A multiple-baseline evaluation of acceptance and commitment therapy focused on repetitive negative thinking in panic disorder

Context and Objectives

Panic disorder is a relatively common condition affecting both clinical and nonclinical populations, with a cross-national lifetime prevalence estimated at 1.7%. Although cognitive-behavioral therapy (CBT) is widely regarded as the treatment of choice for panic disorder, it typically implements interceptive exposure techniques with a standard duration of 12-15 sessions. Despite these effective interventions, significant limitations exist: only 48% of participants in exposure treatment achieve panic remission at post-treatment assessments, and approximately 20% show therapy dropout.

A central problem is the limited acceptability of exposure exercises among clients and therapists. Researchers have identified three complementary alternatives: (a) introducing acceptance components that increase the acceptability of exposure techniques, (b) developing psychological interventions that do not include explicit exposure exercises while maintaining efficacy, and (c) developing and testing brief interventions for panic disorder that produce results similar to CBT but in less than half the therapeutic contact time.

Acceptance and commitment therapy (ACT) is a contextual behavioral therapy that promotes psychological flexibility in relation to internal experiences, including thoughts, feelings, and bodily sensations. Preliminary research suggests that brief ACT protocols focused on dismantling the counterproductive pattern of repetitive negative thinking (RNT) may be as efficacious as standard CBT protocols but without the need to include explicit exposure exercises.

The present study explores the efficacy of a brief 4-session RNT-focused ACT protocol in three individuals with a primary diagnosis of panic disorder. The intervention specifically targets: (a) general RNT in the form of worry and rumination, (b) worry regarding the occurrence of panic attacks, and (c) hypervigilance to bodily sensations. The protocol deliberately excludes explicit exposure exercises as an alternative to the limitations found in exposure-based interventions.

Method

Participants

Recruitment was conducted through social media. Five participants were screened who met the following inclusion criteria: (a) being of legal age, (b) meeting diagnostic criteria for panic disorder as a primary diagnosis according to the Mini-International Neuropsychiatric Interview (MINI), (c) having experienced symptoms for at least the last three months, and (d) agreeing not to initiate additional therapy during the study. Exclusion criteria were: (a) experiencing a psychotic disorder, abuse of psychoactive substances, or severe medical illness, and (b) requiring immediate treatment for severe depression or suicidal behavior as assessed by the MINI.

Two of the five initially recruited participants showed a significant trend of improvement at baseline. As a result, their data were excluded from the study as it would not be determined whether the intervention caused the improvement, resulting in a final sample of three participants.

Participant 1 (P1): 24-year-old female with bachelor's education, working as a photographer. She presented with fear of loneliness and hierarchical triggers for worry/rumination including: worry, rumination, avoiding eating, and staying awake. Her MINI diagnosis included panic disorder and depression (recurrent). P1 also met criteria for unipolar depression.

Participant 2 (P2): 49-year-old male with bachelor's education, working as a flight attendant. His hierarchical trigger for worry and rumination was fear of failure, manifesting in worry, rumination, going out with friends, distraction, and altered physical exercise, along with sleep difficulties. His MINI diagnosis included panic disorder and depression. P2 also met criteria for unipolar depression.

Participant 3 (P3): 30-year-old male with master's education as a graduate student. His hierarchical trigger for worry and rumination was fear of failure, manifesting in worry, rumination, breathing exercises, writing down thoughts, searching for information on the internet, and "ignoring" the thoughts. His MINI diagnosis included panic disorder and depression. P3 also met criteria for unipolar depression.

Design

A randomized, multiple-baseline design across participants was implemented. Participants were randomized to receive the intervention after 3-5 weeks of baseline. Randomization was conducted through www.randomizer.org. The independent variable consisted of a protocol of 4 weekly sessions of approximately 60 minutes each.

The multiple-baseline design is an appropriate approach for evaluating interventions in small case studies, allowing the assessment of causal changes through the staggered introduction of the intervention across participants.

