Behaviour Research and Therapy 49 (2011) 267e274

Contents lists available at ScienceDirect

Behaviour Research and Therapy

Processes of change in psychological exibility in an interdisciplinary group-based treatment for chronic pain based on Acceptance and Commitment Therapy
Lance M. McCracken a, b, *, Olga Gutirrez-Martnez c
a

Centre for Pain Services, Royal National Hospital for Rheumatic Diseases, Upper Borough Walls, Bath BA1 1RL, UK Centre for Pain Research, University of Bath, Bath, UK c Department of Personality, Assessment and Psychological Treatments, University of Barcelona, Spain
b

a r t i c l e i n f o
Article history: Received 10 November 2010 Received in revised form 13 January 2011 Accepted 8 February 2011 Keywords: Chronic pain Acceptance Values Mindfulness Processes of change Cognitive behavioral therapy Acceptance and Commitment Therapy

a b s t r a c t
There are now numerous studies of Acceptance and Commitment Therapy (ACT) for chronic pain. These studies provide growing support for the efcacy and effectiveness of ACT in this context as well as for the role of ACT-specic therapeutic processes, particularly those underlying psychological exibility. The purpose of the present study was to continue to build on this work with a broader focus on these processes, including acceptance of pain, general psychological acceptance, mindfulness, and valuesbased action. Participants included 168 patients who completed an ACT-based treatment for chronic pain and a three-month follow-up. Following treatment and at follow-up, participants reported signicantly reduced levels of depression, pain-related anxiety, physical and psychosocial disability, medical visits, and pain intensity in comparison to the start of treatment. They also showed signicant increases in each of the processes of psychological exibility. Most uncontrolled effect sizes were medium or large at the follow-up. In correlation analyses changes in the four processes measures generally were signicantly related to changes in the measures of depression, anxiety, and disability. In regression analyses the combined processes were related to changes in outcomes above and beyond change in pain intensity. Although in some ways preliminary, these results specically support the unique role of general psychological acceptance in relation to improvements achieved by treatment participants. The current study claries potential processes of change in treatment for chronic pain, particularly those aiming to enhance psychological exibility. 2011 Elsevier Ltd. All rights reserved.

There is a growing body of literature to support Acceptance and Commitment Therapy (ACT) (Hayes, Strosahl, & Wilson, 1999) in the treatment of chronic pain, including treatment outcome studies in both adult (McCracken, Vowles, & Eccleston, 2005; Vowles & McCracken, 2008; Wicksell, Ahlqvist, Bring, Melin, & Olsson, 2008) and pediatric samples (Wicksell, Melin, Lekander, & Olsson, 2009). Treatment outcome studies conducted so far suggest that relatively brief ACT interventions of between three to eight weeks can produce signicant benets in the emotional, physical, and social functioning of people with chronic pain. As further support for applications of ACT to chronic pain, secondary analyses within these studies show that the processes of psychological exibility targeted in ACT appear to account for an appreciable proportion of the benets

* Correspondence to: Lance M. McCracken, PhD, Centre for Pain Services, Royal National Hospital for Rheumatic Diseases, Upper Borough Walls, Bath BA1 1RL, UK. Fax: 44 1225473461. E-mail address: l.mccracken@bath.ac.uk (L.M. McCracken). 0005-7967/$ e see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.brat.2011.02.004

observed. However, a broader focus on processes of change is important to continue to clarify the roles of the separate processes dened within this treatment approach. Simply stated, the assumption behind the application of ACT to chronic pain is that it is not merely the severity of pain or other symptoms in isolation that inuences patient functioning, but also psychological relationships between these symptoms and behavior. Accordingly, ACT is explicitly not aimed at reducing pain or distress, or at changing the frequency or content of thoughts. Instead, ACT seeks to improve functioning for people with chronic pain by modifying the impacts of pain and other symptoms through acceptance and mindfulness methods. It does this by increasing psychological exibility, dened in part as the ability to act effectively in accordance with personal values and goals in the presence of potentially interfering thoughts and feelings (Hayes, Luoma, Bond, Masuda, & Lillis, 2006). In the model underlying ACT psychological exibility entails six interrelated therapeutic processes: acceptance, cognitive defusion, contact with the present moment, self-as-context, values, and committed action.

268

L.M. McCracken, O. Gutirrez-Martnez / Behaviour Research and Therapy 49 (2011) 267e274

