Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Nov 1.
Published in final edited form as: Psychother Res. 2022 Feb 24;32(8):972–983. doi: 10.1080/10503307.2022.2044086

How Does Hostile Resistance Interfere with the Benefits of Cognitive-Behavioral Therapy for Panic Disorder? The Role of Therapist Adherence and Working Alliance

Rachel A Schwartz 1, Kevin S McCarthy 2, Nili Solomonov 3, Dianne L Chambless 4, Barbara Milrod 5, Jacques P Barber 6
PMCID: PMC9399310  NIHMSID: NIHMS1784954  PMID: 35209800

Abstract

Objective:

Although clients’ hostile behavior directed at therapists (hostile resistance) predicts worse outcomes in cognitive-behavioral therapy (CBT) for panic disorder, the process by which this happens remains unknown. This study examines two putative mechanisms: working alliance and therapist adherence.

Method:

Seventy-one adults with primary panic disorder received CBT in a larger trial. Hostile resistance and adherence in Sessions 2 and 10 were reliably coded using observer-rated measures; client- and therapist-rated questionnaires assessed working alliance. Outcome measures were attrition and symptomatic improvement, assessed at multiple timepoints with the Panic Disorder Severity Scale.

Results:

Hostile resistance was significantly related to both preexisting (r = −.36, p = .04) and subsequent declines (r = −.58, p < .0001) in the working alliance. Nevertheless, hierarchical linear modeling revealed that neither a declining alliance nor therapist adherence (whether treated as linear or curvilinear) was independently predictive of symptom change, nor did these factors mediate hostile resistance’s association with worse symptomatic improvement. Exploratory logistic regressions similarly indicated that neither adherence nor alliance moderated whether hostilely resistant clients dropped out.

Conclusion:

This is the first study to establish a bidirectional association between hostile resistance and a declining working alliance. Findings also add to a mixed literature on the adherence-outcome relationship.

Keywords: panic disorder, hostile resistance, working alliance, therapist adherence, process in cognitive-behavioral therapy, resistance

Anger and hostility are widely cited as emotions that can derail psychotherapy and are difficult for therapists to manage (Mayne & Ambrose, 1999). Higher trait levels of these emotions have been shown to interfere with cognitive-behavioral therapy (CBT): In CBT for panic disorder, for instance, higher patient-reported interpersonal aggression is associated with worse therapist competence and adherence (Boswell et al., 2013) and less improvement in anxiety (Cassiello-Robbins et al., 2015). When clients’ hostility is directed at therapists—known as hostile resistance—the challenges therapists must overcome to deliver effective CBT may be even greater. In the CBT literature, clients are said to exhibit hostile resistance when they go against therapists’ direction in an openly combative manner that communicates an attack on the therapists’ competence, methods, or personal qualities (Westra et al., 2009). In this study, expressions of hostile resistance included snapping at the therapist, “This isn’t your area,” and angrily asserting that “just because [the therapist] has a PhD” he is not infallible (“You’re wrong. Your methods are bad because you can’t learn or maybe you don’t want to learn…”).

Hostile resistance is associated with a variety of poor therapy outcomes, such as worse therapist competence in interpersonal therapy for depression (Foley et al., 1987). Most recently, a study by this group found that hostile resistance is a robust predictor of poor outcomes in CBT for adults with panic disorder (Schwartz et al., 2018): Hostility early in therapy predicted drop out, and at mid-treatment predicted less symptom improvement—even after established outcome predictors in this sample (e.g., expectancy; Chambless et al., 2017) were controlled. We subsequently conducted a qualitative investigation to elucidate why patients became hostile in the first place, which revealed that patient, therapist, and treatment factors all play a role (Schwartz et al., 2021). However, the process by which hostile resistance translates into poor outcomes remains unclear. This information could be used to mitigate the negative sequelae of hostility once it occurs, yet no studies have examined the mechanisms driving hostile resistance’s impact on psychotherapy outcomes, and CBT outcomes in particular. Some research in this vein has been conducted in the area of client resistance to CBT, more generally, but has thus far been unsuccessful in identifying mechanisms, determining that neither homework compliance nor impaired therapist empathy plays a mediating role (Aviram & Westra, 2011; Hara et al., 2018).

Especially in the context of structured CBT protocols, one mechanism through which hostile resistance may exert its negative effect is therapist adherence, or the extent to which an intervention is delivered in a way that is prescribed by the treatment manual. Adherence is distinct from competence in that it measures what therapists do, rather than how skillfully they do it. In CBT for panic disorder, therapists show lower treatment fidelity when clients are more resistant generally (Zickgraf et al., 2015) and have higher self-reported aggression (Boswell et al., 2013), suggesting that adherence may play a mediating role in hostile resistance’s deleterious impact. However, although hostile resistance may prompt therapists to be less adherent, it remains uncertain whether worse treatment fidelity is harmful or protective in these moments. On the one hand, hostile attacks may interfere with therapists’ ability to focus the session and deliver the treatment as designed, resulting in worse outcomes because patients receive smaller doses of the intervention. As support, Strunk et al. (2010) found that when depressed patients engaged in behaviors that were associated with less therapist adherence, they showed less symptom improvement in the following CBT session.

On the other hand, poor adherence following hostile resistance might lead to better outcomes if therapists are appropriately deviating from the manual to adjust to the needs of the client. Poor adherence is not necessarily counter-therapeutic, as evidenced by a large yet inconclusive literature examining adherence’s relation to outcome (e.g., Webb et al., 2010). Rather, depending on the circumstances, veering off protocol may be a non-adherent, but nevertheless competent, decision. If a patient were to report acute suicidality, for instance, it would hardly be appropriate for a therapist to respond by asking how the homework went. As such, rigid adherence may be at least as problematic as poor adherence in the context of hostilely resistant clients. With less motivated clients, for instance, Huppert et al. (2006) found that higher adherence produced worse CBT for panic outcome. Growing evidence suggests that the relationship between adherence and outcome may in fact be curvilinear (i.e., quadratic), lending support to the so-called Goldilocks effect, whereby highly rigorous adherence is less beneficial than a moderate level of flexible adherence (e.g., McCarthy et al., 2016). To the extent that rigid adherence contributes to the emergence of hostile resistance (Schwartz et al., 2021), it likewise may follow that responding to such behavior with flexibility may buffer against further hostility and poor treatment response.

