as to
limit the impact of shared method variance by
combining patient- and therapist-report, these
outcome measures were submitted to a data re-
duction procedure that began by calculating stan-
dardized residual gain scores for each measure
based on intake and termination reports and then
conducting a principal components analysis with
varimax rotation. We extracted two factors with
eigenvalues exceeding 1.00 and with a substantial
percentage of the variance accounted for: one
factor that can be interpreted as measuring Axis I
(symptomatology), the other as Axis II (interper-
sonal functioning) of DSM–IV. We then calcu-
lated outcome composites for each interval by
averaging the standardized residual gain scores of
the measures that loaded ?.45 on the respective
factors and by applying the yielded factor scores
as weights. As reported in Muran et al. (2005), no
statistically significant differences between the
treatment conditions were found on these two
outcome factors.
In addition to determining outcome based on
the measures described above, treatment outcome
was also measured by treatment completion (i.e.,
Muran et al.
238

Page 7

completion of all 30 sessions of the treatment
protocol). In this regard, cases were categorized
as completed or dropped out; dropouts were not
included in the analyses of outcome measures.
Dropout status was defined as termination before
the contracted 30 sessions and unilaterally deter-
mined by the patient; premature termination re-
sulting from a change in location of residence
was not considered dropout, and cases that ter-
minated for this reason were not included in this
sample of 128 (see Samstag et al., 1998; Wier-
zbicki & Pekarik, 1993, for rationale regarding
definitional parameters). As reported in Muran et
al. (2005), a statistically significant difference
was found among the three treatment conditions
with regard to dropout: 46% of the patients (19 of
41) dropped from STDP, 37% (17/46) from CBT,
and 20% (8/41) from BRT. Post hoc analyses
indicated that the significant difference was be-
tween STDP and BRT; the difference between
CBT and BRT only approached statistical signif-
icance (p ? .10).
Results
Preliminary Analyses
A number of preliminary analyses were con-
ducted to describe our data set and establish its
consistency with previous research efforts that
have focused on comparable variables.
Table 1 presents the frequencies and percent-
ages of patient- and therapist-reported RPPs for
the three treatment conditions during each of the
first six sessions. A series of multilevel logistic
regression analyses were performed in order to
assess effects due to sessions (within-dyad fac-
tor), patient- versus therapist-report (within-dyad
factor), and treatment condition (between-dyad
factor).1An initial model using all three factors
indicated no session effect: that is, there were no
statistically significant session-to-session differ-
ences in rupture frequencies across the first six
sessions. Therefore, a simpler model analyzing
patient- versus therapist-report and treatment
condition was conducted. The results indicated
statistically significant effects for both factors.
Therapists were significantly more likely to
report ruptures than were their patients
(Wald ?2[1] ? 15.39, p ? .001). There was also
a statistically significant difference among the
treatment conditions (Wald ?2[2] ? 8.19, p ?
.05). Post hoc analyses using a Bonferroni cor-
rection indicated that the CBT condition had
fewer rupture reports than did BRT and STDP,
with no significant difference between BRT and
STDP. Finally, there was a significant interaction
between reporter and treatment condition in
which the effect of the treatment on ruptures
reported was significantly stronger for therapists
than for patients (Wald ?2[2] ? 6.32, p ? .05).
Means and standard deviations for the process
variablesusedinthisstudyarepresentedinTable2.
Statistically significant differences were found
among the treatment conditions for several of
these variables. Post hoc Scheffe ´ tests indicated
that patients reported lower RPI in CBT than
STDP, which replicates in part what was found
with rupture frequency. This was the only
patient-rated process variable that was signifi-
cantly different by treatment—although the
patient-rated alliance as measured by the WAI
approached significance (p ? .084). Therapists
also reported significantly lower rupture intensity
in CBT than in STDP and BRT. Significant dif-
ferences were found among the treatment condi-
tions on therapist-rated RES, WAI, SEQ/S, and
SEQ/D, with post hoc tests indicating that the
CBT therapists reported more rupture resolution,
better working alliance, more session smooth-
ness, and greater depth of exploration than their
counterparts in the other conditions.
Table 3 presents Pearson product–moment cor-
relation coefficients of patient- by therapist-
ratings of all the psychotherapy process measures
administered in this study. A large, significant
correlation was found with the WAI, and a sig-
nificant but modest correlation was found with
the SEQ/D. These findings indicate that early in
the treatment process, patients and therapists
could agree on the quality of their alliance and
the depth of exploration in their work together. It
is interesting to note that patients and therapists
were generally not consistent in their perspec-
tives on rupture intensity and resolution, with
small to medium correlations, only one of
which—the correlation of ratings of rupture
intensity—reached statistical significance.
1In order to analyze these data, the investigators employed
the logistic regression program in STATA, Version 9 (Stata-
Corp, 2007, http://www.stata.com) using a clustering option
and robust standard errors to control for the non-independence
of observations within dyads. Dummy variables for each of
the factors and their interactions were constructed through the
use of the STATA supplementary program DESMAT.
Alliance Ruptures and Resolution
239