Intervention

The RNT-focused ACT protocol consists of 4 weekly sessions of approximately 60 minutes each. The sessions were designed based on previous RNT-focused ACT protocols that have shown a high degree of efficacy in depression and generalized anxiety disorders. The protocol was implemented by the first author, who received training in applying the protocol over ten prior sessions, using role-play, modeling, and feedback, with a total of 128 hours of theoretical and practical ACT training with the last author during her doctoral training. Additionally, the therapist received supervision during the implementation of the interventions by the second author and the last author.

Contents of the RNT-focused ACT protocol:

Session 1: Introduction of the intervention rationale and functional analysis of the inflexible psychological pattern with central focus on worry/rumination. Identification of hierarchical triggers, main contents of the repetitive thought chain, and additional experiential avoidance strategies. Identification and amplification of the consequences of the inflexible pattern and opening of a flexible alternative. Discrimination training on the process of worry/rumination and its contents, contact with its consequences, and training in distancing from triggers. Physical metaphor of "pushing worry/rumination" or "circling the chair." Audio 1: Exercise aimed at developing the skill to differentiate between contacting RNT or taking distance from triggers while choosing valued direction.

Session 2: Review of progress and difficulties experienced since Session 1. Training in multiple examples of distancing from worry/rumination including: (a) Thought zoom exercise, (b) Free association exercise, (c) Conscious fantasy and worry exercise, and (d) "I can't possibly..." exercise. Audio 2: Exercise of observing thoughts in balloons from a hierarchical perspective.

Session 3: Review of progress and difficulties experienced since Session 2. Observer exercise (modified to identify triggers of worry/rumination and contact with values). Clarification of values and identification of valued actions. Garden metaphor. Audio 3: Life movie exercise (contact with values and adversities).

Session 4: Review of progress and difficulties experienced since Session 3. General review of the work done in the intervention. Exploration of the consequences of monitoring vs. not monitoring. Experiential experience of monitoring physical sensations as habitual behavior. Identification of valued actions. Closing of the intervention.

Outcome Measures

Depression, Anxiety and Stress Scale-21 (DASS-21): A 21-item instrument answered on a four-point Likert scale (3 = it happened a lot or most of the time; 0 = it did not happen to me). Contains three subscales: Depression, Anxiety, and Stress. The sum of the scores provides an overall measure of emotional symptoms. The internal consistency of the Colombian validation is excellent with Cronbach's alpha of .93, with a hierarchical factor structure consisting of a general factor and three second-order factors.

Penn State Worry Questionnaire-11 (PSWQ-11): An 11-item instrument answered on a five-point Likert scale (5 = very much; 1 = not at all). Measures the severity of worry associated with generalized anxiety disorder. Shows excellent internal consistency (Cronbach's alpha .95) with unidimensional structure. Scores above 38 are considered high.

Self-report of panic attacks: A self-report was designed to measure the frequency of panic attacks and their associated physiological symptoms (palpitations, sweating, tremors, choking sensation, chest tightness, nausea, abdominal pain, tingling, numbness, chills, fear of dying).

Process Measures

Acceptance and Action Questionnaire-II (AAQ-II): An instrument measuring experiential avoidance, composed of 7 items answered on a seven-point Likert scale (7 = always true; 1 = never true). Higher scores indicate greater experiential avoidance. Showed excellent internal consistency (Cronbach's alpha .90) with unidimensional structure. Scores for nonclinical participants typically range around 18-23 points, while those for clinical participants average above 29 points.

Cognitive Fusion Questionnaire (CFQ): A scale measuring cognitive fusion through 7 items answered on a seven-point Likert scale (7 = always true; 1 = never true). Higher scores indicate greater cognitive fusion. The Colombian validation showed excellent internal consistency with Cronbach's alpha of .93 with unidimensional structure. Scores for nonclinical participants range between 20-24 points, while those for clinical participants typically exceed 29 points.

