Acceptance and commitment training focused on repetitive negative thinking for clinical psychology trainees: A randomized controlled trial

Full reference: Dereix-Calonge, I., Ruiz, F. J., Sierra, M. A., Peña-Vargas, A., & Ramírez, E. S. (2019). Acceptance and commitment training focused on repetitive negative thinking for clinical psychology trainees: A randomized controlled trial. Journal of Contextual Behavioral Science, 12, 81-88. https://doi.org/10.1016/j.jcbs.2019.02.005

Study type: Parallel randomized controlled trial (RCT) with two arms

Background and objectives

Mental health professionals typically experience high levels of emotional symptoms and burnout. Among these professionals, trainees report the highest rates of emotional symptoms, possibly due to multiple stressors characteristic of novice clinical practice: perceived incompetence, disagreements with supervisors, discrepancies between expectations and clinical reality, frequent role changes, and additional academic workload. Previous research has identified that Repetitive Negative Thinking (RNT) focused on clinical practice is a superior longitudinal predictor of increases in emotional symptoms among novice clinical psychology trainees compared to general RNT and experiential avoidance.

Theoretical research from Relational Frame Theory (RFT) suggests that RNT functions as a predominant experiential avoidance strategy that prolongs and amplifies negative affect, extends relational networks of aversive thoughts, develops inflexible behavioral patterns, and obstructs values-based behaviors. Brief RNT-focused ACT interventions have demonstrated high effectiveness for emotional disorders while requiring minimal training for implementation.

The present study evaluated the efficacy of a 6-session RNT-focused ACT training versus a waitlist control (WLC) condition on emotional symptoms and valued living in novice clinical psychology trainees. It was expected that participants in the ACT condition would show fewer emotional symptoms and greater valued living at posttreatment compared to WLC, mediated by reductions in RNT focused on clinical practice by the end of module 2 of the training.

Method

Participants

Ninety-four undergraduate clinical psychology trainees from a Colombian university were invited to participate. Eighty-five trainees (90.43% of those approached) agreed to participate and signed informed consent. Participants were in their ninth semester (of 10) during which they develop their clinical practice at the university's Center of Clinical Psychology. No exclusion criteria were adopted. Mean age was 23.38 years (SD = 2.87; range = 20-33), 79% were women, and most were single (87.4%). Participants were compensated with 20,000 Colombian pesos (approximately 6 US dollars) for completing the study. A sensitivity analysis using G*Power 3.1.9.2 indicated the study was well suited to detect medium effect sizes (d = 0.53).

Design

A parallel, two-arm RCT was conducted. Simple randomization was performed at a 1:1 ratio using the web-based tool Research Randomizer. Participants were randomly allocated to RNT-focused ACT training (N = 43) or waitlist control (N = 42). The second author generated the random allocation sequence, while the first, fourth, and fifth authors enrolled participants. An assistant uninvolved in recruitment and intervention assignment assigned participants to conditions.

Intervention

The RNT-focused ACT program (Sierra & Ruiz, submitted) consisted of 3 modules with approximately 2 hours duration each (6 hours total), originally designed for web-based application. Although designed for web delivery, in this study sessions were delivered face-to-face across 6 weekly sessions of approximately 1 hour each, to prevent access by WLC participants and enhance adherence. Sessions were conducted in groups of approximately 15 participants and led by two psychologists previously trained. Psychologists had a facilitator role as all sessions consisted of viewing videos, listening to audio files, and individually responding to exercises. Participant interaction was minimal.

Module 1, titled "Knowing the problem and finding solutions," aimed to: (a) introduce the intervention rationale, (b) identify the hierarchy of RNT triggers, (c) identify experiential avoidance strategies, (d) introduce and identify values, (e) introduce valued actions, (f) introduce the choice between RNT and valued actions, and (g) experientially differentiate RNT from cognitive defusion. Module 2, "Developing an observation point of your thoughts," included: (a) illustrate RNT processes in detail, (b) practice the difference between judging external events versus taking a nonjudgmental stance, (c) illustrate how private events are context-influenced, (d) practice noticing the flow of thoughts while focusing attention on valued behaviors, and (e) promote a transcendental self-perspective. Module 3, "Focusing on what really matters," aimed to: (a) illustrate that values can change over time, (b) explore long-term consequences of rigid RNT versus flexible value-based behavior, (c) establish goals advancing toward values, (d) explore time management and daily valued activities, (e) identify psychological barriers and establish an advancement agenda, and (f) summarize program contents.

Outcome measures

Primary outcomes (main result measures):

1. Depression Anxiety and Stress Scales-21 (DASS-21; Lovibond & Lovibond, 1995; Spanish version by Daza, Novy, Stanley, & Averill, 2002): A 21-item, 4-point Likert-type scale measuring negative emotional states during the past week. Contains three subscales (Depression, Anxiety, Stress) and permits obtaining a global emotional symptoms score. In this study, DASS-21 obtained an alpha of 0.93 for the total scale, with alphas of 0.89, 0.77, and 0.84 for Depression, Anxiety, and Stress, respectively.

