- Email: [email protected]
Positive affect as a factor of resilience in the pain—negative affect relationship in patients with rheumatoid arthritis
Journal of Psychosomatic Research 60 (2006) 477 – 484
Positive affect as a factor of resilience in the pain—negative affect relationship in patients with rheumatoid arthritis Elin B. Stranda,T, Alex J. Zautrab, Magne Thoresena, Sigrid adeg3rdc, Till Uhligc, Arnstein Finseta a
Department of Behavioural Sciences and Statistics, Institute of Basic Medical Science, University of Oslo, POB 1111 Blindern, N-0317 Oslo, Norway b Department of Psychology, Arizona State University, Tempe, AZ, USA c Department of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway Received 2 May 2005; received in revised form 8 August 2005; accepted 16 August 2005
Abstract Objective: The purpose of this study is to examine positive affect (PA) as a factor of resilience in the relationships between pain and negative affect (NA) in a sample of patients with rheumatoid arthritis. Methods: Forty-three patients (30 women; mean age, 57 years) were interviewed weekly by telephone for 8 weeks. Multilevel modeling was applied to study the within-week relationships among the variables. Results: There was a PainPA interaction effect on NA (b=0.05, Pb.01) indicating a weaker relationship between pain and NA in weeks
with more PA. Pain (b=0.37, Pb.002), interpersonal stress (b=2.42, Pb.001), depression (b=0.26, Pb.01), average perceived stress (b=10.80, Pb.001), and also weekly PA (b=0.1, Pb.01) had a main effect upon NA. Conclusion: Positive affect is most influential in reducing NA during weeks of higher pain and may be a factor of resilience, helping patients experiencing pain fluctuations as less distressful than at lower levels of PA. D 2006 Elsevier Inc. All rights reserved.
Keywords: Negative affect; Pain; Positive affect; Resilience; Rheumatoid arthritis; Stress
Introduction In this paper, we ask whether the stressful impact of chronic pain is lessened by the presence of positive emotions for patients with rheumatoid arthritis (RA). Chronic pain is reported as the most widespread and challenging symptom for patients with RA [1] and a high priority for physician’s attention [2]. Pain in RA is also a potential stressor not only because it is a highly aversive bodily experience but also because the pain intensity and duration fluctuate in a relatively unpredictable and uncontrollable way. Pain varies between individuals and across situations in intensity and duration, and patients differ in the extent that T Corresponding author. Department of Behavioural Sciences and Statistics, Institute of Basic Medical Science, University of Oslo, POB 1111 Blindern, N-0317 Oslo, Norway. Tel.: +47 22 85 10 21; fax: +47 22 85 13 00. E-mail address: [email protected] (E.B. Strand). 0022-3999/06/$ – see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jpsychores.2005.08.010
arthritis pain gives rise to emotional distress. Although there is a well-established association between pain and negative affect (NA) [3,4], both interpersonal stress and also depression play a role in a patient’s vulnerability to pain [5–11]. For RA patients, negative affective responses to pain may influence illness course, increasing the frequency of painful flares, lowering pain thresholds, intensifying pain behaviors, and deteriorating coping [12–19]. Thus, identifying factors that may diminish the established connection between pain and NA may be of considerable value to the health as well as the mental health of RA patients. A shift in focus in pain research to explore an RA patient’s capacities for resilience as well as their vulnerabilities to pain appears warranted at this juncture. The concept of resilience refers to the person’s ability to bounce back from negative emotional experiences and show a flexible adaptation to the changing demands of stressful experiences [20]. These attributes are of considerable importance for sustaining health and well being [21,22]. Positive emotions may be
478
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
essential factors in the process of resilience following adverse events [23–27], including adverse experiences with chronic pain [28]. Tugade and Fredrickson [29] found that highresilient individuals tended to report positive emotions even when under stress, and that these positive emotions contributed to recovery from stress-related negative effects. Their bBroaden and BuildQ theory of positive emotions [30] posits that persons with higher positive affect (PA) have greater capacity to recover psychologically and physiologically to stressful events. Positive emotions may expand the range of cognitions and behaviors to build an individual’s physical, intellectual, and social resources [29–31]. The role of PA in persons with chronic pain has been examined by Zautra et al. [28,32]. In their dynamic model of affect (DMA), they suggest that the relationship between PA and NA changes as a function of context [33–35]. Dynamic model of affect posits that stressful events change the degree of independence between the affective states so they become less differentiated, that is, more bipolar. Thus, according to the DMA, people who can sustain higher levels of PA at the time of the stressor would show significantly less NA [35]. In one longitudinal study of arthritis (RA) patients, they found that the presence of positive affective states reduced the size of the relationship between the patient’s reported weekly baverage painQ and NA [32]. Weekly registrations of pain and affect were employed for a period of 12 to 20 weeks. In this study, they examined the role of PA in the relationship between pain and NA, and with mood clarity as a confounder. So far, no study other than Zautra et al’s [32] has explored this relationship with a combined within- and between-person design, and advanced statistical methods such as multilevel modeling. Stress is highly related to the environmental context— especially interpersonal events [25]. For patients with RA, stress in interpersonal relationship is associated to increased disease activity, depression, and pain [8,17]. In this group of patients, the PA–interpersonal stress interaction on NA has not, as far as we know, been examined. In the current study, we also explore interpersonal stress, that is, the patient’s perceived stress in friends, family, and spouse/partner relationship. The current study is a replication of Zautra et al’s study [32] only in so far as it tests the relationships among pain, PA, and NA. We have also added tests of the role of interpersonal stress, which was not included in the prior study. We also studied bworst painQ during the past week, which was another departure from prior research that relied on reports of last week’s average pain. The bipolarity of the affects and the resilience factor are both stress-related phenomenon. Bipolarity in affects is according to DMA, a consequence of stressful events. For resilience to display, there need to be aversive states to bounce back from, and because the worst pain more closely identifies a stressful event, we rely on this pain rating in our study. To date, the role of PA as a source of resilience has not been studied in patients with RA. This study aims to bridge
that gap and expand our understanding of the affect interrelationships in RA. Norms for experiencing and expressing emotions differ widely between countries, even between countries that on most dimensions may appear similar and share many sociocultural features [36], such as Norway and the United States. It is therefore valuable to test the effects of PA on NA among RA patients who reside in these countries other than the United States. In this paper, we examine data on a sample of Norwegian patients with RA on the association between pain fluctuations, elevations in interpersonal stress, and NAs. We also address the question of how PA influences these relationships. Finally, we explore individual differences in depression and perceived interpersonal stress as vulnerability factors in the experiences of NA during chronic pain.
Method Subjects The sample consisted of 43 patients with RA included in a 10-year follow-up of the Norwegian EURIDISS cohort (European Research in Incapacitating Diseases and Social Support [11]). At entry into the cohort 10 years prior, patients had been diagnosed with RA within the last 4 years. They were asked to take part in the current study when they came to the hospital for the 10-year follow-up examination. Of the 238 patients originally included in the EURIDISS study at T1 (1992), 35 patients had died. Forty-two individuals refused to take part in the follow-up, and 12 patients did not take part in the follow-up for other reasons (could not be located, had moved, etc.). Thus, 149 (63% of 238) patients took part in the 10-year follow-up. Of these 149 patients, 43 participated in the current study. The present sample consisted of 30 (70%) woman and 13 (30%) men. The mean age was 57.5 (S.D.=13.1) with a range from 33 to 80 years. Of the 43 patients, 26 (60%) were married or living with a partner, 33 (75%) had one or more children, and 10 (23%) were in a full-time job, 5 (12%) had a part-time job, and the rest [28 (65%)] were on age or disability pension. A comparison of this sample to the EURIDISS cohort on age, education, sex, Steinbrocker (physical function), Physician global clinical evaluation (VAS), SF-36, Nottingham Health Profile, HAQ, and GHQ revealed significant differences only in age and education. The current sample was younger and had longer education than the EURIDISS cohort. Overall, the current sample was functioning at a relatively high level for RA patients. Of the 43 patients, 80% had a Steinbrocker score (a global measure of function) of 2 on a scale ranging from 1, which is no impairment, to 4, which signifies extensive handicap such as using a wheelchair or staying in bed. None of the RA patients had a score on IV.
