Efficacy and Effectiveness of Cognitive Behaviour Therapy For Chronic Pain: Progress and Some Challenges
Efficacy and Effectiveness of Cognitive Behaviour Therapy For Chronic Pain: Progress and Some Challenges
Stephen Morley
University of Leeds, UK
Correspondence
Stephen Morley PhD, Leeds Institute of Health Sciences, University of Leeds, 101
Published as:
1
Abstract
CBT is a class of treatments that have developed since the 1960s in response to
associated with the experience of pain range from the momentary processes
briefly reviews the main features of contemporary CBT, summarises evidence for its
studies. Finally, attention is given to two areas that require further attention: (1) The
outcome, and (2) procedures for assuring the quality of treatment implementation.
2
A very brief introduction to CBT
psychological treatments for a very wide range of problems including chronic pain.
But what is CBT? Where did it come from? Can we regard it as a unified treatment?
How effective and efficacious is it in the field of chronic pain? What challenges face
practitioners and researchers in the field? The aims of this article are: to provide a
historically based brief review of the status of contemporary CBT for chronic pain; to
note the evidence for its efficacy from randomised controlled trials; to outline the use
of data from RCTs to provide benchmarks for routine clinical treatment and the
practice.
“The term CBT varies widely and may include self instructions … relaxation or biofeedback,
developing coping strategies, changing maladaptive beliefs about pain and goal setting …
management.” Gatchel and colleagues are not alone in noting the broad remit of CBT
and its relative lack of specificity e.g. [12,73]. This definition does capture the
breadth of CBT, the mixture of techniques, aims and the variation in the context in
3
which CBT is practiced. To understand some of the key elements it is useful to see
how the main strands have developed and where they have been incorporated into
TABLE 1 HERE
chronic pain beginning with the application of behaviour analysis by Fordyce in the
1960s [19] and ending with the introduction of Acceptance and Commitment
Therapy in the late 1990’s and from 2000 onwards [37]. The Table also indicates the
origins of each of the strands. Both Fordyce’s application of operant principles and
ACT have their roots in the analysis of respondent behaviour originating with
function of two significant classes of external factors. The first class, reinforcement,
The second class, antecedents, refers to the context in which behaviour occurs and
private experience there are publicly observable expressions of pain that are subject
may be modified in unhelpful ways. Similarly, it could be changed using the same
4
the function of language and rule governed behaviour [22]. The complexity of ACT
is harder to grasp than the Fordyce’s earlier implementation but in essence the aims
are essentially the same, to change the control over behaviour that pain exerts by
into treatment protocols. Biofeedback also had its roots in the analysis of learning.
been for chronic headache which targeted striated muscle [25]. Both biofeedback and
relaxation were incorporated into treatments for pain aimed at modifying the
patients might be construed as a stress response was also developed in the 1970s.
Elements of this analysis were drawn from the work in the preceding decade that
stimulation (stressors) by Lazarus and colleagues. This work established two crucial
ideas in the pain field, appraisal and coping, that are still current e.g. catastrophizing
and coping skills training. Also by the mid 1970’s work on self-control, also with its
roots in behaviour analysis [36] was incorporated into the treatment armamentarium
5
and a definitive text ‘Pain and behavioral medicine: a cognitive-behavioral perspective’
At the turn of the 1970s Beck published a text on the treatment of depression
by cognitive therapy [4]. The significance of this was that a substantive claim was
made that a disorder that had previously been difficult to treat using psychological
methods was treatable. Beck and Ellis had developed separate versions of what
became known as cognitive therapy in the previous two decades. The treatment was
grounded in acute clinical observation and a willingness to devise and test treatment
Perhaps because of the marked overlap between pain and depression elements of
Beck’s therapy were incorporated into pain treatments and the model was adapted
6
Buddhist teaching and 3000 years of practice. Although Kabat-Zinn reported studies
on pain patients it was not until his work was taken up by a group of researchers
looking for a method to prevent relapse after treatment for depression [58] that
The final strand of current CBT is the application of the generalized fear-
avoidance model to chronic pain. The modern experimental and clinical analysis of
fear and avoidance has a long history dating to the work of Masserman, Miller and
Mowrer in the 1940s. This work shaped the development of behavioural treatments
analysis and application to chronic pain was provided in a seminal paper by Vlaeyen
who express fears that engaging in specified movements will result in catastrophic
appraisals by engaging in the very behaviour of which they are frightened. In many
regards the fear-avoidance model captures the essence of CBT which is the careful
7
The literature on the psychology of pain is extensive but much of it can be
[42,44]. Interruption refers to the impact of pain on the disruption of attention and its
their social group. Repeated interference of key tasks impacts upon the person’s
sense of who they are and, perhaps more importantly, whom they might become by
distorting their vision of their future and reshaping their view of the past
[10,44,49,51,]. The importance of each theme varies across people and the duration
of pain. Brief laboratory pain has interruptive effects but unlikely to produce
interference or impact on identity. Acute clinical pain has both interruptive and
pain frequently has a profound effect on all three categories. Repeated interference
with tasks that are essential to achieving various life goals and maintaining a
person’s status in society impact on their sense of self and their future plans [43,57].