Page 8

Pearson correlations between the standard psy-
chotherapy process measures (WAI, SEQ/S, and
SEQ/D) and treatment outcome indices are pre-
sented in Table 4. Generally, the results provide
support for the predictive validity of the standard
process measures, consistent with the literature
(e.g., Horvath & Greenberg, 1989; Stiles, 1980).
More specifically, they indicate that the patient-
and therapist-rated versions of these measures
were especially predictive of change in symptom-
atology (i.e., Factor 1), and that patient- and
therapist-rated WAI were the most predictive
measures, significantly related to both change on
Factor 1 and dropout (dropout status was treated
as a continuous variable, ranging from 1 to 2).
These results support the use of the WAI and the
SEQ as measures of suboutcome in our main
analyses of the RPI and RES.
TABLE 1. Frequency and Percentage of Patient- and Therapist-Reported Ruptures by Treatment Condition and Treatment
Session (N ? 128)
Session
Patient ratedTherapist rated
STDP CBTBRT TotalSTDP CBTBRTTotal
1
2
3
4
5
6
Mean
10/34 (29%)
18/32 (56%)
12/31 (39%)
17/29 (59%)
11/27 (41%)
13/29 (45%)
14/30 (47%)
13/44 (30%)
8/38 (21%)
10/38 (26%)
8/30 (27%)
11/33 (33%)
4/32 (13%)
9/36 (25%)
13/34 (38%)
14/35 (40%)
12/29 (41%)
14/32 (44%)
10/29 (34%)
9/29 (31%)
12/31 (39%)
36/110 (33%)
40/105 (38%)
34/98 (35%)
49/91 (54%)
32/89 (36%)
26/90 (29%)
36/97 (37%)
30/34 (88%)
29/32 (91%)
26/28 (93%)
23/26 (88%)
23/27 (85%)
21/24 (88%)
25/29 (86%)
15/44 (34%)
8/41 (20%)
9/41 (22%)
12/38 (32%)
13/38 (34%)
7/40 (18%)
11/40 (28%)
22/40 (50%)
28/40 (70%)
23/37 (62%)
21/33 (64%)
17/32 (53%)
21/31 (68%)
22/36 (61%)
67/118 (57%)
65/113 (56%)
58/106 (55%)
56/97 (58%)
53/97 (55%)
49/95 (52%)
58/104 (56%)
Note.
submitted. BRT ? brief relational therapy; CBT ? cognitive behavioral therapy; STDP ? short-term dynamic
psychotherapy.
The table includes in each column the number of rupture events reported/number of postsession questionnaires
TABLE 2. Means, Standard Deviations, and Results From Tests of Between-Condition Differences on the Psychotherapy
Process Variables
Process
variables
Treatment condition
Total
(N ? 128),
M (SD)
F value (df)
Alpha
level
Post hoc
Scheffe ´ test
STDP (n ? 41),
M (SD)
CBT (n ? 46),
M (SD)
BRT (n ? 41),
M (SD)
Patient rated
RPI
RES
WAI
SEQ/S
SEQ/D
Therapist rated
RPI
1.31 (.89)
3.03 (1.10)
4.74 (.90)
4.22 (.84)
5.00 (.57)
.65 (.85)
3.02 (1.26)
5.11 (.81)
4.57 (.92)
4.88 (.83)
1.12 (1.11)
3.25 (.93)
4.79 (.74)
4.32 (.75)
4.93 (.69)
1.01 (.98)
3.11 (1.07)
4.90 (.82)
4.38 (.85)
4.93 (.71)
5.47 (2, 121) .005???
.36 (2, 64)
2.53 (2, 114) .084?
1.95 (2, 120)
.32 (2, 120)
STDP ? CBT??
ns
ns
ns
2.81 (.82) 1.83 (1.50)2.52 (1.07)2.37 (1.24)8.05 (2, 125) .001????
STDP ? CBT ?
BRT??
STDP ? CBT ?
BRT??
STDP ? CBT ?
BRT??
STDP ? CBT ?
BRT??
STDP ? CBT ?
BRT??
RES2.52 (.90) 3.41 (1.02)2.54 (.93) 2.76 (.97)9.30 (2, 95).001????
WAI4.41 (1.01)4.86 (.53) 4.32 (.66) 4.50 (.77) 6.88 (2, 118) .001????
SEQ/S 4.17 (.55) 4.52 (.68) 4.01 (.65)4.25 (.66) 7.27 (2, 124) .001????
SEQ/D 4.51 (.51) 4.92 (.65)4.60 (.46) 4.69 (.57)6.83 (2, 124) .001????
Note.
positive sign indicating positive direction for outcome. BRT ? brief relational therapy; CBT ? cognitive behavioral
therapy; RES ? rupture resolution; RPI ? rupture intensity; SEQ/D ? Session Evaluation Questionnaire–Depth of
Exploration Subscale; SEQ/S ? Session Evaluation Questionnaire–Smoothness Subscale; STDP ? short-term dynamic
psychotherapy; WAI ? Working Alliance Inventory (Overall Mean); ns ? not significant.
?p ? .10.
Factor 1 (Symptomatology) and Factor 2 (Interpersonal Functioning) reflect composite residual gain scores with
??p ? .05.
???p ? .01.
????p ? .001.
Muran et al.
240