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
Assessment The patients were invited to the hospital for an initial interview. They were given the first standardized interview and asked to complete baseline questionnaires. The participants were then given weekly, standardized telephone interviews during an 8-week period. In these interviews, they were asked to give a report about last week’s pain, PA, NA, as well as interpersonal stress. A total of 336 weeks of data were obtained, with only 8 weeks of missing data, for a 98.5% completion rate. The missing weeks were distributed on eight different persons: five women and three men. Two of the missing weeks were in the person’s interview week number 3, one in the interview week number 5, another in the interview week number 7, and the other three in the interview week number 8. Two of the missing weeks were due to the interviewer himself, another two were due to hospital stay, forgetting (two), exhaustion (one), and death in the family (one). Most of these missing weeks are to be considered random missing. The proportion of missing weeks is small, and even smaller are the missing weeks due to worsening of health or illness fluctuations (n=3). No one of the patients mentioned pain, PA, NA, interpersonal stress, or depression, all important variables in the study, as reasons for not being able to undertake the telephone interview. Based on this analysis, we do not consider neither the number nor the reasons for the missing data to be of any serious threat to the integrity of the data analytic method. Pain During the phone interview, the participants were asked to rate last week’s pain due to their RA on an 11-point numeric rating scale with points from 0 (no pain) to 10 (worst possible pain) [37]. They separately rated three aspects of the pain: the most intense, the least intense, and the average pain level, over the past week. We provide data on each of the three pain ratings but rely on the rating of most intense pain (MIP) in the test of the effects of PA because this measure most closely identifies weeks when the person’s pain is the most stressful. Research also has shown that peak intensity of pain is less likely to be biased by retrospective recall [47]. Positive affect and negative affect To measure affect weekly, we used Positive and Negative Affect Schedule, which is an established instrument developed to tap the two major dimensions of mood— positive and negative affectivity [38]. One of the distinguishing aspects of the scale’s development was its emphasis on indices of PA and NA that were independent. We used Watson et al.’s own categorizing for PA and NA with 10 items in each affect category. Participants were asked to indicate on a five-point scale, bto what extent did
479
you experience each of the adjectives for positive and for NA during the previous week.Q The five-point scale is labeled 1, very slightly/not at all; 2, a little; 3, moderately; 4, quite a bit; and 5, very much. The adjectives for PA were binterested,Q bstrong,Q binspired,Q battentive,Q benthusiastic,Q bproud,Q balert,Q blively,Q bactive,Q and bdetermined,Q and for NA, the adjectives were bdistressed,Q bupset,Q bnervous,Q bscared,Q bhostile,Q birritable,Q bashamed,Q bjittery,Q bafraid,Q and bguilty.Q PANAS was translated from English into Norwegian and then translated back into Norwegian by a person with English as first language. After discussions among authors on divergences between the versions, the final solutions were used in the study. The translations of PANAS were executed specific for this study, and therefore, there are no other validation of the instruments to compare with. Watson et al. [38] assert that internal consistencies for both scales have been ranging from .86 to .90 for PA and from .84 to .87 for NA in their studies. In the current study, the Cronbach’s a was .90 for NA and .83 for PA. In Watson et al’s studies, the correlation between the NA and PA scales has been invariably low, ranging from .12 to .23. In our study, the correlation between NA and PA was .11 for the weekly scores and .25 for the average affect scores. Interpersonal stress The patients were asked to indicate on a four-point scale the extent interpersonal stress arising over the past week in each of the following areas: friends, family, and spouse/ partner. bOverall, how stressful were your relations with friends (family, partner) this past weekQ (zero=not stressful at all, 1=a little stressful, 2=moderately stressful, and 3=extremely stressed). Each of these questions followed probes regarding number of stressful events in each of the three life domains taken from the Inventory of Small Life Events [39]. Two interpersonal stress variables were computed. An average interpersonal stress variable across the 8 weeks was computed for each participant based on the mean of three weekly ratings: perceived stress with friends, family, and spouse/partner (if present). Another weekly interpersonal stress score was calculated by subtracting each individual’s weekly score from his/her average score. This person-centered method of creating the weekly deviation scores was also used for all the weekly variables used in this paper. Depression Beck Depression Inventory (BDI) was given at the initial interview. Beck Depression Inventory is a widely used instrument to measure depression and consists of 21 items that assess cognitive–affective and vegetative signs of depression. The BDI has been used widely to assess depressive symptoms in psychiatric and nonpsychiatric populations [40].
480
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
Table 1 Means, S.D., and correlation coefficients of predictor variables in the multilevel analyses Variable
n
Mean
S.D.
Range
Criterion NA (raw score)
333
13.8
4.67
23
0.00 0.00 0.00
1.52 4.46 0.27
4.75 31.00 0.22 8.34
2.32 6.00 0.23 4.20
Within-person deviation scores Weekly MIP 336 Weekly PA 328 Weekly IS 333 Between-person variables Average MIP Average PA Average IS Depression
43 43 43 41
NA
MIP
PA
11.3 34.0 2.31
0.12T 0.11 0.15TT
0.06 0.004
0.04
9.50 31.5 0.88 20.0
0.16 0.25 0.52TT 0.32T
0.37T 0.41TT 0.06
0.08 0.35T
IS
0.10
n refers to the number of observations. The scores on the weekly variables are centered by subtracting the observed score from each person’s average across the eight weekly observations. T Pb.05 level (two tailed). TT Pb.01 (two tailed).
Statistical procedure We wanted to investigate the relationship between NA and pain, PA, interpersonal stress (IS), and depression, measured at eight weekly time points, adjusting for individual characteristics of the patients. The assessments provided two levels of data: level 1 consisted of the variables that varied within participants across weeks; level 2 variables identified individual differences that varied between participants but not over weeks. We analyzed the multilevel model with SAS PROC MIXED software (e.g., Ref. [41]). For the level 1 data (the relationship among the variables measured weekly), the equation may be stated as follows: NAij ¼ c0j þ c1j MIPij þ c2j PAij þ c3j IS ij þ c4j PAij MIP þ c5j PAij IS þ c6j week1ij þ eij where the subscripts i and j represent week number and patient number, respectively. The variable bweek1Q indicates whether the measurement is made in the first week or not, and inclusion of this
variable controls for elevations in level commonly found at the first weekly measurement compared with subsequent weekly assessments. The independent variables weekly MIP, weekly PA, and weekly IS are centered at the mean value for each person. These centered variables identify weekly fluctuations in level of pain, PA, and interpersonal stress, independent of the person’s average level across the 8 weeks. To test how PA modulates the relationships between pain, interpersonal stress, and NA, we included interaction terms shown in the equation. The level 2 equation models the between-subject variation in level 1 intercepts: k0j ¼ b00 þ b01 BDI j þ b02 AvgIS j þ b03 AvgPAj þ b04 AvgMIPj þ u0j : This level of analysis is used to explore the extent to which NA was affected by individual differences in depression, average perceived interpersonal stress (AvgIS), average PA (AvgPA), and average MIP (AvgMIP). In addition, separate equations for the interactions between levels 2 and 1 examined individual differences in PA on the
Fig. 1. Interaction of PA on weekly MIP on NA.
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484 Table 2 Multilevel regression predicting weekly NA Random effects Covariance parameter estimated
Subject
Un (1,1) Ar (1) Residuals
ID ID ID
b
S.E. 4.52 0.23 8.57
Z
1.56 0.09 0.95
P 2.89 2.60 8.98
.001 .009 .0001
Fixed effects Predictor variables Level 1 within-person variables Weekly IS Weekly MIP Weekly PA Week1 Weekly PAT Weekly MIP Level 2 between-person variables Average PA Average MIP Average IS Depression
b
S.E.
t
P
2.42 0.37 0.103 1.97 0.05
0.58 0.12 0.04 0.49 0.02
4.17 3.19 2.88 3.98 2.37
b.0001 .002 .004 b.0001 .019
0.11 0.02 10.80 0.26
0.08 0.22 1.93 0.10
1.40 0.11 5.61 2.52
.172 .917 b.0001 .017
For the level 1 analyses, we only report the significant results. n=336 weekly measurements for within-person analyses and n=43 for betweenperson analyses.
weekly pain–weekly NA and on the weekly IS–weekly NA relationships (the last one not shown): MIP1 ¼ c10 þ c11 BDI þ c12 AvgIS þ c13 AvgPA þ c14 AvgMIP þ u1 : In the final model, only the intercept (g0j ) was allowed to vary among patients and was modeled as a random effect. Because of the relatively small number of subjects, we decided not to attempt to estimate random effects of the variables in the model. Further, we modeled the dependence between repeated measurements of NA within patients as an autoregressive process of order 1. The dependent variable looked somewhat skewed. One important assumption behind linear multilevel model is that the residuals are normally distributed [41,42]. The estimated residuals for normality were examined and found satisfactory.