This brief analysis illustrates the breadth of disruption that pain may have on
8
relieve elements of suffering associated at each level of interruption, interference and
identity.
is widely accepted. The major tool for building evidence is the randomised
controlled trial (RCT) and the strategy has been to establish evidence for the efficacy
which the experimental constraints are relaxed so that the intervention can be tested
in routine practice. (In practice this distinction may be somewhat blurred.) The
controlled trials, albeit with some justification, but perhaps to the neglect of evidence
generated in clinical practice [2]. Strategies for using data generated in routine
clinical practice to inform the evidence base for treatments have been relatively
information from the clinic to basic research essentially ‘closing the loop’ and
evidence is needed to validate the utility of interventions and improve the quality of
1
I have chosen to use ‘practice’ as it is a more generic term than ‘medicine’ and includes disciplines outside of
medicine that are sometimes rather disparagingly referred to as para-medical.
9
clinical practice. It may contribute new knowledge and items for the clinical research
agenda in its own right. The remainder of this article (1) illustrates the possible use
of RCTs to generate benchmarks for clinical practice; thereby providing the basis for
practice based evidence of a treatment, and (2) outlines two substantial issues
testing [24,16,45]. There is evidence for the absolute efficacy of CBT procedures i.e.
CBT is superior to no treatment and some suggestion that CBT may be marginally
superior on some measures compared with other treatments i.e. relative efficacy.
This overall conclusion should be placed in the context of the nature of the trials.
Arguably we can have most confidence in the results of a meta-analysis when all the
etc) are homogeneous. This is not the case with respect to CBT for chronic pain
where there is marked heterogeneity across all parameters which then have to be
aggregated [23]. While this is not ideal it does reflect the reality of this complex field
and it is reasonable to conclude that there is evidence for a class effect of CBT
procedures for a range of conditions where chronic pain is the significant feature.
Most outcome measures in these studies are continuous rather than categorical or
binary and the computed effect size is the standardised mean difference between the
10
treatment and control arms (Cohen’s d or Hedge’s g) [53], estimates of absolute
and implementation of best practice. The methods are structured and rely on
metrics that enable one to compare the activity of routine clinical treatment with the
best available evidence and to develop and implement plans to improve current
performance levels where they fall short of best practice [2,33]. For example, data
CBT interventions in clinical practice primarily relies on the comparing pre and post
change, mean difference, pre-post effect size) and to conduct inferential statistical
whether the observed changes are attributable to the intervention or other factors
that have occurred over the same period i.e. natural history. Minami and colleagues
have [40,41] developed a methodology using the estimates of the magnitude of pre-
to post-treatment (or pre-treatment to follow-up) change from both the treated and
estimate the effect sizes and confidence intervals for change in the treated and
11
control arms. These statistics provide two benchmarks. The change in the untreated
group estimates change due the natural history of the disorder and the effects of
benchmark). The change in the treatment arm generates the benchmark for
treatment. The pre-post effect size of routine clinical treatment in a clinical service
setting can be computed and compared to the two benchmarks. The exact statistical
determine whether the service delivered in routine clinical treatment is within the
confidence intervals for the treatment benchmark or the natural history benchmark.
