Cognitive and Behavioral Practice 14 (2007) 2635

www.elsevier.com/locate/cabp

Causes and Management of Treatment-Resistant Panic Disorder and

Agoraphobia: A Survey of Expert Therapists
William C. Sanderson, Hofstra University
Timothy J. Bruce, University of Illinois College of Medicine
Cognitive behavior therapy (CBT) is recognized as an effective psychological treatment for panic disorder (PD). Despite its efficacy, some
clients do not respond optimally to this treatment. Unfortunately, literatures on the prediction, prevention, and management of
suboptimal response are not well developed. Considering this lack of empirical guidance, we decided that it would be useful to survey
expert cognitive behavioral therapists about what they have found in their practices to contribute to a poor treatment response and what
strategies they have found helpful in preventing or managing these problems. Ten factors associated with suboptimal responding
emerged. Listed in order of reported frequency, they were as follows: lack of engagement in behavioral experiments, noncompliance,
comorbidity, inadequate case formulation/misdiagnosis, secondary gain, problems with cognitive restructuring, presence of other
negative life events, medication complications, poor delivery of treatment, and therapeutic relationship barriers. The current paper
discusses these factors and details treatment suggestions to improve outcome provided by the survey participants.

ANIC DISORDER (PD) is a distressing and disabling

anxiety disorder characterized by an onset of recurrent unexpected panic attacks. Panic attacks involve a
sudden rush of intense fear that is accompanied by a
variety of physical (e.g., palpitations, dizziness, sweating)
and cognitive (i.e., fear of dying, losing control, or going
crazy) symptoms (American Psychiatric Association,
2000). Clients with PD fear subsequent attacks and
become preoccupied with potential catastrophic consequences of panic attacks (e.g., the panic attack will cause
a heart attack, stroke, fainting, loss of control). Many
clients suffering from PD develop agoraphobia, which
refers to fear and/or avoidance of activities or situations
that they believe will provoke an attack, where escape may
be difficult (e.g., airplanes, elevators, trains), or where
help may be unavailable in the event of a panic attack
(e.g., being at home alone, in an airplane, far from home).
The severity of panic attacks and agoraphobia can range
from multiple daily attacks and house-boundness to
infrequent attacks and endurance of feared situations
with discomfort, respectively.
Cognitive behavior therapy (CBT) is well established as
an effective psychological treatment for PD. It is a first-line
treatment option according to guidelines of best practice
(cf. American Psychiatric Association, 1998). Although
there are several different CBT packages for PD

1077-7229/07/2635$1.00/0
2006 Association for Behavioral and Cognitive Therapies.
Published by Elsevier Ltd. All rights reserved.

(cf. Margraf, Barlow, Clark, & Telch, 1993), most CBT

treatments include the following components:
psychoeducation about PD and CBT;
panic management strategies such as relaxation and
breathing;
cognitive restructuring of fear-based thought content
and processes;
exposure to feared bodily sensations (interoceptive
exposure);
exposure to feared situations (exteroceptive exposure)
For most clients undergoing treatment, CBT has been
shown to reduce panic attacks, generalized anxiety, agoraphobic avoidance, and depression (e.g., Barlow, Gorman,
Shear, & Woods, 2000). Although results across studies
vary slightly, most show that CBT results in a panic-free
rate of approximately 75% to 90% (Barlow, Raffa, &
Cohen, 2002).
Despite the efficacy of CBT for PD, some clients show
a suboptimal response to it in that they either do not
respond or respond only partially (Rosenbaum, Pollack,
& Pollack, 1996). The literatures on the prediction,
prevention, and management of suboptimal treatment
response are not advanced. Studies of factors associated
with poor outcome were recently reviewed by McCabe
and Antony (2005), who identified three factors with
consistent support: symptom severity, comorbid depression, and a comorbid personality disorder. Although this
information is useful to the practicing clinician in

Panic Disorder
anticipating potential challenges in treatment, empirical
guidance on the types of problems commonly encountered that lead to suboptimal response and how they
might be prevented or managed is lacking. In the
absence of this guidance, we decided that it would be
useful to ask expert clinicians what they have found to
contribute to poor treatment outcome and what
strategies they have found to be useful in the prevention
and management of the problems they see. A similar
method was employed by Scott, Pollack, Otto, Simon, &
Worthington (1999) to evaluate psychiatrists response
to treatment-refractory PD when using pharmacological
interventions.

