A randomized dismantling trial of the open and engaged components of acceptance and commitment therapy in an online intervention for distressed college students

Highlights

• Compared Full ACT website to Open components-only, Engaged-only, or waitlist.

• All 3 ACT conditions improved mental health symptoms relative to waitlist.

• Open condition equivalent to waitlist on reliable change and positive mental health.

• Full ACT improved more than Engaged or Open on cognitive fusion.

Abstract

This dismantling trial compared the effects of a full online Acceptance and Commitment Therapy (ACT) intervention to the isolated effects of the Open (i.e., acceptance, cognitive defusion) and Engaged (i.e., values, committed action) components of ACT. A sample of 181 distressed college students were randomized to one of four conditions: a 12-session full ACT website (Full), a version targeting the Open components (Open), a version targeting the Engaged components (Engaged), or waitlist. Participants in active conditions were also randomized to receive phone coaching or just email prompts to increase program adherence. All three ACT conditions significantly improved over time relative to the waitlist condition on the primary outcome of mental health symptoms. The Engaged and Full conditions had greater rates of reliable change on the primary outcome relative to waitlist, but not the Open condition. Similarly, only Engaged and Full conditions improved on positive mental health relative to waitlist. The Full condition had greater improvements on a few ACT process measures relative to Engaged and Open conditions, particularly cognitive fusion. Overall, results indicate targeting only the Open components of ACT was somewhat less effective, and that including both the Open and Engaged components led to greater decreases in cognitive fusion.

Introduction

Evaluating the treatment components underlying cognitive behavioral therapies (CBTs) is critical for a progressive science of behavior change. Component research can lead to improved efficiency by identifying inert or unnecessary components (possibly making treatment simpler or shorter), and improved efficacy by identifying components that may be emphasized or added to protocols. Beyond these more proximal treatment development goals, component research can serve longer term knowledge development by challenging and refining the underlying theoretical models for how to bring about meaningful clinical outcomes. Such an approach makes it possible to develop a more refined understanding of how a treatment works (including the specific components and processes of change that impact outcomes) rather than simply knowing whether a multicomponent treatment protocol is efficacious. Over time, component research could support a more process-based approach to therapy in which components matched to therapeutic processes can be ideographically applied to given presenting problems (Hofmann & Hayes, 2019).

One CBT that has received considerable attention at the level of treatment components and processes of change is Acceptance and Commitment Therapy (ACT; Hayes, Strosahl, & Wilson, 2012). ACT is a modern CBT, which based on the psychological flexibility model (Hayes, Luoma, Bond, Masuda, & Lillis, 2006), includes a combination of treatment components seeking to alter the impact of maladaptive, internal sources of behavioral regulation such as experiential avoidance and cognitive fusion (through acceptance and cognitive defusion, also called the “Open” components; Hayes, Villatte, Levin, & Hildebrandt, 2011) and to increase more adaptive sources of behavioral regulation (through values and committed action, also called the “Engaged” components; Villatte et al., 2016). ACT also includes components targeting flexible attention to the present moment and a flexible sense of self, called self-as-context (the “Aware” components; Hayes et al., 2011). These Aware components are highly overlapping with other ACT components and are part of the broader foundation and context in which therapy occurs (e.g., attending to internal experiences is necessary to develop openness, attending to the present supports identifying and engaging in valued activities; Villatte et al., 2016).

The components of ACT have been extensively evaluated in laboratory-based studies, with results indicating their isolated efficacy on proximal laboratory outcomes (Levin, Hildebrandt, Lillis, & Hayes, 2012). Similarly, the combination of these components within ACT protocols has been found effective for a wide range of clinical presentations in over 300 randomized controlled trials (RCTs; ACBS, 2019; A-Tjak et al., 2015). Mediational analyses further indicate that clinical improvements through ACT are accounted for by processes of change closely connected to the underlying treatment components (Hayes et al., 2006; Twohig & Levin, 2017) including overall psychological inflexibility (e.g., Twohig, Plumb Vilardaga, Levin, & Hayes, 2015), and more specific processes including acceptance (e.g., Levin, Haeger, Pierce, & Twohig, 2017), cognitive defusion (e.g., Zettle, Rains, & Hayes, 2011), and values (e.g., Gloster et al., 2017).