Valuing Questionnaire (VQ): A 10-item questionnaire that assesses values lived during the previous week, answered on a seven-point Likert scale (6 = completely true; 0 = not at all true). It has two subscales: Progress (clear awareness of what is personally important and perseverance) and Obstruction (disruption of valued life due to avoidance of unwanted experiences and distraction from values). The Spanish version showed adequate psychometric properties with a two-factor structure. The mean scores obtained for the general population in Colombia were 19.5 (SD = 6.43) for Progress and 11.7 (SD = 6.88) for Obstruction, while for a clinical sample (N = 235) they were 17.28 (SD = 6.98) and 15.25 (SD = 7.53), respectively.

Procedure

Phase 1. Design of the intervention protocol and researcher training: The intervention protocol was designed based on previous RNT-focused ACT protocols. The first author was trained in applying the protocol over ten prior sessions using role-play, modeling, and feedback. Previously, the therapist had received theoretical and practical training in ACT equivalent to 128 hours with the last author during her master's studies. Additionally, the therapist received supervision during the implementation of the interventions.

Phase 2. Recruitment of participants: An advertisement explaining the research was distributed through social media. Those who showed interest in participating were interviewed to assess whether they met the study's inclusion and exclusion criteria. This interview consisted of the administration of the MINI and the questionnaires mentioned above. Participants who met the criteria signed an informed consent form in which they provided explicit approval to participate in the research. Participants who did not meet the inclusion criteria were oriented and referred. Baseline data were collected for each participant. Baseline periods ranged from 3 to 5 weeks. Participants were randomly assigned to one of the baseline length options through www.randomizer.org. Additionally, participants were trained in completing the self-report that was completed upon presentation of a panic attack. The questionnaires mentioned above were administered weekly throughout the study.

Phase 3. Application of the intervention protocol: The intervention protocol was conducted weekly through 4 sessions of approximately 60 minutes each.

Phase 4. Closure of the research: Participants responded to measures during the following month each week. Then, they responded each month until completing the 3-month follow-up. Once the follow-up was completed, each participant was interviewed and then the study ended.

Data Analysis

Data were analyzed at the individual level and globally for the three participants. Individual analysis was performed through the nonparametric Tau-U test. Tau-U is a non-overlap effect size between baseline and intervention data. As a nonparametric test, Tau-U does not require compliance with the assumptions of normality, constant variance, and independence of measurements.

Tau-U was derived from Kendall rank correlation and Mann Whitney U and can correct for significant trends during baseline. The range of Tau-U values is between -1 and 1 and can be interpreted as the percentage of data that improves through the baseline and intervention phases. For convenience, all effect sizes favorable to the intervention phase are presented in this study, regardless of whether scores should decrease or increase.

After identifying the magnitude of change in each variable, the presence of clinically significant changes was identified following a proposal similar to the one presented by de Vries et al. (2016). Specifically, to indicate the presence of clinically significant changes, it was required: (a) that the Tau-U value be significantly greater than zero, and (b) to cross a cutoff point on the last treatment measure that placed the participant closer to the nonclinical than the clinical population mean. To test the latter criterion, data obtained across samples in validation studies in Colombia were used.

To obtain an overall appreciation of the treatment effect size, the multiple-baseline effect size developed by Pustejovsky et al. (2014) was calculated using the scnlm package for R. This analysis provides a standardized mean difference that shares metrics similar to Cohen's d effect size frequently used in group designs. This type of analysis yields an effect size comparable across different types of designs. This study used the first mode because some analyses showed convergence problems with the second mode, which is mathematically more complex.

Results

Outcome Measures

Figure 1 presents the evolution of participants' scores on outcome measures throughout the study. Visual analysis reveals relatively stable baseline trends across most participants and measures.

P1 experienced rapid changes in emotional symptoms and pathological worry after introducing the intervention, although there was an increase at some points during the follow-up period. Table 3 shows that effect sizes were close to 1 and statistically significant for all variables (DASS-Total = 0.92, p = .005; DASS-Depression = 0.92, p = .005; DASS-Anxiety = 0.82, p = .012; DASS-Stress = 0.94, p = .004; PSWQ-11 = 1.00, p = .002), showing clinically significant changes in all cases. Regarding the frequency of panic attacks, P1 reported experiencing three attacks during the four weeks of baseline and none after the introduction of the intervention.

