ORIGINAL ARTICLE
A Longitudinal Investigation of Pain-related Social Support Preferences in a Chronic Pain Treatment Sample Lachlan A. McWilliams, PhD,* Kristen S. Higgins, BScH,w Bruce D. Dick, PhD,z and Michelle J. Verrier, BAz
Objectives: Chronic pain patients’ pain-related social support preferences have received little clinical or research attention. However, recent research utilizing the Pain Response Preference Questionnaire (PRPQ) has indicated that such preferences are related to pain-related disability. This study investigated whether pain-related social support preferences: (1) were related to disability levels, (2) predicted changes in pain and disability following a group-based treatment program, and (3) changed from pretreatment to posttreatment. Methods: At pretreatment and posttreatment, chronic pain patients (N = 45) participating in a self-management psychoeducation program completed the PRPQ and measures of pain severity and disability. Results: A multiple regression analysis indicated that at pretreatment the Solicitude scale of the PRPQ was positively associated with disability, and that this association was moderated by the other 2 PRPQ scales (Activity Direction and Suppression). The PRPQ scales did not predict changes in pain or disability and there were no significant changes in the PRPQ scale scores. However, participants did not respond to the Solicitude scale in a consistent manner from pretreatment to posttreatment (51.1% decreased, 35.5% increased, and 13.3% remained the same). These changes were unrelated to changes in pain or disability. Discussion: Additional research aimed at identifying the processes underlying the association between pain-related support preferences and disability is warranted. Given the potentially important role that a desire for solicitous support may play in shaping the social context of pain, understanding the unexpected changes in Solicitude scale scores found in the present study may be particularly important for improving self-management treatments for chronic pain. Key Words: chronic pain treatment, social support, solicitude, activity direction, suppression
(Clin J Pain 2014;30:672–678)
T
here has been a growing research interest in a wide range of interpersonal topics, such as communal coping,1 empathy,2 and self-perceived burden,3 that are relevant to the experience of chronic pain. Given this trend, it is
Received for publication February 7, 2013; revised May 19, 2014; accepted August 13, 2013. From the *Department of Psychology, University of Saskatchewan, Saskatoon, SK; wDepartment of Psychology and Neuroscience, Dalhousie University, Halifax, NS; and zDepartment of Anesthesiology and Pain Medicine, University of Alberta, Edmonton, AB, Canada. The authors declare no conflict of interest. Reprints: Lachlan A. McWilliams, PhD, Department of Psychology, University of Saskatchewan, 9 Campus Drive, 154 Arts, Saskatoon, SK, Canada S7N 5A5 (e-mail: [email protected]). Copyright r 2013 by Lippincott Williams & Wilkins
672 | www.clinicalpain.com
surprising that there has been little research directed toward understanding what forms of social support chronic pain patients find helpful. Newton-John and Williams’4 qualitative research on pain patients and their spouses was the first to address this topic. Through interviews conducted with the spouses, they identified a wide range of possible responses to pain behavior. Interviews with the pain patients revealed that the most positively perceived responses were those involving active coping, such as encouraging task persistence and problem solving. Responses involving solicitous behaviors (viz., providing and offering help) were rated less favorably, with many patients reporting that this type of support made them feel guilty, useless, and burdensome. To study pain-related support preferences further, McWilliams et al5 developed the Pain Response Preference Questionnaire (PRPQ). It assess the degree to which several types of responses to pain are wanted from significant others. The items for this measure were originally designed to capture the forms of support identified by Newton-John and Williams,4 but factor analyses were used to develop empirically derived scales. The most recent of these utilized a chronic pain treatment sample6 and identified scales labeled Solicitude, Activity Direction, and Suppression. McWilliams et al6 used the PRPQ to determine whether pain-related support preferences are related to disability levels. They suggested that pain-support preferences may be related to disability levels because preferences could motivate efforts to obtain specific forms of support. For example, an individual with a strong desire for solicitous support may engage in pain behavior as a method of eliciting such support, and this method of support seeking could result in an elevated level of disability. Alternatively, pain-related support preferences could influence the degree to which an individual is responsive to specific forms of support. For example, solicitous responses to pain behavior would likely not reinforce pain behavior for an individual with a low interest in this form of support. In this case, solicitous support would not be expected to lead to increased pain behavior and disability. McWilliams et al6 found that the correlations between the PRPQ scales and disability ratings were small and nonsignificant. However, a subsequent multiple regression analysis that included sex, pain severity ratings, and all 3 PRPQ scales revealed that each PRPQ scale had a statistically significant association with disability ratings. Solicitude was positively associated with disability, whereas Activity Direction and Suppression were both negatively associated with disability. This pattern of findings indicates that there is a relationship between pain-support preferences and disability, but these relationships are complex rather than simple bivariate relationships. The overall pattern of findings (ie, the PRPQ scales being unrelated to Clin J Pain
Volume 30, Number 8, August 2014
Clin J Pain
Volume 30, Number 8, August 2014
disability in the correlations and all of them being related to disability when they were simultaneously entered in the regression model) suggests that they can be considered suppressor variables for each other (ie, they increase the predictive validity of the other variables when included within the same regression model). The first aim of the present study was to attempt to replicate earlier findings of relationships between pain-support preferences (ie, the PRPQ scales) and disability. Given the complex relationships between the PRPQ scales and disability in past research,6 the current study also utilized a multivariate approach (ie, multiple regression analysis) to explicate such relationships by considering the possibility that the PRPQ scales are suppressor variables for each other and by exploring interactions between the PRPQ scales. The second aim of the study was to investigate whether pain-support preferences predict chronic pain patients’ responses to a group-based psychoeducation program that emphasized cognitive-behavior therapy principles and selfmanagement strategies for coping effectively with chronic pain. Those with a strong desire for solicitous support may be reluctant to adopt a self-management approach. Therefore, it was hypothesized that pretreatment levels of Solicitude would be associated with poorer responses to treatment. Conversely, those wanting encouragement to engage in activities might be particularly responsive to a self-management approach. Thus, it was also hypothesized that pretreatment levels of Activity Direction would be associated with stronger responses to treatment. The third aim of the study was to investigate whether pain-related support preferences changed after the completion of the treatment program. As the self-management program encourages participants to independently practice a variety of strategies to manage their pain and increase their engagement in meaningful activities, it was reasonable to anticipate that participants would experience a reduced desire for solicitous support at posttreatment. However, the self-management program utilized in the present study did not directly address the topic of pain support preferences or include content advocating for specific changes in support preferences, so it was also reasonable to anticipate that participants’ pain-related support preferences would remain relatively stable. Thus, there were no hypotheses regarding changes in pain-related support preferences. Given that pain-related support preferences have been found to be associated with disability,6 it is important to understand whether self-management programs modify such preferences. Changing pain-related support preferences may be a yet to be identified mechanism by which self-management programs produce decreases in disability. Conversely, the persistence of maladaptive patterns of support preferences may interfere with the adoption of selfmanagement strategies. If this is the case, it would suggest that self-management programs for those with chronic pain could be improved by the inclusion of strategies aimed at addressing maladaptive pain-related support preferences.
METHODS Participants Participants were 45 adults (73.3% female) participating in the Pain 101 program at the Multidisciplinary Pain Centre at the University of Alberta Hospital in Edmonton, Alberta, Canada. Participants ranged in age from 19 to 71 years, with a mean age of 47.38 years (SD = r
2013 Lippincott Williams & Wilkins
Pain Support Preferences
11.25). Most of the participants reported being married or in a common law relationship (68.9%); others were divorced or separated (15.6%), in a dating relationship (4.4%), widowed (2.2%), or single (8.9%). A majority of participants were not currently working (33.3% unemployed, 20.0% on sick leave or medical leave, 11.1% students, and 6.7% retired), but approximately 25% of participants were employed (15.6% employed full time, 13.3% employed part time). The average pain duration of the sample was 12.49 years (SD = 12.41). The sample was highly diverse in terms of the type of pain reported. The most common types were neuropathic pain (n = 16), musculoskeletal pain (n = 6), fibromyalgia (n = 6), migraine (n = 3), and headache (n = 2). None of the conditions experienced by the remainder of the sample (n = 12) were reported more than once.