481
interpersonal stress (r=.41, Pb.01). There was also a moderate negative correlation between PA and depression (r=.35, Pb.05). There was no correlation between MIP and depression, average PA and average IS, and IS and depression. These last set of correlations show that our assessments of PA, pain and stress yielded individual differences on key variables that were distinct from depression and also different from one another. To analyze the effects of weekly MIP, weekly PA, weekly IS, and individual differences in depression and interpersonal perceived stress on NA, we used multilevel modeling as outlined earlier. On level 1, weekly MIP had a strong effect upon NA (b=0.37, Pb.0002). On weeks with more pain, NA was significantly higher. This effect was moderated, however, by weekly PA. Weeks of a higher PA moderated the effect of pain on NA. This relationship is illustrated in Fig. 1. Further, there was a significant effect of weekly PA on NA (Table 2). Weekly interpersonal stress was also associated with NA (b=2.42, Pb.0001), but there was no interaction effect between PA and IS in influencing NA. We also executed the same analyses for both weekly average pain and weekly lowest pain level. The same significant result on NA were found for these pain variables as with the MIP level except from the interaction with PA, which was nonsignificant for both the average and lowest pain level. When examining level 2 variables for between-person effects, we found that individual differences, both the average interpersonal stress (b=10.80, Pb.0001) as well as depression (b=0.26, Pb.017), significantly increased NA. In the final model, we included only those variables that were significant. This model (illustrated in Fig. 2) can be formulated as a so-called mixed model: NAij ¼ b0 þ b1 MIPij þ b2 IS ij þ b3 PAij þ b4 PAij MIPij þ b5 BDI j þ b6 AvgIS j þ b7 week1 þ uj þ eij ; where the random effect u j reflects the between-person variability in the intercept.
Results Initial analyses provided descriptive statistics and correlations among study variables. Means, S.D.s, and correlations between NA and the predictor variables are shown in Table 1. Negative affect was correlated with weekly MIP (r=.12, Pb.05) and IS (r=.15, Pb.01), but not with PA. Negative affect was, correlated with between-person differences in IS (r=.52, Pb.01), and depression (r=.32, Pb.01), and not with PA. Among the between-person variables, there was a moderate negative correlation between PA and MIP (r=.37, Pb.05), and MIP was also associated with more
Fig. 2. Final model of associations based on multiple regression analyses.
482
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
Discussion In this study, we found a significant association between weekly pain, weekly stress, and weekly NA. Moreover, we found a weekly PainPA interaction effect, indicating that PA moderated the effects of pain on NA such that higher levels of PA reduced NA during weeks of high pain. Weekly PA also had a direct effect on NA in significantly reducing it. Depression and average perceived stress across weeks were also associated with weekly NA and served as vulnerability factors in relation to NA. Weekly PA did not influence the IS–NA relationship. The average level of PA across the 8 weeks was unrelated to NA directly and also did not influence neither the effects of pain nor the effects of interpersonal stress on NA. Our findings are similar to those of Zautra et al. [32] in their study of a sample of RA and osteoarthritis patients in the United States. In both studies, weekly PA (bstateQ), but not mean PA across weeks (btraitQ), was inversely related to weekly NA, and PA seemed to reduce the effect of pain on NA in weeks of higher pain. Both individual differences in IS, MIP, and depression were in the current sample correlated to NA, where IS had the strongest correlation. In the regression analyses, both average IS and depression were significantly associated to NA, whereas average MIP was not. In the United States study, individual differences in pain across weeks explained a significant proportion of the variance in weekly NA. In the main model of our study (Table 2), between-subject differences in pain across weeks were unrelated to NA. In contrast to the United States study, this study included interpersonal stress in the model. If interpersonal stress was removed from the equation, pain across weeks made a significant contribution to NA similar to what Zautra et al. [32] found. Our findings indicate that both weekly pain and weekly interpersonal stress function as sources of NA, whereas the stable aspect of pain as a stressor (at least in our sample) covaried with interpersonal stress, as a strong predictor of NA. In the United States study, the authors used the patients’ average pain as pain measurement; the Norwegian study used the bMIP.Q The pain level seems to be somewhat lower in the present study (MIP, 4.7, and average pain level, 3.5, on a scale from 0 to 10) than in the United States samples (for RA patients, average pain level was 43 on a scale from 0 to 100); still, we found the same effect of PA on the pain–NA relationship. The differences between the studies appear best explained by the measures of pain used than by cultural differences, although we do not have data that have a direct bearing on that question. Our findings on the pain–affect relationship are in accordance with the DMA postulating a more bipolar relationship between PA and NA during the stress of increased pain fluctuations. In our data, higher average PA did not protect the RA participants from NA associated with weekly pain. Positive affect was most important as a deterrent when pain was at its worst. Thus, the state
dimension of PA may be thought of as a source of resilience during aversive states like with increasing pain. The resilient effect of PA in the pain–NA relation may result from the rise in the person’s well-being, and also through shifts in their cognitive appraisals of self-efficacy and other reframing of the pain and their own coping efforts. More research is needed to identify the mechanisms by which PA is protective during painful episodes. Of interest in particular is the extent to which this influence of PA is due to a narrowing of affective differentiation as proposed by Zautra [35] and/or a boost in affective resources as proposed by Fredrickson [30]. Rheumatoid arthritis participants with more depression and interpersonally stressed overall reported more NA, and also, weeks of high IS were associated with elevations in NA. These findings are in consistence with lots of the research initially mentioned. However, weekly PA did not influence the interaction between weekly interpersonal stress and NA as expected from the DMA. This negative finding may be due to a small sample and a rather low level of reported stress, which means that the stress-related NA is not strong enough to provoke bipolar affective functioning. If this is the case, then the nonsignificant results on the stress–PA interaction here are not inconsistent with the DMA model that we tested. These results may also indicate that stress and NA are two aspects of the same phenomena employed in two different ways through affect rating and questions about perceived stress. The results imply that they may be linked in such a way that not even an elevation in PA can separate them. That PA interacts differently in the pain–NA than in the stress–NA relationship may also indicate that pain and stress are two very different sources of NA and, therefore, demand different sources for bhealing,Q that is, PA may be a better source of resilience for NA caused by pain but not for NA caused by interpersonal stress. This is quite speculative, and the two sources of NA such as pain and interpersonal stress should be further explored. Both PA and NA mean scores in the present study are similar to those seen in other samples both clinical and nonclinical, using PANAS as affect measures [32,43,44]. Whether our results on the affect relationships would generalize to other populations of chronically ill awaits further study. Limitations The patients in our study were recruited from a lager sample of RA patient, and one-third took part in our study. Nonparticipation may be due to several reasons. Given the difference in age and education between the total and our sample, we presume that when presented our follow-up study, it seemed too demanding to the oldest and those with the lowest education. In summary, the sample that completed our study compares favorably with the larger sample from which they were drawn, which
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
gives us some confidence that the group is not overly biased in health and/or well-being. Our sample may still be more resilient than a random sample of the RA patients. It is noteworthy, however, that they did report fairly substantial levels of joint pain in our study and levels of disability comparable to the larger cohort from which they were drawn. We used a once-a-week measurement often used in research, and by consequence, the information is given retrospectively. Studies comparing patient’s evaluations of both momentary and recalled pain have shown that patients accurately recalled the severity of a pain episode for at least 1 week [45] and tended to recall the peak level and the most recent level [46,47]. Thus, the MIP level as used in our study appears to be relatively free of recall bias; nevertheless, some caution within interpretation of the finding is called for given the 7-day time span between assessments. The affective state at the time of a painful experience may shape the nature of the experiences and what is later recalled [48]. For example, NA at recall, may influence an individual’s report of the prior week’s pain and interpersonal stress. For example, depressed persons recall more pain than those who are not in such a mood [49,50]. Still, an important question left unresolved is whether the findings that infer same-time associations refer to influences over the full week, or are best thought of as associations within a shorter time span than that which was measured. The DMA model that was tested makes no assumptions regarding the proper length of time that a same-time association may be observed. Further studies are needed to gauge the time interval within which changes in the nature of the relationships among affective states are most likely to occur. The weekly variables involved in the study may fluctuate frequently; both pain and affect may change several times during a week and even within 1 day. Daily and even within-day assessments are to be encouraged in future studies of this kind to rule out potential effects of biases in retrospective accounts. The main finding from our study is that PA appears to be an important resilience factor protecting patients with RA during pain fluctuations. This potential of the positive affective resources may encourage new approaches to improving the quality of life of patients with chronic pain. Interesting questions remain regarding the differences between persons in their capacity to mobilize and experience positive emotional states when pain increases, and to what degree the person’s social life and interpersonal relations influence in this process. In our analyses, we controlled for individual differences in depression and interpersonal stress to rule out potential confounds to NA and PA. Findings from this study invite further exploration of how differences in individual capacities in emotional awareness and regulation, and the person’s social resources and interpersonal relations influence adaptation to chronic pain conditions such as RA.