Page 9

Main Analyses
As described above, our main analyses con-
cerned examining the relationships between our
measures of RPI and RES and standard measures
of psychotherapy process and outcome.
Pearson correlations of RPI and RES to the stan-
dard psychotherapy process measures, the WAI and
SEQ, are presented in Table 5. The results indicate
statistically significant negative correlations be-
tween patient-rated RPI and patient-rated WAI
and SEQ/S, likewise between therapist-rated RPI
and therapist-rated WAI and SEQ/S, such that
lower RPI was correlated with higher ratings of
the alliance and of session smoothness. Signifi-
cant positive correlations were found between
patient-rated RES and patient-rated WAI and
SEQ/D, as well as between therapist-rated RES
and therapist-rated WAI and SEQ/D, such that
greater rupture resolution was correlated with
higher ratings of the alliance and of session
depth. It should be noted that in these instances,
there was the possibility that shared method vari-
ance might have inflated the correlations. However,
there were also significant negative correlations
between patient-rated RPI and therapist-rated
WAI, SEQ/S and SEQ/D, as well as between
therapist-rated RPI and patient-rated WAI and
SEQ/S. Similarly, there were statistically signif-
icant positive correlations between patient-rated
RES and therapist-rated SEQ/Depth, and between
therapist-rated RES and patient-rated WAI—
analyses in which there was no shared method
variance. A significant negative correlation was
found between therapist-rated RES and therapist-
rated SEQ/S; however, this finding is qualified by
a significant interaction effect of treatment con-
dition, discussed in the additional analyses pre-
sented below.
Table 5 also presents Pearson correlations of
RPI and RES to the outcome measures. The re-
sults indicate that patient-rated RPI significantly
predicted outcome as measured on Factor 2
—such that the greater the intensity of ruptures,
the poorer the outcome with regard to interper-
sonal functioning. Therapist-rated RPI also sig-
nificantly predicted this outcome variable.
Patient- and therapist-rated RES significantly
predicted dropout: the more rupture resolution,
TABLE 3. Pearson Correlations of Patient by Therapist Ratings of Psychotherapy Process Measures (N ? 128)
Patient rated
Therapist rated
RPI RESWAISEQ/SSEQ/D
RPI
RES
WAI
SEQ/S
SEQ/D
.36??
.15
.57??
.10
.25??
Note.
Exploration Subscale; SEQ/S ? Session Evaluation Questionnaire–Smoothness Subscale; WAI ? Working Alliance
Inventory.
?p ? .05.
RES ? rupture resolution; RPI ? rupture intensity; SEQ/D ? Session Evaluation Questionnaire–Depth of
??p ? .01.
TABLE 4. Pearson Correlations for Psychotherapy Process (Standard Measures Only) by Treatment Outcome
Treatment outcome
Psychotherapy process
Patient rated Therapist rated
WAISEQ/SSEQ/D WAISEQ/SSEQ/D
Factor 1 (n ? 84)
Factor 2 (n ? 84)
Dropout (N ? 128)
.45???
.20
?.30???
.30??
.37??
?.04
.54???
.18
?.15
.38???
.13
?.27???
.30??
.09
?.06
.21?
.07
?.14
Note.
in symptomatology; Factor 2 ? change in interpersonal functioning. SEQ/D ? Session Evaluation Questionnaire–Depth
of Exploration Subscale; SEQ/S ? Session Evaluation Questionnaire–Smoothness Subscale; WAI ? Working Alliance
Inventory (Overall Mean).
?p ? .10.
Outcome was established such that the great the change coefficient the better the outcome. Factor 1 ? change
??p ? .05.
???p ? .01.
Alliance Ruptures and Resolution
241

Page 10

the better the treatment retention. As noted
above, dropout status was treated as a continuous
variable, ranging from 1 to 2.
Additional analyses.
ferences by treatment condition on some of our
process measures, we conducted additional anal-
yses to examine the relationships between the
direct rupture and resolution indices and standard
measures of process and outcome with treatment
condition as a moderator. Specifically, we con-
ducted a moderated multiple regression in which
the regression model predicts the dependent vari-
able from the independent variables and yields
results regarding moderator effect (i.e., the inter-
action effect of treatment condition and predictor
or independent variable). A logistic regression
analysis was conducted in a similar fashion when
dropout status was used as the dependent vari-
able. A statistically significant interaction effect
was found between patient-rated RPI and treat-
ment condition on patient-rated SEQ/D, F(2,
117) ? 3.99, p ? .05, on outcome on Factor 1,
F(2, 56) ? 4.97, p ? .05; and on dropout status,
Wald ?2[2] ? 6.16, p ? .05. To further explore
these interactions, Pearson correlations between
patient RPI and SEQ/D, Factor 1, and dropout
status were conducted separately for each treat-
ment condition. Significant negative correlations
Because we found dif-
between RPI and SEQ/D and between RPI and
Factor 1 were found in the CBT condition, such
that for patients in CBT, lower rupture intensity
was associated with greater session depth and
more symptom reduction. Correlations for pa-
tients in the other treatment conditions were not
significant.
A significant interaction effect was also found
between therapist-rated RES and treatment condi-
tion on therapist-rated SEQ/S, F(2, 91) ? 3.97, p ?
.05, outcome on Factor 1, F(2, 43) ? 3.76, p ? .05,
outcome on Factor 2, F(2, 49) ? 4.20, p ? .05, and
dropout status, ?2(2) ? 4.96, p ? .05. Follow-up
Pearson correlations within each treatment condi-
tion found significant positive associations between
therapist-rated RES and therapist-rated SEQ/S for
patients in CBT and BRT. That is, for patients in
CBT and BRT, higher ratings of resolution by the
therapist were correlated with greater session
smoothness per the therapist. No other correlations
were significant.
Discussion
The results of this study demonstrated that the
direct measures of rupture intensity and resolu-
tion were significantly related to standard mea-
sures of psychotherapy process and outcome.
TABLE 5. Pearson Correlations for Rupture Intensity and Resolution by Standard Measures of Psychotherapy Process and
Treatment Outcome (N ? 128)
Variable
Psychotherapy process
Treatment outcome Patient ratedTherapist rated
WAISEQ/SSEQ/D WAISEQ/SSEQ/DFactor 1Factor 2 Dropout
RPI
Patient rated
STDP
CBT
BRT
Therapist rated
RES
Patient rated
Therapist rated
STDP
CBT
BRT
?.32??
?.40??
?.11
?.18
?.38??
?.09
?.08
?.21?
?.31??
?.22?
?.15
?.34
?.49?
.00
?.08
?.35??
.07
.06
.05
.24
.05
?.34??
?.18?
?.38?
?.39??
?.15
?.32??
.48??
.22?
?.12
?.03
.39??
.05
.08
.45??
?.01
?.36??
?.06
.61??
.39?
.26?
.42??
.10
.17
.23
.43
.19
.13
.07
.02
.38
.10
?.29?
?.22?
?.28
?.34
?.16
Note.
positive sign indicating positive direction for outcome. BRT ? brief relational therapy; CBT ? cognitive behavioral
therapy; RES ? rupture resolution; RPI ? rupture intensity; SEQ/S ? Session Evaluation Questionnaire–Smoothness
Subscale; STDP ? Short-Term Dynamic Psychotherapy; SEQ/D ? Session Evaluation Questionnaire–Depth of
Exploration Subscale; WAI ? Working Alliance Inventory (Overall Mean). Correlation coefficients in bold are from
main analyses.
?p ? .05.
Factor 1 (Symptomatology) and Factor 2 (Interpersonal Functioning) reflect composite residual gain scores, with
??p ? .01.
Muran et al.
242

