Cognitive-Motivational Influences on Health Behavior Change in Adults with Chronic Pain

Transcription

1 Pain Medicine 2016; 17: doi: /pme Cognitive-Motivational Influences on Health Behavior Change in Adults with Chronic Pain Ryan J. Anderson, PhD,* Robert W. Hurley, MD, PhD, Roland Staud, MD, and Michael E. Robinson, PhD* *Department of Clinical and Health Psychology, University of Florida, Gainesville, Florida; Department of Anesthesiology, Pain Medicine, Medical College of Wisconsin, Wauwatosa, Wisconsin; Department of Medicine, University of Florida, Gainesville, Florida, USA Correspondence to: Michael E. Robinson, PhD, Department of Clinical and Health, Psychology, Center for Pain Research and Behavioral Health, 101 South Newell Dr, Room 3141, University of Florida, Gainesville, FL 32611, USA. Tel: ; Fax: ; Disclosure and conflict of interest: There are no sources of support to declare for this study. Authors have no conflicts of interest to declare. Abstract Objective. The primary aim was to assess the psychological factors that influence engagement in health behaviors in individuals with chronic pain using a new measure, the Behavioral Engagement Test for Chronic Pain (BET-CP). A secondary aim was to determine preliminary psychometric properties of the BET-CP. Subjects. Participants were 86 adults with chronic musculoskeletal pain recruited from University of Florida pain clinics and the community. Methods. Participants completed a battery of selfreport instruments online, including the BET-CP and measures of related constructs. Items on the BET-CP assessed motivation, self-efficacy, outcome expectations, and the symptom benefit required to engage across four health behaviors: exercise, diet, sleep, and pain self-management (e.g., relaxation and activity pacing). Results. Participants reported modest expectations of pain-related symptom improvement if they practiced the health behaviors (22 26% improvement), but they required twice that (47 54% improvement) to make it worth their while to commit to practicing them. Participants expected to get the most symptom relief from relaxation and activity pacing, but they were most confident and motivated to eat a healthy diet. In a subsample of participants who provided data for psychometric analysis, the BET-CP demonstrated strong test-retest reliability across 7 days and adequate convergent validity. Conclusion. While patients with musculoskeletal pain have outcome expectancies that are nearly in line with research on behavioral pain treatments, their stringent requirements for symptom benefit may impede engagement in the health behaviors recommended for their pain-related symptoms. Additional psychometric study with larger sample sizes is needed to further validate the BET-CP. Key Words. Psychology; Behavior; Expectation; Self-efficacy; Motivation; Exercise Introduction In recent decades, unhealthy behaviors (e.g., poor diet, inactivity, smoking, risky sexual behaviors) have been implicated in many of the major causes of mortality and chronic disease [1]. Promotion of healthy behaviors has therefore played an important role in the prevention and management of chronic health conditions. In chronic pain, behavioral treatments have served as effective adjuncts to biomedical interventions [2]. With the goal of improved functioning and increased engagement in life, traditional components of cognitive-behavioral therapy for pain include relaxation skills training, activity pacing, behavioral activation, and cognitive restructuring [3]. More recently, other behaviorally mediated health problems have been targeted to improve management of chronic pain. Specifically, obesity, physical inactivity, and sleep problems have been shown to increase risk for chronic musculoskeletal pain [4,5], and to worsen pain-related symptom severity and disability once the condition has developed [6 10]. Behavioral sleep interventions [11,12] and lifestyle weight loss treatments (e.g., diet and exercise) [13 15] have shown promise in affecting pain-related outcomes in musculoskeletal pain conditions. VC 2016 American Academy of Pain Medicine. All rights reserved. For permissions, please journals.permissions@oup.com 1079

2 Anderson et al. However, sustained engagement can be challenging for patients as behavioral interventions may conflict with long-standing pain-coping behaviors. Not surprisingly then, dropout or noncompliance to behavioral medicine treatments are common, and attrition can be as high as 60% [16,17]. While some efforts have been made to individualize prescription of exercise in chronic pain patients [14,18], other providers may not take the time to understand a patient s motivation, abilities, or expectations about behavioral treatments, which can potentially also contribute to treatment nonadherence. Through the creation of a clinically relevant self-report measure, the present study seeks to understand the chronic pain patients expectations and beliefs with regard to engaging in health behaviors. This information can be used to understand individual differences in health behavior change, and ultimately may be used as targets for intervention to improve motivation for and adherence to behavioral interventions for pain. The study was informed by two related literatures: patient-centered outcomes and health behavior change in chronic pain. Patient-centered health care strives to take into account the needs, preferences, and values of the individual patient. It does this by improving doctor patient communication, by focusing on biopsychosocial outcomes that are important to the patient, and by understanding a patients expectations and beliefs about their treatment [19]. The subjective nature of chronic pain makes a patient-centered approach to its assessment and treatment essential. Often complete symptom remission is not possible in chronic pain conditions, and patients and providers can differ on the degree of symptom improvement that they consider successful treatment. Thus, increasing attention has been paid to the development of alternate and individualized criteria by which to judge the effectiveness of a treatment. To this end, Robinson et al. created the Patient Centered Outcomes Questionnaire (PCOQ) [20] to assess patients definition of treatment success and expectations for improvement. The instrument has been used in a variety of chronic pain populations, and results demonstrated that the chronic pain patient s definition of treatment success tends to be stringent, and their high expectations for treatment-related improvement