Vous êtes sur la page 1sur 10

PAIN 152 (2011) 27922801


Mediators of change in Acceptance and Commitment Therapy for pediatric

chronic pain
Rikard K. Wicksell a,b,, Gunnar L. Olsson a,c, Steven C. Hayes d
Behavior Medicine Pain Treatment Service, Astrid Lindgren Childrens Hospital, Karolinska University Hospital, Stockholm, Sweden
Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden
Department of Physiology and Pharmacology, Karolinska Institutet, Stockholm, Sweden
Department of Psychology, University of Nevada, Reno, NV, USA

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

a r t i c l e i n f o a b s t r a c t

Article history: Even though psychological interventions are well established in the treatment of pediatric chronic pain,
Received 17 February 2011 there is a clear need for further development, especially with severely disabled patients. However, opti-
Received in revised form 24 August 2011 mizing effectiveness in psychological treatments for pain requires clarication of the mechanisms of
Accepted 8 September 2011
action. Studies addressing change processes are scarce, however, particularly in relation to pediatric
chronic pain. Acceptance and Commitment Therapy (ACT), as an extension of traditional cognitive
behavior therapy, is essentially aimed at improving functioning by increasing the ability to act effec-
tively in the presence of pain and distress, that is, psychological exibility. ACT has shown promising
Chronic pain
results for both adult and pediatric chronic pain. In the present study, the mediators of change in an
Cognitive behavior therapy ACT-oriented treatment for pediatric chronic pain were examined using a bootstrapped cross product
Acceptance and Commitment Therapy of coefcients approach. Pain interference and depression were used as outcome variables. Six different
Mediation analysis variables relevant to theories underlying ACT and cognitive behavior therapy were included in the anal-
Mechanisms of change yses as possible mediators of change: pain impairment beliefs, pain reactivity, self-efcacy, kinesiopho-
bia, catastrophizing, and pain intensity. Results illustrated that pain impairment beliefs and pain
reactivity were the only variables that signicantly mediated the differential effects of treatment on
outcomes at follow-up. Also, these 2 mediators were shown to independently predict effects in out-
come variables at follow-up while controlling for earlier effects in outcome, but only for the ACT con-
dition. Although tentative, the pattern of results suggests that variables consistent with psychological
exibility mediate the effects of ACT-based interventions to improve functioning in patients with
chronic debilitating pain.
2011 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction primary outcome variables, rather than increases in functional

abilities [14,51,79]. Also, evaluations of such interventions have
Pediatric chronic pain is common [16,53], and can be associated largely addressed effectiveness rather than the identication of
with signicant disabilities [25,29,51] that may continue into change processes [69]. Thus, the processes through which psycho-
adulthood [5,78,80]. Psychological treatments, particularly cogni- logical treatments operate are still unclear [48], and the clarica-
tive behavior therapy (CBT), has proven useful for adults with tion of these change processes is considered a central target for
chronic pain [15,24], but randomized controlled trials (RCT) with current clinical research [15,31,32,52]. Although such studies have
pediatric patients are still scarce [14,52]. Psychological treatment recently increased [36,57,65], process analyses of psychological
represents a wide variety of interventions with various objectives treatments of pediatric pain are sparse.
(eg, decreased pain and/or related distress, improved self-efcacy) Recent research has illustrated that greater acceptance of
[63]. Reductions in pain and distress have typically been the chronic pain is associated with less avoidance of important activi-
ties, better emotional well-being, and less health care utilization
Corresponding author at: Behavior Medicine Pain Treatment Service, Astrid
[4143]. Furthermore, the empirical support for acceptance-
Lindgren Childrens Hospital, Karolinska University Hospital, Stockholm 171 76, oriented interventions, primarily Acceptance and Commitment
Sweden. Tel.: +46 0 8 517 79 917; fax: +46 0 8 517 77 265. Therapy (ACT) [22], has increased for both adult [44,46,82] and
E-mail address: Rikard.Wicksell@karolinska.se (R.K. Wicksell). pediatric [83,85,86] chronic pain. ACT was recently listed by Divi-

0304-3959/$36.00 2011 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

R.K. Wicksell et al. / PAIN 152 (2011) 27922801 2793

sion 12 of the American Psychological Association as an empirically indicated by elevated scores on assessments of, for example, dis-
supported treatment for chronic or persistent pain in general ability, kinesiophobia, and depression.
[10]. ACT explicitly aims at improving functioning and quality of life
by teaching patients to experience negative thoughts, emotions, 2.2. Intervention
and bodily sensations in a more open and exible way, in the ser-
vice of engaging in values-oriented behaviors while in the presence 2.2.1. Acceptance and Commitment Therapy (ACT)
of potentially interfering pain and distress. In ACT, this behavioral The ACT intervention was conducted individually and included
pattern is labeled psychological exibility [21]. Studies have indi- approximately 10 weekly sessions with the participant and 12
cated that psychological exibility may moderate and/or mediate sessions with the parents. In total, including parental and follow-
changes in functioning [35,75,87]. However, no study has yet ex- up sessions, the ACT group received between 7 and 20 sessions
plored the mediating effects of variables related to psychological (mean 13, SD 3.5).
exibility in ACT-oriented treatments for pediatric chronic pain. The 2 psychologists involved in the intervention were trained in
We recently reported the results from an RCT with pediatric CBT, and both the psychologists and the physician had experience
longstanding pain, illustrating that ACT was more effective than as well as training in ACT. Treatment delity was maintained
a multidisciplinary treatment including amitriptyline (MDT) [85]. through continuous discussions of treatment processes. We refer
Thus, ACT as a development within CBT appears promising for to previous papers for a more detailed presentation of the clinical
pediatric chronic pain. However, the therapeutic effects of ACT model [82,83,86]. However, a short description of the clinical mod-
may be due to several different change processes. Therefore, this el is provided.
study presents a planned second set of analyses conducted to In ACT [22], avoidance of pain and distress is conceptualized as
investigate these change processes, or mechanisms of action, by a core problem that substantially contributes to disability and re-
evaluating the mediating effects of variables related to psycholog- duced quality of life. According to the theory underlying ACT,
ical exibility in comparison with other variables less associated avoidance occurs primarily when negative thoughts and emotions
with psychological exibility yet commonly used within cognitive have excessive or inappropriate impact on behavior (denoted as
and behavioral theories. cognitive fusion). The core intervention is considered to be acqui-
sition of new behavioral responses during exposure to personally
2. Method important situations and activities that have been previously
avoided due to pain and distress. In contrast to treatments that
2.1. Setting and participants emphasize reduction or control of symptoms, ACT promotes accep-
tance of negative reactions that cannot be directly changed
The setting and participants were thoroughly described in the (thoughts, emotions, bodily sensations) in favor of engaging in
previously presented RCT [85], but a brief presentation will be pro- activities that are meaningful although possibly painful or fear pro-
vided here. Participants were recruited from consecutive patients voking (ie, exposure). As part of this process, the patient is also
with longstanding idiopathic pain referred to the Behavior Medi- trained to distance him/herself from pain and distress in order to
cine Pain Treatment Service at Astrid Lindgren Childrens Hospital, decrease the impact of these experiences on behavior (cognitive
Karolinska University Hospital. Patients aged between 10 and 18 de-fusion). The treatment objective is to improve functioning by
years with pain duration of more than 3 months were considered increasing psychological exibility, dened as the ability to act
eligible for inclusion in the study. Patients were excluded if: (1) effectively in accordance with personal values while in the pres-
pain was explained by an identied pathological process (eg, ence of previously interfering thoughts, emotions, and bodily sen-
arthritis, cancer, inammatory bowel disease); (2) coexisting psy- sations [21].
chiatric or psychosocial issues were considered more relevant than In parental interventions, the shift in perspective from symp-
pain to functioning, including risk for suicide (assessed in the psy- tom alleviation to valued life was emphasized, as well as the
chological screening interview); (3) they had a reduced prociency principles of exposure and operant mechanisms. The parents
in speaking Swedish; (4) they suffered from major cognitive dys- difculties with engaging in effective coaching behaviors due to
functions resulting in difculties following a conversation and/or their own negative reactions were addressed using similar ACT
understanding the description of the study; (5) they were currently processes as described above (ie, values orientation, acceptance,
participating in another rehabilitation program based on cognitive and cognitive defusion).
behavior therapy; and (6) they were previously treated with ami-
triptyline. A total of 32 participants were included in the study 2.2.2. Multidisciplinary treatment and amitriptyline (MDT)
and randomized to 1 of the 2 treatment conditions, and 30 partic- A psychiatrist, a child psychologist, a physiotherapist, and a
ipants completed treatment. Given the exploratory nature of this pain physician, all experienced in working with longstanding
study, only the participants who completed the treatment and pediatric pain, performed the MDT. Participants were seen by the
assessments were included in the analyses. different health care providers based on individual needs. A biobe-
The present sample consisted of 23 girls and 7 boys, with an havioral model of longstanding pain provided a general theoretical
average age of 14.7 years (range 10.818.1) and a mean time since framework for this clinical approach, emphasizing perceived stress
pain onset of 33 months (range 696 months). Of these, 6 partici- in everyday life as an important factor predicting the severity of
pants suffered primarily from headache, 7 were mainly bothered longstanding pain and disability. This approach is supported in
by back and/or neck pain, 6 reported widespread musculoskeletal several articles and summarized in the biobehavioral model of
pain, 6 presented with complex regional pain syndrome, 2 partic- pediatric pain [67,77]. Regarding amitriptyline, doses were in-
ipants reported visceral pain, in 2 cases pain was primarily located creased by 10 mg every week up to 50 mg, and then by 25 mg up
in the lower extremities, and 1 of the participants presented with a to a maximum of 100 mg, with median max doses = 50 mg (mean
postherpetic-type cheek pain. Furthermore, continuous, spontane- 64.3, SD 27.5). The increase of doses was stopped when severe side
ous pain was seen in 21 of the participants, and 9 reported recur- effects appeared (eg, sedation, dry mouth). In total, the MDT group
rent pain. Allodynia or hyperalgesia was present in 15 participants. received between 7 and 59 sessions (mean 22.8, SD 15.4). Amitrip-
As described in the RCT, participants reported a substantial tyline was administered during a period of 1.2 months to 19.6
amount of pain, distress, and difculties with pain adjustment, as months (mean 10.3, SD 5.9) [85].