Page 11

Specifically, results indicated that lower rupture
intensity and higher rupture resolution were as-
sociated with better ratings of the alliance and
session quality. In addition, lower rupture inten-
sity predicted good outcome on measures of in-
terpersonal functioning, while higher rupture res-
olution predicted better treatment retention.
It bears noting at the outset that this study had
a correlational design, and thus did not prove that
ruptures cause poor process and poor outcome, or
that resolution processes cause improvements in
the alliance and prevent dropout. It is possible
that a third variable is responsible for the ob-
served relations: for example, one might expect
that patients with more severe personality pathol-
ogy would report more ruptures and fewer reso-
lutions than higher functioning patients, particu-
larly in the first few sessions of treatment;
personality pathology could also contribute to
weak alliances and poor treatment outcome, in-
dependent of the frequency or intensity of rup-
tures and resolution processes. Future research
could explore ways to address this issue; for
example, by conducting close analyses of the
relation of ruptures and resolution processes to
alliance ratings and intermediate outcome scores
within subjects, so that most patient and therapist
variables are held constant.
The differences found in this study between
patient and therapist perspectives, especially the
modest correlation on reports of rupture intensity
and low correlation on reports of rupture resolu-
tion, underscore that patients and therapists have
somewhat separate views of the therapy relation-
ship. The finding that patients reported fewer
ruptures than therapists is consistent with re-
search on session impact (e.g., Stiles & Snow,
1984) and studies of ratings of the alliance (e.g.,
Fitzpatrick, Iwakabe, & Stalikas, 2005; Hatcher,
Barends, Hansell & Gutfreund, 1995; Kivlighan
& Shaughnessy, 1995; Mallinckrodt & Nelson,
1991), which generally find that patients’ reports
are more positive than those of therapists. Hor-
vath (2000) has observed that differences be-
tween therapists’ and patients’ perspectives of the
therapeutic relationship may stem from the fact
that therapists view the relationship through a
“theoretical lens”: therapists compare the rela-
tionship to what they have been taught a good
therapeutic relationship looks like, whereas cli-
ents compare the relationship to their prior real-
world experiences. This difference in perspective
could be responsible for the more favorable rat-
ings patients generally provide: for the therapist,
a relationship with a patient will usually fall short
of a theoretical ideal, whereas for many patients,
especially those with personality disorders, a re-
lationship with a therapist will compare favorably
to their prior interpersonal experiences.
The possibility that therapists’ views were
shaped by their theoretical lenses is particularly
relevant in a treatment comparison study. Thera-
pists and patients in the CBT condition reported
fewer ruptures than the other conditions, which
suggests that fewer ruptures occurred in CBT
relative to STDP and BRT. However, the inter-
pretation of this finding is qualified by an inter-
action effect in which the effect of treatment
condition was stronger for therapists than for
patients, suggesting that therapists’ theoretical
lenses may have impacted their recognition of
ruptures. Consistent with this possibility, whereas
patient reports of rupture resolution, working al-
liance, and session smoothness and depth did not
differ by treatment condition, there were signifi-
cant differences between therapist reports from
the three conditions, with CBT therapists provid-
ing the most positive reports.
The differences observed between the therapist
reports from the three treatment conditions may
be due to differences in the theoretical orienta-
tions of the therapists. Of course, we cannot as-
sume that therapists in each condition were ho-
mogeneous with respect to how they understood
and were influenced by the theoretical orientation
of that treatment; however, given that therapists
were allowed to choose their treatment condition
and demonstrated adherence to the model they
chose, it is reasonable to consider ways in which
theoretical orientation may have impacted thera-
pists’ responses to alliance ruptures. It is also
important to note that the therapists in this study
were relatively inexperienced. Novice therapists
may be particularly apt to adhere closely to the
theoretical model in which they are being trained,
in contrast to more experienced therapists, who
have been found to become less orthodox and
more integrative over time (Goldfried, 2001;
Goldfried, Raue, & Castonguay, 1998).
How might differences in therapists’ theoreti-
cal lenses lead CBT therapists to report fewer
ruptures and provide more positive assessments
of therapy process than BRT and STDP thera-
pists? BRT therapists are trained to detect rup-
tures and to address them by communicating
openly about what is transpiring between patient
Alliance Ruptures and Resolution
243

