A multiple-baseline evaluation of a brief acceptance and commitment therapy protocol focused on repetitive negative thinking for moderate emotional disorders

Full reference: Ruiz, F. J., Flórez, C. L., García-Martín, M. B., Monroy-Cifuentes, A., Barreto-Montero, K., García-Beltrán, D. M., Riaño-Hernández, D., Sierra, M. A., Suárez-Falcón, J. C., Cardona-Betancourt, V., & Gil-Luciano, B. (2018). A multiple-baseline evaluation of a brief acceptance and commitment therapy protocol focused on repetitive negative thinking for moderate emotional disorders. Journal of Contextual Behavioral Science, 9, 1–14. https://doi.org/10.1016/j.jcbs.2018.04.004

Study type: Empirical research; Multiple-baseline single-case experimental design (SCED); Case study; Therapeutic trial

Background and objectives

Repetitive negative thinking (RNT), in the form of worry and rumination, has been identified as a particularly counterproductive transdiagnostic process implicated in the onset and maintenance of emotional disorders. RNT is characterized by difficulty disengaging from thoughts, memories, and worries across at least two life domains. This transdiagnostic characteristic appears in multiple conditions: generalized anxiety disorder, major depressive disorder, posttraumatic stress disorder, social anxiety disorder, and personality disorders, among others.

Acceptance and commitment therapy (ACT) is a psychotherapeutic approach grounded in relational frame theory and contextual behavioral science, aiming to develop psychological flexibility: the capacity to contact present-moment experiences in an open, curious, and undefended manner, and to act consistently with personal values and goals despite the presence of uncomfortable thoughts, emotions, or sensations. ACT emphasizes acceptance and commitment over symptom control techniques.

Previous studies have suggested ACT is effective in reducing depression and anxiety symptoms. However, most ACT research has examined the effect of complete protocols rather than identifying which specific treatment components are responsible for observed changes. The objective of this study was to evaluate whether a brief RNT-focused ACT protocol would be sufficient to produce significant changes in worry, rumination, and emotional symptoms in individuals experiencing moderate emotional disorders. This research represents an important step in developing and analyzing the effect of specific RNT-focused ACT protocols.

Method

Participants

The study included 10 adult participants (4 men, mean age M = 23.2, SD = 4.24) recruited through social media advertisements with initial screening questions such as "Do you spend too much time distressed thinking about the past or future?" and "Would you like to focus more on things that are important to you?" Inclusion criteria were: (a) age 18 years or older; (b) at least 6 months experiencing thoughts, memories, and worries with significant life interference; (c) significant interference from thoughts, memories, and worries in at least two life domains; (d) General Health Questionnaire-12 (GHQ-12) scores between 15 and 25; (e) Depression, Anxiety and Stress Scales-21 (DASS-21) scores between 15 and 25.

Of 71 potentially interested participants, recruitment screening excluded individuals who did not meet full criteria. One participant showed significant improvement trends during baseline visual analysis and was ethically excluded because presence of significant improvement trends would not establish the intervention effect conclusively. Twenty-six participants met criteria but showed statistically significant improvement during baseline according to Theil-Sen slope testing and were excluded because this study aimed to examine a brief ACT protocol's effect.

The final sample included 10 participants who met all inclusion criteria and completed the multiple-baseline design. One participant showed significant baseline improvement during later analysis and was excluded from statistical analyses, resulting in 9 participants whose data are reported.

Experimental design

The study employed a nonconcurrent multiple-baseline design across participants following recent guidelines. The design involved 5-7 weeks of baseline data collection depending on participant availability. Protocol sessions were separated by one month so first-session effects could be analyzed. A 12-week follow-up was implemented after protocol completion.

Dependent variables were divided into primary and secondary outcome measures. Primary outcomes were emotional symptoms and psychological distress scores. Secondary outcomes were process measures including experiential avoidance, cognitive fusion, pathological worry, repetitive thinking, and valued living.

Materials and apparatus

Depression, Anxiety, and Stress Scales-21 (DASS-21). A 21-item self-report instrument with 4-point Likert scale assessing depression, anxiety, and stress symptoms experienced during the past week. The instrument contains three subscales (Depression, Anxiety, Stress) and has shown good psychometric properties and discriminant validity in clinical and community samples. The Spanish version was used in this study.

General Health Questionnaire-12 (GHQ-12). A 12-item self-report instrument with 4-point Likert scale frequently used for screening psychological distress. Respondents indicate the degree to which they have recently experienced common symptoms of psychological distress. The 0-3 Likert scoring method was used. The GHQ-12 showed good psychometric properties in Colombia.