2794 R.K. Wicksell et al. / PAIN 152 (2011) 27922801

2.3. Assessment 2.4.2. Measures of hypothesized mediators

To examine possible mediators in this ACT-oriented treatment,
Following randomization, all participants completed question- we used 6 measures of constructs that varied in their relevance to
naires and daily ratings during the 2 weeks before treatment, psychological exibility and to other constructs more related to dif-
immediately following the treatment phase (mean 5.3, SD 1.6 ferent CB-oriented theories. Previous research with chronic pain pa-
months after pretreatment assessments), at follow-up 1 (mean 3.5, tients has suggested that behaviors aimed at controlling or avoiding
SD 0.8 months after posttreatment assessments), and at follow-up pain or related distress may be related to poor pain adjustment and
2 (mean 6.8, SD 1.1 months after posttreatment assessments). reduced functioning and quality of life [47]. In contrast, to notice and
accept interfering bodily symptoms, thoughts, and emotions may
2.4. Summary of results from the previously reported RCT facilitate disengagement from unhelpful struggles such as worrying,
and increase engagement in valued activities also in the presence of
In short, signicant improvements with mostly large effect sizes existing symptoms. Accordingly, following an ACT intervention,
were seen for the ACT group over time. Also, the MDT group illus- measures of symptom reduction were hypothesized to be less re-
trated signicant improvements on several measures. The pro- lated to improvements in functioning than instruments related to
longed treatment in the MDT group complicated the comparison pain reactivity and pain impairment beliefs.
between the two conditions. Therefore, we conducted analyses
based on all assessments as well as conned to post assessments Pain impairment beliefs. The Pain and Impairment Relation-
only. When comparing the groups including follow-up assess- ship Scale (PAIRS) assesses patients beliefs concerning the ability to
ments 1 and 2, signicant or nearly signicant differences with function despite pain [58], a key relationship targeted by ACT. The
at least medium effect sizes (in favor of the ACT group) were seen instrument consists of 15 statements that assess to what extent pain
in pain impairment beliefs, pain interference, health-related qual- can prevent engagement in activities, such as As long as I am in
ity of life (mental scale), kinesiophobia, depression, pain intensity, pain, Ill never be able to live as well as I did before. Although PAIRS
and pain reactivity. When analyses were conned to changes from was not explicitly developed to assess psychological in/exibility,
pre- to posttreatment assessments (before extent of treatment the construct assessed by the instrument is highly similar to psycho-
started to diverge between groups), differences between the logical exibility. Psychometric evaluations of the instrument have
groups were generally larger. shown adequate internal consistency [61]. Also, PAIRS could reliably
discriminate between pain and nonpain groups, and the instrument
2.4.1. Measures of outcome variables was signicantly related to impairment even after pain intensity,
In the present study, we used 2 outcome measures (pain inter- duration, and severity of spine dysfunction were controlled for
ference and depression) from the RCT that were related to function- [61]. PAIRS scores have been shown to change signicantly follow-
ing and pain adjustment, and that showed differences between the 2 ing a CBT-oriented treatment, indicating the instruments sensitivity
groups with at least medium effect sizes in analyses when follow-up to change in this type of treatment [19]. The degree to which the par-
assessments were included. It should be noted that also one of the ticipant agreed or disagreed with each statement was rated on a 7-
health-related quality-of-life measures (SF-36, mental scale) ful- point Likert scale (higher scores indicating greater tendency to asso-
lled the above criteria, but since none of the hypothesized media- ciate pain with impairment and to restrict functioning in the pres-
tors were found to be of any functional importance to the relation ence of pain). In one item, changing the wording from work to
between treatment condition and this outcome variable, it was ex- school made an age-appropriate adaptation. In the present set of
cluded from further analyses of relevant change processes. data, Cronbach alpha was .81. Pain interference. The Multidimensional Pain Inventory, Pain reactivity. In the original study we developed an
Interference scale [33] and the Brief Pain Inventory, Pain interfer- instrument appropriate for use in pediatric populations that as-
ence items [6] have been suggested as measures of pain-related sesses the degree of psychological reactivity to longstanding pain.
functioning [11]. A brief inventory, the Pain Interference Index, Pain reactivity (ie, reacting with worrying, anger, and sadness)
was assembled to assess pain interference in adolescents. The 6 rather than pain intensity was hypothesized to predict pain interfer-
questions closely resembled the items in the Multidimensional Pain ence. The instrument contains 5 questions: (1) How often do you
Inventory and Brief Pain Inventory, although age-appropriately for- worry about pain or related symptoms? (2) How often do you think
mulated; for example, Have your symptoms made it difcult for about having pain or other symptoms? (3) How often are you angry
you to manage schoolwork? A composite score of the Pain Interfer- or sad because of pain or related symptoms? (4) How often do you
ence Index was calculated by averaging the 6 items addressing inter- worry about not being able to do things because of pain or related
ference with schoolwork, activities outside school (leisure symptoms? (5) How often do you worry about not being able to do
activities), seeing friends, mood, physical ability, and sleep. The things in the future because of pain or related symptoms? The ques-
items were rated on a 100-mm visual analogue scale from not at tions were rated using a visual analogue scale from never to al-
all to completely. The internal consistency in the RCT was .84. ways. A composite score (ie, mean) was calculated based on the 5
Preliminary analyses based on a large pre-treatment cross-sectional questions, with higher scores indicating more discomfort. Cronbach
sample indicate that the psychometric properties are adequate. alpha in this dataset was .81. Statistical analyses with a large sample
of pediatric pain patients indicate that Pain Reactivity Scale (PRS) Depression. The Center for Epidemiological Studies Depres- has satisfactory psychometric properties, correlates with the Psy-
sion Scale for Children (CES-DC) [81] is designed to assess symp- chological Inexibility in Pain Scale, and explains a substantial
toms of depression. CES can be seen as a measure of pain amount of variance in pain-related functioning [34].
adjustment or emotional functioning in chronic pain populations Previous studies have indicated that worrying about longstand-
and was therefore used in this study as a dependent variable. ing pain is particularly distressing and difcult to dismiss, and may
The reliability and validity of the measure has been established, interfere substantially with everyday life [12]. In ACT, such
especially with adolescents aged 1218 years [17]. The CES-DC negative reactivity is conceptualized as an additional layer of
has been translated to Swedish and has shown adequate reliability discomfort [22] that distinguishes pain sensation from the addi-
[50]. In the present study, the internal consistency was .87. tional distress and reduced behavioral exibility that results from