are greater than what pain treatments can provide [20,21]. This mismatch between patient expectations and treatment effectiveness can contribute to frustration on the part of patient and provider, and may increase the chance for treatment noncompliance or dropout. Behavioral pain researchers have also made efforts to apply theories of health behavior change to chronic pain populations, including the Transtheoretical Model (TTM) [22,23] and Social Cognitive Theory (SCT) [24]. Consistent with the mission of patient centered care, this may lead to a better understanding of individual differences in engagement in behavioral pain coping. Using factors specified in these theories, researchers have found good evidence for the importance of selfefficacy, motivation, and outcome expectancies in explaining pain coping behavior [25 27]. In a clinical trial context, motivation has predicted completion of behavioral treatment for chronic pain [28,29], and all three factors have been shown to mediate the outcomes of behavioral treatment [30 33]. Included in this line of research was the creation of two self-report measures (Pain Stages of Change Questionnaire and the Multidimensional Pain Readiness to Change Questionnaire) that assess patients readiness to adopt pain coping behaviors [27,34]. Use of these questionnaires helped to establish the importance of patient motivation to behavioral treatment outcomes and attrition. In 2003, Jensen et al. organized the components of the behavior change theories into one hybrid model: the Motivational Model of Pain Self-Management [35] (Figure 1). The model posits that motivation is a malleable psychological process that initiates and maintains pain coping behavior, and is influenced by the SCT factors of perceived importance (beliefs about the importance of changing behavior) and self-efficacy (patients confidence in their ability to actually perform the behavior). Results of two cross-sectional studies were supportive of the model [36,37]. The current study extends patient-centered outcomes research, and draws upon the behavior change literature to further understand the psychological factors that influence health behavior engagement in chronic pain. To help achieve this aim, the study developed and tested a new measure of health behavior change [the Behavioral Engagement Test for Chronic Pain (BET- CP)]. The BET-CP represents an operationalization of the Motivational Model for Pain Self-Management as it uses the model as theoretical grounding for the assessed constructs. The BET-CP differs from existing measures in several ways. Existing measures of behavior change in chronic pain assess only motivation in the context of traditional pain coping behaviors [27,34]. The BET-CP assesses motivation and the cognitive determinants of motivation to practice not only the behaviors taught in specialized psychological therapy (i.e., relaxation and activity pacing), but also the health behaviors commonly prescribed by primary care and pain management physicians to address obesity, inactivity, and sleep disturbance in individuals with chronic pain. Additionally, while existing expectancy scales measure expectations about treatment broadly defined [38,39], the BET-CP is specific to behavioral treatment in a chronic pain population, and uses pain-related symptom scales to assess patient beliefs about practicing health behaviors. Also of note, the study introduces a concept novel to the pain behavior change literature the benefit requirement to engage a domain most closely related to a personal cost-benefit analysis. Thus, the primary aim of the present study is to assess the psychological factors that influence health behavior engagement in individuals with chronic pain (e.g., motivation, self-efficacy beliefs, outcome expectations, and 1080

3 Influences on Health Behavior Change Figure 1 Motivational Model of Pain Self- Management [24]. beliefs about the benefit required to engage) across four health behavior domains [physical activity, diet, sleep behaviors, and pain self-management (i.e., relaxation and activity pacing)] using a newly developed measure, the BET-CP. A secondary aim is to perform preliminary psychometric analyses on the BET-CP. Methods Participants Participants were adults with chronic musculoskeletal pain recruited from the outpatient Rheumatology and Pain & Spine Clinics affiliated with the University of Florida in Gainesville, Florida. Patients presenting to the two clinics were screened in-person for eligibility and interest in the study. Participants were also identified by posting study flyers around hospitals and medical clinics in the Gainesville area. Individuals responding to the flyers were screened for eligibility over the phone and over . Inclusion criteria for the study consisted of: 1) adults aged 18 years or older, 2) musculoskeletal pain lasting at least 3 months in duration, 3) the ability to read and comprehend English, and 4) access to a computer with a connection to the internet. Exclusion criteria for the study included participants who were unable to provide informed consent, and those with a current or recent cancer diagnosis. The study was reviewed and approved by the Institutional Review Board of the University of Florida. Prior to enrolling in the study, all participants provided informed consent in accordance with the Institutional Review Board. All participants received financial compensation following their completion of the study, and this may have provided incentive for them to participate. Measures Demographic and Clinical Characteristics Participants provided the following demographic information: age, gender, race, years of education, marital status, and employment status. Additionally, they were asked to provide information regarding their pain condition, including: diagnosis, duration of their pain symptoms, and location of their pain. To indicate the location of their pain, participants examined a body diagram divided into 20 sections, and answered yes or no as to whether or not they are experiencing pain in each of those areas. Behavioral Engagement Test for Chronic Pain Item construction of the Behavioral Engagement Test for Chronic Pain (BET-CP) was informed by the PCOQ [20] and preliminary tests of the motivational model of pain self-management [36,37]. BET-CP content is divided into two sections: a current functioning section that assesses current pain-related symptom levels and recent engagement in four health behavior domains, and a beliefs section that examines patient beliefs and motivation regarding