P2 experienced more gradual changes after the introduction of the intervention. It should be noted that, due to work-related circumstances, Session 3 was conducted three weeks after Session 2. Table 3 shows that effect sizes were statistically significant for all variables except DASS-Anxiety (DASS-Total = 0.97, p = .014; DASS-Depression = 0.83, p = .035; DASS-Anxiety = 0.73, p = .060; DASS-Stress = 0.97, p = .014; PSWQ-11 = 1.00, p = .011). Changes were clinically significant in all variables in which Tau-U values had been statistically significant. Regarding the frequency of panic attacks, P2 reported experiencing two attacks during the two weeks of baseline and only two attacks after introducing the intervention, both occurring in the week between Session 1 and Session 2.

P3 showed changes in all outcome variables that were maintained or increased during the follow-up period. Table 3 shows that effect sizes were statistically significant for all variables except DASS-Depression (DASS-Total = 1.00, p = .005; DASS-Depression = 0.38, p = .290; DASS-Anxiety = 1.00, p = .005; DASS-Stress = 0.95, p = .007; PSWQ-11 = 1.00, p = .005). Changes were clinically significant in all variables in which Tau-U values had been statistically significant. Finally, P3 reported two panic attacks during the three weeks of baseline and none after the introduction of the intervention.

Process Measures

Figure 2 presents the evolution of participants' scores on process measures throughout the study. Again, visual analysis reveals relatively stable baseline trends across most participants and measures.

P1 experienced large and immediate changes in experiential avoidance and cognitive fusion after introducing the intervention. These changes were maintained except at the one-week follow-up. The change in VQ-Obstruction was also immediate and large, although the change in VQ-Progress was moderate. Table 3 shows that effect sizes were close to 1 and statistically significant for all variables (AAQ-II = 1.00, p = .002; CFQ = 1.00, p = .002; VQ-Progress = 0.94, p = .004; VQ-Obstruction = 0.96, p = .003), showing clinically significant changes in all cases.

P2 also improved in all variables, although the effect of the intervention was more gradual. Results remained stable during the follow-up period. Table 3 shows that effect sizes were statistically significant for all variables (AAQ-II = 0.90, p = .023; CFQ = 0.80, p = .043; VQ-Progress = 1.00, p = .011; VQ-Obstruction = 0.93, p = .018), showing clinically significant changes in all variables where Tau-U values had been statistically significant.

P3 showed stepwise changes in all process variables. According to Table 3, all effect sizes were statistically significant (AAQ-II = 1.00, p = .005; CFQ = 0.95, p = .007; VQ-Progress = 0.88, p = .013; VQ-Obstruction = 0.88, p = .013). Changes were clinically significant in all variables except VQ-Progress because the participant was closer to the clinical score than to the nonclinical score at follow-up.

Standardized Mean Difference

Table 4 presents the overall effect sizes obtained by the intervention. For outcome measures, all effect sizes were large and significant (ranging from d = 1.45 for DASS-Depression to d = 2.48 for DASS-Total). Similarly, the intervention obtained very large effect sizes for process measures (ranging from d = 2.43 for VQ-Obstruction to d = 3.58 for CFQ), except for the VQ-Progress variable (d = 0.72). All effect sizes were statistically significant.

Discussion and Conclusions

The present study examined the efficacy of a brief RNT-focused ACT intervention in individuals with a primary diagnosis of panic disorder. A 4-session RNT-focused ACT protocol was designed based on previous protocols that had shown a high degree of efficacy in depression and generalized anxiety disorder. Although the ACT approach is consistent with ACT, the protocol deliberately excluded explicit exposure exercises as an alternative to the limitations found in exposure in terms of acceptability for clients and therapists.