483
Acknowledgments This research was funded by the Research Council of Norway, grant 147831/320. References [1] Katz PP. The stresses of rheumatoid arthritis: appraisals of perceived impact and coping efficacy. Arthritis Care Res 1998;11:9 – 22. [2] Heiberg T, Kvien TK. Preferences for improved health examined in 1024 patients with rheumatoid arthritis. Pain has highest priority. Arthritis Rheum 2002;15:391 – 7. [3] Fernandez E, Milburn TW. Sensory and affective predictors of overall pain and emotions associated with affective pain. Clin J Pain 1994;10: 3 – 9. [4] Fernandez E, Turk DC. The scope and significance of anger in the experience of chronic pain. Pain 1995;61:165 – 75. [5] Banks SM, Kerns RD. Explaining high rates of depression and pain: does the evidence support a relationship? Psychol Bull 1996;97:18 – 34. [6] Blumer D, Heilbronn M. Chronic pain as a variant of depressive diseases. The pain-prone disorder. J Nerv Ment Dis 1982;170:381 – 406. [7] Engel GL. bPsychogenicQ pain and the pain-prone patient. Am J Psychiatry 1959;137:899–918. [8] Kerns RD, Rosenberg R, Otis JD. Self-appraisal problem solving and pain-relevant social support as predictors of the experience of chronic pain. Ann Behav Med 2002;24:100 – 5. [9] Leino P, Magni G. Depressive and distress symptoms as predictors of low back pain, neck–shoulder pain, and other musculoskeletal morbidity: a 10-year follow-up of mental industry employees. Pain 1993;53:89 – 94. [10] Magni G, Moreschi C, Rigatti-Luchini S, Merskey H. Prospective study on the relationship between depressive symptoms and chronic musculoskeletal pain. Pain 1994;56:289 – 97. [11] Smestad LM, Vaglum P, Moum T, Kvien T. The relationship between psychological distress and traditional clinical variables: a 2 year prospective study of 216 patients with early rheumatoid arthritis. Br J Rheumatol 1997;36:1 – 8. [12] Affleck G, Tennen H, Urrows S, Higgins P. Neuroticism and the pain– mood relation in rheumatoid arthritis: insights from a prospective daily study. J Consult Clin Psychol 1992;60:119 – 26. [13] Bradley LA, Alberts KR. Psychological and behavioral approaches to pain management for patients with rheumatic disease. Rheum Dis Clin North Am 1999;25:215 – 332. [14] Janssen SA. Negative affect and sensitization to pain. Scand J Psychol 2002;43:131 – 7. [15] Keefe FJ, Bonk V. Psychosocial assessment of pain in patients having rheumatic diseases. Rheum Dis Clin North Am 1999;25:81 – 103. [16] Lefebvre JC, Keefe FJ, Affleck G, Raezer LB, Starr K, Caldwell DS. The relationship of arthritis self-efficacy to daily pain, daily mood, and daily pain coping in rheumatoid arthritis patients. Pain 1999;80: 425 – 35. [17] Manne S, Zautra A. Spouse criticism and support: their association with coping and adjustment to rheumatoid arthritis. J Pers Soc Psychol 1989;56:608 – 17. [18] Schwartz L, Slater MA, Bircher GR. Interpersonal stress and pain behaviors in patients with chronic pain. J Consult Clin Psychol 1994; 62:861 – 4. [19] Turk DC, Okifuji A. Evaluating the role of physical, operant, cognitive, and affective factors in the pain behaviors of chronic pain patients. Behav Modif 1997;21:259 – 80. [20] Block J, Kremen AM. IQ and ego-resiliency: conceptual and empirical connections and separateness. J Pers Soc Psychol 1996; 70:349 – 61. [21] Jacelon CS. The trait and process of resilience. J Adv Nurs 1997;25: 123 – 9.
484
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
[22] Klohnen EC. Conceptual analysis and measurement of the construct of ego-resiliency. J Pers Soc Psychol 1996;70:1067 – 79. [23] Folkman S, Moskowitz JT. Positive affect and the other side of coping. Am Psychol 2000;55:647 – 54. [24] Fredrickson BL, Tugade MM, Waugh CE, Larkin GR. What good are positive emotions in crises? A prospective study of resilience and emotions following the Terrorist attacks on the United States on September 11th, 2001. J Pers Soc Psychol 2003;84:365 – 76. [25] Lazarus RS. Stress and emotion, a new synthesis. London: Springer Publishing Company, 1999. [26] Taylor SE, Kemeny ME, Reed GM, Bower JE, Gruenewald TL. Psychological resources, positive illusions, and health. Am Psychol 2002;55:99 – 109. [27] Tugade MM, Fredrickson BL, Barrett LF. Psychological resilience and positive emotional granularity: examining the benefits of positive emotions on coping and health. J Pers 2004;72:1162 – 90. [28] Zautra A, Johnson JL, Davies M. Positive Affect as a source of resilience for woman in chronic pain. J Consult Clin Psychol 2005;73: 212 – 20. [29] Tugade MM, Fredrickson BL. Resilient individuals use positive emotions to bounce back from negative emotional experiences. J Pers Soc Psychol 2004;86:320 – 33. [30] Fredrickson BL. The role of positive emotions in positive psychology. The Broaden-and-build theory of positive emotions. Am Psychol 2001; 56:218–26. [31] Isen AM. Positive affect. Handbook Cogn Emot 1999;521 – 39. [32] Zautra AJ, Smith B, Affleck G, Tennen H. Examination of chronic pain and affect relationship: application of a dynamic model of affect. J Consult Clin Psychol 2001;69:786 – 95. [33] Davis MC, Zautra AJ, Smith BW. Chronic pain, stress, and the dynamics of affective differentiation. J Pers 2004;72:1133 – 59. [34] Reich JW, Zautra AJ, Davis M. Dimensions of affect relationship: models and their integrative implications. Rev Gen Psychol 2003;7: 66 – 83. [35] Zautra AJ. Emotion, stress and health. New York: Oxford Press, 2003. [36] Eid M, Diener E. Norms for experiencing emotions in different cultures: inter- and intranational differences. J Pers Soc Psychol 2001; 81:869 – 85.
[37] Farrar JT, Young JP, LaMoreaux L, Werth JL, Poole RM. Clinical importance of changes in chronic pain intensity measurement on an 11-point numerical pain rating scale. Pain 2001;94:149 – 58. [38] Watson D, Clark LA, Tellegen A. Development and validation of brief measures of positive affect: the PANAS Scales. J Pers Soc Psychol 1988;54:1063 – 70. [39] Zautra AJ, Guarnaccia CA, Dohrenwend BP. Measuring small life events. Am J Community Psychol 1986;14:625 – 55. [40] Beck AT, Steer RA, Garbin MG. Psychometric properties of the Beck Depression Inventory: twenty-five years of evaluation. Clin Psychol Rev 1988;8:77 – 100. [41] Singer JD. Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. J Educ Behav Stat 1998;24:323 – 55. [42] Verbeke G, Molenberghs G. Linear mixed model for longitudinal data. New York: Springer, 2000. [43] Berry DS, Hansen JS. Positive affect, negative affect, and social interaction. J Pers Soc Psychol 1996;71:796 – 809. [44] Gable SL, Reis HT, Elliot AJ. Behavioral activation and inhibition in everyday life. J Pers Soc Psychol 2000;78:1135 – 49. [45] Singer AJ, Kowalsko A, Thode HC. Ability of patients to accurately recall the severity of acute painful events. Acad Emerg Med 2001;8: 292 – 5. [46] Reidelmeier D, Kahneman D. Patient’s memories of pain medical treatments: real-time and retrospective evaluations of two minimally invasive procedures. Pain 1996;66:3 – 8. [47] Stone AA, Broderick JE, Kaell AT, Philippe AE, DelesPaul G, Porter LE. Does the peak-end phenomenon observed in laboratory pain studies apply to real-world pain in rheumatoid arthritis? J Pain 2000;1: 212 – 7. [48] Gedney JI. Predictors of short-term and long-term memory of sensory and affective dimensions of pain. J Pain 2003;4:47 – 55. [49] Von Korff M, LeResche L, Dworkin S. First onset of common pain syndrome: a prospective study of depression as a risk factor. Pain 1993;55:251 – 8. [50] Gaskin ME, Greene AF, Robinson ME, Geisser ME. Negative affect and the experience of chronic pain. J Psychosom Res 1992;36: 707 – 13.
