Can Data Modeling Help Identify Chronic Pain Patients Who Will Benefit from ACT?

Chronic pain affects not only physical, but also psychological and social aspects of daily life. At a time when millions of Americans are affected, and its prevalence continues to rise worldwide, effective treatments are sorely lacking. The way people cope with chronic pain can greatly influence its impact on mental health, social interactions, and physical function. Within the health professions, and particularly in Physical Therapy, there is a great deal of interest in psychological interventions, such as ACT, aimed at changing the relationship that chronic pain sufferers have with pain. Evidence suggests that mindful acceptance of the circumstances surrounding the pain experience and consequently adopting value-guided behaviors can significantly improve quality of life even if pain intensity remains unchanged (1). 

This is the main goal of Acceptance and Commitment Therapy (ACT), a contextual form of the more traditional Cognitive Behavioural Therapy (CBT). Instead of trying to restructure and revert maladaptive (pain-feeding) thoughts and behaviors, as CBT does, ACT deemphasizes attempts to control or reduce pain and associated thoughts and feelings, focusing instead on altering the function of these experiences, and the influences they exert on behaviour (2).

Who Might Benefit from Acceptance and Commitment Therapy?

However, because not all patients respond well to these interventions, the issue of whom might benefit from ACT or other psychological approaches remains an open and urgent question. In daily Physical therapy practice we ponder this question before each new patient. Is age a critical factor? What about pain duration? Treatments that succeeded or failed so far? Are there psychological (anxiety, depression) social (relationships), or employment issues compounding the burden of chronic pain? 

To address this problem, Psychology and Physical Therapy research groups are trying to pinpoint specific, evidence-based predictors and moderators of treatment outcome. “Predictors and moderators are both characteristics of the individual (e.g., age, symptom severity, and pain beliefs) that are present prior to the start of treatment and that may influence the individual’s response, regardless of whatever treatment is under study (predictor), or differentially to one treatment versus another (moderator)” (3).

Data Modeling to Predict ACT Success in People With Chronic Pain 

A 2018 study published in the European Journal of Pain by researchers in the UK compared baseline and post-treatment scores on key ACT processes (acceptance, decentering) and pain-related variables, and contrasts them with demographic data to evaluate which factors can predict treatment success (4).

Participants were eligible if their ongoing pain had lasted for at least six months and was having a significant impact in their quality of life or everyday functioning. Patients were considered unlikely to benefit from the treatment approach, and excluded from the study if they had a poorly controlled medical or psychiatric condition, or a severe physical disability or cognitive impairment that might interfere with their ability to engage in a group-based treatment.

The study involved 609 people (195 men, 414 women). Their mean age was 46 years and pain duration was on average 12.7 years.

The ACT-based pain management program was delivered by an interdisciplinary team comprised of psychologists, occupational therapists, physiotherapists, nurses, and physicians. The majority of the sessions were delivered by psychologists, physical therapists, and occupational therapists trained in ACT, who routinely worked within this model. Groups of up to 12 participants received four 8-hour sessions per week over four weeks. Demographic and pain-related information, including age, gender, ethnicity, pain location, pain duration, medications, and health care use was collected on the first day of the program. Several self-reports were also completed on the first and last day of the intervention:

  • Health status was evaluated through the Short-Form Health Survey (SF-36), which includes eight subscales that represent primary aspects of health and functioning.  
  • Depression was measured using the PHQ-9 tool, which assesses the severity of depressive symptoms and response to treatment. This questionnaire was included because a previous review of studies on contextual CBT for chronic pain suggested that depression was as a potentially important outcome predictor (5)
  • Pain intensity was assessed using a standard 10-point scale. 

Participants also completed five questionnaires to assess baseline Psychological Flexibility

  • The Chronic Pain Acceptance Questionnaire (CPAQ), a 20-item measure of acceptance of chronic pain that contains 2 subscales: activity engagement, which gives a measure of the ability to perform desired activities despite the presence of pain, and pain willingness, which measures pain avoidance by addressing attempts to control or reduce pain. 
  • The Cognitive Fusion Questionnaire (CFQ), a 13-item measure of cognitive fusion and defusion, namely, the degree of entanglement or detachment from negative thoughts that influence behavior.
  • The Acceptance and Action Questionnaire (AAQ-II), a 7-item self-report measure of experiential acceptance or avoidance.
  • The Committed Action Questionnaire (CAQ) a self-report measure of committed action developed in people seeking treatment for chronic pain. The original, 18-item version was used.
  • The Experiences Questionnaire (EQ) assesses decentering (“the ability to observe one’s thoughts and feelings as temporary, objective events in the mind, as opposed to reflections of the self that are necessarily true”) and rumination (repeatedly going over a negative thought or a problem) (6).

