Systemic lupus erythematosus (SLE) is a

Transcription

1 PAIN MEDICINE Volume 4 Number Adaptation to Chronic Pain in Systemic Lupus Erythematosus: Applicability of the Multidimensional Pain Inventory Carol M. Greco, PhD,* Thomas E. Rudy, PhD,* and Susan Manzi, MD, MPH Departments of *Anesthesiology, Rehabilitation Science and Technology, Psychiatry, Biostatistics, and Medicine, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania ABSTRACT Objectives. 1) Investigate psychosocial adaptation to chronic pain in patients with systemic lupus erythematosus; 2) Compare pain adaptation of lupus, chronic low back pain, and temporomandibular disorders patients; and 3) Evaluate the validity of Multidimensional Pain Inventorybased profiles of lupus patients. Methods. Two hundred forty females with pain related to systemic lupus erythematosus (N = 80), chronic low back pain (N = 80), or pain related to temporomandibular disorders (N = 80) completed the Multidimensional Pain Inventory, a 60-item psychosocial assessment instrument. All patients were classified into empirically derived profiles based on Multidimensional Pain Inventory responses. Systemic lupus erythematosus profile groups were compared on conceptually related variables external to the classification, including indices of lupus disease activity, pain, distress, and physical function. Results. The scores of lupus patients resembled those of temporomandibular disorder patients, while chronic low back pain patients had higher pain and activity interference. 87.5% of lupus patients could be classified into one of three profiles: Dysfunctional (14%), interpersonally distressed (27.5%), or adaptive coper (46%). The Multidimensional Pain Inventory profiles for lupus patients demonstrated validity, based on external measures. Conclusions. Although many systemic lupus erythematosus patients cope well with their chronic pain, a substantial proportion exhibit pain-related distress, activity interference, or interpersonal difficulties. Key Words. Systemic Lupus Erythematosus; Psychosocial Subgroups; Pain; Low Back Pain; Temporomandibular Disorders Introduction Systemic lupus erythematosus (SLE) is a chronic, unpredictable autoimmune disease that predominantly affects women of childbearing age. Chronic pain is one of the most frequently reported problems among SLE patients, with 85% reporting arthralgia [1], and 32 66% reporting headache of various etiologies [2 4]. Myalgia, Reprint requests to: Carol M. Greco, PhD, Pain Evaluation and Treatment Institute, University of Pittsburgh, 5750 Centre Avenue, Suite 400, Pittsburgh, PA Tel: (412) ; Fax: (412) ; grecocm@msx.upmc.edu. abdominal pain, and chest pain (esophageal pain, pleuritic-type pain, pericarditis, or costochondritis) are also common. SLE patients who report high levels of pain incur substantially more days of diminished productivity in comparison with SLE patients with less pain [5]. Lupus pain is also associated with perceptions of lowered physical functioning [6] and coping resources [7]. Despite the prevalence of pain in this population, SLE patients psychosocial adaptations to chronic pain are not well understood. Recent studies suggest the need for psychosocial interventions to improve pain, distress, coping skills, and other aspects of health status in persons with SLE [8 10]. American Academy of Pain Medicine /03/$15.00/

2 40 Greco et al. However, identifying which patients require interventions and tailoring treatment components to their unique psychosocial needs may be necessary to insure the success and cost-effectiveness of such interventions. The Multidimensional Pain Inventory (MPI) is a 60-item self-report questionnaire designed to assess chronic pain patients perceptions and interpretations of their symptoms. Three profiles of psychosocial adaptation to chronic pain have been identified empirically through cluster analyses of patients responses on the MPI [11 13]. These MPI profiles, or subgroups, are largely independent of physical disease severity [12,14]. Approximately 90% of individuals with diverse pain conditions, including low back pain [12,13], headache [15], temporomandibular disorders (TMD) [14], and fibromyalgia [16], can be classified into one of three profiles. The first profile includes patients who report high levels of pain, affective distress, and interference with activities. This group is labeled dysfunctional (DYS) because pain broadly affects mood and activities. A second group of pain patients is characterized by relationship difficulties and a low sense of support and is labeled interpersonally distressed (ID). Lower levels of disability, distress, and perceived pain characterize the third profile. This profile group is labeled adaptive copers (AC). The external validity of the MPI classification is supported by studies [12,14,16] in which the profile groups responded differently on depression inventories [17,18], marital adjustment [19], and perceptions of control over health [20]. The MPI subgroups have shown differential outcomes following treatment programs [21 23]. For example, pain severity was reduced during multidisciplinary pain treatment in fibromyalgia patients classified as DYS or AC but not in those classified as ID [23], and depression improved in TMD patients in the DYS subgroup, but not in the ID or AC subgroups following dental and psychological treatments [21]. Thus, the information gained from the MPI-based classification system can be useful clinically for adapting treatment to the patient s requirements. Historically, the MPI has been used in populations whose pain etiology is poorly understood or difficult to discern, such as headache, TMD, and low back pain. The MPI classification system has recently been applied to groups with known organic pathology, such as malignant pain, including metastatic and nonmetastatic cancer, in order to identify individuals with problematic pain adaptation [24,25]. To date, the MPI classification has not been applied to individuals with pain related to SLE. In the past, psychosocial aspects of living with life-threatening diseases such as cancer and SLE have been relatively neglected. However, the psychological and behavioral