Just as rigidly applying CBT protocols may result in worse outcomes, there is also evidence that incorporating theoretically “off-brand” interventions can improve outcomes in the context of disruptive clinical processes and unmotivated clients (Constantino et al., 2020; Huppert et al., 2006). Breaking with directive CBT methods in favor of supportive and motivational interviewing (MI) techniques, specifically, has proven protective when patients express resistance broadly in CBT for generalized anxiety (Aviram et al., 2016). If CBT therapists are responding to hostile behavior by incorporating off-protocol techniques from non-CBT theoretical frameworks, this form of non-adherence may be protective against the negative consequences of hostile resistance.

Another possibility altogether is that hostile resistance impedes CBT outcomes by way of degrading the working alliance. When clients are resistant to treatment—even in a way that is not hostile—the alliance and therapists’ positive regard for clients suffer (Westra et al., 2012; Westra & Norouzian, 2018). Further suggesting a potential mediating role, unresolved ruptures in the alliance have been shown to result in worse therapy outcomes both in the wider literature (Eubanks et al., 2018) and in this sample, as worse patient-rated alliance was related to less improvement in panic-related cognitive misinterpretations in this trial (Bagdasarov et al., 2018).

In a qualitative investigation of the factors that contribute to hostile resistance’s emergence, Schwartz et al. (2021) found that a stronger alliance was protective not only against the occurrence of client hostility, but also—in at least one case—against the harmful consequences of such hostility once it occurred. In that case, hostility was so foreign to the dyad’s dynamic that it appeared to function as a wake-up call of sorts to the therapist, who responded by appropriately changing course. Despite the hostile exchange, this patient went on to become a CBT responder—perhaps because the working alliance was quickly restored. Thus, it may be that the clients who go on to have poor outcomes after expressing hostile resistance are those for whom the working alliance becomes irrevocably damaged following the initial rupture.

The aim of this study was to elucidate the mechanisms by which hostile resistance results in worse CBT outcomes in order to guide therapists’ response to such behavior in session, and ultimately improve clinical outcomes. We tested the hypothesis that hostile resistance results in worse outcomes by way of damaging the working alliance. In addition, we also examined whether therapist adherence (both linear and curvilinear) and use of techniques from other theoretical frameworks mediate the association between hostile resistance and poor CBT outcome. Given the uncertainty as to whether poor adherence and deviations from the protocol might function for good or ill in the context of client hostility, and whether the issue may in fact be rigid (rather than low) adherence, we do not have clear hypotheses for these tests. As this is the first study to our knowledge to examine curvilinear adherence in relation to CBT outcome, this study adds to a mixed literature on the relationship between adherence and outcome—which is particularly limited for studies of CBT for panic (Webb et al., 2010)—and permits an examination of whether previous studies might conflict because this relationship is nonlinear.

Method

The present study represents a secondary analysis of data and recordings of sessions that were collected as part of a two-site randomized controlled trial comparing the efficacy of three therapies for adults with panic disorder (Milrod et al., 2016, which contains more details on the parent trial). The present study focuses on this trial’s CBT arm. Treatment was offered gratis from 2006–2011. Informed consent was obtained from all patients, as was approval by the Institutional Review Boards at both sites.

Participants

Seventy-one of the 81 patients randomized to CBT in the parent trial (Schwartz et al., 2018) were included in the present study; six patients withdrew prior to Session 2 and four patients’ recordings were unavailable due to technological or staff error. Excluded patients did not differ from the remaining patients in terms of gender, age, agoraphobia, ethnicity, or baseline panic severity (all ps > .13). All participants met criteria for primary panic disorder on the Anxiety Disorders Interview Schedule for DSM-IV (Brown et al., 2004), with 36 (51%) also meeting for agoraphobia. Patients were excluded from the trial if they were dependent on substances, acutely suicidal, in concurrent psychotherapy, or had a history of psychosis or mania. The sample had a mean age of 39.4 (SD = 12.7) and included 40 (56%) women. Forty-seven (66%) were White, 18 (25%) Black, 5 (7%) Asian, and 1 (1.4%) identified as “other”; 14 (20%) were Hispanic. Thirty-five (49%) were treated Cornell and 36 at the University of Pennsylvania.

Ten (14.1%) of these patients exhibited hostile resistance, as reported in Schwartz et al. (2018). The presence of hostile resistance was reliably coded in this sister study using the Client Resistance Code (Westra et al., 2009), whereby coders watched Sessions 2 and 10 in their entirety and rated 30-second bins for peak resistance severity, from 0 (no resistance) to 3 (hostile resistance). The 10 patients who had codes of 3 in any 30-second bin of a session comprise this study’s hostile resistance group, which has a mean age of 41.6 (SD = 13.2). Patients were 6 men and 4 women: 8 were White, 2 Black (1 Hispanic). Collectively this group expressed 19 instances of hostile resistance during 12 sessions (5 in Session 2, 5 in Session 10, and 2 in both sessions). Due to poor quality video, technological or staff error, or drop out prior to Session 10, not all sessions were available for the full sample; rather, hostile resistance could be coded in 67 Session 2s and 57 Session 10s. Schwartz et al. (2018) contains further details on procedures related to resistance coding.

Treatment

CBT followed a modified version of Panic Control Therapy (PCT; Craske et al., 2000) and entailed psychoeducation, diaphragmatic breathing, correction of dysfunctional beliefs about panic and anxiety, interoceptive and in vivo exposure, and homework. The PCT protocol is highly structured, with each session having a specific agenda. Treatment was administered in up to twenty-four 45-minute biweekly sessions over 12 weeks. Patients who completed less than 16 sessions of treatment were considered treatment dropouts.

Therapists.

CBT was delivered by eight doctoral-level clinicians (75% women, 100% White) with 3–20 years of post-graduate clinical experience (M = 9.5, SD = 6.09). All received specific training in PCT by attending a two-day workshop, and those with less experience with panic treated a supervised pilot case before treating study patients. Throughout the trial, supervisors at both sites monitored adherence by reviewing videos of sessions and regularly providing therapists with adherence-related feedback. Therapists treated between 4 and 18 patients (M = 8.8, SD = 4.7). Five therapists (3 women, 2 men) had at least one client who expressed hostile resistance.

Procedure

Therapist adherence and use of off-protocol techniques were coded in the same two sessions of the CBT protocol as hostile resistance (Schwartz et al., 2018): Session 2, which consisted of psychoeducation and information gathering, and Session 10, which primarily entailed cognitive restructuring. This approach allowed us to examine how hostile resistance affected therapists’ behavior in the moment; moreover, sampling adherence from early- and middle-portions of treatment may be important in light of evidence that adherence levels vary over the course of therapy (Hauke et al., 2014). Just as with resistance coding (Schwartz et al., 2018), Sessions 3 (n = 6) and 9 (n = 3) were rated when Sessions 2 and 10 were unavailable, respectively, based on similarities in content. In all coding procedures, coders were unaware of hypotheses and patient outcomes.