Fenton [18] conducted a proof of concept study for chronic pain. The omnibus
benchmarks for treated and natural history arms over all these trials are shown in
Figure 1. It is possible that routine clinical treatments in services may exceed these
benchmarks [67] when services pay detailed attention to the quality of treatment
implementation. The heterogeneity associated with the trial set used by Fenton
means that services should be cautious in using the omnibus estimates shown in
Figure 1. A better strategy would be to identify high quality trials where the format
and content of treatment closely matches that delivered by the service and the
FIGURE 1 HERE
12
There are three questions we might ask about the measures: (1) What should
we measure? (2) From whose perspective? and, (3) How much change is required?
stimulation, the main aim is to modulate the sensory-intensity qualities of pain. The
primary endpoint is pain and measures of other domains are ancillary. The work of
functioning, and specific measures within each domain, that cover many of the facets
measures [39]. Different instantiations of CBT may legitimately lay claim to specific
Trialists and services might reasonably select robust generic measures, such as those
defined by IMMPACT, and specific measures that are consistent with the exact
model of therapy being given; indeed this is readily observable in many trials [31].
Whose perspective?
Although the majority of measures used in RCTs and clinical settings are
patient reported outcomes who defines what should be measured is rarely debated.
Measures are selected by researchers and clinicians on the grounds of relevance for
13
there particular intervention and their psychometric characteristics. For the most
part they are inherently sound but the content of what is assessed is biased by
improvements in sleep, cognitive function and social role engagement. While the
latter is often measured the former two are rarely incorporated in outcome
The use of the effect size statistic to evaluate the outcome of CBT stems from
disadvantageous when patients and policy makers want yes/no answers to questions
e.g. ‘will this treatment relieve my pain?’ The advantage of the ES (d) statistic is its
capacity to provide a common metric across measures but its interpretation requires
attention to contextual features. Consider the two panels in Figure 2. Each panel
represents data from an RCT in which the initial scores are identical in each group.
At post treatment the ES = 0.5 (a medium effect size) in both trials and we might be
tempted to conclude that they are equivalent. Line C represents a cut-off score
14
denoting an interpretation which has clinical validity e.g. patients with scores below
the cut reduce their use of medication by a factor of 10. The interpretation of the of
the trial results is now entirely different and the treatment in trial B is markedly
preferable. The imposition of cut scores also makes the outcomes more
communicable to patients.
the effectiveness of treatment to patients. For an average effect size of d = 0.5 we can
say that after treatment the averaged treated person is at the 69th percentile of the
referenced to a tangible behavioural marker such as ‘four out of 5 people who take this
treatment are able to do their own supermarket shopping’. One challenge is devise
that have substantive clinical and social validity. Progress has been made in the
settings a change of around 30% or a two point decrement on a 0-10 numerical rating
outcome criteria also has the advantage over the effect size in that the outcome for
FIGURE 2 HERE
15
What change do chronic pain patients see as meeting their treatment
aspirations before treatment for other domains importance? [68] Robinson and
colleagues report two studies in which chronic pain patients were asked to make
ratings of their desired outcome on single items assessing pain, fatigue, distress and
interference [52]. In both studies the percentage decrement required exceeded 50%
and approached 70% in for the interference measure. In their second study patients
re-estimated their desired levels after 2 months of treatment and their estimates were
not known. Thorne and Morley extended Robinson and colleagues’ work and used
This enabled them to use the statistical properties of the scales to generate other
summary data and to define categorical outcome using Jacobson’s reliable change
index (RCI) and clinically significant change (CSC) criteria to develop categorical
outcomes from the continuous scales [27]. The procedure is illustrated graphically in
Figure 3 in which the pre treatment and post treatment scores are plotted.
FIGURE 3 HERE
line of no change is shown by the diagonal line. Knowledge of the scale’s reliability
and the variability of the sample are used to compute the magnitude of change that
can be regarded as reliable (RCI). A data point falling within the two dotted lines
16
parallel to the diagonal indicates that that person has not changed sufficiently for us
to discount the possibility that the change is measurement error. Points outside of
that indicate that the person has made a reliable change. We can now impose a
second criterion onto the scale by determining a cut point2. This is marked by the
vertical and horizontal lines (cross-hairs) on the plot defining patients who fall above
and below the cut point pre and post treatment. Figure 3 shows the various
categories that can be determined from this analysis. Thorne and Morley’s sample
reported desiring percent changes in the outcomes in the same range as Robinson’s
patients i.e. > 50%. The data were also analysed to produce effect size statistics and
these values ranged from 1.38 to 1.70, which are far in excess of values observed in
current RCTs. Finally, the proportion of patients whose scores met RCI and SCS
criteria were also computed. One consequence of this was that for some patients,
those who estimate that a small change would be satisfactory, the change is too small
criteria used i.e. the measurement technology is insensitive to the patient’s criteria.