The intervention showed a high degree of efficacy in reducing emotional symptoms and pathological worry. All participants showed significant changes in these variables except DASS-Depression (P3 showed no significant change) and DASS-Anxiety (P2). Additionally, the frequency of panic attacks was reduced to zero in P1 and P3, while P2 had only two attacks after introducing the intervention. These attacks occurred between Sessions 1 and 2. Therefore, it can be stated that after the intervention, panic attacks were completely suppressed in all three participants during the 3-month follow-up.

Additional evidence for the high efficacy of the intervention was the large effect sizes found in the reduction of emotional symptomatology (d = 2.48, 95% CI [1.50, 3.59]) and pathological worry (d = 2.36, 95% CI [1.01, 3.72]).

Results on process measures were also encouraging. All three participants showed clinically significant changes in experiential avoidance, cognitive fusion, and values obstruction, while P1 and P2 also showed changes in values progress. Overall effect sizes were very large for experiential avoidance (d = 3.26), cognitive fusion (d = 3.58), and values obstruction (d = 2.43). The overall effect size for values progress was close to large (d = 0.72). This difference between the effect size in progress and obstruction in values is consistent with previous studies and could indicate that more extensive work would be needed to identify and engage in valued actions.

The present study replicates the promising findings found in previous studies primarily in depression and generalized anxiety disorders. While previous studies had participants diagnosed with panic disorder, this is the first study to analyze the effect of a brief RNT-focused ACT protocol for treating panic disorder as a primary diagnosis.

As mentioned above, it is noteworthy that the protocol used in this study did not include explicit exposure exercises. Instead, the intervention focused on dismantling dysfunctional patterns of RNT, increasing psychological flexibility, and reducing hypervigilance to bodily sensations as a common form of RNT in people suffering from panic disorder. Thus, the intervention was effective despite not including an explicit exposure component. In this sense, the present research joins initial ACT studies that showed efficacy in anxiety disorders despite not having explicit exposure components in their protocols. The absence of exposure may result in greater acceptability of the intervention for clients and therapists who find it too threatening.

The present study should be interpreted considering some limitations. First, the sample used was small, and further replications are necessary to confirm the efficacy of this brief RNT-focused ACT intervention. In this respect, five participants were initially recruited, but two showed a marked trend of improvement at baseline, making it impossible for us to determine whether the intervention caused the improvement (both ended the study with low symptomatology scores). Second, the effect of the intervention was only evaluated through self-report and psychological instruments. Future research could analyze the effect of the intervention in a clinical interview and behavioral tests of tolerance to physiological sensations similar to those experienced during a panic attack (e.g., CO2-air breathing challenges). Third, a nonconcurrent multiple-baseline design was conducted because the recruitment process was spread over several weeks. This type of design has lower internal validity than concurrent designs, although additional limitations should not be particularly problematic in this type of study. Fourth, the three participants have higher education level, which reduces the generalization of the results found.

Despite the limitations listed above, this initial study shows that RNT-focused ACT interventions may constitute an alternative brief intervention model for panic disorder even without conducting explicit exposure. Furthermore, the non-inclusion of exposure may result in greater acceptability of the intervention for clients and therapists who find it too threatening.

Significance and Contribution

This study provides compelling evidence for a brief, exposure-free ACT intervention for panic disorder. A 4-session protocol targeting psychological flexibility and reduction of maladaptive thought patterns produced clinically significant symptom reduction across all participants, with complete or near-complete suppression of panic attacks and very large effect sizes (DASS-Total d = 2.48; CFQ d = 3.58; AAQ-II d = 3.26). Notably, clinically significant improvements were achieved without including traditional interoceptive exposure exercises, challenging assumptions about exposure necessity and suggesting that psychological flexibility enhancement may represent an effective alternative mechanism. The protocol demonstrated specificity in addressing panic-relevant processes including general worry, panic-specific worry, and hypervigilance to bodily sensations. These findings have important practical implications: demonstrated efficacy in just 4 brief sessions addresses concerns about cost, accessibility, and treatment acceptability that have limited broader implementation of psychological interventions for panic disorder. Changes in process variables parallel clinical improvements, supporting ACT's theoretical mechanisms of change.

---

---