2. Valuing Questionnaire (VQ; Smout, Davies, Burns, & Christie, 2014): A 10-item, 7-point Likert scale assessing general valued living during the past week. Contains two subscales: Progress (enactment of values, including clarity and perseverance) and Obstruction (disruption of valued living due to avoidance of unwanted experiences). In this study, VQ obtained alphas of 0.84 and 0.90 for Progress and Obstruction, respectively.

Process outcomes (RNT measures):

1. Perseverative Thinking Questionnaire (PTQ; Ehring et al., 2011): A 15-item, 5-point Likert scale assessing the tendency to engage in RNT when facing negative experiences or problems in general (content-independent measure). In this study, PTQ obtained an alpha of 0.96.

2. Perseverative Thinking Questionnaire for Clinical Psychology Trainees (PTQ-CPT; Dereix-Calonge, Ruiz, Suárez-Falcón, & Flórez, in press): A 9-item, 5-point Likert scale designed to assess the tendency of clinical psychology trainees to engage in RNT concerning clinical practice issues. Based on the PTQ and also content-independent. In this study, PTQ-CPT obtained an alpha of 0.91.

Data analysis

All variables were explored for data entry accuracy and missing values. Missing item data were estimated using the participant's mean score for the specific scale. Outliers were identified through graphical inspection of raw data and replaced using Winsor's method. Less than 4% of data were modified during outlier correction.

Independent sample t-tests explored condition equivalence at pretreatment. Repeated measures ANOVA (RM ANOVA) analyzed the effects of Time (Pre and Post) and Condition (RNT-focused ACT vs. WLC) factors on all dependent variables. Effect sizes were computed using Morris (2008) formula, interpreted as small (d = 0.20-0.49), medium (d = 0.50-0.79), and large (d > 0.80). Analyses were rerun with only participants with high emotional symptoms (65th percentile on DASS-Total, score > 20). Reliable change index (RCI) and clinically significant change (CSC) were computed per Jacobson and Truax (1991). Simple mediation analyses used nonparametric bootstrapping with PROCESS 3.1 (Hayes, 2018) and 20,000 bootstrapped samples. Condition served as the predictor, PTQ-CPT at module 2 end as the putative mediator, and posttreatment outcome measures as criteria. Indirect effects were considered significant if 95% bias-corrected bootstrap confidence intervals excluded zero.

Results

All participants in the ACT condition attended at least 4 sessions and were considered completers. No data were lost at posttreatment.

Pretreatment equivalence:

ACT condition scores were slightly higher on emotional symptoms and RNT measures, but independent sample t-tests revealed no statistically significant differences (DASS-Total: t(83) = 1.258, p = 0.212; Depression: t(83) = 0.238, p = 0.813; Anxiety: t(83) = 1.360, p = 0.178; Stress: t(83) = 0.155, p = 0.877; VQ-Progress: t(83) = -0.086, p = 0.932; VQ-Obstruction: t(83) = 0.849, p = 0.398; PTQ-CPT: t(83) = 1.288, p = 0.201; PTQ: t(83) = 1.172, p = 0.245).

Pre-post differences in overall sample:

The ACT condition showed slight decreases in DASS-Total, Depression, and Anxiety, while WLC showed increases in all these variables. Regarding Stress, ACT showed less increase than WLC. RM ANOVA revealed statistically significant Time × Condition interactions for DASS-Total (F(1) = 10.35, p = 0.002, d = 0.75) and DASS-Depression (F(1) = 20.77, p < 0.001, d = 0.79), but not for Anxiety (F(1) = 2.76, p = 0.10, d = 0.42) or Stress (F(1) = 1.79, p = 0.19, d = 0.36). Regarding valued living, ACT showed increases in VQ-Progress and decreases in VQ-Obstruction, while WLC showed the opposite pattern. RM ANOVA revealed a statistically significant Time × Condition interaction for VQ-Obstruction (F(1) = 9.99, p = 0.002, d = 0.51), but not for VQ-Progress (F(1) = 1.64, p = 0.231, d = 0.34).

ACT participants showed decreases in PTQ-CPT and PTQ, while WLC showed increases. Paired t-tests indicated changes in ACT were statistically significant (PTQ-CPT: t(42) = 3.76, p < 0.001; PTQ: t(42) = 2.93, p = 0.005). RM ANOVA showed statistically significant Time × Condition interactions for PTQ-CPT (F(1) = 11.60, p = 0.001, d = 0.89) and PTQ (F(1) = 7.01, p = 0.01, d = 0.62).