What Does All That Data Really Tell Us?

While categorical variables (e.g. gender, employment status) can be readily correlated with outcomes of interest, the presence of continuous variables/risk factors (e.g. age, pain duration, etc.) in clinical studies poses a significant challenge when trying to model their influence on treatment outcome. The issue becomes more complex when data is limited (small sample size), and/or multiple variables are considered (multivariable analysis). 

Traditional regression analyses assume most commonly a linear trend, but just like a drug’s effect may be negligible at very low doses, optimal at intermediate doses, and lethal at high doses, or may have short-or long-lasting effects, linear models are seldom reflective of the actual dynamics of the relationship between baseline continuous variables and the observed outcomes. Nonparametric regression models are preferable when the relationship (i.e. functional form) between treatment response and explanatory variables (predictors) is unknown and might be nonlinear. However, selecting among several possible regression modeling strategies is not straightforward, and may critically affect the interpretation of results. 

With these caveats in mind, the study aimed to investigate whether “a theory guided MFP approach to modelling continuous outcomes could predict outcomes in an ACT-based pain management program” (4).

Therefore, multivariable modelling was performed using the multivariable fractional polynomial (MFP) approach, a non-linear regression methodology that allows to determine the impact on the model of candidate variables (predictors). These variables reflected key processes of the psychological flexibility model, i.e. baseline scores on the CFQ, CAQ, AAQ-II, CPAQ, and the decentering subscale of the EQ (all continuous variables), and demographic or pain-related variables such as gender (a binary variable), age, depression (measured by the PHQ-9), and pain duration (all continuous variables), and employment status (a categorical variable).

To identify whether modelling continuous variables using the MFP approach improved the model fit (i.e. how tightly the model fits the actual data distribution) compared to a model that assumed all relationships were linear, model fit assessments were conducted on the 4 regression models generated by the MFP method using Bayesian Information Criterion (BIC) scores. Statistical tests were further conducted to assess model stability and data variability.

Which Are the Best Predictors of ACT Success?

Through this approach, post-treatment outcome scores were predicted across four response variables: 

  1. Physical functioning (SF-36): being unemployed, employment category “other” (retired, carer, housewife/ homemaker, volunteer), decentering, and PHQ-9 significantly predicted post-treatment outcome scores.  Namely, being employed and having lower a baseline decentering score may be associated with better outcomes. 
  1. Social functioning (SF-36): being unemployed and baseline PHQ-9 were the two stronger predictors of social functioning changes post-treatment. Namely, being employed and having lower depression at baseline may be associated with better outcomes.
  1. Emotional functioning (mental health subscale of the SF-36): being unemployed, baseline AAQ-II score, and baseline PHQ-9 score significantly predicted post-treatment score. Namely, being employed, having higher baseline acceptance and lower baseline depression may be associated with better outcomes. 
  1. Pain intensity (rated on a standard scale from 0-10): being unemployed, decentering, and baseline PHQ-9 score significantly predicted post-treatment scores for pain intensity. Namely, being employed, having a lower baseline decentering score, and lower baseline depression may be associated with better outcomes.

The authors note that significant improvement was observed in all domains, and that participants who did not complete treatment (n = 43) scored significantly lower on baseline social functioning and mental health than those who completed treatment. Interestingly, regression analyses indicated that assuming linear functions may be sufficient for all models, except for the one describing pain intensity outcomes, which showed more complex predictor-outcome relationships. 

FREE Download: The 5 Pillars of Pain Care Dr. Tatta’s simple and effective pain assessment tools. Quickly and easily assess pain so you can develop actionable solutions in less time.
Interestingly, regression analyses indicated that assuming linear functions may be sufficient for all models, except for the one describing pain intensity outcomes, which showed more complex predictor-outcome relationships.  Share on X

Social Environment and Mental Health: Good Predictors of ACT Outcomes?

Although the authors warn that unique effects were small, the study results suggested that socio-economic (employment status) and mental health (depression) indicators were the most ubiquitous factors predicting the success of ACT in patients with chronic pain. The relationship between these predictors and treatment efficacy was best described by simple, linear correlations, reinforcing their validity and suggesting possible generalization to other patient populations.  