impacts of SLE disease are receiving increased attention in recent years. For example, high psychological distress is associated with poorer perceived physical and mental functions [9], and low self-esteem and inadequate social support have been linked to higher disease activity and cumulative damage in SLE patients [26]. Identification of potential subgroups of SLE patients whose relationships or activity levels are affected by chronic pain could lead to improved ability to successfully address their unique needs. The aims of this study were: 1) To evaluate SLE patients psychosocial and behavioral adaptations to chronic pain using the MPI; 2) To compare the proportions of SLE individuals in the three psychosocial profiles with the proportions in two distinct chronic pain groups: Chronic low back pain (CLBP) and TMD; and 3) To evaluate the validity of the MPI-based classification system for SLE by comparing the profiles on psychosocial variables external to the MPI classification system. Methods Participants A total of 240 females with pain related to SLE, TMD, or CLBP (80 per group) were included. The SLE sample consisted of patients enrolled in a treatment study at the Pain Evaluation and Treatment Institute (PETI) [27] who had been recruited from the Pittsburgh Lupus Registry and lupus support groups in the Western Pennsylvania and Eastern Ohio area. Of SLE patients approached, 24.6% indicated that they did not experience substantial pain, and therefore, were not interested in participating. Participants completed written informed consent approved by the Institutional Review Board prior to study entry. All SLE participants met 1982 revised American College of Rheumatology criteria for SLE [28] and were females aged years who reported experiencing pain at least three times each week for at least 3 months. SLE patients were excluded if medications and dosages had not been stable for at least 1 month, if they required greater than 15 mg prednisone daily, or if they had current active kidney disease associated with SLE. Because SLE affects women and men at a ratio of approx-

3 Pain Profiles in SLE 41 imately 10: 1, we included only females in this study of SLE pain adaptation. One aim of this study was to compare the psychosocial adaptation of SLE patients with chronic pain with adaptation in other pain conditions that represent a broad spectrum of disability related to pain. TMD and CLBP pain groups were chosen for comparison with SLE because they represent the extremes in a continuum of pain-related disability [13]. Individuals with TMD frequently have only intermittent pain, and impaired function is typically limited to masticatory activities. In contrast, CLBP patients are typically more disabled by their pain. Females with TMD and CLBP were randomly selected, using the SAMPLES program [29], from available research records at the PETI. The TMD and CLBP patients had completed informed consent procedures and completed the MPI prior to entering treatment studies at the same clinic as the SLE patients. These patients were evaluated and diagnosed by a board-certified anesthesiologist (CLBP) or prosthodontist (TMD), were over 18 years of age, did not carry an additional diagnosis of SLE, and had pain duration of 3 months or longer. Procedures The CLBP and TMD patients completed the MPI as part of medical or dental evaluation entry visits for research treatment programs at the PETI. The SLE patients, who were entering another research study at the PETI, underwent a medical evaluation with a rheumatologist (Dr. Manzi) to determine current disease activity and cumulative damage. During this medical evaluation visit, the SLE patients completed the MPI and additional instruments designed to assess perceived pain, distress, and physical functioning. Measures MPI All participants completed the MPI [11], a 60-item instrument containing nine primary scales: Pain severity, interference with activities due to pain, sense of control over life, affective distress, perceived support, the frequencies of 1) punishing responses, 2) solicitous responses, and 3) distracting responses from significant others to the patients pain, and general activity levels. Scale scores range from 0 to 6, with 6 representing the most severe response. This instrument has been demonstrated to have good reliability and validity in a variety of chronic pain conditions [13] and is sensitive to change following multidisciplinary pain treatment [21,23]. Three profiles, DYS, ID, and AC, have been identified based on cluster analysis of MPI responses and replicated across numerous painful conditions [13,14,23 25]. Assessment of SLE Disease Activity and Damage Physician-rated data on SLE activity included three validated and reliable measures. The Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) [30] contains 24 weighted descriptors of signs and symptoms present in the past 10 days, and scores can range from 0 (no activity) to 105. The Systemic Lupus Activity Measure Revised (SLAM-R) [31] measures disease activity in 11 organ systems over the previous month. Scores on the SLAM-R can range from 0 to 81, and a score greater than 8 is considered to represent a level of clinical activity that indicates need for treatment [32]. In addition, the Physician s Global Assessment (PGA) [33] of SLE activity was completed. The PGA is a 10-cm visual analog scale that represents the clinician s judgment of disease activity and has been used in the validation of other instruments. SLE disease damage was measured by the physician-rated Systemic Lupus International Collaborating Clinics/American College of Rheumatology (SLICC/ACR) damage index [34], which reflects cumulative damage in 10 organ systems. Damage may be due to SLE disease, complications of treatment, or presence of diabetes or cancer. Total scores range from 0 (no damage) to 46 (maximum damage). The physician s examination and completion of SLE disease activity and damage indices took place on the same day that SLE participants completed the MPI and the following psychosocial questionnaires. Assessment of Pain, Distress, and Perceived Physical Function Limitations in SLE Two psychometrically validated self-report measures of pain were included. The Arthritis Impact Measurement Scales Revised (AIMS-2) pain scale [35] consists of five averaged items (range: 0 4) that assess frequency and severity of pain and stiffness. The McGill Pain Questionnaire short form (MPQ-SF) total score [36] is designed to measure pain intensity using 15 verbal pain descriptors. Each descriptor is rated on severity from 0 to 3, yielding a possible range of 0 to 45 (highest intensity). Distress was assessed by three well-validated measures. The Center for Epidemiological