Adherence Coding.

Although 10% of CBT tapes were randomly sampled and rated for adherence in the parent trial to ensure treatment fidelity (Milrod et al., 2016), in this study we rated adherence in the same two sessions for all CBT clients to achieve greater homogeneity of session material. Specifically, therapist adherence to Sessions 2 and 10 of the CBT protocol was rated by five advanced undergraduates using session-specific CBT Adherence Scales (Chambless & Sharpless, 2011; available from authors upon request), developed for the parent trial. All scales contained between 3–10 items rated on a Likert scale from 1 (non-adherent/no review) to 7 (adherent/complete review), which were then averaged such that scores ranged from 1 to 7, with higher scores reflecting greater adherence. Items assessed the degree to which therapists adhered to prescribed interventions for that session, such as cognitive restructuring, homework review, breathing skills, and components of psychoeducation and assessment. Each session was watched in its entirety by two coders (randomly assigned), whose ratings were then averaged. Coders were trained by Dianne Chambless, one of the scales’ creators, and achieved excellent interrater reliability (ICC [1,2] = .89 for Session 2, ICC [1,2] = .82 for Session 10). To prevent coder drift, a subset of five videos was rated and reviewed by all coders at weekly calibration meetings (ICC [1,6] = .90). Coders were also trained to detect whether therapists used proscribed interventions (i.e., those from the applied relaxation arm of the trial); however, no use of applied relaxation interventions was reported by any rater.

Off-Protocol Coding.

The 60-item Multitheoretical List of Therapeutic Interventions (MULTI; McCarthy & Barber, 2009) assessed therapists’ use of off-protocol (i.e., non-CBT) techniques. As part of a sister study (Solomonov et al., 2020), 25 doctoral students watched complete recordings of CBT Sessions 2 and 10 and reliably rated items that assessed the degree to which therapists’ behaviors aligned with seven therapeutic orientations: (a) Behavioral; (b) Cognitive; (c) Dialectical-behavioral; (e) Person-centered; (f) Process-experiential; (g) Interpersonal; and (h) Psychodynamic. Items were rated on a 1 (not at all) to 5 (very) scale of how typical each intervention was of the session. Each session was coded by at least two raters, whose ratings were averaged for analysis. An off-protocol summary score of the extent to which therapists used non-CBT techniques was calculated by averaging items from the five non-CBT orientations (i.e., c-h above).1 Any CBT items (i.e., items that appeared in the Behavioral or Cognitive scales) that also appeared on the five non-CBT scales were removed, such that the off-protocol score included 27 items, characterized by good internal consistency (α = .71 Session 2, α = .77 Session 10). The MULTI has previously been shown to be internally consistent and reliable (e.g., McCarthy & Barber, 2009; McCarthy et al., 2016), and interrater reliability for the subscales was acceptable-to-good in a related sample (ICCs = .63-.80; Solomonov et al., 2020).

Non-Coding Measures.

The Panic Disorder Severity Scale (PDSS; Shear et al., 1997) was administered by independent evaluators uninformed as to treatment or therapist assignment to assess panic severity. The semi-structured interview, which was the trial’s primary outcome measure, includes 7 items that are each scored from 0 (none) to 4 (extreme) and then summed. To assess symptom change, the PDSS was administered before, during (at Weeks 1, 5, and 9), and at termination of CBT (Week 12). Prior research has shown good convergent (e.g., r = .55) and discriminant validity (rs = −.01-.33; Shear et al., 1997), and excellent interrater reliability was achieved in the parent study (ICC [2,1] = .95).

The Working Alliance Inventory (WAI; Tracey & Kokotovic, 1989) was administered to therapists (WAI-T) and clients (WAI-C) at Sessions 2, 5, and 10 to assess the strength of the therapeutic alliance. Of 12 items, 9 comprise a factor related to therapist-patient agreement on tasks and goals that has been shown to predict CBT outcome (Agreement/Confidence factor; Webb et al., 2011). These nine items were rated on a 7-point scale and averaged for analyses, such that scores range from 1 (poor) to 7 (strong). Internal consistency of this factor in this sample was excellent at all three times points assessed for both therapist (α = .92-.95) and client (α = .89-.92) versions of the scale.

Statistical Analyses

Putative mediators of the relationship between hostile resistance in Session 10 and less subsequent PDSS change were examined separately using hierarchical linear modeling (HLM). All HLMs predicted rate of PDSS change subsequent to Session 10 (i.e., the time at which hostile resistance was measured), and included the residuals from regressing Session 10 PDSS on baseline PDSS as a covariate to control for the influence of PDSS change prior to Session 10. In Step 1, zero-order correlations between putative mediators and hostile resistance were examined. Step 2 involved examining whether each mediator independently predicted PDSS change (i.e., without including hostile resistance in the model). In Step 3, the mediator and hostile resistance were entered into an HLM together to assess hostile resistance’s effect on outcome when controlling for the mediator. Results in this step would suggest mediation if (a) the Mediator x Time interaction were significant, and (b) the statistically significant relationship between hostile resistance and PDSS change were eliminated (full mediation) or reduced (partial mediation) when the mediator was controlled (i.e., the Hostile Resistance x Time term became less or non-significant). Time (session number) was modeled as a random effect while hostile resistance and mediators were modeled as fixed. Intercepts were modeled as fixed because their variance in this sample approached zero (i.e., patients started treatment with comparable PDSS scores), inconsistent with the assumptions of random effects.

Whereas hostile resistance in Session 10 is related to less improvement on the PDSS, hostility in Session 2 predicts drop out (Schwartz et al., 2018). As such, logistic regressions were used in examinations of Session 2 hostile resistance, with drop out as the binary dependent variable. Because we hypothesized that clients might not drop out if therapists brought in more off-protocol behaviors following hostile resistance, models of attrition tested adherence- and alliance-based factors as moderators, rather than mediators, of outcome. A significant Moderator x Hostile Resistance interaction term would indicate the presence of moderation. However, because only 13 (18.3%) patients in this sample dropped out of treatment, and given that both the independent and dependent variables are dichotomous, attrition analyses are likely underpowered and should be treated as exploratory.

For tests of curvilinear mediators/moderators, all relevant subcomponent terms were included in the models, with the primary interaction of interest being the Mediator-Squared x Hostile Resistance term (or Moderator-Squared x Hostile Resistance, for models of attrition).