Vowles and McCracken (2008) reported the effects of a three to four week intensive treatment for chronic pain based on ACT. Signicant improvements in pain, depression, pain-related anxiety, disability, medical visits, work status and physical performance were found following treatment and at a 3-month follow-up. Two ACT processes were examined in this study: acceptance of pain as measured by the Chronic Pain Acceptance Questionnaire (CPAQ; McCracken, Vowles, & Eccleston, 2004) and values-based action as measured by the Chronic Pain Values Inventory (CPVI; McCracken & Yang, 2006). Changes in acceptance of pain were related to changes in pain, depression, pain-related anxiety, physical and psychosocial disability and physical performance in the pre- to post-treatment interval and changes in values-based action were signicantly associated with change in pain, depression and physical and psychosocial disability in the pre-treatment to followup interval. In a randomized controlled trial Wicksell and colleagues (Wicksell, Ahlqvist, et al., 2008) compared treatment as usual (TAU) to a 10-session ACT-based protocol for patients suffering from whiplash-associated disorder. At a seven-month follow-up, ACT demonstrated better results than TAU in terms of disability, life satisfaction, fear of movement, and depression. Mediation analysis showed signicant indirect effects for psychological inexibility, as measured by the Psychological Inexibility in Pain Scale (PIPS; Wicksell, Renflt, Olsson, Bond, & Melin, 2008), in relation to changes in disability and life satisfaction. The PIPS is regarded as a measure of avoidance and cognitive fusion. The role of a wider range of the components of psychological exibility in the well-being and daily functioning of people with chronic pain has been examined in studies using correlational methods, mostly done at one point in time, without experimental manipulation or application of a treatment. These studies have illustrated the signicant role of processes of general psychological acceptance (McCracken & Velleman, 2010; McCracken & Zhao-OBrien, 2010), acceptance of pain (e.g., Mason, Mathias, & Skevington, 2008; McCracken et al., 2004), mindfulness (McCracken & Velleman, 2010; McCracken, Gauntlett-Gilbert, & Vowles, 2007), value-based action (McCracken & Yang, 2006) and general exibility itself (McCracken, Vowles, & Zhao-OBrien, 2010). The study of these varied processes has not yet been done as comprehensively during the course of treatment. In order to demonstrate that the wider process of psychological exibility as currently conceptualized is useful in the treatment of chronic pain it is necessary eventually to demonstrate that each of its component processes plays a signicant role in treatment outcome. So far process studies of ACT in chronic pain have been limited, mostly constrained by the availability of appropriate validated measures. One way to expand this work is to expand within the process of acceptance. Thus far only specic pain-related acceptance has been studied in this context. An existing measure of general psychological acceptance, the Acceptance and Action Questionnaire (AAQ) (Hayes et al., 2004) could be used to assess acceptance conceived more broadly. It measures acceptance of unwanted thoughts and feelings without a specic focus on pain. Another way to expand this work is to select additional processes not yet examined. Acceptance, cognitive defusion, contact with the present, and self-as-observer are also regarded as the mindfulness processes within ACT (McCracken & Thompson, 2009). Hence, in a situation where there are few specic measures for most of the ACT processes, a measure of mindfulness could be used to reect these. The role of mindfulness in ACT-based treatments has been subject to less study, although recent studies show that increases in mindfulness correlate with treatment effects with ACT (Forman, Butryn, Hoffman, & Herbert, 2009; Kocovski, Fleming, & Rector, 2009).

The purpose of the present study was to investigate a range of treatment processes in ACT for chronic pain that is more comprehensive in comparison to those investigated in previous studies, a range that includes for the rst time general psychological acceptance and mindfulness. The simultaneous examination of multiple specic processes is expected to improve targeting of methods to optimize outcomes and to generally aid in treatment development (Kazdin, 2007; Preacher & Hayes, 2008). The specic aims of the present study were twofold. First, we sought to perform a detailed examination of treatment outcomes following an ACT-based treatment in a sample of chronic pain patients not previously studied for this purpose. Second, we sought to conduct treatment process analysis including four treatment processes: acceptance of pain, general psychological acceptance, mindfulness, and values-based action. Consistent with the ACT model, it was expected that patients would report an increase in the four process variables over the course of therapy. It was also expected that these changes would predict outcome such that patients who improved more on these processes would experience larger improvements in emotional, social, and physical functioning. Finally, we predicted that outcome would be more closely related to changes in the components of psychological exibility than to changes in pain intensity since the focus of treatment was not specically on reducing pain, but rather on changing how one responds to pain. Method Participants Participants were patients who attended treatment at a tertiary care pain rehabilitation unit in southwest England between September 2006 and June 2009. All participants reported persistent pain of 3 months duration or longer and signicant levels of painrelated distress and disability, and agreed with the rehabilitative focus nature of treatment. Participants were excluded from treatment if they required further medical tests or procedures or had conditions sufcient to interfere with participation in a groupbased treatment, such as signicant cognitive impairment or overwhelming psychiatric conditions. Inclusion and exclusion were determined by assessments from a specialist physician and clinical psychologist prior to being offered treatment. These assessments are primarily pragmatic in nature, for purposes of determining the appropriateness of treatment, and not focused on deriving formal medical or psychiatric diagnoses. This study included 168 individuals (112 women, 56 men) between the ages of 18 and 77 years (M 43.5, SD 13.0) who completed a three-or-four-week course of interdisciplinary treatment for chronic pain, as well as the three-month follow-up session. The sample of 168 excluded 57 individuals who completed treatment, but did not attend the follow-up. These 57 were excluded because all of the primary analyses involved an examination of outcomes at the follow-up assessment. The larger proportion of participants was women, 66.7%, as is typically the case in specialty services for chronic pain. Mean age was 46.2 years, SD 10.1. They completed a mean of 13.6 years of education, SD 3.6. They were almost exclusively White European in background, 98.2%. They were mostly married or cohabitating, 58.5%, and the remainder were single, 28.3%, divorced or separated, 11.4%, or widowed 1.8%. They were mostly out of work, 72.3%, and the average time period out of work was 76.5 months, SD 69.3. The largest single group of patients was not working due to pain, 47.6%, followed by retired early specically due to pain, 21.1%, working part time because of pain, 6.6%, working full time, 6.0%, working part time, 4.2% or other, 14.5%. The median

L.M. McCracken, O. Gutirrez-Martnez / Behaviour Research and Therapy 49 (2011) 267e274

269

chronicity of pain was 97.5 months and the usual pain intensity on a scale from zero to ten was 7.0, SD 1.7. The most frequently identied primary pain site was low back, 56.6%, followed by lower extremity, 14.5%, full body, 9.4%, upper extremity, 8.8%, neck, 3.8%, or other, 7.0%. Most participants, 66.4%, reported a diagnosis of chronic unspecic pain such as chronic pain syndrome or non-specic musculoskeletal pain, followed by bromyalgia, 18.6%, post back surgery pain, 7.2%, complex regional pain syndromes, 6.0%, or other. Ethical approval for the study was received from the relevant institutional ethics committee and all participants provided written informed consent prior to their data being used in the study. Measures Participants completed a series of assessment instruments before and after treatment and at the three-month follow-up. Background characteristics were assessed with an ad hoc questionnaire that also included items about pain onset, duration, location, and so forth. Research assistants supervised the assessments to aid in completion. Missing data occurred in fewer than 5.4% of cases on any single measure. The measures administered included measures of the primary process variables of interest in this study: acceptance of chronic pain, general psychological acceptance, mindfulness, and valuesbased action; and outcome measures related to physical, emotional and social functioning and healthcare use. Chronic Pain Acceptance Questionnaire (CPAQ) The CPAQ (McCracken et al., 2004) is a measure of pain-related acceptance widely used in people with chronic pain. The 20 items are rated on a 7-point scale and form two subscales: activity engagement, reecting the pursuit of life activities with pain present, and pain willingness, reecting a relative absence of attempts to avoid or control pain. The total score was used in the present study to enable analyses of acceptance of pain as a single construct. The CPAQ has repeatedly shown to have good psychometric properties (Vowles, McCracken, McLeod, & Eccleston, 2008). The Cronbachs alpha for the total scale based on the current sample was .85. Acceptance and Action Questionnaire-II (AAQ-II) The AAQ-II (Bond et al., submitted for publication) is a 10-item scale developed to assess the same construct as the original AAQ (Hayes et al., 2004). It is a short general measure of psychological acceptance or the willingness to experience unwanted private experiences, such as bodily sensations, emotions, thoughts, memories, in the pursuit of ones values and goals. It is sometimes referred to as a measure of psychological exibility. Patients are asked to rate each statement on a scale from 1 (never true) to 7 (always true). Higher scores represent higher levels of general acceptance. The AAQ has been showed to have good validity and adequate internal consistency (Hayes et al., 2004). The Cronbachs alpha for the AAQ-II based on the current sample was .88. Mindful Attention Awareness Scale (MAAS) The MAAS (Brown & Ryan, 2003) is a 15-item measure of trait mindfulness. Each of the items reects a relative absence of mindfulness and are rated such that higher ratings indicate less frequency of the response indicated. Hence, higher scores on the 0e6 scale (almost always to almost never) indicate higher mindfulness. Items include, I nd myself preoccupied with the future or the past or It seems I am running on automatic without much awareness of what Im doing. The MAAS was chosen as a measure of mindfulness because it is brief and well-validated. The MAAS has

been extensively validated and used in a number of previous studies (Brown & Ryan, 2003; Brown, Ryan, & Creswell, 2007). The Cronbachs alpha for the MAAS based on the current sample was .86. Chronic Pain Values Inventory (CPVI) The CPVI (McCracken & Yang, 2006) is a 12-item measure of values-based action for use with people with chronic pain. It asks respondents to rate the importance of the values they hold in six domains of living: family, intimate relations, friends, work, health, and growth or learning; and their success at living according to them on a scale from 0 (not at all important/successful) to 5 (extremely important/successful), respectively. Previous studies using the CPVI have focused on the mean success rating only. The success items have demonstrated adequate internal consistency and construct validity (McCracken & Yang, 2006; Vowles & McCracken, 2008) and sensitivity to change in ACT-based treatment for chronic pain (Vowles & McCracken, 2008). The Cronbachs alpha for the success scale based on the current sample was .86. British Columbia Major Depression Inventory (BCMDI) The BCMDI (Iverson & Remick, 2004) is a 20-item self-report measure of depression based on the Diagnostic and Statistical Manual of Mental Disorders (4th edition; DSMIV; American Psychiatric Association, 1994) criteria for major depression. The rst 16 items are symptoms that are endorsed if present over the past 2 weeks and then, if present, rated on a scale from 1 (very mild problem) to 5 (very severe problem). The last 4 items measure the impact of these symptoms and problems on day-to-day life with regard to work or school, family and social life activities. These impact scores were not used in the present study. Previous studies support the reliability, validity and clinical usefulness of the test (Iverson & Remick, 2004). The Cronbachs alpha for the BCMDI based on the current sample was .81. Pain Anxiety Symptoms Scale-20 (PASS-20) The PASS-20 (McCracken & Dhingra, 2002) is a 20-item measure of pain-related fear and avoidance. Each item is rated on a frequency scale from 0 (never) to 5 (always). Examples of items include, I think that if my pain gets too severe, it will never decrease, I avoid important activities when I hurt, and I worry when I am in pain. The reliability and validity of the PASS-20 are well established (Carleton, Abrams, Asmundson, Antony, & McCabe, 2009; McCracken & Dhingra, 2002). The Cronbachs alpha for the total scale based on the current sample was .92. Sickness impact prole (SIP) The SIP (Bergner & Bobbitt, 1981) is a behaviorally based measure of health status. It reects perceived health-related limitations in 12 categories of activity, such as sleep and rest, home management, social interaction, and so forth, comprising in total 136 statements. All items in the instrument are weighted and an overall score can be calculated as well as scores on the physical and psychosocial dimensions. Higher scores mean more functional disability. The SIP is widely used in healthcare settings and has repeatedly shown to have good psychometric properties (De Haan, Aaronson, Limburg, Hewer, & Van Crevel, 1993; Kalkman, Schillings, Zwarts, van Engelen, & Bleijenberg, 2007). In the present study we used the physical and psychosocial disability component scores. Medical visits Pain-related medical visits over the past 6 months, including GP, specialist visits, and emergency care, were summed based on patient estimates.

270

L.M. McCracken, O. Gutirrez-Martnez / Behaviour Research and Therapy 49 (2011) 267e274

Pain intensity Average pain intensity over the past week was assessed using a 0 (no pain) to 10 (worst possible pain) numerical rating scale. Treatment All participants in this study received a treatment program that was a form of Acceptance and Commitment Therapy (Hayes et al., 1999) specically designed for delivery to groups of patients, in a specialty care setting, and within a coordinated interdisciplinary team consisting of clinical psychology, physical therapy, occupational therapy, nursing, and medicine (McCracken, 2005). Treatment methods explicitly targeted the key processes of the ACT. The primary process targeted is psychological exibility and the primary goal is improved daily functioning. Particular methods focused on enhancing acceptance of pain and other psychological experiences, contact with the present moment, self-as-observer, cognitive defusion, values, and committed action. Detailed information on treatment philosophy and content can be found in McCracken (2005) and Hayes et al. (1999) and similar methods can be found in Dahl, Wilson, Luciano, and Hayes (2005). As noted, the duration of the active treatment phase was three to four weeks, depending on a psychological assessment of case severity and complexity. Treatment was delivered primarily in a group format during ve days per week for six and one half hours each day. Each treatment day included approximately two and one quarter hours of physical conditioning, one hour of psychological methods, 30 min of mindfulness training, and one hour of activity management, with the remainder of the time devoted to other aspects of skills training and health education. All the methods used by the team of psychologists, physical therapists, occupational therapists, nurses, and physicians were designed not to target pain or other symptoms for removal, but instead to alter the patients relationship to these experiences so that they could reduce their impact and improve functioning. The psychological sessions were designed specically to emphasize experiential methods rather than didactic ones. This treatment generally does not include explicit cognitive restructuring or self-statement analyses exercises, strategies to increase self-efcacy, or training in relaxation or other methods aimed at controlling feelings or thoughts. While in treatment patients lived independently in apartments adjacent to the hospital. To ensure treatment integrity, treatment content and patient progress were discussed in supervision sessions, clinical teams meetings (three times per week) and once-weekly clinical seminar meetings. Statistical analyses The main goal of the present study was to explore the relationship between ACT processes and outcomes beyond whether the treatment produces a positive impact per se. Hence, the primary analyses focused on relations between process change during the treatment phase in relation to outcome at follow-up. Initially, t-tests were used to asses for potential differences among those who attended and did not attend the follow-up appointment. Next, we evaluated treatment outcomes immediately following treatment and at the follow-up. Paired-samples t-tests were used to examine improvements over time for all variables and within-subjects effect sizes were calculated according to Cohen (1988). Finally, we examined if the changes in outcomes were signicantly related to changes in the four ACT process measures with two sets of analyses. In both analyses, to partly address the temporal order of process and outcome, we

computed pre-to-post residualized changes for the hypothesized processes of change and pre-to-follow-up residualized changes for outcome measures. First, we calculated an overall correlation matrix of these change scores. Next, hierarchical multiple regression was used to asses the ability of change scores in the four measures representing psychological exibility to account for variance in the change scores from outcome measures. These analyses also were designed to consider and statistically control the role of relevant patient background variables as well as changes in pain intensity. Results Preliminary analysis From the larger database including the current sample there were no demographic differences between those who attended the followup and those who did not attend, except that follow-up completers had slightly more years of education, t (280) 1.78, p .08. With regard to the primary process and outcome measures, follow-up attenders and non-attenders were not different on any of the process measure or on pain-related anxiety, medical visits, or pain, as measured at pre-treatment, all t < 1.4, all p > .17. However, those who attended follow-up presented with somewhat lower scores on depression, t (286) 2.61, p < .05, physical disability, t (303) 2.48, p < .05, and psychosocial disability, t (302) 2.80, p < .01. Impact of treatment Table 1 summarizes descriptive statistics for all the primary process and outcome measures at pre-, post, and 3-month followup. Based on t-tests, the patients showed signicant improvements in all scores for both time periods, all t (167)  5.17, all p < .001, for pre- to post assessment scores, and all t (167)  3.00, all p < .005, for pre- to 3-month follow-up assessment scores. Fig. 1 shows specic effect size magnitudes for all measures. From pre- to post assessment scores, the overall average effect across all measures was d 0.85 (range: 0.47e1.61). Changes for acceptance of pain, values-based action, depression, pain-related anxiety and psychosocial disability were of a large size. The changes for the remaining variables were medium sized. From pre- to follow-up assessment scores, the Cohens d values were slightly smaller, averaging d 0.68 (range: 0.29e1.42). Nevertheless, improvements over this longer time frame in acceptance of pain, values-based action and pain-related anxiety were of a large size. A medium effect size was seen for psychological acceptance, mindfulness, depression, and physical and psychosocial disability. Changes across time for medical visits and pain intensity were smaller sized.
Table 1 Mean values and standard deviations for all measures over time. Pretreatment M Acceptance of pain Values-based action Psychological acceptance Mindfulness Depression Pain-related anxiety Physical disability Psychosocial disability Medical visits (past 6 months) Pain intensity 48.32 1.64 38.01 3.66 28.07 48.24 0.22 0.31 8.88 7.02 SD 15.68 1.04 12.53 0.86 12.74 18.27 0.12 0.16 12.48 1.54 Posttreatment M 73.53 2.68 44.86 4.08 14.93 32.64 0.14 0.16 e 6.29 SD 19.24 1.13 13.30 0.86 11.40 19.38 0.11 0.14 e 1.77 3-Month follow-up M 70.57 2.42 46.09 4.15 18.96 32.26 0.16 0.20 4.89 6.58 SD 21.39 1.16 12.89 0.93 13.51 18.29 0.11 0.16 5.26 1.94

L.M. McCracken, O. Gutirrez-Martnez / Behaviour Research and Therapy 49 (2011) 267e274

271

1.8 1.6 1.4 1.2 Effect Size 1 0.8 0.6 0.4 0.2 0
ce an -b pt e es c lu Ac Va of in pa ed as ty ty ty y ss on ce its ili ili si et si is ne an xi ab ab en ul es lv pt t s n f r s i a e i n a d p i d c ic in ld in De ed ac oc ed M ca at M Pa si os ch el y r y h yc in Ps Ph Ps Pa n

Pre- to Post Pre- to Follow-up

tio ac

Fig. 1. Within-subject effect sizes (Cohens d) for all measures. Horizontal reference lines in the gure represent small (0.2), medium (0.5) and large (0.8) effect sizes.

Treatment process analysis Simple Pearson correlation coefcients between residualized changes in outcome measures and residualized changes in process measures were calculated. As Table 2 shows, there were signicant relationships between change scores in the four process measures and change scores in depression, pain-related anxiety and physical and psychosocial disability. All but one of these 16 coefcients met criteria for signicance at a level of p < .01. However, no signicant relations between changes on any process variables and change on the number of medical visits were found. Finally, only changes in acceptance of pain were signicantly correlated with changes in pain intensity, with the direction suggesting that greater increases in acceptance of pain were associated with greater reductions in pain intensity. The correlation analyses also showed that none of the changes in the four process measures was correlated with another at a level that would suggest they were redundant or likely to lead to problems of multicollinearity in regression analyses. All correlations were less than .6, including correlations between change in acceptance of pain and change in mindfulness, r .46, change in psychological acceptance, r .55, and change in values-based action, r .40; correlations between change in mindfulness and change in psychological acceptance, r .49, and values-based action, r .33; and between change in psychological acceptance and change in values-based action, r .41; therefore all process variables were retained for the regression analyses. We also tested correlations between age, gender, education, and duration of pain with the residualized change scores for the ve primary outcome measures and the four process measures. Among these 36 correlations only one was signicant at p < .05. Age was negatively correlated with change in acceptance of pain between

pre-treatment and follow-up, r .17, p < .05. Hence, older age was signicantly associated with less improvement in acceptance of pain, albeit weakly so. Next, hierarchical regression analyses are carried out to investigate the unique and combined contributions of change scores in the four process measures from pre-treatment to post-treatment in accounting for change scores in outcome measures from pretreatment to follow-up. First, demographic variables including sex, age, education, and duration of pain were tested and retained in the equations when signicant (p < .05 to enter, p > .10 to remove). Second, change in pain intensity was entered to statistically control its contribution to explained variance in outcome changes. Finally, change scores in the four process measures, acceptance of pain, values-based action, psychological acceptance and mindfulness, were entered into the equation as a block. Variance estimates (DR2), standardized regression coefcients (beta) and squared semi-partial correlation coefcients (sr2) for these analyses are displayed in Table 3. As shown in Table 3, and as expected from the correlation analyses, none of the demographic variables accounted for a signicant amount of variance in change in any of the outcome measures. Changes in pain intensity entered at Step 1 accounted for an average of 6.8% of pre- to follow-up changes in the outcome measures. Addition of the block of the four primary process variables resulted in a signicant increment of the total variance explained by the equations, average total R2 .25, range 10.0% variance in the equation for medical visits to 34.0% in the equation for psychosocial disability. All the overall models were statistically signicant with the exception of the prediction of change in the number of medical visits; all other F > 5.14, all p < .001. The process variables, acceptance of pain, values-based action, psychological

Table 2 Correlations among residualized change scores of process measures (acceptance of pain, values-based action, psychological acceptance and mindfulness) with residualized changes scores of outcome measures. Depression Acceptance of pain Values-based action Psych acceptance Mindfulness *p < .01. **p < .001. .31* .27* .44** .16 Pain-related anxiety .47** .29* .41** .41** Physical disability .37** .25* .40** .28* Psychosocial disability .31* .34** .48** .33** Medical visits .17 .03 .15 .11 Pain intensity .32** .07 .06 .14

272

L.M. McCracken, O. Gutirrez-Martnez / Behaviour Research and Therapy 49 (2011) 267e274

Table 3 Summary of the hierarchical multiple regression analyses predicting pre- to followup change scores in outcomes measures from pre- to post-treatment change scores in process measures. Dependent variable and step Depression Step 1 Pain intensity Step 2 Acceptance of pain Values-based action Psych acceptance Mindfulness Total R2 Pain-related anxiety Step 1 Pain intensity Step 2 Acceptance of pain Values-based action Psych acceptance Mindfulness Total R2 Physical disability Step 1 Pain intensity Step 2 Acceptance of pain Values-based action Psych acceptance Mindfulness Total R2 Psychosocial disability Step 1 Pain intensity Step 2 Acceptance of pain Values-based action Psych acceptance Mindfulness Total R2 Medical visits Step 1 Pain intensity Step 2 Acceptance of pain Values-based action Psych acceptance Mindfulness Total R2 *p < .05. **p < .01. ***p < .001. .10 .02 .03 .06 .13 .02 .0004 .0025 .010 .0004 .34*** .08** .26* .058 .25*** .12 .17 .42*** .09 .0081 .023 .102 .0049 .20*** .09** .28** .073 .18*** .16 .05 .26* .06 .014 .0025 .040 .0025 .30*** .02 .06 .0036 .25*** .24* .05 .15 .19 .032 .0025 .014 .026 .30*** .05* .11 .012 .20*** .01 .12 .44*** .14 .0001 .010 .116 .012 Predictor

DR2

Beta

sr2

.10** .30** .078

acceptance and mindfulness, explained an additional average 18.0% of the variance in outcome measures, even after controlling for change in pain intensity, with the exception of the prediction of the change on medical visits, all other F > 5.80, all p < .001. In the nal equations, at least one of the process variables made the strongest unique contribution in all equations except for medical visits. Psychological acceptance made the strongest unique contribution on three occasions, for depression, physical disability and psychosocial disability, whereas acceptance of pain made a signicant unique contribution to the prediction of pain-related anxiety. The beta values for mindfulness and values-based action were lower, and they did not make signicant unique contributions in accounting for the changes in outcome variables. Discussion This study assessed the outcomes and processes of change in an ACT-based, interdisciplinary, group treatment for chronic pain.

Immediately following treatment and at 3-month follow-up, participants reported signicantly lower levels of depression, pain-related anxiety, physical and psychosocial disability, medical visits and pain intensity in comparison to the start of treatment. Almost all effect sizes relative to treatment onset remained at a medium or large level at the 3-month follow-up, with the exception of pain intensity and number of medical visits, which were of a small size. These results support ndings from previous studies evaluating ACT-based interventions for people with chronic pain (Dahl, Wilson, & Nilsson, 2004; Vowles & McCracken, 2008; Vowles, McCracken, & Zhao-OBrien, 2010; Wicksell Ahlqvist, et al., 2008) and add to the overall base of evidence supporting the effectiveness of this treatment approach (Hayes et al., 2006). The four processes of psychological exibility included here, acceptance of pain, values-based action, psychological acceptance, and mindfulness, improved signicantly over the time periods analyzed. The effect sizes for acceptance of pain and values-based action were large immediately following treatment and at follow-up. In particular, the effect sizes for acceptance of pain were the largest across all assessed variables, which is consistent with one of the explicit goals of treatment. The effect sizes for general acceptance and mindfulness were medium following treatment and at followup. Overall, changes in the four processes of psychological exibility were signicantly related to changes in depression, pain-related anxiety, physical and psychosocial disability in expected directions, such that increases in these processes were associated with improvements in functioning. However, no relations between changes in the process measures and the number of medical visits were found. Across the pre- to follow-up change scores in the various measures of disability and suffering, the pre- to post-change scores in the four process measures, combined to account for much greater variance, average 18.0%, than that explained by changes in pain intensity, average 6.8%. This occurred despite the fact that pain intensity scores were entered at an earlier and statistically advantageous stage in the regression analyses. These ndings provide clear support for the model underlying ACT with its primary focus on psychological exibility (Dahl et al., 2005; McCracken, 2005). Although this study was not designed primarily to compare the processes underlying psychological exibility with each other, pattern of ndings from the regression analyses suggests that general psychological acceptance had a signicant and unique role to play in the improvements achieved by our sample of complex pain sufferers. This nding lends support to the breadth of the ACT model. It is remarkable that change in general psychological acceptance predicted improvements in outcome beyond those accounted for by acceptance of pain. This suggests that improvements following ACT in those suffering from chronic pain result from an increase in a willingness to experience many varied psychological experiences, unwanted emotional experiences, memories, thoughts, urges, other physical symptoms, and so forth. This reects the frequent observation that the suffering and disability experienced by chronic pain patients do not emerge solely from pain and pain avoidance, but of generalized inexible patterns of experiential avoidance (Hayes et al., 2006). Inexible and avoidant responses to pain are presumably specic behavior patterns within a larger functional class of responses coordinated by a context of fusion and control around a wide array of negatively evaluated private experiences, such as anger, depression, fatigue, fear, frustration, guilt and shame, among others. The present data extend preliminary evidence from a cross-sectional study showing that general psychological acceptance had unique relations with the daily functioning of people with chronic pain even when included in a model with other benecial processes, such as pain acceptance and mindfulness (McCracken & Zhao-OBrien, 2010).

L.M. McCracken, O. Gutirrez-Martnez / Behaviour Research and Therapy 49 (2011) 267e274

273

The present results provide additional support for the broad applicability of ACT and its primary process, psychological exibility (Hayes et al., 2006). Not only do these results show impact across a range of outcome measures, which is itself remarkable, but they also add another positive result to the growing body of results across a range of conditions. There results include outcome and processes data, usually including some variant of the AAQ, for diabetes management (Gregg, Callaghan, Hayes, & Glenn-Lawson, 2007), for epilepsy (Lundgren, Dahl, & Hayes, 2008), for anxiety and mood problems (Forman et al., 2009; Lappalainen et al., 2007), for psychological distress in the workplace (Flaxman & Bond, 2010), and smoking cessation (Gifford et al., 2004), among others. As expected, success in valued action improved signicantly over the time periods analyzed. Also, the increase in this success was consistently and signicantly associated with greater reductions in emotional impacts and disability and greater improvements in functioning. Along with recent process studies examining the role of values (Hayes, Orsillo, & Roemer, 2010; Lundgren et al., 2008; Vowles & McCracken, 2008), the present results provide evidence for another component of the ACT model that has been relatively less studied. However, in the regression analyses, changes in valuesbased action did not contribute signicantly to the measured improvements in the outcome measures. This nding contrasts with an earlier study, in which the long-tem benecial impact of an ACT-based program similar to that used in the present study was mediated by changes in values-based action (Vowles & McCracken, 2008; Vowles et al., 2010). In the previous study, values-based action made signicant unique contributions to the prediction of improvements achieved at follow-up. This inconsistency may be in part due to the fact that current study included other process variables from within the same treatment model, general psychological acceptance and mindfulness. Interactions between the different processes in the ACT model over the course of the therapy and during follow-up phases deserve further research. Patients reported higher levels of mindfulness after the intervention and at the follow-up. Changes in mindfulness were signicantly related to changes in pain-related anxiety and physical and psychosocial disability in expected directions. However, the regression analyses did not lend support for mindfulness as a signicant unique predictor of outcome. It is difcult to precisely understand the role of mindfulness with the methods employed here. However, a small number of studies have reported mediational results for mindfulness measures in ACT with mixed results. For example, in an open trial of ACT for social anxiety, Kocovski et al. (2009) found that changes in mindfulness were signicantly correlated with changes in social anxiety, but further regression analyses did not lend further support for mindfulness as possible mediator. In another open trial of ACT for weight loss, Forman et al. (2009) found that mindfulness only emerged as a potential mediator at 6-month follow-up. More robust support for mindfulness per se as a key process within ACT awaits future studies that utilize more precise, specic, and formal, process analyses, or dismantling methodologies. An interesting unpredicted result emerged from a set of somewhat peripheral analyses of age, gender, education, and duration of pain. With one exception none of these variables was signicantly correlated with changes in outcome or process variables. The one exception was a small, signicant, negative correlation between age and acceptance of pain. This pattern of results suggests that ACT as studied here is equally effective regardless of age, gender, education, or duration of pain. Several limitations of the study should be noted. First, the absence of randomization and an appropriate control condition means that we cannot unambiguously attribute treatment effects to the ACTbased treatment. However, the combination of strong outcomes in patients with long term and intractable conditions, and the pattern of