Positive affect as a factor of resilience in the pain—negative affect relationship in patients with rheumatoid arthritis Elin B. Stranda,T, Alex J. Zautrab, Magne Thoresena, Sigrid adeg3rdc, Till Uhligc, Arnstein Finseta a
Department of Behavioural Sciences and Statistics, Institute of Basic Medical Science, University of Oslo, POB 1111 Blindern, N-0317 Oslo, Norway b Department of Psychology, Arizona State University, Tempe, AZ, USA c Department of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway Received 2 May 2005; received in revised form 8 August 2005; accepted 16 August 2005
Abstract Objective: The purpose of this study is to examine positive affect (PA) as a factor of resilience in the relationships between pain and negative affect (NA) in a sample of patients with rheumatoid arthritis. Methods: Forty-three patients (30 women; mean age, 57 years) were interviewed weekly by telephone for 8 weeks. Multilevel modeling was applied to study the within-week relationships among the variables. Results: There was a PainPA interaction effect on NA (b=0.05, Pb.01) indicating a weaker relationship between pain and NA in weeks
with more PA. Pain (b=0.37, Pb.002), interpersonal stress (b=2.42, Pb.001), depression (b=0.26, Pb.01), average perceived stress (b=10.80, Pb.001), and also weekly PA (b=0.1, Pb.01) had a main effect upon NA. Conclusion: Positive affect is most influential in reducing NA during weeks of higher pain and may be a factor of resilience, helping patients experiencing pain fluctuations as less distressful than at lower levels of PA. D 2006 Elsevier Inc. All rights reserved.
Keywords: Negative affect; Pain; Positive affect; Resilience; Rheumatoid arthritis; Stress
Introduction In this paper, we ask whether the stressful impact of chronic pain is lessened by the presence of positive emotions for patients with rheumatoid arthritis (RA). Chronic pain is reported as the most widespread and challenging symptom for patients with RA [1] and a high priority for physician’s attention [2]. Pain in RA is also a potential stressor not only because it is a highly aversive bodily experience but also because the pain intensity and duration fluctuate in a relatively unpredictable and uncontrollable way. Pain varies between individuals and across situations in intensity and duration, and patients differ in the extent that T Corresponding author. Department of Behavioural Sciences and Statistics, Institute of Basic Medical Science, University of Oslo, POB 1111 Blindern, N-0317 Oslo, Norway. Tel.: +47 22 85 10 21; fax: +47 22 85 13 00. E-mail address: [email protected] (E.B. Strand). 0022-3999/06/$ – see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jpsychores.2005.08.010
arthritis pain gives rise to emotional distress. Although there is a well-established association between pain and negative affect (NA) [3,4], both interpersonal stress and also depression play a role in a patient’s vulnerability to pain [5–11]. For RA patients, negative affective responses to pain may influence illness course, increasing the frequency of painful flares, lowering pain thresholds, intensifying pain behaviors, and deteriorating coping [12–19]. Thus, identifying factors that may diminish the established connection between pain and NA may be of considerable value to the health as well as the mental health of RA patients. A shift in focus in pain research to explore an RA patient’s capacities for resilience as well as their vulnerabilities to pain appears warranted at this juncture. The concept of resilience refers to the person’s ability to bounce back from negative emotional experiences and show a flexible adaptation to the changing demands of stressful experiences [20]. These attributes are of considerable importance for sustaining health and well being [21,22]. Positive emotions may be
478
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
essential factors in the process of resilience following adverse events [23–27], including adverse experiences with chronic pain [28]. Tugade and Fredrickson [29] found that highresilient individuals tended to report positive emotions even when under stress, and that these positive emotions contributed to recovery from stress-related negative effects. Their bBroaden and BuildQ theory of positive emotions [30] posits that persons with higher positive affect (PA) have greater capacity to recover psychologically and physiologically to stressful events. Positive emotions may expand the range of cognitions and behaviors to build an individual’s physical, intellectual, and social resources [29–31]. The role of PA in persons with chronic pain has been examined by Zautra et al. [28,32]. In their dynamic model of affect (DMA), they suggest that the relationship between PA and NA changes as a function of context [33–35]. Dynamic model of affect posits that stressful events change the degree of independence between the affective states so they become less differentiated, that is, more bipolar. Thus, according to the DMA, people who can sustain higher levels of PA at the time of the stressor would show significantly less NA [35]. In one longitudinal study of arthritis (RA) patients, they found that the presence of positive affective states reduced the size of the relationship between the patient’s reported weekly baverage painQ and NA [32]. Weekly registrations of pain and affect were employed for a period of 12 to 20 weeks. In this study, they examined the role of PA in the relationship between pain and NA, and with mood clarity as a confounder. So far, no study other than Zautra et al’s [32] has explored this relationship with a combined within- and between-person design, and advanced statistical methods such as multilevel modeling. Stress is highly related to the environmental context— especially interpersonal events [25]. For patients with RA, stress in interpersonal relationship is associated to increased disease activity, depression, and pain [8,17]. In this group of patients, the PA–interpersonal stress interaction on NA has not, as far as we know, been examined. In the current study, we also explore interpersonal stress, that is, the patient’s perceived stress in friends, family, and spouse/partner relationship. The current study is a replication of Zautra et al’s study [32] only in so far as it tests the relationships among pain, PA, and NA. We have also added tests of the role of interpersonal stress, which was not included in the prior study. We also studied bworst painQ during the past week, which was another departure from prior research that relied on reports of last week’s average pain. The bipolarity of the affects and the resilience factor are both stress-related phenomenon. Bipolarity in affects is according to DMA, a consequence of stressful events. For resilience to display, there need to be aversive states to bounce back from, and because the worst pain more closely identifies a stressful event, we rely on this pain rating in our study. To date, the role of PA as a source of resilience has not been studied in patients with RA. This study aims to bridge
that gap and expand our understanding of the affect interrelationships in RA. Norms for experiencing and expressing emotions differ widely between countries, even between countries that on most dimensions may appear similar and share many sociocultural features [36], such as Norway and the United States. It is therefore valuable to test the effects of PA on NA among RA patients who reside in these countries other than the United States. In this paper, we examine data on a sample of Norwegian patients with RA on the association between pain fluctuations, elevations in interpersonal stress, and NAs. We also address the question of how PA influences these relationships. Finally, we explore individual differences in depression and perceived interpersonal stress as vulnerability factors in the experiences of NA during chronic pain.
Method Subjects The sample consisted of 43 patients with RA included in a 10-year follow-up of the Norwegian EURIDISS cohort (European Research in Incapacitating Diseases and Social Support [11]). At entry into the cohort 10 years prior, patients had been diagnosed with RA within the last 4 years. They were asked to take part in the current study when they came to the hospital for the 10-year follow-up examination. Of the 238 patients originally included in the EURIDISS study at T1 (1992), 35 patients had died. Forty-two individuals refused to take part in the follow-up, and 12 patients did not take part in the follow-up for other reasons (could not be located, had moved, etc.). Thus, 149 (63% of 238) patients took part in the 10-year follow-up. Of these 149 patients, 43 participated in the current study. The present sample consisted of 30 (70%) woman and 13 (30%) men. The mean age was 57.5 (S.D.=13.1) with a range from 33 to 80 years. Of the 43 patients, 26 (60%) were married or living with a partner, 33 (75%) had one or more children, and 10 (23%) were in a full-time job, 5 (12%) had a part-time job, and the rest [28 (65%)] were on age or disability pension. A comparison of this sample to the EURIDISS cohort on age, education, sex, Steinbrocker (physical function), Physician global clinical evaluation (VAS), SF-36, Nottingham Health Profile, HAQ, and GHQ revealed significant differences only in age and education. The current sample was younger and had longer education than the EURIDISS cohort. Overall, the current sample was functioning at a relatively high level for RA patients. Of the 43 patients, 80% had a Steinbrocker score (a global measure of function) of 2 on a scale ranging from 1, which is no impairment, to 4, which signifies extensive handicap such as using a wheelchair or staying in bed. None of the RA patients had a score on IV.