Page 12

and therapist. BRT therapists are also taught that
ruptures are an inevitable part of therapy, and that
they provide an important opportunity to better
understand and address patients’ maladaptive re-
lational schemas. Given this perspective on rup-
tures, one would expect that BRT therapists
would be particularly aware and tolerant of rup-
tures. Similarly, STDP therapists’ training in the
importance of transference interpretations may
increase their attention to tensions or strains in
the therapy relationship. Furthermore, for both
BRT and STDP therapists, greater awareness of
ruptures and a belief in the value of addressing
them in the session could inadvertently lead to
more ruptures: by drawing the patient’s attention
to tensions or problems in the patient-therapist
interaction, therapists might increase the patient’s
(and the therapist’s) anxiety and self-
consciousness, which in turn could increase the
likelihood of further misunderstandings or
strains. In contrast, the emphasis in CBT on col-
laboration may lead CBT therapists to focus on
areas of agreement rather than highlighting mo-
ments of tension or strain. As Messer and Wino-
kur (1980) suggested, CBT therapists aim to es-
tablish a “facilitative emotional climate” and
thereby “ensure that things run along smoothly”
(pp. 820–821).
This study found a moderator effect for treat-
ment condition, with lower patient-reported rup-
ture intensity being associated with greater ses-
sion depth and more symptom reduction in CBT,
but not in BRT or STDP. This finding suggests
that success in CBT was associated with a re-
duced intensity of ruptures. The lack of such a
correlation in the other two treatments suggests
that the relationship between rupture intensity
and outcome was more complex, with some rup-
tures possibly contributing positively to treatment
process and outcome. Analyses of rupture reso-
lution also revealed a treatment effect: higher
therapist ratings of resolution were associated
with therapist report of greater session smooth-
ness for patients in CBT and BRT. This might
reflect the particular relief that therapists in these
conditions likely experienced when ruptures were
resolved—for CBT therapists, relief that obsta-
cles were overcome, and for BRT therapists, re-
lief that they had successfully navigated the cen-
tral challenge of the treatment. When evaluating
differences between the three conditions with re-
spect to ruptures and resolution processes, it is
important to bear in mind that outcome among
the three treatments did not differ, with the
exception of the attrition rate, where BRT was
more successful than STDP at retaining pa-
tients in treatment. Thus, despite the finding of
therapeutic equivalence, rupture intensity ap-
parently had different implications for treat-
ment process in CBT, as compared to the two
psychoanalytically oriented treatments that
placed greater emphasis on the importance of
exploring the therapeutic relationship.
Although they demonstrated meaningful rela-
tionships to standard measures of process and
outcome, a limitation of the rupture and resolu-
tion indices used in this study is the fact that they
are single-item measures. Traditionally, psy-
chometrists have discouraged the use of single-
item measures, citing concerns that individual
items usually correlate poorly with the construct
in question and are unreliable (e.g., Nunnally &
Bernstein, 1994). However, Wanous, Reichers,
and Hudy (1997) argued that single-item mea-
sures can be appropriate for measuring psycho-
logical constructs of moderate complexity.
Single-item measures are more efficient, have
greater face validity, and place less of a burden
on participants who may be annoyed by multiple-
item measures that seem tedious and repetitious.
Recent efforts to develop single-item measures
have demonstrated that they can be psychomet-
rically sound. For example, Robins, Hendin, and
Trzesniewski (2001) demonstrated reliability and
convergent validity for a single-item measure of
self-esteem. Similarly, Zimmerman et al. (2006)
found high reliability, and convergent and dis-
criminant validity for single-item measures of
symptom severity, psychosocial functioning, and
quality of life for patients with depression.
Another limitation of the rupture and resolu-
tion indices is that they are self-report measures.
Patients and therapists may be reluctant to recog-
nize ruptures in some instances. Patients, partic-
ularly some Cluster C patients with dependent
traits, may avoid or deny problems in the rela-
tionship because they threaten the patient’s hopes
that the therapist can help him or her. Therapists,
who wish to be successful, and who might be
particularly self-conscious about their perfor-
mance because they are being videotaped for a
research study, may also be tempted to overlook
or deny certain problems in the therapeutic inter-
action. Even when patients and therapists do rec-
ognize that a rupture has occurred, they may be
reluctant to report it. A pilot study in our research
Muran et al.
244