the practice of the four health behaviors. The measure is constructed in this manner to obtain a snapshot of what and how a patient is doing now, and then what it might take for them to change their health behaviors. This information may potentially be used by the pain management clinician when prescribing behavioral treatments, and by the patient to help him or her clarify their beliefs and motivation to practice health behaviors (Appendix 1). In section 1, participants rate their usual levels of pain, fatigue, emotional distress, interference in daily functioning, and sleep disturbance on 101-point numerical rating scales (0 ¼ none, 100 ¼ worst imaginable). Numerical rating scales for pain have demonstrated good convergent validity, are sensitive to treatment effects, and are easy to administer [40]. Robinson and colleagues have also used a one-item numerical rating scale to measure usual levels of fatigue, emotional distress, and interference [20,21,41,42], and Brown et al. demonstrated good concurrent validity of the one-item ratings of usual symptom levels with standardized 1081

4 Anderson et al. measures of the symptoms [41]. Additionally, retrospective ratings of usual pain have shown strong correlations (r ¼ 0.78 or higher) with hourly ratings of pain over the course of two weeks [43,44]. To ascertain the extent to which patients are already practicing health behaviors, section 1 of the BET-CP also asks participants to indicate their recent level of engagement in physical activity, healthy diet, good sleep habits, relaxation, and activity pacing on a 0 (never) to 100 (daily) numerical rating scale. Section 2 of the BET-CP begins by informing patients that engaging in certain health behaviors can improve their pain-related symptoms. It then lists the behaviors (physical activity, sleep, diet, and pain self-management via relaxation and activity pacing), and further defines them by specifying practice recommendations. The practice recommendations provide patients with frequency recommendations and concrete goals to give them a basis for gauging their ratings of self-efficacy, motivation, outcome expectancy, and benefit requirement to engage. For example, the physical activity recommendation of 30 min of moderate intensity physical activity (e.g., brisk walking) on 5 out of 7 days a week is taken from the U.S. Department of Health & Human Services 2008 Physical Activity Guidelines for Americans [45]. Similarly, dietary recommendations were culled from the U.S. Departments of Agriculture and Health and Human Services 2010 dietary guidelines for Americans [46]. The healthy sleep recommendations are the common components of CBT for insomnia [47], and the pain self-management recommendations of relaxation and activity pacing are two of the most common strategies taught in behavioral therapy for pain [3]. Section 2 of the BET-CP then asks participants to rate their beliefs regarding the importance of behavior change (i.e., their outcome expectancy and benefit requirement), their self-efficacy, and their readiness to engage for each of the four health behaviors separately, and for all behaviors as a package. Outcome expectancy and benefit requirement beliefs are assessed for the pain intensity domain for individual health behaviors, and they are assessed across four symptom domains (pain, fatigue, emotional distress, and interference) for all health behaviors combined. Using the same 0 (none) to 100 (worst imaginable) scale used to measure their usual symptom levels, the outcome expectancy question asks participants to rate where they would expect their symptom levels to be if they were to engage in the behavior: If I were to practice (insert behavior) regularly, I would expect that my usual level of (insert symptom) would be. Participants with optimistic expectations might provide symptom ratings that are lower than their usual ratings, while those who are more pessimistic might expect that their symptoms would not change or would even worsen if they engaged in the behavior. The benefit requirement construct is novel and potentially taps into an internal cost-benefit dialogue that the patient may be having in deciding whether to engage in these behaviors. The construct is related to the decisional balance literature in health behavior change, which suggests that individuals weigh the anticipated pros versus anticipated cons in deciding whether or not to engage in a behavior. The benefit requirement question seeks to determine how much symptom benefit is worth it to offset the costs of time and effort that it will take to engage in the health behavior. Using the same 0 (none) to 100 (worst imaginable) scale, the benefit requirement question asks participants to complete the following sentence: I WOULD practice (health behavior) if doing so would improve my usual level of (symptom) to. The questions on outcome expectancy and benefit requirement were positioned directly next to each other (and next to a listing of their usual symptom levels) to make it easier for the participants to gauge current level symptoms vs expected level of symptoms vs symptom level required to make it worth it to practice the behaviors. The question on self-efficacy assesses the participants perceived ability to engage in each behavior by asking, On a scale of 0 (not at all capable) to 100 (fully capable), how capable are you of regularly practicing (insert health behavior). Similarly, to assess participants motivation to engage in each health behavior, they are asked, On a scale of 0 (not at all ready) to 100 (fully ready), how ready are you to regularly practice (insert health behavior). Patient Centered Outcomes Questionnaire Participants completed the patient centered outcomes questionnaire (PCOQ) [20] to assess their definition of treatment success, their expectation for treatment outcome, and importance of seeing improvement across five symptom domains: pain, fatigue, emotional distress, interference in functioning, and sleep disturbance. Items on the PCOQ have demonstrated good construct validity, good test retest reliability, and are sensitive to treatment-related change [20,41]. We expected that items on the BET-CP and PCOQ assess similar constructs involving treatment expectation and success from the patient perspective. So for purposes of this study, the PCOQ served as a measure to establish the convergent validity of the new BET-CP. Pain Self-Efficacy Questionnaire The pain self-efficacy questionnaire PSEQ [48] consists of 10 items assessing the respondent s confidence in their ability to participate in various activities despite their chronic pain. Participants indicated their confidence on a 7-point Likert scale ranging from 0 ( not at all ) to 6 ( completely confident ). A total score (maximum ¼ 60) is obtained by summing the scores from each of the 10 items. Higher scores indicate stronger self-efficacy beliefs. While the types of activities 1082