These studies have used prospective judgements made by chronic pain patients and
reveal the range of desired outcomes and expectations. The magnitude of the
2
Jacobson described statistical criteria for determining cut points but they may also be determined by other
methods such as social (expert) consensus or the use of a ‘gold standard’ index and receiver operating curve
analysis.
17
Morley et al provide an example of the application of RCI/CSC methodology to
The magnitude of the effect size is often interpreted in qualitative terms using
the categories popularised by Cohen [11]: 0.2 = small, 0.5 = medium, 0.8 large. The
fact that Cohen stressed that these qualitative interpretations were dependent on the
context of the study is sometimes forgotten. The effect size may also be considered
explained in a bivariate relationship. In the context of an RCT the effect size between
treatment and control at the end of treatment represents the degree of variance in the
estimate for the absolute efficacy of treatments at around d = 0.8, which is equivalent
to about 14% of the variance [71]. Butler [9] reviewed 16 meta-analysis of CBT for a
range of disorders and reported a weighted grand mean ES(d) = 0.95, or about 18.5%
of the variance. A naïve assumption might be that all of this variance is attributable
to the specific effects of treatment but more nuanced analyses indicate that much the
patients and therapists, the competence of the therapist, and their allegiance to the
specific therapeutic model [71]. There is little research on therapist effects in the
18
cognitive behavioural treatment of chronic pain. The analysis of studies has largely
ignored the effect of groups and variation in therapists [1] but evidence is emerging
constructed ‘in the moment’ and subject to the influence of context such as the
their own ability to reflect on and appraise their behaviour while conducting
therapy. The provision of manuals detailing the structure and content of therapy is
cases and remain in supervision for the period of the trial. The quality scale
developed by Yates et al [74] captures some of these facets of treatment quality. The
scale was used to rate trials 66 trials of psychological therapy published between
1985 and 2009. The resulting analysis indicated that there was no discernable trend
across the 25 years that treatment quality as reported in the trials had significantly
Figure 4. Borelli [6] also developed a measure of treatment fidelity that identifies
receipt of treatment; and enactment of treatment skills. The latter two explicitly
19
assess the patient’s uptake of the treatment and their active use of treatment
period of 10 years and found that the mean proportion of adherence to treatment
fidelity strategies was around one half (0.55): Again there was no discernable
only the power of the specific technical aspects of interventions i.e. specific
behaviour experiments and behaviour change strategies, but to enhance the quality
FIGURE 4 HERE
routine clinical treatment settings and requires significant investment and thought.
fear-avoidance treatment protocol against a protocol for graded activity [30]. The
treatments were concept mapped to describe features that were: (i) essential and
unique; (ii) essential but not unique; (iii) unique but not essential; (iv) compatible but
neither essential nor unique; and (v), prohibited. Videotapes of treatment sessions
integrity can be derived and the degree to which contamination occurs is assessed by
20
the degree to which intrusions of therapeutic activity from categories (i) and (v) into
the target therapy. This very thorough procedure has disadvantages in that it is
costly and the full analysis can only be achieved after treatment is complete. This
makes it unsuitable for routine clinical use where rapid feedback and shaping of the
necessary for randomized clinical trials does not necessarily guarantee that
appropriate changes in the individual patient will occur. The dynamic nature of
response to changes in the patient. These changes are not necessarily effective and
therapists may drift from known effective interventions especially when confronted
with problems and difficulties [69]. Treatments that require active behavioural
change may easily revert to mere talking: therapists take the line of least resistance.
applicability and utility in clinical settings. For example, Lambert and colleagues
[29] have reported studies in which an outcome assessment tool (the OC-45) has
been administered after each session. The establishment of a large database in which
patient progress was tracked during the course of therapy enabled Lambert to devise
the equivalent of a manufacturing control chart and to detect patients who failed to
feedback and clinical support tools both increased the number of patients making
21
reliable and clinically significant changes and decreased the proportion showing
deterioration.