4 42 Greco et al. Studies Depression (CES-D) scale [18] is a 20- item inventory that assesses presence and frequency of depressive symptoms over the past week, yielding a score range of 0 to 60. Cohen s Perceived Stress Scale (STRESS) [37] (4-item version) assesses frequency of feeling out of control or overwhelmed with difficulties during the past week and can range from 0 (none) to 16 (high stress). The Arthritis Self-Efficacy Scales pain and other symptoms [38] consists of 11 items designed to measure confidence in one s ability to manage the pain, fatigue, frustration, and other aspects of disease, reworded to reflect lupus rather than arthritis [8]. Confidence estimates for each item range from 0 100%, and items are averaged, yielding a score range of Perceived physical function was assessed in two ways. The SF-36 Health Survey [39] includes eight reliable and valid subscales that assess various aspects of perceived physical and mental health. This instrument is frequently included in qualityof-life studies of persons with lupus [40,41]. The SF-36 Physical Function (SF36-PF) scale was used in the present study. The SF36-PF consists of 10 questions regarding perceived limitations in performance of various everyday activities. Scores range from 0 100, with higher scores indicating better functioning. Additionally, participants perceptions of hours per day spent resting (HRS- REST) on the couch, bed, or a chair were used to assess physical function. This item can range from 0 to 24. Although this has not been used previously in lupus studies, similar measures (e.g., time spent in bed due to pain) are used in chronic pain populations [12]. Data Analysis The data analyses were designed to assess the applicability of the MPI scale scores and classification system to persons with pain related to SLE. The first objective was to determine whether there were MPI scale score differences among individuals with SLE, CLBP, and TMD. The MPI scales for SLE, TMD, and CLBP were compared using Multivariate Analysis of Variance (MANOVA), with follow-up univariate ANOVAs to detect individual scale differences, since the MANOVA was significant. Tukey honestly significant difference (HSD) post hoc analyses were conducted on any scales with significant ANOVA F values in order to determine specific group differences. The second objective of this study was to determine whether the SLE, CLBP, and TMD samples could be empirically classified into previously identified subgroups of DYS, ID, and AC based on MPI responses. Multivariate generalized distance functions and Bayesian approaches to empirical classification [12] were used to classify SLE, CLBP, and TMD participants MPI scale scores into one of the three profiles (i.e., DYS, ID, or AC). This empirical approach to classification evaluates the goodness of fit of each patient s MPI responses to each of the three profiles. The frequencies of the DYS, ID, and AC clusters or profiles in the different pain groups were compared using chi-square analyses. The third set of analyses was designed to provide external validation of the classification system by comparing MPI profile groups of SLE patients on selected demographic, disease activity, and psychosocial variables. Statistically significant differences were expected on psychosocial variables such as pain severity, distress, and perceived function, whereas the groups were not expected to differ from one another on demographic variables such as age and disease duration. The SLE patients fitting the DYS, ID, and AC classifications were compared using MANOVA, with univariate follow-up analyses and Tukey HSD pairwise post hoc comparisons for significant differences between groups. A P value of <0.05 was used to indicate statistical significance. Results Participant Description Only persons who reported experiencing pain in the past week were included in the study, and therefore, results should not be interpreted as pain prevalence rates for all individuals with SLE. The percentages of SLE patients reporting various types of pain are provided in Table 1. The median number of pain complaints was five (range: 2 9). Daily pain was reported by 85% of SLE participants, and pain lasting throughout the day (12 Table 1 Types of pain symptoms reported by SLE patients with persistent pain (N = 80) Pain type Joint 96.2 Muscle 72.0 Headache 59.5 Raynaud s phenomenon 39.8 Pleuritic-type 36.9 Oral/nasal ulcers 35.7 Abdominal 31.0 Fibromyalgia 21.4 Esophageal 20.2 % of patients

5 Pain Profiles in SLE 43 hours or more) was reported by 56.5%. Cumulative damage due to SLE ranged from 0 to 8 with median of 1 on the SLICC/ACR damage index [34], and current SLE disease activity as measured by the SLEDAI [30] ranged from 0 to 14 with median of 4. The length of time since SLE diagnosis ranged from less than 1 year to over 36 years, with a median of 10.2 years. In this sample of SLE patients, there were no statistically significant correlations between AIMS-2 pain severity and cumulative SLE damage (r = 0.197, P = 0.08), SLE disease activity index (r = 0.175, P = 0.18), or SLE duration (r = 0.066, P = 0.56). Demographic characteristics of the three chronic pain group samples are provided in Table 2. The three groups were compared using ANOVA and chi-square analyses (computed with StatXact 4 [42] for exact probabilities under small sample size conditions). The SLE, TMD, and CLBP patients did not differ significantly in age. Duration of pain in years differed among the groups, with SLE mean pain duration of 10.0 years significantly higher than CLBP at 6.3 years (Tukey HSD P = 0.011). Chi-square analyses indicated significant ethnicity and