Method
Participants
Participants were members of the Association for
Behavioral and Cognitive Therapies (ABCT) who were
selected and invited by the authors to complete an on-line
survey, and who volunteered to participate. Participants
were aware that they would be acknowledged in the
manuscript. To identify experts, both authors independently reviewed the membership list of ABCT and
identified individuals who made significant contributions
to the study or application of CBT for panic disorder. The
authors compared their lists and only individuals who
were identified by both authors were included. Of 30
members who were invited, 20 participated (see the
Appendix for a list of participants).
Procedure
The survey involved two questions to which participants were asked to produce brief answers. Specifically,
invited members were emailed the following message:
We have been invited to write a paper on cognitive
behavioral approaches to treatment resistant panic
disorder. As part of this effort, we are conducting a
brief survey of identified experts in the field. We
hope you agree that empirically informed clinicians, such as the readership of this journal, will
value the opinion of experts in this area, where
direct empirical guidance is lacking. We are asking
if you would give us a few minutes of your time to
answer two brief questions about how you have
come to approach the problem. We will identify
and acknowledge the assistance of all contributing
experts.
By treatment resistant, we generally mean a client
who, in response to conventional CBT for panic
disorder (i.e., psychoeducation, relaxation, breathing retraining, cognitive restructuring, sensation
and situation exposure), continues to exhibit
clinically distressing or disabling features of the

disorder (e.g., panic attacks, agoraphobic avoidance, concern over future attacks, change in activities, avoidance of physical sensations), or otherwise
shows incomplete progress. We are interested in
what your experience with treatment resistance has
been, and how you have come to approach it.
1. In your experience, what have been the primary
reasons that some clients have had incomplete
responses to conventional CBT for panic disorder
(top 3 reasons or fewer)?
2. Would you please briefly explain how you believe
each of the above problems is best approached
therapeutically?
In some literatures (e.g., treatment of infectious
disease), the term treatment resistance has been
defined more narrowly than we did for this survey.
The more narrow definition refers to instances in which
treatment is delivered as intended (i.e., with good
treatment fidelity), received by the client (i.e., compliance is confirmed), but nonetheless results in a poorerthan-expected response. We intentionally broadened the
definition to include any factor that the survey taker
thought accounted for a poorer-then-expected response.
We did this to get a sense of what experienced
therapists commonly encounter in their day-to-day
practices. Although this broader definition is also
termed treatment resistance, we use the terms
suboptimal or incomplete response interchangeably
because they may more accurately reflect what we
assessed.
Analyses
Question 1: In your experience, what have been the primary
reasons that some clients have had incomplete responses to
conventional CBT for panic disorder (top 3 reasons or fewer)?
Answers to Question 1 were listed verbatim for each
participant along with their ranking. The authors then
categorized answers independently. One of us identified
10 categories. The other identified 11 categories, 10 of
which overlapped with the other authors 10. The
remaining category was integrated into an existing one
(i.e., Problems With Cognitive Restructuring), leaving
10 categories of cited reasons for treatment resistance.
The categories were then ranked as follows: A category
ranked 1 by an author received 3 points, a rank of 2
received 2 points, and a rank of 3 received 1 point.
Ranks were summed for each category across respondents. Categories were then ranked from highest to lowest
total points. Alternative methods of ranking did not
change the order.
Question 2: Would you please briefly explain how you believe
each of the above problems is best approached therapeutically? For
each cause of treatment resistance cited, participants were

27

Sanderson & Bruce

28

asked to describe how they have come to approach it

therapeutically. Those responses were listed verbatim for
each problem across participants. Redundancies were
removed, yielding the core recommendations reported
later in this paper.

Results
Table 1 shows the cited causes for treatment resistance
in rank order, as well as the total number of points for
each resulting from the ranking procedure. Four tiers of
causes emerged based on the natural division evident in
the total scores. Lack of Engagement in Behavioral
Experiments was far and away the highest ranked
category with 35 total points. It was cited by more
participants than any other reason (12 of 20), and was
ranked as the top cause of treatment resistance by 9 of
the 12 respondents. Noncompliance ranked a clear
second with 20 points. Eight respondents cited it, 2 of
whom ranked it as a 1. Causes 3 through 6 were
approximately equal in rank, with total rank points
ranging from 10 to 11. They were cited by an average of 5
respondents and ranked as either a 2 or 3 by each
participant. Reasons 7 through 10 were approximately
equal in rank, ranging from 1 to 3 total points. One or 2
respondents cited these and ranked them as a 2 or 3.
Factors Associated with Suboptimal Treatment and
Recommendations
Lack of engagement in behavioral experiments. In CBT for
PD, clients are asked to face and challenge their fears,
mostly through exposure to sensations or situations that
they believe will bring about some feared consequence (cf.
Barlow & Craske, 2000). The unwillingness of clients,
intentional or not, to engage themselves fully in these
behavioral experiments and risk the consequences they
fear was the most frequently cited cause for a suboptimal
response to treatment. Many respondents described this in
terms of subtle avoidance behavior. Examples included
doing only part of an exposure exercise, provoking only
less-feared sensations as opposed to pushing oneself
Table 1
Rank order of reasons cited for treatment resistance
Rank

Cause of treatment resistance (Total Ranking Points )

1
2
3
4
4

Lack of engagement in behavioral experiments (35 pts.)