Given the plethora of outcome measures in the field of chronic pain it may not
valid markers of the change process that could be used to track progress. There are
relatively few change process studies [46] given the volume of trials but several
recent ones [35, 55,56,64] and cohort studies [8] have begun to explore this issue. The
telephone contact [48], and the development of suitable statistical analysis (multi-
level modeling) provide a suite of tools that may help advance our understanding of
self-efficacy and control beliefs; all of which can be measured with relative ease. On
the other hand clinicians and researchers might adopt the ‘old’ technology of single
methodology and more recently Vlaeyen and his colleagues [13,65] have used
that the target outcome measure may also be the process measure [54]. These
Summary
22
CBT is a class of treatments that have developed over 50 years in response to
social roles to aspects of identity construction. A legitimate case can be made for
treatment intervention at any of these levels. To some extent theorists and clinicians
generating large magnitude changes. One danger is that each new conceptualisation
and therapeutic school that shows promise becomes diluted when implemented in
routine clinical practice as elements may simply be added into existing treatment
attention should be given to: (1) developing measures with robust criteria to index
(2) methods for assuring the integrity and quality of treatment implementation.
23
Acknowledgements
I owe a big debt of gratitude to many colleagues and collaborators over the
Frank Keefe, Geert Crombez, Henry McQuay and Andrew Moore, and to my clinical
colleagues in the Leeds Pain Service. Several graduate students have collaborated in
research projects and I thank them for their diligence and generosity: Shona Yates,
Grania Fenton, Fiona Thorne and Sumerra Hussain, whose untimely death in the
24
References
Blackwell, 2010.
4 Beck AT, Rush AJ, Shaw BF, Greenberg RL. Cognitive therapy of depression.
2005;73(5):852-860.
25
7 Brown JL, Edwards PS, Atchison JW, Lafayette-Lucey A, Wittmer VT,
2003;71(1):81-91.
9 Butler AC, Chapman JE, Forman EM, Beck AT. The empirical status of
2006;26(1):17-31.
10 Charmaz K. From the "sick role" to stories of self: understanding the self in
physical health, Vol. 2. Oxford: Oxford University Press, 1999. pp. 209-239.
11 Cohen J. Statistical Power Analysis for the Behavioral Sciences. New York:
12 Dahl JC, Wilson KG, Luciano C, Hayes SC. Acceptance and commitment
26
neck pain: a replicated single-case experimental study of exposure in vivo. J
Pain 2008;9(12):1123-1134.
17 Farrar JT, Young JP, Jr., LaMoreaux L, Werth JL, Poole RM. Clinical
19 Fordyce WE. Behavioral methods for chronic pain and illness. St Louis:
Mosby, 1976.
20 Gatchel RJ, Peng YB, Peters ML, Fuchs PN, Turk DC. The biopsychosocial
Bull 2007;133(4):581-624.
27
21 Grisart J, Van der Linden M, Masquelier E. Controlled processes and
2002;24(8):994-1009.
2001.
2007;26(1):1-9.
25 Holroyd KA, Penzien DB, Hursey KG, Tobin DL, Rogers L, Holm JE, Marcille
PJ, Hall JR, Chila AG. Change mechanisms in EMG biofeedback training:
28
27 Jacobson NS, Roberts LJ, Berns SB, McGlinchey JB. Methods for defining and
447.
29 Lambert MJ, Hansen NB, Harmon SC. Outcome Questionnaire System (the
30 Leeuw M, Goossens ME, de Vet HC, Vlaeyen JW. The fidelity of treatment
31 Leeuw M, Goossens ME, van Breukelen GJ, de Jong JR, Heuts PH, Smeets RJ,
Koke AJ, Vlaeyen JW. Exposure in vivo versus operant graded activity in
chronic low back pain patients: results of a randomized controlled trial. Pain
2008;138(1):192-207.
29
32 Legrain V, Damme SV, Eccleston C, Davis KD, Seminowicz DA, Crombez G.
back pain and neck pain interventions in primary care settings. Euro J Pain in
press.
35 Litt MD, Shafer DM, Ibanez CR, Kreutzer DL, Tawfik-Yonkers Z. Momentary
30
38 McMillan D, Morley S. Single-case quantitative methods. In: M Barkham, GE
Pain 2005;113(1):1-2.
40 Minami T, Serlin RC, Wampold BE, Kircher JC, Brown G. Using clinical trials
41 Minami T, Wampold BE, Serlin RC, Kircher JC, Brown GS. Benchmarks for
2007;75(2):232-243.
31
46 Morley S, Keefe FJ. Editorial: Getting a handle on process and change in CBT
48 Naylor MR, Keefe FJ, Brigidi B, Naud S, Helzer JE. Therapeutic Interactive
Voice Response for chronic pain reduction and relapse prevention. Pain
2008;134(3):335-345.
49 Osborn M, Smith JA. The personal experience of chronic benign lower back
1998;3(Part 1):65-83.