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
Assessment The patients were invited to the hospital for an initial interview. They were given the first standardized interview and asked to complete baseline questionnaires. The participants were then given weekly, standardized telephone interviews during an 8-week period. In these interviews, they were asked to give a report about last week’s pain, PA, NA, as well as interpersonal stress. A total of 336 weeks of data were obtained, with only 8 weeks of missing data, for a 98.5% completion rate. The missing weeks were distributed on eight different persons: five women and three men. Two of the missing weeks were in the person’s interview week number 3, one in the interview week number 5, another in the interview week number 7, and the other three in the interview week number 8. Two of the missing weeks were due to the interviewer himself, another two were due to hospital stay, forgetting (two), exhaustion (one), and death in the family (one). Most of these missing weeks are to be considered random missing. The proportion of missing weeks is small, and even smaller are the missing weeks due to worsening of health or illness fluctuations (n=3). No one of the patients mentioned pain, PA, NA, interpersonal stress, or depression, all important variables in the study, as reasons for not being able to undertake the telephone interview. Based on this analysis, we do not consider neither the number nor the reasons for the missing data to be of any serious threat to the integrity of the data analytic method. Pain During the phone interview, the participants were asked to rate last week’s pain due to their RA on an 11-point numeric rating scale with points from 0 (no pain) to 10 (worst possible pain) [37]. They separately rated three aspects of the pain: the most intense, the least intense, and the average pain level, over the past week. We provide data on each of the three pain ratings but rely on the rating of most intense pain (MIP) in the test of the effects of PA because this measure most closely identifies weeks when the person’s pain is the most stressful. Research also has shown that peak intensity of pain is less likely to be biased by retrospective recall [47]. Positive affect and negative affect To measure affect weekly, we used Positive and Negative Affect Schedule, which is an established instrument developed to tap the two major dimensions of mood— positive and negative affectivity [38]. One of the distinguishing aspects of the scale’s development was its emphasis on indices of PA and NA that were independent. We used Watson et al.’s own categorizing for PA and NA with 10 items in each affect category. Participants were asked to indicate on a five-point scale, bto what extent did
479
you experience each of the adjectives for positive and for NA during the previous week.Q The five-point scale is labeled 1, very slightly/not at all; 2, a little; 3, moderately; 4, quite a bit; and 5, very much. The adjectives for PA were binterested,Q bstrong,Q binspired,Q battentive,Q benthusiastic,Q bproud,Q balert,Q blively,Q bactive,Q and bdetermined,Q and for NA, the adjectives were bdistressed,Q bupset,Q bnervous,Q bscared,Q bhostile,Q birritable,Q bashamed,Q bjittery,Q bafraid,Q and bguilty.Q PANAS was translated from English into Norwegian and then translated back into Norwegian by a person with English as first language. After discussions among authors on divergences between the versions, the final solutions were used in the study. The translations of PANAS were executed specific for this study, and therefore, there are no other validation of the instruments to compare with. Watson et al. [38] assert that internal consistencies for both scales have been ranging from .86 to .90 for PA and from .84 to .87 for NA in their studies. In the current study, the Cronbach’s a was .90 for NA and .83 for PA. In Watson et al’s studies, the correlation between the NA and PA scales has been invariably low, ranging from .12 to .23. In our study, the correlation between NA and PA was .11 for the weekly scores and .25 for the average affect scores. Interpersonal stress The patients were asked to indicate on a four-point scale the extent interpersonal stress arising over the past week in each of the following areas: friends, family, and spouse/ partner. bOverall, how stressful were your relations with friends (family, partner) this past weekQ (zero=not stressful at all, 1=a little stressful, 2=moderately stressful, and 3=extremely stressed). Each of these questions followed probes regarding number of stressful events in each of the three life domains taken from the Inventory of Small Life Events [39]. Two interpersonal stress variables were computed. An average interpersonal stress variable across the 8 weeks was computed for each participant based on the mean of three weekly ratings: perceived stress with friends, family, and spouse/partner (if present). Another weekly interpersonal stress score was calculated by subtracting each individual’s weekly score from his/her average score. This person-centered method of creating the weekly deviation scores was also used for all the weekly variables used in this paper. Depression Beck Depression Inventory (BDI) was given at the initial interview. Beck Depression Inventory is a widely used instrument to measure depression and consists of 21 items that assess cognitive–affective and vegetative signs of depression. The BDI has been used widely to assess depressive symptoms in psychiatric and nonpsychiatric populations [40].
480
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
Table 1 Means, S.D., and correlation coefficients of predictor variables in the multilevel analyses Variable
n
Mean
S.D.
Range
Criterion NA (raw score)
333
13.8
4.67
23
0.00 0.00 0.00
1.52 4.46 0.27
4.75 31.00 0.22 8.34
2.32 6.00 0.23 4.20
Within-person deviation scores Weekly MIP 336 Weekly PA 328 Weekly IS 333 Between-person variables Average MIP Average PA Average IS Depression
43 43 43 41
NA
MIP
PA
11.3 34.0 2.31
0.12T 0.11 0.15TT
0.06 0.004
0.04
9.50 31.5 0.88 20.0
0.16 0.25 0.52TT 0.32T
0.37T 0.41TT 0.06
0.08 0.35T
IS
0.10
n refers to the number of observations. The scores on the weekly variables are centered by subtracting the observed score from each person’s average across the eight weekly observations. T Pb.05 level (two tailed). TT Pb.01 (two tailed).
Statistical procedure We wanted to investigate the relationship between NA and pain, PA, interpersonal stress (IS), and depression, measured at eight weekly time points, adjusting for individual characteristics of the patients. The assessments provided two levels of data: level 1 consisted of the variables that varied within participants across weeks; level 2 variables identified individual differences that varied between participants but not over weeks. We analyzed the multilevel model with SAS PROC MIXED software (e.g., Ref. [41]). For the level 1 data (the relationship among the variables measured weekly), the equation may be stated as follows: NAij ¼ c0j þ c1j MIPij þ c2j PAij þ c3j IS ij þ c4j PAij MIP þ c5j PAij IS þ c6j week1ij þ eij where the subscripts i and j represent week number and patient number, respectively. The variable bweek1Q indicates whether the measurement is made in the first week or not, and inclusion of this
variable controls for elevations in level commonly found at the first weekly measurement compared with subsequent weekly assessments. The independent variables weekly MIP, weekly PA, and weekly IS are centered at the mean value for each person. These centered variables identify weekly fluctuations in level of pain, PA, and interpersonal stress, independent of the person’s average level across the 8 weeks. To test how PA modulates the relationships between pain, interpersonal stress, and NA, we included interaction terms shown in the equation. The level 2 equation models the between-subject variation in level 1 intercepts: k0j ¼ b00 þ b01 BDI j þ b02 AvgIS j þ b03 AvgPAj þ b04 AvgMIPj þ u0j : This level of analysis is used to explore the extent to which NA was affected by individual differences in depression, average perceived interpersonal stress (AvgIS), average PA (AvgPA), and average MIP (AvgMIP). In addition, separate equations for the interactions between levels 2 and 1 examined individual differences in PA on the
Fig. 1. Interaction of PA on weekly MIP on NA.
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484 Table 2 Multilevel regression predicting weekly NA Random effects Covariance parameter estimated
Subject
Un (1,1) Ar (1) Residuals
ID ID ID
b
S.E. 4.52 0.23 8.57
Z
1.56 0.09 0.95
P 2.89 2.60 8.98
.001 .009 .0001
Fixed effects Predictor variables Level 1 within-person variables Weekly IS Weekly MIP Weekly PA Week1 Weekly PAT Weekly MIP Level 2 between-person variables Average PA Average MIP Average IS Depression
b
S.E.
t
P
2.42 0.37 0.103 1.97 0.05
0.58 0.12 0.04 0.49 0.02
4.17 3.19 2.88 3.98 2.37
b.0001 .002 .004 b.0001 .019
0.11 0.02 10.80 0.26
0.08 0.22 1.93 0.10
1.40 0.11 5.61 2.52
.172 .917 b.0001 .017
For the level 1 analyses, we only report the significant results. n=336 weekly measurements for within-person analyses and n=43 for betweenperson analyses.
weekly pain–weekly NA and on the weekly IS–weekly NA relationships (the last one not shown): MIP1 ¼ c10 þ c11 BDI þ c12 AvgIS þ c13 AvgPA þ c14 AvgMIP þ u1 : In the final model, only the intercept (g0j ) was allowed to vary among patients and was modeled as a random effect. Because of the relatively small number of subjects, we decided not to attempt to estimate random effects of the variables in the model. Further, we modeled the dependence between repeated measurements of NA within patients as an autoregressive process of order 1. The dependent variable looked somewhat skewed. One important assumption behind linear multilevel model is that the residuals are normally distributed [41,42]. The estimated residuals for normality were examined and found satisfactory.