marital status differences, with follow-up hierarchical analyses indicating that there were fewer African American and more Caucasian individuals in the TMD sample than in the SLE and CLBP samples (P = 0.001), and more of the TMD patients were married compared with the SLE and CLBP patients (P < 0.001). However, the proportions of SLE, CLBP, and TMD patients reporting on the MPI the responses of spouses/live-in partner versus other relatives or friends to the patients expressions of pain were not significantly different (c 2 (2, N = 240) = 1.71, P = 0.424). Proportions of CLBP and SLE Table 2 Demographic characteristics of chronic pain groups patients receiving disability compensation were not significantly different (37.5% and 36%, respectively), while no TMD patients received such compensation. Before proceeding with comparative analyses among patient groups, basic psychometric analyses were computed to determine whether the MPI performed similarly in SLE and in other pain populations. Cronbach s coefficient of internal consistency (alpha) is reported in Table 3 for each of the nine MPI scales for SLE patients. Internal consistency of the scales ranged from 0.69 to This is comparable with the internal consistency of MPI scales for the chronic pain norm group used in the original MPI validation study, in which Cronbach s alpha ranged from 0.72 to 0.90 [11]. Table 3 also contains the correlations between scales. The correlations were similar in size and direction to those originally reported in validation studies of the MPI [11], thus suggesting the invariance of the instrument with SLE patients. These analyses suggest that the MPI is psychometrically appropriate for SLE patients, and therefore, their MPI scale scores can be compared with those of other pain populations. Comparison of MPI Scale Scores in Three Chronic Pain Conditions: SLE, CLBP, and TMD A MANOVA with pain condition as the independent variable and the nine MPI scale scores as dependent variables was performed to determine whether the MPI was sensitive to differences among the SLE, CLBP, and TMD groups. This MANOVA was significant (F (18,458, N = 240) = 12.26, P < 0.001). Means, standard deviations (SD), univariate F statistic significance levels, and results of post hoc pairwise comparisons are pro- Variable SLE (N = 80) CLBP (N = 80) TMD (N = 80) P value Age in years mean (SD) 47.7 (9.8) 44.9 (13.4) 44.8 (7.9) Pain duration in years mean (SD) 10.0 (6.6) 6.3 (8.5) 7.9 (8.7) Ethnic group (%) Caucasian African American Asian Hispanic Other Marital status (%) <0.001 Married Separated/divorced Single Widowed Disability compensation (%) <0.001

6 44 Greco et al. Table 3 Internal consistency and intercorrelations of MPI scales in 80 females with SLE Cronbach s Pearson MPI scale alpha correlation 1 PS 0.85 PS I LC AD S PR SR DR I LC AD S PR SR DR GA Abbreviations: PS = Pain Severity; I = Interference; LC = Life Control; AD = Affective Distress; S = Support; PR = Punishing Responses; SR = Solicitous Responses; DR = Distracting Responses; GA = General Activity. 1 r 0.22, P < 0.05; r 0.28, P < vided in Table 4. Significant differences among the three chronic pain groups were found on all MPI scales, with the exception of Life Control. All three groups differed significantly on Pain Severity and Interference with life activities due to pain. The CLBP patients reported significantly greater Pain Severity and pain-related Interference with life activities than the other two groups. The mean Interference score of SLE patients was significantly higher than that of TMD patients, even though their ratings of overall Pain Severity were significantly lower. The observed differences in Interference with activities due to pain may be due to the localized nature of TMD pain and the multiple locations and types of SLE pain. CLBP patients reported less Support and General Activity and greater Affective Distress than SLE and TMD patients. SLE patients, on average, reported receiving fewer Punishing Responses (such as anger, irritability, and blame) from significant others regarding pain than did CLBP patients. Although SLE patients reported multiple pain complaints, this symptom in general did not appear to be more disabling for them than for persons with a single source of pain. In contrast, Table 4 Multidimensional Pain Inventory scale score means (SD) by chronic pain groups the CLBP patients reported relatively greater psychosocial and behavioral disruption due to pain than patients in other pain groups. MPI Classification of SLE, CLBP, and TMD Patients The majority of participants (87.5% of SLE and TMD patients and 99% of CLBP patients) could be accurately classified into one of the three MPI pain profiles. Ten patients (12.5%) in each of the SLE and TMD groups and one patient in the CLBP group had MPI responses that did not show an adequate fit statistically to any of the three profiles, and they were classified as anomalous. Anomalous profiles can be the result of random responding or unusual patterns of responding. In the present sample, the mean Pain Severity and Interference scores of the anomalous participants were significantly lower than those of the AC group (F (1,110, N = 240) = 17.5, P < and F (1,110, N = 240) = 15.54, P < 0.001, respectively). Likewise, perceived Support from others regarding pain was lower in the anomalous group relative to the AC group, perhaps indicating that support regarding pain was not needed. MPI scale SLE (N = 80) CLBP (N = 80) TMD (N = 80) P value 1 Tukey HSD 2 Pain Severity 3.03 (1.45) 4.87 (0.94) 3.54 (1.28) <0.001 CLBP > TMD > SLE Interference 3.25 (1.53) 4.86 (0.94) 2.45 (1.66) <0.001 CLBP > SLE > TMD Life Control 3.30 (1.44) 2.99 (1.58) 3.33 (1.13) Affective Distress 3.21 (1.38) 4.05 (1.35) 3.26 (1.32) <0.001 CLBP > SLE&TMD Support 3.99 (1.53) 4.75 (1.38) 3.56 (1.65) <0.001 CLBP > SLE&TMD Punishing Responses 1.26 (1.43) 1.93 (1.71) 1.74 (1.52) CLBP > SLE Solicitous Responses 3.26 (1.69) 3.86 (1.61) 2.93 (1.76) CLBP > TMD Distracting Responses 2.28 (1.43) 2.59 (1.68) 1.78 (1.47) CLBP > TMD General Activity 2.95 (1.06) 1.99 (0.82) 3.03 (0.78) <0.001 SLE&TMD > CLBP 1 ANOVA degrees of freedom are 2, Tukey HSD comparison: P < 0.05.