R.K. Wicksell et al. / PAIN 152 (2011) 27922801 2795

amplifying reactions focused on pain. In the area of anxiety, this physical movement or activity, or kinesiophobia [62,73]. The scale
kind of worry and reactivity occurs as part of a nonfunction pattern consists of 17 items that are rated on a 4-point scale from strongly
of attempting to control and avoid negative emotions [4], and ACT- disagree to strongly agree; higher scores indicate stronger fear
related methods are known to signicantly impact this kind of of (re)injury. The TSK has proven to be a reliable assessment tool
reactivity [20,59]. Thus, interventions aimed at increasing the abil- for longstanding pain in several studies, especially low-back pain
ity to notice and accept pain may well result in less pain reactivity, [7,72]. The internal consistency in this sample was .76.
even if pain intensity remains unchanged, and pain-related reactiv-
ity was therefore considered a possible and theoretically consistent Catastrophizing. Catastrophizing, or internalizing, can be
mediator for ACT. seen as both a maladaptive coping strategy and an expression of
severe pain-related distress. Catastrophizing is frequently sug- Self-efcacy. Self-efcacy represents an important factor in gested as a link between pain and disability [28,60]. Accordingly,
pain-related disability from a traditional CB perspective [9]. Self- the improvements seen in a CB-oriented treatment may be the re-
efcacy can be described as the perceived condence in perform- sult of changes in thought patterns, as conceptualized in the Pain
ing behaviors and overcoming barriers related to, for example, pain Coping Questionnaire [56]. In an ACT treatment, reductions in neg-
[2]. This theoretical construct has been the focus of a large amount ative or catastrophic thoughts are not seen as the primary pro-
of studies, and self-efcacy has repeatedly been shown to be cesses of change, but can occur as values-oriented behaviors and
important in explaining coping styles as well as linking pain to de- psychological exibility increases. Therefore, similar to the Self-
creased functioning and psychological comorbidity [9,27]. The Self- Efcacy Scale and TSK, it was hypothesized that the indirect effects
Efcacy Scale is designed to assess self-efcacy in various activities of catastrophizing should be smaller relative to PAIRS and PRS.
(eg, going shopping, visiting friends) and initially developed for use Catastrophizing is assessed using a subscale of the Pain Coping
with chronic back pain [1]. The instrument consists of 20 items Questionnaire, a self-report instrument for children and adoles-
that are rated on an 11-point scale, with 0 indicating not at all cents from age 8 years that measures how often a particular coping
condent and 10 indicating very condent. Thus, the total range strategy is used, on a scale from 1 (never) to 5 (very often) [56].
is 0200. Following translation to Swedish, the instrument was This subscale has been considered particularly relevant to this pop-
used in a cross-sectional study with subacute, chronic, or recurrent ulation [13]. In the present sample, Cronbach alpha for the subscale
musculoskeletal pain, showing an adequate internal consistency internalizing/catastrophizing was .87.
(alpha .93 and .95 in 2 different samples) [9]. Age adaptations in-
cluded the removal of 2 items, and some items were changed to Pain intensity. The effectiveness of psychological treat-
more age-appropriate activities (eg, driving a car was removed, ments such as CBT is frequently measured by reductions in pain,
go to school and stay the whole day was added). In this data and CB interventions normally address symptom alleviation by,
set, Cronbach alpha was .91. Although self-efcacy is a concept pri- for example, relaxation, biofeedback techniques, and stress man-
marily used in other CB models, its behavioral focus is similar to agement strategies [40,66]. Thus, a well-supported hypothesis is
ACT. However, the self-efcacy model implies that condence that such treatments work by reducing pain intensity in itself
is needed to engage in action. In contrast, from an ACT perspective, [36], which implies that reductions in pain may function as a medi-
it may be possible to act in line with values also in the absence of ating variable. Pain intensity was rated once a day (How much
condence (or similarly, in the presence of pain). Accordingly, pain have you had today?) by the participants on a visual ana-
self-efcacy did not succeed as a mediator in a recent adult ACT logue scale from 0 (Not at all) to 10 (As bad as you can imag-
trial for pain [87], perhaps for this reason. Thus, an ACT treatment ine) during a period of 2 weeks. The mean number of ratings for
may, over time, result in increased condence and self-efcacy as a both groups across all assessment periods was 14.3 (SD 1.8). The
consequence of increased psychological exibility and ability to daily ratings were used to calculate each individuals mean for
engage in activities that were previously avoided. Therefore, in the assessment period, and subsequently the group mean for that
the present study, self-efcacy was hypothesized to be a less rele- period.
vant mediator as compared to pain impairment beliefs and pain
reactivity. 2.5. Mediation analyses Kinesiophobia. Fear of movement and/or reinjury, or kine- Mediators refer to processes through which changes are consid-
siophobia, is a construct that conceptualizes negative thoughts ered to occur [31]. In general, mediation analyses explore the im-
and emotions regarding events associated with pain. The fear- pact of a mediating variable (M) on a relationship between an
avoidance model of chronic pain and disability has gained increas- independent (X) and a dependent (Y) variable. This reects the
ing interest and empirical support during the last decade treatment effect on the outcome measure through a third variable
[37,49,70]. As treatment approaches, there are important similari- (mediator), and mediation effects are therefore referred to as indi-
ties between the fear-avoidance model and ACT, such as the rect effects. Mediation does not show causation, but the functional
emphasis on exposure as the central intervention. However, as importance of the treatments impact on a process (referred to as
with self-efcacy, there are also certain differences between ACT the a path) and that process effect on outcome while controlling
ideas and this more traditional CB conception. In treatment based for treatment (ie, the b path). Mediation is the combination of
on the fear-avoidance model, cognitive interventions may be used these 2 relations, which elevate it above mere correlation by
to correct inaccurate predictions about avoided situations, which requiring that a variable altered by treatment must continue to
may be conceptualized as trying to change the content of thoughts. be functionally relevant over and above that effect.
Also, the interventions (eg, in vivo exposure) are, to an important Analyses of mediators should be based on theoretically relevant
degree, aimed at reducing pain-related fear and anxiety [8]. In con- a priori hypotheses [71]. In order to provide more precise theoret-
trast, an ACT-oriented intervention does not seek to directly reduce ical tests, it is useful, although relatively uncommon, to explore
fear or change the content of thoughts related to kinesiophobia. several plausible mediator variables (ie, specicity), including
Thus, kinesiophobia and psychological exibility may be seen as some not emphasized in the theory underlying the intervention
2 related, but still different, constructs at play in exposure-based [31]. Thus, by considering several different theoretical constructs,
treatment for chronic pain patients. The Tampa Scale of Kinesio- it can be investigated whether the hypothesized mediator is more
phobia (TSK) assesses the participants fear of (re)injury by functionally important than other parallel processes.