481
interpersonal stress (r=.41, Pb.01). There was also a moderate negative correlation between PA and depression (r=.35, Pb.05). There was no correlation between MIP and depression, average PA and average IS, and IS and depression. These last set of correlations show that our assessments of PA, pain and stress yielded individual differences on key variables that were distinct from depression and also different from one another. To analyze the effects of weekly MIP, weekly PA, weekly IS, and individual differences in depression and interpersonal perceived stress on NA, we used multilevel modeling as outlined earlier. On level 1, weekly MIP had a strong effect upon NA (b=0.37, Pb.0002). On weeks with more pain, NA was significantly higher. This effect was moderated, however, by weekly PA. Weeks of a higher PA moderated the effect of pain on NA. This relationship is illustrated in Fig. 1. Further, there was a significant effect of weekly PA on NA (Table 2). Weekly interpersonal stress was also associated with NA (b=2.42, Pb.0001), but there was no interaction effect between PA and IS in influencing NA. We also executed the same analyses for both weekly average pain and weekly lowest pain level. The same significant result on NA were found for these pain variables as with the MIP level except from the interaction with PA, which was nonsignificant for both the average and lowest pain level. When examining level 2 variables for between-person effects, we found that individual differences, both the average interpersonal stress (b=10.80, Pb.0001) as well as depression (b=0.26, Pb.017), significantly increased NA. In the final model, we included only those variables that were significant. This model (illustrated in Fig. 2) can be formulated as a so-called mixed model: NAij ¼ b0 þ b1 MIPij þ b2 IS ij þ b3 PAij þ b4 PAij MIPij þ b5 BDI j þ b6 AvgIS j þ b7 week1 þ uj þ eij ; where the random effect u j reflects the between-person variability in the intercept.
Results Initial analyses provided descriptive statistics and correlations among study variables. Means, S.D.s, and correlations between NA and the predictor variables are shown in Table 1. Negative affect was correlated with weekly MIP (r=.12, Pb.05) and IS (r=.15, Pb.01), but not with PA. Negative affect was, correlated with between-person differences in IS (r=.52, Pb.01), and depression (r=.32, Pb.01), and not with PA. Among the between-person variables, there was a moderate negative correlation between PA and MIP (r=.37, Pb.05), and MIP was also associated with more
Fig. 2. Final model of associations based on multiple regression analyses.
482
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
Discussion In this study, we found a significant association between weekly pain, weekly stress, and weekly NA. Moreover, we found a weekly PainPA interaction effect, indicating that PA moderated the effects of pain on NA such that higher levels of PA reduced NA during weeks of high pain. Weekly PA also had a direct effect on NA in significantly reducing it. Depression and average perceived stress across weeks were also associated with weekly NA and served as vulnerability factors in relation to NA. Weekly PA did not influence the IS–NA relationship. The average level of PA across the 8 weeks was unrelated to NA directly and also did not influence neither the effects of pain nor the effects of interpersonal stress on NA. Our findings are similar to those of Zautra et al. [32] in their study of a sample of RA and osteoarthritis patients in the United States. In both studies, weekly PA (bstateQ), but not mean PA across weeks (btraitQ), was inversely related to weekly NA, and PA seemed to reduce the effect of pain on NA in weeks of higher pain. Both individual differences in IS, MIP, and depression were in the current sample correlated to NA, where IS had the strongest correlation. In the regression analyses, both average IS and depression were significantly associated to NA, whereas average MIP was not. In the United States study, individual differences in pain across weeks explained a significant proportion of the variance in weekly NA. In the main model of our study (Table 2), between-subject differences in pain across weeks were unrelated to NA. In contrast to the United States study, this study included interpersonal stress in the model. If interpersonal stress was removed from the equation, pain across weeks made a significant contribution to NA similar to what Zautra et al. [32] found. Our findings indicate that both weekly pain and weekly interpersonal stress function as sources of NA, whereas the stable aspect of pain as a stressor (at least in our sample) covaried with interpersonal stress, as a strong predictor of NA. In the United States study, the authors used the patients’ average pain as pain measurement; the Norwegian study used the bMIP.Q The pain level seems to be somewhat lower in the present study (MIP, 4.7, and average pain level, 3.5, on a scale from 0 to 10) than in the United States samples (for RA patients, average pain level was 43 on a scale from 0 to 100); still, we found the same effect of PA on the pain–NA relationship. The differences between the studies appear best explained by the measures of pain used than by cultural differences, although we do not have data that have a direct bearing on that question. Our findings on the pain–affect relationship are in accordance with the DMA postulating a more bipolar relationship between PA and NA during the stress of increased pain fluctuations. In our data, higher average PA did not protect the RA participants from NA associated with weekly pain. Positive affect was most important as a deterrent when pain was at its worst. Thus, the state
dimension of PA may be thought of as a source of resilience during aversive states like with increasing pain. The resilient effect of PA in the pain–NA relation may result from the rise in the person’s well-being, and also through shifts in their cognitive appraisals of self-efficacy and other reframing of the pain and their own coping efforts. More research is needed to identify the mechanisms by which PA is protective during painful episodes. Of interest in particular is the extent to which this influence of PA is due to a narrowing of affective differentiation as proposed by Zautra [35] and/or a boost in affective resources as proposed by Fredrickson [30]. Rheumatoid arthritis participants with more depression and interpersonally stressed overall reported more NA, and also, weeks of high IS were associated with elevations in NA. These findings are in consistence with lots of the research initially mentioned. However, weekly PA did not influence the interaction between weekly interpersonal stress and NA as expected from the DMA. This negative finding may be due to a small sample and a rather low level of reported stress, which means that the stress-related NA is not strong enough to provoke bipolar affective functioning. If this is the case, then the nonsignificant results on the stress–PA interaction here are not inconsistent with the DMA model that we tested. These results may also indicate that stress and NA are two aspects of the same phenomena employed in two different ways through affect rating and questions about perceived stress. The results imply that they may be linked in such a way that not even an elevation in PA can separate them. That PA interacts differently in the pain–NA than in the stress–NA relationship may also indicate that pain and stress are two very different sources of NA and, therefore, demand different sources for bhealing,Q that is, PA may be a better source of resilience for NA caused by pain but not for NA caused by interpersonal stress. This is quite speculative, and the two sources of NA such as pain and interpersonal stress should be further explored. Both PA and NA mean scores in the present study are similar to those seen in other samples both clinical and nonclinical, using PANAS as affect measures [32,43,44]. Whether our results on the affect relationships would generalize to other populations of chronically ill awaits further study. Limitations The patients in our study were recruited from a lager sample of RA patient, and one-third took part in our study. Nonparticipation may be due to several reasons. Given the difference in age and education between the total and our sample, we presume that when presented our follow-up study, it seemed too demanding to the oldest and those with the lowest education. In summary, the sample that completed our study compares favorably with the larger sample from which they were drawn, which
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
gives us some confidence that the group is not overly biased in health and/or well-being. Our sample may still be more resilient than a random sample of the RA patients. It is noteworthy, however, that they did report fairly substantial levels of joint pain in our study and levels of disability comparable to the larger cohort from which they were drawn. We used a once-a-week measurement often used in research, and by consequence, the information is given retrospectively. Studies comparing patient’s evaluations of both momentary and recalled pain have shown that patients accurately recalled the severity of a pain episode for at least 1 week [45] and tended to recall the peak level and the most recent level [46,47]. Thus, the MIP level as used in our study appears to be relatively free of recall bias; nevertheless, some caution within interpretation of the finding is called for given the 7-day time span between assessments. The affective state at the time of a painful experience may shape the nature of the experiences and what is later recalled [48]. For example, NA at recall, may influence an individual’s report of the prior week’s pain and interpersonal stress. For example, depressed persons recall more pain than those who are not in such a mood [49,50]. Still, an important question left unresolved is whether the findings that infer same-time associations refer to influences over the full week, or are best thought of as associations within a shorter time span than that which was measured. The DMA model that was tested makes no assumptions regarding the proper length of time that a same-time association may be observed. Further studies are needed to gauge the time interval within which changes in the nature of the relationships among affective states are most likely to occur. The weekly variables involved in the study may fluctuate frequently; both pain and affect may change several times during a week and even within 1 day. Daily and even within-day assessments are to be encouraged in future studies of this kind to rule out potential effects of biases in retrospective accounts. The main finding from our study is that PA appears to be an important resilience factor protecting patients with RA during pain fluctuations. This potential of the positive affective resources may encourage new approaches to improving the quality of life of patients with chronic pain. Interesting questions remain regarding the differences between persons in their capacity to mobilize and experience positive emotional states when pain increases, and to what degree the person’s social life and interpersonal relations influence in this process. In our analyses, we controlled for individual differences in depression and interpersonal stress to rule out potential confounds to NA and PA. Findings from this study invite further exploration of how differences in individual capacities in emotional awareness and regulation, and the person’s social resources and interpersonal relations influence adaptation to chronic pain conditions such as RA.