5 assessed on this questionnaire are more general in nature (e.g., I can enjoy things despite the pain and I can do most of the household chores despite the pain ) than the behaviors assessed on the BET-CP, it represents an established measure of self-efficacy beliefs, and can therefore help to establish the convergent validity of the self-efficacy items on the BET-CP. Multidimensional Pain Readiness to Change Questionnaire Guided by the motivation-related stages of behavior change as outlined in the Transtheoretical Model of Behavior Change, the Multidimensional Pain Readiness to Change Questionnaire (MPRCQ2) [34] assesses the chronic pain patient s readiness to adopt common pain coping strategies taught in cognitive behavioral therapy. It consists of 69 statements about pain coping (e.g., break up tasks into smaller pieces to get more done ) rated on a 7-point Likert scale corresponding to the stages of behavior change [1 ¼ I am not doing this now, am not interested in ever doing it to 7 ¼ I have been doing this for a long time (at least 6 months) ]. Nine subscale scores are obtained, and in the current study, only items from the relevant subscales of exercise (7 items), relaxation (7 items), and pacing (6 items) were completed by participants. These subscales were used to help establish the convergent validity of the BET-CP items assessing participants readiness to engage in the health behaviors. Procedures The study was completed entirely online. Eligible and interested participants were provided with a unique username and password, and instructed to navigate their internet browsers to a secure web address where they inputed their username and password to access the online study. Here, they provided informed consent then completed the battery of self-report measures described above. To address study aim two, a subset of 25 participants returned to the website one week later to complete the BET-CP again to provide test retest reliability data. Statistical Analyses Descriptive means and standard deviations were calculated for continuous demographic and clinical variables, as well as for participants responses to each item on the BET-CP. Frequency statistics were calculated for the categorical variables of gender, race, marital status, employment status, and self-reported diagnosis. A repeated measures analysis of variance (RANOVA) (with Sidak correction for multiple comparisons) was used to test whether the mean ratings of practice frequency (as assessed by 101-point numerical scales) differed across the 4 health behaviors. RANOVAs (with Sidak correction) were also used to test whether the degree of expected or required symptom change differed by type of health behavior or by symptom domain. The RANOVA test was chosen because, in each of these circumstances, we wanted test whether there were differences in mean scores across 3 or more conditions where the sample means are not independent (i.e., the same participants provided ratings across the conditions). A series of paired samples t-tests were used to test whether participants usual symptom levels were different than their expected and benefit required symptom levels. To establish the reliability of the BET-CP over time (i.e., test retest reliability), the Pearson product-moment correlation coefficient was used to correlate participant responses on each item of the BET-CP at time 1 with their responses a week later at time 2. To establish the convergent validity of the BET-CP, participant responses to items on the BET-CP were correlated (using Pearson product-moment correlation coefficients) with their responses on existing measures of the associated phenomenon. Specifically, outcome expectancy and benefit requirement responses on the BET-CP were correlated with the treatment expectation and treatment success questions on the PCOQ. Levels of readiness to engage in exercise and relaxation/activity pacing as assessed by the BET-CP were correlated with the exercise, relaxation, and pacing subscale scores of the MPRCQ2. Finally, ratings of self-efficacy to engage in each of the four health behaviors were correlated with their Pain Self-Efficacy Questionnaire total score. Results Influences on Health Behavior Change Eighty-six adults with chronic musculoskeletal pain provided data for the analyses. Fifty-one participants (59.3%) responded to print advertisements posted in the Gainesville community, 26 (30.2%) were recruited from the University of Florida Pain and Spine Clinic, and 9 (10.5%) were recruited from the University of Florida Rheumatology Clinic. Of the 86 participants, 25 completed the BET-CP a second time to provide BET-CP test retest data. Demographic and Clinical Characteristics Demographic characteristics of the sample are displayed in Table 1. The sample was predominantly middle aged (average age ¼ 49.0 years), female (72.1%), and Caucasian (77.9%). Approximately one-half the sample was married (52.3%) and employed (41.9%), and the sample was, on average, college educated (mean level of education ¼ 15.0 years). Clinically, participants indicated that they had been experiencing pain for an average of 9.5 years (SD ¼ 10.0, range ¼ ). On average, they endorsed 6.7 (SD ¼ 5.2, range ¼ 1 20) areas of pain on the body diagram, suggesting that their pain was often not contained to one area of their body. Of those who reported a diagnosis for their pain, the most common conditions included Fibromyalgia syndrome (N ¼ 18), herniated disc (N ¼ 9), unspecified arthritis (N ¼ 7), osteoarthritis (N ¼ 7), 1083