Procedures Consenting participants were asked to complete a package of questionnaires before the first Pain 101 session. They also completed the same set of self-report measures after the final session. In place of names, codes were assigned to each participant and scores were entered in an encrypted database to assure confidentiality and anonymity. This study was approved by the Human Research Ethics Board at the University of Alberta.
Treatment Program The Pain 101 program is a 10-week group-based psychoeducation treatment for adults with chronic pain. The program is based on a cognitive-behavior therapy model that emphasizes the importance of self-management and living well despite pain, rather than focusing on pain reduction. Each session lasts for 2 hours and provides participants with homework to be completed on a daily basis between the weekly sessions. The homework is based on practical pain management strategies established as a part of a best practice pain management program. Session topics include: Pain Education, Tension and Relaxation, Goal Setting and Pacing, Depression and Managing Negative Thoughts, Sleep and Chronic Pain, Coping with Stress and Anxiety, Communication and Pain Behavior, Mindfulness, Planning for and Dealing with Setbacks, and Life with Chronic Pain. The session that is most relevant to social support preferences is the Pain Behavior and Communication session. This session focuses on helping participants understand what and how they communicate with both verbal and nonverbal messages. A significant goal of that session is to help participants consider what they hope to achieve in their communication with others, including significant others, and the types of messages that are most important to them. Assertiveness training is also included in this session. Many programs present pain behavior in a negative light and work toward eliminating such behavior. The Pain 101 program does not take this approach, but instead aims to help participants to identify and reduce behaviors that are unhelpful and to communicate about their pain more effectively, both verbally and nonverbally. The Pain 101 program does not specifically target social support preference in any of its sessions. However, the social component of the biopsychosocial model upon which this program is based does focus on the effects of chronic pain on relationships and the way that relationship and support factors influence the individual suffering from chronic pain. When participants choose to set goals related to their www.clinicalpain.com |
673
Clin J Pain
McWilliams et al
relationships, it is not uncommon for a discussion of support factors to be included during the process of discussing goals.
Measures Demographic Information Participants were asked to report their age, sex, marital status (single, married or common law, divorced or separated, in a dating relationship, or widow), employment status (employed full time, employed part time, on sick leave or medical leave, unemployed, retired, or student), and location and duration of chronic pain.
PRPQ
The PRPQ5 assesses preferences for different types of pain-related social support from one’s significant other. It presents a range of possible responses to pain that a romantic partner could provide (eg, treat me with extra care and concern), and participants are asked to indicate on a scale from 1 (strongly disagree) to 4 (strongly agree) how much they would like their partner to respond to them in this way when they are in pain. The earliest version of the PRPQ, comprised of 35 items, was used. The scales used in the current study were based on a factor analytic study of the PRPQ conducted with chronic pain patients.6 These included a 10-item Solicitude scale (eg, “treat me with extra care and concern”), a 9-item Activity Direction scale (eg, “tell me I can do things despite pain”), and a 3-item Suppression scale (eg, “help me ignore the pain”). Scores for each scale were created by taking the average score of the items comprising each scale. These scales were found to be internally consistent both at baseline (Solicitude: a = 0.87; Suppression: a = 0.69; Activity Direction: a = 0.87) and at the completion of the Pain 101 program (Solicitude: a = 0.76; Suppression: a = 0.83; Activity Direction: a = 0.80).
Pain Severity Participants rated their pain on an 11-point scale ranging from 0 (no pain) to 10 (worst pain imaginable).
Pain Disability Index (PDI)
The PDI7 assesses the extent to which chronic pain affects a person’s ability to engage in various life activities. The scale is comprised of 7 items which represent the following life activities: family/home responsibilities, recreation, social activity, occupational duties, sexual behavior, self-care, and life-support activity. Participants are asked to rate how much their pain interferes with each activity on a scale from 0 (no disability) to 10 (total disability). Participants’ PDI scores were created by averaging their responses to the 7 items in the scale. The PDI scale was found to be internally consistent in the present study (Pretreatment: a = 0.78; Posttreatment: a = 0.85).
RESULTS Preliminary Analyses Correlations between all of the baseline measures are reported in Table 1. As expected, pain and disability had a statistically significant positive association. On the basis of Cohen’s8 recommendations for describing effect sizes for correlations, this association would be considered moderate (ie, r > 0.30). There were also statistically significant
674 | www.clinicalpain.com
Volume 30, Number 8, August 2014
positive associations between the PRPQ scales. The association between the Activity Direction and Suppression scales was moderate, and the association between the Solicitude and Activity Direction scales was large (ie, r > 0.50).
Predicting Baseline Disability Levels A multiple regression analysis was used to predict PDI scores at baseline. On the basis of Aiken and West’s9 recommendations for testing interaction effects, the continuous variables were centered, and the interaction terms were created with these centered variables. Pain severity ratings and sex were entered in step 1. At this step, only pain severity ratings had a significant association with PDI scores (b = 0.38). The PRPQ scales (Solicitude, Activity Direction, and Suppression) were entered in step 2. When steps 1 and 2 were entered, sex (b = 0.29) and pain severity rating (b = 0.43) had significant relationships with PDI scores. The 2-way interactions between the PRPQ scales were entered in step 3. None of these variables were significantly associated with PDI scores, but sex (b = 0.31) and pain severity ratings (b = 0.41) remained significant. Finally, the 3-way interaction between the PRPQ scales was entered in step 4, and revealed a statistically significant interaction effect. The summary statistics for each model are reported in Table 2, along with the standardized b values for the final model. The final model accounted for 42.0% of the variance in PDI scores. The final model showed that sex, pain ratings, the Solicitude scale, and the 3-way interaction between the PRPQ scales were significant predictors of baseline PDI scores. To explore the 3-way interaction effect involving PRPQ scales, the regression equation from the final model was used to determine plot points. Values of 1 SD above and below the means of each PRPQ scale were plotted. As there was a significant main effect of Solicitude and it had the strongest relationship with disability in past research,6 Solicitude was treated as the focal predictor, whereas Suppression and Activity Direction were treated as moderators. The continuous variables in this analysis were centered, and therefore zeros were entered into the analysis for all of the continuous variables other than the PRPQ scales. Zero was also entered for the categorical sex variable, which represents females. These values are plotted in Figure 1. Preacher et al’s10 procedures were used to determine the statistical significance of the relationships shown in Figure 1. At the high value of Suppression and the low value of Activity Direction, there was a significant positive association between Solicitude and PDI scores (b = 3.65, P = 0.013). At the high values of both Suppression and Activity Direction, there was a marginally significant positive association between Solicitude and PDI scores (b = 1.67, P = 0.097). At the other 2 combinations of high and low scores on Suppression and Activity Direction, the associations between Solicitude and PDI scores were small and not statistically significant. Thus, the overall pattern of findings indicated a positive association between Solicitude and PDI scores (ie, the main effect), and that this relationship was strongest at low levels of Activity Direction and Suppression. Solicitude scores were not related to PDI scores at other combinations of high and low levels of Activity Direction and Suppression.
Longitudinal Analyses Correlations between the pretreatment and posttreatment values of each variable (pain ratings, the PDI, and r
2013 Lippincott Williams & Wilkins
Clin J Pain
Volume 30, Number 8, August 2014
Pain Support Preferences
TABLE 1. Correlations Among Pain Ratings, Pain Disability Index, and PRPQ Scales at Baseline
Measure
1
2
3
4
5
6
Pain rating Pain disability index Solicitude Activity direction Suppression Sex
1
0.38* 1
0.17 0.13 1
0.00 0.13 0.73*** 1
0.04 0.04 0.27 0.45** 1
0.03 0.23 0.15 0.20 0.02 1
*Pr0.05; **Pr0.01; ***Pr0.001 (all 2-tailed).
PRPQ scales) are reported in Table 3. All variables except for the Solicitude scale had statistically significant positive associations between pretreatment and posttreatment values. These relationships were all considered to represent large effect sizes. Paired t tests were conducted to compare the pretreatment and posttreatment means of each variable (Table 3). Pain ratings and PDI scores decreased, and these changes were statistically significant. Cohen’s8 d values were calculated to provide indices of effect size. There was a small effect size regarding pain and disability. There were no statistically significant changes with regard to the PRPQ scales. On the basis of the notion that those with particular pain-related support preferences may be more or less receptive to the treatment provided, regression analyses were conducted to determine whether the baseline PRPQ scales were predictive of treatment outcomes (ie, posttreatment functioning after accounting for pretreatment functioning). Posttreatment PDI scores and pain ratings were used as dependent variables. Given that posttreatment functioning would be expected to be related to baseline levels of the variable of interest, centered pretreatment PDI scores and pain ratings were entered in step 1 of the analyses predicting posttreatment PDI scores and pain ratings, respectively. Sex was also included in step 1. Baseline PRPQ scores were entered in step 2. The 2-way interactions between the PRPQ scales were entered in step 3, and the 3-way interaction between them was entered in step 4. In step 1 of these analyses, baseline PDI scores were positively associated with posttreatment PDI scores (b = 0.65, P < 0.001), and baseline pain ratings were positively associated posttreatment pain ratings (b = 0.56, P < 0.001). In both cases the PRPQ scales and their interaction terms were unrelated to the treatment outcome variables. The longitudinal findings regarding Solicitude were unusual. The null results from the paired-sample t test
indicated that Solicitude scores did not change in a systematic manner (ie, generally increasing or decreasing) and the null findings of the correlation analysis indicated that there was no statistically significant consistency over the course of the study in terms of responding to this measure. Change in Solicitude scale values (calculated by subtracting the posttreatment Solicitude scores from the pretreatment scores) indicated that 23 participants’ Solicitude scores decreased after completing the treatment program, 6 participants’ scores did not change, and 16 participants’ Solicitude scores increased after treatment. The inconsistency in Solicitude scores from pretreatment to posttreatment raised the possibility that the changes in Solicitude may relate to treatment outcomes. Multiple regression analyses were used to investigate whether change in Solicitude predicted posttreatment PDI scores and pain ratings. In the analysis predicting posttreatment PDI scores, sex, pretreatment PDI scores, and pretreatment Solicitude were entered in step 1. Change in Solicitude was entered in step 2. A similar analysis was used to predict posttreatment pain ratings, but pretreatment pain ratings were included in step 1 rather than pretreatment PDI scores. Consistent with earlier analyses, in the models including only step 1, pretreatment PDI scores were positively associated with posttreatment PDI scores and pretreatment pain ratings were positively associated with posttreatment pain ratings. In both cases, the addition of step 2 revealed a nonsignificant positive association between change in Solicitude and the treatment outcome variable.
DISCUSSION The first aim of the present study was to replicate earlier findings of relationships between pain-related support preferences and disability. Consistent with past
TABLE 2. Multiple Regression Analysis Predicting Pain Disability Index Scores at Baseline
Step 1 2 3 4
Predictors
b
F
DR2
Sex Pain rating Solicitude Activity direction Suppression Solicitude activity direction Solicitude suppression Activity direction suppression Solicitude activity direction suppression
0.33* 0.43** 0.58* 0.04 0.12 0.23 0.39 0.26 0.58*
5.17**
0.20**
2.72*
0.06
2.41*
0.09
2.82*
0.07*
Standardized coefficients reported are from the final model. *Pr0.05; **Pr0.01 (all 2-tailed).
r
2013 Lippincott Williams & Wilkins
www.clinicalpain.com |
675
Clin J Pain
McWilliams et al
10 Pain Disability Index score
9 8 7 6 5
AD Low Sup Low
4
AD Low Sup High
3
AD High Sup Low
2
AD High Sup High
1 0 Solicitude Low
Solicitude High
FIGURE 1. Exploration of the 3-way interaction between Solicitude, Activity Direction (AD), and Suppression (Sup) predicting baseline Pain Disability Index scores.
research,6 none of the PRPQ scales had a simple linear association with disability, but significant findings emerged in a multiple regression analysis. In earlier research, Solicitude had a positive association with disability, whereas Activity Direction and Suppression had negative associations with disability. The pattern of findings that emerged in the current study was more complex. In the final model, there was a main effect regarding the Solicitude scale and a 3-way interaction effect involving the 3 PRPQ scales. Thus, at the mean levels of Activity Direction and Suppression, there was a significant positive association between Solicitude and disability. Probing of the 3-way interaction indicated that the relationship between Solicitude and disability ratings was statistically significant at high levels of Suppression and low levels of Activity Direction, but not at the other 3 combinations of high and low responses to the Suppression and Activity Direction scales. Thus, a high desire for Solicitude and Suppression and a low level of interest in Activity Direction is the combination of painrelated support preferences most likely to be associated with heightened disability. Chronic pain patients with this pattern of PRPQ scores would be expected to want those close to them to: (1) draw attention away from their pain (ie, high Suppression), (2) provide a high level of emotional and instrumental support (ie, high Solicitude), and (3) not encourage a high level of activity or persistence with activities despite pain (ie, low Activity Direction). The possibility that some chronic pain patients may both endorse a desire for suppression and solicitous support is counterintuitive. However, it is possible that such
Volume 30, Number 8, August 2014
individuals could want a low level of attention directed toward their pain, and also want a high level of support and attention because of their pain. This is the second study using the PRPQ to demonstrate relationships between pain-related support preferences and self-reports of disability. Thus, additional research regarding pain-related social support preferences is warranted. To date, research has not examined the mechanism (or mechanisms) underlying the associations found. It is possible that disability shapes pain-related support preferences, with more disabled individuals preferring support that involves active assistance from others to compensate for that disability. Alternatively, pain-related preferences could influence the social context of pain in a manner that could exacerbate disability. For example, those who prefer solicitous support may have their pain behavior reinforced more strongly by solicitous responses to such behavior, and that could exacerbate their disability. As well, those with a preference for solicitous support may behave in a manner (eg, increased pain behavior) that is particularly likely to draw solicitous support from others thereby increasing their disability. Research paradigms have been developed for investigating patient-spouse interactions,11 and could be utilized to investigate relationships between pain-support preferences, interaction patterns, and disability levels. Participants’ self-reports indicated that they experienced significantly lower levels of pain and disability at posttreatment than at pretreatment. Given the absence of a control condition in the current study, it is impossible to rule out factors other than the treatment as being responsible for these changes. Nonetheless, the current findings suggest that participants had a positive response to the Pain 101 program. The second aim of the study was to investigate whether pain-related support preferences predicted responses to this group-based psychoeducation program. It was thought that pain-related support preferences would influence the ability of participants to adopt the self-management approach utilized in this program, which would in turn influence their responses to the treatment (ie, posttreatment disability and pain adjusted for pretreatment levels of these variables). Contrary to what was expected, the PRPQ scales were not associated with treatment outcomes. These null finding suggest that the ability to benefit from the treatment program was not compromised by patients’ pretreatment pain-related support preferences. Given that the content of the treatment program largely centered on encouraging individuals to change their own thinking and behavior, issues related to social support and/ or social support preferences may be relatively irrelevant to success in implementing the strategies taught in the
TABLE 3. Correlations and Comparisons Between Pretreatment and Posttreatment Variables (N = 45)
Pretreatment Variables Pain rating Pain disability index Solicitude Activity direction Suppression
Posttreatment
r
M
SD
M
SD
t
P
Cohen’s d
0.56** 0.67** 0.19 0.49** 0.51**
6.26 6.21 3.08 2.84 2.06
1.60 1.63 0.55 0.65 0.64
5.62 5.68 2.98 2.83 2.10
2.03 1.97 0.41 0.51 0.74
2.43* 2.36* 1.08 0.17 0.43
0.019 0.023 0.287 0.866 0.668
0.376 0.362 0.164 0.017 0.059
*Pr0.05; **Pr0.001 (all 2-tailed).
676 | www.clinicalpain.com
r
2013 Lippincott Williams & Wilkins
Clin J Pain
Volume 30, Number 8, August 2014
program. However, in light of the relatively small sample size and the associated limited statistical power, the possibility of a type II error cannot be ruled out. Thus, additional research examining the PRPQ scales as prospective predictors of outcome is certainly warranted. The third aim of the study was to investigate whether pain-related support preferences changed after completion of the treatment program. The differences between the pretreatment and post-treatment scores on Activity Direction and Suppression scales were small and not statistically significant. In addition, participants’ pretreatment scores on these 2 scales had large significant positive associations with their respective posttreatment scores on these scales. This pattern of findings indicates that there was no overall change in Activity Direction and Suppression and that participants’ relative positions on these variables were largely consistent overtime. A different pattern emerged regarding pretreatment and posttreatment scores on the Solicitude scale. As with the other PRPQ scales, a comparison of pretreatment and posttreatment Solicitude scale scores indicated that there was no overall change. However, it is important to note that the pretreatment and posttreatment Solicitude scores were not significantly correlated with each other. This absence of a significant positive association indicates that the relative position of participants on the Solicitude scale changed during the course of the study. A slight majority (51.1%) of participants’ scores indicated a decreased level of endorsement of the Solicitude scale. Given the treatment utilized in the study taught self-management skills, it is not surprising that the majority of participants changed in this direction as it may partially reflect the adoption of selfmanagement approach. Follow-up multiple regression analyses indicated changes in Solicitude scores were unrelated to treatment outcomes. Thus, the findings indicate that a decreased desire for solicitous support was commonly experienced by study participants, and also suggest that this type of change was not a process that led to decreases in pain and disability. It is also noteworthy that a large minority of participants (35.5%) increased their endorsement of the Solicitude scale. Some clinicians might consider an increase of this nature to be a potentially harmful development. However, it is important to recognize the possibility that such changes could be adaptive. Several investigations12 have identified subgroups of chronic pain patients characterized by high levels of thought suppression, the use of distraction, and task persistence despite pain. This rigid pattern of responding to pain is thought to be harmful because it can lead to reinjury. Among this type of patient, an increased interest in solicitous support might reflect a new and adaptive willingness to accept the help and support of others. The data captured in this study did not include variables that would permit us to explore whether this was the case in this population. Future research is needed to explore this possibility. The two main limitations of the current study should be noted. First, the addition of a no-treatment control group would have provided for a more thorough evaluation of the changes in Solicitude scores noted in the present study. It has been assumed that some aspect of the treatment program may have prompted participants to reconsider their interest in solicitous support and that this process could have been responsible for the absence of association between pretreatment and posttreatment r
2013 Lippincott Williams & Wilkins
Pain Support Preferences
Solicitude scores. However, the absence of an association between pretreatment and posttreatment Solicitude scores could also reflect a measurement issue, such as poor testretest reliability of that PRPQ scale. The inclusion of a notreatment control group would have enabled an evaluation of this alternative explanation. Second, the sample used was not large, so the statistical power of the study was limited. From one perspective, this suggests that the associations found reflect substantive relationships that could be detected despite limited statistical power. However, the small sample also makes it more likely that substantive relationships were missed (ie, type II error). Thus, additional research with larger samples is required before completely rejecting the notion that pain-related social support preferences do not have an impact on treatment outcomes. Although not necessarily a limitation, it should be noted that this study was a naturalistic investigation of an existing treatment program that was not specifically aimed at changing participants’ pain-related support preferences. As a result, the change in support preferences observed was minimal. A more direct effort to influence pain-related support preferences would be required to thoroughly examine whether changes in such preferences have the potential to facilitate positive treatment outcomes. Given the potentially important role that patients’ painrelated support preferences may play in shaping the social context of their pain, it is important for future research to examine the changes in Solicitude scale scores found in the present study. A slight majority of participants reported a decreased interest in solicitous support. This may reflect a move toward increased independence that could have been triggered by self-management components of the Pain 101 program. The finding that a sizable minority of the subsample reported an increased desire for solicitous support was unexpected. However, making lifestyle changes, such as those promoted in the Pain 101 program, can be challenging. As a result of such challenges, some participants may have developed a heightened awareness of the value of emotional and instrumental support for facilitating the successful implementation of specific self-management strategies. For example, some participants may have decided that they wanted additional support so that they could apply pacing strategies and also be confident that necessary household tasks would be completed. Additional process-focused research including multiple assessments of pain-related support preferences would be required to evaluate whether specific components of the Pain 101 program are associated with particular patterns of changes in pain-related support preferences. More broadly, previous research6 and the current findings indicate that the constructs captured by the PRPQ are related to chronic pain patients’ reports of disability. Thus, additional research aimed at explaining these relationships is warranted.
REFERENCES 1. Sullivan MJL. The communal coping model of pain catastrophizing: clinical and research implications. Can Psychol. 2012;53:32–41. 2. Cano A, Barterian JA, Heller JB. Empathic and nonempathic interaction in chronic pain couples. Clin J Pain. 2008;24: 678–684. 3. Kowal J, Wilson KG, McWilliams LA, et al. Self-perceived burden in chronic pain: relevance, prevalence, and predictors. Pain. 2012;153:1735–1741.
www.clinicalpain.com |
677
McWilliams et al
4. Newton-John T, Williams DC. Chronic pain couples: perceived marital interactions and pain behaviours. Pain. 2006; 123:53–63. 5. McWilliams LA, Saldanha KM, Dick BD, et al. Development and psychometric evaluation of a new measure of pain-related support preferences: the Pain Response Preference Questionnaire. Pain Res Manag. 2009;14:461–469. 6. McWilliams LA, Dick BD, Bailey K, et al. A psychometric evaluation of the Pain Response Preference Questionnaire in a chronic pain patient sample. Health Psychol. 2012;31:343–351. 7. Pollard CA. Preliminary validity study of the Pain Disability Index. Percept Mot Skills. 1984;59:974. 8. Cohen J. A power primer. Psychol Bull. 1992;112:155–159.