Pre-post differences in participants with high emotional symptoms:

Data from participants with DASS-Total scores > 20 at pretreatment were analyzed (N = 15 ACT; N = 14 WLC). ACT showed decreases in all emotional symptom indicators while WLC showed increases. RM ANOVA revealed statistically significant Time × Condition interactions for DASS-Total (F(1) = 14.99, p < 0.001, d = 2.52), DASS-Depression (F(1) = 11.91, p = 0.002, d = 1.21), DASS-Anxiety (F(1) = 5.60, p = 0.025, d = 1.30), VQ-Progress (F(1) = 5.06, p = 0.033, d = 1.10), and VQ-Obstruction (F(1) = 6.12, p = 0.020, d = 0.88). The PTQ interaction reached marginal significance (F(1) = 4.14, p = 0.052, d = 1.00).

Regarding reliable change (RCI) and clinically significant change (CSC), in the ACT condition, 11 of 15 participants (73.33%) showed RCI and 10 (66.67%) showed CSC. In contrast, only 1 of 14 participants (7.14%) in WLC showed both indicators. Chi-squared tests showed statistically significant differences (RCI: χ²(1) = 13.08, p < 0.001, d = 1.81; CSC: χ²(1) = 10.90, p < 0.001, d = 1.55). Only 6.67% of ACT participants showed deterioration in emotional symptoms compared to 28.57% in WLC.

Mediation analyses:

Mediation analyses with emotional symptoms as dependent variables showed that RNT focused on clinical practice (PTQ-CPT) at module 2 end acted as a statistically significant mediator of the condition effect in all cases. Indirect effects were: DASS-Total (3.787, SE = 1.410, 95% BC CI [1.342, 6.806]); DASS-Depression (0.882, SE = 0.436, 95% BC CI [0.181, 1.884]); DASS-Anxiety (0.909, SE = 0.462, 95% BC CI [0.174, 1.960]); DASS-Stress (1.424, SE = 0.601, 95% BC CI [0.329, 2.678]). Partially standardized mediation effect sizes (abps) were 0.289, 0.187, 0.909, and 1.424, respectively.

For valued living, PTQ-CPT mediated the condition effect only for VQ-Obstruction. For VQ-Progress, the indirect effect was not statistically significant (estimate = -0.522, SE = 0.421, 95% BC CI [-1.523, 0.112], abps = -0.089). For VQ-Obstruction, the indirect effect was statistically significant (1.399, SE = 0.630, 95% BC CI [0.372, 2.835], abps = 0.188).

Discussion and conclusions

This RCT analyzed the effect of a 6-session RNT-focused ACT training versus waitlist control on emotional symptoms and valued living in novice clinical psychology trainees (N = 85). The training was an adaptation of a web-based intervention for emotional disorders and required minimal training for implementation. The training produced statistically significant effects on overall emotional symptoms, depression, and obstruction in valued living, with medium to large effect sizes. Changes in process measures were also statistically significant, particularly for RNT focused on clinical practice, which showed a large effect size. Overall, WLC participants showed emotional symptom increases consistent with previous studies with trainees of similar characteristics.

The effect was substantially greater for participants with high emotional symptom scores, showing large effect sizes for overall emotional symptoms (d = 2.52), depression (d = 1.21), anxiety (d = 1.30), and values (d = 1.10 and 0.88). High-symptom participants in WLC showed deterioration, with 28.57% experiencing emotional symptom deterioration, suggesting that emotional symptoms in trainees are not transitory and that RNT-focused ACT training could be especially beneficial for this population.

This study constitutes one of the few RCTs conducted with clinical psychology trainees. The systematic review by Rudaz et al. (2017) identified 9 studies of mindfulness and acceptance-based interventions with trainees; only one was an RCT, two were nonrandomized trials, and the remainder were open trials. This RCT has the greatest statistical power (largest sample size) of studies with trainees. An additional strength is the longitudinal mediation analysis, which is the first analyzing longitudinal mediators of intervention effects with clinical psychology trainees. Intervention implementation required minimal training, facilitating dissemination. Limitations include the sample consisting of undergraduate trainees (in many countries training is only at postgraduate level), recruitment from a single university, absence of long-term follow-up data, and comparison with WLC rather than an active intervention. Future studies should replicate results with postgraduate trainees, multiple universities, collect longitudinal follow-up, analyze web-based application (original format), compare against nondirective active interventions, and explore other potential mediators such as experiential avoidance and cognitive fusion.

Significance and contribution

This study provides rigorous evidence that brief training based on Relational Frame Theory can significantly reduce emotional symptoms in clinical psychology students during professional formation. Longitudinal mediation analysis demonstrates that reduction of repetitive negative thinking focused on clinical practice is a key change mechanism, suggesting that structured interventions targeting specific psychological processes can be particularly effective during critical periods of professional development. The findings contribute to knowledge about interventions for preventing burnout and emotional symptoms in mental health professionals in training, an important yet relatively unexplored area of research.