Acceptance and Action Questionnaire-II (AAQ-II). A 7-item self-report instrument with 7-point Likert scale reflecting experiential avoidance and inability to be present and behave according to valued-directed actions. The Spanish version of the AAQ-II has shown good psychometric properties and one-factor structure in clinical and nonclinical Colombian samples.

Cognitive Fusion Questionnaire (CFQ). A 7-item instrument with 7-point Likert scale measuring cognitive fusion (the tendency for thoughts to seem "true" or causally generative of behavior). The Spanish version demonstrated excellent psychometric properties in Colombia.

Perseverative Thinking Questionnaire (PTQ). A 15-item instrument with 4-point Likert scale designed to assess rumination tendencies when facing negative psychological experiences or problems. Unlike other measures, the PTQ is content-independent. The PTQ demonstrated excellent internal consistency and adequate convergent and discriminant validity.

Penn State Worry Questionnaire-11 (PSWQ-11). A 16-item self-report instrument with 5-point Likert scale designed to assess the persistent and nonspecific worry characteristic of GAD. An 11-item abbreviated version was used, recommended by Sandín et al. (2009) because reverse-worry items were difficult for Spanish speakers. The PSWQ-11 demonstrated excellent internal consistency and discriminant validity. A threshold score of 37/38 was considered adequate for identifying severe GAD, while 32/33 was adequate for moderate GAD.

Valued Questionnaire (VQ). A 10-item instrument with 6-point Likert scale assessing two subscales: Progress (enactment of values including clear awareness of personal importance and perseverance) and Obstruction (valued living impairment due to avoidance of uncomfortable experiences and distraction from values). The Spanish version demonstrated good psychometric properties.

The ACT protocol was implemented in individual format by 5 therapists. All therapists received approximately 40 hours of ACT training and approximately 12 hours of protocol-specific training. Blinding procedures were not implemented because the study involved only one intervention, and outcome measures were applied through automated internet-based emails.

Procedure

The study was conducted at the Clinical Psychology Clinic of a Colombian university. Study procedures were approved by the Research Ethics Committee. All measures were administered online through Qualtrics. Participants showing interest and meeting initial inclusion criteria were invited to a 40-minute assessment session where intervention rationale was presented, readiness for engagement was evaluated, and informed consent was obtained.

The ACT protocol consisted of two 60-minute sessions. The protocol was based on the RNT account of psychological flexibility and self development. The first session aimed to present intervention rationale, identify main worry/rumination triggers, promote realization of unhelpful experiential avoidance strategies, and identify RNT process and defusion training. The second session reviewed advances in identifying RNT triggers and defusion exercises, and identified valued actions to engage in during the day and beyond. At end of Session 2, participants received 5 audio files (approximately 30 minutes) to practice what was worked on during sessions.

Data analysis

Statistical analyses conducted used a visual analysis approach following single-case experimental design (SCED) guidelines (Parker & Vannest, 2012). Results were first graphically analyzed, then statistical analyses for SCED were selected and computed. Although some participants showed considerable variability in emotional symptoms during baseline, data showed no significant improvement trends. Presence of variability is not problematic but is frequent in clinical settings. This was not the case in the current study.

Theil-Sen slope analysis was conducted using online calculator software (Vannest, Parker, & Gonen, 2011). Only one participant showed statistically significant improvement trends during baseline; therefore, this participant was excluded from protocol implementation.

A Bayesian JZS + AR analysis (Rouder, Speckman, Sun, Morey, & Iverson, 2009) with an autoregressive (AR1) model was executed. This provides a Bayes factor from evidence in both phases. Bayes factors were interpreted according to guidelines by Jeffreys (1961) and Wagenmakers (2007): 1 = no evidence of treatment effect; 1-3 = anecdotal evidence; 3-10 = moderate evidence; 10-30 = strong evidence; 30-100 = very strong evidence; >100 = extreme evidence.

Clinically significant change (CSC) analysis was conducted following de Vries and Morey (2013) and Hedges, Pustejovsky, and Shadish (2012) procedures. The standardized mean difference effect size for SCED was computed for multiple-baseline designs. This standardized mean difference effect facilitates direct comparison and integration of results across single-case designs. The d-statistic was computed at 4-week follow-up (before second session implementation) and at 12-week follow-up.

Results

Within-participant results

Figure 1 shows the evolution of scores on primary outcome measures (DASS-21 total and GHQ-12 scores). Visual inspection shows 9 of 10 participants demonstrated clinically significant improvement in emotional symptoms across baseline, 7-day intervention data, and 12-week follow-up data. Of these 9 participants, those with baseline data that showed stable emotional symptom levels were analyzed. One participant showed significant baseline improvement trends during visual analysis evaluation and was excluded from the study for ethical reasons because presence of deteriorating trends would not establish intervention effect conclusively.