6 Anderson et al. Table 1 Demographic characteristics of the sample (N ¼ 86) N % Years Mean Age in Years 49.0 (SD ¼ 16.9) Female Gender Race/ethnicity Caucasian Hispanic African American Asian Indian American Indian or Alaskan Native Other Marital Status Married Divorced Separated Living with a partner Single Mean Years of Education 15.0 (SD ¼ 2.4) Employment Employed Unemployed Retired Homemaker Disabled Student degeneration of the cervical, thoracic, or lumbar disc (N ¼ 6), and unspecified back pain (N ¼ 4). Study Aim 1: Characterizing Participant Responses to the BET-CP Frequency of Health Behavior Engagement Participant responses to the BET-CP are summarized in Tables 2 and 3. Table 2 displays the self-reported frequency of engagement in the four healthy behaviors over the past 12 months. The RANOVA found a significant main effect for health behavior (F 3,83 ¼ 5.7, P ¼ 0.001), and individual comparisons revealed that participants reported frequency of healthy eating was significantly greater than their engagement in physical activity (P ¼ 0.004). Usual Pain, Expectation, and Benefit Requirement to Practice All Behaviors As displayed in Table 3, participants reported moderate levels of usual pain, fatigue, emotional distress, and interference in daily functioning. Table 3 also displays participants expected symptom levels and their benefit requirement levels to practice all health behaviors. A series of paired samples t-tests revealed that, for each symptom domain, participants expected that practicing all health behaviors would result in a significant degree of symptom improvement compared to their usual levels (P < for all comparisons). Additionally, participants required a significantly greater degree of improvement than they expected to get from practicing all health behaviors (P < for all comparisons). To better characterize these differences, change scores (relative to usual symptom levels) were calculated for both expected symptom levels and benefit requirement levels to practice all health behaviors. Table 3 displays these values. The average expected levels of symptom improvement if participants practiced all health behaviors ranged from 8.7 to 11.7 points ( % improvement), depending on the symptom, and the average required symptom improvement to engage in all health behaviors ranged from 17.8 to 28.3 points ( % improvement). RANOVAs (with Sidak correction for multiple comparisons) were conducted on the difference scores determine whether there were differences by symptom domain. An RANOVA on the expected change scores indicated a significant main effect for symptom domain (F 3,255 ¼ 3.2, P ¼ 0.02), and individual comparisons revealed that the amount of expected improvement in emotional distress was significantly less than the expected amount of expected improvement in fatigue (P ¼ 0.01). Similarly, an RANOVA on the benefit requirement change scores revealed a significant main effect for symptom domain (F 3,77 ¼ 8.5, P < 0.001), and individual comparisons indicated that the amount of required improvement in emotional distress was significantly less than the amount of required improvement in pain, fatigue, and interference in daily functioning (P < for all comparisons). In summary, participants expected to improve by 22 26% if they practiced all health behaviors, and they would require even more improvement than that (47 54%) to make it worth their while to practice all health behaviors. While the absolute magnitude of expected or required improvement was less for emotional distress than it was for the other three symptoms, when viewed as a percentage of the usual symptom levels, the percentage of expected and required improvement is similar for all four symptoms. This pattern of results can be explained by the lower ratings of usual emotional distress compared to usual ratings of pain, fatigue, and interference. The lower levels allowed less room for participants to adjust the levels of expected or required improvement in emotional distress compared to that of the other three symptoms. Expectation and Benefit Requirement to Practice Individual Health Behaviors Expectation and benefit requirement for improvement in pain was also assessed for each health behavior 1084

7 Influences on Health Behavior Change Table 2 Participant responses to BET-CP: self-reported practice of health behaviors over the past year Practice of health behaviors over the past year: 0 (never) to 100 (daily) Mean SD Observed range *Test Retest correlation Physical activity Good sleep habits Healthy diet Pain self-management(relaxation, activity pacing) *Pearson r correlation between participant responses at Time 1 and their retest responses 7 days later at Time 2 (N ¼ 25). P < 0.05 compared to physical activity, good sleep habits, and pain self-management. separately. These values are displayed in Table 3, along with their corresponding change scores from usual pain. When asked to think about practicing the healthy behaviors individually, participants average amount of expected improvement in pain ranged from 11.3 to 17.8 points ( % improvement), depending on the health behavior, and their average amount of required improvement in pain ranged from 21.8 to 23.8 points ( % improvement). RANOVA revealed a significant main effect for health behavior for degree of expected change (F 3,255 ¼ 6.5, P < 0.001), and individual comparisons suggested that the amount of expected improvement in pain was significantly greater for the practice of pain self-management (i.e., relaxation and activity pacing) than it was for physical activity or for eating a healthy diet (P < 0.05 for both comparisons). The RANOVA conducted on the degree of required change did not find a significant main effect for health behavior, indicating that participants required approximately the same degree of improvement in pain across the four health behaviors. In summary, participants expected their pain to improve by 21 34% if they practiced the health behaviors individually, and they would require even more improvement than that (41 45%) to make it worth their while to practice the health behaviors individually. While participants required improvement in pain did not differ by individual health behavior, they expected that relaxation and activity pacing would provide them greater pain relief than the other health behaviors. Motivation and Self-Efficacy Participants were asked to rate how capable (i.e., selfefficacy) and how ready (i.e., motivated) they were to practice each health behavior separately and all of them together on 101 point numerical scales. Table 3 displays these results. Ratings of self-efficacy ranged from 68.3 (physical activity) to 82.6 (healthy diet), and ratings of motivation ranged from 72.3 (physical activity) to 82.4 (healthy diet). RANOVAs were conducted to determine if ratings of self-efficacy or motivation differed by health behavior. In the self-efficacy analysis, there was a significant main effect for health behavior (F 4,82 ¼ 6.7, P < 0.001), and individual comparisons revealed that participants rated their capability to eat a healthy diet significantly higher than their capability to engage in physical activity (P < 0.001), good sleep habits (P ¼ 0.005), and all