Given our focus on whether hostile resistance affects therapists’ behavior in the moments following client hostility, all models included hostile resistance and mediator/moderator variables that were concurrently measured. That is, tests of mediation of the association between hostile resistance in Session 10 and worse symptom change used adherence and off-protocol scores that were also measured in Session 10; tests of moderation of the relationship between hostile resistance in Session 2 and drop out used adherence, off-protocol, and alliance scores measured in Session 2. The exception was for examinations of working alliance as a mediator of PDSS change: Given our hypothesis that hostile resistance in Session 10 leads to a deterioration in working alliance (and, in turn, worse outcome), to rule out the possibility that the alliance had always been poor these models used a residualized alliance variable, generated by regressing alliance in Session 10 on the average of earlier alliance scores (measured in Sessions 2 and 5). Type I error was set at .05 for all analyses. Analyses were conducted using SAS version 9.4.

Results

Overall, the sample was characterized by fairly high working alliance in both Session 2 (M = 5.69, SD = 1.03 client-rated; M = 5.46, SD = 0.84 therapist-rated) and Session 10 (M = 6.05, SD = 0.89 client-rated; M = 5.71, SD = 0.93 therapist-rated). Adherence was also generally high in the subset of sessions rated, with a mean adherence score of 5.31 (SD = 0.92) and 4.89 (SD = 0.98) in Sessions 2 and 10, respectively. Forty-nine (87.5%) of the Session 2s and 48 (85.7%) of the Session 10s had at least adequate adherence (i.e., a score of at least 4). Use of off-protocol interventions was low in both Session 2 (M = 1.66, SD = 0.17) and Session 10 (M = 1.66, SD = 0.24), indicating that resorting to off-protocol techniques was only “slightly typical” of the average session. On average, adherence declined by 0.4 points from Session 2 to Session 10, whereas use of off-protocol techniques did not change over treatment.

Zero-order correlations between hostile resistance and mediators/moderators are presented in Table 1. Only residualized working alliance scores were significantly correlated with hostile resistance: When clients were hostilely resistant in Session 2, both client- (r = −.58, p < .0001) and therapist-rated (r = −.41, p = .002) alliance subsequently declined at medium-large effects. Similarly, expressing hostile resistance in Session 10 was associated with a working alliance that had already been on the decline at a medium-sized effect, whether client- (r = −.36, p = .04) or therapist-rated (r = −.32, p = .03).2 No other adherence or alliance factors were significantly correlated with hostile resistance in either session, all ps > .07.

Table 1.

Pearson Correlations of Hostile Resistance with Mediator/Moderator Variables.

Hostile Resistance, Session 2 Hostile Resistance, Session 10
Adherence, Session 2 < .01 −.18
Adherence, Session 10 −.11 −.25
Adherence, Session 2 (Curvilinear) −.02 −.19
Adherence, Session 10 (Curvilinear) −.12 −.21
Off-Protocol, Session 2 .07 −.01
Off-Protocol, Session 10 .15 −.06
Off-Protocol, Session 2 (Curvilinear) .06 −.01
Off-Protocol, Session 10 (Curvilinear) .17 −.05
Working Alliance-Client, Session 2 −.08 −.15
Working Alliance-Therapist, Session 2 −.17 .18
Working Alliance-Client, Session 10 Residualized −.58** −.36*
Working Alliance-Therapist, Session 10 Residualized −.41** −.32*
Open in a new tab

Note: The residualized working alliance score is the residual of regressing Session 10 alliance on the average of alliance scores from Sessions 2 and 5. As the absence and presence of hostile resistance was coded as 0 and 1, respectively, negative correlations indicate that levels of the mediator/moderator variable were lower in the hostilely resistant group.

*

p < .05.

**

p < .01.

PDSS Symptom Change

As reported in Table 2, HLM analyses revealed that neither adherence (whether linear or curvilinear), nor use of off-protocol techniques (whether linear or curvilinear), nor change in working alliance (whether therapist- or client-rated) was predictive of PDSS change when hostile resistance was not included as a covariate (all ps > .34, all |r|s < .13). When these variables were entered into HLM models with hostile resistance (Table 3), none emerged as significant mediators of the association between hostile resistance in Session 10 and worse symptomatic improvement: All Mediator x Time terms were non-significant (ps > .37, |r|s < .13), and in all but one case, hostile resistance’s effect on PDSS change remained statistically significant, predicting a slower rate of change, even after controlling for the mediating variable. This pattern is inconsistent with mediation. The exception was in the model examining change in client-rated working alliance (WAI-C), in which the statistical significance of hostile resistance was reduced to p = .16. However, given that (a) this analysis’ sample was greatly diminished (n = 31) due to missing questionnaires, (b) the WAI-C x Time term was not significant in this model or when examined as an independent predictor, and that (c) hostile resistance’s effect remained a near medium r of −.26, highly comparable to the median r of −.29 for the significant effect of hostile resistance in analyses with larger samples, we think it probable that hostile resistance’s lesser significance owed to the reduced sample and does not provide evidence of partial mediation.

Table 2.

Mediators’ Prediction of Symptom Change after Session 10 (Hostile Resistance Not Included as a Covariate).

Mediator x Time Term n β SE t (df) p r CI 95%
Adherence 54 0.10 0.26 0.37 (87) .71 .05 −.22, .31
Adherence-Squared 54 −0.05 0.15 −0.34 (86) .74 −.05 −.31, .22
Off-Protocol 56 −0.98 1.00 −0.97 (92) .34 −.13 −.38, .14
Off-Protocol-Squared 56 1.67 3.03 0.55 (91) .58 .07 −.19, .33
Working Alliance-Client, Residualized 36 0.10 0.31 0.32 (62) .75 .05 −.28, .38
Working Alliance-Therapist, Residualized 52 0.22 0.26 0.85 (88) .40 .11 −.16, .38
Open in a new tab

Note: Results were generated using hierarchical linear modeling. All putative mediators were measured at Session 10 with the exception of the two working alliance variables, which are residualized change scores (i.e., residuals of regressing Session 10 alliance on the average of alliance scores from Sessions 2 and 5).

Table 3.

Mediation Results for the Association Between Hostile Resistance in Session 10 and Poor Subsequent Symptom Change.