Between-participant results

Figure 2 shows mean scores across participants on all study variables. After intervention, participants showed better scores than nonclinical samples on all measures. Mean scores were below nonclinical samples on all measures during baseline. Specifically, mean depression and anxiety scores during baseline were in the moderate range per the DASS-21. Mean GHQ-12 scores were also in the clinical range. Protocol sessions were separated by one month so first-session changes could be analyzed. Participants continued improving, although to a lesser degree, after second session implementation 4 weeks later.

Primary outcome measures:

According to Bayesian CSC analysis on DASS-Total scores from baseline to 12-week follow-up, P1, P5, and P10 showed very strong evidence of intervention effect (Bayes Factor > 100). Strong evidence was found for P3 and P4 (Bayes Factor = 8.62, 4.49 respectively). Anecdotal evidence was found for P2, P7, P9 (Bayes Factor = 1.02, 0.72, 0.81 respectively). No participant showed evidence of deterioration.

For GHQ-12 (psychological distress), very strong evidence of intervention effect was found in P1, P3, P4, P6, P7, P8, P9, and P10 (Bayes Factor > 100). Strong evidence was found for P2 (Bayes Factor = 4.08). Very large effect sizes were found on univariate intervention effect metrics (Table 3), with DASS-Total d = 2.44 (95% CI [1.5, 3.5]), GHQ-12 d = 2.68 (95% CI [1.7, 3.8]), pathological worry d = 3.14 (95% CI [1.9, 4.4]), experiential avoidance d = 1.32 (95% CI [0.6, 2.2]), cognitive fusion d = 2.15 (95% CI [1.1, 3.3]), repetitive thinking d = 2.51 (95% CI [1.5, 3.5]), and valued living d = 1.34 (95% CI [0.4, 2.3]). No adverse events were found. RNT-focused protocols deserve further empirical tests.

Secondary outcome measures (process measures):

With respect to secondary outcome measures, all participants showed substantial evidence of intervention effect. Specifically, all participants showed substantial evidence of intervention effect for CFQ (experiential avoidance), 8 participants for valued living progress, and 7 participants for valued living obstruction. No participant showed evidence of deterioration on any primary or secondary outcome measure.

The RNT-focused ACT protocol was effective in decreasing emotional symptoms to some degree across all participants. Table 3 shows effect sizes in the Bayesian JZS + AR analysis and CSC analysis. Table 4 shows mean scores on all measures during baseline, 4-week follow-up, 8-week follow-up, and 12-week follow-up, and effect sizes at 4-week and 12-week follow-ups. After the intervention, all participants showed improvement. Mean scores were below nonclinical samples but after intervention all participants moved closer to nonclinical ranges on most measures.

Discussion and conclusions

This study followed the descriptive conceptualization of RNT in terms of psychological flexibility and the formation of self. To investigate this topic, Riano-Hernández et al. (2016) explored whether a one-session RNT-focused ACT protocol would be sufficient. Results showed significant reductions in at least three of four RNT measures during testing. Given this study's results were larger on three of four RNT measures and emotional symptoms, future research could analyze the effect of a 2-session RNT-focused ACT protocol in participants suffering moderate to severe emotional disorders.

The 10 participants who were entangled with thoughts, memories, and worries for at least 6 months showing interference in at least two life domains participated in this nonconcurrent multiple-baseline design. Participants demonstrated moderate and stable levels of emotional symptoms during the baseline period. Nine of 10 participants were included in final data analysis (one showed significant baseline improvement). One participant showed significant baseline improvement on Theil-Sen slope testing and was excluded, resulting in 9 final participants whose data are reported.

In conclusion, the brief RNT-focused ACT protocol can be used to significantly reduce RNT levels in participants suffering moderate to severe emotional symptoms related to worry. The protocol showed large effect sizes in reducing emotional symptoms, warranting further research. The current study constitutes a second step in developing and analyzing the effect of brief RNT-focused ACT protocols.

Significance and contribution

This study contributes significantly to acceptance and commitment therapy research focused on transdiagnostic processes. By demonstrating that a brief ACT protocol (two 60-minute sessions) focused on repetitive negative thinking is effective in reducing emotional symptoms and psychological distress in participants with moderate emotional disorders, this work validates a more specific and efficient approach to clinical intervention. Findings suggest that complete and prolonged ACT protocols are not necessary to observe significant changes in worry and rumination-related symptoms, which has important implications for clinical implementation and accessibility of ACT-based treatments. The use of a multiple-baseline design with rigorous Bayesian analysis provides strong evidence of causality in small-sample contexts, representing a methodological advance in single-case experimental research in contextual clinical investigation.