52 Robinson ME, Brown JL, George SZ, Edwards PS, Atchison JW, Hirsh AT,
32
and expectations for treatment of chronic pain: the patient perspective. Pain
Med 2005;6(5):336-345.
54 Sidman M. The tactics of scientific research. New York: Basic Books, 1960.
55 Smeets RJEM, Vlaeyen JWS, Kester ADM, Knottnerus JA. Reduction of pain
377.
58 Teasdale JD, Segal Z, Williams JMG. How does cognitive therapy prevent
33
59 Thorn BE. Cognitive therapy for chronic pain. New York: Guilford Press,
2004.
2009;144(2/3):262-269.
61 Turk DC, Dworkin RH, Allen RR, Bellamy N, Brandenburg N, Carr DB,
Cleeland C, Dionne R, Farrar JT, Galer BS. Core outcome domains for chronic
62 Turk DC, Dworkin RH, Revicki D, Harding G, Burke LB, Cella D, Cleeland
CS, Cowan P, Farrar JT, Hertz S, Max MB, Rappaport BA. Identifying
2007;127(3):276-286.
34
experimental design in four patients with chronic low back pain. Behav Res
Ther 2001;39(2):151-166.
2008;76(3):397-407.
2009;47(2):119-127.
72 Williams ACdeC, Potts HW. Group membership and staff turnover affect
35
73 Williams ACdeC. Cognitive-behavioural pain management: lessons learned.
74 Yates S, Morley S, Eccleston C, Williams ACdeC. A scale for rating the quality
36
FIGURES
Figure 1
This figure shows omnibus effect size (d) benchmarks with 95%CIs for CBT (solid)
and waiting list control (open) averaged pre-post treatment changes in randomised
controlled trials for five domains. Data from [18].
37
Figure 2
The two panels in this figure represent idealised pre- to post-treatment changes on
an outcome in two trials, A and B. Each trial has two arms, a control (solid line) and
treatment (dashed line), and the mean at the start of treatment for is identical across
arms and trials. An improvement is indicated by a decrease in the measure. At the
end of treatment difference between the treated and control arms in both trials is
identical and has an effect size [ES(d) = 0.5]; the distribution of the scores shown by
the normal curve for each arm. If the ES(d) values are considered trial A and B are
identical in there effect. If however we have an established cut point denoting the
boundary for a clinically significant change it is clear that very few people in trial A
meet this criterion whereas most people in the treatment arm of trial B may be
regarded as clinically improved.
38
Figure 3
This figure is a tramline display in which the data from individual patients at pre-
treatment (x-axis) and post-treatment (y-axis) are plotted. For a patient who has the
same score on both occasions i.e. no change in score, the data point will fall on the
main diagonal axis (solid line). Any point that is not on the main diagonal indicates
some change. Other information plotted on the display enables us to judge the
significance of the change. The parallel lines either side of the main diagonal are the
95% confidence intervals for the error of measurement and any data point that falls
between these lines is not significantly different from the main diagonal i.e. no
significant change has occurred. Data points outside the confidence interval mean
that the person’s score has reliably changed. Note that this enables us to detect both
reliable improvement and deterioration.
The vertical and horizontal dotted lines plotted for values of x and y = 20
indicate a clinically meaningful cut score on the scale. Patients to the right of the
pre-treatment (vertical) cut score were above the clinical criterion at pre-treatment.
Patients below post-treatment cut score (horizontal) were below the clinical criterion
at post-treatment. As a result we can now classify individual patients on the
additional criterion of whether they have made a significant clinical improvement.
Note that patients below the cut score at the start of treatment may make a clinically
significant deterioration.
39
Figure 4
Plot of ratings for two components of trial quality, study design (circles) and
implementation of treatment (triangles) for 66 trials of psychological treatment for
chronic pain published between 1984 and 2010 based on [16]. Regression analyses
indicate that whereas the design features of trials have improved over time the
reported implementation of treatment has not.
40
Table 1
Timeline outlining the development of CBT applied to the treatment of chronic pain.
The main schools are shown in bold face and their antecedents in italic.
Operant
Behaviour
Operant
Analysis
Operant
Behaviour
Biofeedback
Analysis
1. Cognitive theory
of stress
2. Behavioural Stress
Analysis of self management
control
Buddhism Mindfulness
(1000 BCE) based stress
reduction
Mower-Miller Fear
2- process Avoidance
theory
41