483
Acknowledgments This research was funded by the Research Council of Norway, grant 147831/320. References [1] Katz PP. The stresses of rheumatoid arthritis: appraisals of perceived impact and coping efficacy. Arthritis Care Res 1998;11:9 – 22. [2] Heiberg T, Kvien TK. Preferences for improved health examined in 1024 patients with rheumatoid arthritis. Pain has highest priority. Arthritis Rheum 2002;15:391 – 7. [3] Fernandez E, Milburn TW. Sensory and affective predictors of overall pain and emotions associated with affective pain. Clin J Pain 1994;10: 3 – 9. [4] Fernandez E, Turk DC. The scope and significance of anger in the experience of chronic pain. Pain 1995;61:165 – 75. [5] Banks SM, Kerns RD. Explaining high rates of depression and pain: does the evidence support a relationship? Psychol Bull 1996;97:18 – 34. [6] Blumer D, Heilbronn M. Chronic pain as a variant of depressive diseases. The pain-prone disorder. J Nerv Ment Dis 1982;170:381 – 406. [7] Engel GL. bPsychogenicQ pain and the pain-prone patient. Am J Psychiatry 1959;137:899–918. [8] Kerns RD, Rosenberg R, Otis JD. Self-appraisal problem solving and pain-relevant social support as predictors of the experience of chronic pain. Ann Behav Med 2002;24:100 – 5. [9] Leino P, Magni G. Depressive and distress symptoms as predictors of low back pain, neck–shoulder pain, and other musculoskeletal morbidity: a 10-year follow-up of mental industry employees. Pain 1993;53:89 – 94. [10] Magni G, Moreschi C, Rigatti-Luchini S, Merskey H. Prospective study on the relationship between depressive symptoms and chronic musculoskeletal pain. Pain 1994;56:289 – 97. [11] Smestad LM, Vaglum P, Moum T, Kvien T. The relationship between psychological distress and traditional clinical variables: a 2 year prospective study of 216 patients with early rheumatoid arthritis. Br J Rheumatol 1997;36:1 – 8. [12] Affleck G, Tennen H, Urrows S, Higgins P. Neuroticism and the pain– mood relation in rheumatoid arthritis: insights from a prospective daily study. J Consult Clin Psychol 1992;60:119 – 26. [13] Bradley LA, Alberts KR. Psychological and behavioral approaches to pain management for patients with rheumatic disease. Rheum Dis Clin North Am 1999;25:215 – 332. [14] Janssen SA. Negative affect and sensitization to pain. Scand J Psychol 2002;43:131 – 7. [15] Keefe FJ, Bonk V. Psychosocial assessment of pain in patients having rheumatic diseases. Rheum Dis Clin North Am 1999;25:81 – 103. [16] Lefebvre JC, Keefe FJ, Affleck G, Raezer LB, Starr K, Caldwell DS. The relationship of arthritis self-efficacy to daily pain, daily mood, and daily pain coping in rheumatoid arthritis patients. Pain 1999;80: 425 – 35. [17] Manne S, Zautra A. Spouse criticism and support: their association with coping and adjustment to rheumatoid arthritis. J Pers Soc Psychol 1989;56:608 – 17. [18] Schwartz L, Slater MA, Bircher GR. Interpersonal stress and pain behaviors in patients with chronic pain. J Consult Clin Psychol 1994; 62:861 – 4. [19] Turk DC, Okifuji A. Evaluating the role of physical, operant, cognitive, and affective factors in the pain behaviors of chronic pain patients. Behav Modif 1997;21:259 – 80. [20] Block J, Kremen AM. IQ and ego-resiliency: conceptual and empirical connections and separateness. J Pers Soc Psychol 1996; 70:349 – 61. [21] Jacelon CS. The trait and process of resilience. J Adv Nurs 1997;25: 123 – 9.
484
E.B. Strand et al. / Journal of Psychosomatic Research 60 (2006) 477 – 484
[22] Klohnen EC. Conceptual analysis and measurement of the construct of ego-resiliency. J Pers Soc Psychol 1996;70:1067 – 79. [23] Folkman S, Moskowitz JT. Positive affect and the other side of coping. Am Psychol 2000;55:647 – 54. [24] Fredrickson BL, Tugade MM, Waugh CE, Larkin GR. What good are positive emotions in crises? A prospective study of resilience and emotions following the Terrorist attacks on the United States on September 11th, 2001. J Pers Soc Psychol 2003;84:365 – 76. [25] Lazarus RS. Stress and emotion, a new synthesis. London: Springer Publishing Company, 1999. [26] Taylor SE, Kemeny ME, Reed GM, Bower JE, Gruenewald TL. Psychological resources, positive illusions, and health. Am Psychol 2002;55:99 – 109. [27] Tugade MM, Fredrickson BL, Barrett LF. Psychological resilience and positive emotional granularity: examining the benefits of positive emotions on coping and health. J Pers 2004;72:1162 – 90. [28] Zautra A, Johnson JL, Davies M. Positive Affect as a source of resilience for woman in chronic pain. J Consult Clin Psychol 2005;73: 212 – 20. [29] Tugade MM, Fredrickson BL. Resilient individuals use positive emotions to bounce back from negative emotional experiences. J Pers Soc Psychol 2004;86:320 – 33. [30] Fredrickson BL. The role of positive emotions in positive psychology. The Broaden-and-build theory of positive emotions. Am Psychol 2001; 56:218–26. [31] Isen AM. Positive affect. Handbook Cogn Emot 1999;521 – 39. [32] Zautra AJ, Smith B, Affleck G, Tennen H. Examination of chronic pain and affect relationship: application of a dynamic model of affect. J Consult Clin Psychol 2001;69:786 – 95. [33] Davis MC, Zautra AJ, Smith BW. Chronic pain, stress, and the dynamics of affective differentiation. J Pers 2004;72:1133 – 59. [34] Reich JW, Zautra AJ, Davis M. Dimensions of affect relationship: models and their integrative implications. Rev Gen Psychol 2003;7: 66 – 83. [35] Zautra AJ. Emotion, stress and health. New York: Oxford Press, 2003. [36] Eid M, Diener E. Norms for experiencing emotions in different cultures: inter- and intranational differences. J Pers Soc Psychol 2001; 81:869 – 85.
[37] Farrar JT, Young JP, LaMoreaux L, Werth JL, Poole RM. Clinical importance of changes in chronic pain intensity measurement on an 11-point numerical pain rating scale. Pain 2001;94:149 – 58. [38] Watson D, Clark LA, Tellegen A. Development and validation of brief measures of positive affect: the PANAS Scales. J Pers Soc Psychol 1988;54:1063 – 70. [39] Zautra AJ, Guarnaccia CA, Dohrenwend BP. Measuring small life events. Am J Community Psychol 1986;14:625 – 55. [40] Beck AT, Steer RA, Garbin MG. Psychometric properties of the Beck Depression Inventory: twenty-five years of evaluation. Clin Psychol Rev 1988;8:77 – 100. [41] Singer JD. Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. J Educ Behav Stat 1998;24:323 – 55. [42] Verbeke G, Molenberghs G. Linear mixed model for longitudinal data. New York: Springer, 2000. [43] Berry DS, Hansen JS. Positive affect, negative affect, and social interaction. J Pers Soc Psychol 1996;71:796 – 809. [44] Gable SL, Reis HT, Elliot AJ. Behavioral activation and inhibition in everyday life. J Pers Soc Psychol 2000;78:1135 – 49. [45] Singer AJ, Kowalsko A, Thode HC. Ability of patients to accurately recall the severity of acute painful events. Acad Emerg Med 2001;8: 292 – 5. [46] Reidelmeier D, Kahneman D. Patient’s memories of pain medical treatments: real-time and retrospective evaluations of two minimally invasive procedures. Pain 1996;66:3 – 8. [47] Stone AA, Broderick JE, Kaell AT, Philippe AE, DelesPaul G, Porter LE. Does the peak-end phenomenon observed in laboratory pain studies apply to real-world pain in rheumatoid arthritis? J Pain 2000;1: 212 – 7. [48] Gedney JI. Predictors of short-term and long-term memory of sensory and affective dimensions of pain. J Pain 2003;4:47 – 55. [49] Von Korff M, LeResche L, Dworkin S. First onset of common pain syndrome: a prospective study of depression as a risk factor. Pain 1993;55:251 – 8. [50] Gaskin ME, Greene AF, Robinson ME, Geisser ME. Negative affect and the experience of chronic pain. J Psychosom Res 1992;36: 707 – 13.