health behaviors together (P ¼ 0.001). A significant main effect for health behavior was also found in the RANOVA on ratings of motivation (F 4,82 ¼ 3.6, P ¼ 0.009). Individual comparisons revealed that participants rated their motivation to eat a healthy diet as higher than their motivation to engage in physical activity (P ¼ 0.009). Study Aim 2: Test Retest Reliability and Convergent Validity of the BET-CP Twenty-five participants provided data for the test retest analysis. Tables 2 and 3 display the Pearson r values for each item correlating participant responses at time 1 with their responses at time 2. Test retest reliability was moderate too strong for all items on the BET-CP (Pearson r correlation range ¼ ); correlation values exceeded 0.70 for 27 of the 34 items, and they exceeded 0.60 for all 34 items. Results of the convergent validity analyses are displayed in Table 4 and were based on data provided by a subsample of 55 participants. Responses on established measures of expectation, success criteria, self-efficacy, and motivation showed convergence (via consistently moderate strength correlations) with similar constructs on the BET-CP. Participants expectation of symptom response to practicing all the health behaviors demonstrated moderate correlations (Pearson r ¼ ) with responses to the PCOQ item assessing expectation about symptom response to unspecified treatment for their pain. Similarly, the BET-CP item assessing participants symptom benefit requirement to make it worth their while to practice all the health behaviors demonstrated moderate correlations (Pearson r correlations ¼ ) with their criteria for successful treatment as assessed by the PCOQ. In both analyses relating the BET-CP with the PCOQ, the pain intensity symptom showed the weakest correlations (Pearson r ¼ 0.30 and 0.41, respectively), while emotional distress demonstrated the strongest correlations (Pearson r ¼ 0.47 and 0.47, respectively). 1085

8 Anderson et al. Table 3 Participant responses to BET-CP: usual symptom levels, and expectation, benefit requirement, self-efficacy, and motivation to practice health behaviors Raw values: mean (SD) Observed range Change from usual symptom levels: raw value (% change)* Test Retest Correlation Usual symptom levels Pain 52.6 (24.2) Fatigue 52.5 (27.5) Emotional distress 37.4 (27.6) Interference in daily function 50.2 (27.2) Expectation of symptom levels if practiced all health behaviors Pain 40.9 (24.9) (22.2%) 0.73 Fatigue 38.7 (29.4) (26.3%) 0.65 Emotional Distress 28.7 (25.7) (23.3%) 0.70 Interference in Daily Function 39.1 (28.9) % (22.1%) 0.69 Benefit requirement of symptom levels to practice ALL health behaviors Pain 27.3 (24.0) (47.9%) 0.69 Fatigue 24.2 (23.3) (53.9%) 0.88 Emotional Distress 19.6 (21.4) (47.6%) 0.85 Interference in Daily Function 23.5 (26.6) (53.2%) 0.78 Expectation of pain level if practiced individual health behaviors Physical activity 41.3 (27.0) (21.5%) 0.63 Good sleep habits 36.5 (24.9) (30.6%) 0.69 Healthy diet 39.1 (24.9) (25.7%) 0.89 Pain self-management (relaxation, activity pacing) 34.8 (24.9) (33.8%) 0.85 Benefit requirement of pain level to practice individual health behaviors Physical activity 30.8 (26.3) (41.4%) 0.67 Good sleep habits 30.2 (26.2) (42.6%) 0.81 Healthy diet 28.9 (25.5) (45.1%) 0.82 Pain self-management (relaxation, activity pacing) 28.8 (25.5) (45.2%) 0.86 Self-efficacy (capability) to practice health behaviors Physical activity 68.4 (29.8) Good sleep habits 70.0 (29.1) Healthy diet 82.6 (23.3) Pain self-management (relaxation, activity pacing) 74.9 (27.1) All health behaviors 72.7 (20.9) Motivation (readiness) to practice health behaviors # Physical activity 72.3 (30.2) Good sleep habits 75.6 (29.5) Healthy diet 82.4 (22.0) Pain self-management (relaxation, activity pacing) 76.3 (27.7) All health behaviors 73.7 (24.5) *Percent change values were derived by subtracting participants expected or required symptom levels from their usual levels, then dividing that number by their usual levels and multiplying by 100. Pearson r correlation between participant responses at Time 1 to their re-test responses 7 days later at Time 2 (N ¼ 25). Rating scale: 0 (none) to 100 (worst imaginable). Rating scale: 0 (not at all capable) to 100 (fully capable). # Rating scale: 0 (not at all ready) to 100 (fully ready). 1086

9 Influences on Health Behavior Change Table 4 Convergent validity: participants responses on the BET-CP correlated with their responses on the PCOQ, PSEQ, and MPRCQ-2 (N ¼ 55)* Pearson r value P value BET-CP expectation of symptom levels if practiced all health behaviors correlated with PCOQ expectation item Pain Fatigue Emotional distress 0.47 <0.001 Interference in daily function BET-CP benefit requirement of symptom levels to practice ALL health behaviors correlated with PCOQ success criteria item Pain Fatigue Emotional Distress 0.47 <0.001 Interference in Daily Function BET-CP ratings of self-efficacy (capability) to practice each healthy behavior correlated with PSEQ total score Physical activity 0.55 <0.001 Good sleep habits Healthy diet 0.47 <0.001 Pain self-management (relaxation, activity pacing) 0.36 <0.006 All healthy behaviors 0.46 <0.001 BET-CP ratings of motivation (readiness) to engage in physical activity correlated with MPRCQ-2 exercise subscale # BET-CP ratings of motivation (readiness) to practice pain 0.49 <0.001 self-management correlated with MPRCQ-2 relaxation subscale BET-CP ratings of motivation (readiness) to practice pain self-management correlated with MPRCQ-2 pacing subscale *Responses from 55 participants were used in the convergent validity analyses. Responses from the other 31 participants were not used in these analyses due to the potential confounding effects of a brief psychoeducational intervention that was performed prior to their completion of the validity measures. Description and results of the intervention are beyond the scope of this report. Text of PCOQ expectation item: We would like to know what you expect your treatment to do for you. On a scale of 0 (none) to 10 (worst imaginable), please indicate the levels you expect following treatment. Text of PCOQ success criteria item: On a scale of 0 (none) to 10 (worst imaginable), please indicate the level each of these areas would have to be at for you to consider treatment successful.. Example of a PSEQ item: I can do most of the household chores (e.g., tidying-up, washing dishes, etc.), despite the pain. Items are rated on a 0 (not at all confident) to 6 (completely confident) scale. # Example of an MPRCQ-2 