7 Pain Profiles in SLE 45 Figure 1 MPI-based psychosocial subgroups in three chronic pain conditions. The distributions of the three MPI profiles in the three pain groups are depicted in Figure 1. There were significant differences overall among pain groups in the psychosocial profiles (c 2 (4, N = 219) = 49.0, P < 0.001). Hierarchical follow-up chi-square analyses indicated the profile distributions between SLE and TMD patients were not significantly different (P = 0.91), but that CLBP patients had a significantly different distribution of profile proportions than SLE and TMD patients (P < 0.001). Compared with the TMD and SLE groups, the CLBP group had a higher proportion of DYS profiles, while the TMD and SLE groups had higher proportions of AC profiles. Construct Validation of the MPI in SLE The MPI scale scores of the three SLE patient profiles are displayed in Figure 2. In order to further examine the appropriateness of the MPI in a SLE sample, the three profiles were compared on psychosocial variables conceptually related to the MPI-based classification, including pain, distress, and perceived physical function variables. It was expected that these related variables would differentiate the profiles. In addition, the profiles of SLE patients were compared on categorical and continuous variables that are conceptually unrelated to the psychosocial classification system, namely age, ethnicity, and disease and pain duration, with the expectation that the profiles would not differ on these variables. Cluster Differences on Categorical Demographic Variables A cluster by ethnicity chi-square test was significant (c 2 (4, N = 70) = 14.04, P = 0.002). Post hoc analyses indicated that a significantly higher proportion of African Americans than Caucasians were classed as DYS (P = 0.003), but no differences were found for ID and AC classifications (P = 0.21). Further exploration of the role of ethnicity in the sample indicated that African Americans Figure 2 MPI profiles of 70 SLE patients.

8 46 Greco et al. and Caucasians with SLE did not differ on current disease activity or severity of psychological distress/self-efficacy, but in the area of pain, African Americans tended to report greater pain severity on the AIMS-2 pain scale and MPQ-SF than Caucasians (F (1,67, N = 70) = 4.53, P = 0.014). Significant differences also were found between cluster assignment and marital status (c 2 (6, N = 70) = 16.23, P = 0.013). Post hoc analyses indicated that married individuals were over-represented in the AC profile, and unmarried participants were over-represented in the DYS profile. Profile Comparisons on Demographics and Disease History The mean scores on age, pain duration, disease history variables, disease activity, pain, distress, and perceived function for each SLE pain profile are provided in Table 5. As expected, the three profile groups were not significantly different in age and pain duration or in disease history variables, including the SLE damage index, years since diagnosis, and length of time since symptom onset. Comparing Profiles on Current Disease Activity The MANOVA on the set of three current disease activity indicators (SLAM-R, SLEDAI, and PGA) among clusters was significant (F (6,130, N = 70) = 3.07, P = 0.008). Follow-up univariate analyses indicated that this was primarily due to differences in SLAM-R scores among clusters (Table 5). Tukey HSD post hoc pairwise comparisons indicated that the DYS group had higher SLAM-R scores than the AC group. The SLAM-R relies heavily on patient reports of pain, such as myalgia, joint pain, and headache, thus, it is not surprising that SLAM-R scores differentiated profiles. These three pain report items were then omitted from the SLAM-R scores and the ANOVA was repeated. Nonetheless, the profiles remained significantly different from one another on the SLAM-R when pain items were omitted (F (2,67, N = 70) = 3.59, P = 0.033), which suggests that the SLAM-R differences were not simply due to higher pain report among DYS patients. Validation of Profiles Using External Psychosocial Variables The MANOVAs of pain, distress, and physical function indicated significant differences among the three profiles, and all univariate analyses of individual variables were significant (Table 5). For the AIMS-2 pain scale, Tukey HSD post hoc pairwise comparisons indicated significant differences among all three groups, with DYS having higher pain than ID, and ID reporting higher pain than AC. The DYS patients SF36-PF and Self-Efficacy scores were lower than those of ID and AC Table 5 Comparison of means (SD) of MPI subgroups of SLE patients on demographic and disease variables and measures of pain, distress, and perceived physical function DYS (N = 11) ID (N = 22) AC (N = 37) P value 1 Demographics Age in years (9.73) (11.34) (7.60) Pain duration years (7.73) 9.82 (4.49) 9.54 (7.01) Disease history Damage index 1.82 (2.40) 1.09 (1.44) 1.41 (1.50) Years diagnosed (6.46) (5.59) (8.29) Symptom years (7.24) (8.97) (9.62) SLE activity SLAM-R (4.25) a (3.28) a,b 8.32 (3.37) b SLEDAI 3.09 (3.30) 2.64 (2.11) 4.41 (3.22) PGA (cm) 1.55 (0.84) 1.20 (0.49) 1.26 (0.71) Pain AIMS-2 pain 3.19 (0.25) a 2.53 (0.66) b 2.11 (0.69) c <0.001 MPQ-SF (9.50) a (7.94) a,b (7.88) b <0.001 Distress CES-D (12.64) a (11.52) a,b (10.43) b <0.001 STRESS 9.18 (2.99) a 7.32 (3.37) a,b 5.46 (3.20) b Self-Efficacy (8.97 a (17.85) b (18.43) b Physical function SF36-PF (15.78) a (20.43) b (25.56) b Hours rest 7.36 (4.80) a 4.59 (3.71) a,b 3.30 (2.71) b Results of MANOVAs are presented for each of the six primary domains, with significance levels in bold face. Significant MANOVAs are followed by univariate ANOVAs (df = 2,67). 2 Means with same letters (superscript a or b) for significant F tests are not significantly different from one another at the P < 0.05 level, based on Tukey HSD post hoc tests.