results involving the process measures, suggest that the present ndings would be unlikely except from some specic processes of treatment. Besides, previous randomized controlled trials have shown that ACT for chronic pain (Dahl et al., 2004; Wicksell, Ahlqvist et al., 2008; Wicksell, Olsson, & Hayes, 2010) yields signicant improvements in outcome measures such as life satisfaction, pain, disability and healthcare use, and that increased psychological exibility explains at least some variance in these results. A second weakness of the study was that the effect size for medical visits was somewhat modest and that the addition of the processes of psychological exibility to the corresponding regression equation did not reliably improve the explained variance in this outcome. These results may be in part due to the large baseline standard deviation in medical visits scores, which suggests that the method used to quantify it could need more development. Further experience may lead us to rene this measure, and to better capture direct and indirect costs, as is needed in treatment cost-effectiveness research. It should be noted that the processes of psychological exibility examined are conceptualized as dynamic behavioral patterns and are technically complex to measure. In a sense we remain in our rst generation of instruments development in this area and future improvements are practically inevitable. For many of the measures used here the ability to report on the behavior patterns included in the measures interacts with the behavior patterns assessed. For example, when a person lacks mindfulness skills they are often unaware of how mindless their behavior is. The reporting of acceptance and values tend to have a similar quality e in treatment one learns to be more accurate and to change the quality of ones behavior at the same time. Future studies of ACT and pain also may benet from including measures of additional aspects of the model that were not well-reected in present study, such as cognitive defusion (Gutirrez, Luciano, Rodrguez, & Fink, 2004; Pez-Blarrina et al., 2008). ACT process research may gain from more measurement intensive designs, from using more varied assessment methods, from collecting process data more frequently during the course of treatment and follow-up, and analyzing it in a way that will allow a detailed examination of the relations of process variables and outcome change. Shared method variance is a perennial problem in research designed as done here. Heavy reliance on retrospective selfreport measures ought to be supplemented with more direct measures that are done closer in time and in situation to the behavior patterns of interest and with a mix of methods that reduce the possibility of method effects inating observed relations. Finally, it will be important to examine the repeatability, long-term stability, and generality of these effects. Although, such an analysis of longer term (three-year) follow-up is currently being done by our research group based on a separate patient sample (Vowles et al., 2010). Along with limitations, this study also has several methodological strengths, such as the size of the sample, chronicity of the disorder, clinical signicance, measurement of key process variables, and the inclusion of follow-up assessment. These features and the data we present add some momentum to the empirical progress being made in relation to ACT in general (Gaudiano, 2009; Levin & Hayes, 2009). Despite limitations, the current study adds to existing research into ACT-based treatments for chronic pain, particularly in the broader range of process measures used. We have shown that changes in processes of psychological exibility appear to participate in important ways in patient outcomes, such as their level of emotional suffering and their physical and social functioning, above and beyond change in pain intensity. It seems worthwhile for clinicians to continue to improve their skills with ACT-based methods and for researchers to continue to design treatment studies more creatively, and examine treatment data more carefully, to promote this growing body of work.

274

L.M. McCracken, O. Gutirrez-Martnez / Behaviour Research and Therapy 49 (2011) 267e274 Lappalainen, R., Lehtonen, T., Skarp, E., Taubert, E., Ojanen, M., & Hayes, S. C. (2007). The impact of CBT and ACT models using psychology trainee therapists: a preliminary controlled effectiveness trial. Behavior Modication, 31, 488e511. Levin, M., & Hayes, S. C. (2009). Is acceptance and commitment therapy superior to established treatment comparisons? Psychotherapy & Psychosomatics, 78, 380. Lundgren, T., Dahl, J., & Hayes, S. C. (2008). Evaluation of mediators of change in the treatment of epilepsy with acceptance and commitment therapy. Journal of Behavioral Medicine, 31, 225e235. Mason, V. L., Mathias, B., & Skevington, S. M. (2008). Accepting low back pain: is it related to a good quality of life? Clinical Journal of Pain, 24, 22e29. McCracken, L. M. (2005). Contextual cognitive-behavioral treatment for chronic pain. Seattle: IASP Press. McCracken, L. M., & Dhingra, L. (2002). A short version of the pain anxiety symptoms scale (PASS-20): preliminary development and validity. Pain Research and Management, 7, 45e50. McCracken, L. M., Gauntlett-Gilbert, J., & Vowles, K. E. (2007). The role of mindfulness in a contextual cognitive-behavioral analysis of chronic pain-related suffering and disability. Pain, 131, 63e69. McCracken, L. M., & Thompson, M. (2009). Components of mindfulness in patients with chronic pain. Journal of Psychopathology and Behavioral Assessment, 31, 75e82. McCracken, L. M., & Velleman, S. C. (2010). Psychological exibility in adults with chronic pain: a study of acceptance, mindfulness, and values-based action in primary care. Pain, 148, 141e147. McCracken, L. M., Vowles, K. E., & Eccleston, C. (2004). Acceptance of chronic pain: component analysis and a revised assessment method. Pain, 107, 159e166. McCracken, L. M., Vowles, K. E., & Eccleston, C. (2005). Acceptance-based treatment for persons with complex, long standing chronic pain: a preliminary analysis of treatment outcome in comparison to a waiting phase. Behaviour Research and Therapy, 43, 1335e1346. McCracken, L. M., Vowles, K. E., & Zhao-OBrien, J. (2010). Further development of an instrument to assess psychological exibility in people with chronic pain. Journal of Behavioral Medicine, 33, 346e354. McCracken, L. M., & Yang, S. (2006). The role of values in a contextual cognitivebehavioral approach to chronic pain. Pain, 123, 137e145. McCracken, L. M., & Zhao-OBrien, J. (2010). General psychological acceptance and chronic pain: there is more to accept than the pain itself. European Journal of Pain, 14, 170e175. Pez-Blarrina, M., Luciano, C., Gutirrez-Martnez, O., Valdivia, S., Ortega, J., & Rodrguez-Valverde, M. (2008). The role of values with personal examples in altering the functions of pain: comparison between acceptance-based and cognitive-control-based protocols. Behaviour Research and Therapy, 46, 84e97. Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879e891. Vowles, K. E., & McCracken, L. M. (2008). Acceptance and values-based action in chronic pain: a study of treatment effectiveness and process. Journal of Consulting and Clinical Psychology, 76, 397e407. Vowles, K. E., McCracken, L. M., McLeod, C., & Eccleston, C. (2008). The chronic pain acceptance questionnaire: conrmatory factor analysis and identication of patient subgroups. Pain, 140, 284e291. Vowles, K. E., McCracken, L. M., & Zhao-OBrien, J. (2010). Acceptance and valuesbased action in chronic pain: an analysis of treatment outcomes and processes three years after treatment completion. The Journal of Pain, 11, S55. Wicksell, R. K., Ahlqvist, J., Bring, A., Melin, L., & Olsson, G. L. (2008). Can exposure and acceptance strategies improve functioning and life satisfaction in people with chronic pain and whiplash-associated disorders (WAD)? A randomized controlled trial. Cognitive Behaviour Therapy, 37, 169e182. Wicksell, R. K., Melin, L., Lekander, M., & Olsson, G. L. (2009). Evaluating the effectiveness of exposure and acceptance strategies to improve functioning and quality of life in longstanding pediatric pain e a randomized controlled trial. Pain, 141, 248e257. Wicksell, R. K., Olsson, G. L., & Hayes, S. C. (2010). Psychological exibility as a mediator of improvement in acceptance and commitment therapy for patients with chronic pain following whiplash. European Journal of Pain. European Journal of Pain, 14, 1059.e1e1059.e11. Wicksell, R. K., Renflt, J., Olsson, G. L., Bond, F. W., & Melin, L. (2008). Avoidance and cognitive fusion e central components in pain related disability? Development and preliminary validation of the Psychological inexiblity in pain scale (PIPS). European Journal of Pain, 12, 491e500.