Determining whether a new client may benefit or not from psychologically-informed care can be a difficult task in Physical Therapy practice. However, the fact that psychosocial factors are indelibly tied to the individual pain experience reminds us that concise knowledge and effective implementation of behavioral-cognitive strategies to manage chronic pain is paramount to improve treatment responses. As the study authors point out:  “If a person’s scores on particular facets of psychological flexibility at the start of treatment are indeed associated with outcome, then targeting particular aspects of psychological flexibility early in treatment may be key to maximising change”.

New research on treatment predictors, moderators, and specific processes that trigger change (mediators) is eagerly awaited to help us tailor patient-centered therapies to reduce the impact of chronic pain on physical function, mental stress, and social interactions. 

The fact that psychosocial factors are indelibly tied to the individual pain experience reminds us that concise knowledge and effective implementation of behavioral-cognitive strategies to manage chronic pain is paramount to improve… Share on X

Ready to take your practice to the next level? Check out the Institute’s new course: ACT in Motion (A Pain Exposure Protocol).

REFERENCES

1- Feliu-Soler, A., Montesinos, F., Gutiérrez-Martínez, O., Scott, W., McCracken, L. M., & Luciano, J. V. (2018). Current status of acceptance and commitment therapy for chronic pain: a narrative review. Journal of pain research, 11, 2145–2159. doi:10.2147/JPR.S144631

2- Hayes, S. C., Luoma, J. B., Bond, F. W., Masuda, A., & Lillis, J. (2006). Acceptance and commitment therapy: Model, processes and outcomes. Behaviour research and therapy, 44(1), 1-25.

3- Åkerblom, S. (2018). Predictors and mediators of outcome in CBT for chronic pain: The roles of psychological flexibility and PTSD.

4- Gilpin, H. R., Stahl, D. R., & McCracken, L. M. (2019). A theoretically guided approach to identifying predictors of treatment outcome in Contextual Cognitive Behavioural Therapy for chronic pain. European Journal of Pain, 23(2), 354-366.

5- Gilpin, H. R., Keyes, A., Stahl, D. R., Greig, R., & McCracken, L. M. (2017). Predictors of treatment outcome in contextual cognitive and behavioral therapies for chronic pain: a systematic review. The Journal of Pain, 18(10), 1153-1164.

6- Fresco, D. M., Moore, M. T., van Dulmen, M. H., Segal, Z. V., Ma, S. H., Teasdale, J. D., & Williams, J. M. G. (2007). Initial psychometric properties of the experiences questionnaire: validation of a self-report measure of decentering. Behavior therapy, 38(3), 234-246.

You Might Also Be Interested In

Breaking the Silence: Diagnostic Overshadowing in Physical Therapy and Mental Health

Breaking the Silence: Diagnostic Overshadowing in Physical Therapy and Mental Health Diagnostic overshadowing was initially used to describe a clinician’s tendency to assess individuals with ...
Read More

7 Physical Therapy Interventions That Improve Mental Health

7 Physical Therapy Interventions That Improve Mental Health By Joe Tatta, PT, DPT The relationship between physical therapy and mental health has gained significant attention ...
Read More

What Training Is Required for Physical Therapists To Offer Mental Health Support?

What Training Is Required for Physical Therapists To Offer Mental Health Support? By Joe Tatta, PT, DPT A common question physical therapists may ask is: ...
Read More

Privacy Policy

Effective Date: May, 2018

Your privacy is very important to us. We want to make your experience on the Internet as enjoyable and rewarding as possible, and we want you to use the Internet’s vast array of information, tools, and opportunities with complete confidence.

The following Privacy Policy governs the online information collection practices of Joe Tatta, LLC d/b/a joetatta.co and www.backpainbreakthrough.com ( collectively the “Sites”). Specifically, it outlines the types of information that we gather about you while you are using theSites, and the ways in which we use this information. This Privacy Policy, including our children’s privacy statement, does not apply to any information you may provide to us or that we may collect offline and/or through other means (for example, at a live event, via telephone, or through the mail).

Sign Up for the Integrative Pain Science Institute’s Weekly Newsletter

Enter your email and get the latest in pain science, podcast episodes,
CEU opportunities, and special offers.

You have Successfully Subscribed!

We only send you awesome stuff!