9 Pain Profiles in SLE 47 patients. For all other variables, pairwise comparisons revealed the DYS profile to be significantly higher in pain, distress, or perceived functional limitation than AC, with ID between the other two profiles and not significantly different from either one. Discussion As medical advances improve mortality related to SLE, attention has broadened to include psychosocial factors associated with outcomes in this population. Pain, coping skills, and psychological distress have repeatedly been identified as important factors that are linked with disease activity. The current study is the first to focus specifically on psychosocial adaptation to the chronic pain that is typically associated with SLE. Among female SLE patients with pain, the majority can be classified into distinct groups based on their responses to the MPI. The same empirically derived clusters, DYS, ID, and AC, apply to chronic pain samples of diverse etiologies, such as TMD, CLBP, fibromyalgia, and headache, as well as those whose pain is secondary to malignant process [13,16,25]. Like other individuals with chronic pain, persons with SLE-related pain are not homogeneous, but can be categorized into distinct groups based on psychosocial and behavioral responses. In comparison with samples of CLBP and TMD patients from the same institution, SLE patients appeared most similar to TMD patients in terms of proportions classified as DYS, ID, and AC, with nearly half of these patients classified as AC. The similarity in profile distributions of SLE and TMD patients is surprising given that the SLE patients in the present study had average pain duration of 10 years and a median of five types of pain complaints. In contrast, TMD patients pain, while chronic and potentially disabling, is limited to the head and neck region. Perhaps for many individuals in the SLE sample, particularly the AC group (46% of the sample), pain is not as debilitating as other aspects of lupus, such as fatigue. Future research is needed to assess the incremental psychosocial effects of pain over and above other problems and symptoms associated with SLE. Further investigations into the pain adaptation of SLE patients should include comparison groups that are also struggling with potentially life-threatening and disabling disease, such as persons with pain related to cancer. The construct validity of the MPI in the SLE sample was supported. The clusters differed on pain and distress variables external to the classification and were not significantly different on conceptually unrelated variables of age, disease duration, and cumulative organ damage. However, there were differences among clusters on one of the measures of current disease activity. Participants classified as DYS had higher scores on the SLAM-R, indicating that their disease was more active at the time of assessment. Previous validity studies of the MPI have found profile group to be independent of physical findings; for example, number of dental exam signs of TMD [14]. However, the SLAM-R includes not only counts of clinical signs and symptoms, but also ratings of the current severity of each. Conceivably, the association of profile group with SLE disease activity may be due to good sensitivity of the measure to current status. Our finding that greater disease activity was associated with the DYS profile, which is characterized by higher pain, distress, and disruption in activities, is consistent with several studies linking psychological distress to current SLE disease status [8 10,43]. Despite the serious and potentially lifethreatening nature of SLE disease, many patients with SLE appear to cope adaptively with their pain by maintaining activities and a positive mood and receiving adequate support from significant others. Similarly high proportions of patients fitting the AC profile (47.4%) have been identified in a study of patients with cancer-related pain [25]. The high proportion of patients fitting the AC profile in our sample may be due to the fact that pain is only one of many problematic issues in SLE and other life-threatening conditions. Fatigue and physical or social role changes are aspects of SLE that may influence psychosocial adaptation in addition to pain and are important to address in future investigations. Although many of the SLE patients in this study appeared to cope well with chronic pain, a substantial proportion of the classified patients exhibited pain-related psychosocial disruption. Of the 87.5% who could be classified, 14% fit the DYS profile (higher perceived pain severity and affective distress, lower activity level) and 27.5% were consistent with the ID profile, exhibiting lower support and perceiving negative reactions from significant others. Other investigators report that SLE patients vary in levels of coping, social support, and psychological distress [8,9,44]. This supports the idea that these patients are not a