References
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Bergner, M., & Bobbitt, R. A. (1981). The sickness impact prole: development and nal revision of a health status measure. Medical Care, 19, 787e805. Bond, F. W., Hayes, S. C., Baer, R. A., Carpenter, K. M., Orcutt, H. K., Waltz, T., et al. Preliminary psychometric properties of the acceptance action questionnaire-II: a revised measure of psychological exibility and acceptance, submitted for publication. Brown, K. W., & Ryan, R. M. (2003). The benets of being present: mindfulness and its role in psychological well-being. Journal of Personality and Social Psychology, 84, 822e848. Brown, K. W., Ryan, R. M., & Creswell, J. D. (2007). Mindfulness: theoretical foundations and evidence for its salutary effects. Psychological Inquiry, 18, 211e237. Carleton, R. N., Abrams, M. P., Asmundson, G. J. G., Antony, M. M., & McCabe, R. E. (2009). Pain-related anxiety and anxiety sensitivity across anxiety and depressive disorders. Journal of Anxiety Disorders, 23, 791e798. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Dahl, J., Wilson, K. G., Luciano, M. C., & Hayes, S. C. (2005). Acceptance and commitment therapy for chronic pain. Reno, NV: Context Press. Dahl, J., Wilson, K. G., & Nilsson, A. (2004). Acceptance and commitment therapy and the treatment of persons at risk for long-term disability resulting from stress and pain symptoms: a preliminary randomized trial. Behavior Therapy, 35, 785e801. De Haan, R., Aaronson, N., Limburg, M., Hewer, R. L., & Van Crevel, H. (1993). Measuring quality of life in stroke. Stroke, 24, 320e327. Flaxman, P. E., & Bond, F. W. (2010). A randomised worksite comparison of acceptance and commitment therapy and stress inoculation training. Behaviour Research and Therapy, 48, 816e820. Forman, E. M., Butryn, M. L., Hoffman, K. L., & Herbert, J. D. (2009). An open trial of an acceptance-based behavioral intervention for weight loss. Cognitive and Behavioral Practice, 16, 223e235. Gaudiano, B. A. (2009). sts (2008) methodological comparison of clinical trials of acceptance and commitment therapy versus cognitive behavior therapy: matching apples with oranges? Behaviour Research and Therapy, 47, 1066e1070. Gifford, E. V., Kohlenberg, B. S., Hayes, S. C., Antonuccio, D. O., Piasecki, M. M., Rasmussen-Hall, M. L., et al. (2004). Acceptance-based treatment for smoking cessation. Behavior Therapy, 35, 689e705. Gregg, J. A., Callaghan, G. M., Hayes, S. C., & Glenn-Lawson, J. L. (2007). Improving diabetes self-management through acceptance, mindfulness, and values: a randomized controlled trial. Journal of Consulting and Clinical Psychology, 75, 336e343. Gutirrez, O., Luciano, C., Rodrguez, M., & Fink, B. C. (2004). Comparison between an acceptance-based and a cognitive-control-based protocol for coping with pain. Behavior Therapy, 35, 767e783. Hayes, S. A., Orsillo, S. M., & Roemer, L. (2010). Changes in proposed mechanisms of action during an acceptance-based behavior therapy for generalized anxiety disorder. Behaviour Research and Therapy, 48, 238e245. Hayes, S. C., Luoma, J. B., Bond, F. W., Masuda, A., & Lillis, J. (2006). Acceptance and commitment therapy: model, processes and outcomes. Behaviour Research and Therapy, 44, 1e25. Hayes, S. C., Strosahl, K., & Wilson, K. G. (1999). Acceptance and commitment therapy: An experiential approach to behavior change. New York: Guilford Press. Hayes, S. C., Strosahl, K., Wilson, K. G., Bissett, R. T., Pistorello, J., Toarmino, D., et al. (2004). Measuring experiential avoidance: a preliminary test of a working model. Psychological Record, 54, 553e578. Iverson, G. L., & Remick, R. (2004). Diagnostic accuracy of the British Columbia major depression inventory. Psychological Reports, 95, 1241e1247. Kalkman, J. S., Schillings, M. L., Zwarts, M. J., van Engelen, B. G. M., & Bleijenberg, G. (2007). The development of a model of fatigue in neuromuscular disorders: a longitudinal study. Journal of Psychosomatic Research, 62, 571e579. Kazdin, A. E. (2007). Mediators and mechanisms of change in psychotherapy research. Annual Review of Clinical Psychology, 3, 1e27. Kocovski, N. L., Fleming, J. E., & Rector, N. A. (2009). Mindfulness and acceptancebased group therapy for social anxiety disorder: an open trial. Cognitive and Behavioral Practice, 16, 276e289.