Mediation Model n β SE t (df) p r CI 95%
Adherence x Time
 Hostile Resistance x Time		 52 0.18
−1.91		 0.27
0.89		 0.71 (85)
−2.13 (85)		 .48
.04		 .01
−.29		 −.18, .36
−.52, −.02
Adherence-Squared x Time
 Hostile Resistance x Time		 52 −0.14
−2.11		 0.16
0.93		 −0.86 (84)
−2.27 (84)		 .39
.03		 −.12
−.31		 −.38, .16
−.53, −.04
Off-Protocol x Time
 Hostile Resistance x Time		 53 −0.74
−1.83		 1.02
0.87		 −0.72 (87)
−2.11 (87)		 .47
.04		 −.10
−.28		 −.36, .17
−.51, −.01
Off-Protocol-Squared x Time
 Hostile Resistance x Time		 53 0.16
−1.84		 3.12
0.89		 0.05 (86)
−2.08 (86)		 .96
.04		 .01
−.28		 −.26, .28
−.51, −.01
Working Alliance-Client x Time
 Hostile Resistance x Time		 31 0.22
−1.60		 0.32
1.12		 0.68 (51)
−1.43 (51)		 .50
.16		 .13
−.26		 −.24, .46
−.11, .56
Working Alliance-Therapist x Time
 Hostile Resistance x Time		 46 0.24
−1.80		 0.27
0.90		 0.90 (77)
−2.01 (77)		 .37
.048		 .13
−.29		 −.16, .41
< .001, .54
Open in a new tab

Note: Results were generated using hierarchical linear modeling. Mediation is suggested when the Mediator x Time term is significant and the Hostile Resistance x Time term is non-significant. Hostile resistance and all putative mediators were measured at Session 10 with the exception of the two working alliance variables, which are residualized change scores. In these analyses, a negative Hostile Resistance x Time effect indicates that hostile resistance predicts a slower rate of change on the Panic Disorder Severity Scale (PDSS); a negative Mediator x Time effect indicates that a higher level of that mediator predicts a faster rate of PDSS change.

Drop Out

Results from these exploratory logistic regressions are reported in the supplemental materials. Tests of moderation revealed that neither adherence (whether linear or curvilinear), nor use of off-protocol techniques (whether linear or curvilinear), nor working alliance (whether therapist- or client-rated) moderated whether clients who were hostilely resistant in Session 2 dropped out of CBT, all Moderator x Hostile Resistance ps > .46. Moreover, none of the main effects of these Session 2 factors were significant, all Moderator x Time ps > .50.

Discussion

Using data from a trial of CBT for panic disorder, the present study examined whether hostile resistance leads to poor outcomes by way of degrading the working alliance or altering therapists’ ability to deliver the treatment faithfully. Contrary to hypotheses, therapist adherence—whether conceptualized as linear or curvilinear adherence or as the use of off-protocol techniques—was unrelated to both hostile resistance and outcome, and did not account for hostile resistance’s association with attrition or poor symptomatic improvement. Results were similar for working alliance: Although hostile resistance was significantly related to preexisting and subsequent declines in the alliance, this relation did not explain why clients who were hostilely resistant went on to drop out of therapy and improve less in panic symptoms.

Understanding how hostile resistance translates into poor outcomes might guide therapists’ response to such behavior once it occurs; nevertheless, none of the factors we examined appeared to be operative mechanisms through which hostility exerts its negative effect, consistent with the difficulty others have had in identifying mediators of resistance’s impact, more broadly (Aviram & Westra, 2011; Hara et al., 2018). Given our uncertainty as to whether poor adherence would be protective or harmful in the context of hostility, it is possible that adherence’s role might have been masked if both were true: While in some cases poor adherence may have led to worse outcomes by limiting the dose of the intervention delivered, in others it may have represented a degree of flexibility that was constructive. Were this the case, the positive and negative effects of low adherence may have canceled out. This account is supported by Hauke et al. (2014), who found that adherence was both a force for good and ill for outcome of CBT for panic, depending on the stage of therapy, session content, and patient characteristics.

It is also possible that adherence was unrelated to outcome in this study because we did not test adherence in sessions that entailed exposure, which may be a particularly important ingredient for therapy success compared to other aspects of the treatment protocol (Weck et al., 2016). That said, at odds with this possibility is that a component study showed that cognitive therapy alone, interoceptive exposure alone, and their combination were all equally effective in treating panic (Margraf & Schneider, 1991).

Alternatively, given the heterogeneity in the origins of hostile resistance (Schwartz et al., 2021), it may be that the mechanisms we posited operate only within certain subgroups of hostilely resistant clients but not others. In a grounded theory examination of the moments preceding hostile exchanges, Schwartz et al. (2021) identified two distinct pathways to hostile resistance: In the first, patients’ challenging dispositions (i.e., high levels of narcissism, obsessiveness, anger, or defiance) played a primary role; by contrast, patients who followed the second pathway were not dispositionally challenging, but rather became hostile due to therapist failures—particularly of empathy. These two groups were combined for analyses to increase statistical power. However, as hostile behavior may affect therapists, therapy, and the therapeutic relationship differently in these subgroups, this study’s treatment of hostile resistance as a homogenous phenomenon may have inadvertently obscured an important distinction. Unfortunately, the sample size for each of these two subgroups is too small to permit testing.

Findings add to a mixed literature on whether therapist adherence is related to outcome. Consistent with findings from Webb et al.’s (2010) meta-analysis that adherence has no effect on either psychotherapy outcome (r = .02 based on 32 studies) or CBT outcome specifically (r = .04 based on 9 studies), neither adherence to prescribed interventions nor borrowing techniques from other interventions was significantly related to outcome in this study. Findings also add to a limited literature on the adherence-outcome relationship in panic disorder, specifically. That only one of the CBT studies in this meta-analysis was on panic disorder attests to this gap, especially relative to studies of major depression (Webb et al., 2010). To our knowledge, ours is one of six studies to examine the impact of therapist adherence on CBT for panic disorder outcome, four of which were published subsequent to Webb et al’s (2010) review. Consistent with our findings, none of these studies detected a significant association between adherence and panic symptom change (Boswell et al., 2013; Haug et al., 2016; Hauke et al., 2014; Huppert et al., 2006; Weck et al., 2016). That said, it may be that adherence needs to be investigated in the context of interactions with other variables (e.g., motivation, stage of therapy; Hauke et al., 2014; Huppert et al., 2006).

This study was the first to examine curvilinear adherence in relation to CBT outcome. Contrary to evidence that moderate levels of adherence may be more beneficial than highly rigorous adherence in other therapeutic modalities (the so-called Goldilocks effect; e.g., McCarthy et al., 2016), there was no relationship between curvilinear adherence and outcome in this study—suggesting that the reason previous adherence-outcome studies conflict may not be that the relationship is nonlinear. Rather, perhaps inconsistencies in the literature are so rampant because adherence’s relevance to outcome depends on a variety of contextual factors, such as clinical characteristics, whether treatment was delivered competently, or at what point in treatment adherence was measured (Hauke et al., 2014).