Page 13

program found that patient failure to complete the
entire postsession questionnaire was a better pre-
dictor of dropout than patient scores on the WAI
(Samstag et al., 1998). If patients who are dissat-
isfied with treatment, and perhaps experiencing
ruptures, are less likely to complete postsession
questionnaires, then our findings likely underes-
timate the frequency of ruptures. Patient reluc-
tance to complete questionnaires is also another
possible explanation of the finding that patients
reported fewer ruptures than therapists, discussed
above. Findings from studies using observer mea-
sures of ruptures suggest that relying on patient
and therapist reports leads to an underestimation
of the frequency of ruptures. Using transcripts of
therapy sessions of five patients in psychody-
namic therapy, Sommerfeld, Orbach, Zim, and
Mikulincer (2008) found that 77% of sessions
included at least one rupture marker. A prelimi-
nary analysis of an observer-based measure for
coding ruptures based on videotapes of 30 ther-
apy sessions drawn from a sample of 10 CBT
cases found that every session contained at least
one rupture marker (Eubanks-Carter, Muran, Sa-
fran, & Mitchell, 2008).
Additional potential limitations of this study
include the exclusive focus on ruptures in early
sessions as opposed to later in treatment. Future
research should look at differences in the nature
and predictive value of early and later ruptures.
The decision to average the rupture and resolu-
tion indices across the six sessions, and the re-
duction of the outcome measures to two factors,
reduced the complexity of the data and afforded
some statistical advantages, but also could have
resulted in findings that may be different from
those derived from multilevel, multivariate pro-
cedures that eschew averaging as we did. In ad-
dition to employing statistical methods for nested
designs, future research should also examine dif-
ferences between our direct self-report measures
and indirect measurement based on repeated
postsession ratings of the therapeutic alliance
(Stiles et al., 2004; Strauss et al., 2006). For
example, in a preliminary analysis of the full
30-session protocols of 20 CBT cases (Muran,
Safran, Gorman, Eubanks-Carter, & Banthin,
2008), our research group found that patients
reported ruptures in 11.2% of the sessions and in
60% of the cases. By contrast, indirect measure-
ment of ruptures, based on analysis of patients’
alliance (WAI) ratings using control charting,
identified ruptures in only 8.67% of the sessions,
but in 100% of the cases. Closer examination of
the differences between direct and indirect as-
sessments will increase our understanding of
when and how patients and therapists determine
and report that a rupture has occurred, and the
relationship between ruptures and patients’ and
therapists’ perceptions of the alliance.
Although this study was embedded in an out-
patient mental health clinic in a general hospital
setting and focused on a comorbid patient popu-
lation, the trappings of a psychotherapy research
program, including the completion of question-
naires, videotaping, and supervision, may dimin-
ish the generalizability of our findings to more
naturalistic settings. The fact that the treatment
was time-limited introduces artificial constraints
that limit ecological validity and may have also
impacted the reporting of ruptures. For example,
a 30-session protocol might seem brief for some
patients (and therapists); a sense of limited time
could lead to increased pressure on both members
of the dyad to solve the patient’s problems, which
could contribute to more ruptures. Conversely,
knowledge that the treatment will end at session
30 might lead some patients and therapists to
avoid challenging issues that seem too complex
to deal with adequately in a short time; this could
contribute to more efforts to avoid, minimize, or
deny problems in the alliance.
Other factors that may limit generalizability
include the lack of racial and ethnic diversity
among patients and therapists and the focus on
Cluster C and Personality Disorder NOS diag-
noses, although these are the most prevalent of
the personality disorder classifications (Mattia &
Zimmerman, 2001). Additional limitations in-
clude our modest sample size and sporadic miss-
ing data that limited the power necessary to test
interactions and to assess the strength of effects
within each treatment condition. Finally, the find-
ings should be interpreted in light of the fact that
18 of the most difficult patients were screened out
of the CBT and STDP conditions for another
study (Safran et al., 2005).
By linking ruptures and resolution processes to
process and outcome in three different time-
limited treatments for personality disorders, this
study demonstrated the relevance and importance
of ruptures across different theoretical orienta-
tions. This study also highlighted the challenges
of identifying ruptures, given differences be-
tween patient, therapist, and observer perspec-
tives, and potential differences in the role of
Alliance Ruptures and Resolution
245