678 | www.clinicalpain.com
Clin J Pain
Volume 30, Number 8, August 2014
9. Aiken LS, West SG. Multiple Regression: Testing and Interpreting Interactions. Thousand Oaks, CA: Sage Publications Inc; 1991. 10. Preacher KJ, Curran PJ, Bauer DJ. Computational tools for probing interaction effects in multiple linear regression, multilevel modeling, and latent curve analysis. J Educ Behav Stat. 2006;31:437–448. 11. Romano JM, Turner JA, Jensen MP, et al. Chronic pain patient-spouse behavioral interactions predict patient disability. Pain. 1995;63:353–360. 12. Hasenbring MI, Verbunt JA. Fear-avoidance and endurancerelated responses to pain: new models of behavior and their consequences for clinical practice. Clin J Pain. 2010;26:747–753.
r
2013 Lippincott Williams & Wilkins
A Longitudinal Investigation of Pain-related Social Support Preferences in a Chronic Pain Treatment Sample Lachlan A. McWilliams, PhD,* Kristen S. Higgins, BScH,w Bruce D. Dick, PhD,z and Michelle J. Verrier, BAz
Objectives: Chronic pain patients’ pain-related social support preferences have received little clinical or research attention. However, recent research utilizing the Pain Response Preference Questionnaire (PRPQ) has indicated that such preferences are related to pain-related disability. This study investigated whether pain-related social support preferences: (1) were related to disability levels, (2) predicted changes in pain and disability following a group-based treatment program, and (3) changed from pretreatment to posttreatment. Methods: At pretreatment and posttreatment, chronic pain patients (N = 45) participating in a self-management psychoeducation program completed the PRPQ and measures of pain severity and disability. Results: A multiple regression analysis indicated that at pretreatment the Solicitude scale of the PRPQ was positively associated with disability, and that this association was moderated by the other 2 PRPQ scales (Activity Direction and Suppression). The PRPQ scales did not predict changes in pain or disability and there were no significant changes in the PRPQ scale scores. However, participants did not respond to the Solicitude scale in a consistent manner from pretreatment to posttreatment (51.1% decreased, 35.5% increased, and 13.3% remained the same). These changes were unrelated to changes in pain or disability. Discussion: Additional research aimed at identifying the processes underlying the association between pain-related support preferences and disability is warranted. Given the potentially important role that a desire for solicitous support may play in shaping the social context of pain, understanding the unexpected changes in Solicitude scale scores found in the present study may be particularly important for improving self-management treatments for chronic pain. Key Words: chronic pain treatment, social support, solicitude, activity direction, suppression
(Clin J Pain 2014;30:672–678)
T
here has been a growing research interest in a wide range of interpersonal topics, such as communal coping,1 empathy,2 and self-perceived burden,3 that are relevant to the experience of chronic pain. Given this trend, it is
Received for publication February 7, 2013; revised May 19, 2014; accepted August 13, 2013. From the *Department of Psychology, University of Saskatchewan, Saskatoon, SK; wDepartment of Psychology and Neuroscience, Dalhousie University, Halifax, NS; and zDepartment of Anesthesiology and Pain Medicine, University of Alberta, Edmonton, AB, Canada. The authors declare no conflict of interest. Reprints: Lachlan A. McWilliams, PhD, Department of Psychology, University of Saskatchewan, 9 Campus Drive, 154 Arts, Saskatoon, SK, Canada S7N 5A5 (e-mail: [email protected]). Copyright r 2013 by Lippincott Williams & Wilkins
672 | www.clinicalpain.com
surprising that there has been little research directed toward understanding what forms of social support chronic pain patients find helpful. Newton-John and Williams’4 qualitative research on pain patients and their spouses was the first to address this topic. Through interviews conducted with the spouses, they identified a wide range of possible responses to pain behavior. Interviews with the pain patients revealed that the most positively perceived responses were those involving active coping, such as encouraging task persistence and problem solving. Responses involving solicitous behaviors (viz., providing and offering help) were rated less favorably, with many patients reporting that this type of support made them feel guilty, useless, and burdensome. To study pain-related support preferences further, McWilliams et al5 developed the Pain Response Preference Questionnaire (PRPQ). It assess the degree to which several types of responses to pain are wanted from significant others. The items for this measure were originally designed to capture the forms of support identified by Newton-John and Williams,4 but factor analyses were used to develop empirically derived scales. The most recent of these utilized a chronic pain treatment sample6 and identified scales labeled Solicitude, Activity Direction, and Suppression. McWilliams et al6 used the PRPQ to determine whether pain-related support preferences are related to disability levels. They suggested that pain-support preferences may be related to disability levels because preferences could motivate efforts to obtain specific forms of support. For example, an individual with a strong desire for solicitous support may engage in pain behavior as a method of eliciting such support, and this method of support seeking could result in an elevated level of disability. Alternatively, pain-related support preferences could influence the degree to which an individual is responsive to specific forms of support. For example, solicitous responses to pain behavior would likely not reinforce pain behavior for an individual with a low interest in this form of support. In this case, solicitous support would not be expected to lead to increased pain behavior and disability. McWilliams et al6 found that the correlations between the PRPQ scales and disability ratings were small and nonsignificant. However, a subsequent multiple regression analysis that included sex, pain severity ratings, and all 3 PRPQ scales revealed that each PRPQ scale had a statistically significant association with disability ratings. Solicitude was positively associated with disability, whereas Activity Direction and Suppression were both negatively associated with disability. This pattern of findings indicates that there is a relationship between pain-support preferences and disability, but these relationships are complex rather than simple bivariate relationships. The overall pattern of findings (ie, the PRPQ scales being unrelated to Clin J Pain
Volume 30, Number 8, August 2014
Clin J Pain
Volume 30, Number 8, August 2014
disability in the correlations and all of them being related to disability when they were simultaneously entered in the regression model) suggests that they can be considered suppressor variables for each other (ie, they increase the predictive validity of the other variables when included within the same regression model). The first aim of the present study was to attempt to replicate earlier findings of relationships between pain-support preferences (ie, the PRPQ scales) and disability. Given the complex relationships between the PRPQ scales and disability in past research,6 the current study also utilized a multivariate approach (ie, multiple regression analysis) to explicate such relationships by considering the possibility that the PRPQ scales are suppressor variables for each other and by exploring interactions between the PRPQ scales. The second aim of the study was to investigate whether pain-support preferences predict chronic pain patients’ responses to a group-based psychoeducation program that emphasized cognitive-behavior therapy principles and selfmanagement strategies for coping effectively with chronic pain. Those with a strong desire for solicitous support may be reluctant to adopt a self-management approach. Therefore, it was hypothesized that pretreatment levels of Solicitude would be associated with poorer responses to treatment. Conversely, those wanting encouragement to engage in activities might be particularly responsive to a self-management approach. Thus, it was also hypothesized that pretreatment levels of Activity Direction would be associated with stronger responses to treatment. The third aim of the study was to investigate whether pain-related support preferences changed after the completion of the treatment program. As the self-management program encourages participants to independently practice a variety of strategies to manage their pain and increase their engagement in meaningful activities, it was reasonable to anticipate that participants would experience a reduced desire for solicitous support at posttreatment. However, the self-management program utilized in the present study did not directly address the topic of pain support preferences or include content advocating for specific changes in support preferences, so it was also reasonable to anticipate that participants’ pain-related support preferences would remain relatively stable. Thus, there were no hypotheses regarding changes in pain-related support preferences. Given that pain-related support preferences have been found to be associated with disability,6 it is important to understand whether self-management programs modify such preferences. Changing pain-related support preferences may be a yet to be identified mechanism by which self-management programs produce decreases in disability. Conversely, the persistence of maladaptive patterns of support preferences may interfere with the adoption of selfmanagement strategies. If this is the case, it would suggest that self-management programs for those with chronic pain could be improved by the inclusion of strategies aimed at addressing maladaptive pain-related support preferences.
METHODS Participants Participants were 45 adults (73.3% female) participating in the Pain 101 program at the Multidisciplinary Pain Centre at the University of Alberta Hospital in Edmonton, Alberta, Canada. Participants ranged in age from 19 to 71 years, with a mean age of 47.38 years (SD = r
2013 Lippincott Williams & Wilkins
Pain Support Preferences
11.25). Most of the participants reported being married or in a common law relationship (68.9%); others were divorced or separated (15.6%), in a dating relationship (4.4%), widowed (2.2%), or single (8.9%). A majority of participants were not currently working (33.3% unemployed, 20.0% on sick leave or medical leave, 11.1% students, and 6.7% retired), but approximately 25% of participants were employed (15.6% employed full time, 13.3% employed part time). The average pain duration of the sample was 12.49 years (SD = 12.41). The sample was highly diverse in terms of the type of pain reported. The most common types were neuropathic pain (n = 16), musculoskeletal pain (n = 6), fibromyalgia (n = 6), migraine (n = 3), and headache (n = 2). None of the conditions experienced by the remainder of the sample (n = 12) were reported more than once.
Procedures Consenting participants were asked to complete a package of questionnaires before the first Pain 101 session. They also completed the same set of self-report measures after the final session. In place of names, codes were assigned to each participant and scores were entered in an encrypted database to assure confidentiality and anonymity. This study was approved by the Human Research Ethics Board at the University of Alberta.
Treatment Program The Pain 101 program is a 10-week group-based psychoeducation treatment for adults with chronic pain. The program is based on a cognitive-behavior therapy model that emphasizes the importance of self-management and living well despite pain, rather than focusing on pain reduction. Each session lasts for 2 hours and provides participants with homework to be completed on a daily basis between the weekly sessions. The homework is based on practical pain management strategies established as a part of a best practice pain management program. Session topics include: Pain Education, Tension and Relaxation, Goal Setting and Pacing, Depression and Managing Negative Thoughts, Sleep and Chronic Pain, Coping with Stress and Anxiety, Communication and Pain Behavior, Mindfulness, Planning for and Dealing with Setbacks, and Life with Chronic Pain. The session that is most relevant to social support preferences is the Pain Behavior and Communication session. This session focuses on helping participants understand what and how they communicate with both verbal and nonverbal messages. A significant goal of that session is to help participants consider what they hope to achieve in their communication with others, including significant others, and the types of messages that are most important to them. Assertiveness training is also included in this session. Many programs present pain behavior in a negative light and work toward eliminating such behavior. The Pain 101 program does not take this approach, but instead aims to help participants to identify and reduce behaviors that are unhelpful and to communicate about their pain more effectively, both verbally and nonverbally. The Pain 101 program does not specifically target social support preference in any of its sessions. However, the social component of the biopsychosocial model upon which this program is based does focus on the effects of chronic pain on relationships and the way that relationship and support factors influence the individual suffering from chronic pain. When participants choose to set goals related to their www.clinicalpain.com |
673
Clin J Pain
McWilliams et al
relationships, it is not uncommon for a discussion of support factors to be included during the process of discussing goals.
Measures Demographic Information Participants were asked to report their age, sex, marital status (single, married or common law, divorced or separated, in a dating relationship, or widow), employment status (employed full time, employed part time, on sick leave or medical leave, unemployed, retired, or student), and location and duration of chronic pain.
PRPQ
The PRPQ5 assesses preferences for different types of pain-related social support from one’s significant other. It presents a range of possible responses to pain that a romantic partner could provide (eg, treat me with extra care and concern), and participants are asked to indicate on a scale from 1 (strongly disagree) to 4 (strongly agree) how much they would like their partner to respond to them in this way when they are in pain. The earliest version of the PRPQ, comprised of 35 items, was used. The scales used in the current study were based on a factor analytic study of the PRPQ conducted with chronic pain patients.6 These included a 10-item Solicitude scale (eg, “treat me with extra care and concern”), a 9-item Activity Direction scale (eg, “tell me I can do things despite pain”), and a 3-item Suppression scale (eg, “help me ignore the pain”). Scores for each scale were created by taking the average score of the items comprising each scale. These scales were found to be internally consistent both at baseline (Solicitude: a = 0.87; Suppression: a = 0.69; Activity Direction: a = 0.87) and at the completion of the Pain 101 program (Solicitude: a = 0.76; Suppression: a = 0.83; Activity Direction: a = 0.80).
Pain Severity Participants rated their pain on an 11-point scale ranging from 0 (no pain) to 10 (worst pain imaginable).
Pain Disability Index (PDI)
The PDI7 assesses the extent to which chronic pain affects a person’s ability to engage in various life activities. The scale is comprised of 7 items which represent the following life activities: family/home responsibilities, recreation, social activity, occupational duties, sexual behavior, self-care, and life-support activity. Participants are asked to rate how much their pain interferes with each activity on a scale from 0 (no disability) to 10 (total disability). Participants’ PDI scores were created by averaging their responses to the 7 items in the scale. The PDI scale was found to be internally consistent in the present study (Pretreatment: a = 0.78; Posttreatment: a = 0.85).
RESULTS Preliminary Analyses Correlations between all of the baseline measures are reported in Table 1. As expected, pain and disability had a statistically significant positive association. On the basis of Cohen’s8 recommendations for describing effect sizes for correlations, this association would be considered moderate (ie, r > 0.30). There were also statistically significant
674 | www.clinicalpain.com
Volume 30, Number 8, August 2014
positive associations between the PRPQ scales. The association between the Activity Direction and Suppression scales was moderate, and the association between the Solicitude and Activity Direction scales was large (ie, r > 0.50).
Predicting Baseline Disability Levels A multiple regression analysis was used to predict PDI scores at baseline. On the basis of Aiken and West’s9 recommendations for testing interaction effects, the continuous variables were centered, and the interaction terms were created with these centered variables. Pain severity ratings and sex were entered in step 1. At this step, only pain severity ratings had a significant association with PDI scores (b = 0.38). The PRPQ scales (Solicitude, Activity Direction, and Suppression) were entered in step 2. When steps 1 and 2 were entered, sex (b = 0.29) and pain severity rating (b = 0.43) had significant relationships with PDI scores. The 2-way interactions between the PRPQ scales were entered in step 3. None of these variables were significantly associated with PDI scores, but sex (b = 0.31) and pain severity ratings (b = 0.41) remained significant. Finally, the 3-way interaction between the PRPQ scales was entered in step 4, and revealed a statistically significant interaction effect. The summary statistics for each model are reported in Table 2, along with the standardized b values for the final model. The final model accounted for 42.0% of the variance in PDI scores. The final model showed that sex, pain ratings, the Solicitude scale, and the 3-way interaction between the PRPQ scales were significant predictors of baseline PDI scores. To explore the 3-way interaction effect involving PRPQ scales, the regression equation from the final model was used to determine plot points. Values of 1 SD above and below the means of each PRPQ scale were plotted. As there was a significant main effect of Solicitude and it had the strongest relationship with disability in past research,6 Solicitude was treated as the focal predictor, whereas Suppression and Activity Direction were treated as moderators. The continuous variables in this analysis were centered, and therefore zeros were entered into the analysis for all of the continuous variables other than the PRPQ scales. Zero was also entered for the categorical sex variable, which represents females. These values are plotted in Figure 1. Preacher et al’s10 procedures were used to determine the statistical significance of the relationships shown in Figure 1. At the high value of Suppression and the low value of Activity Direction, there was a significant positive association between Solicitude and PDI scores (b = 3.65, P = 0.013). At the high values of both Suppression and Activity Direction, there was a marginally significant positive association between Solicitude and PDI scores (b = 1.67, P = 0.097). At the other 2 combinations of high and low scores on Suppression and Activity Direction, the associations between Solicitude and PDI scores were small and not statistically significant. Thus, the overall pattern of findings indicated a positive association between Solicitude and PDI scores (ie, the main effect), and that this relationship was strongest at low levels of Activity Direction and Suppression. Solicitude scores were not related to PDI scores at other combinations of high and low levels of Activity Direction and Suppression.
Longitudinal Analyses Correlations between the pretreatment and posttreatment values of each variable (pain ratings, the PDI, and r
2013 Lippincott Williams & Wilkins
Clin J Pain
Volume 30, Number 8, August 2014
Pain Support Preferences
TABLE 1. Correlations Among Pain Ratings, Pain Disability Index, and PRPQ Scales at Baseline
Measure
1
2
3
4
5
6
Pain rating Pain disability index Solicitude Activity direction Suppression Sex
1
0.38* 1
0.17 0.13 1
0.00 0.13 0.73*** 1
0.04 0.04 0.27 0.45** 1
0.03 0.23 0.15 0.20 0.02 1
*Pr0.05; **Pr0.01; ***Pr0.001 (all 2-tailed).
PRPQ scales) are reported in Table 3. All variables except for the Solicitude scale had statistically significant positive associations between pretreatment and posttreatment values. These relationships were all considered to represent large effect sizes. Paired t tests were conducted to compare the pretreatment and posttreatment means of each variable (Table 3). Pain ratings and PDI scores decreased, and these changes were statistically significant. Cohen’s8 d values were calculated to provide indices of effect size. There was a small effect size regarding pain and disability. There were no statistically significant changes with regard to the PRPQ scales. On the basis of the notion that those with particular pain-related support preferences may be more or less receptive to the treatment provided, regression analyses were conducted to determine whether the baseline PRPQ scales were predictive of treatment outcomes (ie, posttreatment functioning after accounting for pretreatment functioning). Posttreatment PDI scores and pain ratings were used as dependent variables. Given that posttreatment functioning would be expected to be related to baseline levels of the variable of interest, centered pretreatment PDI scores and pain ratings were entered in step 1 of the analyses predicting posttreatment PDI scores and pain ratings, respectively. Sex was also included in step 1. Baseline PRPQ scores were entered in step 2. The 2-way interactions between the PRPQ scales were entered in step 3, and the 3-way interaction between them was entered in step 4. In step 1 of these analyses, baseline PDI scores were positively associated with posttreatment PDI scores (b = 0.65, P < 0.001), and baseline pain ratings were positively associated posttreatment pain ratings (b = 0.56, P < 0.001). In both cases the PRPQ scales and their interaction terms were unrelated to the treatment outcome variables. The longitudinal findings regarding Solicitude were unusual. The null results from the paired-sample t test
indicated that Solicitude scores did not change in a systematic manner (ie, generally increasing or decreasing) and the null findings of the correlation analysis indicated that there was no statistically significant consistency over the course of the study in terms of responding to this measure. Change in Solicitude scale values (calculated by subtracting the posttreatment Solicitude scores from the pretreatment scores) indicated that 23 participants’ Solicitude scores decreased after completing the treatment program, 6 participants’ scores did not change, and 16 participants’ Solicitude scores increased after treatment. The inconsistency in Solicitude scores from pretreatment to posttreatment raised the possibility that the changes in Solicitude may relate to treatment outcomes. Multiple regression analyses were used to investigate whether change in Solicitude predicted posttreatment PDI scores and pain ratings. In the analysis predicting posttreatment PDI scores, sex, pretreatment PDI scores, and pretreatment Solicitude were entered in step 1. Change in Solicitude was entered in step 2. A similar analysis was used to predict posttreatment pain ratings, but pretreatment pain ratings were included in step 1 rather than pretreatment PDI scores. Consistent with earlier analyses, in the models including only step 1, pretreatment PDI scores were positively associated with posttreatment PDI scores and pretreatment pain ratings were positively associated with posttreatment pain ratings. In both cases, the addition of step 2 revealed a nonsignificant positive association between change in Solicitude and the treatment outcome variable.
DISCUSSION The first aim of the present study was to replicate earlier findings of relationships between pain-related support preferences and disability. Consistent with past
TABLE 2. Multiple Regression Analysis Predicting Pain Disability Index Scores at Baseline
Step 1 2 3 4
Predictors
b
F
DR2
Sex Pain rating Solicitude Activity direction Suppression Solicitude activity direction Solicitude suppression Activity direction suppression Solicitude activity direction suppression
0.33* 0.43** 0.58* 0.04 0.12 0.23 0.39 0.26 0.58*
5.17**
0.20**
2.72*
0.06
2.41*
0.09
2.82*
0.07*
Standardized coefficients reported are from the final model. *Pr0.05; **Pr0.01 (all 2-tailed).
r
2013 Lippincott Williams & Wilkins
www.clinicalpain.com |
675
Clin J Pain
McWilliams et al
10 Pain Disability Index score
9 8 7 6 5
AD Low Sup Low
4
AD Low Sup High
3
AD High Sup Low
2
AD High Sup High
1 0 Solicitude Low
Solicitude High
FIGURE 1. Exploration of the 3-way interaction between Solicitude, Activity Direction (AD), and Suppression (Sup) predicting baseline Pain Disability Index scores.
research,6 none of the PRPQ scales had a simple linear association with disability, but significant findings emerged in a multiple regression analysis. In earlier research, Solicitude had a positive association with disability, whereas Activity Direction and Suppression had negative associations with disability. The pattern of findings that emerged in the current study was more complex. In the final model, there was a main effect regarding the Solicitude scale and a 3-way interaction effect involving the 3 PRPQ scales. Thus, at the mean levels of Activity Direction and Suppression, there was a significant positive association between Solicitude and disability. Probing of the 3-way interaction indicated that the relationship between Solicitude and disability ratings was statistically significant at high levels of Suppression and low levels of Activity Direction, but not at the other 3 combinations of high and low responses to the Suppression and Activity Direction scales. Thus, a high desire for Solicitude and Suppression and a low level of interest in Activity Direction is the combination of painrelated support preferences most likely to be associated with heightened disability. Chronic pain patients with this pattern of PRPQ scores would be expected to want those close to them to: (1) draw attention away from their pain (ie, high Suppression), (2) provide a high level of emotional and instrumental support (ie, high Solicitude), and (3) not encourage a high level of activity or persistence with activities despite pain (ie, low Activity Direction). The possibility that some chronic pain patients may both endorse a desire for suppression and solicitous support is counterintuitive. However, it is possible that such
Volume 30, Number 8, August 2014
individuals could want a low level of attention directed toward their pain, and also want a high level of support and attention because of their pain. This is the second study using the PRPQ to demonstrate relationships between pain-related support preferences and self-reports of disability. Thus, additional research regarding pain-related social support preferences is warranted. To date, research has not examined the mechanism (or mechanisms) underlying the associations found. It is possible that disability shapes pain-related support preferences, with more disabled individuals preferring support that involves active assistance from others to compensate for that disability. Alternatively, pain-related preferences could influence the social context of pain in a manner that could exacerbate disability. For example, those who prefer solicitous support may have their pain behavior reinforced more strongly by solicitous responses to such behavior, and that could exacerbate their disability. As well, those with a preference for solicitous support may behave in a manner (eg, increased pain behavior) that is particularly likely to draw solicitous support from others thereby increasing their disability. Research paradigms have been developed for investigating patient-spouse interactions,11 and could be utilized to investigate relationships between pain-support preferences, interaction patterns, and disability levels. Participants’ self-reports indicated that they experienced significantly lower levels of pain and disability at posttreatment than at pretreatment. Given the absence of a control condition in the current study, it is impossible to rule out factors other than the treatment as being responsible for these changes. Nonetheless, the current findings suggest that participants had a positive response to the Pain 101 program. The second aim of the study was to investigate whether pain-related support preferences predicted responses to this group-based psychoeducation program. It was thought that pain-related support preferences would influence the ability of participants to adopt the self-management approach utilized in this program, which would in turn influence their responses to the treatment (ie, posttreatment disability and pain adjusted for pretreatment levels of these variables). Contrary to what was expected, the PRPQ scales were not associated with treatment outcomes. These null finding suggest that the ability to benefit from the treatment program was not compromised by patients’ pretreatment pain-related support preferences. Given that the content of the treatment program largely centered on encouraging individuals to change their own thinking and behavior, issues related to social support and/ or social support preferences may be relatively irrelevant to success in implementing the strategies taught in the
TABLE 3. Correlations and Comparisons Between Pretreatment and Posttreatment Variables (N = 45)
Pretreatment Variables Pain rating Pain disability index Solicitude Activity direction Suppression
Posttreatment
r
M
SD
M
SD
t
P
Cohen’s d
0.56** 0.67** 0.19 0.49** 0.51**
6.26 6.21 3.08 2.84 2.06
1.60 1.63 0.55 0.65 0.64
5.62 5.68 2.98 2.83 2.10
2.03 1.97 0.41 0.51 0.74
2.43* 2.36* 1.08 0.17 0.43
0.019 0.023 0.287 0.866 0.668
0.376 0.362 0.164 0.017 0.059
*Pr0.05; **Pr0.001 (all 2-tailed).
676 | www.clinicalpain.com
r
2013 Lippincott Williams & Wilkins
Clin J Pain
Volume 30, Number 8, August 2014
program. However, in light of the relatively small sample size and the associated limited statistical power, the possibility of a type II error cannot be ruled out. Thus, additional research examining the PRPQ scales as prospective predictors of outcome is certainly warranted. The third aim of the study was to investigate whether pain-related support preferences changed after completion of the treatment program. The differences between the pretreatment and post-treatment scores on Activity Direction and Suppression scales were small and not statistically significant. In addition, participants’ pretreatment scores on these 2 scales had large significant positive associations with their respective posttreatment scores on these scales. This pattern of findings indicates that there was no overall change in Activity Direction and Suppression and that participants’ relative positions on these variables were largely consistent overtime. A different pattern emerged regarding pretreatment and posttreatment scores on the Solicitude scale. As with the other PRPQ scales, a comparison of pretreatment and posttreatment Solicitude scale scores indicated that there was no overall change. However, it is important to note that the pretreatment and posttreatment Solicitude scores were not significantly correlated with each other. This absence of a significant positive association indicates that the relative position of participants on the Solicitude scale changed during the course of the study. A slight majority (51.1%) of participants’ scores indicated a decreased level of endorsement of the Solicitude scale. Given the treatment utilized in the study taught self-management skills, it is not surprising that the majority of participants changed in this direction as it may partially reflect the adoption of selfmanagement approach. Follow-up multiple regression analyses indicated changes in Solicitude scores were unrelated to treatment outcomes. Thus, the findings indicate that a decreased desire for solicitous support was commonly experienced by study participants, and also suggest that this type of change was not a process that led to decreases in pain and disability. It is also noteworthy that a large minority of participants (35.5%) increased their endorsement of the Solicitude scale. Some clinicians might consider an increase of this nature to be a potentially harmful development. However, it is important to recognize the possibility that such changes could be adaptive. Several investigations12 have identified subgroups of chronic pain patients characterized by high levels of thought suppression, the use of distraction, and task persistence despite pain. This rigid pattern of responding to pain is thought to be harmful because it can lead to reinjury. Among this type of patient, an increased interest in solicitous support might reflect a new and adaptive willingness to accept the help and support of others. The data captured in this study did not include variables that would permit us to explore whether this was the case in this population. Future research is needed to explore this possibility. The two main limitations of the current study should be noted. First, the addition of a no-treatment control group would have provided for a more thorough evaluation of the changes in Solicitude scores noted in the present study. It has been assumed that some aspect of the treatment program may have prompted participants to reconsider their interest in solicitous support and that this process could have been responsible for the absence of association between pretreatment and posttreatment r
2013 Lippincott Williams & Wilkins
Pain Support Preferences
Solicitude scores. However, the absence of an association between pretreatment and posttreatment Solicitude scores could also reflect a measurement issue, such as poor testretest reliability of that PRPQ scale. The inclusion of a notreatment control group would have enabled an evaluation of this alternative explanation. Second, the sample used was not large, so the statistical power of the study was limited. From one perspective, this suggests that the associations found reflect substantive relationships that could be detected despite limited statistical power. However, the small sample also makes it more likely that substantive relationships were missed (ie, type II error). Thus, additional research with larger samples is required before completely rejecting the notion that pain-related social support preferences do not have an impact on treatment outcomes. Although not necessarily a limitation, it should be noted that this study was a naturalistic investigation of an existing treatment program that was not specifically aimed at changing participants’ pain-related support preferences. As a result, the change in support preferences observed was minimal. A more direct effort to influence pain-related support preferences would be required to thoroughly examine whether changes in such preferences have the potential to facilitate positive treatment outcomes. Given the potentially important role that patients’ painrelated support preferences may play in shaping the social context of their pain, it is important for future research to examine the changes in Solicitude scale scores found in the present study. A slight majority of participants reported a decreased interest in solicitous support. This may reflect a move toward increased independence that could have been triggered by self-management components of the Pain 101 program. The finding that a sizable minority of the subsample reported an increased desire for solicitous support was unexpected. However, making lifestyle changes, such as those promoted in the Pain 101 program, can be challenging. As a result of such challenges, some participants may have developed a heightened awareness of the value of emotional and instrumental support for facilitating the successful implementation of specific self-management strategies. For example, some participants may have decided that they wanted additional support so that they could apply pacing strategies and also be confident that necessary household tasks would be completed. Additional process-focused research including multiple assessments of pain-related support preferences would be required to evaluate whether specific components of the Pain 101 program are associated with particular patterns of changes in pain-related support preferences. More broadly, previous research6 and the current findings indicate that the constructs captured by the PRPQ are related to chronic pain patients’ reports of disability. Thus, additional research aimed at explaining these relationships is warranted.
REFERENCES 1. Sullivan MJL. The communal coping model of pain catastrophizing: clinical and research implications. Can Psychol. 2012;53:32–41. 2. Cano A, Barterian JA, Heller JB. Empathic and nonempathic interaction in chronic pain couples. Clin J Pain. 2008;24: 678–684. 3. Kowal J, Wilson KG, McWilliams LA, et al. Self-perceived burden in chronic pain: relevance, prevalence, and predictors. Pain. 2012;153:1735–1741.
www.clinicalpain.com |
677
McWilliams et al
4. Newton-John T, Williams DC. Chronic pain couples: perceived marital interactions and pain behaviours. Pain. 2006; 123:53–63. 5. McWilliams LA, Saldanha KM, Dick BD, et al. Development and psychometric evaluation of a new measure of pain-related support preferences: the Pain Response Preference Questionnaire. Pain Res Manag. 2009;14:461–469. 6. McWilliams LA, Dick BD, Bailey K, et al. A psychometric evaluation of the Pain Response Preference Questionnaire in a chronic pain patient sample. Health Psychol. 2012;31:343–351. 7. Pollard CA. Preliminary validity study of the Pain Disability Index. Percept Mot Skills. 1984;59:974. 8. Cohen J. A power primer. Psychol Bull. 1992;112:155–159.
678 | www.clinicalpain.com
Clin J Pain
Volume 30, Number 8, August 2014
9. Aiken LS, West SG. Multiple Regression: Testing and Interpreting Interactions. Thousand Oaks, CA: Sage Publications Inc; 1991. 10. Preacher KJ, Curran PJ, Bauer DJ. Computational tools for probing interaction effects in multiple linear regression, multilevel modeling, and latent curve analysis. J Educ Behav Stat. 2006;31:437–448. 11. Romano JM, Turner JA, Jensen MP, et al. Chronic pain patient-spouse behavioral interactions predict patient disability. Pain. 1995;63:353–360. 12. Hasenbring MI, Verbunt JA. Fear-avoidance and endurancerelated responses to pain: new models of behavior and their consequences for clinical practice. Clin J Pain. 2010;26:747–753.
r
2013 Lippincott Williams & Wilkins