item: Exercise for at least 30 min, 3 times per week or more. Items are rated on a 1 (I am not doing this now, and am not interested in ever doing it) to 7 (I have been doing this for a long time (at least 6 months) scale. Participant ratings of self-efficacy to practice the health behaviors demonstrated small to moderate associations with their PSEQ total score. Capability to practice good sleep habits showed the smallest correlation (Pearson r ¼ 0.25) with the PSEQ measure, while capability to engage in physical activity demonstrated the largest correlation (Pearson r ¼ 0.55). Finally, to analyze the convergent validity of the BET-CP assessment of motivation, BET-CP ratings of readiness to engage in physical activity were moderately correlated (Pearson r ¼ 0.44) with the MPRCQ-II exercise subscale, while ratings of readiness to practice pain self-management demonstrated moderate correlations with the relaxation subscale (Pearson r ¼ 0.49) and the pacing subscale (Pearson r ¼ 0.37). Discussion The present study describes an assessment of the cognitive-motivational influences on health behavior change in adults with chronic pain using a new self-report instrument (the BET-CP) designed to measure the level of motivation to practice health behaviors, as well as the beliefs that drive this motivation. Among the beliefs assessed include those of outcome expectancies, and a novel measure of cost-benefit: the symptom benefit that patients require to make it worth their while to engage in the health behaviors. Findings suggest that patients are indeed able to make a distinction between these two concepts, and results revealed the discrepancy between the amount of symptom improvement 1087

10 Anderson et al. expected versus required from practicing the health behaviors. While participants endorsed modest expectations of symptom improvement if they did practice the health behaviors (22 26% improvement), they required twice that (47 54% improvement) to make it worth their while to commit to practicing them. Thus, while participants expectations for symptom improvement are more in line with the reality of behavioral pain treatment effectiveness, their stringent benefit requirement suggests that they may never actually engage in the behaviors, or may drop from behavioral treatment once they have a chance to sample its effectiveness. This result is consistent with previous studies on Robinson s PCOQ [20], which found that patients do not require 100% symptom abatement to consider pain treatment a success, but they may require larger reductions in their pain than existing treatments are able to provide. Participants stringent benefit requirement beliefs did not necessarily translate into an absence of motivation or complete refusal to practice the health behaviors in this sample. On average, participants reported practicing the behaviors at least some of the time. There are several possibilities that may explain this pattern of results. First, the group means reported in the results do not capture individual variability in responses. Second, benefit requirement beliefs (and their relationship to expectations and behavioral practice frequency) may not be an all-or-none phenomenon. A patient with benefit requirements that exceed expectations might practice the behavior, but do so with less frequency or less sustained effort. Third, as detailed in the Motivational Model for Pain Self-Management, benefit requirement beliefs are only one factor that contributes to motivation to practice behaviors. Other factors like self-efficacy or environmental and medical barriers can also account for the variance in practice frequency. Finally, the relationship between patient beliefs and frequency of behavioral engagement may be more apparent when examining these issues longitudinally rather than cross-sectionally. Benefit requirement beliefs may be more relevant in predicting future behavior, like dropout or adherence, particularly when assessed just prior to initiating practice of the health behavior. The BET-CP also allows for the differentiation of beliefs by health behavior. On average, participants expected to benefit most from the practice of relaxation and activity pacing and benefit least from engaging in physical activity. This result may, in part, be explained by the literature on kinesiophobia; individuals with chronic musculoskeletal pain commonly fear that movement will exacerbate their pain and lead to further damage [49,50] and may favor more sedentary activities as a result. Consistent with this fear, in the current sample, 11 (12.7%) of the 86 participants expected that regular exercise would worsen their pain, and an additional 25 participants (29.1%) expected that exercise would result in no change to their pain levels. Extrapolating clinically, psychoeducation on the benefits of exercise (to painrelated symptoms, and to other health variables like weight, mood, and cognitive functioning) may help to counteract fear-avoidance beliefs and potentially increase patients benefit to cost ratio with regard to engaging in physical activity. A graded exposure approach to the prescription of physical activity may also help to correct these beliefs and mitigate dropout from an exercise regimen. While participants expected to get the most symptom relief from practicing relaxation and activity pacing, they were most confident and motivated to eat a healthy diet, and both motivation and self-efficacy correlated with their recent practice frequency of the behaviors. This may be related to a greater familiarity with the actions involved in eating healthy. Dieting is a ubiquitous health behavior discussed in popular culture and prescribed frequently by physicians, while relaxation, activity pacing and sleep behaviors are less commonly prescribed in medical settings and typically reserved for specialty behavioral health treatment. The health benefits associated with healthy eating also extend well beyond pain management, so participants may have been eating healthy for reasons other than pain-related symptom relief. Participants reported moderate to high levels of motivation, self-efficacy, and recent practice frequency for each of the health behaviors. Historically, self-report of health behavior practice is subject to an over-reporting bias and has not correlated highly with objective measures [51]. With this in mind, the relatively high levels of self-efficacy, motivation, and behavioral practice frequency reported in the current study should be interpreted with caution as they may not correlate precisely