Page 14

ruptures across various psychotherapy treat-
ments. Alliance ruptures in the first few ses-
sions of treatment are complex phenomena:
they can serve as early warning indicators of
problems in the therapeutic relationship, but
they can also provide opportunities for the em-
ployment of resolution processes that may help
to strengthen the alliance and retain challeng-
ing patients in treatment. Identifying and dis-
seminating effective rupture resolution strate-
gies is an important next step for improving the
effectiveness of psychotherapy.
References
American Psychiatric Association. (1994). Diagnostic
and statistical manual of mental disorders (4th ed.).
Washington, DC: Author.
BARRETT, M. S., CHUA, W.-J., CRITS-CHRISTOPH, P., GIB-
BONS, M. B., & THOMPSON, D. (2008). Early withdrawal
from mental health treatment: Implications for psycho-
therapy practice. Psychotherapy: Theory, Research,
Practice, Training, 45, 247–267.
BATTLE, C. C., IMBER, S. D., HOEHN-SARIC, R., STONE,
A. R., NASH, E. R., & FRANK, J. D. (1966). Target
complaints as criteria of improvement. American Jour-
nal of Psychotherapy, 20, 184–192.
BECK, A. T., RUSH, A. J., SHAW, B. F., & EMERY, G.
(1979). Cognitive therapy of depression. New York:
Guilford Press.
BENJAMIN, L. S. (1993). Diagnosis and treatment of per-
sonality disorders: A structural approach. New York:
Guilford Press.
BENJAMIN, L. S., & KARPIAK, C. P. (2001). Personality
disorders. Psychotherapy, 38, 487–491.
CLARKIN, J. F., & LEVY, K. N. (2004). The influence of
client variables in psychotherapy. In M. J. Lambert
(Ed.), Bergin and Garfield’s handbook of psychother-
apy and behavior change (5th ed., pp. 194–226). New
York: Wiley.
COADY, N. (1991). The association between client and
therapist interpersonal processes and outcomes in psy-
chodynamic psychotherapy. Research on Social Work
Practice, 1, 122–138.
DEROGATIS, L. R. (1983). SCL-90-R: Administration,
scoring, & procedures manual II. Towson, MD: Clinical
Psychiatric Research.
ENDICOTT, J., SPITZER, R. L., FLEISS, J. L., & COHEN, J.
(1976). The Global Assessment Scale: A procedure for
measuring overall severity of psychiatric disturbance.
Archives of General Psychiatry, 33, 766–771.
EUBANKS-CARTER, C., MURAN, J. C., SAFRAN, J. D., &
MITCHELL, A. (2008, June). Development of an
observer-based rupture resolution rating system. In
J. C. Muran (Chair), Recent developments in rupture
resolution research. Panel conducted at the annual
meeting of the Society for Psychotherapy Research,
Barcelona, Spain.
FIRST, M. B., SPITZER, R. L., GIBBON, M., & WILLIAMS,
J. B. W. (1995). Structured Clinical Interview for DSM–
IV. New York: Biometrics Research Department, New
York Psychiatric Institute.
FITZPATRICK, M. R., IWAKABE, S., & STALIKAS, A. (2005).
Perspective divergence in the working alliance. Psycho-
therapy Research, 15, 69–79.
GARFIELD, S. L. (1994). Research on client variables in
psychotherapy. In A. E. Bergin & S. L. Garfield (Eds.),
Handbook of psychotherapy and behavior change (4th
ed., pp. 190–228). New York: Wiley.
GOLDFRIED, M. R. (2001). How therapists change: Per-
sonal and professional reflections. Washington, DC:
American Psychological Association.
GOLDFRIED, M. R., RAUE, P. J., & CASTONGUAY, L. G.
(1998). The therapeutic focus in significant sessions of
master therapists: A comparison of cognitive-
behavioral and psychodynamic-interpersonal interven-
tions. Journal of Consulting and Clinical Psychology,
66, 803–810.
HATCHER, R. L., BARENDS, A., HANSELL, J., & GUT-
FREUND, M. J. (1995). Patients’ and therapists’ shared
and unique views of the therapeutic alliance: An inves-
tigation using confirmatory factor analysis in a nested
design. Journal of Consulting and Clinical Psychology,
63, 636–643.
HENRY, W. P., SCHACHT, T. E., & STRUPP, H. H. (1986).
Structural analysis of social behavior: Application to a
study of interpersonal process in differential psycho-
therapeutic outcome. Journal of Consulting and Clini-
cal Psychology, 54, 27–31.
HOROWITZ, L. M., ALDEN, L. E., WIGGINS, J. S., & PIN-
CUS, A. L. (2000). Inventory of Interpersonal Problems.
San Antonio, TX: Psychological Corporation.
HORVATH, A. O. (2000). The therapeutic relationship:
From transference to alliance. Journal of Clinical Psy-
chology, 56, 163–173.
HORVATH, A. O., & GREENBERG, L. S. (1989). Develop-
ment and validation of the Working Alliance Inven-
tory. Journal of Counseling Psychology, 36, 223–233.
HORVATH, A. O., & SYMONDS, B. D. (1991). Relation
between working alliance and outcome in psychother-
apy: A meta-analysis. Journal of Counseling Psychol-
ogy, 38, 139–149.
KEOWN, P., HOLLOWAY, F., & KUIPERS, E. (2002). The
prevalence of personality disorders, psychotic disorders
and affective disorders amongst the patients seen by a
community mental health team in London. Social Psy-
chiatry and Psychiatry Epidemiology, 37, 225–229.
KESSLER, R. C., MCGONAGLE, K. A., ZHAO, S., NELSON,
C. B., HUGHES, M., ESHLEMAN, S., ET AL. (1994). Life-
time and 12-month prevalence of DSM–III–R psychi-
atric disorders in the United States: Results from the
National Comorbidity Study. Archives of General Psy-
chiatry, 51, 8–19.
KIESLER, D. J. (1996). Contemporary interpersonal theory
and research: Personality, psychopathology, and psy-
chotherapy. New York: Wiley.
KIVLIGHAN, D. M., & SHAUGHNESSY, P. (1995). Analysis
of the development of the working alliance using hier-
archical linear modeling. Journal of Counseling Psy-
chology, 42, 338–349.
KIVLIGHAN, D. M., & SHAUGHNESSY, P. (2000). Patterns
of working alliance development: A typology of client’s
working alliance ratings. Journal of Counseling Psy-
chology, 47, 362–371.
Muran et al.
246

Page 15

KLEIN, M. H., BENJAMIN, L. S., ROSENFELD, R., TREECE,
C., HUSTED, J., & GREIST, J. H. (1993). The Wisconsin
Personality Disorders Inventory: Development, reli-
ability, and validity. Journal of Personality Disorders, 7,
285–303.
LAMBERT, M. J. (2004). Bergin & Garfield’s handbook of
psychotherapy & behavior change (5th ed.). New York:
Wiley.
LIVESLEY, W. J., (ED.). (2001). Handbook of personality
disorders: Theory, research, and treatment. New York:
Guilford Press.
LUBORSKY, L. (1984). Principles of psychoanalytic psy-
chotherapy: A manual for supportive-expressive treat-
ment. New York: Basic Books.
MALLINCKRODT, B., & NELSON, M. L. (1991). Counselor
training level and the formation of the psychotherapeu-
tic working alliance. Journal of Counseling Psychology,
38, 133–138.
MARTIN, D. J., GARSKE, J. P., & DAVIS, M. K. (2000).
Relation of the therapeutic alliance with outcome and
other variables: A meta-analytic review. Journal of
Consulting and Clinical Psychology, 68, 438–450.
MATTIA, J. I., & ZIMMERMAN, M. (2001). Epidemiology.
In W. J. Livesley (Ed.), Handbook of personality dis-
orders: Theory, research, and treatment. (pp. 107–123).
New York: Guilford Press.
MESSER, S. B., & WINOKUR, M. (1980). Some limits to the
integration of psychoanalytic and behavior therapy.
American Psychologist, 35, 818–827.
MILLON, T., & DAVIS, R. O. (1996). Disorders of person-
ality: DSM–IV and beyond (2nd ed.). New York: Wiley.
MITCHELL, S. A., & ARON, L. (1999). Relational psycho-
analysis: The emergence of a tradition. Hillsdale, NJ:
Analytic Press.
MURAN, J. C. (2002). A relational approach to under-
standing change: Plurality and contextualism in a psy-
chotherapy research program. Psychotherapy Research,
12, 113–138.
MURAN, J. C., & SAFRAN, J. D. (2002). A relational
approach to psychotherapy: Resolving ruptures in the
therapeutic alliance. In F. W. Kaslow (Ed.), Compre-
hensive handbook of psychotherapy (pp. 253–281). New
York: Wiley.
MURAN, J. C., SAFRAN, J. D., GORMAN, B. S., EUBANKS-
CARTER, C., & BANTHIN, D. (2008, June). Identifying
ruptures & their resolution from postsession self-report
measures. In J. C. Muran (Chair), Recent developments
in rupture resolution research. Panel conducted at the
annual meeting of the Society for Psychotherapy Re-
search, Barcelona, Spain.
MURAN, J. C., SAFRAN, J. D., SAMSTAG, L. W., & WIN-
STON, A. (1992). Patient and therapist postsession ques-
tionnaires, Version 1992. Beth Israel Medical Center,
NY.
MURAN, J. C., SAFRAN, J. D., SAMSTAG, L. W., & WIN-
STON, A. (2005). Evaluating an alliance-focused treat-
ment for personality disorders. Psychotherapy, 42, 532–
545.
MURAN, J. C., SEGAL, Z. V., SAMSTAG, L. W., & CRAW-
FORD, C. (1994). Patient pretreatment interpersonal
problems and therapeutic alliance in short-term cogni-
tive therapy. Journal of Consulting and Clinical Psy-
chology, 62,185–190.
NUNNALLY, J. C., & BERNSTEIN, I. H. (1994). Psychomet-
ric theory (3rd ed.). New York: McGraw-Hill.
PATTON, M. J., KIVLIGHAN, D. M., & MULTON, K. D.
(1997). The Missouri Psychoanalytic Counseling Re-
search Project: Relation of changes in counseling pro-
cess to client outcomes. Journal of Counseling Psychol-
ogy, 44, 189–208.
PERSONS, J. B. (1989). Cognitive therapy in practice: A
case formulation approach. New York: Norton.
PIPER, W. E., & JOYCE, A. S. (2001). Psychosocial treat-
ment outcome. In W. J. Livesley (Ed.), Handbook of
personality disorders: Theory, research, and treatment.
(pp. 323–343). New York: Guilford Press.
POLLACK, J., FLEGENHEIMER, W., KAUFMAN, J., & SA-
DOW, J. (1992). Brief adaptive psychotherapy for per-
sonality disorders: A treatment manual. San Diego, CA:
Social & Behavioral Documents.
ROBINS, R. W., HENDIN, H. M., & TRZESNIEWSKI, K. H.
(2001). Measuring global self-esteem: Construct valida-
tion of a single-item measure and the Rosenberg Self-
Esteem Scale. Personality and Social Psychology Bul-
letin, 27, 151–161.
SAFRAN, J. D. (2002). Brief relational psychoanalytic
treatment. Psychoanalytic Dialogues, 12, 171–195.
SAFRAN, J. D., & MURAN, J. C. (1996). The resolution of
ruptures in the therapeutic alliance. Journal of Consult-
ing and Clinical Psychology, 64, 447–458.
SAFRAN, J. D., & MURAN, J. C. (2000). Negotiating the
therapeutic alliance: A relational treatment guide. New
York: Guilford Press.
SAFRAN, J. D., MURAN, J. C., SAMSTAG, L. W., &
STEVENS, C. (2002). Repairing alliance ruptures. In J. C.
Norcross (Ed.), Psychotherapy relationships that work:
Therapist contributions and responsiveness to patients
(pp. 235–254). New York: Oxford University Press.
SAFRAN, J. D., MURAN, J. C., SAMSTAG, L. W., & WIN-
STON, A. (2005). Evaluating an alliance-focused treat-
ment for potential treatment failures. Psychotherapy,
42, 512–532.
SAMSTAG, L. W., BATCHELDER, S. T., MURAN, J. C.,
SAFRAN, J. D., & WINSTON, A. (1998). Early identifica-
tion of treatment failures in short-term psychotherapy:
An assessment of therapeutic alliance and interper-
sonal behavior. Journal of Psychotherapy Practice &
Research, 7, 126–143.
SAMSTAG, L. W., MURAN, J. C., WACHTEL, P. L., SLADE,
A., SAFRAN, J. D., & WINSTON, A. (2008). Evaluating
negative process: A comparison of working alliance,
interpersonal behavior, and narrative coherency among
three psychotherapy outcome conditions. American
Journal of Psychotherapy, 62, 165–194.
SHEA, T. M. (1997). Core battery conference: Assessment
of change in personality disorders. In H. H. Strupp,
L. M. Horowitz, & M. J. Lambert (Eds.), Measuring
patient changes in mood, anxiety, and personality disor-
ders: Toward a core battery (pp. 389–400). Washington,
DC: American Psychological Association.
SLEDGE, W. H., MORAS, K., HARTLEY, D., & LEVINE, M.
(1990). Effect of time-limited psychotherapy on patient
dropout rates. American Journal of Psychiatry, 147,
1341–1347.
SOMMERFELD, E., ORBACH, I., ZIM, S., & MIKULINCER, M.
(2008). An in-session exploration of ruptures in work-
ing alliance and their associations with clients’ core
Alliance Ruptures and Resolution
247