2796 R.K. Wicksell et al. / PAIN 152 (2011) 27922801

CBT normally includes interventions aimed at improving the not contain zero, the indirect effect is signicant at the level spec-
ability to self-manage pain [15]. Thus, in this type of treatment ied in the analysis. Simple mediation model analyses were con-
there are several different but related constructs to consider when ducted to evaluate the importance of each hypothesized
exploring the mechanisms (or processes) through which such mediators. Postassessment scores were used for the hypothesized
treatments work. For example, improvements in functioning may mediators. For the outcome measures, 2 different change scores
be the result of both reductions in pain [36] and catastrophic were calculated; pre to follow-up 1, and pre to follow-up 2. Each
thoughts [64]. Differences in coping styles and strategies may be analysis was based on 5000 bootstrapped samples, as suggested
particularly relevant to understand improvements in pain adjust- by Preacher and Hayes [54]. No imputation was used.
ment and functioning [26]. Some types of behavioral and cognitive For the variables that illustrated indirect effects, subsequent
interventions, such as relaxation, stress management, and decatas- analyses were conducted to address the issue of directionality
trophizing, are typically used to reduce pain intensity and pain-re- (ie, that the functional relationship between process and outcome
lated fear/anxiety, and/or change negative inaccurate predictions variables is opposite to what is dened a priori). Specically, the
about future events [8,40,66]. ACT is not primarily aimed at reduc- outcome variables were entered into the analytic model as media-
ing pain or distress, but at increasing the patients ability to expe- tors while the process variables were used as outcome variables,
rience difcult emotions and thoughts without unnecessary essentially inverting the original analyses.
struggle or avoidance, and to act in accordance with values and The central criteria when discussing mediation are specicity
long-term goals while in the presence of previously interfering and timeline. The ab analyses primarily addressed specicity. Gi-
psychological experiences [21,74,82]. ven the signicant changes in both process and outcome vari-
A number of measures were used to evaluate the possible ables at post assessment, further analyses of mediation are
mechanisms of action in this type of treatment. No single measure required to also address the timeline criteria. Therefore, subse-
of psychological exibility was available for pediatric populations quent analyses with data from each condition were conducted
when the previously reported RCT on ACT was conducted [85], to assess the relevance and impact of the strongest mediator
but the variables pain impairment beliefs and pain reactivity were variables. Specically, it was evaluated whether process mea-
considered most closely related to psychological exibility, that is, sures at post predicted changes in outcome measures from pre
the target construct in ACT. The other measures vary across a to follow-up, when controlling for prior improvements, that is,
range; some concepts such as self-efcacy and kinesiophobia are post assessments, in the outcome measures. A series of hierar-
somewhat related, while others (catastrophizing, pain intensity) chical regression analyses were performed, with the change
are more distant. Thus, it was hypothesized that the pattern of scores between pre and follow-up as the dependent variable.
indirect effects would be stronger for pain impairment beliefs First, post levels of the outcome variable were entered as a pre-
and pain reactivity than for the other variables in both pain inter- dictor (step 1), followed by post data for the process variable
ference and depression, and that these effects would be treatment (step 2). The predictive ability of the process variable at post
specic. was, thus, evaluated while controlling for simultaneous effects
in the outcome variable. These analyses were performed with
2.6. Statistical analyses follow-ups 1 and 2, respectively.
Statistical signicance was interpreted conventionally
Initially, bivariate correlations were calculated to broadly char- (P < 0.05 = signicant; P < 0.10 = marginally signicant) but ex-
acterize the relationship between process and outcome variables. act P-values are reported when available, to facilitate a critical
Although hypotheses regarding possible mediators traditionally interpretation of the data [18]. All analyses were performed using
have been tested using the causal steps approach, as outlined by SPSS version 18.0.3 (SPSS Inc, Chicago, IL, USA).
Baron and Kenny [3], the cross-product of the coefcients approach
is today widely viewed as the best overall test of mediation [38]. In 3. Results
this approach, the coefcient for the differential impact of the 2
treatments on the mediator (the a path) is multiplied by the coef- 3.1. Initial analyses
cient for the relation between the mediator and outcomes con-
trolling for treatment (the b path). This cross-product, ab, is Results from the RCT, with the exception of self-efcacy, have
mathematically equivalent to the difference between the total dif- been presented in a previously published paper [85] and are illus-
ferential impact of treatment (in this case, ACT vs MDT) on out- trated in Supplementary Table 1 (available in the online version
come and that impact adjusting for the mediator (cc0 ). The cross only). Notably, there were no signicant pretreatment differences
product of the coefcients method directly assesses the signi- between the groups in any of the outcome or process measures.
cance of the indirect, or mediating, effect. This approach avoids Bivariate correlations were calculated to analyze broadly the
the analytic difculties provided by the mathematically mutual relationship between the hypothesized mediators (post assess-
relation between the a and b paths: when one coefcient increases ment) and the dependent variables (change scores pre to follow-
the other must decrease and vice versa. Because the a  b distribu- up 1 and pre to follow-up 2). Pain-impairment beliefs at post
tion generally violates the assumption of normality required of assessment were signicantly related to both depression change
parametric approaches [3,39,55] a nonparametric bootstrap ap- scores (pre to follow-up 1, r = .471; pre to follow-up 2, r = .479.
proach to the cross product of the coefcients test was used, that Pain-impairment beliefs also correlated with pain interference at
is also more applicable to smaller data sets because of increased pre to follow-up 2, although this was only marginally signicant
power, as compared to traditional methods [55]. A bias-corrected (r = .370). There was a signicant relationship between pain reac-
condence interval is provided for the tested mediators [54,55] tivity post and both of the pain interference change scores (fol-
by calculating the a  b cross product in n bootstrap samples of low-up 1, r = .450; follow-up 2, r = .642), as well as both of the
the original size, drawn from the original data with replacement depression change scores (follow-up 1, r = .510; follow-up 2,
after each value is drawn. The mean value for the a  b product r = .526). Kinesiophobia was signicantly related to pain interfer-
across the bootstrapped samples provides a point estimate of the ence, but only with regards to changes from pre to follow-up 2,
indirect effect. Condence intervals are derived from the obtained (r = .434). Self-efcacy, catastrophizing, and pain at post did not
distribution of a  b scores, with z score-based corrections for bias show a signicant correlation with pain interference or depression
due to the underlying distribution. If lower and upper bounds do change scores.

R.K. Wicksell et al. / PAIN 152 (2011) 27922801 2797

3.2. Test of indirect effects (mediation analyses) or pain intensity did not illustrate any signicant indirect effects
for either of the pain interference change scores.
All mediators were analyzed in relation to the 2 outcome vari-
ables: pain interference and depression. Post assessments were 3.2.2. Depression
used for mediators, and mediational effects on outcome were ana- A similar pattern of mediation effects was seen for depression.
lyzed for both change scores from pre to follow-up 1 and pre to fol- Pain impairment beliefs signicantly mediated the treatment ef-
low-up 2. Results from the mediation analyses are presented in fect on outcome at follow-up 1, and marginally signicant indirect
Tables 1 and 2, illustrating the indirect effects of treatment condi- effects were seen for changes between pre and follow-up 2. Pain
tion through pain impairment beliefs, pain reactivity, self-efcacy, reactivity showed signicant indirect effects for both depression
kinesiophobia, catastrophizing, and pain intensity. Additional change scores. Self-efcacy, kinesiophobia, catastrophizing, and
information, including normal theory (parametric) tests, can be pain intensity all failed to demonstrate any indirect effects on
found in Supplementary Tables 25 (online version only). depression.

3.2.1. Pain interference 3.2.3. Directionality: mediators as outcome

Regarding pain interference, pain impairment beliefs illustrated Analyses to examine the issue of directionality were performed
marginally signicant indirect effects for change scores between with the 2 process measures that illustrated signicant indirect ef-
pre and follow-up 1, but no mediating effects were seen for fects: pain impairment beliefs and pain reactivity. In this set of
changes between pre and follow-up 2. Pain reactivity signicantly analyses, post assessments for pain interference and depression
mediated the effects of treatment on pain interference for both were used as mediators of the impact of treatment on pain impair-
changes between pre and follow-up 1 and changes from pre to fol- ment beliefs and pain reactivity change scores (pre to follow-up 1;
low-up 2. In contrast, self-efcacy, kinesiophobia, catastrophizing, pre to follow-up 2), essentially reversing the original mediation

Table 1
Results from mediation analyses with pain interference as outcome variable.

Outcome: pain interference (change score) Mediator Indirect effect (point estimate a  b) Bootstrap results for indirect effects: bias corrected
and accelerated condence intervals (BCa)a
(95% CI)b (90% CI)c
Lower Upper Lower Upper
Pre to follow-up 1 Pain impairment beliefs .83 .00 2.40
Pain reactivity 1.10 .08 3.01
Self-efcacy .48 .35 1.79
Kinesiophobia .26 .17 1.43
Catastrophizing -.11 1.04 .09
Pain intensity .60 .02 2.18
Pre to follow-up 2 Pain impairment beliefs 1.16 .41 3.18
Pain reactivity 1.69 .17 4.32
Self-efcacy .47 .27 2.32
Kinesiophobia .71 .15 2.62
Catastrophizing .06 .47 .74
Pain intensity .52 .04 2.35

Note: Number of bootstrap resamples = 5000. The indirect effect is statistically signicant at the chosen level when the condence interval does not include zero.
Bootstrap distribution is adjusted for bias and skewness.
Ninety ve percentage condence intervals (CI) equals P < 0.05 level signicance.
Ninety percentage CI equals P < 0.10 level signicance.

Table 2
Results from mediation analyses with depression as outcome variable.

Outcome: depression (change score) Mediator Indirect effect (point estimate a  b) Bootstrap results for indirect effects: bias corrected and
accelerated condence intervals (BCa)a
(95% CI)b (90% CI)c
Lower Upper Lower Upper
Pre to follow-up 1 Pain impairment beliefs 6.89 1.75 14.59
Pain reactivity 5.43 .01 14.77
Self-efcacy 1.01 3.28 7.42
Kinesiophobia 1.44 .85 7.76
Catastrophizing .43 4.73 .90
Pain intensity .78 1.33 6.08
Pre to follow-up 2 Pain impairment beliefs 11.56 2.46 26.55
Pain reactivity 6.05 3.52 31.96
Self-efcacy .33 10.22 3.09
Kinesiophobia 2.57 1.29 16.77
Catastrophizing .55 11.14 .44
Pain intensity .22 12.16 .33

Note: Number of bootstrap resamples = 5000. The indirect effect is statistically signicant at the chosen level when the condence interval does not include zero.
Bootstrap distribution is adjusted for bias and skewness.
Ninety ve percentage condence intervals (CIs) equal P < 0.05 level signicance.
Ninety percentage CIs equal P < 0.10 level signicance.

2798 R.K. Wicksell et al. / PAIN 152 (2011) 27922801

Table 3
Hierarchical regression analyses to evaluate if the identied mediators at post can predict pre to follow-up changes in outcome.*

Condition Predictor (post): R2 R2 F change Sig F Standardized coefcients beta

Pain impairment beliefs change (df) change (with all variables entered)
Outcome: depression Step Beta t Sig beta
ACT Pre to follow-up 1 1 Depression .04 .04 .59 (1,13) .455 .105 .41 .692
2 Pain impairment beliefs .24 .19 3.03 (1,12) .107 .451 1.74 .107
Pre to follow-up 2 1 Depression .02 .02 .31 (1,13) .588 .016 .066 .949
2 Pain impairment beliefs .36 .33 6.22 (1,12) .028 .594 2.50 .028
MDT Pre to follow-up 1 1 Depression .04 .04 .54 (1,13) .477 .067 .18 .859
2 Pain impairment beliefs .12 .08 1.12 (1,12) .310 .391 1.06 .310
Pre to follow-up 2 1 Depression .17 .17 2.71 (1,13) .123 .193 .56 .587
2 Pain impairment beliefs .23 .06 0.90 (1,12) .363 .327 .95 .363
Outcome: pain interference Predictor (post): Pain reactivity
ACT Pre to follow-up 1 1 Pain interference .00 .00 .00 (1,13) .969 .077 2.79 .016
2 Pain reactivity .56 .56 15.50 (1,12) .002 1.079 3.94 .002
Pre to follow-up 2 1 Pain interference .00 .00 .03 (1,13) .864 .805 2.85 .015
2 Pain reactivity .54 .54 13.93 (1,12) .003 1.053 3.73 .003
MDT Pre to follow-up 1 1 Pain interference .00 .00 .07 (1,13) .939 .023 .05 .965
2 Pain reactivity .00 .00 .01 (1,12) .917 .054 .11 .917
Pre to follow-up 2 1 Pain interference .06 .06 .89 (1,13) .362 .051 .11 .917
2 Pain reactivity .11 .04 .59 (1,12) .458 .370 .77 .458
Outcome: depression Predictor (post): Pain reactivity
ACT Pre to follow-up 1 1 Depression .04 .04 .59 (1,13) .455 .191 .58 .574
2 Pain reactivity .26 .21 3.40 (1,12) .090 .609 1.84 .090
Pre to follow-up 2 1 Depression .02 .02 .31 (1,13) .588 .230 .68 .510
2 Pain reactivity .22 .19 2.96 (1,12) .111 .582 1.72 .111
MDT Pre to follow-up 1 1 Depression .04 .04 .54 (1,13) .477 .279 .78 .452
2 Pain reactivity .05 .01 .13 (1,12) .727 .129 .36 .727
Pre to follow-up 2 1 Depression .17 .17 2.71 (1,13) .123 .436 1.30 .218
2 Pain reactivity .17 .00 .01 (1,12) .923 .033 .10 .923

Note: In evaluating the mediators ability to predict later results in the outcome variable, post levels of the outcome variable are controlled for (step 1 in the model). In step 2,
the post assessments in the mediator are added to the model. Analyses are conducted with both pre to follow-up 1 and pre to follow-up 2 changes in the outcome variable.
Analyses are performed for each condition separately, to investigate the treatment specic effects of the mediator on the outcome.
Analyses are conducted for each condition separately, and only for those variables where mediating effects were seen.

analyses. No mediating effects were seen (P > 0.15 for all variables), Pain reactivity as predictor. Pain reactivity was a consistent
providing incremental support for the directionality of the medita- mediator for pain interference and depression, and thus both these
tional effect illustrated in the original analyses. outcome variables were examined. When controlling for pain
interference at post, a signicant amount (56%) of the variance in
3.2.4. Supplementary analyses: effects of process on outcome, pain interference changes from pre to follow-up 1 was explained
controlling for earlier changes in outcome by pain reactivity at post (b = 1.079, t[1,12] = 3.94, P = 0.002). Sim-
A nal set of analyses were conducted to see if the identied ilarly, 54% of the variance in pre to follow-up 2 changes were ex-
mediators at post could predict changes in outcome from pre to plained by pain reactivity at post, in addition to the variance
follow-up while controlling for earlier effects (post assessments) accounted for by pain interference at post assessment (b = 1.053,
in that outcome variable. A series of hierarchical regression analy- t[1,12] = 3.73, P = 0.003). Pain reactivity at post did not predict fol-
ses were performed, using data from each condition individually. low-up effects in pain interference for the MDT condition when the
Based on the results from the mediation analyses, pain impairment post levels of pain interference were accounted for.
beliefs were used as predictor of depression, and pain reactivity Regarding depression, pain reactivity explained 21.1% of the
was used as predictor of both pain interference and depression. variance in changes from pre to follow-up 1 while controlling for
Postassessment scores for the relevant outcome variable were en- depression at post, a marginally signicant effect (b = .609,
tered as step 1 in the model, followed by the postassessment score t[1,12] = 1.84, P = 0.090). These effects were not signicant at fol-
of the mediator (step 2). In these analyses, missing data were im- low-up 2 (b = .582, t[1,12] = 1.72, P = 0.111). Again, for the MDT
puted using the EstimationMaximization Likelihood Method. condition, pain reactivity was not found to signicantly predict
pre to follow-up changes in depression when post levels of depres- Pain impairment beliefs as predictor. Pain impairment beliefs sion were controlled for.
were a consistent mediator only for depression (Tables 1 and 2) and
this relation was therefore further analyzed. Within the ACT condi- 4. Discussion
tion, pain impairment beliefs at post were not signicantly related to
depression change scores from pre to follow-up 1 when controlling There are still relatively few clinical trials evaluating the effec-
for depression post assessments (b = .451, t[1,12] = 1.74, P = 0.107). tiveness of psychological treatments for pediatric chronic pain, and
However, signicant predictive effects were seen for change scores fewer still that focus on increased physical and emotional func-
pre to follow-up 2 (b = .594, t[1,12] = 2.50, P = 0.028), with 33% of tioning rather than symptom alleviation [52]. Pain intensity is only
the variance in depression explained by pain impairment beliefs at one of many psychological variables that contribute to disability in
post while controlling for depression post assessments. In the pain patients [37,84]. This suggests that improvements in function-
MDT condition, post levels of pain impairment beliefs did not relate ing seen in psychological treatments may occur through a variety
signicantly to pre to follow-up changes in depression when con- of distinct change processes, but studies evaluating possible mech-
trolling for post depression (Table 3). anisms of change are remarkably scarce.

R.K. Wicksell et al. / PAIN 152 (2011) 27922801 2799

To address this concern in relation to ACT for pediatric chronic indicates that changes in pain reactivity and pain impairment be-
pain, data from a previously reported RCT were analyzed to evalu- liefs may be more closely related to decreases in pain interference
ate the mediating function of 6 different and theoretically distinct and depression.
variables. According to ACT theory, variables related to psycholog- The present analyses add to an increasing number of studies
ical exibility should demonstrate stronger mediating effects in that support the relevance of ACT-consistent variables in explain-
this type of treatment than variables associated primarily with ing disability among patients with chronic pain, as well as
symptom reduction. At the time of this study, no validated instru- supporting the role of these processes in exposure and accep-
ment of psychological exibility for pediatric chronic pain was tance-oriented treatment. In a mediation analysis of an RCT with
available. However, pain impairment beliefs and pain reactivity adults, psychological inexibility but neither symptom reduction
were considered proxy variables of psychological exibility and (eg, pain, kinesiophobia) nor increases in self-efcacy mediated
thus were hypothesized to illustrate stronger indirect effects than the effects of ACT on functioning and life satisfaction [87]. Also,
self-efcacy, kinesiophobia, catastrophizing, and pain intensity. tests of indirect effects in cross-sectional studies have indicated
Results supported this hypothesis. No mediating effect of treat- that variables related to psychological exibility (eg, willingness,
ment on the outcome variables (ie, pain interference, depression) activities engagement, avoidance, and fusion) may be an important
was seen for pain, catastrophizing, or kinesiophobia. This clearly link between pain and, for example, disability [75,84]. Variables
suggests that relative improvements in pain interference and consistent with ACT theory have repeatedly been shown to explain
depression were not mainly accomplished through reductions in a substantial amount of variance in, for example, functioning and
pain or distress. Self-efcacy also failed to produce any indirect ef- quality of life among people with chronic debilitating pain
fects. In contrast, differential effects on outcome for ACT as com- [41,42,89]. Notably, although kinesiophobia is a relevant and
pared to MDT were mediated by both pain-impairment beliefs well-supported construct in chronic pain, variables related to psy-
and pain reactivity. Furthermore, subsequent analyses showed that chological exibility have been shown to better predict levels of,
within the ACT but not the MDT, condition pain impairment beliefs for example, functioning, life quality, and pain adjustment [84,88].
and pain reactivity at post predicted a large proportion of changes ACT shares important features with other psychological ap-
in follow-up outcomes after controlling for post outcomes. For proaches, particularly CBT (eg, exposure), but differences also exist
example, more than 50% of the variance in pre to follow-up [76]. For example, acceptance and defusion strategies may be more
changes in pain interference was explained by pain reactivity at or less unique for ACT. On the other hand, ACT is not utilizing cog-
post when controlling for post levels of pain interference. nitive (eg, decatastrophizing) and behavioral (eg, relaxation) inter-
There have long been calls for the use of mediation analysis to ventions to correct inaccurate predictions and to reduce pain, fear,
evaluate mechanisms of action [3], but interest has grown only re- or anxiety as done in other clinical models [8,40,66]. The explora-
cently. For mediational effects to be clearly demonstrated, several tion of change processes in psychological treatments should
requirements need to be met [30]. Analyses clearly benet from an empirically evaluate whether or not conceptual differences be-
experimental design where the treatment variable is manipulated. tween clinical models correspond with results from mediation
Also, proposed mediators should be based on theory, be clearly analyses. Potentially, 2 treatments may appear rather different
operationalized, and have a dened relation to the outcome vari- but work through more or less the same change processes, or vice
ables. The criteria of temporality and specicity are commonly versa. Thus, process analyses are central to empirically establish
seen as critical, but studies that satisfactorily address these are rel- functional similarities and differences between treatments.
atively rare. To ascertain temporality, the change in mediator must Although tentatively, the pattern of results from this and a sim-
be shown to precede changes in outcome [23]. ACT processes have ilar mediation study of ACT [87] contrast some recent ndings
mediated outcomes without temporality being violated in areas from process analyses of CBT. For example, a brief CBT protocol
other than pain [23,68]. In the present study, the use of post assess- for chronic pain was recently shown to work through improve-
ments for the mediators, and the change scores from pre to follow- ments in self-efcacy, catastrophizing, and perceived control
up assessments for the outcome variables, as well as the analyses [65]. In a different study investigating the working mechanisms
of directionality, is in line with the timeline criterion. Also, the pre- of CBT for irritable bowel syndrome, it was concluded that CBT
dictive ability of the mediators when controlling for earlier effects had direct effects on gastrointestinal symptoms and that this was
on outcome variables provides additional support regarding tem- not mainly a consequence of reduced distress [36]. Thus, although
porality. However, to fully ascertain temporality, future studies data are still limited, it appears that ACT may differ from other
should assess these processes earlier and more frequently during interventions within CBT with regards to processes of change.
the course of treatment, to disentangle the key functional path- Different types of CB treatments may work through a variety of
ways and increase the possibility of addressing whether or not mediators and through distinct processes. More research is needed
the proposed mediators are causally related to the outcome to clarify the processes of change also in other types of treatment
variables. for pediatric chronic pain (eg, relaxation, parental interventions).
To meet the criterion of specicity, several potential mediators In addition, direct comparisons of interventions are needed to fur-
should be assessed to ascertain that not just any variable function ther explore differences and similarities among various psycholog-
as mediator, and that change in treatment is consistent with the ical treatments for pain. Differences between clinical approaches
underlying theory [30]. This study contributes to a more general may open up the possibility of detecting moderators for when each
effort aimed at clarifying the processes through which psycholog- type of intervention is likely to be most helpful.
ical interventions are effective [21,45,48]. Primarily, the present This represents the rst study exploring change processes in
study sought to evaluate the specicity of mediational effects in ACT for pediatric chronic pain, and a number of weaknesses should
ACT by examining several possible mediators that were more or be considered. The bootstrapped cross product of coefcients ap-
less consistent with ACT theory. As presented, the more ACT-re- proach facilitates mediation analyses of small data sets, and the
lated variables illustrated the strongest mediating effects. Supple- lack of signicant ndings in some process measures may be re-
mentary analyses illustrated that the mediators ability to explain lated to statistical power. Unfortunately, there is yet no metric
later effects in outcome were seen only for the ACT condition, sug- available for estimating the effect size of mediation in datasets
gesting that these change processes are associated with ACT inter- with fewer than 500 participants [38]. Thus, although the results
ventions. Interestingly, the absence of mediating effects in this from the present study correspond with previous analyses of sim-
study on health-related quality of life (SF-36, mental scale) ilar type [87], the small sample implies that results should be seen