However, neither research on the isolated components of ACT or on full ACT protocols addresses the critical question of whether all of the components of ACT are necessary to produce meaningful clinical outcomes. It may be that some of the components of ACT are redundant and not needed to produce meaningful outcomes, or even to engage targeted processes of change. There is a long list of surprising findings from dismantling and mantling studies of other CBTs indicating that adding or removing components theorized to be critical does not change outcomes (e.g., Dimidjian et al., 2006; Newman et al., 2011; Schmidt et al., 2000). Furthermore, ACT is differentiated in part from some other modern CBTs through its focus on both acceptance/mindfulness and behavior change methods, which can be contrasted with other evidence-based therapies that only emphasize acceptance/mindfulness (e.g., Mindfulness-Based Therapies; Hofmann, Sawyer, Witt, Ashley, & Oh, 2010) or behavior change (e.g., Behavioral Activation; Ekers et al., 2014). Direct research is needed evaluating the efficacy of the acceptance and behavior change components of ACT in isolation versus their combination.

Despite the plethora of component and process research on ACT, there have been very few studies to-date directly comparing ACT components within a treatment context (e.g., Petersen, Krafft, Twohig, & Levin, In Press; Villatte et al., 2016). In a multiple-baseline study, a sample of 15 adults seeking treatment for a mental health concern were randomly assigned to either receive in-person therapy focusing on the Open components of ACT (cognitive defusion, acceptance) or the Engaged components of ACT (values, committed action; Villatte et al., 2016). Of note, Aware components were integrated within both the Open and Engaged conditions, but in the context of targeting distinct ACT processes (e.g., present moment awareness of difficult thoughts and feelings versus valued action). Results indicated that the ACT components functioned differently, with participants receiving the Open components improving on mental health symptoms, acceptance, and defusion more than Engaged, while participants in the Engaged condition improved more on quality of life and valued action. Although results suggested that these ACT components function differently in ways consistent with the underlying theoretical model, they do not yet determine whether treatment is more effective when including the combination of these components in a full ACT protocol.

Based in part on the Villatte et al. (2016) study, we conducted a pilot evaluation comparing web-based versions of ACT that include Full ACT, only the Open, or only the Engaged components (Petersen et al., In Press). As in prior research (Villatte et al., 2016), the Aware components naturally overlapped across conditions. Such online research, in which the intervention is delivered in an automated format, allows for a highly replicable, controlled evaluation of specific treatment procedures without the added variability introduced by in-person therapy. Furthermore, once a website is created, there are minimal costs per end user, thus offering a more cost effective, feasible pathway for conducting dismantling research, which typically requires substantial resources.

The initial pilot trial included a small sample of 55 adults seeking help for a mental health concern in order to test feasibility of the three ACT websites (Petersen et al., In Press). The majority of participants actively engaged in each website condition, although only about a third completed the full program. All conditions showed medium to large improvements on psychological distress and ACT process measures. Although the study was underpowered to detect differential effects, a significant time by condition effect was found indicating that the Full ACT website led to greater improvements than Open or Engaged versions on psychosocial functioning. Overall, this pilot suggests the promise of evaluating ACT components in an online format, and for the specific websites tested in the current study. However, findings were limited due to the lack of a waitlist comparison, low program engagement, and a small, heterogeneous sample of adults interested in mental health resources.

The current dismantling trial sought to compare web-based versions of ACT targeting the Open components, the Engaged components, or Full ACT, relative to a waitlist condition, with a sample of 181 university students reporting elevated distress. To increase program engagement, participants assigned to an ACT website were also randomized to receive phone coaching or just email prompts. We predicted that all three ACT conditions would produce greater effects on mental health relative to the waitlist, but that the Full condition would have greater improvements than the Open or Engaged conditions. We also predicted that the ACT conditions would produce differential effects based on their relevant processes of change: Open and Full would produce greater effects on acceptance and cognitive fusion relative to Engaged and waitlist, while Engaged and Full would produce greater effects on values and committed action relative to Open and waitlist.