with actual practice frequency. Instead the responses might be best viewed as a general guide to the clinician and patient, suggesting generally what the patient has been doing more or less of recently, and what they are confident and motivated to practice in the future. Preliminary psychometric analyses suggest that the constructs measured on the BET-CP are stable and can be reliably assessed over time. Test-retest reliability over 7 days was strong (Pearson r.70) for 27 of the 34 items, and moderate to strong (Pearson r ¼.63.69) for the remaining items. Expectation and benefit requirement items showed some variability in reliability. Generally, these two constructs may be less familiar to participants than the others, and require a certain degree of hypothetical prognostication in gauging their responses. Thus, patients may not be as reliable in their responses to these items. The BET-CP also demonstrated convergence with existing measures of similar constructs via consistently moderate strength correlations. The moderate correlation values were large enough to demonstrate that the BET- CP taps into its hypothesized constructs of expectation of treatment response, self-efficacy, and motivation, but not so large as to suggest redundancy in the measures. Notably, with regard to self-efficacy, even though the 1088

11 PSEQ and BET-CP are not assessing the same behaviors, the two measures nonetheless demonstrated moderate strength correlations. This suggests that behavioral self-efficacy may be a generalizable state for individuals with chronic musculoskeletal pain. As they gain confidence in their ability to participate in one behavior, it may lead to confidence in others as well. Limitations The study has several limitations. First, the sample size is relatively small to establish the psychometric properties of a new measure. As such, findings on the reliability and validity of the BET-CP should be considered preliminary. Further psychometric study is needed with larger sample sizes. Discriminant validity of the BET-CP needs to be established, in addition to examining the test-retest reliability of beliefs using a longer betweentest interval. Second, the design of the study did not allow for corroboration of pain diagnosis by physician or medical chart in the subset of participants who responded to community advertisements. Diagnoses were dependent upon the veracity of self-report which may be subject to error in knowledge or memory. Further, while the sample is diverse with regard to age, setting, and type of musculoskeletal pain, participants were predominantly White Caucasian and welleducated. It is unknown if results are generalizable to other types of pain or to populations with more ethnic or educational diversity. Additionally, our one-item assessment of recent practice of health behaviors is not ideal, and may compromise accuracy in favor of brevity. Self-report is a historically inaccurate method of assessing health behavior practice, subject to biases in memory. A one-item assessment using a prolonged time period of retrospection (i.e., one year) may compound this problem. Finally, the BET-CP is not a measure that captures all factors involved with behavioral treatment adherence. It focuses on patient psychological factors, and does not take into account the environmental influences (finances, safety, travel distance from gym), patient medical factors (type and location of pain, comorbidities), or clinician characteristics (e.g., personality) that may affect adherence to treatment. Conclusion and Future Directions Use of a new measure of the psychological influences on health behavior change revealed that, to make it worth their while to commit to practicing health behaviors, patients with chronic musculoskeletal pain require approximately two times more treatment-related symptom improvement than they expect to get from the practice of the behaviors. This mismatch in behavioral treatment outcome expectancies and benefit requirement may negatively impact adherence to and satisfaction with treatment. Consistent with patient-centered care, the results highlight the need for pain clinicians to open a dialogue with their patients about what they expect and need from their treatment, and how Influences on Health Behavior Change confident and motivated they are to adhere to their prescribed regimen. The BET-CP represents a vehicle for the provider to ask about these issues in a behavioral treatment context. Future work is needed to determine the clinical importance of a discrepancy between patient outcome expectations and benefit requirement beliefs related to the practice of health behaviors. Future work should also test whether or not unrealistic beliefs about the effectiveness of behavioral treatments for pain are able to be altered. With regard to the BET-CP, further psychometric studies are needed using larger sample sizes that establish the discriminant validity of the measure. The BET-CP could also benefit from a revision. This might include improving the specificity of healthy diet practice recommendations, using a smaller retrospective time period to ask patients about their recent practice frequency of the health behaviors, and asking separately about the behaviors of relaxation and activity pacing rather than combining them. References 1 Bauer UE, Briss PA, Goodman RA, Bowman BA. Prevention of chronic disease in the 21st century: Elimination of the leading preventable causes of premature death and disability in the USA. Lancet 2014;384: Morley S, Eccleston C, Williams A. Systematic review and meta-analysis of randomized controlled trials of cognitive behaviour therapy and behaviour therapy for chronic pain in adults, excluding headache. Pain 1999;80: McCracken LM, Turk DC. Behavioral and cognitivebehavioral treatment for chronic pain: Outcome, predictors of outcome, and treatment process. Spine (Phila Pa 1976) 2002;27: Mork PJ, Vik KL, Moe B, et al. Sleep problems, exercise and obesity and risk of chronic musculoskeletal pain: The Norwegian HUNT study. Eur J Public Health 2014;24: Nilsen TI, Holtermann A, Mork PJ. Physical exercise, body mass index, and risk of chronic pain in the low back and neck/shoulders: Longitudinal data from the Nord-Trondelag Health Study. Am J Epidemiol 2011;174: Hitt HC, McMillen RC, Thornton-Neaves T, Koch K, Cosby AG. Comorbidity of obesity and pain in a general population: Results from the Southern Pain Prevalence Study. J Pain 2007;8: Kim CH, Luedtke CA, Vincent A, Thompson JM, Oh TH. Association of body mass index with symptom 1089