2800 R.K. Wicksell et al. / PAIN 152 (2011) 27922801

as tentative. Studies with different and larger samples are needed [14] Eccleston C, Palermo TM, Williams AC, Lewandowski A, Morley S.
Psychological therapies for the management of chronic and recurrent pain in
to assess the validity of these ndings. In the present study, PAIRS
children and adolescents. Cochrane Database Syst Rev 2009;2:CD003968.
(pain impairment beliefs) and PRS (pain reactivity) were used as [15] Eccleston C, Williams AC, Morley S. Psychological therapies for the
proxy measures of ACT-consistent processes. More ACT-specic management of chronic pain (excluding headache) in adults. Cochrane
measures for pediatric populations, such as the Psychological Database Syst Rev 2009;2:CD007407.
[16] El-Metwally A, Salminen JJ, Auvinen A, Kautiainen H, Mikkelsson M. Prognosis
Inexibility in Pain Scale [84], will provide more precision in future of non-specic musculoskeletal pain in preadolescents: a prospective 4-year
studies. Also, the statistical properties of PAIRS should be validated follow-up study till adolescence. Pain 2004;110:5509.
for youths. [17] Fendrich M, Weissman MM, Warner V. Screening for depressive disorder in
children and adolescents: validating the center for epidemiologic studies
In conclusion, further improvements of psychological interven- depression scale for children. Am J Epidemiol 1990;131:53851.
tions for children and adolescents with chronic debilitating pain [18] Greenwald AG, Gonzalez R, Harris RJ, Guthrie D. Effect sizes and p values: what
require that the mechanisms of change in such treatments be clar- should be reported and what should be replicated? Psychophysiology
ied. Acceptance-oriented interventions have recently received [19] Guck TP, Fleischer TD, Willcockson JC, Criscuolo CM, Leibrock LG. Predictive
increasing attention and empirical support, but the present study validity of the pain and impairment relationship scale in a chronic
represents the rst analysis of mediating effects in ACT treatment nonmalignant pain population. Arch Phys Med Rehabil 1999;80:915.
[20] Hayes SA, Orsillo SM, Roemer L. Changes in proposed mechanisms of action
for pediatric pain patients. Results illustrate that improvements oc- during an acceptance-based behavior therapy for generalized anxiety disorder.
cur through changes in pain impairment beliefs and reductions in Behav Res Ther 2010;48:23845.
pain reactivity rather than through changes in self-efcacy, kine- [21] Hayes SC, Luoma JB, Bond FW, Masuda A, Lillis J. Acceptance and
commitment therapy: model, processes and outcomes. Behav Res Ther
siophobia, catastrophizing, or pain intensity. Although more re-
search is needed, the present study provides additional support [22] Hayes SC, Strosahl KD, Wilson KG. Acceptance and commitment therapy: an
for processes related to ACT theory in treatments for chronic pain experiential approach to behavior change. New York: Guilford Press; 1999.
patients aimed at improving physical and emotional functioning. [23] Hesser H, Westin V, Hayes SC, Andersson G. Clients in-session acceptance and
cognitive defusion behaviors in acceptance-based treatment of tinnitus
distress. Behav Res Ther 2009;47:5238.
Conict of interest statement [24] Hoffman BM, Papas RK, Chatkoff DK, Kerns RD. Meta-analysis of psychological
interventions for chronic low back pain. Health Psychol 2007;26:19.
[25] Huguet A, Miro J. The severity of chronic pediatric pain: an epidemiological
There are no nancial or other relationships that might lead to a study. J Pain 2008;9:22636.
conict of interest. [26] Jensen MP, Turner JA, Romano JM. Correlates of improvement in
multidisciplinary treatment of chronic pain. J Consult Clin Psychol
Appendix A. Supplementary data [27] Jensen MP, Turner JA, Romano JM. Self-efcacy and outcome expectancies:
relationship to chronic pain coping strategies and adjustment. Pain
Supplementary data associated with this article can be found, in [28] Jensen MP, Turner JA, Romano JM, Karoly P. Coping with chronic pain: a critical
the online version, at doi:10.1016/j.pain.2011.09.003. review of the literature. Pain 1991;47:24983.
[29] Kashikar-Zuck S, Graham TB, Huenefeld MD, Powers SW. A review of
biobehavioral research in juvenile primary bromyalgia syndrome. Arthritis
References Care Res 2000;13:38897.
[30] Kazdin AE. Mediators and mechanisms of change in psychotherapy research.
[1] Altmaier EM, Lehmann TR, Russell DW, Weinstein JN, Kao CF. The effectiveness Annu Rev Clin Psychol 2007;3:127.
of psychological interventions for the rehabilitation of low back pain: a [31] Kazdin AE, Nock MK. Delineating mechanisms of change in child and
randomized controlled trial evaluation. Pain 1992;49:32935. adolescent therapy: methodological issues and research recommendations. J
[2] Bandura A. Self-efcacy: toward a unifying theory of behavioral change. Child Psychol Psychiatry 2003;44:111629.
Psychol Rev 1977;84:191215. [32] Keefe FJ, Bufngton AL, Studts JL, Rumble ME. Behavioral medicine: 2002 and
[3] Baron RM, Kenny DA. The moderatormediator variable distinction in social beyond. J Consult Clin Psychol 2002;70:8526.
psychological research: conceptual, strategic, and statistical considerations. J [33] Kerns RD, Turk DC, Rudy TE. The west haven-yale multidimensional pain
Pers Soc Psychol 1986;51:117382. inventory (WHYMPI). Pain 1985;23:34556.
[4] Borkovec TD, Alcaine OM, Behar E. Avoidance theory of worry and generalized [34] Klinga M, Olsson G, Wicksell R. Pain related functioning in children with
anxiety disorder. In: Heimberg RG, Turk CL, Mennin DS, editors. Generalized longstanding painrelation to pain related worrying and psychological
anxiety disorders: advances in research and practice. New York: Guilford; exibility. Congress Abstract for Pediatric Pain. Turku, Finland; 2010.
2004. p. 77108. [35] Kratz AL, Davis MC, Zautra AJ. Pain acceptance moderates the relation between
[5] Brattberg G. Do pain problems in young school children persist into early pain and negative affect in female osteoarthritis and bromyalgia patients.
adulthood? A 13-year follow-up. Eur J Pain 2004;8:18799. Ann Behav Med 2007;33:291301.
[6] Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain Inventory. [36] Lackner JM, Jaccard J, Krasner SS, Katz LA, Gudleski GD, Blanchard EB. How
Ann Acad Med Singapore 1994;23:12938. does cognitive behavior therapy for irritable bowel syndrome work? A
[7] Crombez G, Vlaeyen JW, Heuts PH, Lysens R. Pain-related fear is more disabling mediational analysis of a randomized clinical trial. Gastroenterology
than pain itself: evidence on the role of pain-related fear in chronic back pain 2007;133:43344.
disability. Pain 1999;80:32939. [37] Leeuw M, Goossens ME, Linton SJ, Crombez G, Boersma K, Vlaeyen JW. The
[8] de Jong JR, Vlaeyen JW, Onghena P, Goossens ME, Geilen M, Mulder H. Fear of fear-avoidance model of musculoskeletal pain: current state of scientic
movement/(re)injury in chronic low back pain: education or exposure in vivo evidence. J Behav Med 2007;30:7794.
as mediator to fear reduction? Clin J Pain 2005;21:917 [discussion 6972]. [38] MacKinnon DP, Fairchild AJ, Fritz MS. Mediation analysis. Annu Rev Psychol
[9] Denison E, Asenlof P, Lindberg P. Self-efcacy, fear avoidance, and pain 2007;58:593614.
intensity as predictors of disability in subacute and chronic musculoskeletal [39] MacKinnon DP, Lockwood CM, Hoffman JM, West SG, Sheets V. A comparison
pain patients in primary health care. Pain 2004;111:24552. of methods to test mediation and other intervening variable effects. Psychol
[10] American Psychological Association. Society of Clinical Psychology, Division Methods 2002;7:83104.
12. Psychological treatments. Available from: <http://www.div12.org/ [40] Marhold C, Linton SJ, Melin L. A cognitive-behavioral return-to-work program:
PsychologicalTreatments/treatments.html>. (accessed 01.05.2011). effects on pain patients with a history of long-term versus short-term sick
[11] Dworkin RH, Turk DC, Farrar JT, Haythornthwaite JA, Jensen MP, Katz NP, Kerns leave. Pain 2001;91:15563.
RD, Stucki G, Allen RR, Bellamy N, Carr DB, Chandler J, Cowan P, Dionne R, [41] McCracken LM. Learning to live with the pain: acceptance of pain predicts
Galer BS, Hertz S, Jadad AR, Kramer LD, Manning DC, Martin S, McCormick CG, adjustment in persons with chronic pain. Pain 1999;74:217.
McDermott MP, McGrath P, Quessy S, Rappaport BA, Robbins W, Robinson JP, [42] McCracken LM, Eccleston C. Coping or acceptance: what to do about chronic
Rothman M, Royal MA, Simon L, Stauffer JW, Stein W, Tollett J, Wernicke J, pain? Pain 2003;105:197204.
Witter J. Core outcome measures for chronic pain clinical trials: IMMPACT [43] McCracken LM, Eccleston C. A prospective study of acceptance of pain and
recommendations. Pain 2005;113:919. patient functioning with chronic pain. Pain 2005;118:1649.
[12] Eccleston C, Crombez G, Aldrich S, Stannard C. Worry and chronic pain [44] McCracken LM, MacKichan F, Eccleston C. Contextual cognitive-behavioral
patients: a description and analysis of individual differences. Eur J Pain therapy for severely disabled chronic pain sufferers: effectiveness and
2001;5:30918. clinically signicant change. Eur J Pain 2007;11:31422.
[13] Eccleston C, Malleson PN, Clinch J, Connell H, Sourbut C. Chronic pain in [45] McCracken LM, Turk DC. Behavioral and cognitive-behavioral treatment for
adolescents: evaluation of a programme of interdisciplinary cognitive chronic pain: outcome, predictors of outcome, and treatment process. Spine
behaviour therapy. Arch Dis Child 2003;88:8815. 2002;27:256473.

R.K. Wicksell et al. / PAIN 152 (2011) 27922801 2801

[46] McCracken LM, Vowles KE, Eccleston C. Acceptance-based treatment for [68] Varra AA, Hayes SC, Roget N, Fisher G. A randomized control trial examining
persons with complex, long standing chronic pain: a preliminary analysis of the effect of acceptance and commitment training on clinician willingness to
treatment outcome in comparison to a waiting phase. Behav Res Ther use evidence-based pharmacotherapy. J Consult Clin Psychol 2008;76:44958.
2005;43:133546. [69] Vendrig AA, van Akkerveeken PF, McWhorter KR. Results of a multimodal
[47] McCracken LM, Vowles KE, Gauntlett-Gilbert J. A prospective investigation of treatment program for patients with chronic symptoms after a whiplash injury
acceptance and control-oriented coping with chronic pain. J Behav Med of the neck. Spine 2000;25:23844.
2007;30:33949. [70] Vlaeyen JW, Linton SJ. Fear-avoidance and its consequences in chronic
[48] Morley S. Process and change in cognitive behaviour therapy for chronic pain. musculoskeletal pain: a state of the art. Pain 2000;85:31732.
Pain 2004;109:2056. [71] Vlaeyen JW, Morley S. Cognitive-behavioral treatments for chronic pain: what
[49] Nieto R, Miro J, Huguet A. The fear-avoidance model in whiplash injuries. Eur J works for whom? Clin J Pain 2005;21:18.
Pain 2009;13:51823. [72] Vlaeyen JWS, Kole-Snijders AMJ, Boeren RG, van Eek H. Fear of movement/
[50] Olsson G, von Knorring AL. Depression among Swedish adolescents measured (re)injury in chronic low back pain and its relation to behavioral performance.
by the self-rating scale center for epidemiology studies-depression child (CES- Pain 1995;62:36372.
DC). Eur Child Adolesc Psychiatry 1997;6:817. [73] Vlaeyen JWS, Kole-Snijders AMJ, Rotteveel AM, Ruesink R, Heuts P. The role of
[51] Palermo TM. Impact of recurrent and chronic pain on child and family daily fear of movement/(re)injury in pain disability. J Occup Rehabil 1995;5:23552.
functioning: a critical review of the literature. J Dev Behav Pediatr [74] Vowles KE, McCracken LM. Acceptance and values-based action in chronic
2000;21:5869. pain: a study of treatment effectiveness and process. J Consult Clin Psychol
[52] Palermo TM, Eccleston C, Lewandowski AS, Williams AC, Morley S. 2008;76:397407.
Randomized controlled trials of psychological therapies for management of [75] Vowles KE, McCracken LM, Eccleston C. Patient functioning and
chronic pain in children and adolescents: an updated meta-analytic review. catastrophizing in chronic pain: the mediating effects of acceptance. Health
Pain 2010;148:38797. Psychol 2008;27:S13643.
[53] Perquin CW, Hazebroek-Kampschreur AA, Hunfeld JA, Bohnen AM, van [76] Vowles KE, McCracken LM, Eccleston C. Processes of change in treatment for
Suijlekom-Smit LW, Passchier J, van der Wouden JC. Pain in children and chronic pain: the contributions of pain, acceptance, and catastrophizing. Eur J
adolescents: a common experience. Pain 2000;87:518. Pain 2007;11:77987.
[54] Preacher KJ, Hayes AF. Asymptotic and resampling strategies for assessing and [77] Walker LS, Garber J, Smith CA, Van Slyke DA, Claar RL. The relation of daily
comparing indirect effects in multiple mediator models. Behav Res Methods stressors to somatic and emotional symptoms in children with and without
2008;40:87991. recurrent abdominal pain. J Consult Clin Psychol 2001;69:8591.
[55] Preacher KJ, Hayes AF. SPSS and SAS procedures for estimating indirect effects [78] Walker LS, Garber J, Van Slyke DA, Greene JW. Long-term health outcomes in
in simple mediation models. Behav Res Methods Instrum Comput patients with recurrent abdominal pain. J Pediatr Psychol 1995;20:23345.
2004;36:71731. [79] Walker LS, Greene JW. The functional disability inventory: measuring a
[56] Reid GJ, Gilbert CA, McGrath PJ. The pain coping questionnaire: preliminary neglected dimension of child health status. J Pediatr Psychol 1991;16:3958.
validation. Pain 1998;76:8396. [80] Walker LS, Guite JW, Duke M, Barnard JA, Greene JW. Recurrent abdominal
[57] Rhee SH, Parker JC, Smarr KL, Petroski GF, Johnson JC, Hewett JE, Wright GE, pain: a potential precursor of irritable bowel syndrome in adolescents and
Multon KD, Walker SE. Stress management in rheumatoid arthritis: what is the young adults. J Pediatr 1998;132:10105.
underlying mechanism? Arthritis Care Res 2000;13:43542. [81] Weissman MM, Orvaschel H, Padian N. Childrens symptom and social
[58] Riley JF, Ahern DK, Follick MJ. Chronic pain and functional impairment: functioning self-report scales. Comparison of mothers and childrens
assessing beliefs about their relationship. Arch Phys Med Rehabil reports. J Nerv Ment Dis 1980;168:73640.
1988;69:57982. [82] Wicksell RK, Ahlqvist J, Bring A, Melin L, Olsson GL. Can exposure and
[59] Roemer L, Orsillo SM, Salters-Pedneault K. Efcacy of an acceptance-based acceptance strategies improve functioning and life satisfaction in people with
behavior therapy for generalized anxiety disorder: evaluation in a randomized chronic pain and whiplash-associated disorders (WAD)? A randomized
controlled trial. J Consult Clin Psychol 2008;76:10839. controlled trial. Cogn Behav Ther 2008;37:16982.
[60] Severeijns R, Vlaeyen JW, van den Hout MA, Weber WE. Pain catastrophizing [83] Wicksell RK, Dahl J, Magnusson B, Olsson GL. Using acceptance and
predicts pain intensity, disability, and psychological distress independent of commitment therapy in the rehabilitation of an adolescent female with
the level of physical impairment. Clin J Pain 2001;17:16572. chronic pain: a case example. Cogn Behav Pract 2005;12:41523.
[61] Slater MA, Hall HF, Atkinson JH, Garn SR. Pain and impairment beliefs in [84] Wicksell RK, Lekander M, Sorjonen K, Olsson GL. The psychological inexibility in
chronic low back pain: validation of the pain and impairment relationship pain scale (PIPS)statistical properties and model t of an instrument to assess
scale (PAIRS). Pain 1991;44:516. change processes in pain related disability. Eur J Pain 2010;14. 771. e114.
[62] Swinkels-Meewisse EJ, Swinkels RA, Verbeek AL, Vlaeyen JW, Oostendorp RA. [85] Wicksell RK, Melin L, Lekander M, Olsson GL. Evaluating the effectiveness of
Psychometric properties of the Tampa Scale for kinesiophobia and the fear- exposure and acceptance strategies to improve functioning and quality of life
avoidance beliefs questionnaire in acute low back pain. Man Ther in longstanding pediatric paina randomized controlled trial. Pain
2003;8:2936. 2009;141:24857.
[63] Turk DC, Meichenbaum D, Genest M. Pain and behavioral medicine: a [86] Wicksell RK, Melin L, Olsson GL. Exposure and acceptance in the rehabilitation
cognitive-behavioral perspective. New York: Guilford Press; 1983. of adolescents with idiopathic chronic paina pilot study. Eur J Pain
[64] Turk DC, Okifuji A. Psychological factors in chronic pain: evolution and 2007;11:26774.
revolution. J Consult Clin Psychol 2002;70:67890. [87] Wicksell RK, Olsson GL, Hayes SC. Psychological exibility as a mediator of
[65] Turner JA, Holtzman S, Mancl L. Mediators, moderators, and predictors of improvement in acceptance and commitment therapy for patients with
therapeutic change in cognitive-behavioral therapy for chronic pain. Pain chronic pain following whiplash. Eur J Pain 2010;14. 1059. e111.
2007;127:27686. [88] Wicksell RK, Olsson GL, Melin L. The Chronic Pain Acceptance Questionnaire
[66] Turner JA, Mancl L, Aaron LA. Short- and long-term efcacy of brief cognitive- (CPAQ)further validation including a conrmatory factor analysis and a
behavioral therapy for patients with chronic temporomandibular disorder comparison with the Tampa Scale of Kinesiophobia. Eur J Pain 2009;13:7608.
pain: a randomized, controlled trial. Pain 2006;121:18194. [89] Wicksell RK, Renofalt J, Olsson GL, Bond FW, Melin L. Avoidance and cognitive
[67] Varni JW, Rapoff MA, Waldron SA, Gragg RA, Bernstein BH, Lindsley CB. fusion-central components in pain related disability? Development and
Effects of perceived stress on pediatric chronic pain. J Behav Med preliminary validation of the psychological inexibility in pain scale (PIPS).
1996;19:51528. Eur J Pain 2008;12:491500.