Although previous studies have found poor working alliance to predict attrition (Sharf et al., 2010), early working alliance was neither a moderator of hostile resistance’s relation to drop out, nor an independent predictor of drop out in this study. Similarly, a declining alliance neither mediated hostility’s relation to poor symptomatic change nor was independently predictive of such outcome. That therapist-rated alliance failed to predict outcome is perhaps unsurprising in light of meta-analytic evidence that therapist-rated alliance is less predictive of outcomes than client ratings (Horvath & Symonds, 1991). Why client-rated alliance failed to predict outcome, however, is less clear. This result may owe to this variable’s reduced sample, as questionnaires were missing for almost half of the sample; however, that alliance’s effect size was negligible (r = .05) in this analysis militates against this possibility. Given that other studies have similarly failed to detect a relationship between alliance and outcome in CBT for panic (e.g., Ramnerö & Öst, 2007), additional alliance-outcome research in this population may be warranted.

While unrelated to outcome, a declining alliance was significantly related to hostile resistance: Compared to their non-hostile counterparts, clients who were hostilely resistant in Session 2 were significantly more likely to experience subsequent declines in alliance; likewise, exhibiting hostile resistance in Session 10 was associated with an alliance that had already been on the decline. Although previous reports have linked poor alliance with more resistance to treatment (Westra & Norouzian, 2018) and problematic interpersonal behavior (Weck et al., 2016), more generally, to our knowledge this is the first study to establish a bidirectional association between hostile treatment of therapists and a deteriorating working alliance. If poor alliance is both a cause and a consequence of hostile resistance, findings underscore the importance of therapists’ attentiveness to the therapeutic relationship—particularly if clients’ hostile displays provoke emotional reactions in therapists that lead to alliance ruptures (Muran et al., 2021; Schwartz et al., 2021; Westra et al., 2012).

Limitations of the present study include the restricted range of the off-protocol and working alliance variables and the high volume of missing client-rated alliance questionnaires, which may have limited our ability to detect significant effects with these variables. Moreover, given the small number of hostilely resistant patients (n = 10) within the larger sample, mediation and moderation analyses may have been hampered by limited power. Given the low rate of attrition in this sample, tests of drop out may particularly lack sufficient power and should thus be viewed as tentative. Finally, timing issues may pose a constraint on our ability to interpret tests of working alliance as a mediator. Although measurements of alliance and hostile resistance were taken concurrently in that both were measured at Session 10, the study protocol did not specify whether the working alliance questionnaire should be administered before or after this session. Thus, we are unable to establish that working alliance was always assessed after hostile resistance, potentially inconsistent with tests of mediation.

Clinical Implications and Future Directions

Although this study was unable to pinpoint the mechanisms through which hostile resistance leads to poor outcomes, examining other candidates represents an important future direction. Given that client hostility is associated with worse therapist competence (e.g., Boswell et al., 2013; Foley et al., 1987), future studies might examine whether hostile resistance leads to poor outcome by way of affecting therapists’ ability to deliver the treatment skillfully. Alternatively, hostile resistance may interfere with clients’ ability to engage with the treatment, thereby hampering response (Jungbluth & Shirk, 2009).

Though this study’s overall hypotheses were not supported, findings are nevertheless consistent with the recent movement to shift focus away from faithful and adherent delivery of empirically supported treatments (EST), and instead towards flexible incorporations of theory-informed departures from such protocols (e.g., Constantino et al., 2020). Given that adherence is often unrelated to outcome in ESTs—as was the case in this study—these authors have argued that the traditional emphasis on strict adherence should be broadened to include an approach to ESTs that encourages responsively adjusting treatments according to the client’s individual qualities, context, and momentary interactions with therapists. According to such frameworks, hostile resistance might be viewed as a contextual process marker that should potentially prompt therapists to incorporate theoretically “off brand” interventions (Constantino et al., 2020). Although the use of off-protocol techniques was not a significant mediator in this study, such trans-theoretical borrowing was extremely rare—likely because therapists were discouraged from using non-CBT interventions that might contaminate an arm of a clinical trial. Thus, this study may not provide a true test of whether the use of off-protocol techniques attenuates hostile resistance’s deleterious impact on CBT.

Indeed, while our off-protocol variable captured the use of techniques from a variety of theoretical frameworks (e.g., psychodynamic, dialectical-behavioral), it did not include items for every non-CBT orientation; the absence of motivational interviewing (MI) items is particularly notable, given evidence that MI may be especially protective with other forms of resistance (Aviram et al., 2016). It remains unknown whether MI or another type of protocol departure is indicated in the presence of hostile resistance specifically. Given the finding that hostile resistance is a marker of alliance rupture, it may be that therapists ought to respond to clients’ hostile displays as they would to ruptures more generally—with compassion and curiosity, potentially even linking the patient’s momentary hostility to larger interpersonal patterns (Eubanks et al., 2021). Therapists have reported wanting more guidance on how to manage resistant clients (Wolf & Goldfried, 2014). As such, while this study’s findings are consistent with the view that therapists should strive to be flexible and adaptive when confronted with client hostility, an important direction for future research will be to determine which specific strategies are optimally suited for such moments so that therapists feel better equipped to navigate these rare—yet powerful—events in therapy.

Supplementary Material

Supp 1

Clinical or Methodological Significance:

This study suggests that a declining working alliance may be both a cause and a consequence of client hostility, underscoring the importance of therapists’ attentiveness to the therapeutic relationship. Findings are also consistent with recent calls to shift focus away from faithful and adherent delivery of empirically supported treatments, and towards the flexible delivery of such protocols. Methodologically, this study uses hierarchical linear modeling for tests of mediation and is one of few studies to investigate curvilinear adherence in relation to therapy outcome.

Acknowledgements:

The authors thank adherence coders Armen Bagdasarov, Mayra Monreal, Alec Powell, Sibel Sarac, and Jiyoung Song.

This work was supported by National Institute of Mental Health grants R01 MH70918 to Barbara Milrod, R01 MH070664 to Jacques Barber and Dianne Chambless, and K23 MH123864–01 to Nili Solomonov.

Footnotes

1

The MULTI also has an eighth subscale that measures factors pertinent to all theoretical systems (e.g., demonstrating a belief that the treatment will be helpful). This Common Factors subscale was not included in the off-protocol measure because, by definition, it is relevant to CBT treatment and thus not off-protocol.

2

Recall that it is not possible to assess whether hostility in Session 10 is associated with a subsequent decline in the alliance because alliance was not assessed after Session 10.