10 48 Greco et al. homogenous group that responds to SLE and its symptoms in exactly the same way. The current study highlights the importance of comprehensive assessments of psychosocial and behavioral adaptations of SLE patients that include attention to pain-related variability in psychological distress, relationship difficulties, and activity limitations. Sampling issues may limit the generalizability of this study. The results of this study should be replicated with larger samples that include males with SLE as well as persons with SLE who are attending routine rheumatology clinic visits rather than research-based medical evaluations. Although the SLE sample in the present study was representative of the geographic region and the Pittsburgh Lupus Registry with regard to ethnic groups, it included a high proportion of Caucasians, only one Asian, and no Hispanic individuals. In this study, African Americans were over-represented in the DYS group, which was due to higher pain reports among African Americans. This is consistent with other reports of lower tolerance of African Americans to experimentally induced pain, and higher levels of pain related to medical conditions in comparison with other racial/ethnic groups [45]. However, future studies of pain adaptation in SLE should include ethnically diverse samples in order to assess the impact of ethnicity on pain adaptation. The range of disease damage, duration, and current SLE activity in our sample was consistent with other reports [9,26,41,46], with the majority of patients characterized by mild to moderate disease activity. It is possible that a sample including more individuals with severe disease activity could result in a different distribution of MPI profiles. The present study sample does not represent all SLE cases, because some individuals with SLE do not experience pain. Whether SLE patients who do not have pain would report relationship problems and distress due to fatigue or unpredictable symptoms of lupus is unknown. As a first investigation into the pain adaptation of SLE patients, the current study highlights the multidimensional impact of pain in this population. Conclusions The majority of SLE patients could be classified into MPI-based profiles of dysfunctional, interpersonally distressed, and adaptive coper, indicating that persons with SLE are not homogeneous in terms of their psychosocial and behavioral responses to pain. The profile groups differed from one another on external measures of psychological distress, pain, and perceived physical function, thus providing support for the construct validity of the MPI classification in a SLE population. Although all of the SLE patients in this sample reported multiple sources of pain, a substantial proportion appeared to cope well with pain by maintaining activities and a positive mood. A comparison sample of CLBP patients tended to perceive their pain as more disabling than did SLE patients and TMD patients. The MPI may be useful for identifying individuals with SLE who are in need of psychoeducational or behavioral pain management interventions. Acknowledgments Supported by the National Arthritis Foundation: Robert Wood Johnson Clinical Science Grant, USPHS/NIDCR grant R01 DE07514, NIH/NIAMD grant AR38698, and NIH/NCRR/GCRC grant 5-M01-RR References 1 Gladman DD, Urowitz MB. Systemic lupus erythematosus: Clinical and laboratory features. In: Klippel J, Weyand C, Wortman R, editors. Primer on the rheumatic diseases. 11th edition. Atlanta, GA: Arthritis Foundation; p Sfikakis PP, Mitsikostas DD, Manoussakis MN, et al. Headache in systemic lupus erythematosus: A controlled study. Br J Rheumatol 1998;37: Glanz BI, Venkatesan A, Schur PH, et al. Prevalence of migraine in patients with systemic lupus erythematosus. Headache 2001;41: Omdal R, Waterloo K, Koldingsnes W, et al. Somatic and psychological features of headache in systemic lupus erythematosus. J Rheumatol 2001; 28: Clarke AE, Bloch DA, Danoff DS, et al. Decreasing costs and improving outcomes in systemic lupus erythematosus: using regression trees to develop health policy. J Rheumatol 1994;21: Friedman AW, Alarcon GS, McGwin G Jr, et al. Systemic lupus erythematosus in three ethnic groups. IV. Factors associated with self-reported functional outcome in a large cohort study. Arthritis Care Res 1999;12: Akkasilpa S, Minor M, Goldman D, et al. Association of coping responses with fibromyalgia tender points in patients with systemic lupus erythematosus. J Rheumatol 2000;27: Karlson EW, Daltroy LH, Lew RA, et al. The relationship of socioeconomic status, race, and modifiable risk factors to outcomes in patients with systemic lupus erythematosus. Arthritis Rheum 1997;40:47 56.