Noncompliance (20 pts.)
Comorbidity (11 pts.)
Inadequate case formulation or misdiagnosis (10 pts.)
External support of PDA behavior (secondary gain, fear of
disruption) (10 pts.)
Problems with cognitive restructuring (10 pts.)
Presence of other negative life events (3 pts.)
Medication complications (2 pts.)
Poor delivery of CBT (2 pts.)
Therapeutic relationship barriers (1 pt.)

4
7
8
8
10

further, using breathing or relaxation to prevent feared

sensations from emerging, and using various forms of
distraction to minimize anxiety. These various forms of
avoidance were thought to preclude valid testing, restructuring, and eventual extinction of fears. Commonly cited
examples of fears that clients were reluctant to test
included losing control, embarrassing oneself, or simply
not wanting to experience the negative affect engendered
by the experiment for various or sometimes unknown
reasons.
Suggested recommendations for addressing this obstacle were as follows:
1. Education: Use initial and ongoing psychoeducation
that emphasizes the importance of engagement as a
crucial goal of therapy.
2. Graduated exposure: Use individualized graduated
exposure tasks as a means to ease wary clients into full
engagement.
3. Directly observe and assess avoidance: Try to observe
the clients fearful behavior, as opposed to relying
solely on self-report, during initial assessment (e.g., a
behavioral avoidance test) and during initial and other
critical exposures to identify and confront obstacles to
engagement.
4. Consider using motivational enhancement techniques:
Motivational enhancement techniques include validation of the clients particular stage of change through
expression of empathy, identifying discrepancies
between the clients goals and problem behaviors
(e.g., avoidance), rolling with resistance by emphasizing personal control and approaching treatment as an
experiment, and encouraging self-efficacy by focusing
on personal strengths and highlighting positives.
5. Encourage acceptance of negative feelings: Reframe
negative feelings as tolerable, acceptable, and necessary to reduce the clients avoidance (e.g., no pain, no
gain). Use education and support to try to help clients
better accept the experience of negative affect as
unpleasant but not dangerous. Normalize the experience, suggesting that the client take a monitors
(observers) perspective, and encourage the client to
see exposure as an opportunity for building tolerance
of negative affect.
6. Avoid concluding that the client is resistant: Ironically, the label treatment resistant can be its own
obstacle when clients who are not engaging in therapy
are thought of as nonresponders. Therapists may not
be as motivated to work with the client, believing his or
her effort will be futile, and risk creating a self-fulfilling
prophecy. Although it is possible that some clients may
be resistant to treatment, others may merely require
strategies, such as those detailed above, to help them
overcome problems with engagement.

Panic Disorder
Noncompliance. The second most cited reason for
treatment resistance in our survey was noncompliance.
In clients suffering from anxiety disorders, noncompliance is often an expression of avoidance due to fear. The
previous factor, lack of engagement, could justifiably be
described as a form of noncompliance due to fear. We
separated engagement primarily because the respondents did. Here we have categorized all other forms of
noncompliance other than engagement. This category
almost entirely refers to instances of noncompletion of
therapeutic homework, such as not doing self-monitoring tasks, not doing an assigned BAT, not practicing
somatic strategies, not doing tasks related to identifying
or challenging cognitive biases, or not doing scheduled
exposure exercises. Although sometimes fear based, it
also includes noncompliance related to other factors
such as resource limitations, scheduling problems, or
forgetfulness.
Most of the recommendations for dealing with noncompliance aimed at preventing it. Preventative recommendations were as follows:
1. Psychoeducation: One of the most frequently
proposed solutions for this problem was to review
and expand upon psychoeducation. This was based
on the idea that noncompliance sometimes results
from the individuals lack of understanding of the
therapeutic rationale for a task and/or the essential
role that the task plays in promoting improvement.
For example, the following information would be
conveyed to the client to facilitate completion of
restructuring exercise:
The purpose of cognitive restructuring is to learn to
identify and challenge irrational anxiety-provoking
thoughts, thereby reducing anxiety. Frequently,
focusing on these thoughts results in an increase
in anxiety at the moment. There is no other way
to reduce anxiety-provoking thoughts, and they
are very unlikely to improve without using a
deliberate process to correct them such as
cognitive restructuring. If only practiced in the
session the value will be limited because it is
essential to learn to apply it in the majority of
anxiety-provoking situations that occur outside of
the session.
This information should be discussed with the client to
ensure it is understood. Similar explanations for other
treatment strategies were recommended.
2. Consider audiotaping sessions: One way to increase
adherence is to audiotape the sessions and ask the
client to listen to it at least once prior to the next

session. Some clients may benefit from the redundant

presentation of the information. Others may process
the information more clearly when merely listening
(rather than during the session when they are
interacting with the therapist). It also is a way to
apply extra pressure between sessions as they will be
reminded of the assignments while listening to the
session.
3. Increase accountability, if necessary: Some clients may
feel more compelled to engage in treatment strategies
outside the session if they are asked to be more
accountable. For some, simply coming in to the next
session and reporting what they did or did not do is
sufficient. However, for others, asking them to
complete some type of monitoring form (e.g., a record
of relaxation practice sessions, thought records) or
asking them to phone or email when they complete the
exercise may increase motivation.
Several recommendations concerned how homework
tasks where assigned, explained, and carried out, as
follows:
4. Be clear and specific: Write down exactly what the
client is expected to do and whena behavioral
prescription. For example, if a clients assignment is
to drive 2 miles from home three times in between
sessions, record the assignment and then collaborate
with the client to come up with specific days and
times this assignment can be completed (e.g.,
Monday, Wednesday, and Friday during lunch
break).
5. Anticipate obstacles with each assignment: Ask the
client questions regarding facilitators and obstacles
for example, access to the car, having sufficient time
to complete the exercise, hiring a babysitter if necessaryand generating solutions (or modifying the
assignment) to increase the likelihood of the
behavior.
6. Reduce burden: Make homework tasks as easy to carry
out as possible. For example, use checkmarks rather
than narrative or tape record responses rather than
write them, depending on the clients preference.
7. Make tasks clearly relevant to goals: Tie homework
directly to a clearly identified therapeutic goal and
explain its rationale as described previously.
8. Model tasks: When possible, demonstrate assigned
exercises to minimize compliance problems due to a
lack of understanding of the task on the clients
part.
9. Prompt the behavior in the environment and reinforce
it: Use standard behavioral practices for prompting the
behavior in the natural environment, such as making
a phone call or tying it to another high-frequency

29

30

Sanderson & Bruce

activity, and then reward it with something of the
clients choosing.
10. Review the homework: Communicating to the client
that their work is important and valued can facilitate
adherence.
When addressing existing noncompliance, nearly all
respondents recommended doing a functional analysis of
the noncompliant behavior. Some suggested contingency
contracting with persistent noncompliance.
1. Functional analysis of existing noncompliance: Try to
identify factors that are interfering with compliance
or facilitating noncompliance. Most respondents
reported that in their experience noncompliance
was usually an act of avoidance of a feared activity
and was addressed as described under Factor 1. But
other reinforcers of noncompliance were cited as
well, ranging from the subtle, such as not disturbing
existing social relationships, to the obvious, such as
maintaining disability compensation. As noted later,
immediately and directly addressing those factors as
obstacles to continuing therapy was the core
recommendation.
2. Consider contingency contracting: Three respondents suggested addressing noncompliance with
contingency contracting. Evidence from early studies
of contracting in marital therapy suggests that the
process of negotiating the contract is as critical to its
success as its quid pro quo content (Jacobson, 1978).
Recommendations were consistent with this in
emphasizing collaborative exploration of the positive
and negative consequences of compliance and
noncompliance and honestly and directly addressing
the clients motivation and goals.
Comorbidity. The presence of additional mental disorders or clinically significant symptoms (i.e., syndromal
or symptom comorbidity, respectively) was one of four
cited causes of poor treatment response ranked within the
third tier of causes. Although four respondents cited
examples of difficult comorbid anxiety disorders (e.g.,
OCD), the majority discussed comorbid depression.
Interestingly, existing data on the impact of comorbidity
on treatment outcome for clients with a principal
diagnosis of PD suggest that in many cases the presence
of another anxiety disorder does not typically diminish
the efficacy of CBT focused on the panic disorder (Brown,
Antony, & Barlow, 1995). Factors such as the type and
severity of the comorbid anxiety disorder may mediate
that relationship in individual clients. Studies of comorbid
depression, however, suggest that its presence can
interfere with successful outcome (Brown et al., 1995;
McLean, Woody, Taylor, & Koch, 1998).

Recommendations for dealing with comorbidity are

stepwise, as follows:
1. If the comorbid condition is judged severe enough to
interfere with participation in the CBT for panic (e.g.,
hopelessness precluding engagement or other compliance), then the recommendation is to address the
comorbid condition first. In other words, severity of
interference may place the comorbid condition first in
line for treatment.
2. In the more subtle case of conditions or symptoms
that are coexistent but not as interfering, the
recommendation is to treat the panic disorder first,
reevaluating it and the comorbidity along the course
of treatment, and treating residual syndromes or
symptoms next. Many comorbid disorders and symptoms have been shown to remit with successful
treatment of a primary panic disorder (Brown et al.,
1995). Residual conditions or symptoms should be
treated based on initial studies showing that the
presence of some comorbid symptoms places clients
at risk for continued or emergent problems (McLean
et al., 1998).
Inadequate case formulation or misdiagnosis. Nearly every
guideline for the treatment of nearly every health care
condition advises practitioners to revisit their diagnosis
if a client has not responded as expected (cf.
American Psychiatric Association, 1998). Also among
the third tier of cited causes for a poor treatment
response were problems with case formulation. Some
respondents couched recommendations in diagnostic
terms, others in terms of behavior analysis.
Diagnostically speaking, the decision tree for diagnosing PD requires first ruling out symptoms that are
due to the direct physiological effects of a general
medical condition (e.g., hyperthyroidism) or a substance-induced syndrome (e.g., CNS depressant withdrawal, stimulant intoxication) before making the
diagnosis of PD (American Psychiatric Association,
2000). Some of our survey respondents cited examples
in which symptoms of panic that did not respond to
CBT were found subsequently to be products of
medical conditions such as pheochromocytoma or
hyperthyroidism.
Unexpected panic attacks, the diagnostic hallmark of
PD, may occur in other mental disorders such as
posttraumatic stress disorder, generalized anxiety disorder, or depression. Some respondents recounted poor
treatment responders who later were found to be better
described by a different anxiety or mood disorder. For
instance, one respondent described a client who reported
unexpected panic attacks in situations where she felt
alone. The client was initially diagnosed with PD and

Panic Disorder
agoraphobia. The fear of being alone was seen as
reflecting agoraphobic safety and efficacy concerns.
Although benefiting somewhat from exposure to these
situations, her improvement was less than expected. The
client later disclosed that the fear of being alone began
after a sexual assault that she had been unwilling to
disclose earlier. When treatment was oriented around this
context, her response to it improved. This case highlights
therapeutic trust issues as well as diagnostic consequences. Trust issues also appear under Factor 10 later
in the paper.
In behavior analytic terms, the most common error
cited was the misidentification or underestimation of
triggers of panic, resulting in exposures that did not target
the most significant stimuli.
In their recommendations, respondents again
emphasized preventive measures, but recommended
revisiting the initial assessment when nonresponse is
evident:
1. Rule out medical and substance etiology: Require
certain clients to undergo medical and substance
clearance prior to treatment.
2. Assess triggers: Do a thorough and ongoing assessment
of triggers of fear.
3. Educate clients regarding triggers: Place a strong
emphasis on educating the client about triggers and
on identifying them throughout treatment. Assessing
several examples of the sequencing of the clients
thoughts, sensations, and reactions that spiral into
panic may help identify key themes and triggers to
target through exposure.
All of these recommendations are consistent with the
CBT principle of continuing assessment throughout
therapy and adjusting the focus of interventions as
needed.
External support of PD behavior (secondary gain, fear of
disruption). Situations in which the fear and avoidance
of PD are being positively or negatively reinforced
(e.g., disability check, not having to work) or where
overcoming them would cost the client something
personally important (i.e., spousal attention; sympathy
from others) were cited by some respondents as
barriers to improvement. No one cited malingering
as an example of external support; rather, these factors
were seen as obstacles to improvement of real
conditions. Examples included the loss of social
attention, social pressure to remain dependent, and
loss of disability payments, all creating an incentive to
remain ill. In some cases, clients were described as
losing their confidence in being able to be return to
the demands of the roles and responsibilities of work
or other demands.

Recommendations to address this issue were as follows:

1. Conduct a functional assessment: Identify the functional relationship between improvement and its
consequences. Often the factors supporting nonimprovement involve fear and thus nonimprovement
is avoidance. Consider developing a list of pros and
cons of improvement.
2. Problem-solve obstacles: Problem-solve anticipated
changes and problems by specifically identifying
them, generating options for coping, and developing
a plan based on the best options. Gradual exposure
may be needed as well. For example, a first goal may be
to go back to work part-time, then half-time, and then
full-time.
3. Directly address gain as a potential obstacle to
further treatment: Regardless of the factors supporting
nonimprovement, one unanimous recommendation
was to address it directly with the client toward the goal
of deciding whether continued therapy is desired (i.e.,
does the client really want what comes along with
successful treatment?).
Problems with cognitive restructuring. In broad strokes,
cognitive restructuring in the treatment of PD involves
elucidating what the client finds threatening, the biases
in that appraisal of threat, and the generation of
alternative appraisals that correct for those biases.
These identified fears and alternative appraisals are
usually framed as predictions and are tested through
behavioral experiments. The clinician attempts to
make those tests valid to the clientif not at first,
then eventually. Repeated disconfirmation of biased
fear-based predictions helps the client shift from
fearful appraisals to ones more in line with the actual
threat posed by the feared stimulus (e.g., Clark et al.,
1994). Difficulties with cognitive restructuring also
ranked fourth among reasons for a poor treatment
response. The nature of the problems that were cited
varied.
Several respondents noted that they have seen a
poor response when key and often subtle fears are not
elucidated. More than one suggested that this is likely
to occur when assessment is rushed and when clients
are asked to endorse from a list (such as a
questionnaire) rather than produce fearful cognitions.
Other respondents noted that although some fears are
identifiable and testable (e.g., heart attack, going
crazy, embarrassing oneself), others were seen as
more difficult to define and convert into testable
predictions. Examples cited included fading into
nothingness and losing oneself. Still others identified validity problems, most notably difficulty finding
valid alternative appraisals that correct for fearful

31

32

Sanderson & Bruce

biases and the failure of interoceptive exposure
exercises to produce sensations that the client finds
valid. Points made were that some clients expect to
see a shift in the believability of alternatives too
quickly into therapy and that this demoralizes or
otherwise disengages them. Similarly, some clients do
not believe that managing the sensations produced by
interoceptive exercises is similar to managing those encountered during a panic attack, resulting in motivational
problems.
For difficulties elucidating key and subtle fears,
respondents suggested the following: Assess fears comprehensively and monitor change. That is, take the time to do a
comprehensive initial assessment of fearful cognitions,
preferably using guided discovery and possibly questionnaires as an adjunct to this assessment (e.g.,
Agoraphobic Cognitions Questionnaire, Anxiety Sensitivity Index). Periodically revisit this assessment throughout therapy to determine the relevance of cognitions
identified.
For problems with the lack of validity of alternative
rational appraisals, the following solutions were
offered:
1. Create realistic expectations: Help the client
understand and accept that alternatives may seem
invalid and may be doubted at the beginning of
therapy. Encourage clients to focus their efforts on
accomplishing the behavioral goals of the behavioral experiments. Repeated experiential evidence
of safety and manageability drawn through accomplishing those goals will facilitate the cognitive
shift.
To address the invalidity of interoceptive exposure
exercises, the following was recommended:
2. Rule out subtle avoidance: Ensure that the client is
challenging him-or herself with the exercises instead of
subtly avoiding sensations.
3. Use a multipurpose rationale: Explain that the
exercises have multiple purposes and that the client
need not find that the symptoms mimic panic to
benefit from the them. For example, part of the
rationale for sensation exposure can include that it
provides an opportunity for the repeated experience
of raising and lowering of symptoms under controlled
conditions, that it is an arena to practice calming
strategies, and that it allows for repeated testing of
anticipatory fears of what might happen during the
exercise.
4. Use sensations evoked during naturalistic exposures:
Respondents agreed that persistent validity problems
with interoceptive exposure should be addressed by

moving on to naturalistic situational exposures (phobic situations) and working with the symptoms
produced there.
Presence of negative life events/circumstances. Although the
diagnosis of PD may be the reason an individual presents
for treatment, co-existing negative life events may also be
present and need to be addressed to allow the client to
benefit from treatment. Commonly cited negative life
events included relationship distress (e.g., marital problems), job stress, and financial problems. Negative life
events can exacerbate the severity of the primary
condition and distract the client from engagement in
the therapy.
All respondents who highlighted this problem
indicated that they first address the negative life
circumstances before moving to the panic disorder.
For example, for the client with significant marital
distress who finds it difficult to focus on the treatment
of panic, it was suggested to first address the marital
issues, including consideration of marital therapy,
before starting or continuing CBT for PD. Likewise, a
client with job stress may benefit from some level of
stress management training prior to addressing the
panic.
Medication complications. Many clients who begin CBT
for PD are also on medication for PD, often antidepressants, benzodiazepines, or both. The presence of
medication during CBT can interfere with the CBT
through several potential means (see Otto, Smits, &
Reese, 2005, for a review). For example, symptom
attenuation or suppression through the use of medication may reduce motivation to do the work of CBT. Also,
in CBT, sensations of anxiety and panic are treated as
phobic cues and targeted for exposure. The presence of
medication during exposure can place the client at risk
for relapse after discontinuation of the medication
(Marks et al., 1993).
Recommendations for managing the presence of
medication during CBT were as follows:
1. Educate the client: Educate the client regarding the
benefits and risks of concurrent medication use,
discuss discontinuation or continuation options, then
coordinate with the prescriber.
2. Engage the prescriber: Contact the prescriber, educate, and engage their support in the management
plan.
3. Use empirically supported discontinuation protocols:
The respondents who noted this cause indicated that
they follow the protocols used in studies showing that
medication discontinuation can be facilitated and
relapse prevention enhanced by integrating CBT into
medication discontinuation in particular ways (see

Panic Disorder
Spiegel & Bruce, 1997, for a review). For a treatment
manual describing these methods see Otto et al.
(2000).
Poor delivery of treatment. Of course, before determining that a client is not responding to a treatment, it
needs to be determined that the treatment is being
delivered and received. Poor delivery was not a
highly cited obstacle to a good treatment response,
but was mentioned by a three participants. The
common theme of examples cited was that sometimes therapists do not push themselves or clients to
get the most from each phase of treatment. For
instance, it was noted that a therapists reluctance to
have clients challenge themselves strongly during
exposure may make them an unintentional hindrance
to progress.
Recommendations were as follows:
1. Conduct high-fidelity treatment: Attend to delivering
each of the primary emphases of CBT for PD (i.e.,
psychoeducation, somatic skills, cognitive restructuring, and exposure) in a way that maximizes the gain a
client can receive from them.
2. Exposure for the exposer: As the reader might guess,
one recommendation for therapists who find themselves hesitant to encourage clients to challenge
themselves was to expose themselves to asking clients
to do just that. Being a participant model may make
this easier to do.
3. Consult with colleagues: Consulting with a colleague
may reveal suggested changes that could improve
ones treatment delivery.
4. Continuing education: Observing experts through
workshops or videotapes offers opportunities to learn
tried and tested methods.
Therapeutic relationship barriers. In CBT, therapists ask
clients to do things that may make the clients feel
vulnerable to some distress. For this and other reasons
common to all therapies, a clients trust in a therapist
is obviously important. Two respondents cited a lack of
client trust as one reason for treatment resistance.
Fears of being controlled or harmed in one way or
another were cited as examples of impediments to
trust.
The core recommendation was to work on
eliciting, evaluating, and responding to specific fears
and conducting small manageable behavioral experiments that the client is willing to do that test and
ultimately disconfirm the fears. More time than usual
spent on rapport building may be important with
clients who have difficulty trusting or connecting with
others.

33

Discussion
In discussing treatment resistance, we find it helpful
to think first about what an optimal treatment response
looks like. In this, perhaps hypothetical, optimal
response scenario, the clients problems directly related
to their distress and disability are accurately assessed.
This case formulation leads to a treatment plan that is
delivered as intended. The client complies and engages
in the treatment. And, the targeted problems show
improvement that corresponds with reductions in the
distress and disability. It seems that problems at any of
these steps could potentially lead to a suboptimal
response.
Results from this survey of experienced CBT therapists for PD suggest that problems with compliance and
engagement are the most frequently encountered
reasons for a poor treatment response. Participants
Table 2
Guidelines for preventing or managing treatment resistance of panic
disorder
1. Do a thorough initial evaluation.
Rule out general medical and substance conditions that may be
causing the presentation.
Assess for comorbid psychiatric disorders, particularly
depression, substance use, and other anxiety disorders.
Do a comprehensive review of psychosocial systems and life
circumstances that could influence the clinical picture or treatment
plan.
Do a thorough functional assessment of fears, triggers, and
avoidance strategies, facilitated when possible by guided
discovery, direct observation, and psychometric measurement.
2. Deliver CBT with integrity to the model and with sensitivity toward
its difficulty.
Build the trust of the client.
Clarify and emphasize the goals for each phase of therapy.
Explain rationales for prescribed practices.
Emphasize continuing education of key concepts.
Explain, model, and shape the development of the therapeutic
skills being taught.
Emphasize and facilitate engagement in valid behavioral
experiments.
Graduate tasks, as needed.
Assess obstacles and facilitators; functionally analyze and
problem solve barriers.
3. Facilitate treatment adherence.
Make homework specific, clear, and relevant to clearly stated
goals.
Explain, model, and shape the development of homework task
being assigned.
Prompt desired behavior in the natural environment (e.g., by
making a phone call or tying it to another high-frequency
activity), and reinforce it when it has been completed.
Gradate or simplify tasks that are likely or proving to be
difficult.
4. Reconsider the case formulation if progress is not seen when
expected.
Reevaluate the functional assessment of fears and triggers.
Reevaluate diagnostic accuracy.

Sanderson & Bruce

34

described addressing this issue by making engagement a

primary goal of therapy and the facilitation of adherence
a consideration in every therapeutic prescription (cf.
Meichenbaum & Turk, 1987).
Also common were problems with case formulation,
which appeared in several forms: misdiagnosis, missed
comorbid diagnoses, other life circumstances not taken
into consideration, or inaccuracies in behavioral analysis
of triggers. Although not cited as strongly as compliance
and engagement problems, formulation problems were
endorsed by nearly half of respondents. These results
suggest that when treatment response is poorer than
expected but fidelity and compliance appear good,
revisiting the case formulation may be helpful.
Interestingly, poor treatment delivery did not appear
frequently in survey results. That may be a partial product
of sampling bias, in that you might expect to see less of
those problems as therapists become more experienced.
Of course, every therapist should first consider his or her
role in the clients lack of response, before moving to
other considerations.
Table 2 contains a summary of the recommendations
offered by the survey participants in the form of
guidelines for preventing or managing treatment resistance. They ring familiar as the kinds of practices
advocated by every good clinical training program. It is
reassuring to note that our findings overlap with several
nonempirical attempts to provide further information to
clinicians regarding improving treatment outcome for
CBT of PD (e.g., Huppert & Baker-Morissette, 2003;
McCabe & Antony, 2005; Otto & Gould, 1996).

Appendix A. Survey Participants

Anne Marie Albano
David Antonuccio
Marty Antony
Deborah Beidel
Judith Beck
Cheryl Carmin
Bruce Chorpita
Frank M. Dattilio
Thomas Ellis
Steven Friedman
Robert Goisman
James Herbert
Jonathan Huppert
Rolf Jacob
Steven Safren
Lisa Smith
Steven Taylor
David Tolin
Julia Turovsky
Sheila Woody

References
American Psychiatric Association. (1998). Practice guideline for the
treatment of patients with panic disorder. American Journal of
Psychiatry, 144, 134.
American Psychiatric Association. (2000). Diagnostic and Statistical
Manual of Mental Disorders, (4th ed., text revision). Washington,
DC: Author.
Barlow, D. H., & Craske, M. G. (2000). Mastery of Your Anxiety and PanicIII. San Antonio, TX: The Psychological Corporation.
Barlow, D. H., Gorman, J. M., Shear, M. K., & Woods, S. W. (2000).
Cognitive-behavioral therapy, imipramine, or their combination
for panic disorder: a randomized controlled study. Journal of the
American Medical Association, 283, 25292536.
Barlow, D. H., Raffa, S. D., & Cohen, E. M. (2002). Psychosocial
treatments for panic disorders, phobias, and generalized anxiety
disorder. In P. E. Nathan, & J. M. Gorman (Eds.), A Guide to
Treatments that Work (2nd ed.).New York: Oxford University Press.
Brown, T. A., Antony, M. M., & Barlow, D. H. (1995). Diagnostic
comorbidity in panic disorder: effect on treatment outcome and
course of comorbid diagnoses following treatment. Journal of
Consulting and Clinical Psychology, 63, 408418.
Clark, D. M., Salkovskis, P. M., Hackmann, A., Middleton, H.,
Anastasiades, P., & Gelder, M. G. (1994). A comparison of cognitive
therapy, applied relaxation and imipramine in the treatment of
panic disorder. British Journal of Psychiatry, 164, 759769.
Huppert, J. D., & Baker-Morissette, S. L. (2003). Beyond the manual:
the insiders guide to panic control treatment. Cognitive and
Behavioral Practice, 10, 213.
Jacobson, N. S. (1978). A stimulus control model of change in
behavioral couples therapy: implications for contingency contracting. Journal of Marital and Family Therapy, 4, 2935.
Margraf, J., Barlow, D. H., Clark, D. M., & Telch, M. J. (1993).
Psychological treatment of panic: work in progress on outcome,
active ingredients, and follow-up. Behaviour Research and Therapy,
31, 18.
Marks, I. M., Swinson, R. P., Basoglu, M., Kuch, K., Noshirvani, H.,
OSullivan, G. O., Lelliott, P. T., Kirby, M., McNamee, G., &
Sengun, S. (1993). Alprazolam and exposure alone and combined
in panic disorder with agoraphobia. A controlled study in London
and Toronto. British Journal of Psychiatry, 162, 776787.
McCabe, R. E., & Antony, M. M. (2005). Panic disorder and
agoraphobia. In M. Antony, D. R. Ledley, & R. Heimberg (Eds.),
Improving Outcomes and Preventing Relapse in Cognitive Behavioral
Therapy. New York: The Guilford Press.
McLean, P. D., Woody, S., Taylor, S., & Koch, W. J. (1998). Comorbid
panic disorder and major depression: implications for cognitivebehavioral therapy. Journal of Consulting and Clinical Psychology, 66,
240247.
Meichenbaum, D., & Turk, D. C. (1987). Facilitating Treatment Adherence:
A Practitioners guidebook. New York: Plenum Press.
Otto, M. W., Barlow, D. H., Craske, M. G., & Jones, J. C. (2000). Stopping
Anxiety Medication: Panic Control Therapy for Benzodiazepine DiscontinuationTherapist Guide. San Antonio: The Psychological
Corporation.
Otto, M. W., & Gould, R. A. (1996). Maximizing treatment
outcome for panic disorder: cognitive-behavioral strategies.
In M. H. Pollack, M. W. Otto, & J. F. Rosenbaum (Eds.),
Challenges in Clinical Practice: Pharmacologic and Psychosocial
Strategies (pp. 113140). New York: The Guilford Press.
Otto, M. W., Smits, J. A. J., & Reese, H. E. (2005). Combined
psychotherapy and pharmacotherapy for mood and anxiety
disorders in adults: review and analysis. Clinical Psychology: Science
and Practice, 12, 7286.
Rosenbaum, J. F., Pollack, M. H., & Pollack, R. A. (1996). Clinical issues
in the long-term treatment of panic disorder. Journal of Clinical
Psychiatry, 57, 4448.
Scott, E. L., Pollack, M. H., Otto, M. W., Simon, N. M., & Worthington,
J. J. (1999). Clinician response to treatment refractory panic
disorder: a survey of psychiatrists. Journal of Nervous and Mental
Disease, 187, 755757.

Panic Disorder
Spiegel, D. A., & Bruce, T. J. (1997). Benzodiazepines and exposurebased cognitive behavior therapies for panic disorder: conclusions from combined treatment trials. American Journal of
Psychiatry, 154, 773781.

Address correspondence to William C. Sanderson, Department of

Psychology, Hofstra University, Hempstead, NY 11549 USA; e-mail:
psywcs@Hofstra.edu.

The authors would like to thank the survey participants, listed in the
Appendix, who generously offered their valuable time and clinical
perspectives.

Received: December 22, 2005

Accepted: April 6, 2006
Available online 30 November 2006

35