Declaration of Ethics Statement (Competing Interests, Funding): The authors report no conflicts of interest.

Contributor Information

Rachel A. Schwartz, University of Pennsylvania, Dept. of Psychiatry, 3535 Market St., Philadelphia, PA 19104, USA.

Kevin S. McCarthy, Chestnut Hill College, Dept. of Psychology, 9601 Germantown Ave., Philadelphia, PA 19118, USA.

Nili Solomonov, Weill Cornell Medicine, Dept. of Psychiatry, 21 Bloomingdale Rd, White Plains, NY, 10065, USA.

Dianne L. Chambless, University of Pennsylvania, Dept. of Psychology, 425 S. University Ave., Philadelphia, PA 19104, USA.

Barbara Milrod, Albert Einstein College of Medicine, Dept. of Psychiatry and Behavioral Sciences, office: 295 Central Park West Office #1, New York City, NY, 10024, USA.

Jacques P. Barber, Adelphi University, Gordon F. Derner School of Psychology, 158 Cambridge Ave., Garden City, NY 11530-0701, USA.

References

  1. Aviram A, & Westra HA (2011). The impact of motivational interviewing on resistance in cognitive behavioural therapy for generalized anxiety disorder. Psychotherapy Research, 21(6), 698–708. 10.1080/10503307.2011.610832 [DOI] [PubMed] [Google Scholar]
  2. Aviram A, Westra HA, Constantino MJ, & Antony MM (2016). Responsive management of early resistance in cognitive-behavioral therapy for generalized anxiety disorder. Journal of consulting and clinical psychology, 783–794. 10.1037/ccp0000100 [DOI] [PubMed]
  3. Bagdasarov A, Chambless DL, Schwartz RA, McCarthy KS, Milrod B, & Barber JP (2018, November). The role of adherence and alliance in changing cognitive misinterpretations during CBT for panic disorder. Association of Behavioral and Cognitive Therapies, Atlanta, GA. [Google Scholar]
  4. Boswell JF, Gallagher MW, Sauer-Zavala SE, Bullis J, Gorman JM, Shear MK, Woods S, & Barlow DH (2013). Patient characteristics and variability in adherence and competence in cognitive-behavioral therapy for panic disorder. Journal of consulting and clinical psychology, 81(3), 443–454. 10.1037/a0031437 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Brown TA, DiNardo PA, & Barlow DH (2004). Anxiety Disorders Interview Schedule Adult Version (ADIS-IV): Client Interview Schedule. Graywind Publications. [Google Scholar]
  6. Cassiello-Robbins C, Conklin LR, Anakwenze U, Gorman JM, Woods SW, Shear MK, & Barlow DH (2015). The effects of aggression on symptom severity and treatment response in a trial of cognitive behavioral therapy for panic disorder. Comprehensive Psychiatry, 60, 1–8. 10.1016/j.comppsych.2015.04.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chambless DL, Milrod B, Porter E, Gallop R, McCarthy KS, Graf E, Rudden M, Sharpless BA, & Barber JP (2017). Prediction and moderation of improvement in cognitive-behavioral and psychodynamic psychotherapy for panic disorder. Journal of consulting and clinical psychology, 85(8), 803–813. 10.1037/ccp0000224 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chambless DL, & Sharpless BA (2011). Adherence ratings for the Panic Control Therapy protocol for Psychotherapies for Panic Disorder Study [Unpublished measure]. Department of Psychology, University of Pennsylvania. [Google Scholar]
  9. Constantino MJ, Coyne AE, & Muir HJ (2020). Evidence-based therapist responsivity to disruptive clinical process. Cognitive and Behavioral Practice. 10.1016/j.cbpra.2020.01.003 [DOI]
  10. Craske MG, Barlow DH, & Meadows EA (2000). Mastery of Your Anxiety and Panic. Therapist Guide for Anxiety, Panic, and Agoraphobia. TherapyWorks. [Google Scholar]
  11. Eubanks CF, Muran JC, & Safran JD (2018). Alliance rupture repair: A meta-analysis. Psychotherapy, 55(4), 508–519. 10.1037/pst0000185 [DOI] [PubMed] [Google Scholar]
  12. Eubanks CF, Sergi J, Samstag LW, & Muran JC (2021). Commentary: Rupture repair as a transtheoretical corrective experience. Journal of Clinical Psychology, 77(2), 457–466. 10.1002/jclp.23117 [DOI] [PubMed] [Google Scholar]
  13. Foley SH, O’Malley S, Rounsaville B, Prusoff BA, & Weissman MM (1987). The relationship of patient difficulty to therapist performance in interpersonal psychotherapy of depression. Journal of Affective Disorders, 12(3), 207–217. 10.1016/0165-0327(87)90029-2 [DOI] [PubMed] [Google Scholar]
  14. Hara KM, Westra HA, Constantino MJ, & Antony MM (2018). The impact of resistance on empathy in CBT for generalized anxiety disorder. Psychotherapy Research, 28(4), 606–615. 10.1080/10503307.2016.1244616 [DOI] [PubMed] [Google Scholar]
  15. Haug T, Nordgreen T, Öst L-G, Tangen T, Kvale G, Hovland OJ, Heiervang ER, & Havik OE (2016). Working alliance and competence as predictors of outcome in cognitive behavioral therapy for social anxiety and panic disorder in adults. Behaviour Research and Therapy, 77, 40–51. 10.1016/j.brat.2015.12.004 [DOI] [PubMed] [Google Scholar]
  16. Hauke C, Gloster AT, Gerlach AL, Richter J, Kircher T, Fehm L, Stoy M, Lang T, Klotsche J, & Einsle F (2014). Standardized treatment manuals: Does adherence matter? Sensoria, 10(2), 1–13. 10.1016/j.janxdis.2016.05.007 [DOI] [Google Scholar]
  17. Horvath AO, & Symonds BD (1991). Relation between working alliance and outcome in psychotherapy: A meta-analysis. Journal of counseling psychology, 38(2), 139–149. 10.1037/0022-0167.38.2.139 [DOI] [Google Scholar]
  18. Huppert JD, Barlow DH, Gorman JM, Shear MK, & Woods SW (2006). The interaction of motivation and therapist adherence predicts outcome in cognitive behavioral therapy for panic disorder: Preliminary findings. Cognitive and Behavioral Practice, 13(3), 198–204. 10.1016/j.cbpra.2005.10.001 [DOI] [Google Scholar]
  19. Jungbluth NJ, & Shirk SR (2009). Therapist strategies for building involvement in cognitive-behavioral therapy for adolescent depression. Journal of consulting and clinical psychology, 77(6), 1179–1184. 10.1037/a0017325 [DOI] [PubMed] [Google Scholar]
  20. Margraf J, & Schneider S (1991). Outcome and active ingredients of cognitive-behavioral treatments for panic disorder. 25th annual meeting of the Association for the Advancement of Behavior Therapy, New York, [Google Scholar]
  21. Mayne TJ, & Ambrose TK (1999). Research review on anger in psychotherapy. Journal of Clinical Psychology, 55(3), 353–363. [DOI] [PubMed] [Google Scholar]
  22. McCarthy KS, & Barber JP (2009). The multitheoretical list of therapeutic interventions (MULTI): Initial report. Psychotherapy Research, 19(1), 96–113. 10.1080/10503300802524343 [DOI] [PubMed] [Google Scholar]
  23. McCarthy KS, Keefe JR, & Barber JP (2016). Goldilocks on the couch: Moderate levels of psychodynamic and process-experiential technique predict outcome in psychodynamic therapy. Psychotherapy Research, 26(3), 307–317. 10.1080/10503307.2014.973921 [DOI] [PubMed] [Google Scholar]
  24. Milrod B, Chambless DL, Gallop R, Busch FN, Schwalberg M, McCarthy KS, Gross C, Sharpless BA, Leon AC, & Barber JP (2016). Psychotherapies for panic disorder: A tale of two sites. The Journal of Clinical Psychiatry, 77(7), 927–935. 10.4088/JCP.14m09507 [DOI] [PubMed] [Google Scholar]
  25. Muran JC, Eubanks CF, & Samstag LW (2021). One more time with less jargon: An introduction to “Rupture Repair in Practice”. Journal of Clinical Psychology, 77(2), 361–368. 10.1002/jclp.23105 [DOI] [PubMed] [Google Scholar]
  26. Ramnerö J, & Öst L-G (2007). Therapists’ and clients’ perception of each other and working alliance in the behavioral treatment of panic disorder and agoraphobia. Psychotherapy Research, 17(3), 320–328. 10.1080/10503300600650852 [DOI] [Google Scholar]
  27. Schwartz RA, Chambless DL, McCarthy KS, Milrod B, & Barber JP (2018). Client resistance predicts outcomes in cognitive–behavioral therapy for panic disorder. Psychotherapy Research, 1–13. 10.1080/10503307.2018.1504174 [DOI] [PubMed]
  28. Schwartz RA, Chambless DL, Milrod B, & Barber JP (2021). Patient, therapist, and relational antecedents of hostile resistance in cognitive-behavioral therapy for panic disorder: A qualitative investigation. Psychotherapy. 10.1037/pst0000308 [DOI] [PMC free article] [PubMed]
  29. Sharf J, Primavera LH, & Diener MJ (2010). Dropout and therapeutic alliance: A meta-analysis of adult individual psychotherapy. Psychotherapy: Theory, Research, Practice, Training, 47(4), 637–645. 10.1037/a0021175 [DOI] [PubMed] [Google Scholar]
  30. Shear MK, Brown TA, Barlow DH, Money R, Sholomskas DE, Woods SW, Gorman JM, & Papp LA (1997). Multicenter collaborative panic disorder severity scale. American Journal of Psychiatry, 154(11), 1571–1575. 10.1176/ajp.154.11.1571 [DOI] [PubMed] [Google Scholar]
  31. Solomonov N, Falkenström F, Gorman BS, McCarthy KS, Milrod B, Rudden MG, Chambless DL, & Barber JP (2020). Differential effects of alliance and techniques on Panic-Specific Reflective Function and misinterpretation of bodily sensations in two treatments for panic. Psychotherapy Research, 30(1), 97–111. 10.1080/10503307.2019.1585591 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Strunk DR, Brotman MA, & DeRubeis RJ (2010). The process of change in cognitive therapy for depression: Predictors of early inter-session symptom gains. Behaviour Research and Therapy, 48(7), 599–606. 10.1016/j.brat.2010.03.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Tracey TJ, & Kokotovic AM (1989). Factor structure of the Working Alliance Inventory. Psychological Assessment, 1, 207–210. 10.1037/1040-3590.1.3.207 [DOI] [Google Scholar]
  34. Webb CA, DeRubeis RJ, Amsterdam JD, Shelton RC, Hollon SD, & Dimidjian S (2011). Two aspects of the therapeutic alliance: Differential relations with depressive symptom change. Journal of consulting and clinical psychology, 79(3), 279–283. 10.1037/a0023252 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Webb CA, DeRubeis RJ, & Barber JP (2010). Therapist adherence/competence and treatment outcome: A meta-analytic review. Journal of consulting and clinical psychology, 78(2), 200–211. 10.1037/a0018912 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Weck F, Grikscheit F, Höfling V, Kordt A, Hamm AO, Gerlach AL, Alpers GW, Arolt V, Kircher T, & Pauli P (2016). The role of treatment delivery factors in exposure-based cognitive behavioral therapy for panic disorder with agoraphobia. Journal of Anxiety Disorders, 42, 10–18. 10.7790/sa.v0i0.362 [DOI] [PubMed] [Google Scholar]
  37. Westra HA, Aviram A, Connors L, Kertes A, & Ahmed M (2012). Therapist emotional reactions and client resistance in cognitive behavioral therapy. Psychotherapy, 49(2), 163–172. 10.1037/a0023200 [DOI] [PubMed] [Google Scholar]
  38. Westra HA, Aviram A, Kertes A, Ahmed M, & Connors L (2009). Manual for coding resistance in CBT for anxiety [Unpublished manual]. Department of Psychology, York University. [Google Scholar]
  39. Westra HA, & Norouzian N (2018). Using motivational interviewing to manage process markers of ambivalence and resistance in cognitive behavioral therapy. Cognitive Therapy and Research, 1–11. 10.1007/s10608-017-9857-6 [DOI]
  40. Wolf AW, & Goldfried MR (2014). Clinical experiences in using cognitive-behavior therapy to treat panic disorder. Behavior Therapy, 45(1), 36–46. 10.1016/j.beth.2013.10.002 [DOI] [PubMed] [Google Scholar]
  41. Zickgraf HF, Chambless DL, McCarthy KS, Gallop R, Sharpless BA, Milrod BL, & Barber JP (2015). Interpersonal factors are associated with lower therapist adherence in cognitive-behavioural therapy for panic disorder. Clinical Psychology & Psychotherapy, 272–284. 10.1002/cpp.1955 [DOI] [PMC free article] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp 1
NIHMS1784954-supplement-Supp_1.docx (15.4KB, docx)

RESOURCES