11 Pain Profiles in SLE 49 9 Dobkin PL, Fortin PR, Joseph L, et al. Psychosocial contributors to mental and physical health in patients with systemic lupus erythematosus. Arthritis Care Res 1998;11: Da Costa D, Dobkin PL, Pinard L, et al. The role of stress in functional disability among women with systemic lupus erythematosus: A prospective study. Arthritis Care Res 1999;12: Kerns RD, Turk DC, Rudy TE. The West Haven- Yale Multidimensional Pain Inventory (WHYMPI). Pain 1985;23: Turk DC, Rudy TE. Toward an empirically derived taxonomy of chronic pain patients: Integration of psychological assessment data. J Consult Clin Psychol 1988;56: Turk DC, Rudy TE. The robustness of an empirically derived taxonomy of chronic pain patients. Pain 1990;43: Rudy TE, Turk DC, Zaki HS, et al. An empirical taxometric alternative to traditional classification of temporomandibular disorders. Pain 1989;36: Scharff L, Turk DC, Marcus DA. Psychosocial and behavioral characteristics in chronic headache patients: Support for a continuum and dualdiagnostic approach. Cephalalgia 1995;15: Turk DC, Okifuji A, Sinclair JD, et al. Pain, disability, and physical functioning in subgroups of patients with fibromyalgia. J Rheumatol 1996;23: Beck AT, Ward CH, Mendelson M, et al. An inventory for measuring depression. Arch Gen Psychiat 1961;4: Radloff L. The CES-D scale: A self report depression scale for research in the general population. Appl Psychol Meas 1977;1: Locke HJ, Wallace KM. Short-term marital adjustment and prediction tests: Their reliability and validity. J Marriage Fam Liv 1959;21: Wallston KA, Wallston BS, DeVellis R. Development of the Multidimensional Health Locus of Control (MHLC) scales. Health Educ Monogr 1978;6: Rudy TE, Turk DC, Kubinski JA, et al. Differential treatment responses of TMD patients as a function of psychological characteristics. Pain 1995;61: Turk DC, Rudy TE, Kubinski JA, et al. Dysfunctional patients with temporomandibular disorders: Evaluating the efficacy of a tailored treatment protocol. J Consult Clin Psychol 1996;64: Turk DC, Okifuji A, Sinclair JD, et al. Differential responses by psychosocial subgroups of fibromyalgia syndrome patients to an interdisciplinary treatment. Arthritis Care Res 1998;11: Turk DC, Sist TC, Okifuji A, et al. Adaptation to metastatic cancer pain, regional/local cancer pain and non-cancer pain: Role of psychological and behavioral factors. Pain 1998;74: Zaza C, Reyno L, Moulin DE. The Multidimensional Pain Inventory profiles in patients with chronic cancer-related pain: An examination of generalizability. Pain 2000;87: Ward MM, Lotstein DS, Bush TM, et al. Psychosocial correlates of morbidity in women with systemic lupus erythematosus. J Rheumatol 1999; 26: Greco CM, Rudy TE, Manzi S. Stress reduction improves lupus disease activity. Arthritis Rheum 2001;44: Tan EM, Cohen AS, Fries JF et al. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1982;25: Frankel MR, Spencer BD. SAMPLE: A supplementary module for SYSTAT. Evanston, IL: SYSTAT Inc., Bombardier C, Gladman DD, Urowitz MB, et al. Derivation of the SLEDAI. A disease activity index for lupus patients. The Committee on Prognosis Studies in SLE. Arthritis Rheum 1992;35: Liang MH, Socher SA, Larson MG, et al. Reliability and validity of six systems for the clinical assessment of disease activity in systemic lupus erythematosus. Arthritis Rheum 1989;32: Abrahamowicz M, Fortin PR, du Berger R, et al. The relationship between disease activity and expert physician s decision to start major treatment in active systemic lupus erythematosus: A decision aid for development of entry criteria for clinical trials. J Rheumatol 1998;25: Liang MH, Socher SA, Roberts WN, et al. Measurement of systemic lupus erythematosus activity in clinical research. Arthritis Rheum 1988;31: Gladman D, Ginzler E, Goldsmith C, et al. The development and initial validation of the Systemic Lupus International Collaborating Clinics/ American College of Rheumatology damage index for systemic lupus erythematosus. Arthritis Rheum 1996;39: Meenan RF, Gertman PM, Mason JH. Measuring health status in arthritis: The arthritis impact measurement scales. Arthritis Rheum 1980;23: Melzack R. The short-form McGill pain questionnaire. Pain 1987;30: Cohen S, Williamson GM. Perceived stress in a probability sample of the United States. In: Spacapan S, Oskamp S, editors. The social psychology of health. Newberry Park, CA: Sage Publications; p Lorig K, Chastain RL, Ung E, et al. Development and evaluation of a scale to measure perceived selfefficacy in people with arthritis. Arthritis Rheum 1989;32: Ware JE, Snow KK, Kosinski M. SF-36 Health Survey: Manual and interpretation guide. Lincoln, RI: QualityMetric Incorporated, 1993.

12 50 Greco et al. 40 Strand V, Gladman D, Isenberg D, et al. Outcome measures to be used in clinical trials in systemic lupus erythematosus. J Rheumatol 1999;26: Stoll T, Gordon C, Seifert B, et al. Consistency and validity of patient administered assessment of quality of life by the MOS SF-36; its association with disease activity and damage in patients with systemic lupus erythematosus. J Rheumatol 1997; 24: Mehta C, Patel N. StatXact 4 for Windows: Statistical software for exact nonparametric inference. Cambridge, MA: CYTEL Software Corporation, Dobkin PL, Da Costa D, Dritsa M, et al. Quality of life in systemic lupus erythematosus patients during more and less active disease states: Differential contributors to mental and physical health. Arthritis Care Res 1999;12: McCracken LM, Semenchuk EM, Goetsch VL. Cross-sectional and longitudinal analyses of coping responses and health status in persons with systemic lupus erythematosus. Behav Med 1995;20: Edwards CL, Fillingim RB, Keefe F. Race, ethnicity and pain. Pain 2001;94: Bruce IN, Mak VC, Hallett DC, et al. Factors associated with fatigue in patients with systemic lupus erythematosus. Ann Rheum Dis 1999;58: